Interventions to improve adherence to treatment for paediatric tuberculosis in low- and middle-income countries: a systematic review and meta-analysis

Bull World Health Organ 2015;93:700–711B | doi: http://dx.doi.org/10.2471/BLT.14.147231 Systematic reviews 700 Interventions to improve adherence to treatment for paediatric tuberculosis in low- and middle-income countries: a systematic review and meta-analysis Meaghann S Weaver,a Knut Lönnroth,b Scott C Howard,c Debra L Roterd & Catherine G Lama Introduction Paediatric tuberculosis can be controlled or cured if timely and appropriate treatment is completed.1,2 More than 75% of affected patients live in low- and middle-income countries in Asia and Africa and have substantial tuberculosis –related morbidity and mortality.2 Up to 20% of children with tuber- culosis in low- and middle-income countries fail to complete treatment.3 Interrupted tuberculosis treatment poses a public health challenge because it permits the development of drug-resistant disease and allows patients to remain infectious for a relatively long time. Poor adherence results in disease progression, morbidity and death. The most extreme form of incomplete treatment is known as treatment abandonment or treatment default. For tuberculosis, such abandonment is generally represented by a break in treatment of at least two consecu- tive months.1 The barriers to treatment completion in low- and middle- income countries include medical expenses, the indirect costs of transportation and time away from work, the stigmas as- sociated with the illness and/or the treatment, communication breakdowns between providers and patients, limited health literacy, the presence of too few health workers and problems in drug procurement.2 We conducted a systematic review and meta-analysis of interventions designed to reduce such barri- ers to treatment completion among children with tuberculosis in low- and middle-income countries. Our main aim was to appraise the design, delivery and impact of such interventions in such a vulnerable population. Methods Search and selection Using a registered protocol (PROSPERO: CRD42013005800), we searched the PubMed and Cochrane databases for relevant pub- lications that had been published between 1 January 2003 and 1 December 2013. Grey literature was hand-searched. Until 1 May 2014, we attempted to contact the authors of relevant articles and other researchers with experience of tuberculosis in low- and middle-income countries. The search strategy (Box 1; available at: http://www.who.int/bulletin/volumes/93/10/14-147231) was pi- loted by two researchers and reviewed by two medical librarians. To be included in our analyses, a study had to have par- ticipants with active tuberculosis who were younger than 20 years and lived in a country that, according to the World Bank, was low-income or middle-income in December 2013. Studies with adult participants were included only if the cohort out- comes for participants younger than 20 years were available. We were only interested in studies on interventions targeted at the improvement of treatment initiation or completion, the improvement of adherence to medications or appointments, the prevention of treatment refusal or adherence surrogates such as self-efficacy or enablement. Objective To assess the design, delivery and outcomes of interventions to improve adherence to treatment for paediatric tuberculosis in low- and middle-income countries and develop a contextual framework for such interventions. Methods We searched PubMed and Cochrane databases for reports published between 1 January 2003 and 1 December 2013 on interventions to improve adherence to treatment for tuberculosis that included patients younger than 20 years who lived in a low- or middle- income country. For potentially relevant articles that lacked paediatric outcomes, we contacted the authors of the studies. We assessed heterogeneity and risk of bias. To evaluate treatment success – i.e. the combination of treatment completion and cure – we performed random-effects meta-analysis. We identified areas of need for improved intervention practices. Findings We included 15 studies in 11 countries for the qualitative analysis and of these studies, 11 qualified for the meta-analysis – representing 1279 children. Of the interventions described in the 15 studies, two focused on education, one on psychosocial support, seven on care delivery, four on health systems and one on financial provisions. The children in intervention arms had higher rates of treatment success, compared with those in control groups (odds ratio: 3.02; 95% confidence interval: 2.19–4.15). Using the results of our analyses, we developed a framework around factors that promoted or threatened treatment completion. Conclusion Various interventions to improve adherence to treatment for paediatric tuberculosis appear both feasible and effective in low- and middle-income countries. a St Jude Children’s Research Hospital, 262 Danny Thomas Place, MS 721, Memphis, TN 38105, United States of America (USA). b Global Tuberculosis Programme, World Health Organization, Geneva, Switzerland. c World Child Cancer USA, Denver, USA. d Johns Hopkins Bloomberg School of Public Health, Baltimore, USA. Correspondence to Catherine G Lam (email: catherine.lam@stjude.org). (Submitted: 11 September 2014 – Revised version received: 1 May 2015 – Accepted: 7 May 2015 – Published online: 23 June 2015 ) S stematic reviews Bull World Health Organ 2015;93:700–711B| doi: http://dx.doi.org/10.2471/BLT.14.147231 701 Systematic reviews Improving treatment adherence in paediatric tuberculosisMeaghann S Weaver Included studies required a control or comparison population. Retrospec- tive or contemporaneous comparisons from the same region were accepted if the between-population similarities and differences were clearly stated. No language, follow-up or study quality restrictions were imposed. Data extraction By using standardized forms, two investigators independently screened abstracts and extracted data. Discrepan- cies between the two investigators were resolved through discussion (16 records) or by the seeking of clarification from an author of an article of potential interest (three records). We detected 62 studies that met all of our eligibility criteria apart from the provision of explicit outcomes for paediatric patients. Although we at- tempted to determine such outcomes by contacting the authors of the corre- sponding study reports, we successfully obtained outcomes for just 10 additional studies. The other 52 reports provided no current contact information for any author (14 studies), had authors who did not reply to our queries (20 studies) or had authors who stated that the data we wanted were not available (18 studies). From each eligible report, we ex- tracted information on methods, in- terventions, outcomes, participants, settings and co-infection with human immunodeficiency virus (HIV). Treat- ment outcomes were extracted accord- ing to the World Health Organization’s (WHO’s) classifications, with treatment success defined as completion or cure1 – as given in the reports. Risk of bias in the randomized trials was assessed using the Cochrane As- sessment tool4 and reported according to CONSORT standards.5 Quality of the non-randomized trials was assessed us- ing the Effective Public Health Practice Project Quality Assessment tool6 and reported according to TREND stan- dards.5,7 Funding source was recorded as a possible bias source. Studies that integrated qualitative data were assessed using the relevant tools of the Critical Appraisal Skills programme.8 Report- ing of the systematic review adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement.9 Interventions to improve treatment adherence among paediatric patients of tuberculosis were summarized through independent iterative re-reading and organization of the identified themes – with discussion to achieve consensus – in alignment with WHO’s adherence dimensions for long-term therapies.2 For the initial data extraction, interven- tions were divided into five categories: education, psychosocial, care delivery, health systems and social protection or financial (Table 1). We attempted to determine those factors that promoted or threatened treatment completion. These factors might be related to: (i) the patient – e.g. literacy, (ii) the condition, including the presence of comorbidities, (iii) the therapy, including cultural lay beliefs, (iv) the health system, includ- ing accessibility, and (v) socioeconomic status, including family income. Statistical analysis We did a meta-analysis of the treat- ment success rates recorded among paediatric patients. We used the Mantel- Haenszel model and the DerSimonian and Laird random-effects method to calculate odds ratios (ORs) and their 95% confidence intervals (CIs) from