WHO/HIV/2014.24 © World Health Organization 2014
Global Hepatitis Programme Guideline development for Hepatitis C virus Screening, Care and Treatment in low- and middle- income countries
PICO 1: Testing A systematic review and meta-analysis of targeted HCV antibody testing interventions
Conducted by Health Protection Scotland, Glasgow and Burnet Institute, Melbourne 20 June 2013 Review Members Dr Esther Aspinall Dr Stephen Corson Mr David Hunt Dr Joseph Doyle Prof Margaret Hellard Prof Sharon Hutchinson Review Advisory Group Dr Mark Stoove Prof David Goldberg Prof Stanley Luchters Dr Alexander Thompson Dr Stefan Wiktor Mr Tim Nguyen Dr Bryce Smith Dr Yngve Falck-Ytter Ms Rebecca Morgan
Health Protection Scotland, Glasgow University of Strathclyde, Glasgow Burnet Institute, Melbourne, Australia Burnet Institute, Melbourne Burnet Institute, Melbourne Caledonian University, Glasgow
Burnet Institute, Melbourne Health Protection Scotland, Glasgow Burnet Institute, Melbourne St Vincent’s Hospital, University of Melbourne WHO Global Hepatitis Program, Geneva WHO Global Hepatitis Program, Geneva Centre for Disease Control and Prevention, Atlanta Case Western Reserve University, Cleveland Centre for Disease Control and Prevention, Atlanta
PICO 1 (Testing): Targeted HCV antibody testing interventions
© World Health Organization 2014 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 noncommercial 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. Dotted and dashed lines on maps represent approximate border lines for which there may not yet be full agreement. 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.
P a g e |2
PICO 1 (Testing): Targeted HCV antibody testing interventions
Table of Contents BACKGROUND ..................................................................................................................................... 4 METHODS ............................................................................................................................................ 4 RESULTS .............................................................................................................................................. 9 CONCLUSIONS ................................................................................................................................... 12 FIGURES............................................................................................................................................. 13 Figure 1: Flowchart for the Systematic Review ............................................................................ 13 Figure 2: Forest plot of comparison: HCV testing intervention versus no intervention, outcome: HCV antibody tests among those eligible for testing.................................................................... 14 Figure 3: Forest plot of comparison: HCV testing intervention versus no intervention, outcome: HCV antibody positive cases detected among those eligible for testing...................................... 15 Figure 4: Forest plot of comparison: HCV testing intervention versus no intervention, outcome: referrals to a specialist among the estimated HCV antibody positive population ....................... 16 Figure 5: Forest plot of comparison: HCV testing intervention versus no intervention, outcome: attendance at specialist appointment among the estimated HCV antibody positive population 16 Figure 6: Forest plot of comparison: HCV testing intervention versus no intervention, outcome: commencing treatment among the estimated HCV antibody positive population ..................... 17 Figure 7: Forest plot of comparison: HCV testing intervention versus no intervention, outcome: achieving SVR among the estimated HCV antibody positive population ..................................... 17 Figure 8: Forest plot of comparison: HCV testing intervention versus no intervention, outcome: all-cause mortality among the estimated HCV antibody positive population* ............................ 18 TABLES............................................................................................................................................... 19 Table 1: Direct evidence................................................................................................................ 19 Table 2a: GRADE summary of findings: testing for HCV, and HCV positive cases detected among the population eligible for testing ................................................................................................ 35 Table 2b: GRADE summary of findings: management of HCV positive individuals among the estimated HCV antibody positive population ............................................................................... 35 Table 2c: Anticipated absolute effects of HCV testing interventions by population HCV prevalence..................................................................................................................................... 36 REREFENCES ...................................................................................................................................... 37 Appendix 1: Search terms ................................................................................................................. 39 Appendix 2: Sensitivity analyses examining the effect of denominator on pooled effect size ........ 40 Appendix 3: Sensitivity analyses examining the effect of study design on pooled effect size ......... 41 Appendix 4: Sensitivity analysis including/excluding specific studies .............................................. 42 Appendix 5: GRADE outcome assessments ...................................................................................... 43 Appendix 6: Newcastle-Ottawa Scale ............................................................................................... 46 Appendix 7: GRADE approach to assessing the quality of evidence across studies ......................... 47
P a g e |3
PICO 1 (Testing): Targeted HCV antibody testing interventions
BACKGROUND The World Health Organization estimates that between 130 and 150 million people are chronically infected with hepatitis C (HCV) virus worldwide (World Health Organization 2012), most of whom are unaware of their infection (Lavanchy, 2009). People at increased risk of HCV include people who inject drugs (PWID), and people undergoing medical procedures (including the transfusion of infected blood or blood products) in an unsafe setting (Alter et al.,1990, Villano et al., 1997). Intranasal drug use and cosmetic procedures (such as tattooing, body piercing, and manicures) have also been implicated as risk factors for HCV (Karmochine et al., 2006). The relative importance of these risk factors will vary depending on the geographical setting and population studied.
People living with chronic HCV are at increased risk of liver cirrhosis, hepatocellular carcinoma, and liver-related mortality (Grebely, 2011). It has been suggested that testing or screening interventions to detect HCV (in order to facilitate earlier access to HCV treatment and care), may prevent or reduce the risk of liver-related complications (World Health Organization 2012). The purpose of this review was to investigate the effectiveness of targeted HCV testing interventions, in terms of case detection, referral to HCV treatment and care, and the prevention of liver-related morbidity and mortality.
METHODS Narrative review question: Who should be tested for HCV antibodies?
PICO question: Population: People with a history of behaviours or exposures that place them at increased risk of HCV infection Intervention: Targeted HCV antibody testing. “Targeted” means testing of individuals based either on their being part of a defined risk group (e.g. PWID, people living with HIV) or through questions to elicit a history of HCV-risk behaviours Comparison: Symptomatic HCV antibody testing or no targeted HCV antibody testing. “Symptomatic testing” means antibody testing based on the presence of liver-related signs or symptoms. Outcomes: Number of HCV antibody tests, number of HCV antibody positive tests, number of referrals to a specialist*, number attending a specialist**, number commencing treatment for HCV,
P a g e |4
PICO 1 (Testing): Targeted HCV antibody testing interventions
number of cases of HCV transmission, HCV disease progression (liver cirrhosis, hepatocellular carcinoma, decompensated cirrhosis), sustained virological response (SVR), quality of life, all-cause mortality Study type/limits: Experimental or observational studies published between 1st January 1994 and 15th March 2013
* Defined as number of individuals who were referred to a HCV specialist, regardless of whether they actually attended ** Defined as at least one attendance at a HCV specialist appointment during the period of follow-up
Search strategy: A systematic review was carried out for relevant articles in any language, using the following databases and information sources: Ovid MEDLINE, Ovid EMBASE, LILACS, and the Cochrane Library of Systematic reviews The Centre for Reviews and Dissemination Database (which includes the NHS Economic Evaluations Database [NHS EDD], the Health Technology Assessments Database [HTA], and the Database of Abstracts of Reviews of Effects [DARE]) The European Network of Health Economic Evaluations Database (EURONHEED). Reference lists of all relevant articles and reviews Recommendations from Guideline Development Group (GDG) members and other experts in the field Relevant articles identified during the conduct of the other systematic reviews
Search terms are shown in Appendix I, briefly summarized as: Hepatitis C/HCV AND test, casefinding, or screening.
Conduct of the review: The review was carried out using Review Manager (version 5.2; Cochrane Collaboration, Copenhagen), and GRADE profiler (version 3.6; GRADE working group) software, and complied with Cochrane methodology and PRISMA guidelines on reporting (Prospero registration number CRD42013004146). Due to the large number of citations, a single reviewer removed all obviously irrelevant articles on citation screening, and subsequently, two reviewers carried out the abstract and full text screening. A third reviewer was consulted on any points of difference. Review team members are listed in appendix 2.
P a g e |5
PICO 1 (Testing): Targeted HCV antibody testing interventions
-
Although cost-effectiveness was not included as an outcome of this review, relevant cost effectiveness studies were reviewed to check for any unpublished empirical data that met the PICO criteria.
-
Foreign language articles were translated online using Google Translate, with additional interpretation sought from the primary authors as required.
-
Missing data on outcomes of interest were requested from primary authors, with each author contacted twice in the case of non-response.
Quality appraisal: Studies were assessed as having a low, high, or unclear level of bias using the Cochrane Risk of Bias assessment tool.
Data extraction: Data were extracted from each study by the primary reviewer, and checked by the secondary reviewer. The following data were extracted, where available:
-
Study characteristics: country, study design; Study population; Participant characteristics: age, sex, ethnicity, HIV co-infection; Setting: GP, hospital, alcohol and drug services, harm reduction services, sexual health clinics;
-
Inclusion/exclusion criteria for study; Sample size; Intervention group: selection and characteristics of intervention group, prevalence of HCV among those tested;
-
Intervention: type of HCV antibody test used (laboratory versus rapid testing), delivered by healthcare professional/drug support worker/ancillary staff, year or time period of intervention delivered, costs of delivery;
-
Intervention subgroup: ‘Direct’ interventions were defined as interventions where a health or social care provider was given ‘in-practice’ support to offer risk assessment and/or HCV testing. ‘General’ interventions comprised of ‘out-of practice’ interventions, for example, invitations to information sessions for care providers or people at risk, leaflets or posters on HCV testing, or TV/radio awareness-raising campaigns;
-
Comparison group: selection and characteristics of comparison group;
P a g e |6
PICO 1 (Testing): Targeted HCV antibody testing interventions
-
Analysis: number offered intervention, number accepted intervention, reason for refusal, time to follow-up, statistical analysis, primary outcomes of study, secondary outcomes of study;
-
Results: number of HCV antibody tests, number of HCV antibody positive tests, number of referrals to a specialist, number attending a specialist, number commencing treatment for HCV, number of cases of HCV transmission, HCV disease progression (liver cirrhosis, hepatocellular carcinoma, decompensated cirrhosis), sustained virological response (SVR), quality of life, all-cause mortality in the intervention group and the comparison group.
