Organisation mondiale de la santé (OMS) · Technical Documents

WHO recommendations on interventions to improve preterm birth outcomes: evidence base

24 août 2015 Organisation mondiale de la santé

Le texte intégral est hébergé par l’organisation qui le publie. lawenc.com indexe les métadonnées et renvoie vers la source officielle.

Texte intégral

WHO/RHR/15.17

WHO recommendations on interventions to improve preterm birth outcomes: Evidence base

WHO recommendations on interventions to improve preterm birth outcomes: evidence base

WHO/RHR/15.15 © World Health Organization 2015 All rights reserved. Publications of the World Health Organization are available on the WHO website (www.who.int) or can be purchased from WHO Press, World Health Organization, 20 Avenue Appia, 1211 Geneva 27, Switzerland (tel.: +41 22 791 3264; fax: +41 22 791 4857; e-mail: bookorders@who.int). Requests for permission to reproduce or translate WHO publications – whether for sale or for non-commercial distribution – should be addressed to WHO Press through the WHO website (www.who.int/about/ licensing/copyright_form/en/index.html). The designations employed and the presentation of the material in this publication do not imply the expression of any opinion whatsoever on the part of the World Health Organization concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. The mention of specific companies or of certain manufacturers’ products does not imply that they are endorsed or recommended by the World Health Organization in preference to others of a similar nature that are not mentioned. Errors and omissions excepted, the names of proprietary products are distinguished by initial capital letters. All reasonable precautions have been taken by the World Health Organization to verify the information contained in this publication. However, the published material is being distributed without warranty of any kind, either expressed or implied. The responsibility for the interpretation and use of the material lies with the reader. In no event shall the World Health Organization be liable for damages arising from its use.

Contents Table 1a. Table 1b. Table 1c. Table 1d. Table 1e. Table 1f. Table 1g. Table 1h. Table 1i. Table 1j. Table 1k. Table 1l. Table 2a. Table 2b. Table 2c. Table 2d. Table 2e. Table 2f. Table 2g. iii Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women at risk of preterm birth (all women and babies) Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women at risk of preterm birth (gestational age at first dose) Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women at risk of preterm birth (gestational age at birth) Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women at risk of preterm birth (interval to delivery) Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women at risk of preterm birth (singleton and multiple pregnancy subgroups) Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women at risk of preterm birth (preterm prelabour rupture of membranes) Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women at risk of preterm birth (women with chorioamnionitis) Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women undergoing elective caesarean section at late preterm (34—36+6 weeks) Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women at risk of preterm birth (women with hypertension in pregnancy) Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women at risk of preterm birth (women with growth-restricted fetuses) 1 6 12 20 25 28 35 40 42 45

Different corticosteroids and regimens for accelerating fetal lung maturation for women at risk of preterm birth (any regimen of dexamethasone and betamethasone) 49 Repeat course(s) versus single course of antenatal corticosteroids (ACS) for accelerating fetal lung maturation for women at risk of preterm birth Betamimetics for inhibiting preterm labour Calcium channel blockers for inhibiting preterm labour Cyclo-oxygenase (COX) inhibitors for inhibiting preterm labour Magnesium sulfate for inhibiting preterm labour Oxytocin receptor antagonists for inhibiting preterm labour Nitric oxide donors for inhibiting preterm labour Progestational agents for inhibiting preterm labour 52 57 61 62 65 68 71 74

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 2h. Table 2i. Table 2j. Table 2k. Table 2l. Table 2m. Table 3a. Table 3b. Table 3c. Table 3d. Table 3e. Table 4a. Table 4b. Table 5a. Table 5b. Table 5c. Table 5d. Table 6a. Table 7a. Table 7b. Table 7c. Table 7d. Table 7e. Table 8a. Table 8b. Table 9a. Table 9b. Table 9c.

Relaxin for inhibiting preterm labour Hydration for inhibiting preterm labour Maintenance betamimetic therapy for inhibiting preterm labour Magnesium maintenance therapy inhibiting preterm labour Maintenance therapy with calcium channel blockers for inhibiting preterm labour Maintenance therapy with oxytocin antagonists for inhibiting preterm labour Magnesium sulfate for fetal neuroprotection in women at risk of preterm birth (all women and babies) Magnesium sulfate for fetal neuroprotection in women at risk of preterm birth (singleton and multiple pregnancy subgroups) Magnesium sulfate for fetal neuroprotection in women at risk of preterm birth (gestational age at administration) Magnesium sulfate for fetal neuroprotection in women at risk of preterm birth (intention to prevent preterm-birth related neurological complications) Magnesium sulfate for fetal neuroprotection in women at risk of preterm birth (retreatment) Antibiotic prophylaxis for women at risk of preterm birth and with intact membranes (any antibiotics) Antibiotic prophylaxis for women at risk of preterm birth and with intact membranes (antibiotic regimen) Antibiotic prophylaxis for women at risk of preterm birth and ruptured membranes Antibiotic prophylaxis for women at risk of preterm birth and ruptured membranes (antibiotic regimens) Antibiotic prophylaxis for women at risk of preterm birth and ruptured membranes (erythromycin versus co-amoxiclav) Antibiotic prophylaxis for women at risk of preterm birth and ruptured membranes (3-day versus 7-day ampicillin regimens) Mode of delivery for women at risk of preterm birth Kangaroo mother care (KMC) versus conventional care for preterm newborns Continuous Kangaroo mother care (KMC) versus conventional care for preterm newborns Intermittent Kangaroo mother care (KMC) versus conventional care for preterm newborns Radiant warmers versus incubators for care of unstable or sick preterm newborns Plastic bags or wraps versus conventional care immediately after birth in preterm (and some term) newborns Continuous positive airway pressure (CPAP) therapy for preterm newborns with respiratory distress syndrome Timing of initiation (early versus late) of continuous positive airway pressure (CPAP) therapy for preterm newborns with respiratory distress syndrome Surfactant replacement therapy with animal-derived surfactants for preterm newborns with respiratory distress syndrome Surfactant replacement therapy with protein-free synthetic surfactants for preterm newborns with respiratory distress syndrome Protein-free synthetic surfactant treatment or prophylaxis versus natural surfactant therapy for preterm newborns with respiratory distress syndrome

76 89 80 83 84 88 90 94 97 99 103 106 109 117 120 123 125 127 132 134 135 136 138 139 140 142 143 144

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 9d. Table 9e. Table 9f. Table 9g. Table 9h.

Protein-containing synthetic surfactant treatment or prophylaxis versus natural surfactant therapy for preterm newborns with respiratory distress syndrome Prophylactic surfactant replacement therapy versus rescue surfactant therapy with or without continuous positive airway pressure (CPAP) for preterm newborns with respiratory distress syndrome Prophylactic surfactant replacement therapy versus rescue surfactant therapy without continuous positive airway pressure (CPAP) for preterm newborns with respiratory distress syndrome Prophylactic surfactant replacement therapy versus rescue surfactant therapy with continuous positive airway pressure (CPAP) for preterm newborns with respiratory distress syndrome Early surfactant replacement therapy (within 2—3 hours of birth) versus late rescue surfactant therapy (after waiting for symptoms to worsen) with or without continuous positive airway pressure (CPAP) for preterm newborns with respiratory distress syndrome

145 147 149 150 152

Abbreviations: CI: confidence interval; MD: mean difference; OR: odds ratio; RR: relative risk

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 1a. Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women at risk of preterm birth (all women and babies) Source: Roberts D, Dalziel SR. Antenatal corticosteroids for accelerating fetal lung maturation for women at risk of preterm birth. Cochrane Database Syst Rev. 2006;(3):CD004454. (updated for the guideline) Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations No. of patients ACS Placebo or no treatment Relative (95% CI) Effect Absolute Quality Importance

Maternal death 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 1/188 (0.5%) 1/177 (0.6%) RR 0.98 (0.06 to 15.50) 0 fewer per 1000 (from 5 fewer to 82 more) 13 fewer per 1000 (from 37 fewer to 53 more) 8 fewer per 1000 (from 25 fewer to 14 more) 30 more per 1000 (from 6 fewer to 82 more) MD 0.23 higher (1.86 lower to 2.32 higher) 43 fewer per 1000 (from 21 fewer to 62 fewer) 1 fewer per 1000 (from 13 fewer to 15 more) ⊕⊕OO LOW CRITICAL

Maternal admission into intensive care unit 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 6/160 (3.8%) 8/159 (5.0%) RR 0.74 (0.26 to 2.05) ⊕⊕OO LOW CRITICAL

Chorioamnionitis 13 randomized trials serious2 no serious inconsistency no serious indirectness serious3 none 91/1254 (7.3%) 101/1271 (7.9%) RR 0.90 (0.69 to 1.17) ⊕⊕OO LOW CRITICAL

Puerperal sepsis 8 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 57/496 (11.5%) 44/507 (8.7%) RR 1.35 (0.93 to 1.95) ⊕⊕⊕O MODERATE CRITICAL

Mean interval between trial entry and birth (days) (better indicated by higher values) 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 749 764 — ⊕⊕⊕O MODERATE CRITICAL

Fetal and neonatal death 13 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 261/1813 (14.4%) 341/1814 (18.8%) RR 0.77 (0.67 to 0.89) ⊕⊕⊕⊕ HIGH CRITICAL

Fetal death 13 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 86/1813 (4.7%) 89/1814 (4.9%) RR 0.98 (0.73 to 1.30) ⊕⊕⊕O MODERATE CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations

No. of patients ACS Placebo or no treatment Relative (95% CI)

Effect Absolute Quality Importance

Neonatal death 21 randomized trials serious2 no serious inconsistency no serious indirectness no serious imprecision none 210/2218 (9.5%) 306/2190 (14.0%) RR 0.68 (0.58 to 0.80) 45 fewer per 1000 (from 28 fewer to 59 fewer) 14 fewer per 1000 (from 27 fewer to 11 more) 0 fewer per 1000 (from 19 fewer to 34 more) 85 fewer per 1000 (from 65 fewer to 102 fewer) 77 fewer per 1000 (from 49 fewer to 97 fewer) MD 1.42 lower (2.28 to 0.56 lower) MD 2.86 lower (5.51 to 0.21 lower) 38 fewer per 1000 (from 70 fewer to 7 more) ⊕⊕⊕O MODERATE CRITICAL

Childhood death 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 16/537 (3.0%) 20/473 (4.2%) RR 0.68 (0.36 to 1.27) ⊕⊕⊕O MODERATE CRITICAL

Death in adulthood 1 randomized trials serious4 no serious inconsistency no serious indirectness serious3 none 21/493 (4.3%) 21/495 (4.2%) RR 1.00 (0.56 to 1.81) ⊕⊕OO LOW CRITICAL

Respiratory distress syndrome 25 randomized trials serious2 no serious inconsistency no serious indirectness no serious imprecision none 369/2310 (16.0%) 553/2280 (24.3%) RR 0.65 (0.58 to 0.73) ⊕⊕⊕O MODERATE CRITICAL

Moderate/severe respiratory distress syndrome 6 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 81/835 (9.7%) 145/851 (17.0%) RR 0.55 (0.43 to 0.71) ⊕⊕⊕⊕ HIGH CRITICAL

Mean duration of mechanical ventilation/continuous positive airway pressure (days) (better indicated by lower values) 3 randomized trials no serious risk of bias serious5 no serious indirectness no serious imprecision none 264 254 — ⊕⊕⊕O MODERATE CRITICAL

Mean duration of oxygen supplementation (days) (better indicated by lower values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious6 none 28 45 — ⊕⊕⊕O MODERATE CRITICAL

Surfactant use 4 randomized trials serious2 no serious inconsistency no serious indirectness serious3 none 42/392 (10.7%) 56/384 (14.6%) RR 0.74 (0.52 to 1.05) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations

No. of patients ACS Placebo or no treatment Relative (95% CI)

Effect Absolute Quality Importance

Chronic lung disease 6 randomized trials serious2 serious5 no serious indirectness serious3 none 48/413 (11.6%) 50/405 (12.3%) RR 0.86 (0.61 to 1.22) 17 fewer per 1000 (from 48 fewer to 27 more) 50 fewer per 1000 (from 34 fewer to 62 fewer) 36 fewer per 1000 (from 12 fewer to 52 fewer) 34 fewer per 1000 (from 16 fewer to 45 fewer) 10 more per 1000 (from 42 fewer to 80 more) MD 6.93 lower (39.41 lower to 25.55 higher) 48 fewer per 1000 (from 109 fewer to 24 more) MD 0 higher (1.08 lower to 1.09 higher) ⊕OOO VERY LOW CRITICAL

Cerebroventricular haemorrhage 13 randomized trials serious2 no serious inconsistency no serious indirectness no serious imprecision none 88/1445 (6.1%) 155/1427 (10.9%) RR 0.54 (0.43 to 0.69) ⊕⊕⊕O MODERATE CRITICAL

Systemic infection in the first 48 hours of life 6 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 33/685 (4.8%) 57/674 (8.5%) RR 0.57 (0.38 to 0.86) ⊕⊕⊕⊕ HIGH CRITICAL

Necrotizing enterocolitis 8 randomized trials serious2 no serious inconsistency no serious indirectness no serious imprecision none 25/853 (2.9%) 52/822 (6.3%) RR 0.46 (0.29 to 0.74) ⊕⊕⊕O MODERATE CRITICAL

Small for gestational age 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 73/367 (19.9%) 63/331 (19%) RR 1.05 (0.78 to 1.42) ⊕⊕⊕O MODERATE CRITICAL

Mean birth weight (g) (better indicated by higher values) 13 randomized trials serious2 no serious inconsistency no serious indirectness serious3 none 1498 1463 — ⊕⊕OO LOW CRITICAL

Admission to neonatal intensive care unit 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 112/314 (35.7%) 127/315 (40.3%) RR 0.88 (0.73 to 1.06) ⊕⊕⊕O MODERATE IMPORTANT

Mean duration of neonatal hospitalization (days) (better indicated by lower values) 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 323 318 — ⊕⊕⊕O MODERATE CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations

No. of patients ACS Placebo or no treatment Relative (95% CI)

Effect Absolute Quality Importance

Cerebral palsy in childhood 5 randomized trials serious2 no serious inconsistency no serious indirectness serious3 none 20/490 (4.1%) 28/414 (6.8%) RR 0.60 (0.34 to 1.03) 27 fewer per 1000 (from 45 fewer to 2 more) 38 fewer per 1000 (from 57 fewer to 0 more) 75 fewer per 1000 (from 127 fewer to 38 more) 5 fewer per 1000 (from 15 fewer to 134 more) 34 fewer per 1000 (from 81 fewer to 186 more) 6 fewer per 1000 (from 26 fewer to 32 more) 27 fewer per 1000 (from 126 fewer to 212 more) 21 fewer per 1000 (from 107 fewer to 126 more) ⊕⊕OO LOW CRITICAL

Developmental delay in childhood 2 randomized trials serious2 no serious inconsistency no serious indirectness no serious imprecision none 11/266 (4.1%) 19/252 (7.5%) RR 0.49 (0.24 to 1.00) ⊕⊕⊕O MODERATE CRITICAL

Visual impairment in childhood 2 randomized trials serious2 no serious inconsistency no serious indirectness serious3 none 9/100 (9.0%) 11/66 (16.7%) RR 0.55 (0.24 to 1.23) ⊕⊕OO LOW CRITICAL

Hearing impairment in childhood 2 randomized trials serious2 no serious inconsistency no serious indirectness very serious7 none 1/100 (1.0%) 1/66 (1.5%) RR 0.64 (0.04 to 9.87) ⊕OOO VERY LOW CRITICAL

Neurodevelopmental delay in childhood 1 randomized trials serious4 no serious inconsistency no serious indirectness very serious7 none 3/50 (6.0%) 3/32 (9.4%) RR 0.64 (0.14 to 2.98) ⊕OOO VERY LOW CRITICAL

Intellectual impairment in childhood 3 randomized trials serious2 no serious inconsistency no serious indirectness serious3 none 16/409 (3.9%) 17/369 (4.6%) RR 0.86 (0.44 to 1.69) ⊕⊕OO LOW CRITICAL

Behavioural/learning difficulties in childhood 1 randomized trials serious4 no serious inconsistency no serious indirectness very serious7 none 9/54 (16.7%) 7/36 (19.4%) RR 0.86 (0.35 to 2.09) ⊕OOO VERY LOW CRITICAL

Visual impairment in adulthood 1 randomized trials serious4 no serious inconsistency no serious indirectness very serious8 none 18/87 (20.7%) 24/105 (22.9%) RR 0.91 (0.53 to 1.55) ⊕OOO VERY LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations

No. of patients ACS Placebo or no treatment Relative (95% CI)

Effect Absolute Quality Importance

Hearing impairment in adulthood 1 randomized trials serious4 no serious inconsistency no serious indirectness very serious7 none 1/87 (1.1%) 5/105 (4.8%) RR 0.24 (0.03 to 2.03) 36 fewer per 1000 (from 46 fewer to 49 more) 11 fewer per 1000 (from 14 fewer to 57 more) ⊕OOO VERY LOW CRITICAL

Intellectual impairment in adulthood 2 randomized trials serious2 no serious inconsistency no serious indirectness very serious7 none 0/135 (0.0%) 2/138 (1.4%) RR 0.24 (0.01 to 4.95) ⊕OOO VERY LOW CRITICAL

1 2 3 4 5 6 7 8

Wide confidence interval crossing the line of no effect and few events. Most of the pooled effect provided by studies with design limitations. Wide confidence interval crossing the line of no effect. One study with design limitations. Statistical heterogeneity (I2 > 60%). Estimate based on small sample size. Wide confidence interval crossing the line of no effect, few events and small sample size. Wide confidence interval crossing the line of no effect and small sample size.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 1b. Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women at risk of preterm birth (gestational age at first dose) Source: Roberts D, Dalziel SR. Antenatal corticosteroids for accelerating fetal lung maturation for women at risk of preterm birth. Cochrane Database Syst Rev. 2006;(3):CD004454. (updated for the guideline) Quality assessment No. of patients Effect Quality Importance

No. of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

ACS

Placebo or no treatment (subgroups by gestational age at 1st dose) 3/24 (12.5%)

Relative (95% CI)

Absolute

Chorioamnionitis — in women < 26 weeks of gestation at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 6/22 (27.3%) RR 2.18 (0.62 to 7.69) 148 more per 1000 (from 47 fewer to 836 more) 7 more per 1000 (from 56 fewer to 131 more) 56 fewer per 1000 (from 10 fewer to 67 fewer) 21 fewer per 1000 (from 35 fewer to 32 more) 25 fewer per 1000 (from 30 fewer to 71 more) not pooled ⊕⊕OO LOW CRITICAL

Chorioamnionitis — in women between 26 and < 30 weeks at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 17/129 (13.2%) 14/113 (12.4%) RR 1.06 (0.55 to 2.06) ⊕⊕OO LOW CRITICAL

Chorioamnionitis — in women between 30 and < 33 weeks at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 2/150 (1.3%) 10/144 (6.9%) RR 0.19 (0.04 to 0.86) ⊕⊕⊕O MODERATE CRITICAL

Chorioamnionitis — in women between 33 and < 35 weeks at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 3/158 (1.9%) 7/175 (4.0%) RR 0.47 (0.12 to 1.80) ⊕⊕OO LOW CRITICAL

Chorioamnionitis — in women between 35 and < 37 weeks at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 0/81 (0.0%) 3/100 (3.0%) RR 0.18 (0.01 to 3.36) ⊕⊕OO LOW CRITICAL

Chorioamnionitis — in women > 36 weeks at 1st dose 1 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious5 none 0/16 (0.0%) 15/23 (65.2%) 0/24 (0.0%) 17/26 (65.4%) not pooled ⊕⊕OO LOW CRITICAL

Fetal and neonatal deaths — in babies < 26 weeks at 1st dose 1 very serious2 none RR 1.00 (0.66 to 1.50) 0 fewer per 1000 (from 222 fewer to 327 more) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment

No. of patients

Effect Quality Importance

No. of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

ACS

Placebo or no treatment (subgroups by gestational age at 1st dose) 54/121 (44.6%)

Relative (95% CI)

Absolute

Fetal and neonatal deaths — in babies between 26 and < 30 weeks at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 50/140 (35.7%) RR 0.80 (0.59 to 1.08) 89 fewer per 1000 (from 183 fewer to 36 more) 80 fewer per 1000 (from 127 fewer to 2 more) 10 more per 1000 (from 40 fewer to 102 more) 6 more per 1000 (from 21 fewer to 139 more) — ⊕⊕OO LOW CRITICAL

Fetal and neonatal deaths — in babies between 30 and < 33 weeks at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious6 none 19/165 (11.5%) 30/154 (19.5%) RR 0.59 (0.35 to 1.01) ⊕⊕⊕O MODERATE CRITICAL

Fetal and neonatal deaths — in babies between 33 and < 35 weeks at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious6 none 18/168 (10.7%) 18/185 (9.7%) RR 1.10 (0.59 to 2.05) ⊕⊕⊕O MODERATE CRITICAL

Fetal and neonatal deaths — in babies between 35 and < 37 weeks at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 3/87 (3.4%) 3/107 (2.8%) RR 1.23 (0.25 to 5.94) ⊕⊕OO LOW CRITICAL

Fetal and neonatal deaths — in babies > 36 weeks at 1st dose 1 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious1 none 3/18 (16.7%) 8/23 (34.8%) 0/24 (0.0%) 14/26 (53.8%) RR 9.21 (0.51 to 167.82) RR 0.65 (0.33 to 1.25) ⊕⊕OO LOW CRITICAL

Fetal deaths — in babies < 26 weeks at 1st dose 1 very serious1 none 188 fewer per 1000 (from 361 fewer to 135 more) 27 more per 1000 (from 40 fewer to 155 more) 30 fewer per 1000 (from 63 fewer to 42 more) ⊕⊕OO LOW CRITICAL

Fetal deaths — in babies between 26 and < 30 weeks at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 20/140 (14.3%) 14/121 (11.6%) RR 1.23 (0.65 to 2.34) ⊕⊕OO LOW CRITICAL

Fetal deaths — in babies between 30 and < 33 weeks at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 10/165 (6.1%) 14/154 (9.1%) RR 0.67 (0.31 to 1.46) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment

No. of patients

Effect Quality Importance

No. of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

ACS

Placebo or no treatment (subgroups by gestational age at 1st dose) 7/185 (3.8%)

Relative (95% CI)

Absolute

Fetal deaths — in babies between 33 and < 35 weeks at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 7/168 (4.2%) RR 1.10 (0.39 to 3.07) 4 more per 1000 (from 23 fewer to 78 more) 14 more per 1000 (from 7 fewer to 240 more) not pooled ⊕⊕OO LOW CRITICAL

Fetal deaths — in babies between 35 and < 37 weeks at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 2/87 (2.3%) 1/107 (0.9%) RR 2.46 (0.23 to 26.68) ⊕⊕OO LOW CRITICAL

Fetal deaths — in babies > 36 weeks at 1st dose 1 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious5 none 0/18 (0.0%) 7/15 (46.7%) 0/24 (0.0%) 3/12 (25.0%) not pooled ⊕⊕OO LOW CRITICAL

Neonatal deaths — in babies < 26 weeks at 1st dose 1 very serious1 none RR 1.87 (0.61 to 5.72) 218 more per 1000 (from 97 fewer to 1000 more) 123 fewer per 1000 (from 4 fewer to 206 fewer) 56 fewer per 1000 (from 88 fewer to 13 more) 7 more per 1000 (from 32 fewer to 91 more) 7 fewer per 1000 (from 18 fewer to 109 more) ⊕⊕OO LOW CRITICAL

Neonatal deaths — in babies between 26 and < 30 weeks at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 30/120 (25.0%) 40/107 (37.4%) RR 0.67 (0.45 to 0.99) ⊕⊕⊕⊕ HIGH CRITICAL

Neonatal deaths — in babies between 30 and < 33 weeks at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 9/155 (5.8%) 16/140 (11.4%) RR 0.51 (0.23 to 1.11) ⊕⊕OO LOW CRITICAL

Neonatal deaths — in babies between 33 and < 35 weeks at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 11/161 (6.8%) 11/178 (6.2%) RR 1.11 (0.49 to 2.48) ⊕⊕OO LOW CRITICAL

Neonatal deaths — in babies between 35 and < 37 weeks at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 1/85 (1.2%) 2/106 (1.9%) RR 0.62 (0.06 to 6.76) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment

No. of patients

Effect Quality Importance

No. of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

ACS

Placebo or no treatment (subgroups by gestational age at 1st dose) 4/263 (1.5%)

Relative (95% CI)

Absolute

Neonatal deaths — in babies between 34 and < 37 weeks at 1st dose 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 1/248 (0.4%) RR 0.37 (0.06 to 2.26) 10 fewer per 1000 (from 14 fewer to 19 more) — ⊕⊕OO LOW CRITICAL

Neonatal deaths — in babies > 36 weeks at 1st dose 1 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious1 none 3/18 (16.7%) 4/14 (28.6%) 0/24 (0.0%) 1/10 (10.0%) RR 9.21 (0.51 to 167.82) RR 2.86 (0.37 to 21.87) ⊕⊕OO LOW CRITICAL

Respiratory distress syndrome — in babies < 26 weeks at 1st dose 1 very serious1 none 186 more per 1000 (from 63 fewer to 1000 more) 226 fewer per 1000 (from 124 fewer to 292 fewer) 108 fewer per 1000 (from 32 fewer to 157 fewer) 72 fewer per 1000 (from 14 fewer to 106 fewer) 15 fewer per 1000 (from 34 fewer to 87 more) 21 fewer per 1000 (from 34 fewer to 11 more) ⊕⊕OO LOW CRITICAL

Respiratory distress syndrome — in babies between 26 and < 30 weeks at 1st dose 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious7 none 27/129 (20.9%) 50/113 (44.2%) RR 0.49 (0.34 to 0.72) ⊕⊕⊕O MODERATE CRITICAL

Respiratory distress syndrome — in babies between 30 and < 33 weeks at 1st dose 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 25/186 (13.4%) 43/175 (24.6%) RR 0.56 (0.36 to 0.87) ⊕⊕⊕⊕ HIGH CRITICAL

Respiratory distress syndrome — in babies between 33 and < 35 weeks at 1st dose 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 18/212 (8.5%) 34/222 (15.3%) RR 0.53 (0.31 to 0.91) ⊕⊕⊕⊕ HIGH CRITICAL

Respiratory distress syndrome — in babies between 35 and < 37 weeks at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 2/85 (2.4%) 4/104 (3.8%) RR 0.61 (0.11 to 3.26) ⊕⊕OO LOW CRITICAL

Respiratory distress syndrome — in babies between 34 and < 37 weeks at 1st dose 3 randomized trials serious8 no serious inconsistency no serious indirectness very serious4 none 6/298 (2.0%) 13/311 (4.2%) RR 0.49 (0.19 to 1.26) ⊕OOO VERY LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment

No. of patients

Effect Quality Importance

No. of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

ACS

Placebo or no treatment (subgroups by gestational age at 1st dose) 0/24 (0.0%) 2/12 (16.7%)

Relative (95% CI)

Absolute

Respiratory distress syndrome — in babies > 36 weeks at 1st dose 1 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious5 none 0/16 (0.0%) 3/15 (20.0%) not pooled not pooled ⊕⊕OO LOW CRITICAL

Cerebroventricular haemorrhage — in babies < 26 weeks at 1st dose 1 very serious1 none RR 1.20 (0.24 to 6.06) 33 more per 1000 (from 127 fewer to 843 more) 92 fewer per 1000 (from 8 fewer to 132 fewer) 22 fewer per 1000 (from 28 fewer to 29 more) 2 more per 1000 (from 13 fewer to 74 more) not pooled ⊕⊕OO LOW CRITICAL

Cerebroventricular haemorrhage — in babies between 26 and < 30 weeks at 1st dose 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious7 none 9/121 (7.4%) 18/108 (16.7%) RR 0.45 (0.21 to 0.95) ⊕⊕⊕O MODERATE CRITICAL

Cerebroventricular haemorrhage — in babies between 30 and < 33 weeks at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 1/155 (0.6%) 4/140 (2.9%) RR 0.23 (0.03 to 2.00) ⊕⊕OO LOW CRITICAL

Cerebroventricular haemorrhage — in babies between 33 and < 35 weeks at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 3/161 (1.9%) 3/178 (1.7%) RR 1.11 (0.23 to 5.40) ⊕⊕OO LOW CRITICAL

Cerebroventricular haemorrhage — in babies between 35 and < 37 weeks at 1st dose 1 randomized trials randomized trials randomized trials no serious risk of bias no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness no serious indirectness very serious5 none 0/85 (0.0%) 0/18 (0.0%) 23 0/106 (0.0%) 0/24 (0.0%) 26 not pooled ⊕⊕OO LOW CRITICAL

Cerebroventricular haemorrhage — in babies > 36 weeks at 1st dose 1 very serious5 none not pooled not pooled ⊕⊕OO LOW CRITICAL

Mean birth weight (g) — in babies < 26 weeks at 1st dose (better indicated by higher values) 1 very serious2 none — MD 63.14 higher (607.37 lower to 733.65 higher) MD 26.41 higher (215.55 lower to 268.37 higher) ⊕⊕OO LOW CRITICAL

Mean birth weight (g) — in babies between 26 and < 30 weeks at 1st dose (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 140 121 — ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment

No. of patients

Effect Quality Importance

No. of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

ACS

Placebo or no treatment (subgroups by gestational age at 1st dose) 154

Relative (95% CI)

Absolute

Mean birth weight (g) — in babies between 30 and < 33 weeks at 1st dose (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 165 — MD 190.64 lower (359.98 to 21.30 lower) MD 38.72 lower (172.29 lower to 94.85 higher) MD 13.57 lower (175.45 lower to 148.31 higher) MD 3.51 higher (41.98 lower to 49 higher) MD 73.89 higher (270.89 lower to 418.67 higher) ⊕⊕⊕⊕ HIGH CRITICAL

Mean birth weight (g) — in babies between 33 and < 35 weeks at 1st dose (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious6 none 168 185 — ⊕⊕⊕O MODERATE CRITICAL

Mean birth weight (g) — in babies between 35 and < 37 weeks at 1st dose (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 87 107 — ⊕⊕OO LOW CRITICAL

Mean birth weight (g) — in babies between 34 and < 37 weeks at 1st dose (better indicated by higher values) 3 randomized trials serious8 no serious inconsistency no serious indirectness serious6 none 280 287 — ⊕⊕OO LOW CRITICAL

Mean birth weight (g) — in babies > 36 weeks at 1st dose (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 18 24 — ⊕⊕OO LOW CRITICAL

1 Wide confidence interval crossing the line of no effect, few events and small sample size. 2 Wide confidence interval crossing the line of no effect and small sample size. 3 Estimate based on few events and small sample size. 4 Wide confidence interval crossing the line of no effect and few events. 5 No events. 6 Wide confidence interval crossing the line of no effect. 7 Few events and small sample size. 8 Most studies contributing data had design limitations.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 1c.

Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women at risk of preterm birth (gestational age at birth) Source: Roberts D, Dalziel SR. Antenatal corticosteroids for accelerating fetal lung maturation for women at risk of preterm birth. Cochrane Database Syst Rev. 2006;(3):CD004454. (updated for the guideline) Quality assessment No. of patients Imprecision Other considerations ACS Placebo or no treatment (subgroups by gestational age at birth) 11/46 (23.9%) Relative (95% CI) Effect Absolute Quality Importance

No. of studies

Design

Risk of bias

Inconsistency

Indirectness

Chorioamnionitis — in women delivering < 28 weeks of gestation 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 10/45 (22.2%) RR 0.93 (0.44 to 1.97) 17 fewer per 1000 (from 134 fewer to 232 more) 15 more per 1000 (from 75 fewer to 178 more) 36 fewer per 1000 (from 88 fewer to 55 more) 40 fewer per 1000 (from 75 fewer to 15 more) 27 fewer per 1000 (from 53 fewer to 14 more) 11 fewer per 1000 (from 20 fewer to 17 more) ⊕⊕OO LOW CRITICAL

Chorioamnionitis — in women delivering < 30 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 19/91 (20.9%) 18/93 (19.4%) RR 1.08 (0.61 to 1.92) ⊕⊕OO LOW CRITICAL

Chorioamnionitis — in women delivering < 32 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 21/165 (12.7%) 25/154 (16.2%) RR 0.78 (0.46 to 1.34) ⊕⊕OO LOW CRITICAL

Chorioamnionitis — in women delivering < 34 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 25/283 (8.8%) 34/264 (12.9%) RR 0.69 (0.42 to 1.12) ⊕⊕OO LOW CRITICAL

Chorioamnionitis — in women delivering < 36 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 27/401 (6.7%) 37/392 (9.4%) RR 0.71 (0.44 to 1.15) ⊕⊕OO LOW CRITICAL

Chorioamnionitis — in women delivering ≥ 34 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 5/337 (1.5%) 10/391 (2.6%) RR 0.58 (0.20 to 1.68) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (subgroups by gestational age at birth) 2/240 (0.8%) Relative (95% CI)

Effect Absolute Quality Importance

Chorioamnionitis — in women delivering ≥ 36 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 2/202 (1.0%) RR 1.19 (0.17 to 8.36) 2 more per 1000 (from 7 fewer to 61 more) 146 fewer per 1000 (from 269 fewer to 8 more) 97 fewer per 1000 (from 209 fewer to 35 more) 143 fewer per 1000 (from 59 fewer to 212 fewer) 107 fewer per 1000 (from 36 fewer to 166 fewer) 72 fewer per 1000 (from 17 fewer to 112 fewer) 8 more per 1000 (from 20 fewer to 56 more) ⊕⊕OO LOW CRITICAL

Fetal and neonatal deaths — in babies born < 28 weeks 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 39/60 (65%) 53/69 (76.8%) RR 0.81 (0.65 to 1.01) ⊕⊕OO LOW CRITICAL

Fetal and neonatal deaths — in babies born < 30 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 59/99 (59.6%) 71/102 (69.6%) RR 0.86 (0.7 to 1.05) ⊕⊕OO LOW CRITICAL

Fetal and neonatal deaths — in babies born < 32 weeks 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 82/230 (35.7%) 110/223 (49.3%) RR 0.71 (0.57 to 0.88) ⊕⊕⊕⊕ HIGH CRITICAL

Fetal and neonatal deaths — in babies born < 34 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 90/312 (28.8%) 113/286 (39.5%) RR 0.73 (0.58 to 0.91) ⊕⊕⊕⊕ HIGH CRITICAL

Fetal and neonatal deaths — in babies born < 36 weeks 2 randomized trials no serious risk of bias serious2 no serious indirectness no serious imprecision none 107/498 (21.5%) 135/471 (28.7%) RR 0.75 (0.61 to 0.94) ⊕⊕OO MODERATE CRITICAL

Fetal and neonatal deaths — in babies born ≥ 34 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 24/361 (6.6%) 24/409 (5.9%) RR 1.13 (0.66 to 1.96) ⊕⊕OO MODERATE CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (subgroups by gestational age at birth) 3/264 (1.1%) Relative (95% CI)

Effect Absolute Quality Importance

Fetal and neonatal deaths — in babies born ≥ 36 weeks 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 10/234 (4.3%) RR 3.25 (0.99 to 10.66) 26 more per 1000 (from 0 fewer to 110 more) 127 fewer per 1000 (from 221 fewer to 33 more) 41 fewer per 1000 (from 129 fewer to 99 more) 14 fewer per 1000 (from 65 fewer to 65 more) 29 fewer per 1000 (from 71 fewer to 32 more) 17 fewer per 1000 (from 46 fewer to 26 more) 7 fewer per 1000 (from 22 fewer to 27 more) ⊕⊕OO LOW CRITICAL

Fetal deaths — in babies born < 28 weeks 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 15/60 (25.0%) 25/69 (36.2%) RR 0.65 (0.39 to 1.09) ⊕⊕⊕O MODERATE CRITICAL

Fetal deaths — in babies born < 30 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 none 23/99 (23.2%) 28/102 (27.5%) RR 0.85 (0.53 to 1.36) ⊕⊕OO LOW CRITICAL

Fetal deaths — in babies born < 32 weeks 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 37/230 (16.1%) 38/223 (17.0%) RR 0.92 (0.62 to 1.38) ⊕⊕⊕O MODERATE CRITICAL

Fetal deaths — in babies born < 34 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 39/312 (12.5%) 44/286 (15.4%) RR 0.81 (0.54 to 1.21) ⊕⊕⊕O MODERATE CRITICAL

Fetal deaths — in babies born < 36 weeks 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 47/498 (9.4%) 53/471 (11.3%) RR 0.85 (0.59 to 1.23) ⊕⊕⊕O MODERATE CRITICAL

Fetal deaths — in babies born ≥ 34 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 10/361 (2.8%) 14/409 (3.4%) RR 0.81 (0.36 to 1.80) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (subgroups by gestational age at birth) 0/264 (0.0%) 28/44 (63.6%) Relative (95% CI)

Effect Absolute Quality Importance

Fetal deaths — in babies born ≥ 36 weeks 2 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious4 none 2/234 (0.9%) 24/45 (53.3%) RR 5.92 (0.29 to 122.63) RR 0.79 (0.56 to 1.12) — ⊕⊕OO LOW CRITICAL

Neonatal deaths — in babies born < 28 weeks 2 very serious5 none 134 fewer per 1000 (from 280 fewer to 76 more) 105 fewer per 1000 (from 232 fewer to 64 more) 160 fewer per 1000 (from 78 fewer to 222 fewer) 78 fewer per 1000 (from 20 fewer to 120 fewer) 63 fewer per 1000 (from 16 fewer to 98 fewer) 10 fewer per 1000 (from 13 fewer to 38 more) ⊕⊕OO LOW CRITICAL

Neonatal deaths — in babies born < 30 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 none 36/76 (47.4%) 43/74 (58.1%) RR 0.82 (0.60 to 1.11) ⊕⊕OO LOW CRITICAL

Neonatal deaths — in babies born < 32 weeks 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 45/193 (23.3%) 72/185 (38.9%) RR 0.59 (0.43 to 0.80) ⊕⊕⊕⊕ HIGH CRITICAL

Neonatal deaths — in babies born < 34 weeks 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 65/372 (17.5%) 86/343 (25.1%) RR 0.69 (0.52 to 0.92) ⊕⊕⊕⊕ HIGH CRITICAL

Neonatal deaths — in babies born < 36 weeks 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 60/451 (13.3%) 82/418 (19.6%) RR 0.68 (0.50 to 0.92) ⊕⊕⊕⊕ HIGH CRITICAL

Neonatal deaths — in babies born < 37 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 0/163 (0.0%) 2/157 (1.3%) RR 0.19 (0.01 to 3.98) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (subgroups by gestational age at birth) 10/426 (2.3%) Relative (95% CI)

Effect Absolute Quality Importance

Neonatal deaths — in babies born ≥ 34 weeks 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 14/382 (3.7%) RR 1.58 (0.71 to 3.50) 14 more per 1000 (from 7 fewer to 59 more) 18 more per 1000 (from 3 fewer to 87 more) 113 fewer per 1000 (from 252 fewer to 97 more) 213 fewer per 1000 (from 84 fewer to 310 fewer) 203 fewer per 1000 (from 134 fewer to 253 fewer) 130 fewer per 1000 (from 87 fewer to 164 fewer) 117 fewer per 1000 (from 75 fewer to 146 fewer) ⊕⊕OO LOW CRITICAL

Neonatal deaths — in babies born ≥ 36 weeks 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 8/241 (3.3%) 3/273 (1.1%) RR 2.62 (0.77 to 8.96) ⊕⊕OO LOW CRITICAL

Respiratory distress syndrome — in babies born < 28 weeks 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 none 20/48 (41.7%) 29/54 (53.7%) RR 0.79 (0.53 to 1.18) ⊕⊕OO LOW CRITICAL

Respiratory distress syndrome — in babies born < 30 weeks 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious6 none 46/108 (42.6%) 71/110 (64.5%) RR 0.67 (0.52 to 0.87) ⊕⊕⊕O MODERATE CRITICAL

Respiratory distress syndrome — in babies born < 32 weeks 6 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 75/292 (25.7%) 134/291 (46.0%) RR 0.56 (0.45 to 0.71) ⊕⊕⊕⊕ HIGH CRITICAL

Respiratory distress syndrome — in babies born < 34 weeks 5 randomized trials serious7 no serious inconsistency no serious indirectness no serious imprecision none 109/600 (18.2%) 179/577 (31.0%) RR 0.58 (0.47 to 0.72) ⊕⊕⊕O MODERATE CRITICAL

Respiratory distress syndrome — in babies born < 36 weeks 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 66/525 (12.6%) 121/497 (24.3%) RR 0.52 (0.40 to 0.69) ⊕⊕⊕⊕ HIGH CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (subgroups by gestational age at birth) 1/157 (0.6%) Relative (95% CI)

Effect Absolute Quality Importance

Respiratory distress syndrome — in babies born < 37 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 2/163 (1.2%) RR 1.93 (0.18 to 21.03) 6 more per 1000 (from 5 fewer to 128 more) 16 fewer per 1000 (from 29 fewer to 7 more) 9 fewer per 1000 (from 13 fewer to 23 more) 283 fewer per 1000 (from 60 fewer to 369 fewer) 113 fewer per 1000 (from 182 fewer to 26 more) 83 fewer per 1000 (from 2 fewer to 125 fewer) 52 fewer per 1000 (from 6 fewer to 79 fewer) ⊕⊕OO LOW CRITICAL

Respiratory distress syndrome — in babies born ≥ 34 weeks 5 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 19/618 (3.1%) 30/643 (4.7%) RR 0.66 (0.38 to 1.16) ⊕⊕⊕O MODERATE CRITICAL

Respiratory distress syndrome — in babies born ≥ 36 weeks 5 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 1/261 (0.4%) 4/296 (1.4%) RR 0.30 (0.03 to 2.67) ⊕⊕OO LOW CRITICAL

Cerebroventricular haemorrhage — in babies born < 28 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious8 none 5/34 (14.7%) 12/28 (42.9%) RR 0.34 (0.14 to 0.86) ⊕⊕⊕O MODERATE CRITICAL

Cerebroventricular haemorrhage — in babies born < 30 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 11/76 (14.5%) 19/74 (25.7%) RR 0.56 (0.29 to 1.10) ⊕⊕OO LOW CRITICAL

Cerebroventricular haemorrhage — in babies born < 32 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious6 none 13/144 (9.0%) 23/133 (17.3%) RR 0.52 (0.28 to 0.99) ⊕⊕⊕O MODERATE CRITICAL

Cerebroventricular haemorrhage — in babies born < 34 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 16/273 (5.9%) 27/242 (11.2%) RR 0.53 (0.29 to 0.95) ⊕⊕⊕⊕ HIGH CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (subgroups by gestational age at birth) 27/373 (7.2%) Relative (95% CI)

Effect Absolute Quality Importance

Cerebroventricular haemorrhage — in babies born < 36 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 16/394 (4.1%) RR 0.56 (0.31 to 1.02) 32 fewer per 1000 (from 50 fewer to 1 more) 0 more per 1000 (from 2 fewer to 43 more) not pooled ⊕⊕⊕O MODERATE CRITICAL

Cerebroventricular haemorrhage — in babies born ≥ 34 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 1/351 (0.3%) 1/395 (0.3%) RR 1.13 (0.07 to 17.92) ⊕⊕OO LOW CRITICAL

Cerebroventricular haemorrhage — in babies born ≥ 36 weeks 1 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious9 none 0/209 (0.0%) 49 0/250 (0.0%) 51 not pooled ⊕⊕OO LOW CRITICAL

Mean birth weight (g) — in babies born < 28 weeks (better indicated by higher values) 1 very serious5 none — MD 71.2 higher (42.54 lower to 184.94 higher) MD 0.89 higher (98.17 lower to 99.95 higher) MD 1.15 higher (91.77 lower to 94.07 higher) MD 30.28 lower (115.06 lower to 54.5 higher) ⊕⊕OO LOW CRITICAL

Mean birth weight (g) — in babies born < 30 weeks (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 none 99 102 — ⊕⊕OO LOW CRITICAL

Mean birth weight (g) — in babies born < 32 weeks (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 179 168 — CRITICAL

Mean birth weight (g) — in babies born < 34 weeks (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 312 286 — ⊕⊕⊕O MODERATE CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (subgroups by gestational age at birth) 520 Relative (95% CI)

Effect Absolute Quality Importance

Mean birth weight (g) — in babies born < 36 weeks (better indicated by higher values) 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 524 — MD 8.32 lower (51.31 lower to 34.67 higher) MD 13 higher (93.57 lower to 119.57 higher) MD 12 lower (107.48 lower to 83.48 higher) MD 34.84 lower (117.23 lower to 47.55 higher) ⊕⊕⊕O MODERATE CRITICAL

Mean birth weight (g) — in babies born < 37 weeks (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 none 143 130 — ⊕⊕OO LOW CRITICAL

Mean birth weight (g) — in babies born ≥ 34 weeks (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 361 409 — ⊕⊕⊕O MODERATE CRITICAL

Mean birth weight (g) — in babies born ≥ 36 weeks (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 390 367 — ⊕⊕⊕O MODERATE CRITICAL

1 Wide confidence interval crossing the line of no effect, few events and small sample size. 2 Statistical heterogeneity (I2 > 60%). 3 Wide confidence interval crossing the line of no effect. 4 Wide confidence interval crossing the line of no effect and few events. 5 Wide confidence interval crossing the line of no effect and small sample size. 6 Estimate based on small sample size. 7 Most studies contributing data had design limitations. 8 Estimate based on small sample size and few events. 9 No events.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 1d.

Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women at risk of preterm birth (interval to delivery) Source: Roberts D, Dalziel SR. Antenatal corticosteroids for accelerating fetal lung maturation for women at risk of preterm birth. Cochrane Database Syst Rev. 2006;(3):CD004454. (updated for the guideline) Quality assessment No. of patients Imprecision Other considerations ACS Placebo or no treatment (subgroups by interval to delivery) 10/126 (7.9%) Relative (95% CI) Effect Absolute Quality Importance

No. of studies

Design

Risk of bias

Inconsistency

Indirectness

Chorioamnionitis — in women delivering < 24 hours after 1st dose 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 8/113 (7.1%) RR 0.92 (0.38 to 2.27) 6 fewer per 1000 (from 49 fewer to 101 more) 21 fewer per 1000 (from 58 fewer to 57 more) 37 fewer per 1000 (from 61 fewer to 9 more) 18 more per 1000 (from 11 fewer to 91 more) 117 fewer per 1000 (from 17 fewer to 178 fewer) 130 fewer per 1000 (from 44 fewer to 187 fewer) ⊕⊕OO LOW CRITICAL

Chorioamnionitis — in women delivering < 48 hours after 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 11/150 (7.3%) 18/191 (9.4%) RR 0.78 (0.38 to 1.60) ⊕⊕OO LOW CRITICAL

Chorioamnionitis — in women delivering 1—7 days after 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 11/242 (4.5%) 20/240 (8.3%) RR 0.55 (0.27 to 1.11) ⊕⊕⊕O MODERATE CRITICAL

Chorioamnionitis — in women delivering > 7 days after 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 11/229 (4.8%) 7/232 (3.0%) RR 1.59 (0.63 to 4.03) ⊕⊕OO LOW CRITICAL

Fetal and neonatal deaths — in babies born < 24 hours after 1st dose 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious4 none 23/142 (16.2%) 44/151 (29.1%) RR 0.60 (0.39 to 0.94) ⊕⊕⊕O MODERATE CRITICAL

Fetal and neonatal deaths — in babies born < 48 hours after 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 31/165 (18.8%) 66/208 (31.7%) RR 0.59 (0.41 to 0.86) ⊕⊕⊕⊕ HIGH CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (subgroups by interval to delivery) 74/296 (25.0%) Relative (95% CI)

Effect Absolute Quality Importance

Fetal and neonatal deaths — in babies born 1—7 days after 1st dose 3 randomized trials no serious risk of bias serious5 no serious indirectness serious3 none 62/310 (20.0%) RR 0.81 (0.6 to 1.09) 47 fewer per 1000 (from 100 fewer to 23 more) 41 more per 1000 (from 9 fewer to 119 more) 38 fewer per 1000 (from 79 fewer to 45 more) 22 fewer per 1000 (from 61 fewer to 51 more) 1 more per 1000 (from 30 fewer to 54 more) 19 more per 1000 (from 14 fewer to 79 more) 89 fewer per 1000 (from 8 fewer to 134 fewer) ⊕⊕OO LOW CRITICAL

Fetal and neonatal deaths — in babies born > 7 days after 1st dose 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 42/308 (13.6%) 28/290 (9.7%) RR 1.42 (0.91 to 2.23) ⊕⊕⊕O MODERATE CRITICAL

Fetal deaths — in babies born < 24 hours after 1st dose 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 10/142 (7.0%) 18/151 (11.9%) RR 0.68 (0.34 to 1.38) ⊕⊕OO LOW CRITICAL

Fetal deaths — in babies born < 48 hours after 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 13/165 (7.9%) 21/208 (10.1%) RR 0.78 (0.4 to 1.51) ⊕⊕⊕O MODERATE CRITICAL

Fetal deaths — in babies born 1—7 days after 1st dose 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 22/310 (7.1%) 21/296 (7.1%) RR 1.01 (0.58 to 1.76) ⊕⊕⊕O MODERATE CRITICAL

Fetal deaths — in babies born > 7 days after 1st dose 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 22/308 (7.1%) 15/290 (5.2%) RR 1.36 (0.73 to 2.53) ⊕⊕⊕O MODERATE CRITICAL

Neonatal deaths — in babies born < 24 hours after 1st dose 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 14/152 (9.2%) 27/143 (18.9%) RR 0.53 (0.29 to 0.96) ⊕⊕⊕⊕ HIGH CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (subgroups by interval to delivery) 45/187 (24.1%) Relative (95% CI)

Effect Absolute Quality Importance

Neonatal deaths — in babies born < 48 hours after 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 18/152 (11.8%) RR 0.49 (0.30 to 0.81) 123 fewer per 1000 (from 46 fewer to 168 fewer) 50 fewer per 1000 (from 94 fewer to 13 more) 21 more per 1000 (from 12 fewer to 85 more) 37 fewer per 1000 (from 98 fewer to 43 more) 111 fewer per 1000 (from 23 fewer to 171 fewer) 124 fewer per 1000 (from 92 fewer to 150 fewer) 14 fewer per 1000 (from 35 fewer to 21 more) ⊕⊕⊕⊕ HIGH CRITICAL

Neonatal deaths — in babies born 1—7 days after 1st dose 3 randomized trials no serious risk of bias serious5 no serious indirectness serious3 none 40/288 (13.9%) 53/275 (19.3%) RR 0.74 (0.51 to 1.07) ⊕⊕OO LOW CRITICAL

Neonatal deaths — in babies born > 7 days after 1st dose 3 randomized trials no serious risk of bias serious5 no serious indirectness serious3 none 20/286 (7.0%) 13/275 (4.7%) RR 1.45 (0.75 to 2.80) ⊕⊕OO LOW CRITICAL

Respiratory distress syndrome — in babies born < 24 hours after 1st dose 9 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 68/260 (26.2%) 74/257 (28.8%) RR 0.87 (0.66 to 1.15) ⊕⊕⊕O MODERATE CRITICAL

Respiratory distress syndrome — in babies born < 48 hours after 1st dose 3 randomized trials no serious risk of bias serious5 no serious indirectness no serious imprecision none 38/171 (22.2%) 68/203 (33.5%) RR 0.67 (0.49 to 0.93) ⊕⊕⊕O MODERATE CRITICAL

Respiratory distress syndrome — in babies born 1—7 days after 1st dose 9 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 57/563 (10.1%) 126/547 (23.0%) RR 0.46 (0.35 to 0.60) ⊕⊕⊕⊕ HIGH CRITICAL

Respiratory distress syndrome — in babies born > 7 days after 1st dose 8 randomized trials serious6 no serious inconsistency no serious indirectness serious3 none 32/498 (6.4%) 37/490 (7.6%) RR 0.82 (0.53 to 1.28) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (subgroups by interval to delivery) 23/100 (23.0%) Relative (95% CI)

Effect Absolute Quality Importance

Moderate/severe respiratory distress syndrome — in babies born < 24 hours after 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious7 none 13/82 (15.9%) RR 0.69 (0.37 to 1.27) 71 fewer per 1000 (from 145 fewer to 62 more) 151 fewer per 1000 (from 74 fewer to 200 fewer) 123 fewer per 1000 (from 74 fewer to 153 fewer) 22 more per 1000 (from 8 fewer to 104 more) 39 fewer per 1000 (from 66 fewer to 30 more) 75 fewer per 1000 (from 25 fewer to 92 fewer) 35 fewer per 1000 (from 55 fewer to 9 more) ⊕⊕OO LOW CRITICAL

Moderate/severe respiratory distress syndrome — in babies born < 48 hours after 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 18/147 (12.2%) 49/179 (27.4%) RR 0.45 (0.27 to 0.73) ⊕⊕⊕⊕ HIGH CRITICAL

Moderate/severe respiratory distress syndrome — in babies born 1—7 days after 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 17/237 (7.2%) 44/225 (19.6%) RR 0.37 (0.22 to 0.62) ⊕⊕⊕⊕ HIGH CRITICAL

Moderate/severe respiratory distress syndrome — in babies born > 7 days after 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 11/223 (4.9%) 6/223 (2.7%) RR 1.83 (0.69 to 4.87) ⊕⊕OO LOW CRITICAL

Cerebroventricular haemorrhage — in babies born < 24 hours after 1st dose 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 7/133 (5.3%) 11/131 (8.4%) RR 0.54 (0.21 to 1.36) ⊕⊕OO LOW CRITICAL

Cerebroventricular haemorrhage — in babies born < 48 hours after 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious8 none 4/152 (2.6%) 19/187 (10.2%) RR 0.26 (0.09 to 0.75) ⊕⊕⊕O MODERATE CRITICAL

Cerebroventricular haemorrhage — in babies born 1—7 days after 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 9/245 (3.7%) 17/237 (7.2%) RR 0.51 (0.23 to 1.13) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (subgroups by interval to delivery) 2/227 (0.9%) Relative (95% CI)

Effect Absolute Quality Importance

Cerebroventricular haemorrhage — in babies born > 7 days after 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 4/226 (1.8%) RR 2.01 (0.37 to 10.86) 9 more per 1000 (from 6 fewer to 87 more) MD 46.52 higher (94.26 lower to 187.29 higher) MD 5.9 lower (131.95 lower to 120.15 higher) MD 105.92 lower (212.52 lower to 0.68 higher) MD 147.01 lower (291.97 to 2.05 lower) ⊕⊕OO LOW CRITICAL

Mean birth weight (g) — in babies born < 24 hours after 1st dose (better indicated by higher values) 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious7 none 112 130 — ⊕⊕OO LOW CRITICAL

Mean birth weight (g) — in babies born < 48 hours after 1st dose (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 165 208 — ⊕⊕⊕O MODERATE CRITICAL

Mean birth weight (g) — in babies born 1—7 days after 1st dose (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 264 256 — ⊕⊕⊕⊕ HIGH CRITICAL

Mean birth weight (g) — in babies born > 7 days after 1st dose (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 245 241 — ⊕⊕⊕⊕ HIGH CRITICAL

1 Wide confidence interval crossing the line of no effect, few events and small sample size. 2 Wide confidence interval crossing the line of no effect and few events. 3 Wide confidence interval crossing the line of no effect. 4 Estimate based on small sample size. 5 Statistical Heterogeneity (I2 > 60%). 6 Most studies contributing data had design limitations. 7 Wide confidence interval crossing the line of no effect and small sample size. 8 Few events.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 1e.

Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women at risk of preterm birth (singleton and multiple pregnancy subgroups) Source: Roberts D, Dalziel SR. Antenatal corticosteroids for accelerating fetal lung maturation for women at risk of preterm birth. Cochrane Database Syst Rev. 2006;(3):CD004454. (updated for this guideline) Quality assessment No. of patients Imprecision Other considerations ACS Placebo or no treatment (subgroups by singleton/ multiple pregnancy) 61/838 (7.3%) Relative (95% CI) Effect Absolute Quality Importance

No. of studies

Design

Risk of bias

Inconsistency

Indirectness

Chorioamnionitis — in women delivering singleton babies 5 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 50/823 (6.1%) RR 0.82 (0.58 to 1.18) 13 fewer per 1000 (from 31 fewer to 13 more) 34 fewer per 1000 (from 56 fewer to 205 more) 52 fewer per 1000 (from 10 fewer to 87 fewer) 58 fewer per 1000 (from 117 fewer to 44 more) 8 more per 1000 (from 16 fewer to 43 more) 39 fewer per 1000 (from 66 fewer to 33 more) ⊕⊕⊕O MODERATE CRITICAL

Chorioamnionitis — in women delivering multiple babies 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 1/40 (2.5%) 2/34 (5.9%) RR 0.43 (0.04 to 4.49) ⊕⊕OO LOW CRITICAL

Fetal and neonatal deaths — in babies born from singleton pregnancies 3 randomized trials no serious risk of bias serious3 no serious indirectness no serious imprecision none 140/702 (19.9%) 180/723 (24.9%) RR 0.79 (0.65 to 0.96) ⊕⊕⊕O MODERATE CRITICAL

Fetal and neonatal deaths — in babies born from multiple pregnancies 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 19/131 (14.5%) 24/121 (19.8%) RR 0.71 (0.41 to 1.22) ⊕⊕OO LOW CRITICAL

Fetal deaths — in babies born from singleton pregnancies 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 55/702 (7.8%) 51/723 (7.1%) RR 1.12 (0.78 to 1.61) ⊕⊕⊕O MODERATE CRITICAL

Fetal deaths — in babies born from multiple pregnancies 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 6/131 (4.6%) 10/121 (8.3%) RR 0.53 (0.20 to 1.40) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (subgroups by singleton/ multiple pregnancy) 141/968 (14.6%) Relative (95% CI)

Effect Absolute Quality Importance

Neonatal deaths — in babies born from singleton pregnancies 7 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 94/957 (9.8%) RR 0.67 (0.53 to 0.85) 48 fewer per 1000 (from 22 fewer to 68 fewer) 26 fewer per 1000 (from 77 fewer to 77 more) 86 fewer per 1000 (from 64 fewer to 105 fewer) 39 fewer per 1000 (from 105 fewer to 52 more) 46 fewer per 1000 (from 26 fewer to 60 fewer) 41 fewer per 1000 (from 62 fewer to 71 more) MD 16.61 lower (55.45 lower to 22.23 higher) ⊕⊕⊕⊕ HIGH CRITICAL

Neonatal deaths — in babies born from multiple pregnancies 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 13/125 (10.4%) 14/111 (12.6%) RR 0.79 (0.39 to 1.61) ⊕⊕OO LOW CRITICAL

Respiratory distress syndrome — in babies born from singleton pregnancies 12 randomized trials serious5 serious3 no serious indirectness no serious imprecision none 187/1462 (12.8%) 309/1445 (21.4%) RR 0.60 (0.51 to 0.70) ⊕⊕OO LOW CRITICAL

Respiratory distress syndrome — in babies born from multiple pregnancies 4 randomized trials serious5 no serious inconsistency no serious indirectness serious1 none 44/167 (26.3%) 40/153 (26.1%) RR 0.85 (0.6 to 1.2) ⊕⊕OO LOW CRITICAL

Cerebroventricular haemorrhage — in babies born from singleton pregnancies 5 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 35/772 (4.5%) 71/789 (9.0%) RR 0.49 (0.33 to 0.71) ⊕⊕⊕⊕ HIGH CRITICAL

Cerebroventricular haemorrhage — in babies born from multiple pregnancies 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 2/77 (2.6%) 4/60 (6.7%) RR 0.39 (0.07 to 2.06) ⊕⊕OO LOW CRITICAL

Mean birth weight (g) — in babies born from singleton pregnancies (better indicated by higher values) 6 randomized trials serious5 no serious inconsistency no serious indirectness serious1 none 860 867 — ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (subgroups by singleton/ multiple pregnancy) 69 Relative (95% CI)

Effect Absolute Quality Importance

Mean birth weight (g) — in babies born from multiple pregnancies (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 81 — MD 82.36 higher (146.23 lower to 310.95 higher) ⊕⊕OO LOW CRITICAL

1 Wide confidence interval crossing the line of no effect. 2 Wide confidence interval crossing the line of no effect, few events and small sample size. 3 Statistical heterogeneity (I2 > 60%). 4 Wide confidence interval crossing the line of no effect and small sample size. 5 Most studies contributing data had design limitations.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 1f.

Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women at risk of preterm birth (preterm prelabour rupture of membranes) Source: Roberts D, Dalziel SR. Antenatal corticosteroids for accelerating fetal lung maturation for women at risk of preterm birth. Cochrane Database Syst Rev. 2006;(3):CD004454. (updated for the guideline) Quality assessment No. of patients Imprecision Other considerations ACS Placebo or no treatment (subgroups by intact/ruptured membranes) 1/108 (0.9%) Relative (95% CI) Effect Absolute Quality Importance

No. of studies

Design

Risk of bias

Inconsistency

Indirectness

Maternal death — in women with pregnancies not complicated by preterm prelabour rupture of membranes (PPROM) at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 1/110 (0.9%) RR 0.98 (0.06 to 15.50) 0 fewer per 1000 (from 9 fewer to 134 more) not pooled ⊕⊕OO LOW CRITICAL

Maternal death — in women with pregnancies complicated by PPROM at 1st dose 2 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious2 none 0/58 (0.0%) 24/611 (3.9%) 0/45 (0.0%) 30/632 (4.7%) not pooled ⊕⊕OO LOW CRITICAL

Chorioamnionitis — in women with pregnancies not complicated by PPROM at 1st dose 4 serious3 none RR 0.83 (0.50 to 1.40) 8 fewer per 1000 (from 24 fewer to 19 more) 2 fewer per 1000 (from 34 fewer to 44 more) 17 more per 1000 (from 32 fewer to 97 more) 25 fewer per 1000 (from 81 fewer to 84 more) ⊕⊕⊕O MODERATE CRITICAL

Chorioamnionitis — in women with pregnancies complicated by PPROM at 1st dose 7 randomized trials serious4 no serious inconsistency no serious indirectness serious3 none 52/480 (10.8%) 53/479 (11.1%) RR 0.98 (0.69 to 1.40) ⊕⊕OO LOW CRITICAL

Chorioamnionitis — in women with prolonged rupture of membranes > 24 hours 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 30/236 (12.7%) 27/247 (10.9%) RR 1.16 (0.71 to 1.89) ⊕⊕⊕O MODERATE CRITICAL

Chorioamnionitis — in women with prolonged rupture of membranes > 48 hours 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 none 14/122 (11.5%) 16/114 (14.0%) RR 0.82 (0.42 to 1.60) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (subgroups by intact/ruptured membranes) 18/146 (12.3%) Relative (95% CI)

Effect Absolute Quality Importance

Puerperal sepsis — in women with pregnancies not complicated by PPROM at 1st dose 2 randomized trials no serious risk of bias serious6 no serious indirectness very serious5 none 19/143 (13.3%) RR 1.10 (0.61 to 2.00) 12 more per 1000 (from 48 fewer to 123 more) 7 more per 1000 (from 27 fewer to 74 more) 17 fewer per 1000 (from 56 fewer to 113 more) 26 fewer per 1000 (from 61 fewer to 16 more) 92 fewer per 1000 (from 43 fewer to 130 fewer) 37 fewer per 1000 (from 79 fewer to 28 more) 15 fewer per 1000 (from 91 fewer to 108 more) ⊕OOO VERY LOW CRITICAL

Puerperal sepsis — in women with pregnancies complicated by PPROM at 1st dose 4 randomized trials serious4 no serious inconsistency no serious indirectness serious3 none 16/242 (6.6%) 14/235 (6.0%) RR 1.11 (0.55 to 2.25) ⊕⊕OO LOW CRITICAL

Puerperal sepsis — in women with prolonged rupture of membranes > 24 hours 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 4/74 (5.4%) 6/84 (7.1%) RR 0.76 (0.22 to 2.58) ⊕⊕OO LOW CRITICAL

Fetal and neonatal deaths — in babies born from pregnancies not complicated by PPROM at 1st dose 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 116/659 (17.6%) 137/673 (20.4%) RR 0.87 (0.70 to 1.08) ⊕⊕⊕O MODERATE CRITICAL

Fetal and neonatal deaths — in babies born from pregnancies complicated by PPROM at 1st dose 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 55/368 (14.9%) 88/365 (24.1%) RR 0.62 (0.46 to 0.82) ⊕⊕⊕⊕ HIGH CRITICAL

Fetal and neonatal deaths — in babies born following prolonged rupture of membranes > 24 hours 2 randomized trials no serious risk of bias serious6 no serious indirectness serious3 none 33/255 (12.9%) 41/253 (16.2%) RR 0.77 (0.51 to 1.17) ⊕⊕OO LOW CRITICAL

Fetal and neonatal deaths — in babies born following prolonged rupture of membranes > 48 hours 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 none 27/137 (19.7%) 25/118 (21.2%) RR 0.93 (0.57 to 1.51) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (subgroups by intact/ruptured membranes) 42/673 (6.2%) Relative (95% CI)

Effect Absolute Quality Importance

Fetal deaths — in babies born from pregnancies not complicated by PPROM at 1st dose 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 45/659 (6.8%) RR 1.09 (0.73 to 1.64) 6 more per 1000 (from 17 fewer to 40 more) 7 fewer per 1000 (from 26 fewer to 30 more) 12 more per 1000 (from 20 fewer to 74 more) 9 more per 1000 (from 45 fewer to 123 more) 35 fewer per 1000 (from 64 fewer to 5 more) 66 fewer per 1000 (from 29 fewer to 91 fewer) 52 fewer per 1000 (from 81 fewer to 1 more) ⊕⊕⊕O MODERATE CRITICAL

Fetal deaths — in babies born from pregnancies complicated by PPROM at 1st dose 5 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 16/398 (4.0%) 19/392 (4.8%) RR 0.86 (0.46 to 1.61) ⊕⊕⊕O MODERATE CRITICAL

Fetal deaths — in babies born following prolonged rupture of membranes > 24 hours 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 17/255 (6.7%) 13/253 (5.1%) RR 1.23 (0.62 to 2.44) ⊕⊕⊕O MODERATE CRITICAL

Fetal deaths — in babies born following prolonged rupture of membranes > 48 hours 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 14/137 (10.2%) 11/118 (9.3%) RR 1.10 (0.52 to 2.32) ⊕⊕OO LOW CRITICAL

Neonatal deaths — in babies born from pregnancies not complicated by PPROM at 1st dose 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 71/611 (11.6%) 95/625 (15.2%) RR 0.77 (0.58 to 1.03) ⊕⊕⊕O MODERATE CRITICAL

Neonatal deaths — in babies born from pregnancies complicated by PPROM at 1st dose 8 randomized trials serious4 no serious inconsistency no serious indirectness no serious imprecision none 54/519 (10.4%) 85/505 (16.8%) RR 0.61 (0.46 to 0.83) ⊕⊕⊕O MODERATE CRITICAL

Neonatal deaths — in babies born following prolonged rupture of membranes > 24 hours 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 16/238 (6.7%) 28/239 (11.7%) RR 0.56 (0.31 to 1.01) ⊕⊕⊕O MODERATE CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (subgroups by intact/ruptured membranes) 14/107 (13.1%) Relative (95% CI)

Effect Absolute Quality Importance

Neonatal deaths — in babies born following prolonged rupture of membranes > 48 hours 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 13/123 (10.6%) RR 0.81 (0.40 to 1.64) 25 fewer per 1000 (from 79 fewer to 84 more) 103 fewer per 1000 (from 71 fewer to 133 fewer) 102 fewer per 1000 (from 54 fewer to 137 fewer) 83 fewer per 1000 (from 26 fewer to 128 fewer) 44 fewer per 1000 (from 97 fewer to 62 more) 61 fewer per 1000 (from 34 fewer to 80 fewer) 46 fewer per 1000 (from 18 fewer to 62 fewer) ⊕⊕OO LOW CRITICAL

Respiratory distress syndrome — in babies born from pregnancies not complicated by PPROM at 1st dose 5 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 125/752 (16.6%) 211/775 (27.2%) RR 0.62 (0.51 to 0.74) ⊕⊕⊕⊕ HIGH CRITICAL

Respiratory distress syndrome — in babies born from pregnancies complicated by PPROM at 1st dose 12 randomized trials serious4 no serious inconsistency no serious indirectness no serious imprecision none 126/577 (21.8%) 176/552 (31.9%) RR 0.68 (0.57 to 0.83) ⊕⊕⊕O MODERATE CRITICAL

Respiratory distress syndrome — in babies born following prolonged rupture of membranes > 24 hours 6 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 55/311 (17.7%) 82/315 (26.0%) RR 0.68 (0.51 to 0.90) ⊕⊕⊕⊕ HIGH CRITICAL

Respiratory distress syndrome — in babies born following prolonged rupture of membranes > 48 hours 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 none 13/128 (10.2%) 18/119 (15.1%) RR 0.71 (0.36 to 1.41) ⊕⊕OO LOW CRITICAL

Cerebroventricular haemorrhage — in babies born from pregnancies not complicated by PPROM at 1st dose 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 36/597 (6.0%) 74/603 (12.3%) RR 0.50 (0.35 to 0.72) ⊕⊕⊕⊕ HIGH CRITICAL

Cerebroventricular haemorrhage — in babies born from pregnancies complicated by PPROM at 1st dose 5 randomized trials serious4 no serious inconsistency no serious indirectness no serious imprecision none 19/454 (4.2%) 38/441 (8.6%) RR 0.47 (0.28 to 0.79) ⊕⊕⊕O MODERATE CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (subgroups by intact/ruptured membranes) 3/107 (2.8%) Relative (95% CI)

Effect Absolute Quality Importance

Cerebroventricular haemorrhage — in babies born following prolonged rupture of membranes > 48 hours 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 3/123 (2.4%) RR 0.87 (0.18 to 4.22) 4 fewer per 1000 (from 23 fewer to 90 more) 13 fewer per 1000 (from 25 fewer to 25 more) 151 fewer per 1000 (from 45 fewer to 207 fewer) 3 fewer per 1000 (from 46 fewer to 89 more) 3 fewer per 1000 (from 66 fewer to 153 more) 47 fewer per 1000 (from 8 fewer to 74 fewer) 26 more per 1000 (from 14 fewer to 85 more) ⊕⊕OO LOW CRITICAL

Cerebroventricular haemorrhage — in babies born following prolonged rupture of membranes > 24 hours 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious7 none 4/238 (1.7%) 7/239 (2.9%) RR 0.55 (0.16 to 1.84) ⊕⊕OO LOW CRITICAL

Systemic infection in the first 48 hours of life — in babies born from pregnancies not complicated by PPROM at 1st dose 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious8 none 13/100 (13.0%) 28/100 (28.0%) RR 0.46 (0.26 to 0.84) ⊕⊕⊕O MODERATE CRITICAL

Systemic infection in the first 48 hours of life — in babies born from pregnancies complicated by PPROM at 1st dose 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 12/148 (8.1%) 12/143 (8.4%) RR 0.97 (0.45 to 2.06) ⊕⊕OO LOW CRITICAL

Systemic infection in the first 48 hours of life — in babies born following prolonged rupture of membranes > 24 hours 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 8/75 (10.7%) 9/82 (11.0%) RR 0.97 (0.40 to 2.39) ⊕⊕OO LOW CRITICAL

Proven infection while in the neonatal intensive care unit (NICU) — in babies born from pregnancies not complicated by PPROM at 1st dose 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 55/520 (10.6%) 81/537 (15.1%) RR 0.69 (0.51 to 0.95) ⊕⊕⊕⊕ HIGH CRITICAL

Proven infection while in the NICU — in babies born from pregnancies complicated by PPROM at 1st dose 7 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 51/406 (12.6%) 39/390 (10.0%) RR 1.26 (0.86 to 1.85) ⊕⊕⊕O MODERATE CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (subgroups by intact/ruptured membranes) 23/181 (12.7%) Relative (95% CI)

Effect Absolute Quality Importance

Proven infection while in the NICU — in babies born following prolonged rupture of membranes > 24 hours 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 31/182 (17.0%) RR 1.34 (0.82 to 2.21) 43 more per 1000 (from 23 fewer to 154 more) 24 more per 1000 (from 51 fewer to 152 more) 15 fewer per 1000 (from 33 fewer to 57 more) 43 fewer per 1000 (from 10 fewer to 58 fewer) 44 fewer per 1000 (from 81 fewer to 72 more) MD 59.09 lower (157.84 lower to 39.67 higher) MD 42.68 lower (108.91 lower to 23.55 higher) ⊕⊕⊕O MODERATE CRITICAL

Proven infection while in the NICU — in babies born following prolonged rupture of membranes > 48 hours 2 randomized trials no serious risk of bias serious6 no serious indirectness very serious5 none 24/133 (18.0%) 20/125 (16.0%) RR 1.15 (0.68 to 1.95) ⊕OOO VERY LOW CRITICAL

Necrotizing enterocolitis — in babies born from pregnancies not complicated by PPROM at 1st dose 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 3/128 (2.3%) 5/129 (3.9%) RR 0.61 (0.15 to 2.48) ⊕⊕OO LOW CRITICAL

Necrotizing enterocolitis — in babies born from pregnancies complicated by PPROM at 1st dose 4 randomized trials serious4 no serious inconsistency no serious indirectness serious9 none 8/300 (2.7%) 20/283 (7.1%) RR 0.39 (0.18 to 0.86) ⊕⊕OO LOW CRITICAL

Necrotizing enterocolitis — in babies born following prolonged rupture of membranes > 24 hours 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 4/75 (5.3%) 8/82 (9.8%) RR 0.55 (0.17 to 1.74) ⊕⊕OO LOW CRITICAL

Mean birth weight (g) — in babies born from pregnancies not complicated by PPROM at 1st dose (better indicated by higher values) 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 545 562 — ⊕⊕⊕O MODERATE CRITICAL

Mean birth weight (g) — in babies born from pregnancies complicated by PPROM at 1st dose (better indicated by higher values) 5 randomized trials serious4 no serious inconsistency no serious indirectness serious3 none 420 415 — ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (subgroups by intact/ruptured membranes) 169 Relative (95% CI)

Effect Absolute Quality Importance

Mean birth weight (g) — in babies born following prolonged rupture of membranes > 24 hours (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 180 — MD 196.46 lower (335.19 to 57.73 lower) MD 201.79 lower (363.3 to 40.28 lower) ⊕⊕⊕⊕ HIGH CRITICAL

Mean birth weight (g) — in babies born following prolonged rupture of membranes > 48 hours (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious8 none 137 118 — ⊕⊕⊕O MODERATE CRITICAL

Mean duration of mechanical ventilation/continuous positive airway pressure (days) — in babies born from pregnancies not complicated by PPROM at 1st dose (better indicated by lower values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 none 14 19 — MD 3.8 higher (20.79 lower to 28.39 higher) ⊕⊕OO LOW CRITICAL

Mean duration of mechanical ventilation/continuous positive airway pressure (days) — in babies born from pregnancies complicated by PPROM at 1st dose (better indicated by lower values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious8 none 87 78 — MD 3.5 lower (5.12 to 1.88 lower) 11 more per 1000 (from 27 fewer to 86 more) 263 fewer per 1000 (from 126 fewer to 352 fewer) ⊕⊕⊕O MODERATE CRITICAL

Chronic lung disease — in babies born from pregnancies not complicated by PPROM at 1st dose 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 16/218 (7.3%) 15/216 (6.9%) RR 1.16 (0.61 to 2.24) ⊕⊕⊕O MODERATE CRITICAL

Chronic lung disease — in babies born from pregnancies complicated by PPROM at 1st dose 1 randomized trials serious4 no serious inconsistency no serious indirectness serious8 none 23/87 (26.4%) 41/78 (52.6%) RR 0.50 (0.33 to 0.76) ⊕⊕OO LOW CRITICAL

1 Wide confidence interval crossing the line of no effect, few events and small sample size. 2 No events. 3 Wide confidence interval crossing the line of no effect. 4 Most studies contributing data had design limitations. 5 Wide confidence interval crossing the line of no effect and small sample size. 6 Statistical heterogeneity (I2 > 60%). 7 Wide confidence interval crossing the line of no effect and few events. 8 Estimate based on small sample size. 9 Few events.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 1g:

Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women at risk of preterm birth (women with chorioamnionitis) Source: Amiya RM, Mlunde LB, Ota E, Mori R, Oladapo OT. Antenatal corticosteroid therapy for reducing adverse maternal and child outcomes in special populations of women at risk of imminent preterm birth: a systematic review. Plos One. 2015 (review in progress). Quality assessment No. of patients Imprecision Other considerations not serious ACS No ACS Relative (95% CI) OR 0.54 (0.38 to 0.76) Effect Absolute (95% CI) 70 fewer per 1000 (from 35 fewer to 97 fewer) 80 fewer per 1000 (from 40 fewer to 106 fewer) 31 fewer per 1000 (from 77 more to 92 fewer) 119 fewer per 1000 (from 62 fewer to 175 fewer) 135 fewer per 1000 (from 68 fewer to 201 fewer) 77 fewer per 1000 (from 28 more to 172 fewer) Quality Importance

No. of studies 7

Study design

Risk of bias

Inconsistency

Indirectness

Neonatal death — histological chorioamnionitis (HC) and/or clinical chorioamnionitis (CC) observational studies serious1 not serious not serious serious2 81/787 (10.3%) 104/616 (16.9%) ⊕OOO VERY LOW CRITICAL

Neonatal death — HC only 6 observational studies serious1 not serious not serious serious2 not serious 64/638 (10.0%) 89/518 (17.2%) OR 0.49 (0.34 to 0.73) ⊕OOO VERY LOW CRITICAL

Neonatal death — CC only 3 observational studies serious1 not serious not serious very serious3 not serious 17/149 (11.4%) 15/98 (15.3%) OR 0.77 (0.36 to 1.65) ⊕OOO VERY LOW CRITICAL

Respiratory distress syndrome — HC and/or CC 7 observational studies serious1 not serious not serious serious2 not serious 378/789 (47.9%) 384/712 (53.9%) OR 0.62 (0.49 to 0.78) ⊕OOO VERY LOW CRITICAL

Respiratory distress syndrome — HC only 5 observational studies serious1 not serious not serious serious2 not serious 279/580 (48.1%) 285/504 (56.5%) OR 0.58 (0.44 to 0.76) ⊕OOO VERY LOW CRITICAL

Respiratory distress syndrome — CC only 4 observational studies not serious not serious not serious serious2 not serious 99/209 (47.4%) 99/208 (47.6%) OR 0.73 (0.48 to 1.12) ⊕OOO VERY LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 3 Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations not serious

No. of patients ACS No ACS Relative (95% CI)

Effect Absolute (95% CI) 17 fewer per 1000 (from 61 more to 99 fewer) 94 fewer per 1000 (from 58 fewer to 119 fewer) 108 fewer per 1000 (from 53 fewer to 145 fewer) 78 fewer per 1000 (from 21 fewer to 106 fewer) 68 fewer per 1000 (from 36 fewer to 86 fewer) 78 fewer per 1000 (from 26 fewer to 107 fewer) Quality Importance

Surfactant use (HC only) observational studies serious1 not serious not serious serious2 187/316 (59.2%) 244/404 (60.4%) OR 0.93 (0.67 to 1.30) ⊕OOO VERY LOW CRITICAL

Intraventricular haemorrhage — HC and/or CC 6 observational studies not serious not serious not serious serious2 strong association 66/626 (10.5%) 52/313 (16.6%) OR 0.39 (0.25 to 0.61) ⊕⊕OO LOW CRITICAL

Intraventricular haemorrhage — HC only 5 observational studies not serious not serious not serious serious2 strong association 53/463 (11.4%) 32/158 (20.3%) OR 0.41 (0.24 to 0.69) ⊕⊕OO LOW CRITICAL

Intraventricular haemorrhage — CC only 3 observational studies not serious not serious not serious serious2 strong association 13/163 (8.0%) 20/155 (12.9%) OR 0.36 (0.16 to 0.82) ⊕⊕OO LOW CRITICAL

Severe intraventricular haemorrhage — HC and/or CC 5 observational studies not serious not serious not serious serious2 strong association 33/538 (6.1%) 30/271 (11.1%) OR 0.36 (0.20 to 0.65) ⊕⊕OO LOW CRITICAL

Severe intraventricular haemorrhage — HC only 4 observational studies not serious not serious not serious serious2 strong association 28/375 (7.5%) 16/116 (13.8%) OR 0.40 (0.20 to 0.79) ⊕⊕OO LOW CRITICAL

Severe intraventricular haemorrhage — CC only 3 observational studies not serious not serious not serious very serious3 strong association 5/163 (3.1%) 14/155 (9.0%) OR 0.29 (0.10 to 0.89) 62 fewer per 1000 (from 9 fewer to 80 fewer) ⊕OOO VERY LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 4 Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations strong association

No. of patients ACS No ACS Relative (95% CI)

