Improve Perinatal Pakistan

8/13/2019 Improve Perinatal Pakistan

1/10

Articles

www.thelancet.com Published online January 15, 2011 DOI:10.1016/S0140-6736(10)62274-X 1

IntroductionAlthough there has been some reduction in global childmortality, 1 there is little evidence to suggest that newborndeaths have reduced signicantly from the estimatedyearly gure of 4 million deaths in 2000. 2 Pakistan hasone of the highest rates of mortality in children youngerthan 5 years in south Asia (94 deaths per 1000 livebirths),and many (57%) of these deaths occur in the newbornperiod, most in the rst few days after delivery. 3 There aresignicant urbanrural differentials in neonatal mortalityrates (48 newborn deaths per 1000 livebirths in urbanareas vs 55 per 1000 in rural areas) and overall 65% ofbirths take place at home (43% urban vs 74% rural).3 Morethan half (52%) of these births are in the hands oftraditional birth attendants ( Dais),4,5 who are generallyuntrained and who charge for their services.

To help to strengthen primary care and preventive

services, the government of Pakistan introduced theNational Program for Family Planning and PrimaryHealth Care, commonly called the lady health workers(LHW) programme, in 1994. 6 LHWs are mostly youngwomen, resident in the local communities, with at least8 years of formal schooling, who are trained for15 months to deliver care in community settings eitherthrough home visits or from their residences, knownas health homes. Each LHW is responsible for apopulation of about 10001500 and provides antenatalcare, contraceptive advice, growth monitoring, andimmunisation services. 7 The emphasis in the existingcurriculum is on recognition and referral rather thanhome-based management of common neonatalproblems, and two recent evaluations have concluded

Improvement of perinatal and newborn care in ruralPakistan through community-based strategies:a cluster-randomised effectiveness trial Zulqar A Bhutta, Sajid Soo, Simon Cousens, Shah Mohammad, Zahid A Memon, Imran Ali, Asher Feroze, Farrukh Raza, Amanullah Khan,Steve Wall, Jose Martines

Summary Background Newborn deaths account for 57% of deaths in children younger than 5 years in Pakistan. Although a largeprogramme of trained lady health workers (LHWs) exists, the effectiveness of this training on newborn outcomes hasnot been studied. We aimed to evaluate the effectiveness of a community-based intervention package, principallydelivered through LHWs working with traditional birth attendants and community health committees, for reductionof perinatal and neonatal mortality in a rural district of Pakistan.

Methods We undertook a cluster randomised trial between February, 2006, and March, 2008, in Hala and Matiarisubdistricts, Pakistan. Catchment areas of primary care facilities and all affi liated LHWs were used to dene clusters,which were allocated to intervention and control groups by restricted, stratied randomisation. The intervention packagedelivered by LHWs through group sessions consisted of promotion of antenatal care and maternal health education, useof clean delivery kits, facility births, immediate newborn care, identication of danger signs, and promotion ofcareseeking; control clusters received routine care. Independent data collectors undertook quarterly householdsurveillance to capture data for births, deaths, and household practices related to maternal and newborn care. Datacollectors were masked to cluster allocation; those analysing data were not. The primary outcome was perinatal and all-cause neonatal mortality. Analysis was by intention to treat. This trial is registered, ISRCTN16247511.

Findings 16 clusters were assigned to intervention (23 353 households, 12 391 total births) and control groups(23 768 households, 11 443 total births). LHWs in the intervention clusters were able to undertake 4428 (63%) of

7084 planned group sessions, but were only able to visit 2943 neonates (24%) of a total 12 028 livebirths in theircatchment villages. Stillbirths were reduced in intervention clusters (391 stillbirths per 1000 total births) comparedwith control (487 per 1000; risk ratio [RR] 079, 95% CI 068092; p=0006). The neonatal mortality rate was430 deaths per 1000 livebirths in intervention clusters compared with 491 per 1000 in control groups (RR 085,076096; p=002).

Interpretation Our results support the scale-up of preventive and promotive maternal and newborn interventionsthrough community health workers and emphasise the need for attention to issues of programme management andcoverage for such initiatives to achieve maximum potential.

Funding WHO; Saving Newborn Lives Program of Save the Children USA, funded by the Bill & Melinda GatesFoundation.

PublishedOnline January 15, 2011DOI:10.1016/S0140-6736(10)62274-X

See Online/CommentDOI:10.1016/S0140-6736(11)60058-5

Division of Women and ChildHealth, Aga Khan University,Karachi, Pakistan(Prof Z A Bhutta PhD,S Soo FCPS,S Mohammad MPH,Z A Memon MSc, I Ali MSc,A Feroze MSc, F Raza MPH);London School of Hygiene andTropical Medicine, London, UK (S Cousens DipMathStats);Saving Newborn Lives, Savethe Children USA, Washington,DC, USA (A Khan MPH,S Wall MD); and Department ofChild and Adolescent Health,WHO, Geneva, Switzerland (J Martines PhD)