GRADE process: The quality of the body of evidence for each outcome was assessed using Grading of Recommendations Assessment, Development and Evaluation (GRADE) methodology. GRADE rates the quality of evidence for each outcome of interest as high, moderate, low, or very low, depending on a number of criteria, including study design, study quality, study consistency (the similarity of estimates of effect across studies) and study directness (the extent to which the evidence is relevant to the population, intervention, and outcome of interest).
Population denominators: Two different population denominators were used: The population eligible for testing. This varied across studies, from the number of people residing in the region where the intervention took place, to the number of people attending a specific service who were exposed to the intervention. The population who were HCV antibody positive. This was estimated using study-specific HCV antibody prevalence (the number of positive tests/total number of tests conducted in the intervention group within each study). The rationale for using those tested in the intervention group to estimate prevalence was that: o In most cases, there was more HCV testing in the intervention group, therefore providing a better estimate of HCV prevalence o HCV testing in the comparison group was more likely to be due to symptomatic testing, and would therefore lead to an over-estimation of HCV prevalence o For a small number of studies, HCV prevalence in the comparison group was not available. Sensitivity analyses were conducted to assess the effect of changing the population denominator (see below).
P a g e |7
PICO 1 (Testing): Targeted HCV antibody testing interventions
Data synthesis: Calculation of relative risks Random effects meta-analyses were carried out to calculate pooled effect sizes for: being tested for HCV antibodies, testing positive for HCV antibodies, being referred to a specialist, attending a specialist appointment, commencing treatment, achieving an SVR, and all-cause mortality, in populations who were exposed to a HCV testing intervention, compared to populations that were not exposed.
Investigation of heterogeneity: Subgroup meta-analyses were carried out to calculate pooled estimates of: - The pooled relative risk of each outcome among those who were exposed to a ‘direct’ intervention (health or social care service provided with in-practice support to offer risk assessment and HCV testing) compared to populations exposed to a ‘general’ intervention (invitations to information sessions for care providers or people at risk, leaflets or posters on HCV testing, TV/radio awarenessraising campaigns). For outcomes where there were only a small number of studies (referral, attendance, treatment, SVR, and all-cause mortality), results were presented for all studies combined rather than in subgroups.
Sensitivity analyses were carried out as follows: -To assess the effect of using different denominators (the estimated HCV antibody positive population versus the population eligible for testing) on the pooled effect size. -To assess the effect of different study designs (RCTs versus non-RCTs) on the pooled effect size. For outcomes where sensitivity analysis suggested that the inclusion of non-RCT evidence might overestimate the pooled effect size, the analysis was restricted to RCTs only.
Further investigation of heterogeneity was considered inappropriate, given the small number of studies included in the review.
Data synthesis: Calculation of absolute risks GRADE profiler was used to calculate the anticipated absolute effects of HCV testing interventions among the HCV antibody positive population, and among two hypothetical populations with HCV prevalence of 10% and 50%.
P a g e |8
PICO 1 (Testing): Targeted HCV antibody testing interventions
RESULTS The literature search identified a total of 10,024 citations (Figure 1). Initial review excluded 3730 duplicates and 4471 irrelevant citations. Subsequent screening by title and abstract excluded a further 1544 citations which did not meet population, intervention, comparison, and at least one outcome criteria. Two hundred and seventy-nine abstracts advanced to full text review, of which 13 full text articles directly met the PICO criteria. A further 3 full text articles indirectly met the PICO criteria, in that they provided indirect evidence related to one outcome (HCV transmission) that was not covered by the direct evidence.
Direct evidence Of 13 full text articles meeting the PICO, 3 articles (Helsper 2012, Litwin 2012, Sahajian 2011) reported on two different studies within the same article; therefore there were 16 studies in total. Characteristics of each of the 16 studies are described in Table 1. There were five cluster randomised controlled trials (RCT) (Cullen et al., 2006, Hickman et al., 2007, Roudot-Thoraval et al., 2000, Sahajian et al., 2011a, and Sahajian et al.,2011b), four non-randomised controlled trials (Anderson et al., 2006, Cullen et al., 2012, Helsper et al., 2010, Lewis et al., 2012), three before/after studies (Litwin et al., 2012a, Litwin et al., 2012b, Lacey et al., 2007), and four time series analyses (Defossez et al., 2008, Helsper et al., 2012a, Helsper et al., 2012b, Sahajian et al., 2004). There were twelve studies that reported on direct HCV testing interventions (Anderson et al., 2009, Cullen et al., 2006, Cullen et al., 2012, Helsper et al., 2010, Helsper et al., 2012b, Hickman et al., 2008, Lacey et al., 2007, Lewis et al., 2012, Litwin et al., 2012a, Litwin et al., 2012b, Sahajian et al., 2011a, Sahajian et al., 2011b) and four studies that reported on general HCV testing interventions (Defossez et al., 2008, Helsper et al., 2012a, Roudot-Thouraval et al., 2000, Sahajian et al., 2004).
Eight studies compared a population exposed to a HCV testing intervention and a population in a different setting or region (e.g. another health centre or area of residence) during the same time period, who were not exposed to the intervention (Anderson et al., 2006, Cullen et al., 2006, Cullen et al., 2012, Helsper et al., 2010, Hickman et al., 2008, Roudot-Thoraval et al., 2000, Sahajian et al., 2011a, Sahajian et al., 2011b). Seven studies compared a population exposed to a HCV testing intervention and a population in the same setting or region prior to the rollout of the intervention (Defossez et al., 2008, Helsper et al., 2012a, Helsper et al., 2012b, Lacey et al., 2007, Litwin et al., 2012a, Litwin et al., 2012b, Sahajia et al., 2004). One study (Lewis et al., 2012) compared a ‘direct’ intervention (phone and letter invitations to participate in the HCV testing intervention) to a
P a g e |9
PICO 1 (Testing): Targeted HCV antibody testing interventions
‘general’ intervention (leaflets and posters distributed to local mosques). A sensitivity analysis including and excluding Lewis et al.,2012 had minimal impact on the pooled effect size (ES) (Appendix IV). Therefore, the Lewis et al study was included in the meta-analyses, and was assigned to the ‘direct’ intervention subgroup.
Data synthesis Results of sensitivity analyses: The use of different population denominators (the estimated HCV antibody positive population versus the population eligible for testing) had minimal impact on pooled effect size (Appendix II). Therefore, the population denominator that was most clinically appropriate to each outcome was selected, as follows: For the number of HCV antibody tests, and the number of positive HCV antibody tests, the denominator selected was the population eligible for testing. For all outcomes (referral/attendance at specialist appointment, treatment, SVR) where HCV antibody positivity was a pre-requisite for achieving that outcome, the denominator selected was the estimated HCV antibody positive population.
Results of the sensitivity analysis by study design are shown in Appendix III. For two outcomes (referral to specialist, and attendance at specialist), the inclusion of non-RCT evidence potentially over-estimated the effect size, and therefore the meta-analyses for these outcomes were restricted to RCT evidence.
Number of HCV antibody tests: HCV testing interventions were associated with increased testing for HCV (effect size [ES] 2.90, 95% confidence interval [CI] 2.01, 4.17, I2 = 100%) compared to no testing intervention. Direct testing strategies were more successful than general testing strategies in increasing HCV testing uptake (ES 3.47 95% CI 2.52, 4.79, I2 = 94% and ES 1.47, 95% CI 0.71, 3.03, I2 =100% respectively).
Number of HCV antibody positive cases detected: HCV testing interventions were associated with increased detection of HCV antibody positive cases (ES 1.67, 95% CI 1.28, 2.20, I2= 76%). Direct testing strategies were more successful than general testing strategies in increasing the number of HCV antibody positive cases detected (ES 2.31, 95% CI 1.47, 3.62, I2 =79%, and RR 1.26, 95% CI 0.97, 1.64, I2 =58% respectively).
P a g e | 10
PICO 1 (Testing): Targeted HCV antibody testing interventions
Number referred to and attending a specialist for HCV: HCV testing interventions were associated with increased referral to a specialist (ES 3.01, 95% CI 1.79, 5.07, I2= N/A, 1 study), and increased attendance at specialist appointments (ES 3.66, 95% 1.92, 6.99, I2 = N/A, 1 study), compared to no testing intervention.
Number commencing treatment and/or achieving a Sustained Virological Response (SVR): HCV testing interventions were associated with a non-significantly increased chance of commencing treatment (ES 3.02, 95% CI 0.90, 10.15, I2= 0%), and an equal chance of achieving an SVR (ES 1.34, 95% CI 0.25, 7.11, I2=0%), compared to no HCV testing intervention.
All-cause mortality: There was no significant difference in all-cause mortality between populations who were exposed to an HCV testing intervention, and populations that were not (ES 0.89, 95% CI 0.06, 13.95, I2 = N/A, 1 study).
Absolute effects of HCV testing interventions: The anticipated absolute effects of HCV testing interventions are shown in Tables 2a and 2b. The anticipated absolute effects of HCV testing interventions among two hypothetical populations (with HCV prevalence of 10% and 50%), are shown in Table 2c.
Indirect evidence Number of cases of HCV transmission: There were no studies that compared HCV transmission among populations exposed or not exposed to a HCV testing intervention. However, three studies provided indirect evidence on injecting behaviour after receiving a test for HCV. Tsui et al. (2009) prospectively examined a cohort of 112 young PWID after HCV seroconversion. There was no significant change in self-reported injecting, syringe lending, or paraphernalia sharing at 12 months post seroconversion (p=0.75, p=0.07, p=0.45 respectively). Ompad et al. (2002) compared injecting behaviour among 104 young PWID who tested either HCV antibody negative or HCV antibody positive. At six month follow-up, there was no significant difference in self-reported needle, drug cooker, cotton filters, or rinse water sharing between those who tested HCV antibody positive and those who tested HCV antibody negative (p=0.83, p=0.42, p=0.96, p=0.57 respectively).
P a g e | 11
PICO 1 (Testing): Targeted HCV antibody testing interventions
Aitken et al. (2002) prospectively followed 47 PWID who received peer outreach testing and counselling. Of 20 PWID who returned for follow-up (mean 3 months), the number who reported receptive injecting equipment sharing decreased from five to two, and mean use of new needles/syringes increased from 60% to 84%.