Effect Absolute (95% CI) 58 fewer per 1000 (from 10 fewer to 86 fewer) 15 fewer per 1000 (from 43 fewer to 57 more) 95 fewer per 1000 (from 20 fewer to 130 fewer) 3 more per 1000 (from 39 fewer to 55 more) 4 more per 1000 (from 39 fewer to 60 more) 12 fewer per 1000 (from 137 fewer to 174 more) 27 more per 1000 (from 5 fewer to 79 more) Quality Importance

Periventricular leukomalacia — HC and/or CC observational studies not serious not serious not serious serious2 21/480 (4.4%) 30/257 (11.7%) OR 0.47 (0.24 to 0.90) ⊕OOO VERY LOW CRITICAL

Periventricular leukomalacia — HC only 3 observational studies not serious not serious not serious very serious3 not serious 13/317 (4.1%) 6/102 (5.9%) OR 0.74 (0.26 to 2.09) ⊕OOO VERY LOW CRITICAL

Periventricular leukomalacia — CC only 3 observational studies not serious not serious not serious serious2 strong association 8/163 (4.9%) 24/155 (15.5%) OR 0.35 (0.14 to 0.85) ⊕OOO VERY LOW CRITICAL

Neonatal sepsis — HC and/or CC 5 observational studies not serious not serious not serious serious2 not serious 113/684 (16.5%) 92/550 (16.7%) OR 1.02 (0.73 to 1.42) ⊕OOO VERY LOW CRITICAL

Neonatal sepsis — HC only 5 observational studies not serious not serious not serious serious2 not serious 87/580 (15.0%) 80/504 (15.9%) OR 1.03 (0.72 to 1.48) ⊕OOO VERY LOW CRITICAL

Neonatal sepsis — CC only 2 observational studies not serious not serious not serious very serious3 not serious 26/104 (25.0%) 12/46 (26.1%) OR 0.94 (0.40 to 2.18) ⊕OOO VERY LOW CRITICAL

Necrotizing enterocolitis — HC and/or CC 5 observational studies serious1 not serious not serious serious2 not serious 76/684 (11.1%) 33/550 (6.0%) OR 1.49 (0.91 to 2.53) ⊕OOO VERY LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 5 Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations not serious

No. of patients ACS No ACS Relative (95% CI)

Effect Absolute (95% CI) 18 more per 1000 (from 12 fewer to 66 more) 90 more per 1000 (from 17 fewer to 338 more) MD 2 lower (4.23 lower to 0.23 higher) 174 fewer per 1000 (from 34 fewer to 414 fewer) 163 fewer per 1000 (from 6 more to 470 fewer) 0 fewer per 1000 (from 0 fewer to 0 fewer) 47 fewer per 1000 (from 25 more to 99 fewer) Quality Importance

Necrotizing enterocolitis — HC only observational studies serious1 not serious not serious serious2 60/580 (10.3%) 30/504 (5.9%) OR 1.33 (0.78 to 2.26) ⊕OOO VERY LOW CRITICAL

Necrotizing enterocolitis — CC only 2 observational studies serious1 not serious not serious very serious3 not serious 16/104 (15.4%) 3/46 (6.5%) OR 2.63 (0.72 to 9.68) ⊕OOO VERY LOW CRITICAL

Duration of mechanical ventilation, days — HC only 1 observational studies not serious not serious not serious very serious3 not serious 52 36 — ⊕OOO VERY LOW CRITICAL

Use of mechanical ventilation — HC and/or CC 1 observational studies not serious not serious not serious very serious3 strong association 115/153 (75.2%) 58/61 (95.1%) OR 0.18 (0.06 to 0.57) ⊕OOO VERY LOW CRITICAL

Use of mechanical ventilation — HC only 1 observational studies not serious not serious not serious very serious3 not serious 66/89 (74.2%) 29/32 (90.6%) OR 0.30 (0.08 to 1.07) ⊕OOO VERY LOW CRITICAL

Use of mechanical ventilation — CC only 1 observational studies not serious not serious not serious serious4 not serious 49/64 (76.6%) 29/29 (100%) OR 0.05 (0.00 to 0.94) ⊕OOO VERY LOW CRITICAL

Chronic lung disease/bronchopulmonary dysplasia — HC and/or CC 4 observational studies not serious not serious not serious serious2 not serious 80/465 (17.2%) 42/194 (21.6%) OR 0.74 (0.48 to 1.15) ⊕OOO VERY LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 3 Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations not serious

No. of patients ACS No ACS Relative (95% CI)

Effect Absolute (95% CI) 83 fewer per 1000 (from 25 more to 138 fewer) 13 fewer per 1000 (from 91 fewer to 109 more) 127 fewer per 1000 (from 99 more to 196 fewer) MD 6 higher (9.94 lower to 20.94 higher) MD 13 higher (3.75 lower to 29.75 higher) Quality Importance

Chronic lung disease/bronchopulmonary dysplasia — HC only observational studies not serious not serious not serious serious2 55/323 (17.0%) 26/104 (25.0%) OR 0.66 (0.38 to 1.14) ⊕OOO VERY LOW CRITICAL

Chronic lung disease/bronchopulmonary dysplasia — CC only 3 observational studies serious1 not serious not serious very serious3 not serious 25/142 (17.6%) 16/90 (17.8%) OR 0.91 (0.44 to 1.86) ⊕OOO VERY LOW CRITICAL

Cerebral palsy (at 1 and 3 years follow-up) — HC only 1 observational studies serious1 not serious not serious very serious3 not serious 5/58 (8.6%) 3/14 (21.4%) OR 0.35 (0.07 to 1.67) ⊕OOO VERY LOW CRITICAL

General development quotient at 1 years follow-up — HC only 1 observational studies serious1 not serious not serious very serious3 not serious 58 14 — ⊕OOO VERY LOW CRITICAL

General development quotient at 3 years follow-up — HC only 1 observational studies serious1 not serious not serious very serious3 not serious 58 14 — ⊕OOO VERY LOW CRITICAL

HC: histological chorioamnionitis; CC: clinical chorioamnionitis. 1 Evidence heavily based on studies with design limitations including lack of adjustment for potential confounding factors. 2 Estimate based on wide confidence interval crossing the line of no effect. 3 Estimate based on small sample size; wide confidence interval crossing the line of no effect. 4 Estimate based on small sample size.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 1h. Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women undergoing elective caesarean section at late preterm (34—36+6 weeks) Source: Sotiriadis A, Makrydimas G, Papatheodorou S, Ioannidis JP. Corticosteroids for preventing neonatal respiratory morbidity after elective caesarean section at term. Cochrane Database Syst Rev. 2009;(4):CD006614. Quality assessment No. of studies 1 Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations not serious No. of patients ACS No ACS Relative (95% CI) not estimable Effect Absolute (95% CI) 0 fewer per 1000 (from 0 fewer to 0 fewer) 0 fewer per 1000 (from 0 fewer to 0 fewer) 7 fewer per 1000 (from 6 more to 10 fewer) 19 fewer per 1000 (from 4 more to 30 fewer) MD 2.14 lower (5.58 lower to 1.3 higher) 25 fewer per 1000 (from 11 fewer to 29 fewer) Quality Importance

Perinatal death randomized trials not serious not serious very serious1 very serious2 0/467 (0.0%) 0/475 (0.0%) ⊕OOO VERY LOW CRITICAL

Neonatal sepsis 1 randomized trials not serious not serious very serious1 very serious2 not serious 0/467 (0.0%) 0/475 (0.0%) not estimable ⊕OOO VERY LOW CRITICAL

Respiratory distress syndrome 1 randomized trials not serious not serious very serious1 very serious3 not serious 1/467 (0.2%) 5/471 (1.1%) RR 0.32 (0.07 to 1.58) ⊕OOO VERY LOW CRITICAL

Tachypnoea of the neonate 1 randomized trials not serious not serious very serious1 very serious3 not serious 10/467 (2.1%) 19/475 (4.0%) RR 0.52 (0.25 to 1.11) ⊕OOO VERY LOW CRITICAL

Length of stay in neonatal intensive care unit (NICU) 1 randomized trials serious4 not serious very serious1 very serious5 not serious 2 14 — ⊕OOO VERY LOW CRITICAL

Admission to NICU for respiratory complications 1 randomized trials serious4 not serious very serious1 serious6 strong association 2/467 (0.4%) 14/475 (2.9%) RR 0.15 (0.03 to 0.64) ⊕OOO VERY LOW IMPORTANT

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations strong association

No. of patients ACS No ACS Relative (95% CI)

Effect Absolute (95% CI) 28 fewer per 1000 (from 5 fewer to 39 fewer) 13 fewer per 1000 (from 22 more to 34 fewer) 6 more per 1000 (from 1 fewer to 74 more) 92 more per 1000 (from 13 more to 234 more) 27 fewer per 1000 (from 50 more to 72 fewer) Quality Importance

Admission to neonatal special care (all levels) for respiratory complications randomized trials serious4 not serious very serious1 not serious 11/467 (2.4%) 24/475 (5.1%) RR 0.45 (0.22 to 0.90) ⊕OOO VERY LOW IMPORTANT

Admission to neonatal special care (all levels) for any indication 1 randomized trials serious4 not serious very serious1 serious7 not serious 26/467 (5.6%) 32/475 (6.7%) RR 0.81 (0.49 to 1.33) ⊕OOO VERY LOW IMPORTANT

Use of mechanical ventilation 1 randomized trials serious4 not serious very serious1 very serious3 not serious 4/467 (0.9%) 1/475 (0.2%) RR 4.07 (0.46 to 36.27) ⊕OOO VERY LOW CRITICAL

Lower quarter of academic ability at childhood follow-up 1 randomized trials serious8 not serious very serious1 not serious strong association 33/186 (17.7%) 14/164 (8.5%) RR 2.08 (1.15 to 3.74) ⊕OOO VERY LOW CRITICAL

Reported learning difficulty at childhood follow-up 1 randomized trials serious8 not serious very serious1 not serious not serious 25/217 (11.5%) 27/190 (14.2%) RR 0.81 (0.49 to 1.35) ⊕OOO VERY LOW CRITICAL

1 Evidence was derived from a population that does not correspond to the population of interest (i.e. women undergoing elective caesarean section at term rather than in late preterm). 2 No events reported for outcome. 3 Wide confidence interval crossing the line of no effect and few events. 4 For this outcome, the study was at moderate risk of bias due to non-blinded design, with the potential for performance and detection bias. 5 Wide confidence interval crossing line of no effect and small sample size. 6 Few events. 7 Wide confidence interval crossing the line of no effect. 8 Study at risk of attrition bias.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 1i.

Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women at risk of preterm birth (women with hypertension in pregnancy) Source: Roberts D, Dalziel SR. Antenatal corticosteroids for accelerating fetal lung maturation for women at risk of preterm birth. Cochrane Database Syst Rev. 2006;(3):CD004454. (updated for this guideline) Quality assessment No. of patients Imprecision Other considerations ACS Placebo or no treatment (women with hypertension) 1/108 (0.9%) Relative (95% CI) Effect Absolute Quality Importance

No. of studies

Design

Risk of bias

Inconsistency

Indirectness

Maternal death — in women with pregnancies complicated by hypertension syndromes 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 1/110 (0.9%) RR 0.98 (0.06 to 15.50) 0 fewer per 1000 (from 9 fewer to 134 more) 9 more per 1000 (from 4 fewer to 94 more) 39 fewer per 1000 (from 84 fewer to 63 more) 19 fewer per 1000 (from 55 fewer to 78 more) 50 fewer per 1000 (from 126 fewer to 59 more) 58 more per 1000 (from 7 fewer to 182 more) ⊕⊕OO LOW CRITICAL

Chorioamnionitis — in women with pregnancies complicated by hypertension syndromes 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 3/155 (1.9%) 1/156 (0.6%) RR 2.36 (0.36 to 15.73) ⊕⊕OO LOW CRITICAL

Puerperal sepsis — in women with pregnancies complicated by hypertension syndromes 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 9/110 (8.2%) 13/108 (12.0%) RR 0.68 (0.30 to 1.52) ⊕⊕OO LOW CRITICAL

Admission into adult intensive care unit — in women with pregnancies complicated by hypertension syndromes 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 6/110 (5.5%) 8/108 (7.4%) RR 0.74 (0.26 to 2.05) ⊕⊕OO LOW CRITICAL

Fetal and neonatal deaths — in babies born from pregnancies complicated by hypertension syndromes 2 randomized trials no serious risk of bias serious3 no serious indirectness serious4 none 38/156 (24.4%) 46/157 (29.3%) RR 0.83 (0.57 to 1.20) ⊕⊕OO LOW CRITICAL

Fetal deaths — in babies born from pregnancies complicated by hypertension syndromes 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious4 none 22/168 (13.1%) 13/163 (8.0%) RR 1.73 (0.91 to 3.28) ⊕⊕⊕O MODERATE CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (women with hypertension) 33/144 (22.9%) Relative (95% CI)

Effect Absolute Quality Importance

Neonatal deaths — in babies born from pregnancies complicated by hypertension syndromes 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious5 none 16/134 (11.9%) RR 0.50 (0.29 to 0.87) 115 fewer per 1000 (from 30 fewer to 163 fewer) 178 fewer per 1000 (from 100 fewer to 231 fewer) 82 fewer per 1000 (from 17 fewer to 110 fewer) 151 fewer per 1000 (from 45 fewer to 207 fewer) 125 fewer per 1000 (from 36 fewer to 183 fewer) 20 fewer per 1000 (from 36 fewer to 67 more) MD 131.72 lower (319.68 lower to 56.24 higher) ⊕⊕⊕O MODERATE CRITICAL

Respiratory distress syndrome — in babies born from pregnancies complicated by hypertension syndromes 5 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 33/191 (17.3%) 68/191 (35.6%) RR 0.50 (0.35 to 0.72) ⊕⊕⊕⊕ HIGH CRITICAL

Cerebroventricular haemorrhage — in babies born from pregnancies complicated by hypertension syndromes 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious6 none 7/134 (5.2%) 19/144 (13.2%) RR 0.38 (0.17 to 0.87) ⊕⊕⊕O MODERATE CRITICAL

Systemic infection in the first 48 hours of life — in babies born from pregnancies complicated by hypertension syndromes 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious5 none 13/100 (13.0%) 28/100 (28.0%) RR 0.46 (0.26 to 0.84) ⊕⊕⊕O MODERATE CRITICAL

Proven infection while in the neonatal intensive care unit — in babies born from pregnancies complicated by hypertension syndromes 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious5 none 21/134 (15.7%) 40/144 (27.8%) RR 0.55 (0.34 to 0.87) ⊕⊕⊕O MODERATE CRITICAL

Necrotizing enterocolitis — in babies born from pregnancies complicated by hypertension syndromes 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 2/100 (2.0%) 4/100 (4.0%) RR 0.50 (0.09 to 2.67) ⊕⊕OO LOW CRITICAL

Mean birth weight (g) — in babies born from pregnancies complicated by hypertension syndromes (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious7 none 46 49 — ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations ACS

No. of patients Placebo or no treatment (women with hypertension) 5/100 (5.0%) Relative (95% CI)

Effect Absolute Quality Importance

Chronic lung disease — in babies born from pregnancies complicated by hypertension syndromes 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 1/100 (1.0%) RR 0.20 (0.02 to 1.68) 40 fewer per 1000 (from 49 fewer to 34 more) ⊕⊕OO LOW CRITICAL

1 Wide confidence interval crossing the line of no effect, few events and small sample size. 2 Wide confidence interval crossing the line of no effect and few events. 3 Statistical Heterogeneity (I2 > 60%). 4 Wide confidence interval crossing the line of no effect. 5 Estimate based on small sample size. 6 Estimate based on small sample size and few events. 7 Wide confidence interval crossing the line of no effect and small sample size.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 1j.

Antenatal corticosteroids (ACS) versus placebo or no treatment for accelerating fetal lung maturation for women at risk of preterm birth (women with growth-restricted fetuses) Source: Amiya RM, Mlunde LB, Ota E, Mori R, Oladapo OT. Antenatal corticosteroid therapy for reducing adverse maternal and child outcomes in special populations of women at risk of imminent preterm birth: a systematic review and meta-analysis. 2014 (unpublished). Quality assessment No. of patients Imprecision Other considerations not serious ACS No ACS Relative (95% CI) OR 0.48 (0.03 to 8.68) Effect Absolute (95% CI) 0 fewer per 1000 (from 0 fewer to 0 fewer) 41 fewer per 1000 (from 94 more to 126 fewer) 30 fewer per 1000 (from 6 more to 68 fewer) 1 fewer per 1000 (from 0 fewer to 0 fewer) 52 fewer per 1000 (from 26 more to 128 fewer) 1 fewer per 1000 (from 0 fewer to 0 fewer) Quality Importance

No. of studies 1

Study design

Risk of bias

Inconsistency

Indirectness

Mode of delivery — caesarean section (small for gestational age or SGA) observational studies not serious not serious not serious very serious1 139/146 (95.2%) 19/19 (100.0%) ⊕OOO VERY LOW IMPORTANT

Chorioamnionitis — histological and/or clinical (SGA) 1 observational studies serious2 not serious not serious very serious1 not serious 11/63 (17.5%) 34/157 (21.7%) OR 0.77 (0.36 to 1.63) ⊕OOO VERY LOW CRITICAL

Perinatal death — fetal death or neonatal death (intrauterine growth-restricted or IUGR) 4 observational studies not serious not serious not serious serious3 not serious 41/324 (12.7%) 33/179 (18.4%) OR 0.81 (0.58 to 1.04) ⊕OOO VERY LOW CRITICAL

Perinatal death — fetal death or neonatal death (SGA) 6 observational studies not serious not serious not serious serious3 not serious —4 —4 OR 0.78 (0.58 to 1.04) ⊕OOO VERY LOW CRITICAL

Respiratory distress syndrome (IUGR) 4 observational studies serious2 not serious not serious serious3 not serious 142/324 (43.8%) 88/179 (49.2%) OR 0.81 (0.59 to 1.11) ⊕OOO VERY LOW CRITICAL

Respiratory distress syndrome (SGA) 8 observational studies serious2 not serious not serious serious3 not serious —4 —4 OR 0.83 (0.66 to 1.05) ⊕OOO VERY LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations not serious

No. of patients ACS No ACS Relative (95% CI)

Effect Absolute (95% CI) 25 fewer per 1000 (from 121 more to 132 fewer) 44 more per 1000 (from 27 fewer to 173 more) 13 fewer per 1000 (from 65 fewer to 92 more) 1 fewer per 1000 (from 0 fewer to 0 fewer) 43 fewer per 1000 (from 112 more to 166 fewer) 0 fewer per 1000 (from 66 fewer to 94 more) 2 fewer per 1000 (from 50 fewer to 216 more) Quality Importance

Surfactant use (IUGR) observational studies serious2 not serious not serious very serious1 19/53 (35.8%) 13/34 (38.2%) OR 0.90 (0.37 to 2.20) ⊕OOO VERY LOW CRITICAL

Surfactant use (SGA) 3 observational studies serious2 not serious not serious serious3 not serious 81/262 (30.9%) 47/210 (22.4%) OR 1.39 (0.85 to 2.28) ⊕OOO VERY LOW CRITICAL

Major brain lesion — intraventricular haemorrhage (IVH), intracranial haemorrhage (ICH), periventricular haemorrhage (PVH) or periventricular leukomalacia (PVL) (IUGR) 2 observational studies not serious not serious not serious very serious1 not serious 12/116 (10.3%) 10/96 (10.4%) OR 0.86 (0.35 to 2.10) ⊕OOO VERY LOW CRITICAL

Major brain lesion (IVH, ICH, PVH or PVL) (SGA) 5 observational studies not serious not serious not serious serious3 serious5 —4 —4 OR 0.57 (0.41 to 0.78) ⊕OOO VERY LOW CRITICAL

Neonatal sepsis (IUGR) 2 observational studies serious2 not serious not serious very serious1 not serious 45/115 (39.1%) 36/96 (37.5%) OR 0.83 (0.44 to 1.58) ⊕OOO VERY LOW CRITICAL

Neonatal sepsis (SGA) 3 observational studies serious2 not serious not serious serious3 not serious 51/178 (28.7%) 45/253 (17.8%) OR 1.00 (0.58 to 1.73) ⊕OOO VERY LOW CRITICAL

Necrotizing enterocolitis (IUGR) 1 observational studies serious2 not serious not serious very serious1 not serious 3/53 (5.7%) 2/34 (5.9%) OR 0.96 (0.15 to 6.07) ⊕OOO VERY LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 3 Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations not serious

No. of patients ACS No ACS Relative (95% CI)

Effect Absolute (95% CI) 3 fewer per 1000 (from 20 fewer to 66 more) 70 fewer per 1000 (from 14 more to 138 fewer) 71 fewer per 1000 (from 14 more to 139 fewer) 13 more per 1000 (from 27 fewer to 255 more) 51 more per 1000 (from 14 fewer to 161 more) 202 fewer per 1000 (from 39 more to 416 fewer) MD 1.09 higher (from 0.86 lower to 3.05 higher) Quality Importance

Necrotizing enterocolitis (SGA) observational studies serious2 not serious not serious very serious1 4/116 (3.4%) 5/191 (2.6%) OR 0.90 (0.22 to 3.76) ⊕OOO VERY LOW CRITICAL

Chronic lung disease/bronchopulmonary dysplasia (IUGR) 3 observational studies serious2 not serious not serious serious3 not serious 47/211 (22.3%) 44/151 (29.1%) OR 0.69 (0.43 to 1.13) ⊕OOO VERY LOW CRITICAL

Chronic lung disease/bronchopulmonary dysplasia (SGA) 4 observational studies serious2 not serious not serious serious3 not serious 81/357 (22.7%) 50/170 (29.4%) OR 0.69 (0.44 to 1.07) ⊕OOO VERY LOW CRITICAL

Patent ductus arteriosus (IUGR) 1 observational studies serious2 not serious not serious very serious1 not serious 10/53 (18.9%) 6/34 (17.6%) OR 1.09 (0.35 to 3.32) ⊕OOO VERY LOW CRITICAL

Patent ductus arteriosus (SGA) 2 observational studies serious2 not serious not serious serious3 not serious 19/116 (16.4%) 16/191 (8.4%) OR 1.70 (0.82 to 3.54) ⊕OOO VERY LOW CRITICAL

Low birth weight < 3rd percentile for gestational age (SGA) 1 observational studies not serious not serious not serious very serious1 not serious 63/146 (43.2%) 12/19 (63.2%) OR 0.44 (0.16 to 1.19) ⊕OOO VERY LOW CRITICAL

Duration of mechanical ventilation, days (IUGR) 2 observational studies not serious not serious not serious very serious1 not serious 115 96 — ⊕OOO VERY LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 2 Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations not serious

No. of patients ACS No ACS Relative (95% CI)

Effect Absolute (95% CI) 54 more per 1000 (from 80 fewer to 185 more) 10 more per 1000 (from 105 fewer to 125 more) 177 more per 1000 (from 24 more to 269 more) 214 more per 1000 (from 25 more to 530 more) Quality Importance

Use of mechanical ventilation (IUGR) observational studies not serious not serious not serious very serious1 61/115 (53.0%) 45/96 (46.9%) OR 1.24 (0.72 to 2.14) ⊕OOO VERY LOW CRITICAL

Use of mechanical ventilation (SGA) 3 observational studies not serious not serious not serious serious3 not serious 127/261 (48.7%) 56/115 (48.7%) OR 1.04 (0.65 to 1.66) ⊕OOO VERY LOW CRITICAL

Survival without handicap at 2 years corrected age (IUGR) 1 observational studies not serious not serious not serious serious6 not serious 51/62 (82.3%) 40/62 (64.5%) OR 2.55 (1.11 to 5.87) ⊕OOO VERY LOW CRITICAL

Growth < 10th percentile in early childhood (follow up to school age) (IUGR) 1 observational studies not serious not serious not serious very serious6 strong association 14/49 (28.6%) 3/42 (7.1%) OR 5.20 (1.38 to 19.62) ⊕OOO VERY LOW CRITICAL

1 Wide confidence interval crossing the line of no effect and small sample size. 2 Evidence based heavily or entirely on studies with design limitations including lack of adjustment for potential confounding factors. 3 Wide confidence interval crossing the line of no effect. 4 Raw data unavailable for one of the included studies (only ORs and 95% CIs reported); generic inverse variance method used for meta-analysis. 5 Funnel plot suggests the presence of some degree of publication bias. 6 Wide confidence interval crossing the line of no effect, small sample size, and few events.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 1k.

Different corticosteroids and regimens for accelerating fetal lung maturation for women at risk of preterm birth (any regimen of dexamethasone and betamethasone) Source: Brownfoot FC, Gagliardi DI, Bain E, Middleton P, Crowther CA. Different corticosteroids and regimens for accelerating fetal lung maturation for women at risk of preterm birth. Cochrane Database Syst Rev. 2013;(8):CD006764. Quality assessment No. of patients Imprecision Other considerations none Dexamethasone (any regimen) 120 Betamethasone (any regimen) 120 Relative (95% CI) — Effect Absolute Quality Importance

No. of studies 1

Design

Risk of bias

Inconsistency

Indirectness

Interval between admission and birth (days) (better indicated by higher values) randomized trials serious1 serious2 no serious indirectness very serious3 MD 3.48 higher (3.38 lower to 10.34 higher) MD 0 higher (0.99 lower to 0.99 higher) MD 7 higher (5.56 to 8.44 higher) 8 more per 1000 (from 9 fewer to 50 more) 18 more per 1000 (from 36 fewer to 82 more) 21 fewer per 1000 (from 30 fewer to 8 more) ⊕OOO VERY LOW CRITICAL

Interval between admission and birth (days) — dexamethasone vs betamethasone; ruptured membranes (better indicated by higher values) 1 randomized trials serious1 no serious inconsistency no serious indirectness serious4 none 60 60 — ⊕⊕OO LOW CRITICAL

Interval between admission and birth (days) — dexamethasone vs betamethasone (intact membranes) (better indicated by higher values) 1 randomized trials serious1 no serious inconsistency no serious indirectness serious4 none 60 60 — ⊕⊕OO LOW CRITICAL

Neonatal death 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious5 none 8/278 (2.9%) 6/318 (1.9%) RR 1.41 (0.54 to 3.67) ⊕⊕⊕O MODERATE CRITICAL

Respiratory distress syndrome 5 randomized trials serious6 no serious inconsistency no serious indirectness serious5 none 122/354 (34.5%) 121/399 (30.3%) RR 1.06 (0.88 to 1.27) ⊕⊕OO LOW CRITICAL

Severe intraventricular haemorrhage 4 randomized trials serious6 no serious inconsistency no serious indirectness serious7 none 4/257 (1.6%) 10/292 (3.4%) RR 0.40 (0.13 to 1.24) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 4 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Dexamethasone (any regimen) 9/257 (3.5%) Betamethasone (any regimen) 21/292 (7.2%) Relative (95% CI)

Effect Absolute Quality Importance

Intraventricular haemorrhage (all grades) randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision RR 0.44 (0.21 to 0.92) 40 fewer per 1000 (from 6 fewer to 57 fewer) 26 more per 1000 (from 19 fewer to 104 more) 4 more per 1000 (from 8 fewer to 45 more) 9 fewer per 1000 (from 53 fewer to 61 more) 72 fewer per 1000 (from 228 fewer to 157 more) MD 0.01 higher (0.11 lower to 0.12 higher) 152 more per 1000 (from 119 fewer to 1000 more) ⊕⊕⊕⊕ HIGH CRITICAL

Neonatal sepsis 2 randomized trials serious8 no serious inconsistency no serious indirectness serious5 none 29/254 (11.4%) 23/262 (8.8%) RR 1.30 (0.78 to 2.19) ⊕⊕OO LOW CRITICAL

Necrotizing enterocolitis 3 randomized trials very serious9 no serious inconsistency no serious indirectness very serious7 none 5/294 (1.7%) 4/304 (1.3%) RR 1.29 (0.38 to 4.40) ⊕OOO VERY LOW CRITICAL

Retinopathy of prematurity 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious5 none 31/254 (12.2%) 34/262 (13.0%) RR 0.93 (0.59 to 1.47) ⊕⊕⊕O MODERATE CRITICAL

Low birth weight 1 randomized trials serious1 no serious inconsistency no serious indirectness very serious2 none 21/36 (58.3%) 45/69 (65.2%) RR 0.89 (0.65 to 1.24) ⊕OOO VERY LOW CRITICAL

Birth weight (kg) (better indicated by higher values) 5 randomized trials serious6 no serious inconsistency no serious indirectness no serious imprecision none 348 386 — ⊕⊕⊕O MODERATE CRITICAL

Neonatal intensive care unit admission 2 randomized trials serious6 serious2 no serious indirectness serious5 none 42/156 (26.9%) 40/189 (21.2%) RR 1.72 (0.44 to 6.72) ⊕OOO VERY LOW IMPORTANT

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Dexamethasone (any regimen) 34 Betamethasone (any regimen) 36 Relative (95% CI) —

Effect Absolute Quality Importance

Neonatal intensive care unit stay (days) (better indicated by lower values) randomized trials serious1 no serious inconsistency no serious indirectness serious4 MD 0.91 lower (1.77 to 0.05 lower) — ⊕⊕OO LOW CRITICAL

Neurosensory disability as a child (18 months) 1 randomized trials randomized trials serious1 no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious10 none 1/8 (12.5%) 4/330 (1.2%) 0/4 (0.0%) 5/373 (1.3%) RR 1.67 (0.08 to 33.75) RR 0.83 (0.23 to 3.03) ⊕OOO VERY LOW CRITICAL

Periventricular leukomalacia 4 no serious risk of bias very serious7 none 2 fewer per 1000 (from 10 fewer to 27 more) 162 more per 1000 (from 97 fewer to 1000 more) ⊕⊕OO LOW CRITICAL

Bronchopulmonary dysplasia 2 randomized trials no serious risk of bias serious2 no serious indirectness serious5 none 22/214 (10.3%) 27/250 (10.8%) RR 2.50 (0.10 to 61.34) ⊕⊕OO LOW CRITICAL

1 One study with design limitations. 2 Statistical heterogeneity (I2 > 60%). 3 Wide confidence interval crossing the line of no effect and small sample size. 4 Estimate based on small sample size. 5 Wide confidence interval crossing the line of no effect. 6 Most studies contributing data had design limitations. 7 Wide confidence interval crossing the line of no effect and few events. 8 One of the studies contributing data had serious design limitations. 9 Most studies contributing data had design limitations, with more than 40% of weight from a study with serious design limitations. 10 Wide confidence interval crossing the line of no effect, few events and small sample size.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 1l.

Repeat course(s) versus single course of antenatal corticosteroids (ACS) for accelerating fetal lung maturation for women at risk of preterm birth Source: Crowther CA, McKinlay CJ, Middleton P, Harding JE. Repeat doses of prenatal corticosteroids for women at risk of preterm birth for improving neonatal health outcomes. Cochrane Database Syst Rev. 2011;(6):CD003935. (updated for this guideline) Quality assessment No. of patients Imprecision Other considerations Repeat course(s) of ACS 106/818 (13.0%) Single course of ACS Relative (95% CI) Effect Absolute Quality Importance

No. of studies

Design

Risk of bias

Inconsistency

Indirectness

Birth < 28 weeks of gestation 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 99/814 (12.2%) RR 1.07 (0.83 to 1.38) 9 more per 1000 (from 21 fewer to 46 more) 7 more per 1000 (from 34 fewer to 47 more) 25 fewer per 1000 (from 67 fewer to 17 more) MD 0.09 lower (0.33 lower to 0.15 higher) 6 more per 1000 (from 7 fewer to 24 more) 9 more per 1000 (from 4 fewer to 26 more) ⊕⊕⊕O MODERATE CRITICAL

Birth < 34 weeks 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 717/1058 (67.8%) 728/1082 (67.3%) RR 1.01 (0.95 to 1.07) ⊕⊕⊕⊕ HIGH CRITICAL

Preterm birth < 37 weeks 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 475/585 (81.2%) 501/596 (84.1%) RR 0.97 (0.92 to 1.02) ⊕⊕⊕⊕ HIGH CRITICAL

Mean gestational age at birth (weeks) (better indicated by higher values) 8 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 1586 1593 — ⊕⊕⊕O MODERATE CRITICAL

Puerperal sepsis 5 randomized trials serious2 no serious inconsistency no serious indirectness serious1 none 72/1565 (4.6%) 61/1526 (4.0%) RR 1.15 (0.83 to 1.60) ⊕⊕OO LOW CRITICAL

Chorioamnionitis 6 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 140/2152 (6.5%) 118/2109 (5.6%) RR 1.16 (0.92 to 1.46) ⊕⊕⊕O MODERATE CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations

No. of patients Repeat course(s) of ACS 96/2791 (3.4%) Single course of ACS Relative (95% CI)

Effect Absolute Quality Importance

Fetal and neonatal mortality 9 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 102/2763 (3.7%) RR 0.94 (0.71 to 1.23) 2 fewer per 1000 (from 11 fewer to 8 more) 1 fewer per 1000 (from 3 fewer to 7 more) 3 fewer per 1000 (from 15 fewer to 13 more) 60 fewer per 1000 (from 32 fewer to 88 fewer) 5 fewer per 1000 (from 22 fewer to 16 more) 2 more per 1000 (from 4 fewer to 10 more) 6 fewer per 1000 (from 11 fewer to 2 more) ⊕⊕⊕O MODERATE CRITICAL

Fetal death 7 randomized trials serious2 no serious inconsistency no serious indirectness very serious3 none 4/1375 (0.3%) 5/1380 (0.4%) RR 0.82 (0.24 to 2.84) ⊕OOO VERY LOW CRITICAL

Neonatal death 7 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 47/1352 (3.5%) 52/1361 (3.8%) RR 0.91 (0.62 to 1.34) ⊕⊕⊕O MODERATE CRITICAL

Respiratory distress syndrome 8 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 463/1603 (28.9%) 565/1603 (35.2%) RR 0.83 (0.75 to 0.91) ⊕⊕⊕⊕ HIGH CRITICAL

Intraventricular haemorrhage 6 randomized trials serious2 no serious inconsistency no serious indirectness no serious imprecision none 129/1533 (8.4%) 137/1532 (8.9%) RR 0.94 (0.75 to 1.18) ⊕⊕⊕O MODERATE CRITICAL

Intraventricular haemorrhage — grade 3 or 4 6 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 32/2419 (1.3%) 28/2400 (1.2%) RR 1.13 (0.69 to 1.86) ⊕⊕⊕O MODERATE CRITICAL

Necrotizing enterocolitis 8 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 45/2709 (1.7%) 60/2685 (2.2%) RR 0.74 (0.51 to 1.08) ⊕⊕⊕O MODERATE CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations

No. of patients Repeat course(s) of ACS 140/2446 (5.7%) Single course of ACS Relative (95% CI)

Effect Absolute Quality Importance

Retinopathy of prematurity 7 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 137/2437 (5.6%) RR 1.02 (0.81 to 1.28) 1 more per 1000 (from 11 fewer to 16 more) 51 fewer per 1000 (from 12 fewer to 82 fewer) 17 fewer per 1000 (from 52 fewer to 27 more) 15 more per 1000 (from 2 fewer to 35 more) MD 75.79 lower (117.63 to 33.96 lower) 5 more per 1000 (from 25 fewer to 35 more) 4 more per 1000 (from 8 fewer to 19 more) ⊕⊕⊕O MODERATE CRITICAL

Use of surfactant 9 randomized trials no serious risk of bias serious4 no serious indirectness no serious imprecision none 514/2772 (18.5%) 643/2753 (23.4%) RR 0.78 (0.65 to 0.95) ⊕⊕⊕O MODERATE CRITICAL

Early systemic neonatal infection (variously defined) 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 177/763 (23.2%) 193/781 (24.7%) RR 0.93 (0.79 to 1.11) ⊕⊕⊕⊕ HIGH CRITICAL

Small for gestational age at birth 7 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 191/1996 (9.6%) 163/1979 (8.2%) RR 1.18 (0.97 to 1.43) ⊕⊕⊕O MODERATE CRITICAL

Mean birth weight (g) (better indicated by higher values) 9 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 2820 2806 — ⊕⊕⊕⊕ HIGH CRITICAL

Admission to the neonatal intensive care unit 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 872/1731 (50.4%) 863/1717 (50.3%) RR 1.01 (0.95 to 1.07) ⊕⊕⊕⊕ HIGH IMPORTANT

Chronic lung disease 8 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 181/2709 (6.7%) 170/2684 (6.3%) RR 1.06 (0.87 to 1.30) ⊕⊕⊕O MODERATE CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations

No. of patients Repeat course(s) of ACS 20/2453 (0.8%) Single course of ACS Relative (95% CI)

Effect Absolute Quality Importance

Periventricular leukomalacia 7 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 26/2435 (1.1%) RR 0.77 (0.43 to 1.37) 2 fewer per 1000 (from 6 fewer to 4 more) 8 more per 1000 (from 23 fewer to 39 more) 8 more per 1000 (from 68 fewer to 93 more) 7 fewer per 1000 (from 61 fewer to 58 more) 5 fewer per 1000 (from 27 fewer to 22 more) 2 more per 1000 (from 4 fewer to 14 more) 1 fewer per 1000 (from 3 fewer to 6 more) ⊕⊕⊕O MODERATE CRITICAL

Survival free of any disability to early childhood follow-up 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 1241/1584 (78.3%) 1215/1571 (77.3%) RR 1.01 (0.97 to 1.05) ⊕⊕⊕⊕ HIGH CRITICAL

Survival free of major neurosensory disability to early childhood follow-up 2 randomized trials serious2 serious4 no serious indirectness no serious imprecision none 557/642 (86.8%) 572/675 (84.7%) RR 1.01 (0.92 to 1.11) ⊕⊕OO LOW CRITICAL

Disability at early childhood follow-up 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 175/495 (35.4%) 182/504 (36.1%) RR 0.98 (0.83 to 1.16) ⊕⊕⊕⊕ HIGH CRITICAL

Developmental delay at early childhood follow-up 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 260/1603 (16.2%) 269/1599 (16.8%) RR 0.97 (0.84 to 1.13) ⊕⊕⊕⊕ HIGH CRITICAL

Blindness at early childhood follow-up 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 24/1590 (1.5%) 20/1561 (1.3%) RR 1.17 (0.65 to 2.10) ⊕⊕⊕O MODERATE CRITICAL

Deafness at early childhood follow-up 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious3 none 6/1710 (0.4%) 7/1695 (0.4%) RR 0.85 (0.29 to 2.52) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations

No. of patients Repeat course(s) of ACS 56/1948 (2.9%) Single course of ACS Relative (95% CI)

Effect Absolute Quality Importance

Cerebral palsy at early childhood follow-up 5 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 54/1935 (2.8%) RR 1.03 (0.71 to 1.49) 1 more per 1000 (from 8 fewer to 14 more) 6 fewer per 1000 (from 164 fewer to 627 more) ⊕⊕⊕O MODERATE CRITICAL

Any maternal side-effects of therapy 2 randomized trials serious2 serious4 no serious indirectness serious1 none 115/739 (15.6%) 159/735 (21.6%) RR 0.97 (0.24 to 3.90) ⊕OOO VERY LOW IMPORTANT

1 Wide confidence interval crossing the line of no effect. 2 Most studies contributing data had design limitations. 3 Wide confidence interval crossing the line of no effect and few events. 4 Statistical heterogeneity (I2 > 60%).

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 2a.