the unadjusted raw data, with the as- sumption that intervention effects on treatment success in one setting might differ from those in other settings. We did sensitivity analyses that included only randomized or quasi-randomized studies or excluded studies with com- parison population estimates derived from another setting (available from the corresponding author). Heterogeneity across studies was assessed using the I2 statistic. We summarized the main meta-analysis results as a forest plot but used funnel plots to assess publication bias. Analyses were conducted using Review Manager version 5.2 (Cochrane Collaboration, Copenhagen, Denmark). Results We initially identified 413 articles of potential interest. Of these, 164 quali- fied for full-text review and we included 15 articles in our qualitative synthesis (Fig. 1).10–24 The articles were on 15 sepa- rate studies (Table 2). Three of the studies were published in Portuguese11,13,14 and the remainder in English. Five studies were based in the upper-middle-income countries of Brazil11,13,14 and Thailand,15,16 three in the lower-middle-income coun- tries of India,12 Lesotho10 and Pakistan,17 and seven in the low-income countries of Bangladesh,23 Ethiopia,18,19 Kenya,22 Myanmar,24 South Sudan21 and the United Republic of Tanzania.20 Four settings were urban outpatient,11,13,16,17 three rural outpatient,12,18,19 two subur- Table 1. Categorization of interventions aimed at improving tuberculosis treatment adherence Intervention category Components Examples Education Behavioural and cognitive Teaching of patients, family members and community members Psychosocial Behavioural and affective Counselling Contracts Cultural competence contextualization Social support to include communication relevant to patient efficacy or enablement Care delivery Behavioural, affective, biological and structural Treatment regimen interventions in the form of combination pills or easier dosing Convenience of visits timed with medication refills Staff training – including provider-targeted interventions related to communication Decentralization of health contact via home visits or community health workers Health systems Behavioural, biological, cognitive and structural Management processes Tracer systems Referral support Direct accountability in the form of direct observation of therapy Social protection or financial Behavioural and structural Financial support for – or provision of – food, transportation and housing Free health services or reimbursement of costs Bull World Health Organ 2015;93:700–711B| doi: http://dx.doi.org/10.2471/BLT.14.147231702 Systematic reviews Improving treatment adherence in paediatric tuberculosis Meaghann S Weaver ban outpatient,23,24 one rural camp.21 The remaining studies were done in variable settings.10,14,15,20,22 The payment system for health servic- es was not described in nine studies11,12,14–20 but the reports on four studies described capped fees24 or clinic fee coverage.16,23,24 In seven studies, drug expenses were covered for one intervention group only,12 for both the intervention and comparison groups, as part of a national scheme,16,22–24 or for at least the intervention group – with un- clear indication if the drug expenses of the comparison group were also covered.10,21 The included studies were con- ducted between 1996 and 2011 and reported – including the unpublished data supplied by authors – between 2003 and 2014. The median duration of the investigated interventions was 24 months (range: 9–96). The number of participants younger than 20 years – which had to be clarified through author contact for six studies and excluded population-based comparison samples – varied from four to 308 (mean: 106; median: 61) and totalled 1587 across all 15 studies. Such paediatric patients represented between 3% and 100% of the patients investigated (mean: 22%; median: 11%). The prevalence of HIV co-infection, which was only reported for six studies, ranged from less than 5% to 74%.10,11,13,15,20,21 Interventions The timing of interventions either in- cluded referral10 or induction15 or ran just from treatment initiation to treat- ment completion.11–14,16–24 Health be- haviour models informing intervention design were mentioned in two studies – the precede-proceed model was used to help engage patients in one study16 while social franchising was used to help engage providers in another study.24 Many studies involved several cat- egories and subcategories of interven- tions (Table 3). Some used interventions combining cognitive and behavioural components, as exemplified by educa- tion for patients,10,12,16,18,19,21–24 family members,10,12,20,21 or community lead- ers.12,18,19,21,24 Educational curricula ad- dressed the administration11,16,18–20,22,23 and adverse effects of medication,16,19,23,24 the personal or public health conse- quences of early treatment discon- tinuation16,19,21–23 and overall health or hygiene.16,18,19,21 Eleven studies incorporated af- fective and behavioural components, through psychosocial support with therapeutic alliances (i.e. relationship- building between providers and patie nts),10,16,18,19,21–23 patient empowerment to select a treatment supporter or loca- tion,12,15,16,18,20,22 counselling,10,16,21,22 prob- lem-solving,16 decreasing stigma14,19,21 and peer support.14,16,19,21,22 Care delivery interventions included health provider training,10,16–19,21,22,24 con- venient appointment scheduling,10,12,19,23 migration-sensitive therapy duration21 and easier dosing schedules.10,12,18,24 Health system interventions included the directly observed treatment, short- course strategy,10–12,15,16,18–22 referral support,17,19 patient tracers10,12,16–19,21,22,24 – including tracing within 24 hours12,21 – and home visiting.16,22 Social protection or financial sup- port interventions included weekly food rations,10,21 monthly food bas- kets,13 housing,21 medication cover- age,12,21 recognition of the importance of employment14,16 or school,10 essential supplies for daily life,16 transport reim- bursement10 and income-generation support.10 One study required a deposit that was refundable upon treatment completion.12 Treatment adherence Adherence-related measures included those extracted from self-reports,16,24 pharmacy refill data,23 medication records maintained by treatment sup- porters,12,19 clinic attendance records,23 confirmation of referrals17 and medical records.10,11,13–15,18,20–22,24 Terminology describing unfa- vourable outcomes included defau lt,10,12,15,16,18–22,24 drop-out,11,14 abandon- ment13,14 and treatment interruption.19 Three of 10 studies used the term default and, in defining their default criteria, were consistent with WHO defini- tions.10,19,24 Drop-out was defined in one study as treatment interruption for more than 30 days.11 Treatment abandonment was not defined in the two studies using the term.13,14 In addition to treatment success – i.e. completion or cure – positive out- comes were defined in the study reports as successful referral – i.e. confirmed arrival at the referral facility,17 continu- ous attendance at scheduled visits,22,23 more than 90% medication adherence23 or self-reported beneficial health be- haviours.16 Fig. 1. Flowchart for the selection of studies on interventions to improve adherence to treatment for paediatric tuberculosis in low- and middle-income countries Search of PubMed and Cochrane databases (n = 375) Articles after duplicate removal (n = 413) Titles and abstracts screened (n = 413) Full-text articles assessed for eligibility (n = 164) Articles included in qualitative analysis (n = 15) Articles included in meta-analysis (n = 11) Secondary bibliographic searches (n = 44) Expert recommendation (n = 14) Articles excluded (n = 249) • Reviews, letters or editorial comments (n = 16) • Unrelated to adherence to tuberculosis treatment (n = 82) • In high-income country (n = 9) • Lacking control or comparison group (n = 8) • No intervention investigated (n = 58) • No outcome related to treatment adherence (n = 75) • No paediatric participants (n = 1) Full-text articles excluded (n = 149) • Reviews, letters or editorial comments (n = 4) • In high-income country (n = 7) • Lacking control or comparison group (n = 11) • No intervention investigated (n = 44) • No outcome related to treatment adherence (n = 24) • No paediatric participants (n = 53) • Only investigating latent tuberculosis (n = 6) Articles found ineligible for meta-analysis (n = 4) • Rate of treatment success within paediatric cohort not reported and not traceable (n = 4) Bull World Health Organ 2015;93:700–711B| doi: http://dx.doi.org/10.2471/BLT.14.147231 703 Systematic reviews Improving treatment adherence in paediatric tuberculosisMeaghann S Weaver Ta bl e 2. St ud ie s o n in te rv en tio ns to im pr ov e tr ea tm en t a dh er en ce fo r p ae di at ric tu be rc ul os is in lo w - a nd m id dl e- in co m e co un tr ie s, 19 96 –2 01 1 St ud y Co un tr y a nd st ud y de sig n Ca re se tt in g Pa rt ici pa nt d es cr ip tio n Du ra tio n, m on th s Pe rio d St ud y a rm s In te rv en tio n Co m pa ris on N on -r an do m iz ed An uw at no nt ha ka te e t a l.1 5 Th ai la nd , p ro sp ec tiv e ob se rv at io na l c oh or ta Re gi on – a ll pu bl ic a nd pr iv at e fa ci lit ie s i n fo ur pr ov in ce s D iv er se p at ie nt p op ul at io n in cl ud in g ur ba n, ru ra l a nd m ig ra nt p op ul at io ns . H IV co -in fe ct io n ra te 2 0% . O f t he pa rt ic ip an ts , 2 23 (3 % ) w er e ag ed < 1 5 ye ar sb . 