Khatunetal.BMCPublicHealth2012,12:84 http://www.biomedcentral.com/1471-2458/12/84 RESEARCH ARTICLE Open Access Causes of neonatal and maternal deaths in Dhaka slums: Implications for service delivery Fatema Khatun1*, Sabrina Rasheed1, Allisyn C Moran1, Ashraful M Alam1, Mohammad Sohel Shomik1, Munira Sultana2, Nuzhat Choudhury1,2, Mohammad Iqbal1 and Abbas Bhuiya1 Abstract Background: Bangladesh has about 5.7 million people living in urban slums that are characterized by adverse living conditions, poor access to healthcare services and health outcomes. In an attempt to ensure safe maternal, neonatal and child health services in the slums BRAC started a programme, MANOSHI, in 2007. This paper reports the causes of maternal and neonatal deaths in slums and discusses the implications of those deaths for Maternal Neonatal and Child Health service delivery. Methods: Slums in three areas of Dhaka city were selected purposively. Data on causes of deaths were collected during 2008-2009 using verbal autopsy form. Two trained physicians independently assigned the cause of deaths. Results: A total of 260 newborn and 38 maternal deaths were identified between 2008 and 2009. The majority (75%) of neonatal deaths occurred during 0-7 days. The main causes of deaths were birth asphyxia (42%), sepsis (20%) and birth trauma (7%). Post partum hemorrhage (37%) and eclampsia (16%) were the major direct causes and hepatic failure due to viral hepatitis was the most prevalent indirect cause (11%) of maternal deaths. Conclusion: Delivery at a health facility with child assessment within a day of delivery and appropriate treatment could reduce neonatal deaths. Maternal mortality is unlikely to reduce without delivering at facilities with basic Emergency Obstetric Care (EOC) and arrangements for timely referral to EOC. There is a need for a comprehensive package of services that includes control of infectious diseases during pregnancy, EOC and adequate after delivery care. Background asphyxia(45%),lowbirthweight/prematurity(15%)and Highmaternalandneonatalmortalityratesarestillmajor neonatal sepsis (12%) as the main causes of neonatal challenges facing Bangladesh. Despite significant reduc- deaths[4]. tion over the last two decades, the maternal mortality Urbanslumsdeservespecialattentionastheyhostover ratio and neonatal mortality rate are high at 194 per 5.7 million people, roughly 3.8% of the total national 100,000 live births and 37 per 1000 live births [1,2] population[3].Recent datashowthatmostoftheglobal respectively. The situation is worse in the country’s urbangrowthisoccurringinthelowincomecountriesof urban slums with neonatal mortality rate of 43.7 per Asian citiessuchasDhaka [5]. Ifone only considersthe 1000live births[3]. Themaincausesofmaternaldeaths greatpopulationdensitieswithinthe cities,cities clearly are hemorrhage (31%), eclampsia (20%) according to present a very different health environment than the 2010National survey [1]. There is no informationavail- moresparselypopulatedruralareas.Thoughrapidurba- able related to maternal cause of death in the urban nization is driven mostly by a more dynamic economic slums. There are no national surveys of exploring the environment in the cities, it also attracts many of the causes of neonatal deaths in Bangladesh. However, a poorest anddisadvantaged membersofrural society [3]. study conducted selected rural area reported birth This population presents an enormous challenge to the urbaninfrastructure,includingpublichealthandhealth- *Correspondence:[email protected] care systems already stretched beyond capacity to meet 1ICDDR,B,68ShaheedTajuddinAhmedSarani,Mohakhali,Dhaka1212, the need of the present population. Key elements of Bangladesh urban infrastructure, including water and sanitation are Fulllistofauthorinformationisavailableattheendofthearticle ©2011Khatunetal;licenseeBioMedCentralLtd.ThisisanOpenAccessarticledistributedunderthetermsoftheCreativeCommons AttributionLicense(http://creativecommons.org/licenses/by/2.0),whichpermitsunrestricteduse,distribution,andreproductionin anymedium,providedtheoriginalworkisproperlycited. Khatunetal.BMCPublicHealth2012,12:84 Page2of9 http://www.biomedcentral.com/1471-2458/12/84 already inadequate for the need of the population and Identification of deaths effects of these inadequacies are likely to be felt at the In the three slums deaths of adult female and under five level of population health [3]. The size of slum popula- childrenwereidentified through the existingMANOSHI tion and the unique challenges of providing health ser- program Management Information System (MIS). Study vices to the residentsofurban slumsmake urban health team visitedhouseholds to collect information ondeath. amajorissueofconcernforBangladeshandothersimilar Thesevisitsweremademonthlyusingastandardform.In nations.Amongthemanychallenges,designingappropri- addition, information on the deaths was collected from ate service delivery programs, especially for reducing traditional birth attendants, drug sellers and local health maternal and neonatal mortality, is critical on many providers.Oncethefinallistwascompiled,thehouseholds counts.Reachingtheslumdwellerswithappropriateser- werevisitedto collectfurtherinformation.Thedetailsof vices in terms of service components and delivery strat- allthedeathsinvestigatedareprovidedinFigure1. egy is essential. Service components should be effective In this paper we discussed the maternal and neonatal in reducing mortality from common causes of deaths. deaths. Maternal deaths was defined as any death of a Keeping thisin mind,a studytocollectcausesof deaths woman (aged 15-49 years) while pregnant or within 42 of adult females and children < 5 years of age in slums days of delivery or termination of pregnancy, irrespec- was conducted between 2008 and 2009 as a part of a tive of the duration and site of the pregnancy, from any large scale maternal,neonatalandchildhealth(MNCH) cause related to or aggravated by the pregnancy or its service delivery program MANOSHI (Maa Nobojatok O management, but not from accidental or incidental Shishu orMother, NewbornandChild)implementedby causes. Neonatal deaths were defined as death of new- BRAC [6]. This paper reports maternal and neonatal born within 28 days of birth due to any cause. deathsinthestudy area anddiscussestheirimplications inthe contextoftheexistingandfuturepossible service Data collection deliveryprogramsintheurbansetting. Afteridentificationofdeath,atrainedfemaleinterviewer visitedthehouseholdtoconductaverbalautopsyinterview Methods within15to30daysofthedeath.Theverbalautopsyques- Study area tionnaireusedwasbasedonthosedevelopedbyINDEPTH ThisstudywasconductedinthreeselectedareasofDhaka (http://www.indepth-network.org) and WHO [8]. The city which contained slums: Gulshan (Korail, Shat tola), questionnaire was adapted to local customs and culture Uttara(north and south Arichpur) and Kamrnagir Char andwastranslatedintoBengali.Thequestionnaireincluded covering approximately 300,000 residents in total. All sectionsonbackground,eventsleadingtothedeath,signs deathsamongadultfemalesandchildren<5yearsofage andsymptomsofillnessleadingtothedeath,pregnancy whowereresidentsoftheslums(basedonprogramidenti- history,careseekingbehaviorandperceptionoftherespon- ficationnumbers)inthestudyareabetweenJanuary2008 dentsaboutthecauseofdeath.Wealsocollecteddescrip- andDecember2009wereeligibleforthestudy. tionsofsignsandsymptomsofcomplications.Abriefstory To reduce morbidity and mortality of mothers, new- oftheeventsleadingtothedeathwasrecorded.Thestory borns and children in the urban slums of Bangladesh, comprised of hospital records