CONCLUSIONS HCV testing interventions were effective in increasing the number of HCV antibody positive cases detected, and the number of attendances and referrals to specialist care. Testing interventions were associated with a borderline significant increase in HCV treatment commencement, but this was not reflected in an increased chance of SVR. This may be due to the short period of follow-up used by most of the studies.
Implications for practice Direct HCV testing strategies were more effective than general strategies in increasing the number of tests, and in detecting HCV antibody positive cases. However, decisions regarding which testing strategy to use should also take into account the relative workload and costs involved in direct compared to general strategies.
Implications for research There is a lack of evidence for the effectiveness of HCV testing interventions with regards to treatment outcomes and HCV-related morbidity and mortality. Further research in this area should focus on the longer-term outcomes of testing interventions for HCV.
P a g e | 12
PICO 1 (Testing): Targeted HCV antibody testing interventions
FIGURES
Figure 1: Flowchart for the Systematic Review
P a g e | 13
PICO 1 (Testing): Targeted HCV antibody testing interventions
Figure 2: Forest plot of comparison: HCV testing intervention versus no intervention, outcome: HCV antibody tests among those eligible for testing
Note: sensitivity analyses investigating effect size (ES) by study design produced the following pooled estimates: Restricted to RCT: ES 2.70 (95% CI 1.44, 5.05) I2 = 97%; restricted to non-RCTs: 3.08 (95% CI 1.90, 4.98) I2 = 100%. Given the similarity in pooled effect sizes and heterogeneity in each group, all study designs were included for this outcome.
P a g e | 14
PICO 1 (Testing): Targeted HCV antibody testing interventions
Figure 3: Forest plot of comparison: HCV testing intervention versus no intervention, outcome: HCV antibody positive cases detected among those eligible for testing
Note: sensitivity analyses investigating effect size (ES) by study design produced the following pooled estimates: Restricted to RCT: ES 1.58 (95% CI 1.14, 2.18) I2 = 42%; restricted to non-RCTs: 1.85 (95% CI 1.23, 2.79) I2 = 84%. Given the similarity in pooled effect sizes and heterogeneity in each group, all study designs were included for this outcome.
P a g e | 15
PICO 1 (Testing): Targeted HCV antibody testing interventions
Figure 4: Forest plot of comparison: HCV testing intervention versus no intervention, outcome: referrals to a specialist among the estimated HCV antibody positive population
Note: sensitivity analyses investigating effect size (ES) by study design produced the following pooled estimates: All studies: ES 6.28 (95% CI 1.81, 21.76) I2 = 67%; Restricted to RCT: ES 3.01 (95% CI 1.79, 5.07) I2 =N/A, 1 study; restricted to non-RCTs: 12.06 (95% CI 4.07, 35.71) I2 = 0%. Given that the RCT evidence was judged to be high quality, and that the inclusion of non-RCT evidence risked over-estimating the effect size, the analysis for this outcome was restricted to RCT evidence only.
Figure 5: Forest plot of comparison: HCV testing intervention versus no intervention, outcome: attendance at specialist appointment among the estimated HCV antibody positive population
Note: sensitivity analyses investigating effect size (ES) by study design produced the following pooled estimates: All studies: ES 6.30 (95% CI 2.24, 17.71) I2 = 42%; Restricted to RCT: ES 3.66 (95% CI 1.92, 6.99) I2 = N/A; restricted to non-RCTs: 13.03 (95% CI 3.61, 47.00) I2 = 0%. Given that the RCT evidence was judged to be high quality, and that the inclusion of non-RCT evidence risked overestimating the effect size, the analysis for this outcome was restricted to RCT evidence only.
P a g e | 16
PICO 1 (Testing): Targeted HCV antibody testing interventions
Figure 6: Forest plot of comparison: HCV testing intervention versus no intervention, outcome: commencing treatment among the estimated HCV antibody positive population
Note: sensitivity analyses investigating effect size (ES) by study design produced the following pooled estimates: Restricted to RCT: ES 4.45 (95% CI 0.53, 37.12) I2 = N/A, 1 study; restricted to non-RCTs: 2.50 (95% CI 0.57, 10.96) I2 = 0%. Given the similarity in pooled effect sizes across the two groups, all studies were included in the analysis for this outcome.
Figure 7: Forest plot of comparison: HCV testing intervention versus no intervention, outcome: achieving SVR among the estimated HCV antibody positive population
Note: non-RCT evidence only was available for this outcome.
P a g e | 17
PICO 1 (Testing): Targeted HCV antibody testing interventions
Figure 8: Forest plot of comparison: HCV testing intervention versus no intervention, outcome: all-cause mortality among the estimated HCV antibody positive population*
*No deaths were observed, so the relative risk calculation approximates events to =1 in both the intervention and comparison arms. Note: RCT evidence only was available for this outcome
P a g e | 18
PICO 1 (Testing): Targeted HCV antibody testing interventions
TABLES Table 1: Direct evidence STUDY PARTICIPANTS
Anderson et al., 2006 Setting and location: Two General Practices in Scotland, UK Intervention period: November 2003 – April 2004 Population: The eligible population was 2079 individuals aged 30-54 years registered with two General Practices serving an area of high socio-economic deprivation. Inclusion/exclusion criteria: Individuals with an existing diagnosis of HCV, or where offering a test was not clinically appropriate, were excluded. Study population characteristics: 53% male 12.8% among those tested in the intervention group Study design and duration: Non-randomised controlled trial, four years of follow-up Loss to follow-up: 7% of those tested in the intervention group People attending non-urgent GP appointments were offered an HCV test and given an information leaflet. Those accepting the offer could immediately attend the testing and counselling appointment, or return at a later date. Routine venepuncture or oral fluid testing were offered. Individuals testing HCV RNA positive were referred to a single Hepatology Unit for further management. Individuals attending a comparison GP practice (which was situated within the same Health Centre building) were not exposed to the intervention, and received routine care.
HCV ANTIBODY PREVALENCE METHODS
INTERVENTIONS
OUTCOMES
Of 1165 eligible patients in the intervention group, 117 were tested, 15 were HCV antibody positive, 11 attended a specialist appointment, and 1 achieved an SVR. Of 914 eligible patients in the comparison group, 0 were tested. While non-targeted HCV screening in the general practice setting can detect infected former people who inject drugs (PWID), the low diagnostic yield among non-PWID limited the effectiveness of the intervention. The low uptake of treatment among chronically infected persons four years after diagnosis demonstrates the difficulties in clinically managing these individuals. Sequence generation? Not randomised Allocation concealment? No Blinding? Outcome is explicitly defined, and therefore the lack of blinding was unlikely to affect outcome Incomplete outcome data addressed? Yes Free of selective reporting? Yes Free of other bias? Yes HIGH RISK OF BIAS
HEADLINE FINDING OF STUDY
COCHRANE RISK OF BIAS ASSESSMENT
SUMMARY
P a g e | 19
PICO 1 (Testing): Targeted HCV antibody testing interventions
STUDY PARTICIPANTS
Cullen et al., 2006 Setting and location: 25 General Practices in Ireland Intervention period: Not stated Population: 196 individuals receiving methadone treatment from their GP. Inclusion/exclusion criteria: A small number of patients where consent could not be obtained to access medical notes were excluded. Study population characteristics: 72 % male 70.2% among the intervention group. It was not possible to calculate prevalence among those tested in the intervention group, because the total number of tests was not recorded. Study design and duration: Cluster randomised controlled trial, six months duration Loss to follow-up: None – data collection was from GP records Liaison support nurse discussed screening practice guidelines with all practice staff, provided clinical and administrative support, liaised with specialist hepatology and addiction services, and carried out counselling and testing at practices. Control practices continued with routine care. Of 104 eligible patients in the intervention group, 51 were tested within the GP practice (plus an unknown number who were tested at locations other than their GP practice and were referred back to their GP if positive), 73 in total tested positive, 44 were referred, 37 attended a specialist appointment, of which 5 commenced HCV treatment. Of 92 eligible patients in the comparison group, 25 were tested within the GP practice (plus an unknown number who were tested at locations other than their GP practice and were referred back to their GP if positive), 41 in total tested positive, 13 were referred, 9 attended a specialist appointment, of which 1 commenced HCV treatment. General Practice has an important role in the care of people at risk of hepatitis C, and when appropriately supported can effectively implement current best practice. Sequence generation? Yes Allocation concealment? Yes Blinding? Yes Incomplete outcome data addressed? Yes Free of selective reporting? Yes Free of other bias? Yes LOW RISK OF BIAS
HCV ANTIBODY PREVALENCE
METHODS
INTERVENTIONS
OUTCOMES
HEADLINE FINDING OF STUDY
COCHRANE RISK OF BIAS ASSESSMENT
SUMMARY
P a g e | 20
PICO 1 (Testing): Targeted HCV antibody testing interventions
STUDY PARTICIPANTS
Cullen et al., 2012 Setting and location: 16 General Practices in Scotland, UK Intervention period: February – October 2007 Population: Estimated 1013 individuals aged 30-54 years (with GP records that suggested previous history of PWID) who attended 16 GPs serving an area of high socio-economic deprivation during the intervention period. Inclusion/exclusion criteria: individuals for whom offering a test was not clinically appropriate were excluded. Study population characteristics: 51% men 70.5% among those tested in the intervention group Study design/duration: Non-randomised controlled trial, 3 years of follow-up Loss to follow-up: effectively zero, as follow-up information was derived from routine data Eligible individuals attending non-urgent appointments were offered an HCV test and given an information leaflet. Participants returned to the practice to receive their results and post-test discussion from their GP. One GP practice received a staff seminar, and the remaining 7 GPs received HCV information. The control practices continued routine practice, and were not aware of their participation in the trial. Of 485 eligible patients in the intervention group, 105 were tested, 74 were HCV antibody positive, 43 were PCR positive, 31 were referred, 22 attended a specialist appointment, 4 commenced treatment and 2 achieved an SVR. Of an estimated 528 eligible patients in the comparison group, 36 were tested, 8 were HCV antibody positive, 5 were PCR positive, 3 were referred, 2 attended, 2 commenced treatment and 2 achieved an SVR. Targeted case-finding in primary care achieved higher test uptake and diagnosis rates. However, to optimise diagnosis and referral of chronically infected individuals, alternative means of diagnosis (e.g. dried blood spots) and retention in care (e.g. outreach services) need to be explored. Sequence generation? No Allocation concealment? No Blinding? No, but unlikely to affect outcome Incomplete outcome data addressed? Yes Free of selective reporting? Yes Free of other bias? Yes HIGH RISK OF BIAS
HCV ANTIBODY REVALENCE METHODS
INTERVENTIONS
OUTCOMES
HEADLINE FINDING OF STUDY
COCHRANE RISK OF BIAS ASSESSMENT
SUMMARY
P a g e | 21
PICO 1 (Testing): Targeted HCV antibody testing interventions
STUDY PARTICIPANTS
Defossez et al., 2008 Setting and location: Poitou-Charentes Region, France Intervention period: June 1999 onwards Population: People residing in a single region of France, population 1.5 million Inclusion/exclusion criteria: People who had tested HCV antibody positive prior to the intervention were excluded from the numerator 1.5% among those tested in the intervention population Study design and duration: Time series analysis comparing pre-intervention (Nov- Dec 1997) with post-intervention (Feb-May 2003). Test data were obtained from >80% of HCV testing laboratories in the region. HCV clinical management data were obtained from questionnaires to the GP or specialist who had requested HCV positive tests. The questionnaire response rate was 6070%. The clinical management questionnaire was carried out six to twelve months after the original laboratory survey. Loss to follow-up: N/A: series of cross-sectional studies National anti-hepatitis C programme commenced in June 1999, which included implementation of a targeted screening programme, and improved care for HCV infected patients. The comparison group is the population of the same region prior to the rollout of the programme. Of a population of 1,677,855 post-intervention, 20,920 were tested, 307 were HCV antibody positive. Survey data were available for 216 HCV antibody positive patients. Of these, 96 were newly diagnosed during 2003, of whom 14 commenced treatment, and 6 died. Of a population of 1,640,453 pre-intervention, 6168 were tested and 196 were HCV antibody positive. Survey data were available for 130 HCV antibody positive patients. Of these, 69 were newly diagnosed during 1997, of whom 8 commenced treatment, and 3 died. To account for the change in denominator between the laboratory-based data and the survey responses for the newly diagnosed individuals, a proportion (96/307) of the area population was used for the intervention, and 69/196 for the comparison group. This equated to 524,671 in the intervention group, and 577,506 in the control group.