24 20 04 –2 00 6 D O T su pe rv ise d by fa m ily m em be r or H CW Se lf- ad m in ist er ed th er ap y H ec k et a l.1 1 Br az il, re tro sp ec tiv e ob se rv at io na l c ro ss - se ct io na la Ci ty – 1 8 ur ba n ou tp at ie nt pr im ar y he al th u ni ts an d fiv e re fe rra l u ni ts su pe rv ise d by M un ic ip al Tu be rc ul os is Co nt ro l Pr og ra m m e So ci oe co no m ic a nd e du ca tio n su m m ar y no t p ro vi de d; H IV co -in fe ct io n ra te 1 6% . O f t he pa rt ic ip an ts , 5 7 (9 % ) w er e ag ed ≤ 1 9 ye ar s. 96 20 00 –2 00 4 an d 20 05 –2 00 8 D ec en tra liz at io n of tu be rc ul os is pr og ra m m e ac tio ns fo r p rim ar y ca re a nd im pl em en ta tio n of D O T SO C be fo re de ce nt ra liz at io n in iti at iv es Le e et a l.2 3 Ba ng la de sh , pr os pe ct iv e be fo re - an d- af te r s tu dy a Cl in ic – su bu rb an p rim ar y he al th c lin ic in in du st ria l co m pl ex n ea r c ap ita l Pa rt ic ip an ts h ad lo w so ci oe co no m ic st at us , l im ite d ed uc at io n an d hi gh le ve l o f ill ite ra cy . O f t he p ar tic ip an ts , 26 (7 % ) w er e ag ed < 1 8 ye ar sb . 33 20 05 –2 00 6 an d 20 06 –2 00 7 Pa tie nt e du ca tio n on th e im po rt an ce o f t re at m en t ad he re nc e pr ov id ed , b y a ph ys ic ia n, w ee kl y fo r 1 m on th , fo rt ni gh tly fo r n ex t m on th , t he n m on th ly . V isi ts sc he du le d to co in ci de w ith m ed ic at io n re fil ls SO C, w ith n o st an da rd ize d pa tie nt ed uc at io n an d re tu rn vi sit s n ot ti m ed to co in ci de w ith re fil ls M ar qu es a nd d a Cu nh a1 4 Br az il, re tro sp ec tiv e be fo re -a nd -a fte ra H os pi ta l – u rb an h os pi ta l In di ge no us p op ul at io n su ffe rin g ex tre m e po ve rt y, m al nu tri tio n an d cu ltu ra l a nd so ci oe co no m ic b ar rie rs to ex te nd ed h os pi ta liz at io n. O f th e pa rt ic ip an ts , 2 44 (4 1% ) w er e ag ed < 1 5 ye ar sb . 35 19 96 –1 99 8 an d 19 98 –1 99 9 O ut pa tie nt tr ea tm en t w ith h om e- ba se d D O T vi a in di ge no us h ea lth ag en ts Sy st em at ic ho sp ita liz at io n of pa tie nt s f or u p to 6 m on th s O ng ’a ng ’o e t a l.2 2 Ke ny a, re tro sp ec tiv e co ho rt a Re gi on – sa m pl e of fo ur ur ba n an d ru ra l p ub lic he al th fa ci lit ie s, us in g an d no t u sin g CH W s M en tio n of st ig m a to w ar ds tu be rc ul os is an d cu ltu ra l be lie fs a ga in st c on ve nt io na l tre at m en t o f t he d ise as e in ru ra l s et tin g. O f t he pa rt ic ip an ts , 2 98 (1 1% ) w er e ag ed < 1 4 ye ar sb . 72 20 05 –2 01 1 Pe rs on al ize d ed uc at io n fro m C H W , on tr ea tm en t a nd ri sk s i nv ol ve d in la ck o f a dh er en ce , p lu s C H W - su pe rv ise d D O T at h ou se ho ld le ve l w ith o ng oi ng C H W e du ca tio na l su pp or t N ur se a t h ea lth fa ci lit y ad vi se d pa tie nt s o f tre at m en t s ch ed ul e, ne ed fo r a dh er en ce a nd ne ed fo r f am ily su pp or t. W ee kl y D O T at h ea lth fa ci lit y (c on tin ue s. . . ) Meaghann S WeaverImproving treatment adherence in paediatric tuberculosis Systematic reviews 704 Bull World Health Organ 2015;93:700–711B| doi: http://dx.doi.org/10.2471/BLT.14.147231 St ud y Co un tr y a nd st ud y de sig n Ca re se tt in g Pa rt ici pa nt d es cr ip tio n Du ra tio n, m on th s Pe rio d St ud y a rm s In te rv en tio n Co m pa ris on Sa tt i e t a l.1 0 Le so th o, re tro sp ec tiv e co ho rt Co m m un ity – m ou nt ai no us ru ra l a nd ur ba n, in pa tie nt a nd ou tp at ie nt se tt in g N in et ee n pa tie nt s w ith su sp ec te d or c on fir m ed M D R tu be rc ul os is, o f w ho m 14 (7 4% ) w er e co -in fe ct ed w ith H IV , 1 2 (6 3% ) w er e m al no ur ish ed a nd a ll w er e ag ed < 1 6 ye ar s. 42 20 07 –2 01 1 Co m pr eh en siv e ap pr oa ch to c ar e fo r M D R tu be rc ul os is, w ith o r w ith ou t H IV c o- in fe ct io n, u sin g so ci al su pp or t, cl os e m on ito rin g by CH W s a nd c lin ic ia ns a nd in pa tie nt ca re w he n w ar ra nt ed Pa tie nt s o f M D R tu be rc ul os is w ith h ig h ra te s o f H IV c o- in fe ct io n in n ei gh bo ur in g So ut h Af ric a va n de n Bo og aa rd e t a l.2 0 U ni te d Re pu bl ic of Ta nz an ia , re tro sp ec tiv e ob se rv at io na l c oh or ta Re gi on – u rb an a nd ru ra l di st ric ts w ith n at io na l re fe rra l h os pi ta l, r eg io na l ho sp ita l a nd p rim ar y he al th c lin ic s So ci oe co no m ic a nd e du ca tio n su m m ar y no t p ro vi de d. H IV co -in fe ct io n ra te 3 1% . O f t he pa rt ic ip an ts , 3 08 (1 1% ) w er e ag ed < 1 5 ye ar s. 12 20 07 Pa tie nt -c en tre d tre at m en t t ha t al lo w ed p at ie nt s t o ch oo se be tw ee n co m m un ity a nd fa ci lit y- ba se d D O T Co nv en tio na l f ac ili ty - ba se d D O T su pe rv ise d by fa ci lit y- ba se d pr ov id er Ba da r e t a l.1 7 Pa ki st an , p ro sp ec tiv e ob se rv at io na l c oh or t Pr ov in ce – u rb an , no ng ov er nm en t o ut pa tie nt te rt ia ry c ar e ho sp ita l a s re fe rri ng c en tre So ci oe co no m ic a nd e du ca tio n su m m ar y no t p ro vi de d. O f t he pa rt ic ip an ts , 1 50 (3 4% ) w er e ag ed ≤ 1 9 ye ar s. 9 20 09 El ec tro ni c da ta ba se re gi st er , de sig na te d ov er sig ht o f r ef er ra ls, st aff re fe rra l o rie nt at io n, tra ck in g vi a 1– 3 ph on e ca lls , co m m un ic at io n be tw ee n ce nt re s vi a ex ch an ge s o f p re -s ta m pe d m ai l, s ch ed ul ed m ee tin gs a nd ph on e co nt ac t a nd p at ie nt s re fe rre d to c lo se st fa ci lit y Pa tie nt re sp on sib le fo r re tu rn to re fe rri ng u ni t Ca nt al ic e Fi lh o1 3 Br az il, be fo re -a nd - af te ra Cl in ic – u rb an p rim ar y he al th c ar e ou tp at ie nt cl in ic So ci oe co no m ic a nd e du ca tio n su m m ar y no t p ro vi de d. H IV co -in fe ct io n ra te < 5 % . O f t he pa rt ic ip an ts , 8 (6 % ) w er e ag ed < 1 8 ye ar sb . 57 20 01 –2 00 3 an d 20 04 –2 00 6 St an da rd tr ea tm en t r eg im en p lu s m on th ly fo od b as ke t St an da rd tr ea tm en t re gi m en , i nc lu di ng se lf- ad m in ist er ed th er ap y Ke us e t a l.2 1 So ut h Su da n, pr os pe ct iv e ob se rv at io na l c oh or ta Pr og ra m m e – hu m an ita ria n ru ra l tu be rc ul os is ca m p lo ca te d in “t ra ns iti on al ” z on e be tw ee n m ili tia a nd lo ca l fa ct io ns Pa st or al , m ig ra to ry p op ul at io n liv in g in c on fli ct c on di tio ns w ith n o he al th in fra st ru ct ur e. H IV c o- in fe ct io n ra te < 5 % . O f t he p ar tic ip an ts , 8 4 (5 2% ) w er e ag ed < 1 5 ye ar s. 9 20 01 Vi lla ge -b as ed tr ea tm en t i n a co nfl ic t z on e of S ou th S ud an Tr ea tm en t i n a le ss in se cu re a re a – M an ya tt a Re gi on – w ith 2 -m on th su pe rv ise d th en 3- m on th u ns up er vi se d re gi m en Lö nn ro th e t a l.2 4 M ya nm ar , pr os pe ct iv e co ho rt Cl in ic s – m ul tip le to w ns hi p ou tp at ie nt c lin ic s s er vi ng lo w -in co m e po pu la tio n M os tly p at ie nt s w ith lo w so ci oe co no m ic st at us , f ro m to w ns hi ps in w hi ch m an y us ed p riv at e he al th c ar e as th e fir st p oi nt o f c on ta ct . O f t he pa rt ic ip an ts , 6 6 (2 6% ) w er e ag ed 1 6 ye ar s. 14 20 04 –2 00 5 So ci al fr an ch ise e ng ag in g pr iv at e ge ne ra l p ra ct iti on er s t o de liv er qu al ity c on tro lle d tu be rc ul os is ca re , i nc lu di ng se rv ic e br an di ng , de fin ed tr ea tm en t s up po rt er a nd de fa ul t t ra ci ng m ec ha ni sm Co nt in ua tio n of p re vi ou s SO C, w ith p at ie nt ut ili za tio n of e xi st in g tre at m en t c en tre s a nd th e pu bl ic se ct or ’s D O T lo go b ra nd in g (. . . co nt in ue d) (c on tin ue s. . . ) Meaghann S Weaver Improving treatment adherence in paediatric tuberculosis Systematic reviews 705Bull World Health Organ 2015;93:700–711B| doi: http://dx.doi.org/10.2471/BLT.14.147231 St ud y Co un tr y a nd st ud y de sig n Ca re se tt in g Pa rt ici pa nt d es cr ip tio n Du ra tio n, m on th s Pe rio d St ud y a rm s In te rv en tio n Co m pa ris on Ra nd om iz ed o r q ua si - ra nd om iz ed D at ik o an d Li nd tjø rn 18 Et hi op ia , p ro sp ec tiv e ra nd om ize d Cl in ic s – ru ra l o ut pa tie nt se tt in