if available and of the BRACstartedaslum-basedMNCHprogramcalledMAN- descriptionoftheeventsgivenbytherespondents.Inaddi- OSHI[6].EachBRACdeliverycentrecoveredabout2000 tion,hospitalrecords,laboratorytestsanddeathcertificates households and was manned by two urban birth atten- werephotocopiedandincludedinthereviewprocess.For dants(UBA)andoneCommunityHealthWorker(CHW). verbalautopsyinterviewsofdeceasedwomen,weselecteda One community midwife oversaw the operations of 4-5 closefamilymemberhadbeenpresentduringtheillness delivery centers. At the delivery centre, services such as thatledtodeath,andwasabletodescribethesymptoms normal delivery, post natal care (PNC), assessment of and medical consultations prior to death. For newborn mothers and the neonate for identification of maternal deaths,mothersorcloserelativesweretheprimaryrespon- and neonatal danger signs, referral when complications dents.Forneonataldeaths,78%oftherespondentswere warrantedandmanagementandreferralforbirthasphyxia parentsandformaternaldeaths,21%ofrespondentswere and low birth weight cases were provided. The program husbands,29%wereparentsandparentsinlaws,26%were used CHWs to identify pregnancies, provide Ante Natal siblings, 5% were adult sons or daughters and18% were Care (ANC), provide counseling related to maternal and neighborsofthedeceased. childhealth,bringmotherstoslum-baseddeliverycenters for normal delivery and provide appropriate referral lin- Assigning causes of death kagestosecondaryandtertiaryfacilitiesfordeliverycom- Two registered physicians independently reviewed each plications[6].Furtherdetailsaboutthestaffinganddesign questionnaire and assigned a cause of death based on ofMANOSHIcanbefoundelsewhere[7]. International Classification of Diseases version 10 (ICD Khatunetal.BMCPublicHealth2012,12:84 Page3of9 http://www.biomedcentral.com/1471-2458/12/84 Total number of death (adult female and children < 5 yrs) N=576 Adult female Children <5 yrs n=93 n=483 Non-maternal Maternal Neonate Infant Children 2-4 yrs n=55 n=38 n=260 n=147 n=76 Figure1Numbersofdeathsbycategories. 10) [9]. When there was a disagreement between the performed using Statistical Package for Social Sciences two reviewers, the verbal autopsy questionnaire was (SPSS,Chicago,IL)version12.0. reviewed by a third physician. Final cause of death was assigned based on consensus of at least two physicians. Ethical consideration If disagreement remained, the cause was assigned a “not Ethical Review Committee of the ICDDR,B provided determined” code. If there were not enough information approval. Informed verbal consent was taken from all to assign a definite cause, the code “unknown” was interviewees and confidentiality and anonymity were given by the physicians. After independent review by ensured. two physicians, 92% of assigned causes of death were concordant and for 8% of the cases arrived at jointly Results based on consensus of the two primary physicians. Between2008and 2009, there were 260neonataland 38 maternal deaths. Of the neonatal deaths, 36% occurred Quality assurance during the first day and 75% during the first week of life Identification and verification of deaths (Figure2).Fortysevenpercentofneonatewerelowbirth AlldeathsthatwereidentifiedfromMANOSHIprogram weight (< 2500 g) and 52% were pre-term (< 37 weeks). MISandotherlocalsourceswere verifiedby studyteam Thirty seven percent of mothers with a neonatal death throughhouseholdvisitsusingastandardform.Duplica- had no education, 60% of mothers availed ANC from tion of deaths from different sources was avoided MANOSHIand42%deliveredathome(Table1). throughusingnamesofthehouseholdheadsandMAN- The three main causes of neonatal deaths were birth OSHIprogramidentificationnumbers. asphyxia(42%),neonatalsepsisandpneumonia(27%)and Training of field workers birth trauma (7%) (Figure 