Betamimetics for inhibiting preterm labour Source: Neilson JP, West HM, Dowswell T. Betamimetics for inhibiting preterm labour. Cochrane Database Syst Rev. 2014;(2):CD004352. Quality assessment No. of patients Imprecision Other considerations none All betamimetics 404/654 (61.8%) Placebo Relative (95% CI) RR 0.95 (0.88 to 1.03) Effect Absolute Quality Importance

No. of studies 10

Design

Risk of bias

Inconsistency

Indirectness

Delivery < 37 weeks of gestation randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision 383/558 (68.6%) 34 fewer per 1000 (from 82 fewer to 21 more) not pooled ⊕⊕⊕⊕ HIGH CRITICAL

Maternal death 2 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious1 none 0/502 (0.0%) 77/590 (13.1%) 0/405 (0.0%) 5/491 (1.0%) not pooled ⊕⊕OO LOW CRITICAL

Cessation of treatment due to adverse drug reaction 5 no serious imprecision none RR 11.38 (5.21 to 24.86) 106 more per 1000 (from 43 more to 243 more) 125 fewer per 1000 (from 47 fewer to 184 fewer) 104 fewer per 1000 (from 10 fewer to 182 fewer) 5 fewer per 1000 (from 17 fewer to 18 more) 2 fewer per 1000 (from 16 fewer to 44 more) ⊕⊕⊕⊕ HIGH CRITICAL

Birth within 48 hours of treatment 10 randomized trials serious2 no serious inconsistency no serious indirectness no serious imprecision none 151/652 (23.2%) 218/557 (39.1%) RR 0.68 (0.53 to 0.88) ⊕⊕⊕O MODERATE CRITICAL

Delivery within 7 days 5 randomized trials serious2 no serious inconsistency no serious indirectness no serious imprecision none 184/454 (40.5%) 238/457 (52.1%) RR 0.80 (0.65 to 0.98) ⊕⊕⊕O MODERATE CRITICAL

Perinatal death (7 days) 11 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 16/712 (2.2%) 20/620 (3.2%) RR 0.84 (0.46 to 1.55) ⊕⊕⊕O MODERATE CRITICAL

Neonatal death 6 randomized trials serious2 no serious inconsistency no serious indirectness serious3 none 19/629 (3.0%) 12/545 (2.2%) RR 0.90 (0.27 to 3.00) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients All betamimetics 1/370 (0.3%) Placebo Relative (95% CI)

Effect Absolute Quality Importance

Infant death randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 2/380 (0.5%) RR 0.51 (0.05 to 5.64) 3 fewer per 1000 (from 5 fewer to 24 more) 31 fewer per 1000 (from 69 fewer to 19 more) 33 fewer per 1000 (from 40 fewer to 26 more) 24 fewer per 1000 (from 38 fewer to 72 more) 168 more per 1000 (from 79 fewer to 1000 more) 467 more per 1000 (from 40 more to 1000 more) 60 more per 1000 (from 44 fewer to 610 more) ⊕⊕OO LOW CRITICAL

Respiratory distress syndrome 8 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 123/664 (18.5%) 136/575 (23.7%) RR 0.87 (0.71 to 1.08) ⊕⊕⊕⊕ HIGH CRITICAL

Cerebral palsy 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 none 1/125 (0.8%) 5/121 (4.1%) RR 0.19 (0.02 to 1.63) ⊕⊕OO LOW CRITICAL

Necrotizing enterocolitis 2 randomized trials serious2 no serious inconsistency no serious indirectness very serious5 none 1/75 (1.3%) 3/74 (4.1%) RR 0.42 (0.06 to 2.78) ⊕OOO VERY LOW CRITICAL

Neonatal sepsis or infection 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 45/399 (11.3%) 40/410 (9.8%) RR 2.72 (0.19 to 39.63) ⊕⊕⊕O MODERATE CRITICAL

Fetal tachycardia 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious6 none 12/15 (80.0%) 5/15 (33.3%) RR 2.40 (1.12 to 5.13) ⊕⊕⊕O MODERATE CRITICAL

Fetal hypoglycaemia 3 randomized trials serious2 serious7 no serious indirectness serious3 none 143/427 (33.5%) 29/430 (6.7%) RR 1.89 (0.35 to 10.04) ⊕OOO VERY LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 3 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients All betamimetics 1/425 (0.2%) 6/54 (11.1%) 214/592 (36.1%) Placebo Relative (95% CI)

Effect Absolute Quality Importance

Maternal pulmonary oedema randomized trials randomized trials randomized trials no serious risk of bias serious8 no serious inconsistency no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness no serious indirectness serious4 0/427 (0.0%) 0/52 (0.0%) 19/497 (3.8%) RR 3.03 (0.12 to 74.23) RR 12.53 (0.72 to 216.91) RR 9.91 (6.46 to 15.20) — ⊕⊕⊕O MODERATE CRITICAL

Myocardial ischaemia 1 very serious5 none — ⊕OOO VERY LOW CRITICAL

Palpitation 5 no serious risk of bias no serious imprecision none 341 more per 1000 (from 209 more to 543 more) 300 more per 1000 (from 291 fewer to 1000 more) 14 more per 1000 (from 1 fewer to 89 more) 76 more per 1000 (from 21 more to 239 more) 161 more per 1000 (from 84 more to 280 more) 17 more per 1000 (from 26 fewer to 593 more) ⊕⊕⊕⊕ HIGH CRITICAL

Tachycardia 2 randomized trials no serious risk of bias serious7 no serious indirectness very serious9 none 65/165 (39.4%) 19/64 (29.7%) RR 2.01 (0.02 to 252.89) ⊕OOO VERY LOW CRITICAL

Cardiac arrhythmias 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 7/352 (2.0%) 2/356 (0.6%) RR 3.54 (0.74 to 16.92) ⊕⊕OO LOW CRITICAL

Chest pain 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 39/406 (9.6%) 3/408 (0.7%) RR 11.29 (3.81 to 33.46) ⊕⊕⊕⊕ HIGH CRITICAL

Headaches 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 98/516 (19.0%) 22/420 (5.2%) RR 4.07 (2.60 to 6.35) ⊕⊕⊕⊕ HIGH CRITICAL

Hypotension 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 none 4/69 (5.8%) 2/67 (3.0%) RR 1.56 (0.12 to 20.86) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients All betamimetics 106/352 (30.1%) Placebo Relative (95% CI)

Effect Absolute Quality Importance

Hyperglycaemia randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision 37/356 (10.4%) RR 2.9 (2.05 to 4.09) 197 more per 1000 (from 109 more to 321 more) 328 more per 1000 (from 194 more to 530 more) 98 more per 1000 (from 42 more to 198 more) 356 more per 1000 (from 190 more to 642 more) MD 1.30 higher (2.74 lower to 5.34 higher) MD 5.20 higher (0.56 to 9.84 higher) ⊕⊕⊕⊕ HIGH CRITICAL

Hypokalaemia 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 138/352 (39.2%) 23/356 (6.5%) RR 6.07 (4.00 to 9.2) ⊕⊕⊕⊕ HIGH CRITICAL

Dyspnoea 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 55/406 (13.5%) 14/408 (3.4%) RR 3.86 (2.21 to 6.77) ⊕⊕⊕⊕ HIGH CRITICAL

Tremor 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 138/352 (39.2%) 13/356 (3.7%) RR 10.74 (6.20 to 18.59) ⊕⊕⊕⊕ HIGH CRITICAL

Infant long-term neurological development (Bayley score: psychomotor development) (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious6 none 125 121 — ⊕⊕⊕O MODERATE CRITICAL

Infant long-term neurological development (Bayley score: mental development) (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious6 none 125 121 — ⊕⊕⊕O MODERATE CRITICAL

1 No events. 2 Most studies contributing data had design limitations. 3 Wide confidence interval crossing the line of no effect. 4 Wide confidence interval crossing the line of no effect and few events. 5 Wide confidence interval crossing the line of no effect, few events and small sample size. 6 Estimate based on small sample size. 7 Statistical heterogeneity (I2 > 60%). 8 One study with design limitations. 9 Wide confidence interval crossing the line of no effect and small sample size.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 2b.

Calcium channel blockers for inhibiting preterm labour Source: Flenady V, Wojcieszek AM, Papatsonis DNM, Stock OM, Murray L, Jardine LA, Carbonne B. Calcium channel blockers for inhibiting preterm labour and birth. Cochrane Database Syst Rev. 2014;(6):CD002255. Quality assessment No. of patients Imprecision Other Any calcium considerations channel blocker none 65/101 (64.4%) Placebo Relative (95% CI) RR 0.65 (0.18 to 2.43) Effect Absolute Quality Importance

No. of studies 2

Design

Risk of bias

Inconsistency

Indirectness

Preterm birth (< 37 weeks of gestation) randomized trials no serious risk of bias serious1 no serious indirectness very serious2 69/72 (95.8%) 335 fewer per 1000 (from 786 fewer to 1000 more) 603 fewer per 1000 (from 491 fewer to 680 fewer) — ⊕OOO VERY LOW CRITICAL

Birth < 48 hours after trial entry 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 27/101 (26.7%) 62/72 (86.1%) RR 0.30 (0.21 to 0.43) ⊕⊕⊕O MODERATE CRITICAL

Maternal adverse drug reaction 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 25/45 (55.6%) 0/44 (0%) RR 49.89 (3.13 to 795.02) ⊕⊕⊕O MODERATE CRITICAL

1 Statistical heterogeneity (I2 > 60%). 2 Wide confidence interval crossing the line of no effect and small sample size. 3 Estimate based on small sample size.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 2c.

Cyclo-oxygenase (COX) inhibitors for inhibiting preterm labour Source: King JF, Flenady V, Cole S, Thornton S. Cyclo-oxygenase (COX) inhibitors for treating preterm labour. Cochrane Database Syst Rev. 2005;(2):CD001992. (updated for this guideline) Quality assessment No. of patients Imprecision Other considerations none Any Cox inhibitors 3/18 (16.7%) Placebo Relative (95% CI) RR 0.21 (0.07 to 0.62) Effect Absolute Quality Importance

No. of studies 1

Design

Risk of bias

Inconsistency

Indirectness

Preterm birth < 37 weeks of gestation randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 14/18 (77.8%) 614 fewer per 1000 (from 296 fewer to 723 fewer) 489 fewer per 1000 (from 593 fewer to 171 more) 442 fewer per 1000 (from 675 fewer to 495 more) MD 3.53 higher (1.13 to 5.92 higher) 68 more per 1000 (from 40 fewer to 327 more) 57 more per 1000 (from 34 fewer to 461 more) ⊕⊕⊕O MODERATE CRITICAL

Delivery within 48 hours of initiation of treatment 2 randomized trials no serious risk of bias serious2 no serious indirectness very serious3 none 4/34 (11.8%) 22/36 (61.1%) RR 0.20 (0.03 to 1.28) ⊕OOO VERY LOW CRITICAL

Delivery within 7 days of initiation of treatment 2 randomized trials no serious risk of bias serious2 no serious indirectness very serious4 none 11/34 (32.4%) 27/36 (75.0%) RR 0.41 (0.10 to 1.66) ⊕OOO VERY LOW CRITICAL

Gestation at birth (weeks) (better indicated by higher values) 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 34 33 — ⊕⊕⊕O MODERATE CRITICAL

Maternal adverse drug reaction 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious3 none 9/50 (18.0%) 6/51 (11.8%) RR 1.58 (0.66 to 3.78) ⊕⊕OO LOW CRITICAL

Chorioamnionitis or endometritis 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious3 none 4/31 (12.9%) 2/33 (6.1%) RR 1.94 (0.44 to 8.60) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Any Cox inhibitors 7/16 (43.8%) Placebo Relative (95% CI)

Effect Absolute Quality Importance

Postpartum haemorrhage randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious3 2/18 (11.1%) RR 3.94 (0.95 to 16.29) 327 more per 1000 (from 6 fewer to 1000 more) 19 fewer per 1000 (from 71 fewer to 149 more) — ⊕⊕OO LOW CRITICAL

Perinatal mortality 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious3 none 4/53 (7.5%) 5/53 (9.4%) RR 0.80 (0.25 to 2.58) ⊕⊕OO LOW CRITICAL

Intraventricular haemorrhage — grade 3 or 4 1 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious3 none 1/19 (5.3%) 0/35 (0.0%) 0/20 (0.0%) 1/35 (2.9%) RR 3.15 (0.14 to 72.88) RR 0.31 (0.01 to 7.15) ⊕⊕OO LOW CRITICAL

Neonatal sepsis 2 very serious3 none 20 fewer per 1000 (from 28 fewer to 176 more) 60 more per 1000 (from 96 fewer to 669 more) 3 fewer per 1000 (from 68 fewer to 294 more) 0 fewer per 1000 (from 91 fewer to 225 more) ⊕⊕OO LOW CRITICAL

Patent ductus arteriosus 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious3 none 4/19 (21.1%) 3/20 (15.0%) RR 1.40 (0.36 to 5.46) ⊕⊕OO LOW CRITICAL

Necrotizing enterocolitis 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious3 none 3/35 (8.6%) 3/35 (8.6%) RR 0.97 (0.21 to 4.43) ⊕⊕OO LOW CRITICAL

Respiratory distress syndrome 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious3 none 8/53 (15.1%) 8/53 (15.1%) RR 1.00 (0.40 to 2.49) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 2 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Any Cox inhibitors 5/35 (14.3%) Placebo Relative (95% CI)

Effect Absolute Quality Importance

Chronic neonatal lung disease randomized trials no serious risk of bias serious2 no serious indirectness very serious3 4/35 (11.4%) RR 0.96 (0.07 to 12.37) 5 fewer per 1000 (from 106 fewer to 1000 more) not pooled ⊕OOO VERY LOW CRITICAL

Premature closure of the ductus arteriosus 3 randomized trials randomized trials randomized trials no serious risk of bias no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness no serious indirectness very serious5 none 0/53 (0.0%) 0/53 (0.0%) 34 0/53 (0.0%) 0/53 (0.0%) 33 not pooled ⊕⊕OO LOW CRITICAL

Persistent pulmonary hypertension of the newborn 3 very serious5 none not pooled not pooled ⊕⊕OO LOW CRITICAL

Birth weight (better indicated by higher values) 2 serious1 none — MD 716.34 higher (425.52 to 1007.16 higher) 170 fewer per 1000 (from 374 fewer to 127 more) ⊕⊕⊕O MODERATE CRITICAL

Admission to neonatal intensive care nursery 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 13/19 (68.4%) 17/20 (85.0%) RR 0.80 (0.56 to 1.15) ⊕⊕OO LOW CRITICAL

1 Estimate based on small sample size. 2 Statistical heterogeneity (I2 > 60%). 3 Wide confidence interval crossing the line of no effect, few events and small sample size. 4 Wide confidence interval crossing the line of no effect and small sample size. 5 No events.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 2d.

Magnesium sulfate for inhibiting preterm labour Source: Crowther CA, Hiller JE, Doyle LW. Magnesium sulphate for preventing preterm birth in threatened preterm labour. Cochrane Database Syst Rev. 2002;(4):CD001060. (updated for this guideline) Quality assessment No. of patients Imprecision Other considerations none Magnesium sulfate 18/30 (60.0%) No tocolytic treatment 34/35 (97.1%) Relative (95% CI) RR 0.62 (0.46 to 0.83) Effect Absolute Quality Importance

No. of studies 1

Design

Risk of bias

Inconsistency

Indirectness

Preterm birth (< 37 weeks of gestation) randomized trials serious1 no serious inconsistency no serious indirectness seriousv 369 fewer per 1000 (from 165 fewer to 525 fewer) not pooled ⊕⊕OO LOW CRITICAL

Serious maternal outcome 1 randomized trials randomized trials serious1 no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious3 none 0/45 (0.0%) 8/151 (5.3%) 0/45 (0.0%) 5/159 (3.1%) not pooled ⊕OOO VERY LOW CRITICAL

Maternal adverse effects leading to discontinuation of treatment 4 serious4 very serious5 none RR 1.31 (0.01 to 221.68) 10 more per 1000 (from 31 fewer to 1000 more) 317 fewer per 1000 (from 531 fewer to 111 more) 14 more per 1000 (from 99 fewer to 204 more) MD 0.08 higher (4.08 lower to 4.24 higher) MD 0.78 lower (1.4 to 0.17 lower) ⊕OOO VERY LOW CRITICAL

Birth < 48 hours after trial entry 3 randomized trials serious4 serious6 no serious indirectness very serious7 none 36/91 (39.6%) 73/99 (73.7%) RR 0.57 (0.28 to 1.15) ⊕OOO VERY LOW CRITICAL

Birth < 24 hours after trial entry 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious7 none 22/76 (28.9%) 22/80 (27.5%) RR 1.05 (0.64 to 1.74) ⊕⊕OO LOW CRITICAL

Interval between trial entry and birth (days) (better indicated by higher values) 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious8 none 137 144 — ⊕⊕⊕O MODERATE CRITICAL

Gestational age at birth (better indicated by higher values) 3 randomized trials serious9 serious6 no serious indirectness serious2 none 137 144 — ⊕OOO VERY LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 2 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Magnesium sulfate 8/123 (6.5%) No tocolytic treatment 2/134 (1.5%) Relative (95% CI)

Effect Absolute Quality Importance

Total deaths (fetal, neonatal and infant) randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 RR 4.56 (1.00 to 20.86) 53 more per 1000 (from 0 more to 296 more) — ⊕⊕⊕O MODERATE CRITICAL

Fetal deaths 2 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency very serious11 no serious indirectness no serious indirectness very serious10 none 2/123 (1.6%) 7/137 (5.1%) 0/134 (0.0%) 6/153 (3.9%) RR 5.70 (0.28 to 116.87) RR 1.37 (0.48 to 3.97) ⊕⊕OO LOW CRITICAL

Neonatal/infant deaths 3 very serious10 none 15 more per 1000 (from 20 fewer to 116 more) 29 more per 1000 (from 15 fewer to 148 more) 51 more per 1000 (from 11 fewer to 124 more) — ⊕OOO VERY LOW CRITICAL

Serious infant outcome 3 randomized trials no serious risk of bias very serious11 no serious indirectness very serious10 none 9/139 (6.5%) 6/153 (3.9%) RR 1.74 (0.63 to 4.77) ⊕OOO VERY LOW CRITICAL

Respiratory distress syndrome 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 80/136 (58.8%) 86/153 (56.2%) RR 1.09 (0.98 to 1.22) ⊕⊕⊕O MODERATE CRITICAL

Proven neonatal infection (variously defined) 1 randomized trials randomized trials randomized trials no serious risk of bias serious1 no serious inconsistency no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness no serious indirectness very serious10 none 2/15 (13.3%) 0/45 (0.0%) 5/136 (3.7%) 0/19 (0.0%) 0/45 (0.0%) 7/153 (4.6%) RR 6.25 (0.32 to 121.14) not pooled ⊕⊕OO LOW CRITICAL

Severe intraventricular haemorrhage (IVH) (grade 3 or 4) or periventricular leukomalacia (PVL) 1 very serious3 none not pooled ⊕OOO VERY LOW CRITICAL

IVH (any) 3 no serious risk of bias very serious10 none RR 0.86 (0.28 to 2.62) 6 fewer per 1000 (from 33 fewer to 74 more) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 3 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Magnesium sulfate 4/136 (2.9%) No tocolytic treatment 4/153 (2.6%) Relative (95% CI) RR 1.19 (0.33 to 4.29)

Effect Absolute Quality Importance

Necrotizing enterocolitis randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious10 5 more per 1000 (from 18 fewer to 86 more) — ⊕⊕OO LOW CRITICAL

Respiratory arrest 1 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious10 none 1/76 (1.3%) 5/76 (6.6%) 0/80 (0.0%) 12/89 (13.5%) RR 3.16 (0.13 to 76.3) RR 0.49 (0.18 to 1.32) ⊕⊕OO LOW CRITICAL

Admission to neonatal intensive care unit 1 very serious10 none 69 fewer per 1000 (from 111 fewer to 43 more) 29 more per 1000 (from 66 fewer to 209 more) 12 more per 1000 (from 57 fewer to 130 more) — ⊕⊕OO LOW CRITICAL

Need for assisted ventilation 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious7 none 15/76 (19.7%) 15/89 (16.9%) RR 1.17 (0.61 to 2.24) ⊕⊕OO LOW CRITICAL

Caesarean section 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious7 none 22/136 (16.2%) 22/144 (15.3%) RR 1.08 (0.63 to 1.85) ⊕⊕OO LOW CRITICAL

Hypotension (variously defined) 1 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious10 none 1/76 (1.3%) 0/16 (0.0%) 0/80 (0.0%) 0/19 (0.0%) RR 3.16 (0.13 to 76.3) not pooled ⊕⊕OO LOW CRITICAL

Tachycardia (variously defined) 1 very serious3 none not pooled ⊕⊕OO LOW CRITICAL

1 One study with design limitations. 2 Estimate based on small sample size. 3 No events. 4 Two of the studies contributing data had design limitations. 5 Wide confidence interval crossing the line of no effect and few events. 6 Statistical heterogeneity (I2 > 60%). Variation in size of effect. 7 Wide confidence interval crossing the line of no effect and small sample size. 8 Wide confidence interval crossing the line of no effect. 9 More than 40% of weight from a study with design limitations. 10 Wide confidence interval crossing the line of no effect, few events and small sample size. 11 Statistical heterogeneity (I2 > 60%). Variation in size and direction of effect.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 2e.

Oxytocin receptor antagonists for inhibiting preterm labour Source: Flenady V, Reinebrant HE, Liley HG, Tambimuttu EG, Papatsonis DNM. Oxytocin receptor antagonists for inhibiting preterm labour. Cochrane Database Syst Rev. 2014;(6):CD004452. Quality assessment No. of patients Imprecision Other considerations Oxytocin receptor antagonist (atosiban) 12/246 (4.9%) Placebo Relative (95% CI) Effect Absolute Quality Importance

No. of studies

Design

Risk of bias

Inconsistency

Indirectness

Extremely preterm birth (< 28 weeks of gestation) 1 randomized trials serious1 no serious inconsistency no serious indirectness no serious imprecision none 4/255 (1.6%) RR 3.11 (1.02 to 9.51) 33 more per 1000 (from 0 more to 133 more) 85 more per 1000 (from 5 fewer to 186 more) not pooled ⊕⊕⊕O MODERATE CRITICAL

Preterm birth (< 37 weeks) 1 randomized trials serious1 no serious inconsistency no serious indirectness serious2 none 144/246 (58.5%) 128/255 (50.2%) RR 1.17 (0.99 to 1.37) ⊕⊕OO LOW CRITICAL

Maternal death 1 randomized trials randomized trials serious1 no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious3 none 0/246 (0.0%) 40/306 (13.1%) 0/255 (0.0%) 10/307 (3.3%) not pooled ⊕OOO VERY LOW CRITICAL

Maternal adverse drug reaction requiring cessation of treatment 2 serious1 no serious imprecision none RR 4.02 (2.05 to 7.85) 98 more per 1000 (from 34 more to 223 more) 3 more per 1000 (from 56 fewer to 423 more) MD 0.5 lower (1.56 lower to 0.56 higher) 21 more per 1000 (from 4 fewer to 92 more) ⊕⊕⊕O MODERATE CRITICAL

Birth within 48 hours of initiation of treatment 2 randomized trials serious4 no serious inconsistency no serious indirectness very serious5 none 6/76 (7.9%) 5/76 (6.6%) RR 1.05 (0.15 to 7.43) ⊕OOO VERY LOW CRITICAL

Gestational age (weeks) (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious6 none 57 57 — ⊕⊕⊕O MODERATE CRITICAL

Perinatal mortality 1 randomized trials serious1 no serious inconsistency no serious indirectness very serious7 none 11/288 (3.8%) 5/295 (1.7%) RR 2.25 (0.79 to 6.40) ⊕OOO VERY LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations

No. of patients Oxytocin receptor antagonist (atosiban) 5/365 (1.4%) Placebo Relative (95% CI)

Effect Absolute Quality Importance

Stillbirth 3 randomized trials serious4 no serious inconsistency no serious indirectness very serious7 none 8/372 (2.2%) RR 0.63 (0.22 to 1.84) 8 fewer per 1000 (from 17 fewer to 18 more) 21 more per 1000 (from 1 fewer to 123 more) 35 more per 1000 (from 3 more to 178 more) 43 more per 1000 (from 11 fewer to 118 more) 12 fewer per 1000 (from 42 fewer to 48 more) 14 fewer per 1000 (from 17 fewer to 13 more) MD 138.31 lower (248.76 to 27.86 lower) ⊕OOO VERY LOW CRITICAL

Neonatal death 1 randomized trials serious1 no serious inconsistency no serious indirectness very serious7 none 8/288 (2.8%) 2/295 (0.7%) RR 4.10 (0.88 to 19.13) ⊕OOO VERY LOW CRITICAL

Infant death (up to 12 months) 1 randomized trials serious1 no serious inconsistency no serious indirectness no serious imprecision none 12/288 (4.2%) 2/295 (0.7%) RR 6.15 (1.39 to 27.22) ⊕⊕⊕O MODERATE CRITICAL

Respiratory distress syndrome 2 randomized trials serious4 no serious inconsistency no serious indirectness serious2 none 67/340 (19.7%) 54/349 (15.5%) RR 1.28 (0.93 to 1.76) ⊕⊕OO LOW CRITICAL

Intraventricular haemorrhage 1 randomized trials serious1 no serious inconsistency no serious indirectness serious2 none 16/243 (6.6%) 19/246 (7.7%) RR 0.85 (0.45 to 1.62) ⊕⊕OO LOW CRITICAL

Necrotizing enterocolitis 1 randomized trials serious1 no serious inconsistency no serious indirectness very serious7 none 1/283 (0.4%) 5/292 (1.7%) RR 0.21 (0.02 to 1.76) ⊕OOO VERY LOW CRITICAL

Birth weight (g) (better indicated by higher values) 2 randomized trials serious4 no serious inconsistency no serious indirectness no serious imprecision none 343 349 — ⊕⊕⊕O MODERATE CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations

No. of patients Oxytocin receptor antagonist (atosiban) 115/274 (42.0%) Placebo Relative (95% CI)

Effect Absolute Quality Importance

Admission to neonatal intensive care 1 randomized trials serious1 no serious inconsistency no serious indirectness serious2 none 110/286 (38.5%) RR 1.09 (0.89 to 1.34) 35 more per 1000 (from 42 fewer to 131 more) 89 more per 1000 (from 39 fewer to 373 more) 56 more per 1000 (from 2 more to 138 more) ⊕⊕OO LOW CRITICAL

Caesarean section 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 none 13/56 (23.2%) 8/56 (14.3%) RR 1.62 (0.73 to 3.61) ⊕⊕OO LOW CRITICAL

Maternal drug reaction 2 randomized trials serious4 no serious inconsistency no serious indirectness no serious imprecision none 49/306 (16.0%) 32/307 (10.4%) RR 1.54 (1.02 to 2.32) ⊕⊕⊕O MODERATE CRITICAL

1 One study with design limitations. 2 Wide confidence interval crossing the line of no effect. 3 No events. 4 Most studies contributing data had design limitations. 5 Wide confidence interval crossing the line of no effect, few events and small sample size. 6 Estimate based on small sample size. 7 Wide confidence interval crossing the line of no effect and few events.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 2f.

Nitric oxide donors for inhibiting preterm labour Source: Duckitt K, Thornton S, O’Donovan OP, Dowswell T. Nitric oxide donors for treating preterm labour. Cochrane Database Syst Rev. 2014;(5):CD002860. Quality assessment No. of patients Imprecision Other considerations none Nitric oxide donors 8/74 (10.8%) Placebo or no treatment 16/79 (20.3%) Relative (95% CI) RR 0.53 (0.24 to 1.17) Effect Absolute Quality Importance

No. of studies 1

Design

Risk of bias

Inconsistency

Indirectness

Delivery < 28 completed weeks randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 95 fewer per 1000 (from 154 fewer to 34 more) 27 fewer per 1000 (from 148 fewer to 156 more) 181 fewer per 1000 (from 355 fewer to 426 more) 128 more per 1000 (from 175 fewer to 606 more) 8 fewer per 1000 (from 13 fewer to 96 more) 18 fewer per 1000 (from 30 fewer to 60 more) MD 327 higher (272.13 lower to 926.13 higher) ⊕⊕OO LOW CRITICAL

Delivery < 34 completed weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 26/74 (35.1%) 30/79 (38%) RR 0.93 (0.61 to 1.41) ⊕⊕OO LOW CRITICAL

Delivery < 37 completed weeks 2 randomized trials no serious risk of bias2 serious3 no serious indirectness serious4 none 44/149 (29.5%) 65/154 (42.2%) RR 0.57 (0.16 to 2.01) ⊕⊕OO LOW CRITICAL

Prolongation of pregnancy > 48 hours 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious5 none 67/91 (73.6%) 64/95 (67.4%) RR 1.19 (0.74 to 1.90) ⊕⊕⊕O MODERATE CRITICAL

Death in utero unrelated to congenital abnormalities 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 0/74 (0.0%) 1/79 (1.3%) RR 0.36 (0.01 to 8.59) ⊕⊕OO LOW CRITICAL

Death in first 28 days of life unrelated to congenital abnormalities 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 1/91 (1.1%) 3/95 (3.2%) RR 0.43 (0.06 to 2.89) ⊕⊕OO LOW CRITICAL

Birth weight (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 none 17 16 — ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Nitric oxide donors 3/17 (17.6%) Placebo or no treatment 6/16 (37.5%) Relative (95% CI)

Effect Absolute Quality Importance

Respiratory distress syndrome randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 RR 0.47 (0.14 to 1.57) 199 fewer per 1000 (from 322 fewer to 214 more) 14 more per 1000 (from 10 fewer to 279 more) 75 fewer per 1000 (from 87 fewer to 19 more) 222 more per 1000 (from 63 more to 425 more) 277 more per 1000 (from 90 more to 553 more) 46 more per 1000 (from 30 fewer to 249 more) 16 fewer per 1000 (from 94 fewer to 146 more) ⊕⊕OO LOW CRITICAL

Intraventricular haemorrhage 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 2/74 (2.7%) 1/79 (1.3%) RR 2.14 (0.20 to 23.06) ⊕⊕OO LOW CRITICAL

Chronic lung disease 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 1/74 (1.4%) 7/79 (8.9%) RR 0.15 (0.02 to 1.21) ⊕⊕OO LOW CRITICAL

Adverse drug reactions 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious6 none 61/91 (67.0%) 43/95 (45.3%) RR 1.49 (1.14 to 1.94) ⊕⊕⊕O MODERATE CRITICAL

Headache 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious6 none 42/74 (56.8%) 23/79 (29.1%) RR 1.95 (1.31 to 2.90) ⊕⊕⊕O MODERATE CRITICAL

Dizziness 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 9/74 (12.2%) 6/79 (7.6%) RR 1.60 (0.60 to 4.28) ⊕⊕OO LOW CRITICAL

Flushing 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 11/74 (14.9%) 13/79 (16.5%) RR 0.90 (0.43 to 1.89) ⊕⊕OO LOW IMPORTANT

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Nitric oxide donors 9/74 (12.2%) Placebo or no treatment 8/79 (10.1%) Relative (95% CI)

Effect Absolute Quality Importance

Hypotension randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 RR 1.20 (0.49 to 2.95) 20 more per 1000 (from 52 fewer to 197 more) 31 more per 1000 (from 78 fewer to 156 more) ⊕⊕OO LOW CRITICAL

Completion of course of maternal steroids 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 74/91 (81.3%) 74/95 (77.9%) RR 1.04 (0.90 to 1.20) ⊕⊕OO LOW IMPORTANT

1 Wide confidence interval crossing the line of no effect, few events and small sample size. 2 Most of the pooled effect provided by studies with low risk of bias. 3 Statistical heterogeneity (I2 = 90%). 4 Wide confidence interval crossing the line of no effect. 5 Wide confidence interval crossing the line of no effect and small sample size. 6 Estimate based on small sample size.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 2g.

Progestational agents for inhibiting preterm labour Source: Su LL, Samuel M, Chong YS. Progestational agents for treating threatened or established preterm labour. Cochrane Database Syst Rev. 2014;(1):CD006770. Quality assessment No. of patients Imprecision Other considerations none Progestational agents 8/31 (25.8%) Placebo Relative (95% CI) RR 0.62 (0.30 to 1.27) Effect Absolute Quality Importance

No. of studies 1

Design

Risk of bias

Inconsistency

Indirectness

Preterm birth < 34 weeks of gestation randomized trials serious1 no serious inconsistency no serious indirectness very serious2 13/31 (41.9%) 159 fewer per 1000 (from 294 fewer to 113 more) 133 fewer per 1000 (from 205 fewer to 117 more) 200 fewer per 1000 (from 11 fewer to 322 fewer) 64 fewer per 1000 (from 166 fewer to 134 more) 17 fewer per 1000 (from 25 fewer to 160 more) — ⊕OOO VERY LOW CRITICAL

Preterm delivery < 35 weeks 1 randomized trials serious1 no serious inconsistency no serious indirectness very serious2 none 3/30 (10%) 7/30 (23.3%) RR 0.43 (0.12 to 1.50) ⊕OOO VERY LOW CRITICAL

Preterm delivery < 37 weeks 4 randomized trials serious3 no serious inconsistency no serious indirectness no serious imprecision none 51/147 (34.7%) 77/146 (52.7%) RR 0.62 (0.39 to 0.98) ⊕⊕⊕O MODERATE CRITICAL

Delivery within 48 hours of intervention 1 randomized trials serious1 no serious inconsistency no serious indirectness very serious2 none 11/54 (20.4%) 15/56 (26.8%) RR 0.76 (0.38 to 1.50) ⊕OOO VERY LOW CRITICAL

Perinatal mortality 1 randomized trials serious1 no serious inconsistency no serious indirectness very serious2 none 0/43 (0.0%) 1/40 (2.5%) RR 0.31 (0.01 to 7.41) ⊕OOO VERY LOW CRITICAL

Intraventricular haemorrhage 1 randomized trials randomized trials no serious risk of bias serious1 serious1 no serious indirectness no serious indirectness very serious2 none 1/51 (2.0%) 1/51 (2.0%) 0/53 (0.0%) 1/53 (1.9%) RR 3.12 (0.13 to 74.76) RR 1.04 (0.07 to 16.18) ⊕OOO VERY LOW CRITICAL

Necrotizing enterocolitis 1 no serious inconsistency very serious2 none 1 more per 1000 (from 18 fewer to 286 more) 2 fewer per 1000 (from 24 fewer to 335 more) ⊕OOO VERY LOW CRITICAL

Respiratory distress syndrome 1 randomized trials serious1 no serious inconsistency no serious indirectness very serious2 none 1/43 (2.3%) 1/40 (2.5%) RR 0.93 (0.06 to 14.38) ⊕OOO VERY LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Progestational agents 19/51 (37.3%) Placebo Relative (95% CI) RR 1.01 (0.61 to 1.65)

Effect Absolute Quality Importance

Low birth weight (< 2.5 kg) randomized trials serious1 no serious inconsistency no serious indirectness very serious2 20/54 (37%) 4 more per 1000 (from 144 fewer to 241 more) MD 324.7 higher (155.05 to 494.34 higher) 14 more per 1000 (from 71 fewer to 167 more) 12 more per 1000 (from 38 fewer to 153 more) 35 fewer per 1000 (from 89 fewer to 148 more) 55 fewer per 1000 (from 87 fewer to 99 more) ⊕OOO VERY LOW CRITICAL

Birth weight (g) (better indicated by higher values) 2 randomized trials serious3 no serious inconsistency no serious indirectness serious4 none 73 70 — ⊕⊕OO LOW CRITICAL

Admission to neonatal intensive care unit 2 randomized trials serious3 no serious inconsistency no serious indirectness very serious5 none 17/94 (18.1%) 16/93 (17.2%) RR 1.08 (0.59 to 1.97) ⊕OOO VERY LOW CRITICAL

Mechanical ventilation 2 randomized trials serious3 no serious inconsistency no serious indirectness very serious2 none 7/94 (7.4%) 6/93 (6.5%) RR 1.18 (0.41 to 3.37) ⊕OOO VERY LOW CRITICAL

Oxygen requirement on day 7 of life 1 randomized trials serious1 no serious inconsistency no serious indirectness very serious2 none 4/51 (7.8%) 6/53 (11.3%) RR 0.69 (0.21 to 2.31) ⊕OOO VERY LOW CRITICAL

Oxygen requirement on day 28 of life 1 randomized trials serious1 no serious inconsistency no serious indirectness very serious2 none 2/51 (3.9%) 5/53 (9.4%) RR 0.42 (0.08 to 2.05) ⊕OOO VERY LOW CRITICAL

1 One study with design limitations. 2 Wide confidence interval crossing the line of no effect, few events and small sample size. 3 Most studies contributing data had design limitations. 4 Estimate based on small sample size. 5 Wide confidence interval crossing the line of no effect and small sample size.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 2h.

Relaxin for inhibiting preterm labour Source: Bain E, Heatley E, Hsu K, Crowther CA. Relaxin for preventing preterm birth. Cochrane Database Syst Rev. 2013;(8):CD010073. Quality assessment No. of patients Imprecision Other considerations none Relaxin No relaxin Relative (95% CI) RR 0.92 (0.81 to 1.05) Effect Absolute Quality Importance

No. of studies 1

Design

Risk of bias

Inconsistency

Indirectness

Preterm birth randomized trials very serious1 no serious inconsistency no serious indirectness very serious2 33/37 (89.2%) 31/32 (96.9%) 77 fewer per 1000 (from 184 fewer to 48 more) 467 fewer per 1000 (from 121 fewer to 663 fewer) 410 fewer per 1000 (from 690 fewer to 90 more) 0 fewer per 1000 (from 410 fewer to 690 more) 267 fewer per 1000 (from 560 fewer to 860 more) 890 fewer per 1000 (from 990 fewer to 780 more) 68 fewer per 1000 (from 272 fewer to 460 more) ⊕OOO VERY LOW CRITICAL

Birth within 7 days of treatment 1 randomized trials very serious1 no serious inconsistency no serious indirectness serious3 none 7/15 (46.7%) 14/15 (93.3%) RR 0.50 (0.29 to 0.87) ⊕OOO VERY LOW CRITICAL

Birth within 7 days of treatment (subgroups) — premature rupture of membranes 1 randomized trials very serious1 no serious inconsistency no serious indirectness very serious4 none 4/7 (57.1%) 11/11 (100.0%) RR 0.59 (0.31 to 1.09) ⊕OOO VERY LOW CRITICAL

Birth within 7 days of treatment (subgroups) — placental pathology 1 randomized trials very serious1 no serious inconsistency no serious indirectness very serious4 none 3/3 (100.0%) 3/3 (100.0%) RR 1.00 (0.59 to 1.69) ⊕OOO VERY LOW CRITICAL

Birth within 7 days of treatment (subgroups) — maternal complications 1 randomized trials very serious1 no serious inconsistency no serious indirectness very serious4 none 2/5 (40.0%) 2/3 (66.7%) RR 0.60 (0.16 to 2.29) ⊕OOO VERY LOW CRITICAL

Birth within 7 days of treatment (subgroups) — uncomplicated premature labour 1 randomized trials very serious1 no serious inconsistency no serious indirectness very serious4 none 0/5 (0.0%) 1/1 (100.0%) RR 0.11 (0.01 to 1.78) ⊕OOO VERY LOW CRITICAL

Perinatal mortality 1 randomized trials very serious1 no serious inconsistency no serious indirectness very serious4 none 5/15 (33.3%) 6/15 (40.0%) RR 0.83 (0.32 to 2.15) ⊕OOO VERY LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Relaxin No relaxin Relative (95% CI) RR 1.18 (0.37 to 3.76)

Effect Absolute Quality Importance

Perinatal mortality (subgroups) — premature rupture of membranes randomized trials very serious1 no serious inconsistency no serious indirectness very serious4 3/7 (42.9%) 4/11 (36.4%) 65 more per 1000 (from 229 fewer to 1000 more) 333 fewer per 1000 (from 613 fewer to 1000 more) 133 more per 1000 (from 347 fewer to 1000 more) 890 fewer per 1000 (from 990 fewer to 780 more) 67 fewer per 1000 (from 243 fewer to 470 more) 0 fewer per 1000 (from 62 fewer to 903 more) — ⊕OOO VERY LOW CRITICAL

Perinatal mortality (subgroups) — placental pathology 1 randomized trials very serious1 no serious inconsistency no serious indirectness very serious4 none 1/3 (33.3%) 2/3 (66.7%) RR 0.50 (0.08 to 2.99) ⊕OOO VERY LOW CRITICAL

Perinatal mortality (subgroups) — maternal complications 1 randomized trials very serious1 no serious inconsistency no serious indirectness very serious4 none 4/5 (80.0%) 2/3 (66.7%) RR 1.20 (0.48 to 2.99) ⊕OOO VERY LOW CRITICAL

Perinatal mortality (subgroups) — uncomplicated premature labour 1 randomized trials very serious1 no serious inconsistency no serious indirectness very serious4 none 0/5 (0.0%) 1/1 (100.0%) RR 0.11 (0.01 to 1.78) ⊕OOO VERY LOW CRITICAL

Neonatal death 1 randomized trials very serious1 no serious inconsistency no serious indirectness very serious4 none 4/15 (26.7%) 5/15 (33.3%) RR 0.80 (0.27 to 2.41) ⊕OOO VERY LOW CRITICAL

Fetal death 1 randomized trials very serious1 no serious inconsistency no serious indirectness very serious4 none 1/15 (6.7%) 1/15 (6.7%) RR 1.00 (0.07 to 14.55) ⊕OOO VERY LOW CRITICAL

Intrapartum fever 1 randomized trials randomized trials very serious1 no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious4 none 2/7 (28.6%) 17/25 (68.0%) 0/11 (0.0%) 18/25 (72.0%) RR 7.50 (0.41 to 136.52) RR 0.94 (0.66 to 1.36) ⊕OOO VERY LOW CRITICAL

Labour stopped 1 very serious1 very serious4 none 43 fewer per 1000 (from 245 fewer to 259 more) ⊕OOO VERY LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Relaxin No relaxin Relative (95% CI) RR 2.00 (0.43 to 9.32)

Effect Absolute Quality Importance

Birth weight < 2500 g randomized trials very serious1 no serious inconsistency no serious indirectness very serious4 4/15 (26.7%) 2/15 (13.3%) 133 more per 1000 (from 76 fewer to 1000 more) ⊕OOO VERY LOW CRITICAL

1 One study with serious design limitations. 2 Wide confidence interval crossing the line of no effect and small sample size. 3 Few events and small sample size. 4 Wide confidence interval crossing the line of no effect, few events and small sample size.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 2i.