HCV ANTIBODY PREVALENCE
METHODS
INTERVENTIONS
OUTCOMES
HEADLINE FINDING OF STUDY
National Campaigns targeting the general public and healthcare professionals seem to have had no impact on patient management: in particular, drug users still do not receive adequate follow-up. Sequence generation? No Allocation concealment? No Blinding? No, but unlikely to affect outcome Incomplete outcome data addressed? Yes, good coverage of regional data Free of selective reporting? Yes Free of other bias? It is not possible to know what proportion of the ‘intervention group’ was actually exposed to the intervention. Further, the intervention survey period was 4 months, compared to 2 months for the comparison survey period. However, it was unclear from the text whether this was done to correct for seasonal variations in testing. A sensitivity analysis was carried out to assess the impact of including and excluding the study. Including Defossez impacted on the effect size for the ‘commence treatment’ and ‘allcause mortality’ outcomes, and therefore was removed from these analyses.
COCHRANE RISK OF BIAS ASSESSMENT
P a g e | 22
PICO 1 (Testing): Targeted HCV antibody testing interventions
SUMMARY STUDY PARTICIPANTS
HIGH RISK OF BIAS Helsper et al., 2010 Setting and location: General Practices in two regions of the Netherlands Intervention period: October 2007-January 2008 Population: People at increased risk of HCV attending study GP practices were targeted for the intervention. However, the denominator provided by the authors is the total population resident in the intervention and comparison regions (540,000) rather than the target population. Inclusion/exclusion criteria: GPs linked to shelters for drug and alcohol use were excluded. Study population characteristics: N/A 1.7% among those tested in the intervention group Study design and duration: Non randomised controlled trial, intervention period lasted four months. Loss to follow-up: Effectively none, due to use of routine data The intervention comprised a support campaign for GPs, which included information, education sessions, and in-practice support from practice facilitators to carry out HCV risk assessment. Control practices continued with routine care. A public campaign to raise awareness of HCV, consisting of radio and newspapers advertisements, was carried out in both intervention and control regions during the same time period. Of a population of 269,125 residing in the intervention area, 172 were tested, and 3 were positive. Of a population of 266,678 residing in the comparison area, 118 were tested, and 1 was positive. A campaign of support for primary care was cost-effective. Sequence generation? No Allocation concealment? No Blinding? No, but unlikely to affect outcome Incomplete outcome data addressed? Yes Free of selective reporting? Yes Free of other bias? Yes HIGH RISK OF BIAS
HCV ANTIBODY PREVALENCE METHODS
INTERVENTIONS
OUTCOMES
HEADLINE FINDING OF STUDY
COCHRANE RISK OF BIAS ASSESSMENT
SUMMARY
P a g e | 23
PICO 1 (Testing): Targeted HCV antibody testing interventions
STUDY PARTICIPANTS
Helsper et al., 2012a Setting and location: A single region of the Netherlands Intervention period: October 2007 – January 2008 Population: General population and at-risk groups residing in the Gelre-IJssel region, population 166,315 Inclusion/exclusion criteria: N/A Study population characteristics: N/A
HCV ANTIBODY PREVALENCE
0% among those tested in the intervention group. In order to estimate the total HCV antibody positive population, a prevalence of 1% among the study population was assumed. Study design and duration: Time series analysis (plus a cost-effectiveness study based on the results of the time series analysis). Loss to follow-up: N/A Local broadcasting of radio advertisements, publishing advertisements in newspapers, distribution of specially designed posters and brochures in public areas where risk groups were expected to congregate. The comparison group were the population living in the same region prior to rollout of the intervention. Of an estimated population of 166,315 residing in the area post-intervention, 118 were tested, and 0 were positive. Of an estimated population of 166,315 residing in the area pre-intervention, 86 were tested, and 0 were positive. A campaign aimed at the general public without support for primary care did not improve case-finding and was therefore not cost-effective. Sequence generation? No Allocation concealment? No Blinding? No, but unlikely to affect outcome Incomplete outcome data addressed? Yes Free of selective reporting? Yes Free of other bias? No – it is not possible to know what proportion of the ‘intervention group’ were actually exposed to the intervention HIGH RISK OF BIAS
METHODS
INTERVENTIONS
OUTCOMES
HEADLINE FINDING OF STUDY
COCHRANE RISK OF BIAS ASSESSMENT
SUMMARY
P a g e | 24
PICO 1 (Testing): Targeted HCV antibody testing interventions
STUDY PARTICIPANTS
Helsper et al., 2012b Setting and location: Drug services in Rotterdam, the Netherlands Intervention period: May 2007 – September 2008 Population: Current or former ‘hard drug users’ (HDUs): total population of HDUs in geographical area of intervention was estimated at 5000 Inclusion/exclusion criteria: N/A Study population characteristics: N/A
HCV PREVALENCE METHODS
30.6% among those tested in the intervention group Study design and duration: Time series analysis Loss to follow-up: Effectively none, due to use of routine data 26 addiction care professionals were trained to provide HCV counselling, and three information meetings were held that were attended by 180 HDUs. The comparison group were the same population prior to the rollout of the intervention. Of a population of 5000 HDUs, 186 were tested, and 57 were HCV antibody positive during the intervention. Of a population of 5000 HDUs, there was zero testing prior to the intervention. This was an assumption based on a review of Drug Services practice, as well as local Health Service Organization records. Because the number of tests was based on an assumption, no further assumptions (e.g. the number of positive tests) were made about the comparison arm. A campaign aimed at hard drug users was cost-effective. Sequence generation? No Allocation concealment? No Blinding? No, but unlikely to affect outcome Incomplete outcome data addressed? Yes Free of selective reporting? Yes Free of other bias? No – it is not possible to know what proportion of the ‘intervention group’ were actually exposed to the intervention HIGH RISK OF BIAS
INTERVENTIONS
OUTCOMES
HEADLINE FINDING OF STUDY
COCHRANE RISK OF BIAS ASSESSMENT
SUMMARY
P a g e | 25
PICO 1 (Testing): Targeted HCV antibody testing interventions
STUDY PARTICIPANTS
Hickman et al., 2007 Setting and location: 14 specialist drug clinics and 6 prisons in the UK Intervention period: June 2004 – June 2005 Population: 12,350 prison inmates or drug users attending specialist drug clinics Inclusion/exclusion criteria: N/A Study population characteristics: N/A 27.3% among those tested in the intervention group Study design and duration: Cluster randomised controlled trial Loss to follow-up: Effectively none, because of use of laboratory data HCV testing using dried blood spot (DBS) testing. Half day introduction to the intervention that included staff training and information on DBS, plus on-going support from local specialist HCV nurses. The comparison group (matched prisons or drug services) received routine care. Of a population of 6550 in the intervention group, 791 were tested, and 216 were HCV antibody positive. Of a population of 5800 in the comparison group, 243 were tested, and 104 were HCV antibody positive. This study provides preliminary evidence that DBS may increase the uptake of HCV diagnostic testing, by increasing the opportunities for patients to be offered testing. Sequence generation? Yes Allocation concealment? Yes Blinding? No, but unlikely to affect outcome Incomplete outcome data addressed? Yes Free of selective reporting? Yes Free of other bias? Yes LOW RISK OF BIAS
HCV ANTIBODY PREVALENCE
METHODS
INTERVENTIONS
OUTCOMES
HEADLINE FINDING OF STUDY
COCHRANE RISK OF BIAS ASSESSMENT
SUMMARY
P a g e | 26
PICO 1 (Testing): Targeted HCV antibody testing interventions
STUDY PARTICIPANTS
Lacey et al., 2007 Setting and location: Inpatient psychiatry unit in a tertiary hospital, Australia Intervention period: August 2002 – January 2003 Population: 832 psychiatric inpatients admitted to a single psychiatric unit during the intervention and comparison periods. Inclusion criteria: Diagnosis of serious mental disorder, able to provide informed consent, inpatient stay of more than two days. Exclusion criteria: Known HCV infection Study population characteristics: N/A 19.7% among those tested in the intervention group Study design and duration: Before/after study (cohort) Loss to follow-up: None – use of laboratory data During the intervention period, a leaflet providing information on HCV was distributed, and a research assistant facilitated education/discussion groups, and carried out counselling and testing. The comparison group comprised psychiatric inpatients admitted during the same length time period (Feb 2002July 2002) prior to the rollout of the intervention. Of 402 individuals in the intervention group, 71 were tested, and 14 were HCV antibody positive and referred to Infectious Diseases. Of 430 individuals in the comparison group, 40 were tested. An education and counselling programme can increase HCV testing among psychiatric inpatients. Sequence generation? Not randomised Allocation concealment? No Blinding? No, but unlikely to affect outcome Incomplete outcome data addressed? Yes Free of selective reporting? Yes Free of other bias? Yes HIGH RISK OF BIAS