g in so ut h of c ou nt ry Pa tie nt s w ith p oo r a cc es s, po ve rt y an d lo w h ea lth - se ek in g be ha vi ou rs . O f t he pa rt ic ip an ts , 3 2 (1 0% ) w er e ag ed < 1 4 ye ar s. 19 20 06 –2 00 8 Lo ca l t re at m en t b y H EW s. H EW tra in in g in a dh er en ce su pp or t, di ag no sis , r ef er ra l w ith e nh an ce d ca se fi nd in g an d th e pr ob le m s of n on -a dh er en ce . C om m un ity m ob ili za tio n an d ed uc at io n H EW s d id n ot re ce iv e tra in in g on d ia gn os tic te ch ni qu es o r a dh er en ce su pp or t. H EW s e ng ag e in c om m un ity e du ca tio n on sy m pt om s o f tu be rc ul os is. D O T pr ov id ed a t h ea lth fa ci lit y in st ea d of w ith in lo ca l n ei gh bo ur ho od D em iss ie e t a l.1 9 Et hi op ia , p ro sp ec tiv e qu as i-r an do m ize da Cl in ic s – ru ra l o ut pa tie nt ce nt re s i n no rt h of c ou nt ry Tu be rc ul os is as so ci at ed w ith st ro ng c om m un ity st ig m a, to th e ex te nt th at p at ie nt s m ay lo se th ei r w or k if em pl oy er is aw ar e of d ia gn os is. O f t he pa rt ic ip an ts , 7 (5 % ) w er e ag ed < 1 5 ye ar s. 12 19 98 –1 99 9 Pa tie nt s o rg an ize d ac co rd in g to re sid en tia l a re a in to c lu bs , e ac h w ith 3 –1 0 m em be rs , a n el ec te d le ad er a nd th e sa m e ap po in tm en t da te s. W ee kl y cl ub m ee tin gs w ith e m ph as is on so ci al su pp or t to w ar ds tr ea tm en t c om pl et io n. Co nt in ua tio n of p re vi ou s SO C. N o tu be rc ul os is cl ub s b ut o th er w ise sim ila r t re at m en t re gi m en a nd p ac ka ge s of h ea lth e du ca tio n as in th e in te rv en tio n ar m Kh or tw on g an d Ka ew ku ng w al 16 Th ai la nd , p ro sp ec tiv e qu as i-r an do m ize da Cl in ic s – u rb an o ut pa tie nt ho sp ita l c lin ic s M ar gi na liz ed m ig ra nt po pu la tio n liv in g in cr ow de d co nd iti on s, w ith hi gh m ob ili ty . L ac k of le ga l st at us o r r eg ist ra tio n m ad e m os t i ne lig ib le fo r r ou tin e he al th -c ar e se rv ic es . O f t he pa rt ic ip an ts , 4 (4 % ) w er e ag ed < 1 8 ye ar sb . 16 20 09 –2 01 0 M ig ra nt p op ul at io n pr ov id ed w ith in te ns iv e ed uc at io n m od ul es , ho m e an d w or kp la ce v isi ts a nd ph on e- ca ll re m in de rs , w ith em ph as is on th er ap eu tic h ea lth te am re la tio ns hi ps M ig ra nt p op ul at io n re ce iv ed c on tin ua tio n of p re vi ou s S O C, w hi ch in cl ud ed o pt io na l tre at m en t s up er vi sio n by a vi lla ge h ea lth v ol un te er M at he w e t a l.1 2 In di a, re tro sp ec tiv e qu as i-r an do m ize d ob se rv at io na l c oh or t Cl in ic – o ut pa tie nt c lin ic ba se d in ru ra l s ec on da ry - le ve l m iss io n ho sp ita l i n no rt h of c ou nt ry In o ne o f t he p oo re st re gi on s in In di a, w ith h ig h ra te o f ill ite ra cy . T rib al p op ul at io n en ga ge d in sm al l-s ca le fa rm in g, w ith p oo r r oa d ac ce ss . O f t he p ar tic ip an ts , 9 4 (1 4% ) w er e ag ed < 1 5 ye ar s bu t d at a w er e on ly re po rt ed fo r 6 1 of th es e. 30 20 01 –2 00 3 Fr ee d ru gs , v isi ts m ad e to th e pa tie nt b y th e D O T su pe rv iso r – a co m m un ity m em be r – m on th ly du rin g in te ns iv e ph as e an d ev er y 2 m on th s t he re af te r. Ad he re nc e ch ec ks . P at ie nt a sk ed to v isi t c lin ic th re e tim es d ur in g th er ap y D ru gs p ro vi de d at co st , f am ily m em be r su pp or te d D O T an d ac co m pa ni ed p at ie nt to ap po in tm en ts . M on th ly cl in ic v isi ts in in te ns iv e ph as e an d cl in ic vi sit s e ve ry 2 m on th s th er ea fte r CH W : c om m un ity h ea lth w or ke r; D OT : d ire ct ly o bs er ve d th er ap y; H CW : h ea lth -c ar e w or ke r; HE W : h ea lth -e xt en sio n w or ke r; HI V: h um an im m un od efi ci en cy v iru s; M D R: m ul tid ru g- re sis ta nt ; S O C: st an da rd o f c ar e. a An a ut ho r o f t he re le va nt a rti cl e ha d to b e co nt ac te d to c la rif y th e ra te o f t re at m en t s uc ce ss in th e pa ed ia tri c pa rti ci pa nt s a nd /o r t he d efi ni tio n us ed fo r t re at m en t a ba nd on m en t. b Th e siz e of th e pa ed ia tri c sa m pl e ha s n ot b ee n pu bl ish ed p re vi ou sly a nd h ad to b e ob ta in ed b y di re ct c on ta ct w ith a n au th or o f t he re le va nt a rti cl e. (. . . co nt in ue d) Bull World Health Organ 2015;93:700–711B| doi: http://dx.doi.org/10.2471/BLT.14.147231706 Systematic reviews Improving treatment adherence in paediatric tuberculosis Meaghann S Weaver Risk of bias The benefits of the investigated inter- ventions may be overestimated because of short follow-up and failure to assess adherence after the interventions were discontinued. Confounders, such as the extra attention given to participants during educational interventions,16,23 complicate our analyses. Although one study report details how controls – who did not receive the educational inter- vention – were supervised by health volunteers,16 it failed to give any idea of the corresponding contact time. The concurrent use of several interventions makes it hard to determine the main reason for successful outcomes. Social feedback loops – in which successful interventions foster a dynamic for more community adherence – were subjec- tively recognized by several research teams.16,18,19,21,24 Intervention complexity increased as attention expanded beyond the patient to include the provider,23 the family,13–15 both the provider and family10–12,16,17,20 or the provider, family and community.18,19,21,22,24 Complexity was characterized by contextual inter- actions that were susceptible to policy timing,13,18,20,21,24 staffing capabilities and attitudes,12,16,17,19,22,23 relationships13,16,19,23 and resources.18,19,23,24 No empiric qual- ity measures of implementation fidelity were described. Two studies incorporated qualita- tive data from focus groups and in-depth interviews.19,22 Although context, sam- pling and data collection were outlined and the findings appeared supported by data, there was no discussion of reflexivity and no detailed description of the analyses. None of the studies we investigated incorporated long-term ob- servational or ethnographic approaches. In one prospective randomized con- trolled trial, the study communities were randomly allocated to intervention and control groups to limit selection bias.18 Three quasi-randomized trials deter- mined assignment by residence.12,16,19 No before-and-after studies used controls to account for any secular change. None of the articles described blinding measures and three specified a lack of blinding for assessors11,24 or participants.20 All of the results reported in thir- teen studies were apparently defined a priori.10–16,18–20,22–24 The remaining two studies accounted for modification Table 3. Interventions to improve adherence to treatment for paediatric tuberculosis in low- and middle-income countries, 1996–2011 Main category of primary intervention, reference Intervention categories and subcategories included in study Educational Psychosocial Care delivery Health systems Social protection or financial Pr ov id er Pa tie nt Fa m ily Co m m un ity Th er ap eu tic a lli an ce a Pe er su pp or t Co un se lli ng St ig m a ad dr es se d St aff su pp or t Pa tie nt -c en tr ed ch oi ce s Sc he du lin g De ce nt ra liz at io n St aff tr ai ni ng Ca re q ua lit y a ss ur an ce Tr ea tm en t c on ve ni en ce Di re ct ly o bs er ve d tr ea tm en t Re gi st ry Tr ac in g Fo od Tr an sp or t Li vi ng e nv iro nm en t In co m e ge ne ra tio n Su bs id ize d tr ea tm en t Educational Khortwong and Kaewkungwal16 – + – – + + + – + + – + + – – + – + + – + – – Lee et al.23 – + – – + – – – + – + – – – – – – – – – – – – Psychosocial Demissie et al.19 – + – + + + – + + – + – + – – + – + – – – – – Care delivery Anuwatnonthakate et al.15 – – – – – – – – + + – + – – – + – – – – – – – Datiko and Lindtjørn18 – + – + + – – – – + – + + + + + – + – – – – – Heck et al.11 – – – – – – – – – – – + – – – + – – – – – – – Keus et al.21 + + + + + + + + + – – + + + + + – + + – + – – Marques and da Cunha14 – – – – – + – + – – – + – – – – – – – – – – – Satti et al.10 – + + – + – + – + – + + + – + + – + + + + + – van den Boogaard et al.20 + – + – – – – – + + – + – – – + – – – – – – – Health systems Badar et al.17 – – – – – – – – – – – – + – – – + + – – – – – Lönnroth et al.24 + + – + – – – – – – – – + + + – + + – – – – + Mathew et al.12 – + + + – – – – + + + + – + + + – + – – – – + Ong’ang’o et al.22 – + – – + + + – + + – + + – – + – + – – – – – Social protection or financial Cantalice Filho13 – – – – – – – – – – – – – – – – – – + – – – – a Refers to relationship-building between providers and patients. Bull World Health Organ 2015;93:700–711B| doi: http://dx.doi.org/10.2471/BLT.14.147231 707 Systematic reviews Improving treatment adherence in paediatric