Correspondence to:Prof Zulqar A Bhutta, HuseinLalji Dewraj Professor andFounding Chair, Division ofWomen and Child Health,Aga Khan University,Karachi 74800, [email protected]


2/10

Articles

2 www.thelancet.com Published online January 15, 2011 DOI:10.1016/S0140-6736(10)62274-X

that they do provide reasonable primary care promotiveand preventive services. 8,9

The potential of community-based interventions toreduce newborn morbidity and mortality is wellrecognised. 10,11 Such interventions include communityhealth workers (CHWs) delivering preventive andtherapeutic interventions such as antibiotics at home, 12,13 community mobilisation through womens supportgroups 14,15 or community mobilisers working through

individual and group sessions, 16 and community-based

interventions delivered through non-governmentalorganisations 17 or community volunteers. 18 Commonfeatures of these interventions include civil societyengagement, exibility of approaches, community volun-teers, social mobilisers, or CHWs dedicated to thedesignated tasks through home visits or group sessions.However, despite the success of these projects (largelyundertaken as effi cacy trials), translation of theseinterventions into packages of care and complex inter-ventions that can be delivered within public healthsystems at scale remains a major challenge. 19 Most ofthese studies were fairly small and none principally usedthe public sector, making translation of this evidence topublic health systems diffi cult.

We undertook the rst effectiveness trial of a package ofpreventive maternal and newborn care strategies in ruralPakistan, delivered through public sector LHWs incollaboration with voluntary community health commit-tees (CHCs) and Dais. We have previously reportedndings from the development and pilot testing phase ofthe project, 16 which showed the feasibility of delivering thepackage of care through government sector LHWs.

MethodsStudy designWe undertook a cluster randomised trial in rural Sindhin southern Pakistan. The Hala and Matiari subdistricts(hereafter called Hala ) are located 250 km north ofKarachi and include two towns and 1400 villages, withan offi cial population of 06 million. We undertook asurvey of all facilities (22 basic health units [BHUs], tworural health centres [RHCs] of which one was upgradedto a referral hospital status in 2006, and a districtreferral hospital) and LHWs in the area. 437 LHWs wereavailable in the district, of 500 approved positions. Thedistribution of these LHWs was variable, with severalvillages covered by non-residential LHWs who livedelsewhere and commuted to work. There were twodistrict referral hospitals located in Hala town andMatiari, with one paediatrician, two obstetricians, andseveral medical offi cers. Several mid wives were available

in the area, with most working within the private sector.Webappendix p 1 shows further details of the study areaand local health systems.

We developed an intervention package for promotiveand preventive newborn care in collaboration with theDirectorate of Health, Government of Sindh, which hasbeen described previously. 16 The intervention consistedof training of LHWs and Dais and promotion of liaisonbetween them, together with facilitation of the creationof voluntary CHCs to promote maternal and newborncare in their villages. The intervention package waspilot tested in four clusters (with four control clusters)in the district from 2003 to 2005 to rene methods andto assess acceptability by the population. Somemodications of the LHW training curriculum,

Panel : Changes made to the training curriculum for lady health workers, communityengagement, data collection, and surveillance on the basis of the pilot phase 16

Training in preventive newborn careAn augmented training package was developed for intervention lady health workers(LHWs) in consultation with the LHW programme (table 1) and implemented by regularLHW programme supervisors and trainers. The standard LHW training took place during aperiod of 15 months, including 3 months of didactic training and monthly refreshersessions of 1 day each. The study intervention package only added an extra 6 days to LHWtraining. This extra training encouraged LHWs to identify all pregnant women in their area,provide basic antenatal care (including rest and nutrition counselling, screening forcommon illnesses, iron folate and tetanus toxoid administration) and work with traditionalbirth attendants (Dais) to identify births. LHWs were trained in mouth-to-mouthresuscitation, but no resuscitation equipment or injectable antibiotics were provided tothem. Clean delivery kits were provided to LHWs in both intervention and control clustersand were also available for purchase in local pharmacies. LHWs were encouraged to visitmothers twice during pregnancy and within 24 h of birth. Additionally, visits wereencouraged on days 3, 7, 14, and 28 after birth. LHWs were reimbursed for any additionaltravel costs arising from the intervention and training, but did not receive any additionalsalary or other nancial motivation.

Dai training for basic newborn careWith the assistance of Directorate of Health staff, we developed a 3-day training programmefor Dais in basic newborn care including basic resuscitation (through skin rubbing, soleicking, and immediate newborn care). No remuneration, commodities, or monetaryincentives were provided to theDais for these training sessions apart from transportation tobasic health units and meals or refreshments. Attendance was entirely voluntary(webappendix p 2).Daiswere informed about these training sessions through the LHWs andcommunity health committees (CHCs) and were also encouraged to attend the LHW-ledcommunity education sessions. No training sessions were held in the control clusters.