HCV ANTIBODY PREVALENCE METHODS
INTERVENTIONS
OUTCOMES
HEADLINE FINDING OF STUDY
COCHRANE RISK OF BIAS ASSESSMENT
SUMMARY
P a g e | 27
PICO 1 (Testing): Targeted HCV antibody testing interventions
STUDY PARTICIPANTS
Lewis et al., 2012 Setting and location: Mosques and GP practices serving the Pakistani population in the UK Intervention period: Not available Population: 2297 Pakistani patients attending local mosques in London, UK Inclusion/exclusion criteria: Not available Study population characteristics: Not available 4.9% among those tested in the intervention group Study design and duration: Non-randomised controlled trial Loss to follow-up: None Patients were contacted by letter and phone to invite them to attend screening clinics at their GP surgery. This was compared to an opportunistic approach where leaflets were distributed to mosques advising them to attend clinics for testing. The inclusion/exclusion of this study in the meta-analyses was tested in the sensitivity analysis. Of a population of 1134 in the intervention group, 223 were tested, and 11 were HCV antibody positive. Of a population of 1163 in the control group, 17 were tested, and 0 were HCV antibody positive. Community awareness campaigns do not directly lead to testing for viral hepatitis in at risk immigrant groups. A direct screening approach is more effective than an opportunistic screening approach in primary care. Sequence generation? No Allocation concealment? No Blinding? No, but unlikely to affect outcome Incomplete outcome data addressed? Yes Free of selective reporting? Yes Free of other bias? Yes HIGH RISK OF BIAS
HCV ANTIBODY PREVALENCE METHODS
INTERVENTIONS
OUTCOMES
HEADLINE FINDING OF STUDY
COCHRANE RISK OF BIAS ASSESSMENT
SUMMARY
P a g e | 28
PICO 1 (Testing): Targeted HCV antibody testing interventions
STUDY PARTICIPANTS
Litwin et al., 2012a Setting and location: Three community-based primary care clinics located in urban areas of New York, USA Intervention period: Intervention ran from November 2008 to March 2009 Population: 15,572 patients who attended one of the study practices during the intervention or comparison period. Inclusion criteria: Age > 18 years Exclusion criteria: previously tested for HCV antibodies Study population characteristics: 25% male, Race/ethnicity: 30% black 5.3% among those tested in the intervention group Study design and duration: Before/after study (cohort), ‘before’ period was two months, and ‘after’ period was five months. The difference in time periods in the two arms was accounted for in the denominator (the number of eligible individuals attending the practice during the study period). Loss to follow-up: None (use of routine data) Risk based screening facilitated by providing a 'risk sticker' on notes, prompting physician to ask about HCV risk factors and to offer testing if a risk factor was present. The comparison group were individuals attending the practice prior to the intervention who received routine care. Of 8981 individuals in the intervention group, 1179 were tested, and 62 were HCV antibody positive. Of 6591 individuals in the intervention group, 394 were tested, and 36 were HCV antibody positive. A ‘risk-based’ strategy increased the number of tests and the number of positive tests. Sequence generation? No Allocation concealment? No Blinding? No, but unlikely to affect outcome Incomplete outcome data addressed? Yes Free of selective reporting? Yes Free of other bias? Yes HIGH RISK OF BIAS
HCV ANTIBODY PREVALENCE METHODS
INTERVENTIONS
OUTCOMES
HEADLINE FINDING OF STUDY
COCHRANE RISK OF BIAS ASSESSMENT
SUMMARY
P a g e | 29
PICO 1 (Testing): Targeted HCV antibody testing interventions
STUDY PARTICIPANTS
Litwin et al., 2012b Setting and location: Three community-based primary care clinics located in urban areas of New York, USA Intervention period: Intervention ran from March-June 2009 Population: 16,756 patients who attended one of the study practices during the intervention or comparison period. This population refers to all patients who attended the clinics at least once during the intervention and comparison periods, not the number of patients who were in the correct age category. Inclusion criteria: Age > 18 years Exclusion criteria: record of previous test for HCV antibodies in medical notes. Study population characteristics: 25% male, Race/ethnicity: 30% black 5.9% among those tested in the intervention group Study design and duration: Before/after study (cohort): ‘before’ period was two months, and ‘after’ period was four months. Loss to follow-up: None (use of routine data) Birth cohort-based screening, facilitated by adding a ‘Birth cohort’ reminder sticker on the top of each set of progress notes, asking GPs to offer HCV testing to all patients born between 1945 and 1964. The comparison group were individuals attending the practice in a period prior to the intervention. Of 10,165 patients in the intervention group, 1008 were tested, and 59 were HCV antibody positive during the intervention. Of 6591 patients in the comparison group, 394 were tested, and 36 were HCV antibody positive during the intervention. A birth cohort strategy increased the number of tests, and the number of positive tests. Sequence generation? No Allocation concealment? No Blinding? No, but unlikely to affect outcome Incomplete outcome data addressed? Yes Free of selective reporting? Yes Free of other bias? Yes. Note: the birth cohort screening took place after the risk-based screening in the same practices (Litwin et al., 2012a). However, the authors carried out further sensitivity analyses to investigate whether this would have affected the outcomes of the birth cohort intervention. These sensitivity analyses were non-significant. HIGH RISK OF BIAS
HCV ANTIBODY PREVALENCE METHODS
INTERVENTIONS
OUTCOMES
HEADLINE FINDING OF STUDY
COCHRANE RISK OF BIAS ASSESSMENT
SUMMARY
P a g e | 30
PICO 1 (Testing): Targeted HCV antibody testing interventions
STUDY PARTICIPANTS
Roudot-Thoraval et al., 2000 Setting and location: 184 General Practitioners in the Creteil region of France Intervention period: April 1997 – November 1998 Population: People attending GP appointments. Given that the only denominator information provided by the authors was the number of General Practitioners in each arm of the trial, an assumption was made that each GP had 1000 patients. Therefore, there the population of the intervention arm was estimated to be 94,000 (94 GPs), and the population of the comparison arm was estimated to be 90,000 (90 GPs). Inclusion/exclusion criteria: N/A Study population characteristics: N/A 3.4% among those tested in the intervention group Study design and duration: Randomised controlled trial Loss to follow-up: Effectively zero, as follow-up data obtained from routine sources Provision of study posters and leaflets at General Practice surgeries, informing patients of the risk factors of HCV. Comparison General Practice surgeries provided routine care. All GPs (both intervention and comparison group) were invited to an information evening on hepatitis C. Of an estimated 94,000 individuals in the intervention arm, there were 294 tests, of which 10 were HCV antibody positive. Of an estimated 90,000 individuals in the comparison arm, there were 323 tests, of which 15 were HCV antibody positive. The HCV antibody testing intervention had no impact on the number of tests, or the number of positive tests. GPs requested very few HCV tests, suggesting a low rate of identified risk factors in general practice. Sequence generation? ‘Random allocation’ Allocation concealment? Unclear Blinding? No, but unlikely to affect outcome Incomplete outcome data addressed? Yes Free of selective reporting? Yes Free of other bias? Yes LOW RISK OF BIAS
HCV ANTIBODY PREVALENCE METHODS
INTERVENTIONS
OUTCOMES
HEADLINE FINDING OF STUDY
COCHRANE RISK OF BIAS ASSESSMENT
SUMMARY
P a g e | 31
PICO 1 (Testing): Targeted HCV antibody testing interventions
STUDY PARTICIPANTS
Sahajian et al., 2004 Setting and location: 3052 General Practitioners and specialists in private practice in the Lyon area in France Intervention period: Pre-campaign period was April 1999 to March 2000, and campaign period was April 2000 to March 2001 Population: Population of region was approximately 1.5 million. The population attending GPs and private specialists was not provided. Inclusion/exclusion criteria: N/A Study population characteristics: N/A 1.7% among those tested in the intervention group Study design and duration: time series analysis, 2 years Loss to follow-up: N/A During the intervention period, a guide on screening was mailed to private practitioners, and workshops and training sessions on screening for GPs and laboratory physicians were offered. The aim of the campaign was to encourage practitioners to offer HCV tests to people at increased risk (due to a history of PWID, blood products prior to 1991, or elevated ALT levels) There was also an information campaign to the general public that included a press conference and media advertisements. The comparison group was the population of the same region prior to the screening campaign. In the intervention population (approximately 1.5 million), there were 15,952 tests, and 276 positive tests. In the comparison group, there were 13,799 tests, and 231 positive tests. The HCV antibody testing intervention increased the number of tests, and there was also a small increase in the number of positive tests. Sequence generation? No Allocation concealment? No Blinding? No, but unlikely to affect outcome Incomplete outcome data addressed? Yes Free of selective reporting? Yes Free of other bias? No – it is not possible to know what proportion of the ‘intervention group’ were actually exposed to the intervention HIGH RISK OF BIAS