tuberculosisMeaghann S Weaver of the results reported due to limited follow-up data, which had impaired the assessment of cure21 or treatment outcome beyond referrals.17 Funding sources included nongov- ernmental organizations,10,11,20–24 health departments18 or international15,17,19 or local16 academic institutes or were not specified.11 Table 4 and Table 5 show the re- sults on study-specific biases (available at: http://www.who.int/bulletin/vol/- umes/93/10/14-147231). Meta-analysis Treatment success rates for the paedi- atric participants in both the treatment and comparison groups were reported for 11 studies.10–12,14–16,18–20,22,23 These studies were included in the meta-anal- ysis and together represented 1279 chil- dren – excluding those in any external comparison groups. In three of the four studies excluded from the meta-analysis, the interventions investigated appeared to bring improved rates of treatment success, for all age groups.13,21,24 The results of the other excluded study17 indicated that the intervention led to increased referral rates. Meta-analysis revealed a threefold improvement in odds of treatment success for children receiving the in- terventions (Fig. 2; OR: 3.02; 95% CI: 2.19–4.15). There was no evidence of statistical heterogeneity (I2: 0%). A fun- nel plot showed symmetry for the large, high-powered studies but potential publication bias for the smaller studies (Fig. 3; available at: http://www.who. int/bulletin/volumes/93/09/14-147231). Sensitivity analysis did not modify the overall results (available from the cor- responding author). Baseline risk factors reported for poor adherence outcomes are outlined in Box 2. Discussion In our review of interventions to pro- mote paediatric tuberculosis treatment adherence in low- and middle-income countries, we found evidence that such interventions can result in clinically important improvements in tuberculosis treatment success. Diverse interventions addressing education, psychosocial support, care delivery, health system strengthening and social protection are reportedly feasible and effective in facilitating treatment completion. Several studies followed collabora- tive strategies. For example, there was evidence of social franchise programmes communicating with the media, tuber- culosis villages communicating with local leaders, tuberculosis clubs com- municating with neighbours, health centres communicating with referral facilities and health providers engaging in motivational communication with patients. We used systematic methods to identify and analyse a broad range of studies, without language limitations and with solicitation of input from the authors of relevant articles in an attempt to minimize search bias. We provided detailed descriptions and syntheses of interventions – which were often multi- component and complex – that had been implemented among children in low- and middle-income countries. Our summary findings may help guide future interven- tion planning and evaluation. Our reviews did, however, have several limitations. For example, few studies included specific details on the nature of their paediatric programme, and no data on individual patients were available. Given the gen- erally small sample sizes, the reported confidence intervals for the effects of individual interventions were often broad. Despite this, all but one of the 11 studies included in the meta-analysis had odds ratios that indicated that the investigated intervention improved the rate of treat- ment success, and the four largest of these studies provided unequivocal evidence of such benefit. Heterogeneity in the context and measurement of adherence, outcome definition and reporting limit the value of between-study comparisons. In high- income countries, multi-component interventions are common and often found to be superior to single-compo- nent interventions.26 Several of the rele- vant studies included in our reviews also attempted to target several adherence Fig. 2. Effect on the odds of treatment success of interventions to improve adherence to treatment for paediatric tuberculosis in low- and middle-income countries, 1996–2011 Study or Odds Ratio Odds Ratio Subgroup Weight (%) M-H, Random, 95% CI M-H, Random, 95% CI Ong’ang’o 2014 30.2 2.37 (1.32–4.24) Knortwong 2013 0.7 5.00 (0.11–220.62) Lee 2013 2.2 0.57 (0.07–4.88) Satti 2012 5.7 1.79 (0.47–6.79) Heck 2011 1.2 9.34 (0.48–182.06) Datiko 2009 1.6 1.31 (0.10–16.56) van den Boogaards 2009 30.6 4.12 (2.31–7.35) Anuwatnonthakate 2008 7.8 3.64 (1.16–11.43) Mathew 2005 1.2 15.19 (0.85–270.99) Demissie 2003 0.8 11.00 (0.28–433.80) Marques 2003 17.9 2.95 (1.39–6.29) Total (95% CI) 100.0 3.02 (2.19–4.15) Heterogeneity: Tau2 = 0.00; Chi2 = 7.54, df = 10 (P = 0.67); l2 = 0% Test for overall effect: Z = 6.76 (P < 0.00001) Favours control Favours intervention 0.01 0.1 1 10 100 CI: confidence interval; M-H: Mantel-Haenszel model. Note: In a random effects meta-analysis, odds ratios were derived from individual studies (squares) or as summary value (diamond). The size of the square data marker for individual studies is proportional to the number of patients in the study. Box 2. Reported risk factors for poor tuberculosis treatment adherence outcomes in paediatric patients Patient-related • Female sex12 • Male sex14,23 Condition-related • Human immunodeficiency virus- positive20 • Smear-negative tuberculosis20,23 Treatment-related • Tuberculosis retreatment24 Social and/or economic related • Low-socioeconomic level24 Health system related • Distance from care source12 Bull World Health Organ 2015;93:700–711B| doi: http://dx.doi.org/10.2471/BLT.14.147231708 Systematic reviews Improving treatment adherence in paediatric tuberculosis Meaghann S Weaver factors simultaneously, by using com- plex interventions. Such complex inter- ventions make it difficult to attribute the results to particular intervention catego- ries or components. One of the studies we reviewed was of an intervention that included education, improved dosing and appointment convenience, patient tracing, reduction of out-of-pocket costs and a deposit that was refunded on treat- ment completion.12 It may be that only when implemented together do these elements succeed. Recognizing the interconnected nature of WHO’s five adherence di- mensions and intervention categories for long-term therapies,2 we have summarized contextual factors affect- ing the adherence interventions we investigated in a framework (Fig. 4). The themes highlighted in this figure are intended to be illustrative across dimensions and intervention categories. For instance, factors that may adversely affect tuberculosis treatment adherence that span psychosocial and educational categories – e.g. low literacy and limited self-efficacy – are shown in the figure alongside adherence-promoting factors such as family education and patient em- powerment. The contextual framework may aid further collaborative studies and analyses of adherence-targeted interventions. Through qualitative analysis, we identified three areas where stud- ies described – or failed to describe – children’s unique features that can affect adherence intervention delivery. First, few studies described paediatric- specific disease epidemiology and use of paediatric-inclusive outcomes. Several authors reported an unexpectedly high prevalence of paediatric tuberculosis that warranted management as a public health problem.14,18,21 However, most of the studies that we screened simply excluded children and 54 studies that would otherwise have been eligible for our analyses had to be excluded because they failed to report paediatric outcomes separately. Even for the eligible studies, adherence outcomes were not explicitly adapted for paediatric patients – al- though paediatric-specific treatment toxicity was recognized in one study.10 Second, several reports noted challenges in paediatric tuberculosis diagnosis and care. Children can pose diagnostic dilemmas that complicate epidemiological and outcome esti- mates.10,21 One study noted that paedi- atric lymph-node biopsies could not be safely performed locally.21 Another considered how children’s difficulty with sputum production may contribute to low detection rates18 while a different study specified distinct sputum collec- tion techniques for younger children.10 Dosing instructions that were adapted for paediatric treatment were also recommended.10 Key comorbidities in children – e.g. malnutrition21 – may benefit from dedicated attention. Third, several studies acknowledged the need to consider the preferences and social role of children and adolescents, who may need tailored interventions. In one study involving the use of directly observed, short-term treatment, chil- dren and women were more likely than men to select community-based over facility-based treatment, when given the option.20 