Community organisation, mobilisation, and group education sessionsThree community mobilisers from Aga Khan University (Karachi, Pakistan) assisted theLHWs in identifying community volunteers in intervention communities. Communityvolunteers were encouraged to form CHCs in liaison with LHWs with the aim ofpromoting maternal and newborn care in their villages. In addition to advocacy work withcommunity elders and local political leaders, the CHCs were encouraged to organise anemergency transport fund and use of vehicles using local resources. The CHCs facilitatedthe LHWs in accessing women and in conducting group education sessions in theintervention villages. These sessions, attended by women of reproductive age, adolescentgirls, and older women, were held on a quarterly basis in a local household. The LHWs,often assisted byDais, facilitated these group sessions using standard materials and ipcharts developed for this purpose. Additionally, a two-part video docudrama onpregnancy and newborn care was produced and made available to intervention clusterLHWs for use in group sessions using available village resources (a home television andvideo cassette player). These sessions and related materials were in the local language(Sindhi) and focused on key knowledge and behaviours listed in table 1.

See Online for webappendix


3/10

Articles


community engagement, data collection procedures,and surveillance were made on the basis of the pilotphase and are summarised in panel 1. In the controlclusters, the LHW programme continued to function asusual and no additional attempt was made to link LHWswith the Dais or communities. They were, however,provided with regular refresher training according tothe standard national LHW programme curriculumincluding monthly debrieng sessions in publicsector health facilities. Table 1 shows details of the

intervention package.The Hala trial was approved by the ethics review

committees of Aga Khan University (Karachi, Pakistan),London School of Hygiene and Tropical Medicine(London, UK), and WHO (Geneva, Switzerland).

Randomisation and maskingThe key care provider within our intervention strategywas the LHW. Since LHWs are trained and supervisedby staff in BHUs and RHCs and use only thesefacilities for regular replenishment of supplies andreporting, we used the catchment areas of individualfunctional primary care facilities (BHUs, RHCs) andall affi liated LHWs as units of randomisation or clustersfor this trial. Altogether, 26 such clusters with

available LHWs were identied in the district, eight ofwhich were involved in the pilot study. 16 Two furtherclusters were excluded because they had very few LHWsand were largely in the riverine Katcha areas withpoor access. The full trial was thus implementedin the remaining 16 clusters (gure 1) betweenFebruary, 2006, and March, 2008. There were noexclusion criteria for LHWs, women, households, orclusters after randomisation.

To ensure reasonable balance between the two groupswe used restricted, stratied randomisation to allocateclusters to the intervention and control groups. 20 Threestrata (consisting of two, six, and eight clusters) wereidentied on the basis of their size and the number ofLHWs per 1000 population. We identied 126 randomallocations that resulted in similar population sizes inthe two groups (difference


4/10

Articles


Data collectors and their supervisors were masked tocluster allocation. Anthropologists undertaking verbaland social autopsies were masked to cluster allocationand nature of training of LHWs in their area. Dataanalysts were not masked to the cluster allocation.

Procedures

We established 13 independent data collection teams whoundertook quarterly visits to all villages in interventionand control clusters. The household surveillance systemwas designed on the basis of geographic boundaries andcontiguity of villages, rather than clusters. Data wereobtained from each household for all births, deaths,inmigrations, and outmigrations. From the secondsurveillance round onwards, women reporting livebirthsin the 4 weeks preceding the visit were interviewed with astructured questionnaire to obtain information aboutknowledge and practices relating to newborn care andLHW visits. Additionally, LHWs in the interventionclusters were asked to record information about homevisits, newborn illnesses, referrals, and deaths on specialproformas. A separate team obtained data for births,newborn referrals, and outcomes from the registers ofpublic and private sector health facilities in the area.Verbal and social autopsies of stillbirths and neonataldeaths were done by a separate team of trainedanthropologists within 1216 weeks of the event. Theverbal autopsy instruments used for obtaininginformation about stillbirths and newborn deaths werebased on adaptations of WHO recommended instrumentsused for assessing neonatal deaths and stillbirths. 2123 Theprimary outcome of the trial was perinatal and all-causeneonatal mortality.

Statistical analysisAfter the pilot phase, a complete household and healthfacility survey was done between May and August, 2005,in the 16 clusters selected for the full trial, to measuresocioeconomic characteristics and document baselineperinatal and neonatal mortality rates on the basis ofrecall of all births and deaths in the preceding year. Usingdata from the baseline census and the method describedby Hayes and Bennett, 24 we estimated the coeffi cient ofvariation in NMRs between clusters to be 016. With thisestimate of the coeffi cient, and assuming that during a2-year period there would be an average of 1400 livebirthsper cluster, we estimated that we would have close to90% power to detect a 30% reduction in NMR from

50 deaths per 1000 livebirths to 35 deaths per1000 livebirths.5 The power to detect a reduction of25% was close to 75%.

Since the number of randomised clusters was small wechose to analyse the primary outcome data (perinatal andneonatal mortality) at cluster level as recommended byHayes and Moulton, 25 and to adjust for baseline(preintervention) mortality rates. Analysis was by intentionto treat. For each cluster, the NMR during the interventionphase was calculated and the logarithm of the cluster-levelNMRs was then used as the independent variable in alinear regression model to provide an estimate of theNMR ratio associated with the intervention and its 95% CI,while accounting for the cluster randomisation. Stratumwas included as a xed effect in the model. The logarithm

Figure :Trial prole*Pairs of twins included: 109 both livebirths, ve both stillbirths, and 12 one livebirth and one stillbirth. Pairs oftwins included: 115 both livebirths, two both stillbirths, and eight one livebirth and one stillbirth.