Hydration for inhibiting preterm labour Source: Stan CM, Boulvain M, Pfister R, Hirsbrunner-Almagbaly P. Hydration for treatment of preterm labour. Cochrane Database Syst Rev. 2013;(11):CD003096. Quality assessment No. of patients Imprecision Other considerations none Hydration (all women) 9/73 (12.3%) No treatment/ bed rest alone 6/37 (16.2%) Relative (95% CI) RR 0.76 (0.29 to 1.97) Effect Absolute Quality Importance

No. of studies 1

Design

Risk of bias serious1

Inconsistency

Indirectness

Delivery < 32 weeks of gestation randomized trials no serious inconsistency no serious indirectness very serious2 39 fewer per 1000 (from 115 fewer to 157 more) 25 fewer per 1000 (from 71 fewer to 139 more) 23 more per 1000 (from 75 fewer to 175 more) MD 0.99 lower (7.85 lower to 5.87 higher) 2 fewer per 1000 (from 96 fewer to 207 more) 53 fewer per 1000 (from 134 fewer to 63 more) MD 39 higher (26.11 lower to 104.11 higher) ⊕OOO VERY LOW CRITICAL

Delivery < 34 weeks 1 randomized trials serious1 no serious inconsistency no serious indirectness very serious2 none 4/62 (6.5%) 5/56 (8.9%) RR 0.72 (0.20 to 2.56) ⊕OOO VERY LOW CRITICAL

Delivery < 37 weeks 2 randomized trials serious1 no serious inconsistency no serious indirectness very serious3 none 38/135 (28.1%) 24/93 (25.8%) RR 1.09 (0.71 to 1.68) ⊕OOO VERY LOW CRITICAL

Time to delivery (days) (better indicated by lower values) 2 randomized trials serious4 no serious inconsistency no serious indirectness very serious3 none 135 93 — ⊕OOO VERY LOW CRITICAL

Admission to neonatal intensive care unit 1 randomized trials serious1 no serious inconsistency no serious indirectness very serious2 none 11/62 (17.7%) 10/56 (17.9%) RR 0.99 (0.46 to 2.16) ⊕OOO VERY LOW CRITICAL

Use of tocolytic drugs 2 randomized trials serious4 no serious inconsistency no serious indirectness very serious3 none 41/138 (29.7%) 30/96 (31.3%) RR 0.83 (0.57 to 1.20) ⊕OOO VERY LOW IMPORTANT

Cost of treatment (first 24 hours, in US$) (better indicated by lower values) 1 randomized trials serious1 no serious inconsistency no serious indirectness very serious3 none 54 49 — ⊕OOO VERY LOW IMPORTANT

1 One study with design limitations. 2 Wide confidence interval crossing the line of no effect, few events and small sample size. 3 Wide confidence interval crossing the line of no effect and small sample size. 4 All studies contributing data had design limitations.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 2j.

Maintenance betamimetic therapy for inhibiting preterm labour Source: Dodd JM, Crowther CA, Middleton P. Oral betamimetics for maintenance therapy after threatened preterm labour. Cochrane Database Syst Rev. 2012;(12):CD003927. (updated for this guideline) Quality assessment No. of patients Imprecision Other considerations none Any betamimetic 3/62 (4.8%) Placebo/no treatment 1/58 (1.7%) Relative (95% CI) RR 2.81 (0.30 to 26.22) Effect Absolute Quality Importance

No. of studies 1

Design

Risk of bias

Inconsistency

Indirectness

Very preterm birth (< 34 weeks of gestation) randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 31 more per 1000 (from 12 fewer to 435 more) 35 more per 1000 (from 29 fewer to 111 more) 43 fewer per 1000 (from 115 fewer to 342 more) 20 fewer per 1000 (from 63 fewer to 91 more) 64 fewer per 1000 (from 116 fewer to 25 more) — ⊕⊕OO LOW CRITICAL

Preterm birth (< 37 weeks) 6 randomized trials serious2 no serious inconsistency no serious indirectness serious3 none 111/336 (33.0%) 98/308 (31.8%) RR 1.11 (0.91 to 1.35) ⊕⊕OO LOW CRITICAL

Preterm birth within 24 hours of therapy 1 randomized trials serious4 no serious inconsistency no serious indirectness very serious1 none 2/23 (8.7%) 3/23 (13.0%) RR 0.67 (0.12 to 3.62) ⊕OOO VERY LOW CRITICAL

Preterm birth within 48 hours of therapy 1 randomized trials serious4 no serious inconsistency no serious indirectness very serious1 none 7/100 (7.0%) 9/100 (9.0%) RR 0.78 (0.30 to 2.01) ⊕OOO VERY LOW CRITICAL

Preterm birth within 1 week of therapy 2 randomized trials serious5 no serious inconsistency no serious indirectness serious3 none 19/150 (12.7%) 28/145 (19.3%) RR 0.67 (0.40 to 1.13) ⊕⊕OO LOW CRITICAL

Side-effects sufficient to stop therapy 2 randomized trials randomized trials no serious risk of bias6 serious2 no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious1 none 1/73 (1.4%) 11/349 (3.2%) 0/68 (0.0%) 4/332 (1.2%) RR 2.71 (0.11 to 64.79) RR 2.41 (0.86 to 6.74) ⊕⊕OO LOW CRITICAL

Perinatal mortality 6 very serious7 none 17 more per 1000 (from 2 fewer to 69 more) 5 more per 1000 (from 19 fewer to 49 more) ⊕OOO VERY LOW CRITICAL

Respiratory distress syndrome 6 randomized trials serious8 no serious inconsistency no serious indirectness serious3 none 20/388 (5.2%) 19/382 (5.0%) RR 1.10 (0.61 to 1.98) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 2 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Any betamimetic 3/212 (1.4%) Placebo/no treatment 3/204 (1.5%) Relative (95% CI) RR 0.98 (0.22 to 4.28)

Effect Absolute Quality Importance

Necrotizing enterocolitis randomized trials serious8 no serious inconsistency no serious indirectness very serious7 0 fewer per 1000 (from 11 fewer to 48 more) 1 fewer per 1000 (from 13 fewer to 45 more) 71 fewer per 1000 (from 82 fewer to 21 more) MD 4.13 higher (91.89 lower to 100.16 higher) 31 more per 1000 (from 36 fewer to 157 more) 172 more per 1000 (from 79 more to 301 more) 80 more per 1000 (from 6 more to 296 more) 121 more per 1000 (from 18 more to 297 more) ⊕OOO VERY LOW CRITICAL

Intraventricular haemorrhage 3 randomized trials serious8 no serious inconsistency no serious indirectness very serious7 none 4/237 (1.7%) 4/229 (1.7%) RR 0.97 (0.27 to 3.58) ⊕OOO VERY LOW CRITICAL

Low birth weight (< 2500 g) 1 randomized trials serious4 no serious inconsistency no serious indirectness very serious1 none 1/80 (1.3%) 5/60 (8.3%) RR 0.15 (0.02 to 1.25) ⊕OOO VERY LOW CRITICAL

Birth weight (better indicated by higher values) 7 randomized trials serious8 no serious inconsistency no serious indirectness serious3 none 395 385 — ⊕⊕OO LOW CRITICAL

Neonatal intensive care unit admission 2 randomized trials serious5 no serious inconsistency no serious indirectness very serious9 none 19/134 (14.2%) 14/126 (11.1%) RR 1.28 (0.68 to 2.41) ⊕OOO VERY LOW CRITICAL

Tachycardia 4 randomized trials serious2 no serious inconsistency no serious indirectness no serious imprecision none 68/210 (32.4%) 31/204 (15.2%) RR 2.13 (1.52 to 2.98) ⊕⊕⊕O MODERATE CRITICAL

Tachypnoea 2 randomized trials no serious risk of bias10 no serious inconsistency no serious indirectness serious11 none 15/134 (11.2%) 4/126 (3.2%) RR 3.52 (1.20 to 10.33) ⊕⊕⊕O MODERATE CRITICAL

Hypotension 2 randomized trials serious5 no serious inconsistency no serious indirectness serious11 none 21/85 (24.7%) 11/81 (13.6%) RR 1.89 (1.13 to 3.19) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Any betamimetic 12/72 (16.7%) Placebo/no treatment 2/68 (2.9%) Relative (95% CI) RR 5.67 (1.32 to 24.40)

Effect Absolute Quality Importance

Palpitations randomized trials serious4 no serious inconsistency no serious indirectness serious11 137 more per 1000 (from 9 more to 688 more) — ⊕⊕OO LOW CRITICAL

Headache 1 randomized trials randomized trials no serious risk of bias serious2 no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious1 none 1/50 (2.0%) 40/167 (24.0%) 0/45 (0.0%) 36/168 (21.4%) RR 2.71 (0.11 to 64.79) RR 1.11 (0.76 to 1.62) ⊕⊕OO LOW CRITICAL

Maternal antenatal readmission to hospital 4 serious3 none 24 more per 1000 (from 51 fewer to 133 more) 1 fewer per 1000 (from 16 fewer to 235 more) 161 more per 1000 (from 70 fewer to 694 more) ⊕⊕OO LOW CRITICAL

Need for mechanical ventilation 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 1/62 (1.6%) 1/58 (1.7%) RR 0.94 (0.06 to 14.61) ⊕⊕OO LOW CRITICAL

Neonatal jaundice 1 randomized trials serious4 no serious inconsistency no serious indirectness very serious9 none 10/25 (40.0%) 6/25 (24.0%) RR 1.67 (0.71 to 3.89) ⊕OOO VERY LOW CRITICAL

1 Wide confidence interval crossing the line of no effect, few events and small sample size. 2 Most studies contributing data had design limitations. 3 Wide confidence interval crossing the line of no effect. 4 One study with design limitations. 5 More than 40% of weight from a study with design limitations. 6 One study contributing data rated low risk of bias. 7 Wide confidence interval crossing the line of no effect and few events. 8 All studies contributing data had design limitations. 9 Wide confidence interval crossing the line of no effect and small sample size. 10 More than 50% of weight from studies at low risk of bias. 11 Estimate based on small sample size.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 2k.

Magnesium maintenance therapy inhibiting preterm labour Source: Han S, Crowther CA, Moore V. Magnesium maintenance therapy for preventing preterm birth after threatened preterm labour. Cochrane Database Syst Rev. 2013;(5):CD000940. Quality assessment No. of patients Imprecision Other considerations none Magnesium Placebo or no treatment 30/49 (61.2%) Relative (95% CI) RR 1.05 (0.80 to 1.40) Effect Absolute Quality Importance

No. of studies 2

Design

Risk of bias serious1

Inconsistency

Indirectness

Birth < 37 weeks of gestation randomized trials no serious inconsistency no serious indirectness very serious2 31/50 (62.0%) 31 more per 1000 (from 122 fewer to 245 more) MD 0.55 lower (1.34 lower to 0.25 higher) — ⊕OOO VERY LOW CRITICAL

Gestational age at delivery (weeks) (better indicated by higher values) 2 randomized trials serious1 no serious inconsistency no serious indirectness very serious2 none 90 93 — ⊕OOO VERY LOW CRITICAL

Perinatal mortality (death before discharge among live-born infants) 1 randomized trials randomized trials randomized trials randomized trials serious3 no serious inconsistency no serious inconsistency no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness no serious indirectness no serious indirectness very serious4 none 2/25 (8.0%) 1/25 (4.0%) 1/25 (4.0%) 87 0/25 (0.0%) 0/25 (0.0%) 0/25 (0.0%) 93 RR 5 (0.25 to 99.16) RR 3.00 (0.13 to 70.30) RR 3.00 (0.13 to 70.3) — ⊕OOO VERY LOW CRITICAL

Respiratory distress syndrome 1 serious3 very serious4 none — ⊕OOO VERY LOW CRITICAL

Periventricular haemorrhage 1 serious3 very serious4 none — ⊕OOO VERY LOW CRITICAL

Neonatal length of stay (days) (better indicated by lower values) 2 serious1 very serious2 none MD 1.18 higher (0.46 lower to 2.82 higher) 84 more per 1000 (from 35 fewer to 329 more) 118 fewer per 1000 (from 308 fewer to 213 more) ⊕OOO VERY LOW CRITICAL

Neonatal intensive care unit admissions 1 randomized trials serious3 no serious inconsistency no serious indirectness very serious2 none 15/65 (23.1%) 10/68 (14.7%) RR 1.57 (0.76 to 3.24) ⊕OOO VERY LOW CRITICAL

Maternal readmission for threatened preterm labour 1 randomized trials serious3 no serious inconsistency no serious indirectness very serious2 none 11/25 (44.0%) 14/25 (56.0%) RR 0.79 (0.45 to 1.38) ⊕OOO VERY LOW CRITICAL

1 Both studies contributing data had design limitations. 2 Wide confidence interval crossing the line of no effect and small sample size. 3 One study with design limitations. 4 Wide confidence interval crossing the line of no effect, few events and small sample size.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 2l.

Maintenance therapy with calcium channel blockers for inhibiting preterm labour Source: Naik Gaunekar N, Raman P, Bain E, Crowther CA. Maintenance therapy with calcium channel blockers for preventing preterm birth after threatened preterm labour. Cochrane Database Syst Rev. 2013;(10):CD004071. Quality assessment No. of patients Imprecision Other considerations none Calcium channel blockers 3/29 (10.3%) Control Relative (95% CI) RR 3.21 (0.35 to 29.11) Effect Absolute Quality Importance

No. of studies 1

Design

Risk of bias

Inconsistency

Indirectness

Birth < 28 weeks of gestation randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 1/31 (3.2%) 71 more per 1000 (from 21 fewer to 907 more) 28 more per 1000 (from 49 fewer to 122 more) 19 fewer per 1000 (from 82 fewer to 57 more) MD 0.32 higher (0.61 lower to 1.25 higher) 25 fewer per 1000 (from 42 fewer to 91 more) 7 more per 1000 (from 64 fewer to 198 more) MD 5.35 higher (0.49 to 10.21 higher) not pooled ⊕⊕OO LOW CRITICAL

Birth < 34 weeks 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 116/267 (43.4%) 111/273 (40.7%) RR 1.07 (0.88 to 1.30) ⊕⊕⊕O MODERATE CRITICAL

Birth < 37 weeks 5 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 208/337 (61.7%) 218/344 (63.4%) RR 0.97 (0.87 to 1.09) ⊕⊕⊕⊕ HIGH CRITICAL

Gestation at birth (better indicated by higher values) 5 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 337 344 — ⊕⊕⊕O MODERATE CRITICAL

Birth within 48 hours of treatment 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 1/62 (1.6%) 3/66 (4.5%) RR 0.46 (0.07 to 3.00) ⊕⊕OO LOW CRITICAL

Birth within 7 days of treatment 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 7/62 (11.3%) 7/66 (10.6%) RR 1.07 (0.40 to 2.87) ⊕⊕OO LOW CRITICAL

Pregnancy prolongation (days) (better indicated by higher values) 4 randomized trials serious3 no serious inconsistency no serious indirectness no serious imprecision none 136 139 — ⊕⊕⊕O MODERATE CRITICAL

Maternal death 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 0/230 (0.0%) 0/236 (0.0%) not pooled ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Calcium channel blockers 13/201 (6.5%) Control Relative (95% CI) RR 0.88 (0.43 to 1.81)

Effect Absolute Quality Importance

Maternal intrauterine infection randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 15/205 (7.3%) 9 fewer per 1000 (from 42 fewer to 59 more) 0 more per 1000 (from 4 fewer to 64 more) not pooled ⊕⊕OO LOW CRITICAL

Maternal admission to intensive care unit 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 none 1/230 (0.4%) 1/236 (0.4%) RR 1.02 (0.06 to 16.19) ⊕⊕OO LOW CRITICAL

Maternal adverse drug reaction causing treatment cessation 1 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious4 none 0/33 (0.0%) 6/230 (2.6%) 0/35 (0.0%) 4/236 (1.7%) not pooled ⊕⊕OO LOW CRITICAL

Perinatal mortality 2 very serious5 none RR 1.48 (0.45 to 4.86) 8 more per 1000 (from 9 fewer to 65 more) not pooled ⊕⊕OO LOW CRITICAL

Stillbirth 1 randomized trials randomized trials no serious risk of bias serious3 no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious4 none 0/29 (0.0%) 1/66 (1.5%) 0/31 (0.0%) 2/67 (3%) not pooled ⊕⊕OO LOW CRITICAL

Neonatal death 2 very serious1 none RR 0.75 (0.05 to 11.76) 7 fewer per 1000 (from 28 fewer to 321 more) 5 more per 1000 (from 47 fewer to 83 more) 3 fewer per 1000 (from 38 fewer to 62 more) 13 fewer per 1000 (from 44 fewer to 41 more) ⊕OOO VERY LOW CRITICAL

Composite neonatal morbidity 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 40/249 (16.1%) 38/248 (15.3%) RR 1.03 (0.69 to 1.54) ⊕⊕⊕O MODERATE CRITICAL

Neonatal sepsis 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 18/238 (7.6%) 19/241 (7.9%) RR 0.96 (0.52 to 1.79) ⊕⊕⊕O MODERATE CRITICAL

Respiratory distress syndrome 3 randomized trials serious3 no serious inconsistency no serious indirectness serious2 none 19/275 (6.9%) 23/279 (8.2%) RR 0.84 (0.47 to 1.50) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 3 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Calcium channel blockers 3/275 (1.1%) Control Relative (95% CI) RR 0.41 (0.12 to 1.42)

Effect Absolute Quality Importance

Intraventricular haemorrhage randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 8/278 (2.9%) 17 fewer per 1000 (from 25 fewer to 12 more) 24 fewer per 1000 (from 39 fewer to 73 more) 14 fewer per 1000 (from 22 fewer to 26 more) 10 more per 1000 (from 7 fewer to 63 more) 27 more per 1000 (from 39 fewer to 403 more) 58 fewer per 1000 (from 221 fewer to 174 more) 21 more per 1000 (from 46 fewer to 99 more) MD 0.14 lower (3.25 lower to 2.96 higher) MD 14 higher (4.21 to 23.79 higher) ⊕⊕OO LOW CRITICAL

Intraventricular haemorrhage — any 2 randomized trials serious3 no serious inconsistency no serious indirectness very serious1 none 1/74 (1.4%) 3/73 (4.1%) RR 0.42 (0.06 to 2.78) ⊕OOO VERY LOW CRITICAL

Intraventricular haemorrhage — grade 3 or 4 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 none 2/201 (1.0%) 5/205 (2.4%) RR 0.41 (0.08 to 2.08) ⊕⊕OO LOW CRITICAL

Necrotizing enterocolitis 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 none 7/275 (2.5%) 4/278 (1.4%) RR 1.68 (0.53 to 5.35) ⊕⊕OO LOW CRITICAL

Small for gestational age 1 randomized trials serious6 no serious inconsistency no serious indirectness very serious1 none 3/37 (8.1%) 2/37 (5.4%) RR 1.50 (0.27 to 8.46) ⊕OOO VERY LOW CRITICAL

Low birth weight 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious7 none 25/48 (52.1%) 25/43 (58.1%) RR 0.90 (0.62 to 1.3) ⊕⊕OO LOW CRITICAL

Neonatal intensive care unit admission 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 130/348 (37.4%) 128/361 (35.5%) RR 1.06 (0.87 to 1.28) ⊕⊕⊕O MODERATE CRITICAL

Length of neonatal intensive care unit stay (days) (better indicated by lower values) 3 randomized trials serious3 no serious inconsistency no serious indirectness very serious7 none 70 62 — ⊕OOO VERY LOW CRITICAL

Length of neonatal hospital stay (days) (better indicated by lower values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious8 none 29 31 — ⊕⊕⊕O MODERATE CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 2 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Calcium channel blockers 5/238 (2.1%) Control Relative (95% CI) RR 0.74 (0.25 to 2.20)

Effect Absolute Quality Importance

Chronic lung disease randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 7/241 (2.9%) 8 fewer per 1000 (from 22 fewer to 35 more) not pooled ⊕⊕OO LOW CRITICAL

Periventricular leukomalacia 2 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious4 none 0/238 (0.0%) 38/282 (13.5%) 0/241 (0.0%) 37/294 (12.6%) not pooled ⊕⊕OO LOW CRITICAL

Mechanical ventilation 2 serious2 none RR 1.07 (0.70 to 1.64) 9 more per 1000 (from 38 fewer to 81 more) 0 fewer per 1000 (from 185 fewer to 288 more) ⊕⊕⊕O MODERATE CRITICAL

Neonatal jaundice 1 randomized trials serious6 no serious inconsistency no serious indirectness very serious7 none 19/37 (51.4%) 19/37 (51.4%) RR 1.00 (0.64 to 1.56) ⊕OOO VERY LOW CRITICAL

1 Wide confidence interval crossing the line of not effect, few events and small sample size. 2 Wide confidence interval crossing the line of no effect. 3 Most studies contributing data had design limitations. 4 No events. 5 Wide confidence interval crossing the line of no effect and few events. 6 One study with design limitations. 7 Wide confidence interval crossing the line of no effect and small sample size. 8 Estimate based on small sample size.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 2m. Maintenance therapy with oxytocin antagonists for inhibiting preterm labour Source: Papatsonis DN, Flenady V, Liley HG. Maintenance therapy with oxytocin antagonists for inhibiting preterm birth after threatened preterm labour. Cochrane Database Syst Rev. 2013;(10):CD005938. Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none No. of patients Atosiban Placebo Relative (95% CI) RR 0.75 (0.28 to 2.01) Effect Absolute Quality Importance

Birth < 28 weeks of gestation randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 7/45 (15.6%) 6/29 (20.7%) 52 fewer per 1000 (from 149 fewer to 209 more) 21 fewer per 1000 (from 75 fewer to 78 more) 42 fewer per 1000 (from 110 fewer to 45 more) not pooled ⊕⊕OO LOW CRITICAL

Birth < 32 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 19/158 (12.0%) 18/127 (14.2%) RR 0.85 (0.47 to 1.55) ⊕⊕OO LOW CRITICAL

Birth < 37 weeks 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 90/267 (33.7%) 92/243 (37.9%) RR 0.89 (0.71 to 1.12) ⊕⊕⊕O MODERATE CRITICAL

Maternal death 1 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious4 none 0/261 (0.0%) 4/261 (1.5%) 0/251 (0.0%) 5/251 (2.0%) not pooled ⊕⊕OO LOW CRITICAL

Perinatal death 1 very serious5 none RR 0.77 (0.21 to 2.83) 5 fewer per 1000 (from 16 fewer to 36 more) — ⊕⊕OO LOW CRITICAL

Fetal death 1 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious5 none 1/261 (0.4%) 3/261 (1.1%) 0/251 (0.0%) 5/251 (2.0%) RR 2.89 (0.12 to 70.50) RR 0.58 (0.14 to 2.39) ⊕⊕OO LOW CRITICAL

Neonatal death 1 very serious5 none 8 fewer per 1000 (from 17 fewer to 28 more) 5 fewer per 1000 (from 15 fewer to 32 more) ⊕⊕OO LOW CRITICAL

Infant death (up to 12 months) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 none 4/289 (1.4%) 5/269 (1.9%) RR 0.74 (0.20 to 2.74) ⊕⊕OO LOW CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Atosiban Placebo Relative (95% CI) RR 1.06 (0.66 to 1.70)

Effect Absolute Quality Importance

Respiratory distress syndrome randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 33/288 (11.5%) 29/269 (10.8%) 6 more per 1000 (from 37 fewer to 75 more) 10 more per 1000 (from 4 fewer to 81 more) MD 0.10 higher (131.78 lower to 131.98 higher) 41 fewer per 1000 (from 97 fewer to 36 more) ⊕⊕⊕O MODERATE CRITICAL

Necrotizing enterocolitis 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious5 none 5/288 (1.7%) 2/269 (0.7%) RR 2.34 (0.46 to 11.93) ⊕⊕OO LOW CRITICAL

Birth weight (g) (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 289 269 — ⊕⊕⊕O MODERATE CRITICAL

Neonatal intensive care unit admission 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 61/284 (21.5%) 68/266 (25.6%) RR 0.84 (0.62 to 1.14) ⊕⊕⊕O MODERATE CRITICAL

1 Wide confidence interval crossing the line of no effect, few events and small sample size. 2 Wide confidence interval crossing the line of no effect and small sample size. 3 Wide confidence interval crossing the line of no effect. 4 No events. 5 Wide confidence interval crossing the line of no effect and few events.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 3a.

Magnesium sulfate for fetal neuroprotection in women at risk of preterm birth (all women and babies) Source: Doyle LW, Crowther CA, Middleton P, Marret S, Rouse D. Magnesium sulphate for women at risk of preterm birth for neuroprotection of the fetus. Cochrane Database Syst Rev. 2009;(1):CD004661. (updated for this guideline) Quality assessment No. of patients Imprecision Other considerations none Magnesium sulfate 10/2682 (0.4%) 1/2682 (0.0%) 1/2682 (0.0%) 28/1300 (2.2%) No active treatment 8/2729 (0.3%) 3/2729 (0.1%) 1/2729 (0.0%) 32/1306 (2.5%) Relative (95% CI) RR 1.25 (0.51 to 3.07) RR 0.34 (0.04 to 3.26) RR 1.02 (0.06 to 16.25) RR 0.89 (0.54 to 1.47) Effect Absolute Quality Importance

No. of studies 4

Design

Risk of bias

Inconsistency

Indirectness

Maternal mortality randomized trials randomized trials randomized trials randomized trials no serious risk of bias no serious risk of bias no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness no serious indirectness no serious indirectness very serious1 1 more per 1000 (from 1 fewer to 6 more) 1 fewer per 1000 (from 1 fewer to 2 more) 0 more per 1000 (from 0 fewer to 6 more) 3 fewer per 1000 (from 11 fewer to 12 more) 55 more per 1000 (from 36 more to 81 more) 3 more per 1000 (from 14 fewer to 21 more) 2 fewer per 1000 (from 10 fewer to 9 more) 3 more per 1000 (from 14 fewer to 25 more) 3 more per 1000 (from 16 fewer to 26 more) ⊕⊕OO LOW CRITICAL

Maternal cardiac arrest 4 very serious1 none ⊕⊕OO LOW CRITICAL

Maternal respiratory arrest 4 very serious1 none ⊕⊕OO LOW CRITICAL

Mother admitted to intensive care unit 2 serious2 none ⊕⊕⊕O MODERATE CRITICAL

Cessation of maternal therapy 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 192/2396 (8.0%) 60/2451 (2.4%) RR 3.26 (2.46 to 4.31) ⊕⊕⊕⊕ HIGH CRITICAL

Paediatric mortality (fetal mortality and mortality occurring later) 5 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness no serious imprecision no serious imprecision none 435/2997 (14.5%) 128/2997 (4.3%) 430/3042 (14.1%) 133/3042 (4.4%) RR 1.02 (0.90 to 1.15) RR 0.96 (0.77 to 1.21) ⊕⊕⊕⊕ HIGH CRITICAL

Fetal death 5 none ⊕⊕⊕⊕ HIGH CRITICAL

Deaths among live-borns (during primary hospitalization) 4 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness serious2 none 267/2967 (9.0%) 307/2997 (10.2%) 258/3013 (8.6%) 297/3042 (9.8%) RR 1.04 (0.84 to 1.29) RR 1.03 (0.84 to 1.27) ⊕⊕⊕O MODERATE CRITICAL

Deaths among live-borns (to latest age of follow-up) 5 serious2 none ⊕⊕⊕O MODERATE CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 5 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Magnesium sulfate 539/2997 (18.0%) No active treatment 580/3042 (19.1%) Relative (95% CI) RR 0.92 (0.78 to 1.09)

Effect Absolute Quality Importance

Death or cerebral palsy randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision 15 fewer per 1000 (from 42 fewer to 17 more) 0 fewer per 1000 (from 31 fewer to 38 more) 14 fewer per 1000 (from 43 fewer to 21 more) 5 more per 1000 (from 27 fewer to 41 more) 9 fewer per 1000 (from 30 fewer to 17 more) 8 fewer per 1000 (from 18 fewer to 6 more) 3 fewer per 1000 (from 11 fewer to 9 more) 4 more per 1000 (from 12 fewer to 26 more) 1 more per 1000 (from 20 fewer to 27 more) ⊕⊕⊕⊕ HIGH CRITICAL

Death or any neurological impairment 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 499/1427 (35.0%) 495/1421 (34.8%) RR 1.00 (0.91 to 1.11) ⊕⊕⊕⊕ HIGH CRITICAL

Death or substantial gross motor dysfunction 4 randomized trials no serious risk of bias serious3 no serious indirectness no serious imprecision none 490/2967 (16.5%) 523/3013 (17.4%) RR 0.92 (0.75 to 1.12) ⊕⊕⊕O MODERATE CRITICAL

Death or major neurological disability 2 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness no serious imprecision no serious imprecision none 394/1427 (27.6%) 454/2169 (20.9%) 386/1421 (27.2%) 482/2218 (21.7%) RR 1.02 (0.90 to 1.15) RR 0.96 (0.86 to 1.08) ⊕⊕⊕⊕ HIGH CRITICAL

Intraventricular haemorrhage 3 none ⊕⊕⊕⊕ HIGH CRITICAL

Intraventricular haemorrhage — grade 3 or 4 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 72/1817 (4.0%) 88/1882 (4.7%) RR 0.83 (0.62 to 1.13) ⊕⊕⊕O MODERATE CRITICAL

Periventricular leukomalacia 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 70/2169 (3.2%) 76/2218 (3.4%) RR 0.92 (0.67 to 1.26) ⊕⊕⊕O MODERATE CRITICAL

Major neurological disability 2 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness serious2 none 98/1427 (6.9%) 203/1427 (14.2%) 91/1421 (6.4%) 200/1421 (14.1%) RR 1.07 (0.82 to 1.40) RR 1.01 (0.86 to 1.19) ⊕⊕⊕O MODERATE CRITICAL

Any neurological impairment 2 no serious imprecision none ⊕⊕⊕⊕ HIGH CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 4 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Magnesium sulfate 57/2967 (1.9%) No active treatment 94/3013 (3.1%) Relative (95% CI) RR 0.61 (0.44 to 0.85)

Effect Absolute Quality Importance

Substantial gross motor dysfunction randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision 12 fewer per 1000 (from 5 fewer to 17 fewer) 1 fewer per 1000 (from 2 fewer to 5 more) 1 fewer per 1000 (from 5 fewer to 11 more) 2 fewer per 1000 (from 20 fewer to 20 more) 15 fewer per 1000 (from 5 fewer to 22 fewer) 29 more per 1000 (from 25 fewer to 91 more) 24 more per 1000 (from 10 fewer to 65 more) 6 fewer per 1000 (from 14 fewer to 4 more) 1 more per 1000 (from 16 fewer to 25 more) ⊕⊕⊕⊕ HIGH CRITICAL

Blindness 3 randomized trials randomized trials randomized trials no serious risk of bias no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness no serious indirectness serious2 none 3/1779 (0.2%) 9/1779 (0.5%) 647/2967 (21.8%) 4/1757 (0.2%) 12/1757 (0.7%) 670/3013 (22.2%) RR 0.74 (0.17 to 3.3) RR 0.79 (0.24 to 2.56) RR 0.99 (0.91 to 1.09) ⊕⊕⊕O MODERATE CRITICAL

Deafness 3 serious2 none ⊕⊕⊕O MODERATE CRITICAL

Developmental delay or intellectual impairment 4 no serious imprecision none ⊕⊕⊕⊕ HIGH CRITICAL

Cerebral palsy 5 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 104/2997 (3.5%) 151/3042 (5.0%) RR 0.70 (0.55 to 0.89) ⊕⊕⊕⊕ HIGH CRITICAL

Chronic lung disease (infant requires oxygen at age 28 days) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 280/629 (44.5%) 260/626 (41.5%) RR 1.07 (0.94 to 1.22) ⊕⊕⊕⊕ HIGH CRITICAL

Chronic lung disease (infant requires oxygen at 36 weeks of age) 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 220/981 (22.4%) 195/962 (20.3%) RR 1.12 (0.95 to 1.32) ⊕⊕⊕O MODERATE CRITICAL

Neonatal convulsions 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 55/2169 (2.5%) 70/2218 (3.2%) RR 0.80 (0.56 to 1.13) ⊕⊕⊕O MODERATE CRITICAL

Neonatal hypotonia 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 85/1188 (7.2%) 88/1256 (7.0%) RR 1.02 (0.77 to 1.36) ⊕⊕⊕O MODERATE CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 2 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Magnesium sulfate 1418 No active treatment 1410 Relative (95% CI) —

Effect Absolute Quality Importance

Duration of primary hospital stay for newborns (days) (better indicated by lower values) randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness serious2 MD 0.52 lower (4.15 lower to 3.11 higher) 29 fewer per 1000 (from 53 fewer to 0 more) 6 more per 1000 (from 3 fewer to 20 more) 33 more per 1000 (from 6 more to 70 more) 36 more per 1000 (from 2 more to 88 more) ⊕⊕⊕O MODERATE CRITICAL

Ongoing respiratory support 3 no serious imprecision none 980/2169 (45.2%) 1069/2218 (48.2%) RR 0.94 (0.89 to 1.00) ⊕⊕⊕⊕ HIGH CRITICAL

Maternal respiratory depression 2 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness serious2 none 41/1631 (2.5%) 80/821 (9.7%) 31/1672 (1.9%) 52/805 (6.5%) RR 1.31 (0.83 to 2.07) RR 1.51 (1.09 to 2.09) ⊕⊕⊕O MODERATE CRITICAL

Maternal hypotension 2 no serious imprecision none ⊕⊕⊕⊕ HIGH CRITICAL

Maternal tachycardia 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 56/535 (10.5%) 36/527 (6.8%) RR 1.53 (1.03 to 2.29) ⊕⊕⊕⊕ HIGH CRITICAL

1 Wide confidence interval crossing the line of no effect and few events. 2 Wide confidence interval crossing the line of no effect. 3 Statistical heterogeneity (I2 > 60%).

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 3b.

Magnesium sulfate for fetal neuroprotection in women at risk of preterm birth (singleton and multiple pregnancy subgroups) Source: Doyle LW, Crowther CA, Middleton P, Marret S, Rouse D. Magnesium sulphate for women at risk of preterm birth for neuroprotection of the fetus. Cochrane Database Syst Rev. 2009;(1):CD004661. (updated for this guideline) Quality assessment No. of patients Imprecision Other considerations none Magnesium sulfate 395/2468 (16.0%) No active treatment 388/2516 (15.4%) Relative (95% CI) RR 1.04 (0.85 to 1.26) Effect Absolute Quality Importance

No. of studies 3

Design

Risk of bias

Inconsistency

Indirectness

Paediatric mortality (fetal mortality and mortality occurring later) — both singleton and multiple pregnancies randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 6 more per 1000 (from 23 fewer to 40 more) 2 more per 1000 (from 23 fewer to 31 more) 36 more per 1000 (from 52 fewer to 193 more) 7 fewer per 1000 (from 58 fewer to 58 more) 7 fewer per 1000 (from 44 fewer to 34 more) 31 more per 1000 (from 123 fewer to 431 more) 0 fewer per 1000 (from 49 fewer to 56 more) 0 fewer per 1000 (from 34 fewer to 41 more) ⊕⊕⊕O MODERATE CRITICAL

Paediatric mortality (fetal mortality and mortality occurring later) — singleton pregnancy subgroup 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 329/2113 (15.6%) 327/2143 (15.3%) RR 1.01 (0.85 to 1.20) ⊕⊕⊕⊕ HIGH CRITICAL

Paediatric mortality (fetal mortality and mortality occurring later) — multiple pregnancy subgroup 3 randomized trials no serious risk of bias serious2 no serious indirectness serious1 none 66/355 (18.6%) 61/373 (16.4%) RR 1.22 (0.68 to 2.18) ⊕⊕OO LOW CRITICAL

Death or cerebral palsy — both singleton and multiple pregnancies 2 randomized trials no serious risk of bias serious2 no serious indirectness no serious imprecision none 334/1427 (23.4%) 344/1421 (24.2%) RR 0.97 (0.76 to 1.24) ⊕⊕⊕O MODERATE CRITICAL

Death or cerebral palsy — singleton pregnancy subgroup 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 277/1163 (23.8%) 285/1158 (24.6%) RR 0.97 (0.82 to 1.14) ⊕⊕⊕⊕ HIGH CRITICAL

Death or cerebral palsy — multiple pregnancy subgroup 2 randomized trials no serious risk of bias serious2 no serious indirectness serious1 none 57/264 (21.6%) 59/263 (22.4%) RR 1.14 (0.45 to 2.92) ⊕⊕OO LOW CRITICAL

Death or neurological impairment — both singleton and multiple pregnancies 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 499/1427 (35.0%) 495/1421 (34.8%) RR 1 .00 (0.86 to 1.16) ⊕⊕⊕⊕ HIGH CRITICAL

Death or neurological impairment — singleton pregnancy subgroup 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 405/1163 (34.8%) 399/1158 (34.5%) RR 1 .00 (0.90 to 1.12) ⊕⊕⊕⊕ HIGH CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 2 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Magnesium sulfate 94/264 (35.6%) No active treatment 96/263 (36.5%) Relative (95% CI) RR 1.21 (0.56 to 2.65)

Effect Absolute Quality Importance

Death or neurological impairment — multiple pregnancy subgroup randomized trials no serious risk of bias serious2 no serious indirectness serious1 77 more per 1000 (from 161 fewer to 602 more) 5 more per 1000 (from 41 fewer to 60 more) 6 more per 1000 (from 30 fewer to 44 more) 51 more per 1000 (from 120 fewer to 436 more) 7 fewer per 1000 (from 16 fewer to 7 more) 2 fewer per 1000 (from 12 fewer to 14 more) 26 fewer per 1000 (from 42 fewer to 13 more) 1 more per 1000 (from 21 fewer to 27 more) ⊕⊕OO LOW CRITICAL

Death or major neurological disability — both singleton and multiple pregnancies 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 394/1427 (27.6%) 386/1421 (27.2%) RR 1.02 (0.85 to 1.22) ⊕⊕⊕⊕ HIGH CRITICAL

Death or major neurological disability — singleton pregnancy 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 326/1163 (28.0%) 319/1158 (27.5%) RR 1.02 (0.89 to 1.16) ⊕⊕⊕⊕ HIGH CRITICAL

Death or major neurological disability — multiple pregnancy subgroup 2 randomized trials no serious risk of bias serious2 no serious indirectness serious1 none 68/264 (25.8%) 67/263 (25.5%) RR 1.20 (0.53 to 2.71) ⊕⊕OO LOW CRITICAL

Cerebral palsy — both singleton and multiple pregnancies 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 38/1427 (2.7%) 47/1421 (3.3%) RR 0.80 (0.53 to 1.22) ⊕⊕⊕O MODERATE CRITICAL

Cerebral palsy — singleton pregnancy subgroup 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 31/1163 (2.7%) 33/1158 (2.8%) RR 0.92 (0.57 to 1.49) ⊕⊕⊕O MODERATE CRITICAL

Cerebral palsy — multiple pregnancy subgroup 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious3 none 7/264 (2.7%) 14/263 (5.3%) RR 0.52 (0.21 to 1.25) ⊕⊕OO LOW CRITICAL

Neurological impairment — both singleton and multiple pregnancies 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 203/1427 (14.2%) 200/1421 (14.1%) RR 1.01 (0.85 to 1.19) ⊕⊕⊕⊕ HIGH CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 2 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Magnesium sulfate 159/1163 (13.7%) No active treatment 147/1158 (12.7%) Relative (95% CI) RR 1.06 (0.88 to 1.28)

Effect Absolute Quality Importance

Neurological impairment — singleton pregnancy subgroup randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 8 more per 1000 (from 15 fewer to 36 more) 28 fewer per 1000 (from 79 fewer to 42 more) 4 more per 1000 (from 12 fewer to 26 more) 10 more per 1000 (from 8 fewer to 34 more) 21 fewer per 1000 (from 51 fewer to 34 more) ⊕⊕⊕O MODERATE CRITICAL

Neurological impairment — multiple pregnancy subgroup 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 44/264 (16.7%) 53/263 (20.2%) RR 0.86 (0.61 to 1.21) ⊕⊕⊕O MODERATE CRITICAL

Major neurological disability — both singleton and multiple pregnancies 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 98/1427 (6.9%) 91/1421 (6.4%) RR 1.07 (0.82 to 1.40) ⊕⊕⊕O MODERATE CRITICAL

Major neurological disability — singleton pregnancy subgroup 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 80/1163 (6.9%) 67/1158 (5.8%) RR 1.17 (0.87 to 1.59) ⊕⊕⊕O MODERATE CRITICAL

Major neurological disability — multiple pregnancy subgroup 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 18/264 (6.8%) 24/263 (9.1%) RR 0.77 (0.44 to 1.37) ⊕⊕⊕O MODERATE CRITICAL

1 Wide confidence interval crossing the line of no effect. 2 Statistical heterogeneity (I2 > 60%). 3 Wide confidence interval crossing the line of no effect and few events.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 3c.