HCV ANTIBODY PREVALENCE METHODS
INTERVENTIONS
OUTCOMES
HEADLINE FINDING OF STUDY
COCHRANE RISK OF BIAS ASSESSMENT
SUMMARY
P a g e | 32
PICO 1 (Testing): Targeted HCV antibody testing interventions
STUDY PARTICIPANTS
Sahajian et al., 2011a Setting and location: 12 homeless hostels and shelters providing long-term accommodation in the Lyon area in France Intervention period: October 2007 – April 2009 Population: 1852 residents of homeless hostels or shelters Inclusion/exclusion criteria: Not available Study population characteristics: Not available 3.2% among those tested in the intervention group Study design and duration: Cluster randomised controlled trial, 18-month duration Loss to follow-up: None (used routine data) In the intervention hostels and shelters, group information sessions were followed by referral, if interested, to a Health Centre where a medical check-up and HCV testing were carried out. The comparison hostels and shelters received routine care. Of 1041 individuals in the intervention group, there were 95 tests, and 3 positive tests. Of 811 individuals in the comparison group, there were 12 tests, and 0 positive tests. Health authorities must ensure widespread screening of underprivileged people, which is more effective when conducted in shelters than in screening centres (see also Sahajian 2011b). Sequence generation? Yes Allocation concealment? Yes Blinding? No, but unlikely to affect outcome Incomplete outcome data addressed? Yes Free of selective reporting? Yes Free of other bias? Yes LOW RISK OF BIAS
HCV ANTIBODY PREVALENCE METHODS
INTERVENTIONS
OUTCOMES
HEADLINE FINDING OF STUDY
COCHRANE RISK OF BIAS ASSESSMENT
SUMMARY
P a g e | 33
PICO 1 (Testing): Targeted HCV antibody testing interventions
STUDY PARTICIPANTS
Sahajian et al., 2011b Setting and location: 12 homeless hostels and shelters providing long-term accommodation in the Lyon area in France Intervention period: October 2007 – April 2009 Population: 1595 residents of homeless hostels or shelters Inclusion/exclusion criteria: Not available Study population characteristics: Not available 2.8% among those tested in the intervention group Study design and duration: Cluster randomised controlled trial, 18-month duration Loss to follow-up: Effectively zero, as use of routine data In the intervention hostels and shelters, group information sessions were followed by on-site medical check-ups and HCV testing for those who were interested. The comparison hostels and shelters received routine care. Of 784 individuals in the intervention group, there were 145 tests, and 4 positive tests. Of 811 individuals in the comparison group, there were 12 tests, and 0 positive tests. Health authorities must ensure widespread screening of underprivileged people, which is more effective when conducted in shelters than in screening centres (see also Sahajian 2011a). Sequence generation? Yes Allocation concealment? Yes Blinding? No, but unlikely to affect outcome Incomplete outcome data addressed? Yes Free of selective reporting? Yes Free of other bias? Yes LOW RISK OF BIAS
HCV ANTIBODY PREVALENCE METHODS
INTERVENTIONS
OUTCOMES
HEADLINE FINDING OF STUDY
COCHRANE RISK OF BIAS ASSESSMENT
SUMMARY
P a g e | 34
PICO 1 (Testing): Targeted HCV antibody testing interventions
Table 2a: GRADE summary of findings: testing for HCV, and HCV positive cases detected among the population eligible for testing Outcome Subgroup Eligible population (studies) GRADE quality* Effect size (95% CI) Baseline risk per 10,000 eligible population 59 tests conducted 60 tests conducted 44 tests conducted 2 cases detected 1 case detected Anticipated absolute effects per 10,000 eligible population (95% CI)
Tested for HCV
All strategies General strategy Direct strategy
7,435,283 (16 studies) 6,834,938 (4 studies) 600,345 (12 studies) 7,424,451 (14 studies) 6,834,938 (4 studies) 589,513 (10 studies)
MODERATE
a
2.90 (2.01, 4.17) 1.47 (0.71, 3.03)
112 more HCV antibody tests (from 59 more to 186 more) 28 more HCV antibody tests (from 17 fewer to 122 more) 108 more HCV antibody tests (from 66 more to 166 more) 1 more case detected (from 0 more to 2 more) 0 more cases (from 0 fewer to 1 more) 10 more cases detected (from 4 more to 20 more)
LOW
b
MODERATE
c
3.47 (2.52, 4.79) 1.67 (1.28, 2.20) 1.26 (0.97, 1.64) 2.31 (1.47, 3.62)
HCV positive cases detected
All strategies General strategy Direct strategy
MODERATE
d
MODERATE
e
MODERATE
f
8 cases detected
*Marked down for: a) inconsistency; b) inconsistency and imprecision; c) inconsistency; d) inconsistency; e) inconsistency; f) inconsistency.
Table 2b: GRADE summary of findings: management of HCV positive individuals among the estimated HCV antibody positive population Outcome (mean follow-up) Population (studies) GRADE quality* a
Effect size (95% CI)
Baseline risk per 10,000 HCV antibody positive population 2,000 referrals to specialist
Anticipated absolute effects per 10,000 HCV antibody positive population (95% CI) 4020 more referrals to a specialist (from 1580 more to 8140 more)
Referral to HCV specialist (Six months) Attendance at HCV specialist (Six months) Commenced treatment (Two years) SVR (Three years six months) Morbidity All-cause mortality (Six months) Quality of Life
138 (1 study)
MODERATE
3.01 (1.79, 5.07)
138 (1 study)
MODERATE
b
3.66 (1.92, 6.99)
1,380 attending a specialist
3683 more cases attending a specialist (from 1274 more to 8294 more) 109 more commencing treatment (from 5 fewer to 495 more)
17,263 (4 studies)
MODERATE
c
3.02 (0.90, 10.15)
54 commencing treatment
980 (2 studies)
VERY LOW
d
1.34 (0.25, 7.11)
41 achieving SVR
14 more achieving SVR (from 31 fewer to 250 more)
No data available 138 (1 study) LOW e
0.89 (0.06, 13.95)
Insufficient data
Insufficient data
No data available
*Marked down for: a) sparse data; b) sparse data; c) sparse data; d) observational studies and sparse data; e) directness and sparse data.
P a g e | 35
PICO 1 (Testing): Targeted HCV antibody testing interventions
Table 2c: Anticipated absolute effects of HCV testing interventions by population HCV prevalence Outcome GRADE quality MODERATE Subgroup HCV prevalence 10% 50% Direct strategy 10% 50% General strategy 10% 50% Referral to specialist MODERATE All strategies 10% 50% Attendance at specialist MODERATE All strategies 10% 50% Commenced treatment MODERATE All strategies 10% 50% SVR VERY LOW All strategies 10% 50% All-cause mortality LOW All strategies 10% 50% Anticipated absolute effects of HCV testing interventions per 10,000 population (95% CI) 5 more cases detected (from 2 more to 8 more) 23 more cases detected (from 9 more to 40 more) 8 more cases detected (from 3 more to 15 more) 39 more cases detected (from 14 more to 77 more) 2 more cases detected (from 0 fewer to 5 more) 10 more cases detected (from 1 fewer to 26 more) 330 more cases (from 649 more to 1370 more) 2812 more cases (from 1017 more to 5935 more) 289 more cases (from 100 more to 651 more) 1735 more cases (from 600 more to 3907 more) 13 more commencing treatment (from 1 fewer to 60 more) 39 more commencing treatment (from 2 fewer to 179 more) 0 more SVRs (from 0 fewer to 4 more) 5 more SVRs (from 11 fewer to 86 more) Insufficient data Insufficient data
HCV antibody positive cases detected
All strategies
P a g e | 36
PICO 1 (Testing): Targeted HCV antibody testing interventions