16 clusters randomised

51409 participants (married women)

8 clusters assigned intervention

26 892 participants

14152 total pregnancies1536 pregnancies miscarried

before 7 months

225 lost to follow-up and outmigration

8 clusters analysed

12391 total deliveries12 517 births (12265 single and 126*

twin births)12 028 livebirths

489 stillbirths517 neonatal deaths113 postneonatal deaths

8 clusters assigned control

24517 participants

12835 total pregnancies1233 pregnancies miscarried

before 7 months

159 lost to follow-up and outmigration

8 clusters analysed

11443 total deliveries11568 births (11318 single and 125

twin births)11 005 livebirths

563 stillbirths540 neonatal deaths131 postneonatal deaths

Allocation

Follow-up

Intervention

(8 clusters)

Control

8 clusters)Total population 158 393 159 833Households 23 353 23 768

Mean number of people per household 68 (34) 67 (34)Mean number of children younger than 5 years per household 13 (13) 12 (13)

Maternal education*Illiterate 19 674 (83%) 18 776 (80%)

Able to read and write 1117 (5%) 893 (4%)Primary and middle school 1877 (8%) 2318 (10%)

Higher secondary school 789 (3%) 1123 (5%)Graduate and above 160 (1%) 374 (2%)

Not reported 98 (


5/10

Articles


of the baseline (preintervention) cluster-level NMR was

included as a covariate and the regression was weightedon the basis of the number of events in each cluster. Asimilar approach was used to analyse stillbirth rates. Allanalyses were done with Stata (version 10). When analysingreported practices before, during, and after delivery, weused the svy commands within Stata to account for theclustered nature of the data. No interim analyses ofmortality were done.

The Hala trial is registered, ISRCTN16247511

Role of the funding sourceThe funding bodies provided clearance for the projectdesign, but apart from eld visits to review progress, didnot inuence the eld trial or the data analysis procedures.The corresponding author had full access to all the datain the study and had nal responsibility for the decisionto submit for publication.

ResultsFigure 2 shows the trial prole and outcomes for all16 clusters. Table 2 shows baseline householdcharacteristics. Most mothers (>80%) were illiterate andalthough most families owned their own house, less thanhalf had access to a toilet and most used rewood forcooking. In 38 villages (17 in intervention and 21 incontrol clusters), there were pre-existing committeesprincipally focused on initiatives facilitating education.These committees were at various levels of functionalityand none were undertaking activities related to maternaland newborn care. The overall population covered by theLHW programme in the study area at baseline was 77%,and the number of resident LHWs per 10 000 populationwas similar in intervention and control clusters (table 3).Baseline stillbirth and neonatal mortality rates wereslightly lower in the intervention group than in thecontrol group (table 3; webappendix pp 34). Roughly 40%of all births occurred in local facilities (both public andprivate), with the remainder of births at home (table 3).

488 (96%) of 506 villages in the intervention clustersestablished CHCs during the study, of whom249 (51%) also established emergency transport funds.

Most CHCs met at least quarterly and had regular liaisonswith the LHW. No new LHWs were recruited during thestudy, and apart from two increments in salaries, in 2006and 2009, no signicant changes were introduced in themain curriculum and training programme for LHWs.

LHWs in the intervention areas were able to undertake4428 (63%) of 7084 planned quarterly community groupsessions in villages and self-reported visiting andexamining 2943 neonates (24%) of a total 12 028 livebirthsin the intervention clusters (1213 and 1730 in years 1and 2 of the study, respectively), but none were able toimplement all four suggested postnatal home visits.LHWs in the intervention clusters also reportedexamining 248 unwell neonates (42 in year 1 and 206 inyear 2). Of these, 72 (29%) babies with suspected severe

illnesses were referred to health facilities or localphysicians, and others with non-severe problems were

managed at home. There were no reported deaths inthis subgroup. The independent household surveillancedocumented 651 home visits by LHWs in the interventionclusters compared with 212 in control clusters. Therewas a trend of increasing home visits by LHWs duringpregnancy reported by women in the interventionclusters during the trial (p


6/10

Articles


by additional family members and at a quarter of the

sessions, some husbands also attended.The major pregnancy outcomes (miscarriages, still-births, neonatal deaths, and livebirths) were determinedthrough self-reporting in the quarterly householdsurveillance system. The baseline rates of miscarriagewere similar in both groups (76 reported miscarriagesper 1000 pregnancies in intervention clusters vs 71 incontrol clusters). Subsequently, during the interventionperiod, reported miscarriage rates were higher in both

groups, particularly in the intervention clusters

(109 reported miscarriages per 1000 pregnancies inintervention clusters vs 96 in control clusters; risk ratio[RR] 112, 95% CI 089140; p=031). No clear timetrend for miscarriages was discernible. Rates of stillbirth(RR 079, 068092; p=0006) and neonatal mortality(RR 085, 076096; p=002) were signicantly lowerin intervention clusters than control (table 4). Stillbirthrates in the intervention clusters seemed to decreasewith time, while remaining roughly constant in thecontrol clusters (web appendix pp 56). Within theintervention clusters, neonatal mortality seemed lowerin the areas covered by LHWs (376 per 1000 livebirths)than in areas that were not covered (483 deaths per1000 livebirths), whereas no major difference wasapparent in the control clusters (478 vs 504 per1000 livebirths, respectively; test for interaction p=004).There were seven fewer maternal deaths in theintervention clusters than in the control clusters (71 vs 78)during the study.