Magnesium sulfate for fetal neuroprotection in women at risk of preterm birth (gestational age at administration) Source: Doyle LW, Crowther CA, Middleton P, Marret S, Rouse D. Magnesium sulphate for women at risk of preterm birth for neuroprotection of the fetus. Cochrane Database Syst Rev. 2009;(1):CD004661. (updated for this guideline) Quality assessment No. of patients Imprecision Other considerations none Magnesium sulfate 391/2573 (15.2%) No active treatment 399/2619 (15.2%) Relative (95% CI) RR 0.98 (0.84 to 1.14) Effect Absolute Quality Importance

No. of studies 4

Design

Risk of bias

Inconsistency

Indirectness

Paediatric mortality (fetal mortality and mortality occurring later) — < 34 weeks of gestation at randomization randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision 3 fewer per 1000 (from 24 fewer to 21 more) 8 fewer per 1000 (from 84 fewer to 105 more) 19 fewer per 1000 (from 42 fewer to 6 more) 9 fewer per 1000 (from 96 fewer to 118 more) 9 fewer per 1000 (from 49 fewer to 36 more) 15 more per 1000 (from 70 fewer to 121 more) 3 fewer per 1000 (from 41 fewer to 37 more) 14 more per 1000 (from 50 fewer to 87 more) ⊕⊕⊕⊕ HIGH CRITICAL

Paediatric mortality (fetal mortality and mortality occurring later) — < 30 weeks at randomization 2 randomized trials no serious risk of bias serious1 no serious indirectness serious2 none 187/769 (24.3%) 196/768 (25.5%) RR 0.97 (0.67 to 1.41) ⊕⊕OO LOW CRITICAL

Death or cerebral palsy — < 34 weeks at randomization 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 492/2573 (19.1%) 547/2619 (20.9%) RR 0.91 (0.80 to 1.03) ⊕⊕⊕⊕ HIGH CRITICAL

Death or cerebral palsy — < 30 weeks at randomization 2 randomized trials no serious risk of bias serious1 no serious indirectness serious2 none 224/769 (29.1%) 239/768 (31.1%) RR 0.97 (0.69 to 1.38) ⊕⊕OO LOW CRITICAL

Death or neurological impairment — < 34 weeks at randomization 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 452/1033 (43.8%) 459/1027 (44.7%) RR 0.98 (0.89 to 1.08) ⊕⊕⊕⊕ HIGH CRITICAL

Death or neurological impairment — < 30 weeks at randomization 2 randomized trials no serious risk of bias serious1 no serious indirectness no serious imprecision none 383/769 (49.8%) 386/768 (50.3%) RR 1.03 (0.86 to 1.24) ⊕⊕⊕O MODERATE CRITICAL

Death or major neurological disability — < 34 weeks at randomization 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 347/1033 (33.6%) 350/1027 (34.1%) RR 0.99 (0.88 to 1.11) ⊕⊕⊕⊕ HIGH CRITICAL

Death or major neurological disability — < 30 weeks at randomization 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 278/769 (36.2%) 277/768 (36.1%) RR 1.04 (0.86 to 1.24) ⊕⊕⊕⊕ HIGH CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 4 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Magnesium sulfate 101/2573 (3.9%) No active treatment 149/2619 (5.7%) Relative (95% CI) RR 0.69 (0.54 to 0.88)

Effect Absolute Quality Importance

Cerebral palsy — < 34 weeks at randomization randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision 18 fewer per 1000 (from 7 fewer to 26 fewer) 8 fewer per 1000 (from 25 fewer to 17 more) 4 more per 1000 (from 26 fewer to 38 more) 7 more per 1000 (from 32 fewer to 52 more) 7 more per 1000 (from 14 fewer to 36 more) 13 more per 1000 (from 16 fewer to 51 more) ⊕⊕⊕⊕ HIGH CRITICAL

Cerebral palsy — < 30 weeks at randomization 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 37/769 (4.8%) 43/768 (5.6%) RR 0.86 (0.56 to 1.31) ⊕⊕⊕O MODERATE CRITICAL

Neurological impairment — < 34 weeks at randomization 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 198/1033 (19.2%) 194/1027 (18.9%) RR 1.02 (0.86 to 1.20) ⊕⊕⊕⊕ HIGH CRITICAL

Neurological impairment — < 30 weeks at randomization 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 196/769 (25.5%) 190/768 (24.7%) RR 1.03 (0.87 to 1.21) ⊕⊕⊕⊕ HIGH CRITICAL

Major neurological disability — < 34 weeks at randomization 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 93/1033 (9.0%) 85/1027 (8.3%) RR 1.09 (0.83 to 1.43) ⊕⊕⊕O MODERATE CRITICAL

Major neurological disability — < 30 weeks at randomization 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 91/769 (11.8%) 81/768 (10.5%) RR 1.12 (0.85 to 1.48) ⊕⊕⊕O MODERATE CRITICAL

1 Statistical heterogeneity (I2 > 60%). 2 Wide confidence interval crossing the line of no effect.

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 3d.

Magnesium sulfate for fetal neuroprotection in women at risk of preterm birth (intention to prevent preterm-birth related neurological complications Source: Doyle LW, Crowther CA, Middleton P, Marret S, Rouse D. Magnesium sulphate for women at risk of preterm birth for neuroprotection of the fetus. Cochrane Database Syst Rev. 2009;(1):CD004661. (updated for this guideline) Quality assessment No. of patients Imprecision Other considerations none Magnesium sulfate 226/2199 (10.3%) No active treatment 242/2247 (10.8%) Relative (95% CI) RR 0.95 (0.80 to 1.12) Effect Absolute Quality Importance

No. of studies 4

Design

Risk of bias

Inconsistency

Indirectness

Paediatric mortality (fetal mortality and mortality occurring later) — neuroprotective intent randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision 5 fewer per 1000 (from 22 fewer to 13 more) 26 more per 1000 (from 17 fewer to 73 more) 2 fewer per 1000 (from 6 fewer to 5 more) 0 fewer per 1000 (from 31 fewer to 38 more) 4 fewer per 1000 (from 23 fewer to 18 more) 26 more per 1000 (from 4 fewer to 66 more) 3 fewer per 1000 (from 21 fewer to 20 more) ⊕⊕⊕⊕ HIGH CRITICAL

Paediatric mortality (fetal mortality and mortality occurring later) — other intent (maternal neuroprotective — pre-eclampsia) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 209/798 (26.2%) 188/795 (23.6%) RR 1.11 (0.93 to 1.31) ⊕⊕⊕O MODERATE CRITICAL

Fetal death — neuroprotective intent 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 17/2199 (0.8%) 22/2247 (1.0%) RR 0.78 (0.42 to 1.46) ⊕⊕⊕O MODERATE CRITICAL

Fetal death — other intent (maternal neuroprotective — pre-eclampsia) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 111/798 (13.9%) 111/795 (14.0%) RR 1.00 (0.78 to 1.27) ⊕⊕⊕O MODERATE CRITICAL

Deaths among live-borns — to latest age of follow-up — neuroprotective intent 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 209/2199 (9.5%) 220/2247 (9.8%) RR 0.96 (0.77 to 1.18) ⊕⊕⊕⊕ HIGH CRITICAL

Deaths among live-borns — to latest age of follow-up — other intent (maternal neuroprotective — pre-eclampsia) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 98/798 (12.3%) 77/795 (9.7%) RR 1.27 (0.96 to 1.68) ⊕⊕⊕O MODERATE CRITICAL

Deaths among live-borns — during primary hospitalization — neuroprotective intent 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 187/2169 (8.6%) 195/2218 (8.8%) RR 0.97 (0.76 to 1.23) ⊕⊕⊕O MODERATE CRITICAL

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Magnesium sulfate 80/798 (10.0%) No active treatment 63/795 (7.9%) Relative (95% CI) RR 1.27 (0.92 to 1.73)

Effect Absolute Quality Importance

Deaths among live-borns — during primary hospitalization — other intent (maternal neuroprotective — pre-eclampsia) randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 21 more per 1000 (from 6 fewer to 58 more) 26 fewer per 1000 (from 3 fewer to 45 fewer) 22 more per 1000 (from 19 fewer to 70 more) 23 fewer per 1000 (from 75 fewer to 33 more) 23 more per 1000 (from 20 fewer to 71 more) 24 fewer per 1000 (from 44 fewer to 0 more) 26 more per 1000 (from 14 fewer to 76 more) 15 fewer per 1000 (from 59 fewer to 38 more) 20 more per 1000 (from 20 fewer to 68 more) ⊕⊕⊕O MODERATE CRITICAL

Death or cerebral palsy — neuroprotective intent 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 328/2199 (14.9%) 387/2247 (17.2%) RR 0.85 (0.74 to 0.98) ⊕⊕⊕⊕ HIGH CRITICAL

Death or cerebral palsy — other intent (maternal neuroprotective — pre-eclampsia) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 211/798 (26.4%) 193/795 (24.3%) RR 1.09 (0.92 to 1.29) ⊕⊕⊕O MODERATE CRITICAL

Death or any neurological impairment — neuroprotective intent 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 280/629 (44.5%) 294/626 (47.0%) RR 0.95 (0.84 to 1.07) ⊕⊕⊕⊕ HIGH CRITICAL

Death or any neurological impairment — other intent (maternal neuroprotective — pre-eclampsia) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 219/798 (27.4%) 201/795 (25.3%) RR 1.09 (0.92 to 1.28) ⊕⊕⊕O MODERATE CRITICAL

Death or substantial gross motor dysfunction — neuroprotective intent 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 280/2169 (12.9%) 335/2218 (15.1%) RR 0.84 (0.71 to 1.00) ⊕⊕⊕⊕ HIGH CRITICAL

Death or substantial gross motor dysfunction — other intent (maternal neuroprotective — pre-eclampsia) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 210/798 (26.3%) 188/795 (23.6%) RR 1.11 (0.94 to 1.32) ⊕⊕⊕O MODERATE CRITICAL

Death or major neurological disability — neuroprotective intent 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 176/629 (28.0%) 185/626 (29.6%) RR 0.95 (0.80 to 1.13) ⊕⊕⊕⊕ HIGH CRITICAL

Death or major neurological disability — other intent (maternal neuroprotective — pre-eclampsia) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 218/798 (27.3%) 201/795 (25.3%) RR 1.08 (0.92 to 1.27) ⊕⊕⊕O MODERATE CRITICAL

100

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 3 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Magnesium sulfate 639/2169 (29.5%) No active treatment 660/2218 (29.8%) Relative (95% CI) RR 1.00 (0.91 to 1.09)

Effect Absolute Quality Importance

Developmental delay or intellectual impairment — neuroprotective intent randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision 0 fewer per 1000 (from 27 fewer to 27 more) 3 fewer per 1000 (from 9 fewer to 13 more) 17 more per 1000 (from 17 fewer to 64 more) 5 fewer per 1000 (from 11 fewer to 10 more) 9 fewer per 1000 (from 16 fewer to 1 more) 7 fewer per 1000 (from 14 fewer to 6 more) 12 fewer per 1000 (from 3 fewer to 18 fewer) 2 fewer per 1000 (from 9 fewer to 11 more) 4 fewer per 1000 (from 6 fewer to 7 more) ⊕⊕⊕⊕ HIGH CRITICAL

Developmental delay or intellectual impairment — other intent (maternal neuroprotective — pre-eclampsia) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 8/798 (1.0%) 10/795 (1.3%) RR 0.80 (0.32 to 2.01) ⊕⊕OO LOW CRITICAL

Major neurological disability — neuroprotective intent 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 89/629 (14.1%) 78/626 (12.5%) RR 1.14 (0.86 to 1.51) ⊕⊕⊕O MODERATE CRITICAL

Major neurological disability — other intent (maternal neuroprotective — pre-eclampsia) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 9/798 (1.1%) 13/795 (1.6%) RR 0.69 (0.30 to 1.6) ⊕⊕OO LOW CRITICAL

Cerebral palsy — neuroprotective intent: mild cerebral palsy 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 54/2169 (2.5%) 74/2218 (3.3%) RR 0.74 (0.52 to 1.04) ⊕⊕⊕O MODERATE CRITICAL

Cerebral palsy — neuroprotective intent: moderate cerebral palsy 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 14/981 (1.4%) 21/962 (2.2%) RR 0.66 (0.34 to 1.28) ⊕⊕⊕O MODERATE CRITICAL

Cerebral palsy — neuroprotective intent: moderate/severe cerebral palsy 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 45/2169 (2.1%) 72/2218 (3.2%) RR 0.64 (0.44 to 0.92) ⊕⊕⊕⊕ HIGH CRITICAL

Cerebral palsy — neuroprotective intent: severe cerebral palsy 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 11/981 (1.1%) 13/962 (1.4%) RR 0.82 (0.37 to 1.82) ⊕⊕OO LOW CRITICAL

Cerebral palsy — other intent (maternal neuroprotective — pre-eclampsia) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 2/798 (0.3%) 5/795 (0.6%) RR 0.40 (0.08 to 2.05) ⊕⊕OO LOW CRITICAL

101

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Magnesium sulfate 193/629 (30.7%) No active treatment 187/626 (29.9%) Relative (95% CI) RR 1.03 (0.87 to 1.21)

Effect Absolute Quality Importance

Any neurological impairment — neuroprotective intent randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision 9 more per 1000 (from 39 fewer to 63 more) 4 fewer per 1000 (from 11 fewer to 12 more) 17 fewer per 1000 (from 7 fewer to 24 fewer) — ⊕⊕⊕⊕ HIGH CRITICAL

Any neurological impairment — other intent (maternal neuroprotective — pre-eclampsia) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 10/798 (1.3%) 13/795 (1.6%) RR 0.77 (0.34 to 1.74) ⊕⊕OO LOW CRITICAL

Substantial gross motor dysfunction — neuroprotective intent 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 56/2169 (2.6%) 94/2218 (4.2%) RR 0.60 (0.43 to 0.83) ⊕⊕⊕⊕ HIGH CRITICAL

Substantial gross motor dysfunction — other intent (maternal neuroprotective — pre-eclampsia) 1 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious2 none 1/798 (0.1%) 8/981 (0.8%) 0/795 (0.0%) 11/962 (1.1%) RR 2.99 (0.12 to 73.26) RR 0.51 (0.05 to 4.96) ⊕⊕OO LOW CRITICAL

Deafness — neuroprotective intent 2 very serious2 none 6 fewer per 1000 (from 11 fewer to 45 more) 0 fewer per 1000 (from 1 fewer to 19 more) 0 fewer per 1000 (from 2 fewer to 12 more) 1 fewer per 1000 (from 2 fewer to 11 more) ⊕⊕OO LOW CRITICAL

Deafness — other intent (maternal neuroprotective — pre-eclampsia) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 1/798 (0.1%) 1/795 (0.1%) RR 1.00 (0.06 to 15.90) ⊕⊕OO LOW CRITICAL

Blindness — neuroprotective intent 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 2/981 (0.2%) 2/962 (0.2%) RR 0.97 (0.14 to 6.9) ⊕⊕OO LOW CRITICAL

Blindness — other intent (maternal neuroprotective — pre-eclampsia) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 1/798 (0.1%) 2/795 (0.3%) RR 0.50 (0.05 to 5.48) ⊕⊕OO LOW CRITICAL

1 Wide confidence interval crossing the line of no effect. 2 Wide confidence interval crossing the line of no effect and few events.

102

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 3e.

Magnesium sulfate for fetal neuroprotection in women at risk of preterm birth (retreatment) Source: Doyle LW, Crowther CA, Middleton P, Marret S, Rouse D. Magnesium sulphate for women at risk of preterm birth for neuroprotection of the fetus. Cochrane Database Syst Rev. 2009;(1):CD004661. (updated for this guideline) Quality assessment No. of patients Imprecision Other considerations none Retreatment No retreatment 429/3013 (14.2%) Relative (95% CI) RR 1.00 (0.84 to 1.18) Effect Absolute Quality Importance

No. of studies 4

Design

Risk of bias

Inconsistency

Indirectness

Paediatric mortality (fetal mortality and mortality occurring later) randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision 433/2967 (14.6%) 0 fewer per 1000 (from 23 fewer to 26 more) 10 more per 1000 (from 10 fewer to 37 more) 9 fewer per 1000 (from 47 fewer to 36 more) 15 fewer per 1000 (from 40 fewer to 12 more) 14 fewer per 1000 (from 37 fewer to 14 more) 21 fewer per 1000 (from 61 fewer to 30 more) 0 fewer per 1000 (from 31 fewer to 38 more) 0 fewer per 1000 (from 31 fewer to 38 more) ⊕⊕⊕⊕ HIGH CRITICAL

Paediatric mortality (fetal mortality and mortality occurring later) — retreatment permitted 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 103/1188 (8.7%) 96/1256 (7.6%) RR 1.13 (0.87 to 1.48) ⊕⊕⊕O MODERATE CRITICAL

Paediatric mortality (fetal mortality and mortality occurring later) — retreatment not permitted 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 330/1779 (18.5%) 333/1757 (19.0%) RR 0.95 (0.75 to 1.19) ⊕⊕⊕⊕ HIGH CRITICAL

Death or cerebral palsy 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 534/2967 (18.0%) 579/3013 (19.2%) RR 0.92 (0.79 to 1.06) ⊕⊕⊕⊕ HIGH CRITICAL

Death or cerebral palsy — retreatment permitted 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 144/1188 (12.1%) 170/1256 (13.5%) RR 0.90 (0.73 to 1.10) ⊕⊕⊕O MODERATE CRITICAL

Death or cerebral palsy — retreatment not permitted 3 randomized trials no serious risk of bias serious2 no serious indirectness serious1 none 390/1779 (21.9%) 409/1757 (23.3%) RR 0.91 (0.74 to 1.13) ⊕⊕OO LOW CRITICAL

Death or neurological impairment 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 499/1427 (35.0%) 495/1421 (34.8%) RR 1.00 (0.91 to 1.11) ⊕⊕⊕⊕ HIGH CRITICAL

Death or neurological impairment — retreatment not permitted 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 499/1427 (35.0%) 495/1421 (34.8%) RR 1.00 (0.91 to 1.11) ⊕⊕⊕⊕ HIGH CRITICAL

103

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 2 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Retreatment No retreatment 386/1421 (27.2%) Relative (95% CI) RR 1.02 (0.90 to 1.15)

Effect Absolute Quality Importance

Death or major neurological disability randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision 394/1427 (27.6%) 5 more per 1000 (from 27 fewer to 41 more) 5 more per 1000 (from 27 fewer to 41 more) 4 more per 1000 (from 12 fewer to 26 more) 4 more per 1000 (from 12 fewer to 26 more) 1 more per 1000 (from 20 fewer to 27 more) 1 more per 1000 (from 20 fewer to 27 more) 16 fewer per 1000 (from 7 fewer to 24 fewer) 24 fewer per 1000 (from 9 fewer to 35 fewer) ⊕⊕⊕⊕ HIGH CRITICAL

Death or major neurological disability — retreatment not permitted 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 394/1427 (27.6%) 386/1421 (27.2%) RR 1.02 (0.90 to 1.15) ⊕⊕⊕⊕ HIGH CRITICAL

Major neurological disability 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 98/1427 (6.9%) 91/1421 (6.4%) RR 1.07 (0.82 to 1.40) ⊕⊕⊕O MODERATE CRITICAL

Major neurological disability — retreatment not permitted 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 98/1427 (6.9%) 91/1421 (6.4%) RR 1.07 (0.82 to 1.40) ⊕⊕⊕O MODERATE CRITICAL

Neurologic impairment 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 203/1427 (14.2%) 200/1421 (14.1%) RR 1.01 (0.86 to 1.19) ⊕⊕⊕⊕ HIGH CRITICAL

Neurologic impairment — retreatment not permitted 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 203/1427 (14.2%) 200/1421 (14.1%) RR 1.01 (0.86 to 1.19) ⊕⊕⊕⊕ HIGH CRITICAL

Cerebral palsy 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 101/2967 (3.4%) 151/3013 (5.0%) RR 0.68 (0.53 to 0.87) ⊕⊕⊕⊕ HIGH CRITICAL

Cerebral palsy — retreatment permitted 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 41/1188 (3.5%) 74/1256 (5.9%) RR 0.59 (0.40 to 0.85) ⊕⊕⊕⊕ HIGH CRITICAL

104

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 3 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Retreatment No retreatment 77/1757 (4.4%) Relative (95% CI) RR 0.76 (0.55 to 1.06)

Effect Absolute Quality Importance

Cerebral palsy — retreatment not permitted randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 60/1779 (3.4%) 11 fewer per 1000 (from 20 fewer to 3 more) ⊕⊕⊕O MODERATE CRITICAL

1 Wide confidence interval crossing the line of no effect. 2 Statistical heterogeneity (I2 > 60%).

105

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 4a. Antibiotic prophylaxis for women at risk of preterm birth and with intact membranes (any antibiotics) Source: Flenady V, Hawley G, Stock OM, Kenyon S, Badawi N. Prophylactic antibiotics for inhibiting preterm labour with intact membranes. Cochrane Database Syst Rev. 2013;(12):CD000246. Quality assessment No. of studies 10 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none No. of patients Any antibiotic 1973/5251 (37.6%) No antibiotics 871/2136 (40.8%) Relative (95% CI) RR 0.98 (0.92 to 1.05) Effect Absolute Quality Importance

Birth < 36 or < 37 weeks of gestation randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision 8 fewer per 1000 (from 33 fewer to 20 more) 4 more per 1000 (from 11 fewer to 23 more) 3 fewer per 1000 (from 23 fewer to 17 more) MD 5.59 higher (0.31 to 10.87 higher) MD 0.53 higher (0 to 1.06 higher) 29 fewer per 1000 (from 16 fewer to 41 fewer) 42 more per 1000 (from 10 fewer to 117 more) 5 more per 1000 (from 2 fewer to 14 more) ⊕⊕⊕⊕ HIGH CRITICAL

Birth within 48 hours of randomization 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 509/4959 (10.3%) 183/1841 (9.9%) RR 1.04 (0.89 to 1.23) ⊕⊕⊕⊕ HIGH CRITICAL

Birth within 7 days of randomization 8 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 817/5091 (16.0%) 342/1962 (17.4%) RR 0.98 (0.87 to 1.10) ⊕⊕⊕⊕ HIGH CRITICAL

Interval between randomization and birth (days) (better indicated by higher values) 6 randomized trials randomized trials randomized trials no serious risk of bias no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness no serious indirectness serious1 none 1773 726 — ⊕⊕⊕O MODERATE CRITICAL

Gestational age at birth (better indicated by higher values) 10 serious1 none 495 491 — ⊕⊕⊕O MODERATE CRITICAL

Maternal infection 10 no serious imprecision none 458/5246 (8.7%) 236/2125 (11.1%) RR 0.74 (0.63 to 0.86) ⊕⊕⊕⊕ HIGH CRITICAL

Maternal adverse drug reaction requiring cessation of treatment 5 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 54/313 (17.3%) 41/313 (13.1%) RR 1.32 (0.92 to 1.89) ⊕⊕⊕O MODERATE CRITICAL

Perinatal mortality 10 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 141/5213 (2.7%) 43/2091 (2.1%) RR 1.22 (0.88 to 1.69) ⊕⊕⊕O MODERATE CRITICAL

106

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Any antibiotic 39/5105 (0.8%) No antibiotics 19/1975 (1.0%) Relative (95% CI) RR 0.73 (0.43 to 1.26)

Effect Absolute Quality Importance

Stillbirth 8 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 3 fewer per 1000 (from 5 fewer to 3 more) 7 more per 1000 (from 0 more to 16 more) 1 more per 1000 (from 7 fewer to 14 more) 1 fewer per 1000 (from 15 fewer to 15 more) 1 more per 1000 (from 5 fewer to 10 more) 5 fewer per 1000 (from 13 fewer to 6 more) 4 fewer per 1000 (from 8 fewer to 3 more) 2 more per 1000 (from 12 fewer to 19 more) ⊕⊕⊕O MODERATE CRITICAL

Neonatal death 9 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 101/5183 (1.9%) 24/2065 (1.2%) RR 1.57 (1.03 to 2.40) ⊕⊕⊕⊕ HIGH CRITICAL

Infant death (> 28 days) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 78/3508 (2.2%) 24/1146 (2.1%) RR 1.06 (0.68 to 1.67) ⊕⊕⊕O MODERATE CRITICAL

Respiratory distress syndrome 9 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 463/5159 (9%) 197/2041 (9.7%) RR 0.99 (0.84 to 1.16) ⊕⊕⊕⊕ HIGH CRITICAL

Necrotizing enterocolitis 6 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 62/5004 (1.2%) 25/1876 (1.3%) RR 1.06 (0.64 to 1.73) ⊕⊕⊕O MODERATE CRITICAL

Neonatal sepsis 10 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 127/5252 (2.4%) 76/2134 (3.6%) RR 0.86 (0.64 to 1.16) ⊕⊕⊕O MODERATE CRITICAL

Intraventricular haemorrhage 5 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 59/4968 (1.2%) 30/1845 (1.6%) RR 0.76 (0.48 to 1.19) ⊕⊕⊕O MODERATE CRITICAL

Neonatal mechanical ventilation 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 371/4685 (7.9%) 121/1556 (7.8%) RR 1.02 (0.84 to 1.24) ⊕⊕⊕⊕ HIGH CRITICAL

107

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 5 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Any antibiotic 1438/4882 (29.5%) No antibiotics 524/1746 (30.0%) Relative (95% CI) RR 0.97 (0.81 to 1.15)

Effect Absolute Quality Importance

Birth weight < 2500 g randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision 9 fewer per 1000 (from 57 fewer to 45 more) MD 58.38 higher (26.24 lower to 143 higher) 47 fewer per 1000 (from 99 fewer to 26 more) 12 more per 1000 (from 19 fewer to 47 more) 3 more per 1000 (from 4 fewer to 14 more) 13 more per 1000 (from 0 fewer to 36 more) 37 more per 1000 (from 4 fewer to 86 more) ⊕⊕⊕⊕ HIGH CRITICAL

Birth weight (better indicated by higher values) 12 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 5327 2204 — ⊕⊕⊕O MODERATE CRITICAL

Admission to neonatal intensive or special care nursery 5 randomized trials no serious risk of bias serious2 no serious indirectness serious1 none 1301/4992 (26.1%) 493/1883 (26.2%) RR 0.82 (0.62 to 1.10) ⊕⊕OO LOW CRITICAL

Moderate/severe functional impairment at 7 years of age 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 417/2317 (18.0%) 124/735 (16.9%) RR 1.07 (0.89 to 1.28) ⊕⊕⊕O MODERATE CRITICAL

Chronic neonatal lung disease 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 102/4685 (2.2%) 29/1556 (1.9%) RR 1.17 (0.78 to 1.76) ⊕⊕⊕O MODERATE CRITICAL

Cerebral palsy at 7 years of age 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious1 none 68/2403 (2.8%) 12/770 (1.6%) RR 1.82 (0.99 to 3.34) ⊕⊕⊕O MODERATE CRITICAL

Any functional impairment at 7 years of age 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 957/2317 (41.3%) 275/735 (37.4%) RR 1.10 (0.99 to 1.23) ⊕⊕⊕⊕ HIGH CRITICAL

1 Wide confidence interval crossing the line of no effect. 2 Statistical heterogeneity (I2 > 60%).

108

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 4b. Antibiotic prophylaxis for women at risk of preterm birth and with intact membranes (antibiotic regimen) Source: Flenady V, Hawley G, Stock OM, Kenyon S, Badawi N. Prophylactic antibiotics for inhibiting preterm labour with intact membranes. Cochrane Database Syst Rev. 2013;(12):CD000246. Quality assessment No. of studies 5 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none No. of patients A particular type of antibiotic 643/1721 (37.4%) Placebo or no treatment 288/709 (40.6%) Relative (95% CI) RR 0.99 (0.89 to 1.10) Effect Absolute Quality Importance

Birth < 36 or < 37 weeks of gestation — betalactam antibiotics alone randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision 4 fewer per 1000 (from 45 fewer to 41 more) 8 more per 1000 (from 35 fewer to 58 more) 4 fewer per 1000 (from 45 fewer to 41 more) 109 fewer per 1000 (from 302 fewer to 193 more) 1 more per 1000 (from 25 fewer to 35 more) 6 more per 1000 (from 22 fewer to 41 more) 12 more per 1000 (from 14 fewer to 44 more) ⊕⊕⊕⊕ HIGH CRITICAL

Birth < 36 or < 37 weeks — macrolide antibiotics alone 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 622/1658 (37.5%) 223/577 (38.6%) RR 1.02 (0.91 to 1.15) ⊕⊕⊕⊕ HIGH CRITICAL

Birth < 36 or < 37 weeks — macrolide and betalactam antibiotics 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 683/1813 (37.7%) 326/800 (40.8%) RR 0.99 (0.89 to 1.10) ⊕⊕⊕⊕ HIGH CRITICAL

Birth < 36 or < 37 weeks — antibiotics active against anaerobic bacteria 2 randomized trials no serious risk of bias serious1 no serious indirectness serious2 none 63/117 (53.8%) 70/109 (64.2%) RR 0.83 (0.53 to 1.30) ⊕⊕OO LOW CRITICAL

Birth within 48 hours of randomization — betalactam antibiotics alone 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 152/1534 (9.9%) 51/519 (9.8%) RR 1.01 (0.75 to 1.36) ⊕⊕⊕O MODERATE CRITICAL

Birth within 48 hours of randomization — macrolide antibiotics alone 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 166/1600 (10.4%) 51/519 (9.8%) RR 1.06 (0.78 to 1.42) ⊕⊕⊕O MODERATE CRITICAL

Birth within 48 hours of randomization — macrolide and betalactam antibiotics 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 192/1767 (10.9%) 74/753 (9.8%) RR 1.12 (0.86 to 1.45) ⊕⊕⊕O MODERATE CRITICAL

109

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients A particular type of antibiotic 5/58 (8.6%) Placebo or no treatment 8/51 (15.7%) Relative (95% CI)

Effect Absolute Quality Importance

Birth within 48 hours of randomization — antibiotics active against anaerobic bacteria randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious3 RR 0.55 (0.19 to 1.57) 71 fewer per 1000 (from 127 fewer to 89 more) MD 0.09 lower (2.96 lower to 2.78 higher) MD 4.26 higher (2.88 lower to 11.41 higher) MD 0.27 lower (2.95 lower to 2.41 higher) MD 10.5 higher (4.95 to 16.06 higher) 29 fewer per 1000 (from 3 fewer to 49 fewer) 20 fewer per 1000 (from 43 fewer to 9 more) 26 fewer per 1000 (from 3 fewer to 45 fewer) ⊕⊕OO LOW CRITICAL

Interval between randomization and birth (days) — betalactam antibiotics alone (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 1534 519 — ⊕⊕⊕O MODERATE CRITICAL

Interval between randomization and birth (days) — macrolide antibiotics alone (better indicated by higher values) 3 randomized trials no serious risk of bias very serious4 no serious indirectness serious2 none 1691 611 — ⊕OOO VERY LOW CRITICAL

Interval between randomization and birth (days) — macrolide and betalactam antibiotics (better indicated by higher values) 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 1629 592 — ⊕⊕⊕O MODERATE CRITICAL

Interval between randomization and birth (days) — antibiotics active against anaerobic bacteria (better indicated by higher values) 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 154 139 — ⊕⊕⊕⊕ HIGH CRITICAL

Maternal infection — betalactam antibiotics alone 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 146/1696 (8.6%) 76/689 (11.0%) RR 0.74 (0.56 to 0.97) ⊕⊕⊕⊕ HIGH CRITICAL

Maternal infection — macrolide antibiotics alone 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 157/1653 (9.5%) 64/569 (11.2%) RR 0.82 (0.62 to 1.08) ⊕⊕⊕O MODERATE CRITICAL

Maternal infection — macrolide and betalactam antibiotics 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 165/1790 (9.2%) 97/773 (12.5%) RR 0.79 (0.64 to 0.98) ⊕⊕⊕⊕ HIGH CRITICAL

110

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 3 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients A particular type of antibiotic 5/155 (3.2%) Placebo or no treatment 6/139 (4.3%) Relative (95% CI) RR 0.66 (0.11 to 3.92)

Effect Absolute Quality Importance

Maternal infection — antibiotics active against anaerobic bacteria randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious3 15 fewer per 1000 (from 38 fewer to 126 more) — ⊕⊕OO LOW CRITICAL

Maternal adverse drug reaction requiring cessation of treatment — betalactam antibiotics alone 1 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious5 none 1/40 (2.5%) 15/53 (28.3%) 0/42 (0.0%) 16/50 (32.0%) RR 3.15 (0.13 to 75.05) RR 0.88 (0.49 to 1.59) ⊕⊕OO LOW CRITICAL

Maternal adverse drug reaction requiring cessation of treatment — macrolide antibiotics alone 1 very serious5 none 38 fewer per 1000 (from 163 fewer to 189 more) 69 more per 1000 (from 10 fewer to 198 more) 7 more per 1000 (from 69 fewer to 140 more) 3 more per 1000 (from 8 fewer to 22 more) 4 more per 1000 (from 8 fewer to 25 more) 7 more per 1000 (from 4 fewer to 26 more) ⊕⊕OO LOW CRITICAL

Maternal adverse drug reaction requiring cessation of treatment — macrolide and betalactam antibiotics 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 34/161 (21.1%) 24/170 (14.1%) RR 1.49 (0.93 to 2.40) ⊕⊕⊕O MODERATE CRITICAL

Maternal adverse drug reaction requiring cessation of treatment — antibiotics active against anaerobic bacteria 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 19/112 (17.0%) 17/101 (16.8%) RR 1.04 (0.59 to 1.83) ⊕⊕⊕O MODERATE CRITICAL

Perinatal mortality — betalactam antibiotics alone 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 42/1668 (2.5%) 14/655 (2.1%) RR 1.13 (0.64 to 2.01) ⊕⊕⊕O MODERATE CRITICAL

Perinatal mortality — macrolide antibiotics alone 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 45/1653 (2.7%) 13/569 (2.3%) RR 1.17 (0.64 to 2.11) ⊕⊕⊕O MODERATE CRITICAL

Perinatal mortality — macrolide and betalactam antibiotics 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 52/1790 (2.9%) 14/779 (1.8%) RR 1.39 (0.79 to 2.43) ⊕⊕⊕O MODERATE CRITICAL

111

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 3 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients A particular type of antibiotic 4/155 (2.6%) Placebo or no treatment 2/139 (1.4%) Relative (95% CI)

Effect Absolute Quality Importance

Perinatal mortality — antibiotics active against anaerobic bacteria randomized trials serious6 no serious inconsistency no serious indirectness very serious RR 1.63 (0.36 to 7.39) 9 more per 1000 (from 9 fewer to 92 more) 1 fewer per 1000 (from 7 fewer to 12 more) 5 fewer per 1000 (from 8 fewer to 5 more) 2 fewer per 1000 (from 7 fewer to 8 more) not pooled ⊕OOO VERY LOW CRITICAL

Stillbirth — betalactam antibiotics alone 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious7 none 16/1668 (1.0%) 7/655 (1.1%) RR 0.91 (0.39 to 2.14) ⊕⊕OO LOW CRITICAL

Stillbirth — macrolide antibiotics alone 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious7 none 10/1653 (0.6%) 6/569 (1.1%) RR 0.54 (0.20 to 1.48) ⊕⊕OO LOW CRITICAL

Stillbirth — macrolide and betalactam antibiotics 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious7 none 13/1684 (0.8%) 6/663 (0.9%) RR 0.73 (0.28 to 1.90) ⊕⊕OO LOW CRITICAL

Stillbirth — antibiotics active against anaerobic bacteria 3 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious8 none 0/155 (0.0%) 26/1668 (1.6%) 0/139 (0.0%) 7/655 (1.1%) not pooled ⊕⊕OO LOW CRITICAL

Neonatal death — betalactam antibiotics alone 4 serious2 none RR 1.32 (0.61 to 2.86) 3 more per 1000 (from 4 fewer to 20 more) 8 more per 1000 (from 3 fewer to 32 more) 9 more per 1000 (from 1 fewer to 30 more) 9 more per 1000 (from 9 fewer to 92 more) ⊕⊕⊕O MODERATE CRITICAL

Neonatal death — macrolide antibiotics alone 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 35/1653 (2.1%) 7/569 (1.2%) RR 1.68 (0.77 to 3.64) ⊕⊕⊕O MODERATE CRITICAL

Neonatal death — macrolide and betalactam antibiotics 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 38/1760 (2.2%) 8/753 (1.1%) RR 1.83 (0.88 to 3.82) ⊕⊕⊕O MODERATE CRITICAL

Neonatal death — antibiotics active against anaerobic bacteria 3 randomized trials serious6 no serious inconsistency no serious indirectness very serious3 none 4/155 (2.6%) 2/139 (1.4%) RR 1.63 (0.36 to 7.39) ⊕OOO VERY LOW CRITICAL

112

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients A particular type of antibiotic 17/1133 (1.5%) Placebo or no treatment 8/382 (2.1%) Relative (95% CI)

Effect Absolute Quality Importance

Infant death (> 28 days) — betalactam antibiotics alone randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 RR 0.72 (0.31 to 1.65) 6 fewer per 1000 (from 14 fewer to 14 more) 3 more per 1000 (from 10 fewer to 31 more) 6 more per 1000 (from 8 fewer to 38 more) 7 fewer per 1000 (from 23 fewer to 15 more) 5 fewer per 1000 (from 22 fewer to 16 more) 4 more per 1000 (from 14 fewer to 25 more) 24 fewer per 1000 (from 53 fewer to 139 more) 3 more per 1000 (from 5 fewer to 23 more) ⊕⊕⊕O MODERATE CRITICAL

Infant death (> 28 days) — macrolide antibiotics alone 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 29/1204 (2.4%) 8/382 (2.1%) RR 1.15 (0.53 to 2.49) ⊕⊕⊕O MODERATE CRITICAL

Infant death (> 28 days) — macrolide and betalactam antibiotics 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 32/1171 (2.7%) 8/382 (2.1%) RR 1.30 (0.61 to 2.81) ⊕⊕⊕O MODERATE CRITICAL

Respiratory distress syndrome — betalactam antibiotics alone 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 142/1628 (8.7%) 154/1650 (9.3%) RR 0.93 (0.75 to 1.16) ⊕⊕⊕⊕ HIGH CRITICAL

Respiratory distress syndrome — macrolide antibiotics alone 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 133/1600 (8.3%) 138/1556 (8.9%) RR 0.94 (0.75 to 1.18) ⊕⊕⊕⊕ HIGH CRITICAL

Respiratory distress syndrome — macrolide and betalactam antibiotics 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 153/1682 (9.1%) 149/1700 (8.8%) RR 1.04 (0.84 to 1.29) ⊕⊕⊕O MODERATE CRITICAL

Respiratory distress syndrome — antibiotics active against anaerobic bacteria 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious3 none 2/58 (3.4%) 3/51 (5.9%) RR 0.59 (0.10 to 3.37) ⊕⊕OO LOW CRITICAL

Necrotizing enterocolitis — betalactam antibiotics alone 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 20/1621 (1.2%) 6/606 (1.0%) RR 1.31 (0.52 to 3.32) ⊕⊕⊕O MODERATE CRITICAL

113

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients A particular type of antibiotic 16/1600 (1.0%) Placebo or no treatment 4/519 (0.8%) Relative (95% CI)

Effect Absolute Quality Importance

Necrotizing enterocolitis — macrolide antibiotics alone randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 RR 1.30 (0.44 to 3.86) 2 more per 1000 (from 4 fewer to 22 more) 5 more per 1000 (from 5 fewer to 29 more) 59 fewer per 1000 (from 66 fewer to 1 more) 2 fewer per 1000 (from 9 fewer to 13 more) 2 fewer per 1000 (from 9 fewer to 13 more) 0 fewer per 1000 (from 7 fewer to 14 more) 80 fewer per 1000 (from 96 fewer to 45 more) 5 more per 1000 (from 42 fewer to 69 more) ⊕⊕⊕O MODERATE CRITICAL

Necrotizing enterocolitis — macrolide and betalactam antibiotics 2 randomized trials serious9 no serious inconsistency no serious indirectness serious2 none 26/1682 (1.5%) 9/663 (1.4%) RR 1.36 (0.60 to 3.11) ⊕⊕OO LOW CRITICAL

Necrotizing enterocolitis — antibiotics active against anaerobic bacteria 2 randomized trials serious9 no serious inconsistency no serious indirectness very serious3 none 0/101 (0.0%) 6/89 (6.7%) RR 0.13 (0.02 to 1.01) ⊕OOO VERY LOW CRITICAL

Intraventricular haemorrhage — betalactam antibiotics alone 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 19/1628 (1.2%) 9/613 (1.5%) RR 0.84 (0.38 to 1.87) ⊕⊕⊕O MODERATE CRITICAL

Intraventricular haemorrhage — macrolide antibiotics alone 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 18/1600 (1.1%) 7/519 (1.3%) RR 0.83 (0.35 to 1.99) ⊕⊕⊕O MODERATE CRITICAL

Intraventricular haemorrhage — macrolide and betalactam antibiotics 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 21/1682 (1.2%) 8/663 (1.2%) RR 0.97 (0.43 to 2.19) ⊕⊕⊕O MODERATE CRITICAL

Intraventricular haemorrhage — antibiotics active against anaerobic bacteria 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious3 none 1/58 (1.7%) 5/51 (9.8%) RR 0.18 (0.02 to 1.46) ⊕⊕OO LOW CRITICAL

Moderate/severe functional impairment at 7 years of age — betalactam antibiotics alone 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 131/763 (17.2%) 41/245 (16.7%) RR 1.03 (0.75 to 1.41) ⊕⊕⊕O MODERATE CRITICAL

114

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients A particular type of antibiotic 142/785 (18.1%) Placebo or no treatment 41/245 (16.7%) Relative (95% CI)

Effect Absolute Quality Importance

Moderate/severe functional impairment at 7 years of age — macrolide antibiotics alone randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 RR 1.08 (0.79 to 1.48) 13 more per 1000 (from 35 fewer to 80 more) 20 more per 1000 (from 30 fewer to 89 more) 3 more per 1000 (from 9 fewer to 41 more) 7 more per 1000 (from 8 fewer to 49 more) 28 more per 1000 (from 0 more to 107 more) 15 more per 1000 (from 49 fewer to 94 more) 49 more per 1000 (from 23 fewer to 131 more) ⊕⊕⊕O MODERATE CRITICAL

Moderate/severe functional impairment at 7 years of age. — macrolide and betalactam antibiotics 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 144/769 (18.7%) 41/245 (16.7%) RR 1.12 (0.82 to 1.53) ⊕⊕⊕O MODERATE CRITICAL

Cerebral palsy at 7 years of age — betalactam antibiotics alone 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 15/792 (1.9%) 4/257 (1.6%) RR 1.22 (0.41 to 3.63) ⊕⊕⊕O MODERATE CRITICAL

Cerebral palsy at 7 years of age — macrolide antibiotics alone 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 18/816 (2.2%) 4/257 (1.6%) RR 1.42 (0.48 to 4.15) ⊕⊕⊕O MODERATE CRITICAL

Cerebral palsy at 7 years of age — macrolide and betalactam antibiotics 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 35/795 (4.4%) 4/257 (1.6%) RR 2.83 (1.02 to 7.88) ⊕⊕⊕⊕ HIGH CRITICAL

Any functional impairment at 7 years of age — betalactam antibiotics alone 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 299/763 (39.2%) 92/245 (37.6%) RR 1.04 (0.87 to 1.25) ⊕⊕⊕⊕ HIGH CRITICAL

Any functional impairment at 7 years of age — macrolide antibiotics alone 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 333/785 (42.4%) 92/245 (37.6%) RR 1.13 (0.94 to 1.35) ⊕⊕⊕O MODERATE CRITICAL

115

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients A particular type of antibiotic 325/769 (42.3%) Placebo or no treatment 92/245 (37.6%) Relative (95% CI)

Effect Absolute Quality Importance

Any functional impairment at 7 years of age — macrolide and betalactam antibiotics randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 RR 1.13 (0.94 to 1.35) 49 more per 1000 (from 23 fewer to 131 more) ⊕⊕⊕O MODERATE CRITICAL

1 Statistical heterogeneity (I2 > 60%). 2 Wide confidence interval crossing the line of no effect. 3 Wide confidence interval crossing the line of no effect, few events and small sample size. 4 Statistical heterogeneity (I2 > 60%). Variation in size and direction of effect. 5 Wide confidence interval crossing the line of no effect and small sample size. 6 One study with design limitations contributed 80% of the weight. 7 Wide confidence interval crossing the line of no effect and few events. 8 No events. 9 One study with design limitations contributed > 40% of the weight.