3). We were able to assign Threegraduatefemaleresearchassistantsand1supervisor underlyingcausefor60%ofthedeathsandfoundthatthe constitutedthedatacollectionteam.Theteamwastrained major underlying cause of neonataldeaths was low birth fortwo weeks onthe study questionnaire andonhowto weight(74%)(Datanotshown).Wecollectedinformation dealwithsituationwhileconductinginterviewsonasensi- about lay perception of cause of death. From lay beliefs, tiveissuesuchasdeaths.Thetrainingincludedfieldtest- the leading reason for neonatal death was supernatural ing. Physicians received two-week training on how to causes (evil spirit/evil eyes), believed by 22% of respon- reviewthequestionnaireinorderto assignICD 10codes dents,premature birth (9%)and other biomedical causes forcauseofdeath. (Table2). Quality control of the data The median age of women who died due to maternal For every 10 respondent questionnaires, a quality con- causes was 25 years. Of the women who died due to trol field officer re-interviewed one respondent to verify maternal causes, 42% did not receive any formal educa- data quality. tion,82%receivedantenatalcare,86%weremulti-gravida, 31% delivered at home, 10% delivered at BRAC delivery Data analysis centreand75%hadnormalvaginaldelivery(Table3). For reporting purposes neonatal sepsis and pneumonia Themajorcausesofmaternaldeathswerepostpartum werecombinedtoonecategory[10].Descriptiveanalysis hemorrhage (PPH) (37%), eclampsia (16%), and hepatic was done as appropriate. Statistical analyses were failure (11%) (Figure 4). We were able to assign an Khatunetal.BMCPublicHealth2012,12:84 Page4of9 http://www.biomedcentral.com/1471-2458/12/84 40 35 30 ) % ( 25 r e b m 20 u n al 15 t o T 10 5 0 1 2 3 4 5 6 7 8 9 101112131415161718192021222324252627 Age at death (days) Figure2Ageofneonateatdeath. underlying cause for 37% maternal deaths. The major complicationsandresultsinbirthasphyxiaarewellknown underlying cause of maternal death was hepatic failure [14], however,implementationofpreventivemeasuresto due to viral hepatitis (43%) (Data not shown). Respon- avoidasphyxiaduringperinatalperiodremainschallenging dents believedthat maternal deathswere caused by var- inlowincomecountries[15]andislikelyexacerbatedby ious biomedical conditions (Table 4). From the lay living conditions in the slums. MANOSHI program perspective 19% of the maternal deaths occurred due to trainedandincentivizedCHWstoidentifybirthasphyxia, bleeding followed by 16% due to jaundice (Table 4). No toprovidebasicmanagementthroughbagandmask,and supernaturalcauseswerereported. togivereferralsifasphyxiationofthenewborncontinues after basic resuscitation techniques are applied at the Discussion BRACdeliverycenters[6].However,45%ofalldeliveries National data on causes of maternal and newborn tookplaceathomeandBRACmadenoprovisionforpro- deaths are available [1-3,11]. However, mortality data for vidinganytrainedpersonnelforhomedeliveries[16].So, urban slums is not usually segregated. In describing the forthelargenumberofhomedeliveries(45%ofalldeliv- causes of neonatal and maternal deaths in selected eries)therearenotrainedpersonnelinthecommunityto slums of urban Dhaka, this paper will provide useful help with identification or timely referral for birth information for program planners who are involved in asphyxia. Thus the program is less likely to have signifi- providing healthcare to ensure maternal, neonatal and cantimpactonreductionofdeathfrombirthasphyxiafor child survival in urban slums. In addition, because the substantialnumberofcasesremainsoutofprogramcov- data were collected from a large-scale MNCH program erage.TrainingCHWswhowouldmakehomevisitsdur- area, an understanding may be gleaned of areas of focus ing delivery to identify and make referrals for birth where emphasis could reduce the number of neonatal