Information about household practices was availablefor 4474 pregnancies resulting in a livebirth, representing19% of all 23 033 livebirths during that period and 58% ofall livebirths in the 28 days preceding the surveillancevisit (table 5). The most common reason for lack of accessto women who had delivered in the past 28 days was thecommon cultural practice of women delivering in theirparents house and residing with them for 40 days afterchildbirth. The ratio of women with recent deliveries(within 28 days) who were surveyed was similar inintervention (2339 [19%] of 12 028 livebirths) and control(2135 [19%] of 11 005 livebirths) clusters.

More women in the intervention clusters than controlreported attendance for antenatal care in facilities (table 5;gure 4), but this difference was small and not signicantand could reect sampling variation. However, women inintervention clusters reported signicantly more frequentcontact with their LHW during pregnancy than did thosein the control clusters (44% vs 26%; p=005). Amongwomen delivering at home, the use of clean delivery kits

Interventionclusters

Controlclusters

Mortality riskratio (95% CI)*

p value

Livebirths identied 12 028 11 005

Postneonatal infant deaths 113 131 Miscarriages

Number 1536 1233 Rate per 1000 known pregnancies 109 96 112 (089140) 031

StillbirthsNumber 489 563

Rate per 1000 total births 391 487 079 (068092) 0006Early neonatal mortality

Number 391 409 Rate per 1000 livebirths 325 372 086 (075098) 003

Late neonatal mortalityNumber 126 131

Rate per 1000 livebirths 105 119 083 (064107) 013

Neonatal mortalityNumber 517 540 Rate per 1000 livebirths 430 491 085 (076096) 002

Perinatal mortalityNumber 880 972

Rate per 1000 total births 703 840 083 (074093) 0004

*All parameter estimates, CIs, and p values estimated by (weighted) analysis of variance at the cluster level; dependentvariable=log (rate); stratication used in the randomisation included as a xed effect; log (baseline neonatal mortalityrate) included as a covariate; weights used were based on the number of events reported in each cluster. Neonataldened as age 028 days; postneonatal, 29365 days; early neonatal, 07 days; and late neonatal, 828.

Table : Summary outcomes of quarterly surveillance (rounds 18)

Intervention group (n=2339) Control group (n=2135) p valuen/N % (95% CI) n/N % (95% CI)

Women attending at least one antenatal consultation in a facility 1616/2339 69% (5682) 1230/2135 58% (3976) 029Women attending four or more antenatal consultations in a facility 302/2339 13% (917) 191/2135 9% (216) 036

Women having contact with LHW during pregnancy 1019/2334 44% (3156) 553/2123 26% (1537) 005Women delivering in a facility 1272/2339 54% (4861) 936/2135 44% (3453) 007

Home deliveries using a clean delivery kit 302/867 35% (2743) 34/1102 3% (25)


7/10

Articles


was much more common in the intervention clusters(35% vs 3%; p


8/10

Articles


the classication of fetal deaths as miscarriages orstillbirths, nor in classication of stillbirths as very earlyneonatal deaths or vice versa. However, current methodsof verbal autopsies are at best fairly crude in differentiationof late miscarriages, stillbirths, and early neonatal deaths,and misclassication cannot be entirely excluded. 29,30

The reductions in stillbirths and neonatal deaths weresmaller than the 34% and 28% reductions reported in thepilot phase, 16 and were lower than those reported fromother recent large cluster randomised trials in southAsia.1215 The scale-up phase of the Hala effectiveness trialwas three times larger than the initial pilot and was much

bigger than other studies in the region. It also differedfrom other studies in the region 1215,17,18 in that theintervention was principally delivered through thegovernment health system rather than by workersemployed directly by the research team. Referral healthfacilities serving both intervention and control clusterswere strengthened and the population in the controlclusters received basic maternal and newborn carethrough the existing LHW programme.