116

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 5a.

Antibiotic prophylaxis for women at risk of preterm birth and ruptured membranes Source: Kenyon S, Boulvain M, Neilson JP. Antibiotics for preterm rupture of membranes. Cochrane Database Syst Rev. 2013;(12):CD001058. Quality assessment No. of patients Imprecision Other considerations none Any antibiotic 3104/3642 (85.2%) Placebo Relative (95% CI) RR 1.00 (0.98 to 1.03) Effect Absolute Quality Importance

No. of studies 3

Design

Risk of bias

Inconsistency

Indirectness

Birth < 37 weeks of gestation randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision 1102/1289 (85.5%) 0 fewer per 1000 (from 17 fewer to 26 more) not pooled ⊕⊕⊕⊕ HIGH CRITICAL

Maternal death 3 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious1 none 0/369 (0.0%) 126/767 (16.4%) 0/394 (0.0%) 196/792 (24.7%) not pooled ⊕⊕OO LOW CRITICAL

Chorioamnionitis 11 no serious imprecision none RR 0.66 (0.46 to 0.96) 84 fewer per 1000 (from 10 fewer to 134 fewer) not pooled ⊕⊕⊕⊕ HIGH CRITICAL

Major adverse drug reaction 3 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious1 none 0/3913 (0.0%) 1296/4128 (31.4%) 0/1574 (0.0%) 717/1799 (39.9%) not pooled ⊕⊕OO LOW CRITICAL

Birth within 48 hours of randomization 7 no serious imprecision none RR 0.71 (0.58 to 0.87) 116 fewer per 1000 (from 52 fewer to 167 fewer) 141 fewer per 1000 (from 74 fewer to 195 fewer) 8 fewer per 1000 (from 20 fewer to 6 more) 5 fewer per 1000 (from 17 fewer to 10 more) ⊕⊕⊕⊕ HIGH CRITICAL

Birth within 7 days of randomization 7 randomized trials no serious risk of bias serious2 no serious indirectness no serious imprecision none 2388/4145 (57.6%) 1221/1820 (67.1%) RR 0.79 (0.71 to 0.89) ⊕⊕⊕O MODERATE CRITICAL

Perinatal death/death before discharge (all studies: placebo and no treatment) 18 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 299/4604 (6.5%) 172/2268 (7.6%) RR 0.89 (0.74 to 1.08) ⊕⊕⊕O MODERATE CRITICAL

Perinatal death/death before discharge (sensitivity analysis: placebo-controlled trials only) 12 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 276/4315 (6.4%) 138/1986 (6.9%) RR 0.93 (0.76 to 1.14) ⊕⊕⊕O MODERATE CRITICAL

117

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 11 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Any antibiotic 100/4273 (2.3%) Placebo Relative (95% CI) RR 1.09 (0.65 to 1.83)

Effect Absolute Quality Importance

Neonatal necrotizing enterocolitis randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 58/1956 (3.0%) 3 more per 1000 (from 10 fewer to 25 more) 14 fewer per 1000 (from 47 fewer to 25 more) 30 fewer per 1000 (from 7 fewer to 50 fewer) not pooled ⊕⊕⊕O MODERATE CRITICAL

Neonatal respiratory distress syndrome 12 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 965/4303 (22.4%) 551/1984 (27.8%) RR 0.95 (0.83 to 1.09) ⊕⊕⊕⊕ HIGH CRITICAL

Treatment with surfactant 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 526/3584 (14.7%) 217/1225 (17.7%) RR 0.83 (0.72 to 0.96) ⊕⊕⊕⊕ HIGH CRITICAL

Neonatal encephalopathy 1 randomized trials randomized trials serious4 no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious1 none 0/30 (0.0%) 234/3654 (6.4%) 0/30 (0.0%) 104/1307 (8.0%) not pooled ⊕OOO VERY LOW CRITICAL

Positive neonatal blood culture 3 no serious risk of bias no serious imprecision none RR 0.79 (0.63 to 0.99) 17 fewer per 1000 (from 1 fewer to 29 fewer) 54 fewer per 1000 (from 25 fewer to 79 fewer) 18 fewer per 1000 (from 2 fewer to 30 fewer) 0 fewer per 1000 (from 29 fewer to 29 more) MD 53.83 higher (7.06 to 100.6 higher) ⊕⊕⊕⊕ HIGH CRITICAL

Neonatal infection including pneumonia 12 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 85/823 (10.3%) 141/857 (16.5%) RR 0.67 (0.52 to 0.85) ⊕⊕⊕⊕ HIGH CRITICAL

Major cerebral abnormality on ultrasound before discharge 12 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 240/4303 (5.6%) 184/1986 (9.3%) RR 0.81 (0.68 to 0.98) ⊕⊕⊕⊕ HIGH CRITICAL

Birth weight < 2500 g 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 2605/3614 (72.1%) 911/1262 (72.2%) RR 1.00 (0.96 to 1.04) ⊕⊕⊕⊕ HIGH CRITICAL

Birth weight (better indicated by higher values) 12 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 4355 2019 — ⊕⊕⊕⊕ HIGH CRITICAL

118

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Any antibiotic 938/2375 (39.5%) Placebo Relative (95% CI) RR 1.01 (0.91 to 1.12)

Effect Absolute Quality Importance

Serious childhood disability at 7 years of age randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision 311/796 (39.1%) 4 more per 1000 (from 35 fewer to 47 more) MD 5.05 lower (9.77 to 0.33 lower) 15 fewer per 1000 (from 117 fewer to 95 more) 23 fewer per 1000 (from 46 fewer to 5 more) ⊕⊕⊕⊕ HIGH CRITICAL

Days in neonatal intensive care unit (NICU) (better indicated by lower values) 3 randomized trials randomized trials serious5 no serious inconsistency serious2 no serious indirectness no serious indirectness serious6 none 110 115 — ⊕⊕OO LOW CRITICAL

Admission to NICU 4 no serious risk of bias no serious imprecision none 2583/3687 (70.1%) 975/1336 (73%) RR 0.98 (0.84 to 1.13) ⊕⊕⊕O MODERATE CRITICAL

Number of newborns requiring ventilation 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 757/3641 (20.8%) 292/1283 (22.8%) RR 0.90 (0.80 to 1.02) ⊕⊕⊕⊕ HIGH CRITICAL

1 No events. 2 Statistical heterogeneity (I2 > 60%). 3 Wide confidence interval crossing the line of no effect. 4 One study with design limitations. 5 Half the weight from a study with design limitations. 6 Estimate based on small sample size.

119

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 5b.

Antibiotic prophylaxis for women at risk of preterm birth and ruptured membranes (antibiotic regimens) Source: Kenyon S, Boulvain M, Neilson JP. Antibiotics for preterm rupture of membranes. Cochrane Database Syst Rev. 2013;(12):CD001058. Quality assessment No. of patients Imprecision Other considerations none Antibiotic prophylaxis 0/40 (0.0%) 0/329 (0.0%) 7/165 (4.2%) Placebo Relative (95% CI) not pooled Effect Absolute Quality Importance

No. of studies 1

Design

Risk of bias

Inconsistency

Indirectness

Maternal death: subgroup analysis by type of antibiotic — all penicillin (excluding co-amoxiclav) randomized trials randomized trials randomized trials no serious risk of bias no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness no serious indirectness very serious 0/45 (0.0%) 0/349 (0.0%) 10/167 (6.0%) not pooled ⊕⊕OO LOW CRITICAL

Maternal death: subgroup analysis by type of antibiotic — other antibiotic 2 very serious1 none not pooled not pooled ⊕⊕OO LOW CRITICAL

Perinatal death/death before discharge: subgroup analysis by type of antibiotic (placebo-controlled trials only) — all penicillin (excluding co-amoxiclav) 4 very serious2 none RR 0.73 (0.30 to 1.80) 16 fewer per 1000 (from 42 fewer to 48 more) 6 fewer per 1000 (from 26 fewer to 21 more) 7 fewer per 1000 (from 25 fewer to 18 more) 9 more per 1000 (from 21 fewer to 59 more) 7 fewer per 1000 (from 33 fewer to 86 more) 17 more per 1000 (from 3 more to 62 more) ⊕⊕OO LOW CRITICAL

Perinatal death/death before discharge: subgroup analysis by type of antibiotic (placebo-controlled trials only) — betalactam (including co-amoxiclav) 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 80/1236 (6.5%) 46/644 (7.1%) RR 0.91 (0.64 to 1.30) ⊕⊕⊕O MODERATE CRITICAL

Perinatal death/death before discharge: subgroup analysis by type of antibiotic (placebo-controlled trials only) — macrolide (including erythromycin) 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 84/1354 (6.2%) 56/784 (7.1%) RR 0.90 (0.65 to 1.25) ⊕⊕⊕O MODERATE CRITICAL

Perinatal death/death before discharge: subgroup analysis by type of antibiotic (placebo-controlled trials only) — other antibiotic 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 28/371 (7.5%) 26/391 (6.6%) RR 1.13 (0.68 to 1.88) ⊕⊕⊕O MODERATE CRITICAL

Necrotizing enterocolitis: subgroup analysis by type of antibiotic — all penicillin (excluding co-amoxiclav) 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 5/124 (4.0%) 6/138 (4.3%) RR 0.85 (0.25 to 2.97) ⊕⊕OO LOW CRITICAL

Necrotizing enterocolitis: subgroup analysis by type of antibiotic — betalactam (including co-amoxiclav) 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 29/1236 (2.3%) 3/644 (0.5%) RR 4.72 (1.57 to 14.23) ⊕⊕⊕⊕ HIGH CRITICAL

120

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 3 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Antibiotic prophylaxis 21/1322 (1.6%) Placebo Relative (95% CI) RR 0.88 (0.45 to 1.69)

Effect Absolute Quality Importance

Necrotizing enterocolitis: subgroup analysis by type of antibiotic — macrolide (including erythromycin) randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 19/754 (2.5%) 3 fewer per 1000 (from 14 fewer to 17 more) 8 fewer per 1000 (from 33 fewer to 33 more) 67 fewer per 1000 (from 30 fewer to 83 fewer) 22 fewer per 1000 (from 32 fewer to 222 more) 31 fewer per 1000 (from 80 fewer to 54 more) 67 fewer per 1000 (from 11 fewer to 108 fewer) 85 fewer per 1000 (from 7 fewer to 125 fewer) 13 fewer per 1000 (from 29 fewer to 10 more) ⊕⊕⊕O MODERATE CRITICAL

Necrotizing enterocolitis: subgroup analysis by type of antibiotic — other antibiotic 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 25/402 (6.2%) 30/421 (7.1%) RR 0.89 (0.54 to 1.47) ⊕⊕⊕O MODERATE CRITICAL

Neonatal infection including pneumonia: subgroup analysis by type of antibiotic — all penicillin (excluding co-amoxiclav) 5 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 6/258 (2.3%) 25/263 (9.5%) RR 0.30 (0.13 to 0.68) ⊕⊕⊕⊕ HIGH CRITICAL

Neonatal infection including pneumonia: subgroup analysis by type of antibiotic — betalactam (including co-amoxiclav) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious4 none 0/31 (0.0%) 1/31 (3.2%) RR 0.33 (0.01 to 7.88) ⊕⊕OO LOW CRITICAL

Neonatal infection including pneumonia: subgroup analysis by type of antibiotic — macrolide (including erythromycin) 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 19/163 (11.7%) 25/171 (14.6%) RR 0.79 (0.45 to 1.37) ⊕⊕⊕O MODERATE CRITICAL

Neonatal infection including pneumonia: subgroup analysis by type of antibiotic — other antibiotic 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 60/371 (16.2%) 90/392 (23.0%) RR 0.71 (0.53 to 0.95) ⊕⊕⊕⊕ HIGH CRITICAL

Major cerebral abnormality on ultrasound before discharge: subgroup analysis by type of antibiotic — all penicillin (excluding co-amoxiclav) 3 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 10/124 (8.1%) 23/138 (16.7%) RR 0.49 (0.25 to 0.96) ⊕⊕⊕⊕ HIGH CRITICAL

Major cerebral abnormality on ultrasound before discharge: subgroup analysis by type of antibiotic — betalactam (including co-amoxiclav) 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 53/1236 (4.3%) 39/644 (6.1%) RR 0.78 (0.52 to 1.16) ⊕⊕⊕O MODERATE CRITICAL

121

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 4 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Antibiotic prophylaxis 68/1352 (5.0%) Placebo Relative (95% CI) RR 0.93 (0.60 to 1.44)

Effect Absolute Quality Importance

Major cerebral abnormality on ultrasound before discharge: subgroup analysis by type of antibiotic — macrolide (including erythromycin) randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 47/784 (6.0%) 4 fewer per 1000 (from 24 fewer to 26 more) 27 fewer per 1000 (from 99 fewer to 116 more) ⊕⊕⊕O MODERATE CRITICAL

Major cerebral abnormality on ultrasound before discharge: subgroup analysis by type of antibiotic — other antibiotic 4 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none 63/402 (15.7%) 76/421 (18.1%) RR 0.85 (0.45 to 1.64) ⊕⊕⊕O MODERATE CRITICAL

1 No events. 2 Wide confidence interval crossing the line of no effect and few events. 3 Wide confidence interval crossing the line of no effect. 4 Wide confidence interval crossing the line of no effect, few events and small sample size.

122

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 5c.

Antibiotic prophylaxis for women at risk of preterm birth and ruptured membranes (erythromycin versus co-amoxiclav) Source: Kenyon S, Boulvain M, Neilson JP. Antibiotics for preterm rupture of membranes. Cochrane Database Syst Rev. 2013;(12):CD001058. Quality assessment No. of patients Imprecision Other considerations none Erythromycin Co-amoxiclav Relative (95% CI) RR 0.99 (0.96 to 1.03) Effect Absolute Quality Importance

No. of studies 1

Design

Risk of bias

Inconsistency

Indirectness

Birth < 37 weeks of gestation randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision 1006/1190 (84.5%) 1025/1205 (85.1%) 9 fewer per 1000 (from 34 fewer to 26 more) not pooled ⊕⊕⊕⊕ HIGH CRITICAL

Major adverse drug reaction 1 randomized trials randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious1 none 0/1190 (0.0%) 414/1190 (34.8%) 0/1205 (0.0%) 367/1205 (30.5%) not pooled ⊕⊕OO LOW CRITICAL

Birth within 48 hours of randomization 1 no serious imprecision none RR 1.14 (1.02 to 1.28) 43 more per 1000 (from 6 more to 85 more) 35 more per 1000 (from 6 fewer to 75 more) 7 fewer per 1000 (from 22 fewer to 15 more) 11 fewer per 1000 (from 1 fewer to 15 fewer) 2 fewer per 1000 (from 32 fewer to 32 more) 3 fewer per 1000 (from 29 fewer to 29 more) ⊕⊕⊕⊕ HIGH CRITICAL

Birth within 7 days of randomization 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 725/1190 (60.9%) 695/1205 (57.7%) RR 1.06 (0.99 to 1.13) ⊕⊕⊕⊕ HIGH CRITICAL

Perinatal death/death before discharge 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 70/1190 (5.9%) 79/1205 (6.6%) RR 0.90 (0.66 to 1.23) ⊕⊕⊕O MODERATE CRITICAL

Neonatal necrotizing enterocolitis 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 11/1190 (0.9%) 24/1205 (2.0%) RR 0.46 (0.23 to 0.94) ⊕⊕⊕⊕ HIGH CRITICAL

Neonatal respiratory distress syndrome 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 236/1190 (19.8%) 241/1205 (20.0%) RR 0.99 (0.84 to 1.16) ⊕⊕⊕⊕ HIGH CRITICAL

Treatment with surfactant 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 176/1190 (14.8%) 182/1205 (15.1%) RR 0.98 (0.81 to 1.19) ⊕⊕⊕⊕ HIGH CRITICAL

123

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Erythromycin Co-amoxiclav Relative (95% CI) RR 0.84 (0.62 to 1.15)

Effect Absolute Quality Importance

Positive neonatal blood culture randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 68/1190 (5.7%) 82/1205 (6.8%) 11 fewer per 1000 (from 26 fewer to 10 more) 4 more per 1000 (from 10 fewer to 24 more) 0 fewer per 1000 (from 36 fewer to 36 more) MD 19 higher (41.92 lower to 79.92 higher) 46 fewer per 1000 (from 88 fewer to 4 more) 0 fewer per 1000 (from 35 fewer to 35 more) 0 fewer per 1000 (from 30 fewer to 36 more) ⊕⊕⊕O MODERATE CRITICAL

Major cerebral abnormality on ultrasound before discharge 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 50/1190 (4.2%) 46/1205 (3.8%) RR 1.10 (0.74 to 1.63) ⊕⊕⊕O MODERATE CRITICAL

Birth weight < 2500 g 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 863/1190 (72.5%) 877/1205 (72.8%) RR 1.00 (0.95 to 1.05) ⊕⊕⊕⊕ HIGH CRITICAL

Birth weight (better indicated by higher values) 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious2 none 1190 1205 — ⊕⊕⊕O MODERATE CRITICAL

Serious childhood disability at 7 years of age 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 293/788 (37.2%) 344/824 (41.7%) RR 0.89 (0.79 to 1.01) ⊕⊕⊕⊕ HIGH CRITICAL

Neonatal intensive care 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 836/1190 (70.3%) 848/1205 (70.4%) RR 1.00 (0.95 to 1.05) ⊕⊕⊕⊕ HIGH CRITICAL

Number of newborns requiring ventilation 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness no serious imprecision none 251/1190 (21.1%) 254/1205 (21.1%) RR 1 .00 (0.86 to 1.17) ⊕⊕⊕⊕ HIGH CRITICAL

1 No events. 2 Wide confidence interval crossing the line of no effect.

124

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 5d. Antibiotic prophylaxis for women at risk of preterm birth and ruptured membranes (3-day versus 7-day ampicillin regimens) Source: Kenyon S, Boulvain M, Neilson JP. Antibiotics for preterm rupture of membranes. Cochrane Database Syst Rev. 2013;(12):CD001058. Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations No. of patients 3-day ampicillin regimen 8/42 (19.0%) 7-day ampicillin regimen 11/42 (26.2%) Relative (95% CI) Effect Absolute Quality Importance

Chorioamnionitis 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none RR 0.73 (0.33 to 1.63) 71 fewer per 1000 (from 175 fewer to 165 more) 23 more per 1000 (from 90 fewer to 312 more) 0 fewer per 1000 (from 179 fewer to 250 more) 37 fewer per 1000 (from 58 fewer to 119 more) 26 fewer per 1000 (from 43 fewer to 86 more) 15 fewer per 1000 (from 146 fewer to 188 more) 21 fewer per 1000 (from 30 fewer to 65 more) ⊕⊕OO LOW CRITICAL

Birth within 48 hours of randomization 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 8/42 (19.0%) 7/42 (16.7%) RR 1.14 (0.46 to 2.87) ⊕⊕OO LOW CRITICAL

Birth within 7 days of randomization 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 25/42 (59.5%) 25/42 (59.5%) RR 1 .00 (0.7 to 1.42) ⊕⊕OO LOW CRITICAL

Perinatal death/death before discharge 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 1/65 (1.5%) 4/65 (6.2%) RR 0.40 (0.05 to 2.94) ⊕⊕OO LOW CRITICAL

Neonatal necrotizing enterocolitis 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 1/65 (1.5%) 3/65 (4.6%) RR 0.43 (0.07 to 2.86) ⊕⊕OO LOW CRITICAL

Neonatal respiratory distress syndrome 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious2 none 24/65 (36.9%) 25/65 (38.5%) RR 0.96 (0.62 to 1.49) ⊕⊕OO LOW CRITICAL

Neonatal intraventricular haemorrhage 2 randomized trials no serious risk of bias no serious inconsistency no serious indirectness very serious1 none 0/65 (0.0%) 2/65 (3.1%) RR 0.33 (0.04 to 3.12) ⊕⊕OO LOW CRITICAL

125

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations

No. of patients 3-day ampicillin regimen 36/42 (85.7%) 7-day ampicillin regimen 36/42 (85.7%) Relative (95% CI)

Effect Absolute Quality Importance

Neonatal intensive care 1 randomized trials no serious risk of bias no serious inconsistency no serious indirectness serious3 none RR 1.00 (0.84 to 1.19) 0 fewer per 1000 (from 137 fewer to 163 more) ⊕⊕⊕O MODERATE CRITICAL

126

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 6a.

Mode of delivery for women at risk of preterm birth Source: Alfirevic Z, Milan SJ, Livio S. Caesarean section versus vaginal delivery for preterm birth in singletons. Cochrane Database Syst Rev. 2013;(9):CD000078. Quality assessment No. of patients Imprecision Other considerations none Caesarean section 7/58 (12.1%) 7/35 (20.0%) 0/23 (0.0%) 11/46 (23.9%) Vaginal delivery 0/58 (0.0%) 0/43 (0.0%) 0/15 (0.0%) 4/43 (9.3%) Relative (95% CI) RR 7.21 (1.37 to 38.08) RR 7.21 (1.37 to 38.08) not pooled Effect Absolute Quality Importance

No. of studies 4

Design

Risk of bias

Inconsistency

Indirectness

Major maternal postpartum complications randomized trials randomized trials randomized trials randomized trials serious2 no serious inconsistency no serious inconsistency no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness no serious indirectness no serious indirectness serious1 — ⊕⊕OO LOW CRITICAL

Major maternal postpartum complications — breech 3 serious2 serious1 none — ⊕⊕OO LOW CRITICAL

Major maternal postpartum complications — cephalic 1 serious3 very serious4 none not pooled ⊕OOO VERY LOW CRITICAL

Maternal puerperal pyrexia 3 serious2 serious1 none RR 2.98 (1.18 to 7.53) 184 more per 1000 (from 17 more to 607 more) 283 more per 1000 (from 26 more to 933 more) not pooled ⊕⊕OO LOW CRITICAL

Maternal puerperal pyrexia — breech 2 randomized trials serious2 no serious inconsistency no serious indirectness serious1 none 11/23 (47.8%) 4/28 (14.3%) RR 2.98 (1.18 to 7.53) ⊕⊕OO LOW CRITICAL

Maternal puerperal pyrexia — cephalic 1 randomized trials randomized trials serious3 no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious4 none 0/23 (0.0%) 1/53 (1.9%) 0/15 (0.0%) 1/50 (2.0%) not pooled ⊕OOO VERY LOW CRITICAL

Maternal wound infection 3 serious2 very serious5 none RR 1.16 (0.18 to 7.70) 3 more per 1000 (from 16 fewer to 134 more) 5 more per 1000 (from 23 fewer to 191 more) not pooled ⊕OOO VERY LOW CRITICAL

Maternal wound infection — breech 2 randomized trials serious2 no serious inconsistency no serious indirectness very serious5 none 1/30 (3.3%) 1/35 (2.9%) RR 1.16 (0.18 to 7.70) ⊕OOO VERY LOW CRITICAL

Maternal wound infection — cephalic 1 randomized trials serious3 no serious inconsistency no serious indirectness very serious4 none 0/23 (0.0%) 0/15 (0.0%) not pooled ⊕OOO VERY LOW CRITICAL

127

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 3 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Caesarean section 10/53 (18.9%) Vaginal delivery 4/50 (8.0%) Relative (95% CI) RR 2.63 (1.02 to 6.78)

Effect Absolute Quality Importance

Other maternal infection randomized trials serious2 no serious inconsistency no serious indirectness serious1 130 more per 1000 (from 2 more to 462 more) 186 more per 1000 (from 2 more to 661 more) not pooled ⊕⊕OO LOW CRITICAL

Other maternal infection — breech 2 randomized trials serious2 no serious inconsistency no serious indirectness serious1 none 10/30 (33.3%) 4/35 (11.4%) RR 2.63 (1.02 to 6.78) ⊕⊕OO LOW CRITICAL

Other maternal infection — cephalic 1 randomized trials randomized trials serious3 no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious4 none 0/23 (0.0%) 0/58 (0.0%) 0/15 (0.0%) 4/58 (6.9%) not pooled ⊕OOO VERY LOW CRITICAL

Cord prolapse 4 serious2 very serious5 none RR 0.25 (0.03 to 1.92) 52 fewer per 1000 (from 67 fewer to 63 more) 70 fewer per 1000 (from 90 fewer to 86 more) not pooled ⊕OOO VERY LOW CRITICAL

Cord prolapse — breech 3 randomized trials serious2 no serious inconsistency no serious indirectness very serious5 none 0/35 (0.0%) 4/43 (9.3%) RR 0.25 (0.03 to 1.92) ⊕OOO VERY LOW CRITICAL

Cord prolapse — cephalic 1 randomized trials randomized trials randomized trials randomized trials randomized trials serious3 no serious inconsistency no serious inconsistency no serious inconsistency no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness no serious indirectness no serious indirectness no serious indirectness very serious4 none 0/23 (0.0%) 0/58 (0.0%) 0/35 (0.0%) 0/23 (0.0%) 22/23 (95.7%) 0/15 (0.0%) 0/58 (0.0%) 0/43 (0.0%) 0/15 (0.0%) 28/28 (100.0%) not pooled ⊕OOO VERY LOW CRITICAL

Head entrapment 4 serious2 very serious4 none not pooled not pooled ⊕OOO VERY LOW CRITICAL

Head entrapment — breech 3 serious2 very serious4 none not pooled not pooled ⊕OOO VERY LOW CRITICAL

Head entrapment — cephalic 1 serious3 very serious4 none not pooled not pooled ⊕OOO VERY LOW CRITICAL

Delivery < 7 days after entry — breech 2 serious2 very serious6 none RR 0.95 (0.73 to 1.24) 50 fewer per 1000 (from 270 fewer to 240 more) ⊕OOO VERY LOW CRITICAL

128

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 3 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Caesarean section 2/46 (4.3%) Vaginal delivery 8/43 (18.6%) Relative (95% CI) RR 0.29 (0.07 to 1.14)

Effect Absolute Quality Importance

Perinatal death randomized trials serious2 no serious inconsistency no serious indirectness very serious5 132 fewer per 1000 (from 173 fewer to 26 more) 154 fewer per 1000 (from 204 fewer to 105 more) 89 fewer per 1000 (from 129 fewer to 305 more) 360 more per 1000 (from 91 fewer to 1000 more) 37 fewer per 1000 (from 47 fewer to 158 more) 144 fewer per 1000 (from 234 fewer to 32 more) 147 fewer per 1000 (from 257 fewer to 103 more) 136 fewer per 1000 (from 232 fewer to 235 more) — ⊕OOO VERY LOW CRITICAL

Perinatal death — breech 2 randomized trials serious2 no serious inconsistency no serious indirectness very serious5 none 1/23 (4.3%) 6/28 (21.4%) RR 0.28 (0.05 to 1.49) ⊕OOO VERY LOW CRITICAL

Perinatal death — cephalic 1 randomized trials serious3 no serious inconsistency no serious indirectness very serious5 none 1/23 (4.3%) 2/15 (13.3%) RR 0.33 (0.03 to 3.29) ⊕OOO VERY LOW CRITICAL

Birth asphyxia — breech 1 randomized trials serious3 no serious inconsistency no serious indirectness very serious6 none 5/5 (100.0%) 4/7 (57.1%) RR 1.63 (0.84 to 3.14) ⊕OOO VERY LOW CRITICAL

Neonatal fitting/seizures — breech 3 randomized trials serious2 no serious inconsistency no serious indirectness very serious5 none 0/35 (0.0%) 2/42 (4.8%) RR 0.22 (0.01 to 4.32) ⊕OOO VERY LOW CRITICAL

Respiratory distress syndrome 3 randomized trials serious2 no serious inconsistency no serious indirectness very serious6 none 9/53 (17.0%) 16/50 (32.0%) RR 0.55 (0.27 to 1.10) ⊕OOO VERY LOW CRITICAL

Respiratory distress syndrome — breech 2 randomized trials serious2 no serious inconsistency no serious indirectness very serious6 none 6/30 (20.0%) 12/35 (34.3%) RR 0.57 (0.25 to 1.30) ⊕OOO VERY LOW CRITICAL

Respiratory distress syndrome — cephalic 1 randomized trials serious3 no serious inconsistency no serious indirectness very serious6 none 3/23 (13.0%) 4/15 (26.7%) RR 0.49 (0.13 to 1.88) ⊕OOO VERY LOW CRITICAL

Hypoxic ischaemic encephalopathy — breech 1 randomized trials serious3 no serious inconsistency no serious indirectness very serious5 none 1/5 (20.0%) 0/7 (0.0%) RR 4.00 (0.20 to 82.01) ⊕OOO VERY LOW CRITICAL

129

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 4 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Caesarean section 4/56 (7.1%) Vaginal delivery 4/54 (7.4%) Relative (95% CI) RR 0.92 (0.27 to 3.14)

Effect Absolute Quality Importance

Intracranial pathology randomized trials serious2 no serious inconsistency no serious indirectness very serious5 6 fewer per 1000 (from 54 fewer to 159 more) 44 fewer per 1000 (from 95 fewer to 462 more) 32 fewer per 1000 (from 68 fewer to 143 more) 64 more per 1000 (from 52 fewer to 1000 more) — ⊕OOO VERY LOW CRITICAL

Birth injury to baby — breech 1 randomized trials serious3 no serious inconsistency no serious indirectness very serious5 none 1/18 (5.6%) 2/20 (10.0%) RR 0.56 (0.05 to 5.62) ⊕OOO VERY LOW CRITICAL

Intracranial pathology — breech 3 randomized trials serious2 no serious inconsistency no serious indirectness very serious5 none 1/33 (3.0%) 3/39 (7.7%) RR 0.58 (0.12 to 2.86) ⊕OOO VERY LOW CRITICAL

Intracranial pathology — cephalic 1 randomized trials serious3 no serious inconsistency no serious indirectness very serious5 none 3/23 (13.0%) 1/15 (6.7%) RR 1.96 (0.22 to 17.1) ⊕OOO VERY LOW CRITICAL

Necrotizing enterocolitis — breech 1 randomized trials randomized trials serious3 no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness very serious5 none 2/5 (40.0%) 4/53 (7.5%) 0/7 (0.0%) 5/50 (10.0%) RR 6.67 (0.39 to 114.78) RR 0.76 (0.12 to 4.66) ⊕OOO VERY LOW CRITICAL

Neonatal infection (proven) 3 serious2 very serious6 none 24 fewer per 1000 (from 88 fewer to 366 more) 9 more per 1000 (from 80 fewer to 1000 more) 89 fewer per 1000 (from 129 fewer to 305 more) ⊕OOO VERY LOW CRITICAL

Neonatal infection (proven) — breech 2 randomized trials serious2 no serious inconsistency no serious indirectness very serious5 none 3/30 (10.0%) 3/35 (8.6%) RR 1.10 (0.07 to 17.74) ⊕OOO VERY LOW CRITICAL

Neonatal infection (proven) — cephalic 1 randomized trials serious3 no serious inconsistency no serious indirectness very serious6 none 1/23 (4.3%) 2/15 (13.3%) RR 0.33 (0.03 to 3.29) ⊕OOO VERY LOW CRITICAL

130

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 1 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Caesarean section 5 Vaginal delivery 7 Relative (95% CI) —

Effect Absolute Quality Importance

Ventilation (days) — breech (better indicated by lower values) randomized trials serious3 no serious inconsistency no serious indirectness very serious6 MD 18.26 higher (19.9 lower to 56.42 higher) 373 more per 1000 (from 124 fewer to 1000 more) MD 3.71 higher (20.85 lower to 28.27 higher) ⊕OOO VERY LOW CRITICAL

Need for mechanical ventilation — breech 1 randomized trials serious3 no serious inconsistency no serious indirectness very serious6 none 4/5 (80.0%) 3/7 (42.9%) RR 1.87 (0.71 to 4.88) ⊕OOO VERY LOW CRITICAL

Supplemental oxygen (days) — breech (better indicated by lower values) 1 randomized trials serious3 no serious inconsistency no serious indirectness very serious6 none 5 7 — ⊕OOO VERY LOW CRITICAL

1 Estimate based on small sample size. 2 All studies contributing data had design limitations. 3 One study with design limitations. 4 No events. 5 Wide confidence interval crossing the line of no effect, few events and small sample size. 6 Wide confidence interval crossing the line of no effect and small sample size.

131

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 7a.

Kangaroo mother care (KMC) versus conventional care for preterm newborns Source: Conde-Agudelo A, Belizan JM, Diaz-Rossello J. Kangaroo mother care to reduce morbidity and mortality in low birthweight infants. Cochrane Database Syst Rev. 2011;(3):CD002771. Quality assessment No. of patients Imprecision Other considerations none KMC Conventional care 45/848 (5.3%) Relative (95% CI) RR 0.60 (0.39 to 0.92) Effect Absolute (95% CI) 21 fewer per 1000 (from 4 fewer to 32 fewer) 23 fewer per 1000 (from 6 fewer to 34 fewer) 24 more per 1000 (from 31 fewer to 596 more) 21 fewer per 1000 (from 3 fewer to 33 fewer) 23 fewer per 1000 (from 5 fewer to 36 fewer) 12 more per 1000 (from 34 fewer to 210 more) 53 fewer per 1000 (from 27 fewer to 73 fewer) 50 fewer per 1000 (from 22 fewer to 66 fewer) Quality Importance

No. of studies 8

Study design

Risk of bias

Inconsistency

Indirectness

Overall mortality at discharge or at 40—41 weeks postmenstrual age randomized trials not serious not serious not serious not serious 28/888 (3.2%) ⊕⊕⊕⊕ HIGH CRITICAL

Mortality at discharge or at 40—41 weeks postmenstrual age for studies in low- and middle-income countries 7 randomized trials not serious not serious not serious not serious none 26/855 (3.0%) 44/821 (5.4%) RR 0.57 (0.37 to 0.89) ⊕⊕⊕⊕ HIGH CRITICAL

Mortality at discharge or at 40—41 weeks postmenstrual age for studies in high-income countries 1 randomized trial not serious serious1 not serious serious2 none 2/33 (6.1%) 1/27 (3.7%) RR 1.64 (0.16 to 17.09) ⊕⊕OO LOW CRITICAL

Overall mortality at latest follow-up 11 randomized trials not serious not serious not serious not serious none 46/1088 (4.2%) 69/1079 (6.4%) RR 0.67 (0.48 to 0.95) ⊕⊕⊕⊕ HIGH CRITICAL

Overall mortality at last follow-up for studies in low- and middle-income countries 9 randomized trials not serious not serious not serious not serious none 42/1020 (4.1%) 66/1016 (6.5%) RR 0.65 (0.45 to 0.93) ⊕⊕⊕⊕ HIGH CRITICAL

Overall mortality at last follow-up for studies in high-income countries 2 randomized trials not serious not serious not serious very serious3 none 4/68 (5.9%) 3/63 (4.8%) RR 1.25 (0.29 to 5.42) ⊕⊕OO LOW CRITICAL

Severe infection at last follow-up 7 randomized trials not serious not serious not serious not serious none 47/685 (6.9%) 80/658 (12.2%) RR 0.56 (0.40 to 0.78) ⊕⊕⊕⊕ HIGH CRITICAL

Nosocomial infection at discharge or at 40—41 weeks postmenstrual age 3 randomized trials not serious not serious not serious not serious none 19/469 (4.1%) 40/444 (9.0%) RR 0.45 (0.27 to 0.76) ⊕⊕⊕⊕ HIGH CRITICAL

132

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies 6 Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients KMC Conventional care 95/344 (27.6%) Relative (95% CI) RR 0.34 (0.17 to 0.67)

Effect Absolute (95% CI) 182 fewer per 1000 (from 91 fewer to 229 fewer) 61 fewer per 1000 (from 15 more to 119 fewer) 25 fewer per 1000 (from 4 more to 42 fewer) Quality Importance

Hypothermia randomized trials not serious not serious not serious not serious 32/354 (9.0%) ⊕⊕⊕⊕ HIGH CRITICAL

Hyperthermia 4 randomized trials not serious not serious not serious serious1 none 52/228 (22.8%) 64/220 (29.1%) RR 0.79 (0.59 to 1.05) ⊕⊕⊕O MODERATE CRITICAL

Readmission to hospital at latest follow-up 2 randomized trials not serious not serious not serious serious1 none 18/474 (3.8%) 30/472 (6.4%) RR 0.60 (0.34 to 1.06) ⊕⊕⊕O MODERATE CRITICAL

1 Only one study conducted, hence consistency could not be assessed. 2 Wide confidence intervals for the outcome. 3 Very wide confidence intervals because of very few events.