REREFENCES Aitken CK, Kerger M, Crofts N. Peer-delivered hepatitis C testing and counselling: a means of improving the health of injecting drug users. Drug and Alcohol Review, 2002; 21:33-37 Alter MJ, Hadler SC, Judson FN, Mares A, Alexander J, Hu PY, Miller JK, Moyer LA, Fields HA, Bradley DW, Margolis HS. Risk Factors for Acute Non-A, Non-B Hepatitis in the United States and Association with Hepatitis C Virus Infection. Journal of the American Medical Association, 1990; 264: 2231-2235 Anderson EM, Mandeville RP, Hutchinson SJ, Cameron SO, Mills PR, Fox R, Ahmed S, Taylor A, Spence E, Goldberg DJ. Evaluation of a general practice based hepatitis C virus screening intervention. Scottish medical journal, 2009; 54:3-7 Cullen W, Stanley J, Langton D, Kelly Y, Staines A, Bury G. Hepatitis C infection among injecting drug users in general practice: A cluster randomised controlled trial of clinical guidelines' implementation. British Journal of General Practice, 2006; 56: 848-56 Cullen BL, Hutchinson SJ, Cameron SO, Anderson E, Ahmed S, Spence E, et al. Identifying former injecting drug users infected with hepatitis C: An evaluation of a general practice-based case-finding intervention. Journal of Public Health, 2012; 34: 14-23 Defossez G, Verneau A, Ingrand I, Silvain C, Ingrand P, Beauchant M. Evaluation of the French national plan to promote screening and early management of viral hepatitis C, between 1997 and 2003: A comparative cross-sectional study in Poitou-Charentes region. European Journal of Gastroenterology and Hepatology, 2008; 20: 367-72 Grebely J, Dore GJ. What Is Killing People with Hepatitis C Virus Infection? Seminars in Liver Disease, 2011; 31:331-339 Helsper CW, van Essen GA, Bonten MJM, de Wit NJ. A support programme for primary care leads to substantial improvements in the effectiveness of a public hepatitis C campaign. Family Practice, 2010; 27:328-32 Helsper CW, Borkent-Raven BA, De Wit NJ, Van Essen GA, Bonten MJM, Hoepelman AIM, et al. Costeffectiveness of targeted screening for hepatitis C in the Netherlands. Epidemiology and Infection, 2012; 140:58-69. Hickman M, McDonald T, Judd A, Nichols T, Hope V, Skidmore S, et al. Increasing the uptake of hepatitis C virus testing among injecting drug users in specialist drug treatment and prison settings by using dried blood spots for diagnostic testing: A cluster randomized controlled trial. Journal of Viral Hepatitis, 2008; 15:250-4 Karmochine M, Carrat F, Dos Santos O, Cacoub P, Raguin G for the GERMIVIC Study Group. A casecontrol study of risk factors for hepatitis C infection in patients with unexplained routes of infection. Journal of Viral Hepatitis, 2006; 13: 775-782 Lacey C, Ellen S, Devlin H, Wright E, Mijch A. Hepatitis C in psychiatry inpatients: Testing rates, prevalence and risk behaviours. Australasian Psychiatry, 2007; 15:315-9 Lavanchy D. The global burden of hepatitis C. Liver International, 2009; 29: 74-81
P a g e | 37
PICO 1 (Testing): Targeted HCV antibody testing interventions
Lewis H, Burke K, Begum S, Ushiro-Limb I, Foster G. What is the best method of case finding for chronic viral Hepatitis in at-risk migrant communities? Journal of Hepatology, 2012; 56:S351. Litwin AH, Smith BD, Drainoni ML, McKee D, Gifford AL, Koppelman E, et al. Primary care-based interventions are associated with increases in hepatitis C virus testing for patients at risk. Digestive and Liver Disease, 2012; 44: 497-503 Ompad DC, Fuller CM, Vlahov D, Thomas D, Strathdee SA. Lack of Behavior Change after Disclosure of Hepatitis C Virus Infection among Young Injection Drug Users in Baltimore, Maryland. CID, 2002; 35: 783-788 Roudot-Thoraval F, Monnet E, Mercet P, Bastie A, Dhumeaux D, Miguet JP. Strategies of hepatitis C screening in general practice. Results of a two-center randomized trial. Gastroenterologie clinique et biologique, 2000; 24:1037-41 Sahajian F, Excler G, Bailly F, Caillat-Vallet E, Trepo C, Sepetjan M, et al. Hepatitis C screening practices among private practitioners: Impact of an information campaign. Gastroenterologie Clinique et Biologique, 2004; 28:714-9 Sahajian F, Bailly F, Vanhems P, Fantino B, Vannier-Nitenberg C, Fabry J, et al. A randomized trial of viral hepatitis prevention among underprivileged people in the Lyon area of France. Journal of Public Health, 2011;33:182-92 Tsui JI, Vittinghoff E, Hahn JA, Evans JL, Davidson PJ, Page K. Risk behaviours after hepatitis C virus seroconversion in young injection drug users in San Francisco. Drug and Alcohol Dependence, 2009; 105:160-63 Villano SA, Vlahov D, Nelson KE, Lyles CM, Cohn S, Thomas DL. Incidence and Risk Factors for Hepatitis C among Injection Drug Users in Baltimore, Maryland. Journal of Clinical Microbiology, 1997; 35: 3274-3277 World Health Organization, Fact sheet No. 164, July 2012. Available at (accessed 1st February 2013): http://www.who.int/mediacentre/factsheets/fs164/en/
P a g e | 38
PICO 1 (Testing): Targeted HCV antibody testing interventions
APPENDICIES Appendix 1: Search terms Medline and Embase 1. *Hepatitis C/ 2. HCV.ti,ab. 3. hepatitis c.ti,ab. 4. 1 or 2 or 3 5. *Mass Screening/ 6. case finding.ti,ab. 7. case identification.ti,ab. 8. case detection.ti,ab. 9. screening.ti,ab. 10. test* for hepatitis C.ti,ab. 11. test* for HCV.ti,ab. 12. hepatitis C testing.ti,ab. 13. HCV testing.ti,ab. 14. 5 or 6 or 7 or 8 or 9 or 10 or 11 or 12 or 13 15. 4 and 14 16. limit 15 to humans 17. limit 16 to yr="1994 -Current" 18. remove duplicates from 17
P a g e | 39
PICO 1 (Testing): Targeted HCV antibody testing interventions
Appendix 2: Sensitivity analyses examining the effect of denominator on pooled effect size Outcome Subgroup Denominator used Pooled effect size I2
Referrals to specialist
Total
Population eligible for testing HCV antibody positive population
6.23 (1.84, 21.09) 6.28 (1.81, 21.76) 6.23 (1.84, 21.09) 6.28 (1.81, 21.76) No studies No studies 6.26 (2.25, 17.38) 6.30 (2.24, 17.71) 6.26 (2.25, 17.38) 6.30 (2.24, 17.71) No studies No studies 3.01 (0.89, 10.15) 3.02 (0.90, 10.15) 3.01 (0.89, 10.15) 3.02 (0.90, 10.15) No studies No studies 1.34 (0.25, 7.13) 1.34 (0.25, 7.11) 1.34 (0.25, 7.13) 1.34 (0.25, 7.11) No studies No studies 0.88 (0.06, 13.94) 0.89 (0.06, 13.95) 0.88 (0.06, 13.94) 0.89 (0.06, 13.95) No studies No studies
65% 67% 65% 67% 40% 42% 40% 42% 0% 0% 0% 0% 0% 0% 0% 0% 1 study 1 study 1 study 1 study -
Direct strategy
Population eligible for testing HCV antibody positive population
General strategy
Population eligible for testing HCV antibody positive population
Attendances at specialist
Total
Population eligible for testing HCV antibody positive population
Direct strategy
Population eligible for testing HCV antibody positive population
General strategy
Population eligible for testing HCV antibody positive population
Commenced treatment
Total
Population eligible for testing HCV antibody positive population
Direct strategy
Population eligible for testing HCV antibody positive population
General strategy
Population eligible for testing HCV antibody positive population
SVR
Total
Population eligible for testing HCV antibody positive population
Direct strategy
Population eligible for testing HCV antibody positive population
General strategy
Population eligible for testing HCV antibody positive population
Deaths
Total
Population eligible for testing HCV antibody positive population
Direct strategy
Population eligible for testing HCV antibody positive population
General strategy
Population eligible for testing HCV antibody positive population
P a g e | 40
PICO 1 (Testing): Targeted HCV antibody testing interventions
Appendix 3: Sensitivity analyses examining the effect of study design on pooled effect size Outcome Number of HCV antibody tests Subgroup All strategies Study design All studies RCT Non-RCT Direct strategy All studies RCT Non-RCT General strategy All studies RCT Non-RCT Number of HCV antibody cases detected All strategies All studies RCT Non-RCT Direct strategy All studies RCT Non-RCT General strategy All studies RCT Non-RCT Referral to HCV specialist among antibody positive population Attendance at HCV specialist All strategies All studies RCT Non-RCT All strategies All studies RCT Non-RCT HCV treatment commenced All strategies All studies RCT Non-RCT SVR All strategies All studies RCT Non-RCT All-cause mortality All strategies All studies RCT Non-RCT Pooled effect size* 2.90 (2.01, 4.17) 2.70 (1.44, 5.05) 3.08 (1.90, 4.98) 3.47 (2.52, 4.79) 3.38 (1.92, 5.93) 3.73 (2.34, 5.96) 1.47 (0.71, 3.03) 0.87 (0.74, 1.02) 1.74 (0.75, 4.07) 1.67 (1.28, 2.20) 1.58 (1.14,2.18) 1.85 (1.23, 2.79) 2.31 (1.47, 3.62) 1.74 (1.47, 2.07) 3.65 (1.22, 10.93) 1.26 (0.97, 1.64) 0.64 (0.29, 1.42) 1.35 (1.09, 1.66) 6.28 (1.81, 21.76) 3.01 (1.79, 5.07) 12.06 (4.07, 35.71) 6.30 (2.24, 17.71) 3.66 (1.92, 6.99) 13.03 (3.61, 47.00) 3.02 (0.90, 10.15) 4.45 (0.53, 37.12) 2.50 (0.57, 10.96) 1.34 (0.25, 7.11) No studies 1.34 (0.25, 7.11) 0.89 (0.06, 13.95) 0.89 (0.06, 13.95) No studies I2 100% 97% 100% 94% 94% 95% 100% 1 study 100% 76% 42% 84% 79% 0% 90% 58% 1 study 48% 67% 1 study 0% 42% 1 study 0% 0% 1 study 0% 100% 0% 1 study 1 study Only one study (an RCT) reported on this outcome: use ‘all studies’. Inclusion of non-RCTs has no effect on heterogeneity and produces a more conservative effect size: include all study designs for this outcome. Only non-RCT evidence available for this outcome, therefore include non-RCTs The inclusion of non-RCTs changes the direction of the pooled effect size, but both results are non-significant: include all study designs in the data synthesis for this outcome. Inclusion of non-RCTs may over-estimate the pooled effect size and increases heterogeneity: include RCTs only in the data synthesis for this outcome. Inclusion of non-RCTs may over-estimate the pooled effect size and increases heterogeneity: include RCTs only in the data synthesis for this outcome. Comment/Decision Stratification by study design has minimal impact on effect size or heterogeneity: include all study designs in the data synthesis for this outcome. Stratification by study design has minimal impact on effect size or heterogeneity: include all study designs in the data synthesis for this outcome. The inclusion of non-RCTs changes the direction of the pooled effect size, but both results remain non-significant: include all study designs in the data synthesis for this outcome. The inclusion of non-RCTs increases heterogeneity but has minimal impact on the pooled effect size: include all study designs in the data synthesis for this outcome. The inclusion of non-RCTs has a minimal effect on the pooled effect size: include all study designs in the data synthesis for this outcome.