Other contextual factors should be considered. Theintervention was complex and was delivered throughpublic sector LHWs and the government health system,and was thus subject to human resource constraints, thecompeting demands of other routine activities, andgeneral weaknesses in health system functionality. These

factors affect LHW job stress and performance. 9,31 The

investigators had little control over local managers,several of whom were transferred into and out of thedistrict several times. Even the robust federal LHWprogramme saw three changes of leadership at the federaland four at the provincial level between 2004 and 2008.Although the changes did not prevent the introduction ofthe project in the district, several agreed actions by theDirectorate of Health, such as deployment of additionalLHWs to cover the entire area, replacement of LHWswho were not locally based, and provision of keycommodities such as newborn weighing scales, were notimplemented. Almost a quarter of the target populationand villages of the area remained uncovered by LHWs,who in turn often had to multitask and provide logisticalsupport to other government programmes, as per policy.From work registers we estimate that during the courseof the intervention, study LHWs spent an average 30% oftheir time on the periodic polio eradication campaigns,which interrupted regular maternal, neonatal, and childhealth activities. The fairly low coverage of theintervention overall should therefore be seen in thecontext of a busy functional primary care CHWprogramme being charged with implementing a complexpackage of community-based advocacy and education.

Despite these limitations and the known reduction ineffectiveness when scaling up from effi cacy trials, 32 theHala trial provides encouragement that a public sectorprogramme promoting preventive maternal and newborncare can lead to behavioural change and careseeking formothers during pregnancy and childbirth with resultanthealth benets. The observed reduction in stillbirthsparalleled the increase in facility births and skilledattendance during delivery in the intervention clusters.The two main public sector hospitals in the districtalready had trained obstetricians and midwives withadequate facilities for emergency obstetric care.

We are unable to ascribe improvements in perinatal andneonatal outcomes to any single component of the threeelements of the intervention package. The LHWs played akey part in the implementation of all three components,although overall coverage rates of various components of

the intervention package varied. The LHWs were able toliaise with CHCs and deliver the community groupsessions with greater effi ciency than targeted postnatalhome visits and overall coverage of some components ofLHW supported activities, such as presence duringchildbirth, immediate postnatal visits, and examination ofsick neonates, remained low. We speculate therefore thatthe pathway for effectiveness of the Hala intervention waslargely through improved antenatal contact with LHWsleading to improved childbirth care for mothers, includingincreased facility births in the hands of skilled attendants,and improvements in some elements of immediatenewborn care. 33

The lack of improvement in careseeking for newbornillnesses in the intervention clusters could be related to

Panel :Research in context

Systematic reviewGrowing evidence exists of the effectiveness of various approaches to community-baseddelivery strategies and platforms to address neonatal mortality and morbidity. We havedocumented that various community-based interventions work and have the potential toreduce neonatal mortality.10,26 Similarly Lewin and colleagues27 have shown the benets ofusing lay health workers or traditional birth attendants in improving a range of childhealth outcomes, and Sibley and co-workers28 have made the case for training oftraditional birth attendants to reduce perinatal mortality. We have also shown thatcommunity support groups using health workers or other community mobilisers couldaffect household behaviours and improve newborn outcomes and careseeking.19 Others11 have evaluated the effect of home visits on neonatal mortality and shown signicantbenets. However, few studies have evaluated packages of care and none have done so ineffectiveness settings.

InterpretationThe results of our study suggest that trained public sector community health workers inrural Pakistan can deliver a package of preventive and promotive health care messages tocommunity members. Despite limitations of time and competing tasks, lady healthworkers (LHWs) were able to build a rapport with community members and implement apackage of promotive and preventive maternal and newborn care interventions. Althoughthe overall coverage achieved by the LHW-supported intervention was low, the effect oncrucial household behaviours and careseeking patterns was promising. The LHWs were alsoable to liaise with local traditional birth attendants (Dais) and volunteer community healthcommittees, and provide community education and advocacy for facility births. Thesendings add to the growing evidence base for the effi cacy and effectiveness ofcommunity-based approaches to address newborn mortality in diffi cult-to-reach areas andsupport use of strategies involving outreach workers in such settings.


9/10

Articles


the poor quality of care for newborns in the local public

sector facilities and the reluctance among families totravel long distances to seek newborn care. 5,34 Theinformation available about the subset of newborninfants with illnesses also suggested that careseekingwas largely in the formal and informal private sector.

Urgent attention is needed for the provision of adequatebasic and emergency newborn care facilities in the healthsystem, and the LHW programme might also considerinclusion of interventions for immediate newborn caresuch as emergency resuscitation, kangaroo mother care,and oral antibiotic treatment for suspected respiratoryinfections. By contrast with others, 12,13 we did not provideany home-based bag and mask resuscitation or antibiotictreatment. Despite indications that these interventions areeffi cacious, further trials are needed in effectivenesssettings, especially those that integrate interventionpackages across the continuum of care. 35 Recent systematicreviews of community health workers suggest that they areeffective in delivering a range of interventions to affectnewborn health and child survival. 27,36,37 Such program maticinterventions and implementation research are a priority.