Another study adapted an in- tervention, for use among children, ac- cording to household and social needs. This intervention included supporting the children in returning to school.10 As one study commented, tuberculosis – and tuberculosis treatment – can cut the economic productivity of adolescents and young adults, who tend to have relatively high burdens of the disease.12 Based on our review and identi- fied themes, future studies need to: (i) assess interventions in low- and middle-income countries that ex- plicitly analyse paediatric-inclusive and paediatric-distinct needs and outcomes, (ii) use mixed-method ap- proaches that can assess the pathways linking context-dependent factors with outcomes, (iii) use longitudinal evalua- tions that investigate the sustainability of the effectiveness and benefits of interventions and the potential burdens posed by interventions, and (iv) incor- porate and address cost–effectiveness, resource implications and potential scalability. Our findings indicate the potential usefulness of diverse interventions to in- crease the rate of treatment completion among paediatric tuberculosis patients and improve outcomes in resource-poor settings. ■ Acknowledgements We thank members of the International Society of Paediatric Oncology’s Paedi- atric Oncology in Developing Countries Abandonment of Treatment Working Group (SIOP PODC). Competing interests: None declared. Fig. 4. Contextual framework showing factors that may promote or threaten adherence to treatment for paediatric tuberculosis in low- and middle-income countries Patient Therapy Condition Social/Economic Health system Social support; mobilized community resources; coordinated multidisciplinary care Family education; adherence counselling; contracting; patient empowerment Support for food, transportation, housing and daily living; lower out-of-pocket expenses Low literacy; limited education; limited self-efficacy Food insecurity; distance from health centre; limited and/or missed income Patient-centred service locations and times; tracer systems; adherence-sensitive staff and management process Limited provider knowledge; inadequate communication or engagement; complex regimens Social stigma of condition or treatment; inadequate therapeutic alliance; mistrust Education Psychosocial Health system Care delivery Social protection Notes: The central circle, which contains the adherence dimensions used by the World Health Organization, is surrounded by the five main categories of relevant interventions. The factors that may promote treatment adherence are shown in green boxes and factors that may threaten treatment adherence are shown in white boxes. Therapeutic alliance refers to relationship-building between providers and patients. Bull World Health Organ 2015;93:700–711B| doi: http://dx.doi.org/10.2471/BLT.14.147231 709 Systematic reviews Improving treatment adherence in paediatric tuberculosisMeaghann S Weaver صخلم ةطسوتمو ةدودمح نادلبلا في لافطلأا ىدل لسلا ضرمب صالخا جلاعلاب مازتللاا ىوتسم ينستح لىإ ةيعاسلا تلاخدتلا يعيمتج ليلتحو ةيجهنم ةعجارم :لخدلا تايلمع لىع ةبتترلما جئاتنلاو ميدقتلاو ميمصتلا مييقت ضرغلا ّلسلا تلاالح مدقلما جلاعلاب مازتللاا ىوتسم عفر ضرغب لخدتلا راطإ عضوو لخدلا ةطسوتمو ةدودمح نادلبلا في لافطلأا ىدل .عونلا اذه نم لخدتلا تايلمعل يقايس و PubMed تانايب دعاوق في اًثحب انيرجأ دقل ةقيرطلا /رياني لوأ ينب ام ةترفلا في ةروشنلما ريراقتلا دايجلإ Cochrane ماع نم لولأا نوناك/برمسيد لوأو 2003 ماع نم نياثلا نوناك مازتللاا ىوتسم ينستح لىإ ةيعاسلا لخدتلا تايلمع لوح 2013 مهرماعأ لقت ضىرم تنمضت يتلاو ّلسلا ضرلم مدقلما جلاعلاب ةضفخنم وأ ةدودمح نادلب في نوشيعي اوناك نمم اًماع 20 نع ةلص تاذ نوكت نأ لمتيح يتلا تلااقلما دايجإ متي يكلو .لخدلا انيلوت دقف ،لافطلأا نم ضىرلماب ةطبترلما تلاصحلما اهصقني نكل سناجتلا مدع صرنع مييقتب انمق ماك .تاساردلا يفلؤمب لاصتلاا لىع فوقولا لجأ نمو .زايحنلاا رطخو )Heterogeneity( قيقتحو جلاعلا لماكإ ينب ام ةفيلوتلا يأ – جلاعلا حاجن ىوتسم راثلآل )Meta-analysis( اًيعيمتج ًلايلتح انيرجأ دقف – ءافشلا تاءارجإ ريوطتل جايتحلاا بناوج ديدحتب لياتلاب انمقو .ةيئاوشعلا .لخدتلا ليلحتلا لىع اًصرح اًدلب 11 في ةسارد 15 انّمض دقل جئاتنلا 11 تلهأت دقف تاسردلا هذه ينب نمو ،جئاتنلا هذله يعونلا تلاخدتلا ينب نمو .ًلافط 1279 تلثم يعيمجتلا ليلحتلل ةسارد ناتنثا تزكر ،ةسارد 15 اهددع غلابلا تاساردلا في ةفوصولما سيفنلا معدلا لىع ةدحاو ةسارد زيكرت بصنا مايف ،ةيعوتلا لىع ميدقت لىع زيكترلا لىإ تا�سارد عبس تهتجاو ،يعماتجلااو تدرفناو ،ةيحصلا مظنلا لىع تاسارد عبرأ تزكر مايف ،ةياعرلا لافطلأا ققحو .ةيلالما تادماتعلاا لىع زيكترلاب ةدحاو ةسارد نم لىعأ اًبسن يجلاعلا لخدتلا ةبرجتل عضالخا قيرفلا في ينلجسلما ةيعجرلما تاعومجلما في ينلجسلما لافطلأاب ًةنراقم جلاعلا حاجن :95% اهرادقم ةيحجرأ ةبسنبو ؛3.02 تغلب لماتحا ةبسنب( ،انتلايلتح نم ةصلختسلما جئاتنلا لىع دماتعلاابو .)4.15–2.19 لماتكا لىع تعجش يتلا لماوعلا لىع موقي اًيقايس اًراطإ انعضو دقف .هلماتكا تدده وأ جلاعلا ىوتسم ينستح لىإ ةيعاسلا ةفلتخلما تلاخدتلا نأ ينبتي جاتنتسلاا نادلبلا في لافطلأا ىدل ّلسلا ضرمب صالخا جلاعلاب مازتللاا .ىودج تاذو ةلاعف تناك لخدلا ةطسوتمو ةدودمح 摘要 在中低收入国家采取干预措施以提高儿科结核病的治疗依从性 ： 系统评审和元分析 目的 针对在中低收入国家提高儿科结核病的治疗依从 性，评估干预措施的设计、完成和效果，并为该种干 预措施制定一个关联性框架。 方 法 我们在 PubMed 和 Cochrane 数据库中搜索了 于 2003 年 1 月 1 日至 2013 年 12 月 1 日之间发表的报 告，这些报告与提高结核病治疗依从性的干预措施有 关，其中包括居住在中低收入国家且年龄不满 20 岁 的患者。 相关文章可能缺少关于儿科的结果，为此， 我们联系了研究的作者。 我们对异质性和偏倚风险进 行了评估。 为了评估治疗成功率——即，完治率与治 愈率相结合——我们采用随机效应开展了元分析。 我 们确认了有必要对干预实践作出改善的领域。 结果 我们将 11 个国家的 15 项研究纳入定向分析，其 中 11 项符合元分析的条件——其代表 1279 名儿童。 关于 15 项研究中描述的干预措施，其中两项侧重于 教育，一项侧重于社会心理支持，七项侧重于医疗保 健服务，四项侧重于卫生体系，另外一项侧重于财政 拨款。 与对照组的儿童相比，经干预措施庇佑的儿童 具有更高的治疗成功率（比值比： 3.02；95% 置信区间： 2.19–4.15）。通过我们分析得出的结果，我们围绕促 进或威胁完治率的因素制定了框架。 结论 在中低收入国家为提高儿科结核病的治疗依从性 而采取的多种干预措施显得可行、有效。 Résumé Interventions pour améliorer l’observance thérapeutique dans le traitement de la tuberculose chez l’enfant dans les pays à revenu faible et intermédiaire: revue systématique et méta-analyse Objectif Évaluer la conception, la mise en œuvre et les résultats des interventions visant à améliorer l’observance thérapeutique dans le traitement de la tuberculose chez l’enfant dans les pays à revenu faible et intermédiaire et élaborer un cadre contextuel pour ce type d’interventions. Méthodes Nous avons fait des recherches dans les bases de données PubMed et Cochrane pour trouver des rapports publiés entre le 1er janvier 2003 et le 1er décembre 2013 sur des interventions visant à améliorer l’observance thérapeutique dans le traitement de la tuberculose chez des patients de moins de vingt ans vivant dans des pays à revenu faible ou intermédiaire. Pour les articles potentiellement pertinents qui omettaient de citer spécifiquement les résultats chez l’enfant, nous avons contacté les auteurs. Nous avons évalué l’hétérogénéité et le risque de biais. Pour évaluer la réussite d’un traitement (c’est-à-dire: combinaison de l’achèvement du traitement et de la guérison), nous avons effectué une méta-analyse à effets aléatoires. Nous avons également identifié les points à améliorer en vue d’optimiser les programmes d’intervention. Résultats Pour notre analyse qualitative, nous avons intégré quinze études, menées dans onze pays. Sur ces études, onze ont pu être retenues pour la méta-analyse (représentant 1.279 enfants). Concernant les interventions décrites dans les quinze études, deux ciblaient l’éducation, une le soutien psychosocial, sept la délivrance des soins, quatre les systèmes de santé et une le soutien financier. Des taux de succès du traitement plus élevés ont été constatés chez les enfants qui ont bénéficié des interventions, comparativement aux enfants des Bull World Health Organ 2015;93:700–711B| doi: http://dx.doi.org/10.2471/BLT.14.147231710 Systematic reviews Improving treatment adherence in paediatric tuberculosis Meaghann S Weaver groupes témoins (rapport des cotes: 3,02; intervalle de confiance de 95%: 2,19-4,15). À partir des résultats de nos analyses, nous avons conçu un cadre autour des facteurs ayant favorisé ou entravé l’achèvement du traitement. Conclusion Plusieurs interventions visant à améliorer l’observance thérapeutique dans le traitement de la tuberculose chez l’enfant semblent être à la fois réalisables et efficaces dans les pays à revenu faible et intermédiaire. Резюме Медицинские вмешательства, способствующие соблюдению режима лечения туберкулеза у детей в странах с низким и средним уровнем доходов: систематический обзор и метаанализ Цель Оценить структуру, реализацию и результаты медицинских вмешательств, способствующих соблюдению режима лечения детского туберкулеза в странах с низким и средним уровнем доходов населения, и разработать контекстуальную схему таких вмешательств. Методы Был проведен поиск отчетов по медицинским вмешательствам, способствующим соблюдению режима лечения детского туберкулеза, опубликованным с 1 января 2003 г. по 1 декабря 2013 г. Поиск проводился в базах данных PubMed и Кокрановской библиотеки. Нас интересовали пациенты моложе 20 лет, проживающие в странах с низким или средним уровнем