*The pooled effect size selected for use in the analyses is shown in bold type
P a g e | 41
PICO 1 (Testing): Targeted HCV antibody testing interventions
Appendix 4: Sensitivity analysis including/excluding specific studies Outcome Subgroup Inclusion Pooled effect size I2
Number of HCV antibody tests
All strategies
Lewis et al., 2012 included Lewis et al., 2012 excluded
2.90 (2.01, 4.17) 2.60 (1.79, 3.77) 3.47 (2.52, 4.79) 2.91 (2.18, 3.88) 1.67 (1.28, 2.20) 1.63 (1.25, 2.12) 2.31 (1.47, 3.62) 2.16 (1.39, 3.36) 2.90 (2.01, 4.17) 2.65 (1.99, 3.52) 1.67 (1.28, 2.20) 1.77 (1.25, 2.49) 2.25 (1.11, 4.56) 3.02 (0.90, 10.15) 1.84 (0.53, 6.34) 0.89 (0.06, 13.95)
100% 100% 94% 93% 76% 79% 79% 79% 100% 98% 76% 78% 0% 0% 0% 1 study
Direct strategy
Lewis et al., 2012 included Lewis et al., 2012 excluded
HCV antibody positive cases detected
All strategies
Lewis et al., 2012 included Lewis et al., 2012 excluded
Direct strategy
Lewis et al., 2012 included Lewis et al., 2012 excluded
Number of HCV antibody tests
All strategies
Defossez et al., 2008 included Defossez et al., 2008 excluded
HCV antibody positive cases detected Commenced treatment
All strategies
Defossez et al., 2008 included Defossez et al., 2008 excluded
All strategies
Defossez et al., 2008 included Defossez et al., 2008 excluded
All-cause mortality
All strategies
Defossez et al., 2008 included Defossez et al., 2008 excluded
*The pooled effect size selected for use in the analyses is shown in bold type
P a g e | 42
PICO 1 (Testing): Targeted HCV antibody testing interventions
Appendix 5: GRADE outcome assessments Testing for HCV antibodies: all strategies GRADE assessment criteria Type of evidence Decrease Serious limitation to study quality? grade if Important inconsistency? Uncertainty about directness? Imprecise or sparse data? High probability of reporting bias? Strong evidence of association? Evidence of a dose response gradient? Plausible confounders would reduce effect? Assessment 5 RCTs, 9 non-RCTs No Yes - individual study effect sizes are inconsistent, I2 = 78% No No No No No No GRADE score High Mark down Moderate GRADE running total
Increase grade if
Moderate
Testing for HCV antibodies: general strategy GRADE assessment criteria Type of evidence Decrease Serious limitation to study quality? grade if Important inconsistency? Uncertainty about directness? Imprecise or sparse data? High probability of reporting bias? Strong evidence of association? Evidence of a dose response gradient? Plausible confounders would reduce effect? Assessment One RCT and 3 non-RCTs No Yes –individual study effect sizes are inconsistent, I2 = 100% No Yes, wide confidence interval around effect size No No No No GRADE score High Mark down Moderate GRADE running total
Mark down
Low
Increase grade if
Low
Testing for HCV antibodies: direct strategy GRADE assessment criteria Type of evidence Decrease Serious limitation to study quality? grade if Important inconsistency? Uncertainty about directness? Imprecise or sparse data? High probability of reporting bias? Strong evidence of association? Evidence of a dose response gradient? Plausible confounders would reduce effect? Assessment 4 RCTs, 6 non-RCTs No Yes - individual study effect sizes are inconsistent, I2 = 94% No No No No No No GRADE score High Mark down Moderate GRADE running total
Increase grade if
Moderate
Testing HCV antibody positive: all strategies GRADE assessment criteria Type of evidence Assessment 5 RCTs, and 9 non-RCTs. Sensitivity analysis showed little impact on effect size of including non-RCTs. No Yes - individual study effect sizes are inconsistent, I2 = 76% No No No No No No GRADE score High GRADE running total
Decrease grade if
Serious limitation to study quality? Important inconsistency? Uncertainty about directness? Imprecise or sparse data? High probability of reporting bias? Strong evidence of association? Evidence of a dose response gradient? Plausible confounders would reduce effect?
Mark down
Moderate
Increase grade if
Moderate
P a g e | 43
PICO 1 (Testing): Targeted HCV antibody testing interventions
Testing HCV antibody positive: general strategy GRADE assessment criteria Type of evidence Assessment 1 RCT and 3 non-RCTs. Sensitivity analysis showed little impact of including nonRCTs. No Yes - individual study effect sizes are inconsistent, I2 = 58% No No No No No No GRADE score High GRADE running total
Decrease grade if
Serious limitation to study quality? Important inconsistency? Uncertainty about directness? Imprecise or sparse data? High probability of reporting bias? Strong evidence of association? Evidence of a dose response gradient? Plausible confounders would reduce effect?
Mark down
Moderate
Increase grade if
Moderate
Testing HCV antibody positive: direct strategy GRADE assessment criteria Type of evidence Assessment 4 RCTs and 6 non-RCTs. Sensitivity analysis showed little impact of including nonRCTs. No No Yes - individual study effect sizes are inconsistent, I2 = 79% No No No No No GRADE score High GRADE running total
Decrease grade if
Serious limitation to study quality? Important inconsistency? Uncertainty about directness? Imprecise or sparse data? High probability of reporting bias? Strong evidence of association? Evidence of a dose response gradient? Plausible confounders would reduce effect?
Mark down
Moderate
Increase grade if
Moderate
Referral to a HCV specialist: all strategies GRADE assessment criteria Type of evidence Decrease grade if Serious limitation to study quality? Important inconsistency? Uncertainty about directness? Imprecise or sparse data? High probability of reporting bias? Strong evidence of association? Evidence of a dose response gradient? Plausible confounders would reduce effect? Assessment 1 RCT (Inclusion of non-RCT evidence increased heterogeneity and effect size). No No No Yes – only 1 RCT, due to results of sensitivity analysis No No No No GRADE score High GRADE running total
Mark down
Moderate
Increase grade if
Moderate
Attendance at a HCV specialist: all strategies GRADE assessment criteria Type of evidence Decrease grade if Serious limitation to study quality? Important inconsistency? Uncertainty about directness? Imprecise or sparse data? High probability of reporting bias? Strong evidence of association? Evidence of a dose response gradient? Plausible confounders would reduce effect? Assessment 1 RCT (Inclusion of non-RCT evidence increased heterogeneity and effect size). No No No Yes – only 1 RCT, due to results of sensitivity analysis No No No No GRADE score High GRADE running total
Mark down
Moderate
Increase grade if
Moderate
P a g e | 44
PICO 1 (Testing): Targeted HCV antibody testing interventions
Commenced HCV treatment: all strategies GRADE assessment criteria Type of evidence Assessment 1 RCT and 2 observational studies. Inclusion of non-RCT evidence had minimal impact on effect size and heterogeneity. No No No Yes – there was a very small number of events in both arms No No No No GRADE score High GRADE running total
Decrease grade if
Serious limitation to study quality? Important inconsistency? Uncertainty about directness? Imprecise or sparse data? High probability of reporting bias? Strong evidence of association? Evidence of a dose response gradient? Plausible confounders would reduce effect?
Mark down
Moderate
Increase grade if
Moderate
Sustained Virological Response (SVR): all strategies GRADE assessment criteria Type of evidence Decrease Serious limitation to study quality? grade if Important inconsistency? Uncertainty about directness? Imprecise or sparse data? High probability of reporting bias? Strong evidence of association? Evidence of a dose response gradient? Plausible confounders would reduce effect? Assessment 2 observational studies No No No Yes – there was a very small number of events in both arms No No No No GRADE score Low GRADE running total
Mark down
Very low
Increase grade if
Very low
All-cause mortality: all strategies GRADE assessment criteria Type of evidence Decrease Serious limitation to study quality? grade if Important inconsistency? Uncertainty about directness? Assessment 1 RCT No No Yes – six months follow-up is inadequate to compare mortality in the two arms Yes – only 1 study, small n No No No No GRADE score High GRADE running total
Mark down
Moderate
Increase grade if
Imprecise or sparse data? High probability of reporting bias? Strong evidence of association? Evidence of a dose response gradient? Plausible confounders would reduce effect?
Mark down
Low
Low
P a g e | 45
PICO 1 (Testing): Targeted HCV antibody testing interventions
Appendix 6: Newcastle-Ottawa Scale NEWCASTLE - OTTAWA QUALITY ASSESSMENT SCALE FOR COHORT STUDIES Note: A study can be awarded a maximum of one star for each numbered item within the Selection and Outcome categories. A maximum of two stars can be given for Comparability Selection 1) Representativeness of the exposed cohort a) truly representative of the average _______________ (describe) in the community * b) somewhat representative of the average ______________ in the community * c) selected group of users eg nurses, volunteers d) no description of the derivation of the cohort 2) Selection of the non exposed cohort a) drawn from the same community as the exposed cohort * b) drawn from a different source c) no description of the derivation of the non exposed cohort 3) Ascertainment of exposure a) secure record (eg surgical records) * b) structured interview * c) written self report d) no description 4) Demonstration that outcome of interest was not present at start of study a) yes * b) no Comparability 1) Comparability of cohorts on the basis of the design or analysis a) study controls for _____________ (select the most important factor) * b) study controls for any additional factor * (This criteria could be modified to indicate specific control for a second important factor.) Outcome 1) Assessment of outcome a) independent blind assessment * b) record linkage * c) self report d) no description 2) Was follow-up long enough for outcomes to occur a) yes (select an adequate follow up period for outcome of interest) * b) no 3) Adequacy of follow up of cohorts a) complete follow up - all subjects accounted for * b) subjects lost to follow up unlikely to introduce bias - small number lost - > ____ % (select an adequate %) follow up, or description provided of those lost) * c) follow up rate < ____% (select an adequate %) and no description of those lost d) no statement
P a g e | 46
PICO 1 (Testing): Targeted HCV antibody testing interventions
Appendix 7: GRADE approach to assessing the quality of evidence across studies Quality of Evidence (summary score) High (4) =Further research is very unlikely to change our confidence in the estimate of effect. Moderate (3) = Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low (2) = Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Study Design Randomized trials or valid accuracy studies for diagnostic tests begin with a score of High (4) Downgrading Factors Study Limitations: -1 Serious -2 Very serious Upgrading Factors
Consistency: -1 Serious -2 Very serious Observational studies or indirect accuracy studies for diagnostic tests begin with a score of low (2).
Large effect +1 Large +2 Very large
Directness: -1 Serious -2 Very serious
Plausible confounding would change the effect +1
Precision: -1 Serious Very low (1) = Any estimate of effect is very uncertain. -2 Very serious Dose-response gradient +1 if Present Publication Bias: -1 Serious -2 Very serious
P a g e | 47