Previous projections have suggested that communityand outreach interventions, if implemented at scale, havethe potential of reducing newborn deaths by 36%. 26 Despite much lower coverage in our trial than in othereffi cacy trials, the intervention package was associatedwith a reduction in perinatal and newborn mortality of1520%. More importantly, we found that LHWs couldwork effectively with existing Dais in the area. Sincetraining of traditional birth attendants has little effect onreducing perinatal mortality, 28 these ndings suggest away to link various public and private sector healthworkers in health systems with promotion of skilled careand in-facility births for families. Future strategies couldcombine CHW-based programmes with additionalmodalities for promotion of facility births throughcommunity education, 34 publicprivate partnerships, 38 and scal incentive schemes. 39 However, to be effective,such health workers and programmes need closeoversight and dedicated activities.Contributors

ZAB drew up the project proposal and was the principal investigator forthe Hala trial. ZAB also wrote the rst draft of the manuscript withcontributions from SS and SC. SC, AF, and IA undertook the statisticalanalyses. All authors contributed to the implementation, data review,and writing process.

Conicts of interestWe declare that we have no conicts of interest.

AcknowledgmentsThe Hala trial was funded by grants from WHO and the SavingNewborn Lives programme funded by the Bill & Melinda GatesFoundation. We thank Department of Health, Sindh Government, andthe National Program for Family Planning and Primary Health Care fortheir encouragement and support towards undertaking this trial, all staffof the Hala project offi ce for their hard work and support, and notably,the excellent support provided by Usman Chachar, the districtcoordinating offi cer, and Hasan Murad Shah, the executive districtoffi cer for health for Matiari district .

References1 Rajaratnam JK, Marcus JR, Flaxman AD, et al. Neonatal,

postneonatal, childhood, and under-5 mortality for 187 countries,19702010: a systematic analysis of progress towards MillenniumDevelopment Goal 4. Lancet 2010;375: 19882008.

2 Lawn JE, Cousens S, Zupan J, for the Lancet Neonatal SurvivalSteering Team. 4 million neonatal deaths: When? Where? Why?Lancet 2005; 365: 891900.

3 Bhutta ZA, Cross A, Raza F, Zaheer Z. Chapter 8. Infant and childmortality. In: Pakistan demographic and health survey 200607.Islamabad, Pakistan: National Institute of Population Studies,Pakistan, and Macro International Inc, 2008.

4 Jafarey SN. Characteristics and practices of traditional birthattendants (Dais) a preliminary survey. J Pak Med Assoc 1981;31: 28891.

5 Bhutta ZA, Ali N, Hyder AA, Wajid A. Perinatal and newborn carein Pakistan: seeing the unseen. In: Bhutta ZA, ed. Maternal andchild health in Pakistan: challenges and opportunities. Karachi,Pakistan: Oxford University Press, 2004.

6 National Programme for Family Planning and Primary Health Care.Ministry of Health, Government of Pakistan. http://www.phc.gov.pk(accessed May 2, 2010).

7 Douthwaite M, Ward P. Increasing contraceptive use in ruralPakistan: an evaluation of the Lady Health Worker Programme.Health Policy Plan 2005; 20: 11723.

8 Oxford Policy Management. Evaluation of the Prime MinistersLady Health Worker Programme. 2003. http://www.opml.co.uk/policy_areas/statistics/quantitative_information_for_policy_and_programme/cn1064_lhw.html (accessed Dec 23, 2010).

9 Oxford Policy Management. External evaluation of the Lady HealthWorker Programme, Pakistan. 2009. http://www.opml.co.uk/policy_areas/statistics/quantitative_information_for_policy_and_programme/6241_lhw.html (accessed Dec 23, 2010).

10 Bhutta ZA, Darmstadt GL, Hasan BS, Haws RA. Community-basedinterventions for improving perinatal and neonatal health outcomesin developing countries: a review of the evidence. Pediatrics 2005;115 (2 suppl) : 519617.

11 Gogia S, Sachdev HPS. Home visits by community health workersto prevent neonatal deaths in developing countries: a systematicreview. Bull World Health Organ 2010;88: 65866.

12 Bang AT, Bang RA, Baitule SB, Reddy MH, Deshmukh MD. Effectof home-based neonatal care and management of sepsis on neonatalmortality: eld trial in rural India. Lancet 1999;354: 195561.

13 Baqui AH, El-Arifeen S, Darmstadt GL, et al, for the ProjahnmoStudy Group. Effect of community-based newborn-care interventionpackage implemented through two service-delivery strategies inSylhet district, Bangladesh: a cluster-randomised controlled trial.Lancet 2008; 371: 193644.

14 Manandhar DS, Osrin D, Shrestha BP, et al, and members of theMIRA Makwanpur trial team. Effect of a participatory interventionwith womens groups on birth outcomes in Nepal:cluster-randomised controlled trial. Lancet 2004; 364: 97079.

15 Tripathy P, Nair N, Barnett S, et al. Effect of a participatoryintervention with womens groups on birth outcomes and maternaldepression in Jharkhand and Orissa, India: a cluster-randomised

controlled trial. Lancet 2010;375: 118292.16 Bhutta ZA, Memon ZA, Soo S, Salat MS, Cousens S, Martines J.

Implementing community-based perinatal care: results from a pilotstudy in rural Pakistan. Bull World Health Organ 2008; 86: 45259.

17 Baqui AH, Rosecrans AM, Williams EK, et al. NGO facilitationof a government community-based maternal and neonatal healthprogramme in rural India: improvements in equity.Health Policy Plan 2008; 23: 23443.