133

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 7b.

Continuous Kangaroo mother care (KMC) versus conventional care for preterm newborns Source: Conde-Agudelo A, Belizan JM, Diaz-Rossello J. Kangaroo mother care to reduce morbidity and mortality in low birthweight infants. Cochrane Database Syst Rev. 2011;(3):CD002771. Quality assessment No. of patients Imprecision Other considerations Continuous KMC Conventional care Relative (95% CI) Effect Absolute (95% CI) Quality Importance

No. of studies

Study design

Risk of bias

Inconsistency

Indirectness

Overall mortality at discharge or at 40—41 weeks postmenstrual age 3 randomized trials not serious not serious not serious not serious none 23/575 (4.0%) 37/542 (6.8%) RR 0.60 (0.38 to 0.96) 27 fewer per 1000 (from 3 fewer to 42 fewer) ⊕⊕⊕⊕ HIGH CRITICAL

Overall mortality at latest follow-up 4 randomized trials not serious not serious not serious not serious none 39/692 (5.6%) 59/692 (8.5%) RR 0.67 (0.46 to 0.98) 28 fewer per 1000 (from 2 fewer to 46 fewer) ⊕⊕⊕⊕ HIGH CRITICAL

Severe infection at latest follow-up 1 randomized trials not serious serious1 not serious serious2 none 26/343 (7.6%) 35/320 (10.9%) RR 0.69 (0.43 to 1.12) 34 fewer per 1000 (from 13 more to 62 fewer) ⊕⊕OO LOW CRITICAL

Nosocomial infection at discharge or at 40—41 weeks postmenstrual age 1 randomized trials not serious serious1 not serious not serious none 13/343 (3.8%) 25/320 (7.8%) RR 0.49 (0.25 to 0.93) 40 fewer per 1000 (from 5 fewer to 59 fewer) ⊕⊕⊕O MODERATE CRITICAL

1 Only one trial, hence consistency could not be assessed. 2 Wide confidence intervals crossing the line of no effect.

134

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 7c.

Intermittent Kangaroo mother care (KMC) versus conventional care for preterm newborns Source: Conde-Agudelo A, Belizan JM, Diaz-Rossello J. Kangaroo mother care to reduce morbidity and mortality in low birthweight infants. Cochrane Database Syst Rev. 2011;(3):CD002771. Quality assessment No. of patients Imprecision Other considerations none Intermittent KMC 5/313 (1.6%) Conventional care 8/306 (2.6%) Relative (95% CI) RR 0.59 (0.19 to 1.81) Effect Absolute (95% CI) 11 fewer per 1000 (from 21 fewer to 21 more) 8 fewer per 1000 (from 19 fewer to 20 more) 73 fewer per 1000 (from 36 fewer to 96 fewer) 74 fewer per 1000 (from 5 fewer to 102 fewer) 182 fewer per 1000 (from 91 fewer to 229 fewer) 61 fewer per 1000 (from 15 more to 119 fewer) Quality Importance

No. of studies 5

Study design

Risk of bias

Inconsistency

Indirectness

Overall mortality at discharge or at 40—41 weeks postmenstrual age randomized trials not serious not serious not serious serious1 ⊕⊕⊕O MODERATE CRITICAL

Overall mortality at latest follow-up 7 randomized trials not serious not serious not serious serious1 none 7/396 (1.8%) 10/387 (2.6%) RR 0.68 (0.26 to 1.77) ⊕⊕⊕O MODERATE CRITICAL

Severe infection 6 randomized trials not serious not serious not serious not serious none 21/342 (6.1%) 45/338 (13.3%) RR 0.45 (0.28 to 0.73) ⊕⊕⊕⊕ HIGH CRITICAL

Nosocomial infection at discharge or at 40—41 weeks postmenstrual age 2 randomized trials not serious not serious not serious not serious none 6/124 (4.8%) 15/124 (12.1%) RR 0.39 (0.16 to 0.96) ⊕⊕⊕⊕ HIGH CRITICAL

Hypothermia 6 randomized trials not serious not serious not serious not serious none 320/354 (90.4%) 95/344 (27.6%) RR 0.34 (0.17 to 0.67) ⊕⊕⊕⊕ HIGH CRITICAL

Hyperthermia 4 randomized trials not serious not serious not serious serious1 none 52/228 (22.8%) 64/220 (29.1%) RR 0.79 (0.59 to 1.05) ⊕⊕⊕O MODERATE CRITICAL

1 Wide confidence intervals crossing the line of no effect and few events.

135

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 7d.

Radiant warmers versus incubators for care of unstable or sick preterm newborns Source: Flenady VJ, Woodgate PG. Radiant warmers versus incubators for regulating body temperature in newborn infants. Cochrane Database Syst Rev. 2003;(4):CD000435. (updated for this guideline)

No. of studies 2 Design Risk of bias

Quality assessment Inconsistency Indirectness Imprecision Other considerations none

No. of patients Radiant warmer 1/47 (2.1%) Incubator Relative (95% CI) RR 0.27 (0.05 to 1.59)

Effect Absolute Quality Importance

Neonatal mortality randomized trials no serious risk of bias1 no serious inconsistency serious2 very serious3,4 5/47 (10.6%) 78 fewer per 1000 (from 101 fewer to 63 more) 67 fewer per 1000 (from 140 fewer to 215 more) 53 fewer per 1000 (from 66 fewer to 200 more) 15 fewer per 1000 (from 22 fewer to 153 more) MD 1.06 higher (0.94 lower to 3.06 higher) MD 0.86 higher (1.49 lower to 3.21 higher) ⊕OOO VERY LOW CRITICAL

Culture positive sepsis (assessed with positive blood culture) 1 randomized trials serious5 serious6 serious7 serious3 none 3/30 (10.0%) 5/30 (16.7%) RR 0.60 (0.16 to 2.29) ⊕OOO VERY LOW CRITICAL

Bronchopulmonary dysplasia 1 randomized trials very serious5,8 serious6 serious7 serious3 none 0/30 (0.0%) 2/30 (6.7%) RR 0.20 (0.01 to 4.00) ⊕OOO VERY LOW CRITICAL

Severe intraventricular haemorrhage (IVH) (grade 3 or 4) (assessed by ultrasound ) 2 randomized trials very serious9,10 no serious inconsistency serious2 very serious3,4 none 0/45 (0.0%) 1/45 (2.2%) RR 0.33 (0.01 to 7.87) ⊕OOO VERY LOW CRITICAL

Weight gain (better indicated by higher values) 2 randomized trials very serious5,8 no serious inconsistency serious2 serious3 none 43 43 — ⊕OOO VERY LOW IMPORTANT

Time to regain birth weight (better indicated by lower values) 2 randomized trials very serious11,12 no serious inconsistency serious2 serious3 none 45 45 — ⊕OOO VERY LOW IMPORTANT

136

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

No. of studies 3 Design Risk of bias

Quality assessment Inconsistency Indirectness Imprecision Other considerations none

No. of patients Radiant warmer 26 Incubator Relative (95% CI) —

Effect Absolute Quality Importance

Insensible water losses (better indicated by lower values) randomized trials very serious8,13 no serious inconsistency serious2 no serious imprecision 27 MD 0.94 higher (0.47 to 1.41 higher) ⊕OOO VERY LOW IMPORTANT

1 Majority of evidence from the study with no blinding but the outcome is objective. 2 All the studies were from high-income countries. 3 95% CI around the pooled estimate includes both: (1) no effect and (2) appreciable benefit or appreciable harm. 4 Event rate very low. 5 Post-randomization exclusions. 6 Single study. 7 Study from high-income country. 8 No blinding of outcome assessment. 9 All the evidence from the study with post-randomization exclusions. 10 All the evidence from the study with no blinding of outcome assessment. 11 Majority of evidence from the study with post-randomization exclusions (one infant excluded because of refusal of consent following randomization). 12 Majority of evidence from the study with no blinding of outcome assessment. 13 Unclear allocation concealment in all the studies.

137

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 7e.

Plastic bags or wraps versus conventional care immediately after birth in preterm (and some term) newborns Source: Oatley H, Blencowe H, Lawn JE. Systematic review of the effect of coverings including plastic bags and wraps on mortality and morbidity in preterm and term neonates. 2014 (unpublished). Quality assessment No. of patients Imprecision Other considerations none Covering in plastic bags or wraps 27/166 (16.3%) Conventional care 34/175 (19.4%) Relative (95% CI) RR 0.84 (0.54 to 1.30) Effect Absolute (95% CI) 31 fewer per 1000 (from 58 more to 89 fewer) 183 more per 1000 (from 32 fewer to 970 more) 330 fewer per 1000 (from 220 fewer to 425 fewer) 711 more per 1000 (from 101 fewer to 1000 more) 151 fewer per 1000 (from 209 fewer to 345 more) Quality Importance

No. of studies 7

Study design

Risk of bias

Inconsistency

Indirectness

All-cause neonatal mortality including neonates born ≤ 29 weeks of gestation randomized trials not serious not serious serious1 very serious2 ⊕OOO VERY LOW CRITICAL

All-cause mortality including neonates born 26—36 weeks of gestation 2 randomized trials not serious serious3 serious1 very serious3 none 12/99 (12.1%) 13/115 (11.3%) RR 2.62 (0.72 to 9.58) ⊕OOO VERY LOW CRITICAL

Hypothermia 2 randomized trials not serious serious4 serious1 not serious3 none 51/112 (45.5%) 92/117 (78.6%) RR 0.58 (0.46 to 0.72) ⊕⊕OO LOW CRITICAL

Necrotizing enterocolitis 1 randomized trials5 serious6 serious7 serious1 very serious3 none 34/180 (18.9%) 29/203 (14.3%) RR 5.98 (0.29 to 121.80) ⊕OOO VERY LOW CRITICAL

Intraventricular haemorrhage 2 randomized trials5 not serious not serious serious1 serious2 none 32/219 (14.6%) 52/241 (21.6%) RR 0.30 (0.03 to 2.60) ⊕⊕OO LOW CRITICAL

1 All facility-based studies conducted in high-income settings. 2 Wide confidence intervals crossing the line of no effect. 3 Very wide confidence intervals crossing the line of no effect. 4 Some heterogeneity. 5 There were two other observational studies. 6 Methodological inconsistencies. 7 No explanation was provided.

138

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 8a. Continuous positive airway pressure (CPAP) therapy for preterm newborns with respiratory distress syndrome Source: Ho JJ, Subramaniam P, Henderson-Smart DJ, Davis PG. Continuous distending pressure for respiratory distress syndrome in preterm infants. Cochrane Database Syst Rev 2002;(2):CD002271 (updated for this guideline) Quality assessment No. of studies 6 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none No. of patients CPAP Oxygen by head box or cannula 32/179 (17.9%) Relative (95% CI) RR 0.52 (0.32 to 0.87) Effect Absolute Quality Importance

In-hospital mortality (assessed with: mortality during initial hospital stay ) randomized trials serious1 no serious inconsistency serious2 no serious imprecision 16/176 (9.1%) 86 fewer per 1000 (from 23 fewer to 122 fewer) 10 more per 1000 (from 25 fewer to 107 more) 147 fewer per 1000 (from 47 fewer to 231 fewer) 153 fewer per 1000 (from 237 fewer to 129 more) 87 more per 1000 (from 16 more to 224 more) ⊕⊕OO LOW CRITICAL

Bronchopulmonary dysplasia (assessed with: oxygen requirement at 28 days of age ) 3 randomized trials serious3 no serious inconsistency serious2 serious4 none 6/126 (4.8%) 6/134 (4.5%) RR 1.22 (0.44 to 3.39) ⊕OOO VERY LOW CRITICAL

Respiratory failure warranting mechanical ventilation 5 randomized trials serious3 no serious inconsistency serious2 no serious imprecision none 56/154 (36.4%) 84/160 (52.5%) RR 0.72 (0.56 to 0.91) ⊕⊕OO LOW CRITICAL

Need for surfactant 1 randomized trials serious3 serious5 serious6 serious4 none 3/26 (11.5%) 7/26 (26.9%) RR 0.43 (0.12 to 1.48) ⊕OOO VERY LOW CRITICAL

Any air leak 6 randomized trials serious3 no serious inconsistency serious2 no serious imprecision none 25/172 (14.5%) 11/179 (6.1%) RR 2.42 (1.26 to 4.65) ⊕⊕OO LOW CRITICAL

1 Allocation concealment unclear in two studies with combined weight of > 50%. 2 All studies are from high-income countries. 3 Neither outcome assessors nor treatment team was blinded to group allocation. 4 95% CI around the pooled estimate includes both: (1) no effect and (2) appreciable benefit or appreciable harm. 5 Single study. 6 Study from high-income country.

139

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 8b. Timing of initiation (early versus late) of continuous positive airway pressure (CPAP) therapy for preterm newborns with respiratory distress syndrome Source: Ho JJ, Henderson-Smart DJ, Davis PG. Early versus delayed initiation of continuous distending pressure for respiratory distress syndrome in preterm infants. Cochrane Database Syst Rev. 2002;(2):CD002975. (updated for this guideline) Quality assessment No. of studies 2 Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none No. of patients Early CPAP Late CPAP Relative (95% CI) RR 0.93 (0.13 to 6.81) Effect Absolute Quality Importance

Neonatal mortality randomized trials serious1 no serious inconsistency serious2 serious3 1/23 (4.3%) 2/38 (5.3%) 4 fewer per 1000 (from 46 fewer to 306 more) 56 fewer per 1000 (from 112 fewer to 45 more) 16 fewer per 1000 (from 48 fewer to 163 more) 142 fewer per 1000 (from 13 fewer to 214 fewer) 280 fewer per 1000 (from 54 fewer to 436 fewer) 24 fewer per 1000 (from 93 fewer to 135 more) ⊕OOO VERY LOW CRITICAL

In-hospital mortality 7 randomized trials no serious risk of bias4 no serious inconsistency serious5 serious3 none 15/109 (13.8%) 24/128 (18.8%) RR 0.70 (0.40 to 1.24) ⊕⊕OO LOW CRITICAL

Bronchopulmonary dysplasia (assessed with: oxygen requirement at 28 days of age) 2 randomized trials serious6 no serious inconsistency no serious indirectness7 very serious3,8 none 2/53 (3.8%) 3/55 (5.5%) RR 0.70 (0.12 to 3.98) ⊕OOO VERY LOW CRITICAL

Need for mechanical ventilation 6 randomized trials very serious1,6 no serious inconsistency serious2 no serious imprecision none 13/73 (17.8%) 29/92 (31.5%) — ⊕OOO VERY LOW CRITICAL

Need for surfactant therapy 1 randomized trials serious6 serious9 no serious indirectness no serious imprecision none 18/36 (50.0%) 28/36 (77.8%) RR 0.64 (0.44 to 0.93) ⊕⊕OO LOW CRITICAL

Air leaks 5 randomized trials very serious1,6 no serious inconsistency serious2 serious3 none 8/63 (12.7%) 12/81 (14.8%) RR 0.84 (0.37 to 1.91) ⊕OOO VERY LOW CRITICAL

140

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations none

No. of patients Early CPAP Late CPAP Relative (95% CI) RR 0.46 (0.27 to 0.79)

Effect Absolute Quality Importance

Sepsis 1 randomized trials serious6 serious9 no serious indirectness no serious imprecision 11/36 (30.6%) 24/36 (66.7%) 360 fewer per 1000 (from 140 fewer to 487 fewer) ⊕⊕OO LOW CRITICAL

1 Allocation concealment unclear in most/all studies. 2 Studies from high-income countries. 3 95% CI around the pooled estimate includes both: (1) no effect and (2) appreciable benefit or appreciable harm. 4 Allocation concealment mentioned in two studies with combined weight of > 50%. 5 All studies except one with weight of evidence < 50% from high-income countries. 6 Neither treatment team nor outcome assessors were masked to group allocation. 7 > 50% weight of evidence from the study from a low- and middle-income country setting. 8 Only two and three events in the two studies (both groups combined). 9 Single study.

141

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 9a.

Surfactant replacement therapy with animal-derived surfactants for preterm newborns with respiratory distress syndrome Source: Seger N, Soll R. Animal derived surfactant extract for treatment of respiratory distress syndrome. Cochrane Database Syst Rev. 2009;(2):CD007836. (updated for this guideline) Quality assessment No. of patients Imprecision Other considerations Surfactant replacement therapy 145/744 (19.5%) No therapy or placebo Relative (95% CI) Effect Absolute Quality Importance

No. of studies

Design

Risk of bias

Inconsistency

Indirectness

Neonatal mortality 10 randomized trials no serious risk of bias no serious inconsistency serious1 no serious imprecision none 206/725 (28.4%) RR 0.68 (0.57 to 0.82) 9 fewer per 100 (from 5 fewer to 12 fewer) 18 fewer per 1000 (from 59 fewer to 30 more) 165 fewer per 1000 (from 130 fewer to 189 fewer) 16 more per 1000 (from 13 fewer to 63 more) 23 more per 1000 (from 21 fewer to 79 more) 19 fewer per 1000 (from 58 fewer to 28 more) ⊕⊕⊕O MODERATE CRITICAL

Bronchopulmonary dysplasia (assessed with: use of supplemental oxygen at 36 weeks postmenstrual age) 9 randomized trials no serious risk of bias serious2 serious1 no serious imprecision none 278/796 (34.9%) 285/772 (36.9%) RR 0.95 (0.84 to 1.08) ⊕⊕OO LOW CRITICAL

Air leaks (assessed with: any air leak syndromes such as pulmonary interstitial emphysema, pneumothorax, pneumomediastinum, etc.) 7 randomized trials no serious risk of bias no serious inconsistency serious1 no serious imprecision none 102/694 (14.7%) 213/686 (31%) RR 0.47 (0.39 to 0.58) ⊕⊕⊕O MODERATE CRITICAL

Pulmonary haemorrhage 2 randomized trials no serious risk of bias no serious inconsistency serious1 serious3 none 32/457 (7.0%) 24/441 (5.4%) RR 1.29 (0.77 to 2.15) ⊕⊕OO LOW CRITICAL

Sepsis (assessed with: culture proven bacterial sepsis) 4 randomized trials no serious risk of bias no serious inconsistency serious1 serious3 none 95/513 (18.5%) 82/499 (16.4%) RR 1.14 (0.87 to 1.48) ⊕⊕OO LOW CRITICAL

Severe intraventricular haemorrhage (IVH) (assessed with: grade 3 or 4 IVH detected by ultrasound or computerized tomography [CT] scan of the head) 10 randomized trials no serious risk of bias no serious inconsistency serious1 serious3 none 195/758 (25.7%) 206/743 (27.7%) RR 0.93 (0.79 to 1.10) ⊕⊕OO LOW CRITICAL

1 All the studies were done in level-3 NICUs in high-income countries. 2 There was significant heterogeneity. 3 Confidence intervals were wide and crossed the line of no effect.

142

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 9b. Surfactant replacement therapy with protein-free synthetic surfactants for preterm newborns with respiratory distress syndrome Source: Soll R. Synthetic surfactant for respiratory distress syndrome in preterm infants. Cochrane Database Syst Rev. 2000;(2):CD001149. Quality assessment No. of studies Study design Other considerations No. of patients Protein-free synthetic surfactant treatment or prophylaxis 149/1176 (12.7%) Natural surfactant Relative (95% CI) Effect Absolute (95% CI) Quality Importance

Risk of bias

Inconsistency

Indirectness

Imprecision

Overall neonatal mortality 6 randomized trials not serious not serious serious1 not serious non applicable 200/1176 (17.0%) RR 0.73 (0.61 to 0.88) 43 fewer per 1000 (from 1 fewer to 1 fewer) 42 fewer per 1000 (from 1 fewer to 1 fewer) 88 fewer per 1000 (from 1 fewer to 1 fewer) 34 fewer per 1000 (from 1 fewer to 1 fewer) 13 fewer per 1000 (from 1 fewer to 1 fewer) ⊕⊕⊕O MODERATE CRITICAL

In-hospital mortality 6 randomized trials not serious not serious serious1 not serious non applicable 201/1178 (17.1%) 251/1174 (21.3%) RR 0.79 (0.68 to 0.92) ⊕⊕⊕O MODERATE CRITICAL

Air leaks 5 randomized trials not serious not serious serious1 not serious non applicable 186/1161 (16.0%) 289/1167 (24.8%) RR 0.64 (0.55 to 0.76) ⊕⊕⊕O MODERATE CRITICAL

Bronchopulmonary dysplasia 5 randomized trials not serious not serious serious1 not serious non applicable 123/1123 (11.0%) 162/1125 (14.4%) RR 0.75 (0.61 to 0.92) ⊕⊕⊕O MODERATE CRITICAL

Severe intraventricular haemorrhage 5 randomized trials not serious not serious serious1 serious2 non applicable 80/1161 (6.8%) 95/1167 (8.1%) RR 0.84 (0.63 to 1.12) ⊕⊕OO LOW CRITICAL

1 All trials conducted in high-income countries. 2 Wide confidence intervals crossing the line of no effect.

143

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 9c.

Protein-free synthetic surfactant treatment or prophylaxis versus natural surfactant therapy for preterm newborns with respiratory distress syndrome Source: Soll RF, Blanco F. Natural surfactant extract versus synthetic surfactant for neonatal respiratory distress syndrome. Cochrane Database Syst Rev. 2001;(2):CD000144. Quality assessment No. of patients Imprecision Other considerations Protein-free synthetic surfactant 765/2838 (27.0%) Animal derived surfactant extract 553/2609 (21.2%) Relative (95% CI) Effect Absolute Quality Importance

No. of studies

Design

Risk of bias

Inconsistency

Indirectness

Neonatal mortality 12 randomized trials no serious risk of bias no serious inconsistency serious1 no serious imprecision none RR 1.07 (0.99 to 1.17) 15 more per 1000 (from 2 fewer to 36 more) 0 fewer per 1000 (from 24 fewer to 30 more) 36 more per 1000 (from 19 more to 56 more) 2 fewer per 1000 (from 24 fewer to 19 more) 7 fewer per 1000 (from 23 fewer to 12 more) ⊕⊕⊕O MODERATE CRITICAL

Bronchopulmonary dysplasia (assessed with: use of supplemental oxygen at 36 weeks postmenstrual age) 7 randomized trials serious2 no serious inconsistency serious1 no serious imprecision none 688/2123 (32.4%) 569/1883 (30.2%) RR 1.00 (0.92 to 1.10) ⊕⊕OO LOW CRITICAL

Pneumothorax 11 randomized trials no serious risk of bias no serious inconsistency serious1 no serious imprecision none 313/2804 (11.2%) 187/2577 (7.3%) RR 1.49 (1.26 to 1.77) ⊕⊕⊕O MODERATE CRITICAL

Sepsis 10 randomized trials serious2 no serious inconsistency serious1 no serious imprecision none 735/2776 (26.5%) 594/2468 (24.1%) RR 0.99 (0.90 to 1.08) ⊕⊕OO LOW CRITICAL

Severe intraventricular haemorrhage 9 randomized trials serious2 no serious inconsistency serious1 no serious imprecision none 349/2590 (13.5%) 316/2379 (13.3%) RR 0.95 (0.83 to 1.09) ⊕⊕OO LOW CRITICAL

1 All the studies were done in level-3 NICUs in high-income countries. 2 Subjective outcome; blinding of outcome assessment not done in most studies.

144

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 9d. Protein-containing synthetic surfactant treatment or prophylaxis versus natural surfactant therapy for preterm newborns with respiratory distress syndrome Source: Pfister RH, Soll RF, Wiswell T. Protein-containing synthetic surfactant versus animal derived surfactant extract for the prevention and treatment of respiratory distress syndrome. Cochrane Database Syst Rev. 2007;(3):CD006069. Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations No. of patients Proteincontaining synthetic surfactant 114/646 (17.6%) Animalderived surfactant Relative (95% CI) Effect Absolute Quality Importance

Neonatal mortality 2 randomized trials no serious risk of bias no serious inconsistency serious1 serious2 none 81/382 (21.2%) RR 0.79 (0.61 to 1.02) 45 fewer per 1000 (from 83 fewer to 4 more) 3 fewer per 1000 (from 53 fewer to 60 more) 0 fewer per 1000 (from 36 fewer to 49 more) 33 fewer per 1000 (from 59 fewer to 7 more) 4 more per 1000 (from 66 fewer to 83 more) 55 fewer per 1000 (from 19 fewer to 80 fewer) ⊕⊕OO LOW CRITICAL

Bronchopulmonary dysplasia at 36 weeks of gestation 2 randomized trials no serious risk of bias serious3 serious1 no serious imprecision none 235/646 (36.4%) 127/382 (33.2%) RR 0.99 (0.84 to 1.18) ⊕⊕OO LOW CRITICAL

Air leaks 2 randomized trials no serious risk of bias serious3 serious1 serious2 none 93/646 (14.4%) 51/382 (13.4%) RR 1.00 (0.73 to 1.37) ⊕OOO VERY LOW CRITICAL

Pulmonary haemorrhage 2 randomized trials no serious risk of bias no serious inconsistency serious1 serious2 none 61/646 (9.4%) 46/382 (12%) RR 0.73 (0.51 to 1.06) ⊕⊕OO LOW CRITICAL

Sepsis (culture proven) 1 randomized trials no serious risk of bias serious4 serious1 no serious imprecision none 232/527 (44.0%) 113/258 (43.8%) RR 1.01 (0.85 to 1.19) ⊕⊕OO LOW CRITICAL

Necrotizing enterocolitis 2 randomized trials no serious risk of bias serious5 serious1 no serious imprecision none 50/646 (7.7%) 53/382 (13.9%) RR 0.60 (0.42 to 0.86) ⊕⊕OO LOW CRITICAL

145

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations

No. of patients Proteincontaining synthetic surfactant 16/119 (13.4%) Animalderived surfactant Relative (95% CI)

Effect Absolute Quality Importance

Severe intraventricular haemorrhage (IVH) (assessed with: grade 3 or 4 IVH detected by ultrasound or computerized tomography [CT] scan of the head) 1 randomized trials no serious risk of bias serious4 serious1 serious2 none 11/124 (8.9%) RR 1.52 (0.73 to 3.13) 46 more per 1000 (from 24 fewer to 189 more) ⊕OOO VERY LOW CRITICAL

1 Both the studies were done in level-3 NICUs in high-income countries. 2 95% CI around the pooled estimate of effect includes both: (1) no effect and (2) increased risk. 3 Effect size of the two studies in different directions. 4 Single study. 5 Effect size of the two studies in same direction but I2 > 60%.

146

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 9e.

Prophylactic surfactant replacement therapy versus rescue surfactant therapy with or without continuous positive airway pressure (CPAP) for preterm newborns with respiratory distress syndrome Source: Rojas-Reyes MX, Morley CJ, Soll R. Prophylactic versus selective use of surfactant in preventing morbidity and mortality in preterm infants. Cochrane Database Syst Rev. 2012;(3):CD000510. Quality assessment No. of patients Imprecision Other considerations Prophylactic surfactant replacement therapy 246/2256 (10.9%) Selective surfactant therapy Relative (95% CI) Effect Absolute Quality Importance

No. of studies

Design

Risk of bias

Inconsistency

Indirectness

Neonatal mortality 10 randomized trials no serious risk of bias serious1 serious2 no serious imprecision none 274/2251 (12.2%) RR 0.89 (0.76 to 1.04) 13 fewer per 1000 (from 29 fewer to 5 more) 36 fewer per 1000 (from 63 fewer to 0 more) 5 more per 1000 (from 20 fewer to 32 more) 13 fewer per 1000 (from 27 fewer to 4 more) 1 more per 1000 (from 6 fewer to 15 more) 16 fewer per 1000 (from 34 fewer to 7 more) ⊕⊕OO LOW CRITICAL

In-hospital mortality 5 randomized trials no serious risk of bias serious1 serious2 serious3 none 101/728 (13.9%) 125/730 (17.1%) RR 0.79 (0.63 to 10) ⊕OOO VERY LOW CRITICAL

Bronchopulmonary dysplasia 10 randomized trials no serious risk of bias no serious inconsistency serious2 no serious imprecision none 362/1607 (22.5%) 357/1584 (22.5%) RR 1.02 (0.91 to 1.14) ⊕⊕⊕O MODERATE CRITICAL

Air leaks 9 randomized trials no serious risk of bias serious5 serious2 Serious3 none 165/2044 (8.1%) 189/2032 (9.3%) RR 0.86 (0.71 to 1.04) ⊕OOO VERY LOW CRITICAL

Pulmonary haemorrhage 4 randomized trials no serious risk of bias no serious inconsistency serious2 very serious3 none 13/1015 (1.3%) 12/1008 (1.2%) RR 1.05 (0.49 to 2.22) ⊕OOO VERY LOW CRITICAL

Sepsis 6 randomized trials no serious risk of bias Serious1 serious2 Serious3 none 95/1227 (7.7%) 113/1211 (9.3%) RR 0.83 (0.64 to 1.08) ⊕OOO VERY LOW CRITICAL

147

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations

No. of patients Prophylactic surfactant replacement therapy 211/2170 (9.7%) Selective surfactant therapy Relative (95% CI)

Effect Absolute Quality Importance

Severe intraventricular haemorrhage 10 randomized trials no serious risk of bias no serious inconsistency serious2 Serious3 none 241/2177 (11.1%) RR 0.87 (0.74 to 1.04) 14 fewer per 1000 (from 29 fewer to 4 more) ⊕⊕OO LOW CRITICAL

1 Significant heterogeneity: P < 0.05; I2 > 50%. 2 All the studies were done in level-3 neonatal intensive care units in high-income countries. 3 Wide confidence interval around the pooled estimate of effect crossing the line of no effect.

148

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 9f.

Prophylactic surfactant replacement therapy versus rescue surfactant therapy without continuous positive airway pressure (CPAP) for preterm newborns with respiratory distress syndrome Source: Rojas-Reyes MX, Morley CJ, Soll R. Prophylactic versus selective use of surfactant in preventing morbidity and mortality in preterm infants. Cochrane Database Syst Rev. 2012;(3):CD000510. (updated for this guideline) Quality assessment No. of patients Imprecision Other considerations Prophylactic surfactant replacement therapy 122/1394 (8.8%) Rescue surfactant therapy without CPAP 172/1367 (12.6%) Relative (95% CI) Effect Absolute (95% CI) Quality Importance

No. of studies

Design

Risk of bias

Inconsistency

Indirectness

Overall neonatal mortality 8 randomized trials serious1 not serious serious2 not serious none RR 0.69 (0.56 to 0.85) 39 fewer per 1000 (from 19 fewer to 55 fewer) 64 fewer per 1000 (from 16 fewer to 100 fewer) 22 fewer per 1000 (from 2 fewer to 39 fewer) 3 fewer per 1000 (from 8 fewer to 10 more) 9 fewer per 1000 (from 19 more to 33 fewer) 14 fewer per 1000 (from 9 more to 33 fewer) 31 fewer per 1000 (from 8 fewer to 47 fewer) ⊕⊕OO LOW CRITICAL

In-hospital mortality 4 randomized trials serious1 not serious serious2 not serious none 86/520 (16.5%) 116/510 (22.7%) RR 0.72 (0.56 to 0.93) ⊕⊕OO LOW CRITICAL

Air leaks 8 randomized trials serious1 not serious serious2 not serious none 117/1391 (8.4%) 144/1369 (10.5%) RR 0.79 (0.63 to 0.98) ⊕⊕OO LOW CRITICAL

Pulmonary haemorrhage 3 randomized trials serious1 not serious serious2 serious3 none 7/806 (0.9%) 9/786 (1.1%) RR 0.73 (0.28 to 1.87) ⊕OOO VERY LOW CRITICAL

Bronchopulmonary dysplasia 9 randomized trials serious1 not serious serious2 serious3 none 235/1411 (16.7%) 242/1378 (17.6%) RR 0.95 (0.81 to 1.11) ⊕OOO VERY LOW CRITICAL

Severe intraventricular haemorrhage 8 randomized trials serious1 not serious serious2 serious3 none 127/1339 (9.5%) 143/1317 (10.9%) RR 0.87 (0.70 to 1.08) ⊕OOO VERY LOW CRITICAL

Sepsis 5 randomized trials serious1 not serious serious2 not serious none 68/1022 (6.7%) 96/991 (9.7%) RR 0.68 (0.51 to 0.92) ⊕⊕OO LOW CRITICAL

1 No blinding in the assessment of outcomes except one study. 2 All trials from level 3 neonatal Intensive care units in high-income countries. 3 Wide confidence intervals including no effect.

149

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 9g. Prophylactic surfactant replacement therapy versus rescue surfactant therapy with continuous positive airway pressure (CPAP) for preterm newborns with respiratory distress syndrome Source: Rojas-Reyes MX, Morley CJ, Soll R. Prophylactic versus selective use of surfactant in preventing morbidity and mortality in preterm infants. Cochrane Database Syst Rev. 2012;(3):CD000510. (updated for this guideline) Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations No. of patients Prophylactic surfactant replacement therapy 124/862 (14.4%) Rescue surfactant therapy with CPAP 172/884 (19.5%) Relative (95% CI) Effect Absolute (95% CI) Quality Importance

Overall neonatal mortality 2 randomized trials serious1 not serious serious2 serious3 none RR 1.24 (0.97 to 1.58) 47 more per 1000 (from 6 fewer to 113 more) 31 more per 1000 (from 9 fewer to 120 more) 5 more per 1000 (from 18 fewer to 41 more) 15 more per 1000 (from 6 fewer to 100 more) 89 more per 1000 (from 6 fewer to 201 more) 18 fewer per 1000 (from 24 more to 50 fewer) ⊕OOO VERY LOW CRITICAL

In-hospital mortality 1 randomized trials serious1 serious4 serious2 serious3 none 15/208 (7.2%) 9/220 (4.1%) RR 1.76 (0.79 to 3.94) ⊕OOO VERY LOW CRITICAL

Air leaks 1 randomized trials serious1 serious4 serious2 serious3 none 48/653 (7.4%) 45/663 (6.8%) RR 1.08 (0.73 to 1.60) ⊕OOO VERY LOW CRITICAL

Pulmonary haemorrhage 1 randomized trials serious1 serious4 serious2 serious3 none 6/209 (2.9%) 3/222 (1.4%) RR 2.12 (0.54 to 8.39) ⊕OOO VERY LOW CRITICAL

Bronchopulmonary dysplasia 1 randomized trials serious1 serious4 serious2 serious3 none 127/196 (64.8%) 115/206 (55.8%) RR 1.16 (0.99 to 1.36) ⊕OOO VERY LOW CRITICAL

Severe intraventricular haemorrhage 2 randomized trials serious1 not serious serious2 serious3 none 12/72 (16.7%) 14/92 (15.2%) RR 0.88 (0.67 to 1.16) ⊕OOO VERY LOW CRITICAL

150

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations

No. of patients Prophylactic surfactant replacement therapy 27/205 (13.2%) Rescue surfactant therapy with CPAP 17/220 (7.7%) Relative (95% CI)

Effect Absolute (95% CI) Quality Importance

Sepsis 1 randomized trials serious1 serious4 serious2 serious3 none RR 1.70 (0.96 to 3.03) 54 more per 1000 (from 3 fewer to 157 more) ⊕OOO VERY LOW CRITICAL

1 No blinding in the assessment of outcomes except one study. 2 No trial from low- and middle-income countries. 3 Wide confidence intervals crossing the line of no effect. 4 Only one study, hence consistency could not be assessed.

151

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Table 9h. Early surfactant replacement therapy (within 2—3 hours of birth) versus late rescue surfactant therapy (after waiting for symptoms to worsen) with or without continuous positive airway pressure (CPAP) for preterm newborns with respiratory distress syndrome Source: Bahadue FL, Soll R. Early versus delayed selective surfactant treatment for neonatal respiratory distress syndrome. Cochrane Database Syst Rev. 2012;(11):CD001456. (updated for this guideline) Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations No. of patients Early surfactant replacement therapy 353/1782 (19.8%) Late rescue surfactant therapy 424/1795 (23.6%) Relative (95% CI) Effect Absolute Quality Importance

Neonatal mortality 6 randomized trials no serious risk of bias no serious inconsistency serious1 no serious imprecision none RR 0.84 (0.74 to 0.95) 38 fewer per 1000 (from 12 fewer to 61 fewer) 33 fewer per 1000 (from 3 fewer to 60 fewer) 38 fewer per 1000 (from 18 fewer to 53 fewer) 178 fewer per 1000 (from 100 fewer to 237 fewer) 7 fewer per 1000 (from 30 fewer to 20 more) ⊕⊕⊕O MODERATE CRITICAL

In-hospital mortality 5 randomized trials no serious risk of bias no serious inconsistency serious1 no serious imprecision none 376/1570 (23.9%) 431/1587 (27.2%) RR 0.88 (0.78 to 0.99) ⊕⊕⊕O MODERATE CRITICAL

Bronchopulmonary dysplasia (assessed with: use of supplemental oxygen at 36 weeks postmenstrual age) 4 randomized trials serious3 no serious inconsistency serious4 no serious imprecision none 118/1135 (10.4%) 177/1547 (11.4%) RR 0.67 (0.54 to 0.84) ⊕⊕OO LOW CRITICAL

Air leaks (assessed with: any air leak syndromes such as pulmonary interstitial emphysema, pneumothorax, pneumomediastinum, etc.) 2 randomized trials no serious risk of bias no serious inconsistency serious4 no serious imprecision none 65/233 (27.9%) 105/230 (45.7%) RR 0.61 (0.48 to 0.78) ⊕⊕⊕O MODERATE CRITICAL

Severe intraventricular haemorrhage 3 randomized trials serious3 no serious inconsistency serious4 serious5 none 245/1519 (16.1%) 257/1531 (16.8%) RR 0.96 (0.82 to 1.12) ⊕OOO VERY LOW CRITICAL

152

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

Quality assessment No. of studies Design Risk of bias Inconsistency Indirectness Imprecision Other considerations

No. of patients Early surfactant replacement therapy 24/35 (68.6%) Late rescue surfactant therapy 24/40 (60%) Relative (95% CI)

Effect Absolute Quality Importance

Confirmed bacterial sepsis 1 randomized trials no serious risk of bias serious6 no serious indirectness7 serious5 none RR 1.14 (0.81 to 1.60) 84 more per 1000 (from 114 fewer to 360 more) ⊕⊕OO LOW CRITICAL

1 All the studies were done in level-3 neonatal intensive care units (NICUs) in high-income countries except one. 2 We used the data from the studies with the lowest and highest risk in the control group to estimate the “low” and “high” control risk. 3 Subjective outcome; blinding of outcome assessment unclear (not mentioned) in all the studies. 4 All the studies were done in level-3 NICUs in high-income countries. 5 95% CI around the pooled estimate of effect includes both: (1) no effect and (2) increased risk. 6 Single study. 7 Conducted in a low- or middle-income country.

153

WHO RECOMMENDATIONS ON INTERVENTIONS TO IMPROVE PRETERM BIRTH OUTCOMES: EVIDENCE BASE

For more information, please contact: Department of Reproductive Health and Research E-mail: reproductivehealth@who.int www.who.int/reproductivehealth Department of Maternal, Newborn, Child and Adolescent Health E-mail: mncah@who.int www.who.int/maternal_child_adolescent World Health Organization Avenue Appia 20, CH-1211 Geneva 27 Switzerland

Informations clés

Organisation Organisation mondiale de la santé (OMS)

Type de document Technical Documents

Date d'adoption 24 août 2015

Source Organisation mondiale de la santé