доходов. Если в потенциально релевантной статье было недостаточно данных по результатам лечения пациентов детского возраста, мы обращались за сведениями к авторам исследований. Была выполнена оценка гетерогенности данных и риска системной ошибки. Для оценки успешности лечения, т. е. завершения курса лечения и выздоровления пациента, был проведен метаанализ случайных воздействий. Были выявлены области, в которых практики медицинских вмешательств нуждаются в улучшении. Результаты В количественный анализ было включено 15 исследований, выполненных в 11 странах. Из них 11 исследований было отобрано для метаанализа. В этих исследованиях было представлено 1279 детей. Среди описанных в них вмешательств две программы делали акцент на обучении, одна — на психологической поддержке, семь — на эффективности ухода, четыре касались систем здравоохранения и одна — предоставления финансирования. В тех группах, где проводилось вмешательство, дети имели более высокий показатель успешного лечения по сравнению с детьми из контрольных групп (отношение шансов: 3,02; 95% доверительный интервал: 2,19–4,15). Используя результаты наших анализов, мы разработали модель на базе факторов, которые способствуют или препятствуют завершению лечения. Вывод Различного рода вмешательства, способствующие соблюдению режима лечения туберкулеза у детей, оказались осуществимой и эффективной стратегией для стран с низким и средним уровнем доходов населения. Resumen Intervenciones para mejorar el cumplimiento del tratamiento de la tuberculosis pediátrica en los países de ingresos bajos y medios: una revisión sistemática y un metanálisis Objetivo Evaluar el diseño, la prestación y los resultados de las intervenciones para mejorar el cumplimiento del tratamiento de la tuberculosis pediátrica en países de ingresos bajos y medios y desarrollar un marco contextual para tales intervenciones. Métodos Se realizaron búsquedas en las bases de datos PubMed y Cochrane para encontrar informes sobre las intervenciones para mejorar el cumplimiento del tratamiento de la tuberculosis publicados entre el 1 de enero de 2003 y el 1 de diciembre de 2013 que incluyeran pacientes menores de 20 años que vivieran en países de ingresos bajos o medios. Se contactó con los autores de los estudios con artículos relevantes que carecían de resultados pediátricos. Se evaluó la heterogeneidad y el riesgo de sesgo. Se llevaron a cabo metanálisis de efectos aleatorios para evaluar el éxito del tratamiento, es decir, la combinación de finalización del tratamiento y cura. Se identificaron áreas que necesitaban una mejora de las prácticas de intervención. Resultados Se incluyeron 15 estudios en 11 países para el análisis cualitativo y, de esos estudios, 11 cumplieron los requisitos para el metanálisis: una representación de 1279 niños. De las intervenciones descritas en los 15 estudios, dos se centraban en la educación, uno en el apoyo psicosocial, siete en la prestación de asistencia, cuatro en los sistemas de salud y uno en las dotaciones financieras. Los niños en el brazo de intervención tenían una tasa mayor de éxito del tratamiento en comparación con aquellos en grupos de control (razón de posibilidades: 3,02; intervalo de confianza del 95%: 2.19–4.15). Utilizando los resultados en los análisis, se desarrolló un marco alrededor de los factores que promovían o amenazaban la finalización del tratamiento. Conclusión Varias intervenciones para mejorar el cumplimiento del tratamiento de la tuberculosis pediátrica parecen tanto viables como efectivas en países de ingresos bajos y medios. References 1. Guidance for national tuberculosis programmes on the management of tuberculosis in children. 2nd ed. Geneva: World Health Organization; 2014. 2. Adherence to long-term therapies: evidence for action. Geneva: World Health Organization; 2003. 3. Pefura Yone EW, Kengne AP, Kuaban C. Incidence, time and determinants of tuberculosis treatment default in Yaounde, Cameroon: a retrospective hospital register-based cohort study. BMJ Open. 2011;1(2):e000289. doi: http://dx.doi.org/10.1136/bmjopen-2011-000289 PMID: 22116091 4. Higgins JPT, Green S, editors. Cochrane handbook for systematic reviews of interventions version 5.1.0 [updated March 2011]. Oxford: The Cochrane Collaboration; 2011. Available from: www.cochrane-handbook.org [cited 2015 May 13]. 5. Enhancing the quality and transparency of health research [Internet]. Oxford: Equator Network; 2013. Available from: http://www.equator- network.org/ [cited 2014 Apr 25]. 6. Armijo-Olivo S, Stiles CR, Hagen NA, Biondo PD, Cummings GG. Assessment of study quality for systematic reviews: a comparison of the Cochrane Collaboration Risk of Bias Tool and the Effective Public Health Practice Project Quality Assessment Tool: methodological research. J Eval Clin Pract. 2012 Feb;18(1):12–8. doi: http://dx.doi.org/10.1111/j.1365- 2753.2010.01516.x PMID: 20698919 Bull World Health Organ 2015;93:700–711B| doi: http://dx.doi.org/10.2471/BLT.14.147231 711 Systematic reviews Improving treatment adherence in paediatric tuberculosisMeaghann S Weaver 7. Petticrew M, Anderson L, Elder R, Grimshaw J, Hopkins D, Hahn R, et al. Complex interventions and their implications for systematic reviews: a pragmatic approach. J Clin Epidemiol. 2013 Nov;66(11):1209–14. doi: http:// dx.doi.org/10.1016/j.jclinepi.2013.06.004 PMID: 23953085 8. Critical appraisal skills programme [Internet]. Oxford: CASP UK; 2014. Available from: http://www.casp-uk.net/ [cited 2014 Apr 22]. 9. Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med. 2009 Jul 21;6(7):e1000097. doi: http://dx.doi. org/10.1371/journal.pmed.1000097 PMID: 19621072 10. Satti H, McLaughlin MM, Omotayo DB, Keshavjee S, Becerra MC, Mukherjee JS, et al. Outcomes of comprehensive care for children empirically treated for multidrug-resistant tuberculosis in a setting of high HIV prevalence. PLoS ONE. 2012;7(5):e37114. doi: http://dx.doi.org/10.1371/journal. pone.0037114 PMID: 22629356 11. Heck MA, da Costa JS, Nunes MF. Prevalência de abandono do tratamento da tuberculose e fatores associados no município de Sapucaia do Sul (RS), Brasil, 2000-2008. Rev Bras Epidemiol. 2011 Sep;14(3):478–85. Portuguese. doi: http://dx.doi.org/10.1590/S1415-790X2011000300012 PMID: 15675553 12. Mathew A, Binks C, Kuruvilla J, Davies PD. A comparison of two methods of undertaking directly observed therapy in a rural Indian setting. Int J Tuberc Lung Dis. 2005 Jan;9(1):69–74. PMID: 15675553 13. Cantalice Filho JP. Efeito do incentivo alimentício sobre o desfecho do tratamento de pacientes com tuberculose em uma unidade primária de saúde no município de Duque de Caxias, Rio de Janeiro. J Bras Pneumol. 2009 Oct;35(10):992–7. Portuguese.doi: http://dx.doi.org/10.1590/S1806- 37132009001000008 PMID: 19918632 14. Marques AM, da Cunha RV. A medicação assistida e os índices de cura de tuberculose e de abandono de tratamento na população indígena Guaraní-Kaiwá no Município de Dourados, Mato Grosso do Sul, Brasil. Cad Saude Publica. 2003 Sep-Oct;19(5):1405–11. Portuguese.doi: http://dx.doi. org/10.1590/S0102-311X2003000500019 PMID: 14666222 15. Anuwatnonthakate A, Limsomboon P, Nateniyom S, Wattanaamornkiat W, Komsakorn S, Moolphate S, et al. Directly observed therapy and improved tuberculosis treatment outcomes in Thailand. PLoS ONE. 2008;3(8):e3089. doi: http://dx.doi.org/10.1371/journal.pone.0003089 PMID: 18769479 16. Khortwong P, Kaewkungwal J. Thai health education program for improving TB migrant’s compliance. J Med Assoc Thai. 2013 Mar;96(3):365–73. PMID: 23539943 17. Badar D, Ohkado A, Naeem M, Khurshid-ul-Zaman S, Tsukamoto M. Strengthening tuberculosis patient referral mechanisms among health facilities in Punjab, Pakistan. Int J Tuberc Lung Dis. 2011 Oct;15(10):1362–6. doi: http://dx.doi.org/10.5588/ijtld.10.0620 PMID: 22283896 18. Datiko DG, Lindtjørn B. Health extension workers improve tuberculosis case detection and treatment success in southern Ethiopia: a community randomized trial. PLoS ONE. 2009;4(5):e5443. doi: http://dx.doi.org/10.1371/ journal.pone.0005443 PMID: 19424460 19. Demissie M, Getahun H, Lindtjørn B. Community tuberculosis care through “TB clubs” in rural North Ethiopia. Soc Sci Med. 2003 May;56(10):2009–18. doi: http://dx.doi.org/10.1016/S0277-9536(02)00182-X PMID: 12697193 20. van den Boogaard J, Lyimo R, Irongo CF, Boeree MJ, Schaalma H, Aarnoutse RE, et al. Community vs. facility-based directly observed treatment for tuberculosis in Tanzania’s Kilimanjaro Region. Int J Tuberc Lung Dis. 2009 Dec;13(12):1524–9. PMID: 19919771 21. Keus K, Houston S, Melaku Y, Burling S. Field research in humanitarian medical programmes. Treatment of a cohort of tuberculosis patients using the Manyatta regimen in a conflict zone in South Sudan. Trans R Soc Trop Med Hyg. 2003 Nov-Dec;97(6):614–8. doi: http://dx.doi.org/10.1016/S0035- 9203(03)80048-2 PMID: 16134258 22. Ong’ang’o JR, Mwachari C, Kipruto H, Karanja S. The effects on tuberculosis treatment adherence from utilising community health workers: a comparison of selected rural and urban settings in Kenya. PLoS ONE. 2014;9(2):e88937. doi: http://dx.doi.org/10.1371/journal.pone.0088937 PMID: 24558452 23. Lee S, Khan OF, Seo JH, Kim DY, Park KH, Jung SI, et al. Impact of physician’s education on adherence to tuberculosis treatment for patients of low socioeconomic status in Bangladesh. Chonnam Med J. 2013 Apr;49(1):27– 30. doi: http://dx.doi.org/10.4068/cmj.2013.49.1.27 PMID: 23678474 24. Lönnroth K, Aung T, Maung W, Kluge H, Uplekar M. Social franchising of TB care through private GPs in Myanmar: an assessment of treatment results, access, equity and financial protection. Health Policy Plan. 2007 May;22(3):156–66. doi: http://dx.doi.org/10.1093/heapol/czm007 PMID: 17434870 25. Higgins JPT, Altman DG, Sterne JAC, on behalf of the Cochrane Statistical Methods Group and the Cochrane Bias Methods Group. Chapter 8: Assessing risk of bias in included studies. In: Higgins JPT, Green S, editors. Cochrane handbook for systematic reviews of interventions. London: The Cochrane Collaboration; 2011. pp. 8.13–8.18. 