18 Kumar V, Mohanty S, Kumar A, et al, for the Saksham Study Group.Effect of community-based behaviour change management onneonatal mortality in Shivgarh, Uttar Pradesh, India:a cluster-randomised controlled trial. Lancet 2008; 372: 115162.

19 Bhutta ZA, Lassi ZS. Empowering communities for maternaland newborn health. Lancet 2010;375: 114244.

20 Hayes RJ, Moulton LH. Cluster randomised trials. London, UK:Chapman and Hall/CRC Interdisciplinary Statistics, 2008.

21 WHO. Standard neonatal verbal autopsy questionnaire, revisedversion: WHO/JHU/SNL. Geneva, Swizerland: World HealthOrganization, 2003.


10/10

Articles

10 www thelancet com Published online January 15 2011 DOI:10 1016/S0140 6736(10)62274 X

22 Baqui AH, Darmstadt GL, Williams EK, et al. Rates, timing andcauses of neonatal deaths in rural India: implications for neonatalhealth programmes. Bull World Health Organ 2006; 84: 70613.

23 Kalter HD, Khazen RR, Barghouthi M, Odeh M. Prospectivecommunity-based cluster census and case control study ofstillbirths and neonatal deaths in the West Bank and Gaza Strip.Paediatr Perinat Epidemiol 2008; 22: 32133.

24 Hayes RJ, Bennett S. Simple sample size calculation forcluster-randomized trials. Int J Epidemiol 1999;28: 31926.

25 Moulton LH. Covariate-based constrained randomizationof group-randomized trials. Clin Trials 2004; 1: 297305.

26 Darmstadt GL, Bhutta ZA, Cousens S, Adam T, Walker N,de Bernis L, for the Lancet Neonatal Survival Steering Team.Evidence-based, cost-effective interventions: how many newbornbabies can we save? Lancet 2005; 365: 97788.

27 Lewin S, Munabi-Babigumira S, Glenton C, et al. Lay healthworkers in primary and community health care for maternaland child health and the management of infectious diseases.Cochrane Database Syst Rev 2010;3: CD004015.

28 Sibley LM, Sipe TA, Brown CM, Diallo MM, McNatt K, Habarta N.Traditional birth attendant training for improving health behavioursand pregnancy outcomes. Cochrane Database Syst Rev 2007;3: CD005460.

29 Edmond KM, Quigley MA, Zandoh C, et al. Diagnostic accuracyof verbal autopsies in ascertaining the causes of stillbirths andneonatal deaths in rural Ghana. Paediatr Perinat Epidemiol 2008;22: 41729.

30 Carlo WA, Goudar SS, Jehan I, et al, for the First Breath StudyGroup. Newborn-care training and perinatal mortality in developingcountries. N Engl J Med 2010;362: 61423.

31 Haq Z, Iqbal Z, Rahman A. Job stress among community healthworkers: a multi-method study from Pakistan. Int J Ment Health Syst 2008; 2: 15.

32 Tugwell P, de Savigny D, Hawker G, Robinson V. Applying clinicalepidemiological methods to health equity: the equity effectivenessloop. BMJ 2006; 332: 35861.

33 Baqui AH, Ahmed S, El Arifeen S, et al, for the Projahnmo 1 StudyGroup. Effect of timing of rst postnatal care home visit onneonatal mortality in Bangladesh: a observational cohort study.BMJ 2009; 339: b2826.

34 Midhet F, Becker S. Impact of community-based interventions onmaternal and neonatal health indicators: Results from a communityrandomized trial in rural Balochistan, Pakistan. Reprod Health 2010;7: 30.

35 Bhutta ZA, Chopra M, Axelson H, et al. Countdown to 2015 decadereport (200010): taking stock of maternal, newborn, and childsurvival. Lancet 2010;375: 203244.

36 Bhutta ZA, Lassi ZS, Pariyo G, Huicho L. Global experienceof community health workers for delivery of health relatedMillennium Development Goals: a systematic review, country casestudies, and recommendations for scaling up. Geneva, Switzerland:Global Health Workforce Alliance, 2010. http://www.who.int/workforcealliance/knowledge/publications/CHW_FullReport_2010.pdf (accessed Dec 23, 2010).

37 Lassi Z, Haider B, Bhutta ZA. Community-based interventionpackages for preventing maternal morbidity and mortality andimproving neonatal outcomes. New Delhi, India: InternationalInitiative for Impact Evaluation, 2010.

38 Lim SS, Dandona L, Hoisington JA, James SL, Hogan MC,Gakidou E. Indias Janani Suraksha Yojana, a conditional cashtransfer programme to increase births in health facilities: an impactevaluation. Lancet 2010;375: 200923.

39 Mavalankar D, Singh A, Patel SR, Desai A, Singh PV. Savingmothers and newborns through an innovative partnership withprivate sector obstetricians: Chiranjeevi scheme of Gujarat, India.Int J Gynaecol Obstet 2009; 107: 27176.

Improve Perinatal Pakistan

Documents

Transcript of Improve Perinatal Pakistan