26. Roter DL, Hall JA, Merisca R, Nordstrom B, Cretin D, Svarstad B. Effectiveness of interventions to improve patient compliance: a meta-analysis. Med Care. 1998 Aug;36(8):1138–61. doi: http://dx.doi.org/10.1097/00005650- 199808000-00004 PMID: 9708588 Bull World Health Organ 2015;93:700–711B| doi: http://dx.doi.org/10.2471/BLT.14.147231 711A Systematic reviews Improving treatment adherence in paediatric tuberculosisMeaghann S Weaver Box 1. Search strategy to identify studies on interventions to improve adherence to treatment for paediatric tuberculosis in low- and middle-income countries (“low income economies” OR “lower middle income economies” OR “middle income economies” OR “developing countries”[MeSH Terms] OR (“developing”[All Fields] AND “countries”[All Fields]) OR “developing countries”[All Fields]) OR (“developing countries”[MeSH Terms] OR (“developing”[All Fields] AND “countries”[All Fields]) OR “developing countries”[All Fields] OR (“developing”[All Fields] AND “country”[All Fields]) OR “developing country”[All Fields]) OR (“developing countries”[MeSH Terms] OR (“developing”[All Fields] AND “countries”[All Fields]) OR “developing countries”[All Fields] OR (“underdeveloped”[All Fields] AND “countries”[All Fields]) OR “underdeveloped countries”[All Fields]) OR (“developing countries”[MeSH Terms] OR (“developing”[All Fields] AND “countries”[All Fields]) OR “developing countries”[All Fields] OR (“underdeveloped”[All Fields] AND “country”[All Fields]) OR “underdeveloped country”[All Fields]) OR (emergent[All Fields] AND countries[All Fields]) OR (emergent[All Fields] AND country[All Fields]) OR (“developing countries”[MeSH Terms] OR (“developing”[All Fields] AND “countries”[All Fields]) OR “developing countries”[All Fields] OR (“developing”[All Fields] AND “nation”[All Fields]) OR “developing nation”[All Fields]) OR (underdeveloped[All Fields] AND “nation”[All Fields])) OR (emergent[All Fields] AND “nation”[All Fields]) OR ((“poverty”[MeSH Terms] OR “poverty”[All Fields] OR (“low”[All Fields] AND “income”[All Fields]) OR “low income”[All Fields]) AND countries[All Fields]) OR ((“poverty”[MeSH Terms] OR “poverty”[All Fields] OR (“low”[All Fields] AND “income”[All Fields]) OR “low income”[All Fields]) AND country[All Fields]) OR angola OR Fij OR palau OR albania OR gabon OR panama OR algeria OR grenada OR peru OR american samoa OR hungary OR romania OR argentina OR iran OR serbia OR azerbaijan OR iraq OR seychelles OR belarus OR jamaica OR south africa OR belize OR jordan OR st. lucia OR bosnia and herzegovina OR kazakhstan OR st. vincent and the grenadines OR botswana OR lebanon OR suriname OR brazil OR libya OR thailand OR bulgaria OR macedonia, fyr OR tonga OR china OR malaysia OR tunisia OR colombia OR maldives OR turkey OR costa rica OR marshall islands OR turkmenistan OR cuba OR mauritius OR tuvalu OR dominica OR mexico OR venezuela, rb OR dominican republic OR montenegro OR ecuador OR namibia OR armenia OR india OR samoa OR bhutan OR kiribati OR sao tome and principe OR bolivia OR kosovo OR senegal OR cameroon OR Lao OR solomon islands OR cape verde OR lesotho OR sri lanka OR congo OR mauritania OR sudan OR cote d’ivoire OR ivory coast OR micronesia OR swaziland OR djibouti OR moldova OR syria OR egypt OR mongolia OR timor OR el salvador OR morocco OR ukraine OR georgia OR nicaragua OR uzbekistan OR ghana OR nigeria OR vanuatu OR guatemala OR pakistan OR vietnam OR guyana OR papua new guinea OR west bank OR gaza OR honduras OR paraguay OR yemen OR indonesia OR philippines OR zambia OR afghanistan OR gambia OR myanmar OR bangladesh OR guinea OR nepal OR benin OR niger OR burkina faso OR haiti OR rwanda OR burundi OR kenya OR sierra leone OR cambodia OR korea OR somalia OR central african republic OR kyrgyz OR sudan OR chad OR liberia OR tajikistan OR comoros OR madagascar OR tanzania OR congo OR malawi OR togo OR eritrea OR mali OR uganda OR ethiopia OR mozambique OR zimbabwe)) AND tuberculosis[MeSH Major Topic] AND (“Health Education”[Mesh] OR “Counseling”[Mesh] OR “Directive Counseling”[Mesh] OR “Health Promotion”[Mesh] OR “Reminder Systems”[Mesh] OR “Directly Observed Therapy”[Mesh] OR “Social Support”[Mesh] OR “Contracts”[Mesh] OR “Decision Support Techniques”[Mesh] OR intervention OR treatment OR outcome) AND (study OR trial) AND (“Treatment Refusal”[Mesh] OR “Patient Participation”[Mesh] OR “Patient Dropouts”[Mesh] OR “Patient Compliance”[Mesh] OR “Motivation”[Mesh] OR “Cooperative Behavior”[Mesh]) OR “Refusal to Treat”[Mesh]) OR “Medication Adherence”[Mesh] OR medication adherence OR nonadherence OR non-adherence OR compliance OR noncompliance OR abandonment of treatment OR abandonment of therapy OR treatment abandonment OR therapy abandonment OR treatment default OR lost to follow-up OR loss to follow up OR default* OR against medical advice OR abscond* OR refusal OR stop* treatment OR (interrupt* AND treatment) OR (treatment AND discontinu*) OR (treatment AND continu*) OR failure to complete treatment OR incomplete treatment OR treatment maintenance OR no show OR retention of care OR run away OR attrition)) AND (“last 10 years”[PDat] AND Humans[Mesh] AND (infant[MeSH] OR child[MeSH] OR adolescent[MeSH] OR “young adult”[MeSH]) NOT “case reports”[Publication Type]) NOT “review”[Publication Type] Table 4. Assessment of non-randomized studies on interventions to improve adherence to treatment for paediatric tuberculosis in low- and middle-income countries Study Selection bias Study design Confounders Blinding Data collec- tion method Withdrawals and dropouts Global rating Anuwatnonthakate et al.15 Moderate Moderate Strong Weak Weak Strong Weak Heck et al.11 Moderate Weak Weak Weak Weak Moderate Weak Lee et al.23 Moderate Moderate Strong Not clear Weak Moderate Moderate Marques and da Cunha14 Not clear Moderate Weak Not clear Weak Weak Weak Ong’ang’o et al.22 Moderate Moderate Strong Moderate Weak Strong Moderate Satti et al.10 Moderate Weak Weak Not clear Weak Strong Weak van den Boogaard et al.20 Moderate Moderate Moderate Weak Weak Moderate Weak Badar et al.17 Not clear Weak Weak Weak Weak Weak Weak Cantalice Filho13 Moderate Moderate Moderate Not clear Weak Weak Weak Keus et al.21 Moderate Weak Weak Moderate Weak Strong Weak Lönnroth et al.24 Weak Weak Weak Not clear Weak Strong Weak Note: Assessed by using Effective Public Health Practice Project Quality Assessment Tool for Quantitative Studies. Bull World Health Organ 2015;93:700–711B| doi: http://dx.doi.org/10.2471/BLT.14.147231711B Systematic reviews Improving treatment adherence in paediatric tuberculosis Meaghann S Weaver Table 5. Risk of bias in randomized control and quasi-randomized control studies on interventions to improve adherence to treatment for paediatric tuberculosis in low- and middle-income countries Study Random sequence generation Allocation concealment Blinding of participants and personnel Blinding of outcome assessors Incomplete outcome data Selective reporting Other bias Datiko and Lindtjørn18 Low High Low High Low Low Low Demissie et al.19 High Unclear Unclear High Low Unclear Low Khortwong and Kaewkungwal16 High Unclear Unclear High Low Unclear Low Mathew et al.12 High Unclear High High High Unclear Low Note: Assessed by using Cochrane criteria for judging risk of bias.25 Fig. 3. Funnel plot to evaluate publication bias of studies on interventions to improve adherence to treatment for paediatric tuberculosis in low- and middle-income countries OR 0.01 0.1 1 10 100 SE (lo g[ OR ]) 0 0.5 1 1.5 2 OR: odds ratio; SE: standard error. Note: The dashed line represents the summary odds ratio derived from the meta-analysis. Odds ratios have been plotted on a logarithmic scale.