asphyxiacouldhelpreducedeaths. and maternal deaths in similar programs. Thesecondlargestcauseofneonataldeathswasneona- talsepsisincludingpossibleseriousbacterialinfectionand Neonatal deaths pneumonia.Thisechoesthefindingsofasimilarstudyin Inourstudy, birthasphyxia (42%)isthedominantcause ruralBangladesh[4].AccordingtoWHOIMCIguidelines ofneonatal deaths. Thisfinding was supported by verbal neonatalinfectionsshouldbetreatedwithinjectableanti- autopsy data from a rural area of Bangladesh and neigh- biotics[10].InseveralinterventionsinIndiaand Bangla- boringcountryNepal[4,12].Thehighproportionofbirth desh, CHWs were trained to recognize and manage asphyxia in the urban slums and rural Bangladesh com- suspectedseriousneonatalinfectionswithinjectablepro- pared to global average of 23% indicates a lack of appro- cainepenicillinandgentamicin[17,18]whichresultedin priate resuscitation after birth and a lack of immediate 60%and34%reductioninneonatalmortalityrespectively. referral to hospitals [13]. Factors that lead to delivery MANOSHI CHWs were not trained to provide such Khatunetal.BMCPublicHealth2012,12:84 Page5of9 http://www.biomedcentral.com/1471-2458/12/84 Table 1Reproductive and socio-demographic Table 1 Reproductive and socio-demographic characteris- characteristics ofmothers ofthe deceased neonates tics of mothers of the deceased neonates (2008-9) (2008-9) (Continued) Socio-demographiccharacteristics N=260 Unknown 8(3.1) (%) PNCtakenfrom(n=152) Mother’sage Skilled(MBBS,Nurse) 76(50.0) 15-19y 52(20.2) BRAC(CHWs) 70(46.1) 20-24y 90(35.0) Unskilled(villagedoctors,drugsellersCSBA.trained 6(3.9) 25-29y 68(26.5) TBA) 30-35y 21(8.2) Birthorder 35-39y 19(7.4) 1st 82(31.5) >40y 7(2.7) 2nd 67(25.8) Monthlyhouseholdincome 3rd 53(20.4) <5000Tk.(US$66) 124(47.7) 4thandabove 58(22.3) 5000-10000Tk.(US$67-133) 120(46.2) >10,000Tk.(US$134) 16(6.2) Mothers’education interventionsandtherefore,itisunlikelysuchsteepreduc- None 95(37.0) tion in mortality can be expected from MANOSHI. The MANOSHI CHWs were trained toonly identify cases of 1-5years 107(41.6) neonatal infections during PNC home visits and were 6+years 55(21.4) given incentivesformakingreferrals. The measureswere Mothers’occupation undermined, however as, according to the MANOSHI Housewife 209(80.4) midline survey [16], only 13% of mothers received PNC Maid 15(5.8) visits at 3-28 days which may have left many neonatal Garmentsworker 22(8.5) infectionsundetected.And,evenforthosemothersvisited Others 14(5.4) by MANOSHI workers for PNC, it is unclear whether Reproductivecharacteristicsofmothers accuratedetectionandreferralforsepsisoccurred.There- Receivedante-natalcare fore,toreducedeathsfromneonatalinfectionsfuturepro- Yes 223(85.8) gramsshouldfindawaytofirstreachtheneonatesduring thecriticalperiodoffirstdayafterdeliveryandthenreli- No 31(11.9) ablyprovidetreatmentdirectlyoraccesstotreatmentfor Unknown 6(2.3) neonatalinfections. ANCtakenfrom(n=223) Seventy-fivepercentofneonataldeathsinthestudyarea BRAC(CHWs) 157(70.4) occurred within 7 days after birth, a finding in concur- MBBSdoctors 36(16.1) rence with other studies from Bangladesh [4,19]. Other Nurse,Midwives,CSBA 24(10.8) researchershaveshownthatthetimingofvisitbytrained Drugsellers 2(0.9) personnel is crucial for child survival and that receiving Others 4(1.8) visitonthedayofbirthreducedtheriskofneonatalmor- Placeofdelivery talitybytwo-thirdsamongneonateswhosurvivedthefirst Home 108(41.5) day [20].However, only 34% ofmothers and neonatesin theMANOSHIprogramareareceivedPNCvisitwithina BRACdeliverycentre 63(24.2) day of delivery [16]. Although the proportion of women Otherfacility 81(31.2) receiving PNC increased with the MANOSHI interven- Onthewaytofacility 5(1.9) tion,betteremphasisonPNCvisitsonthedayofdelivery Unknown 3(1.2) with CHWs trained to provide management, treatment Modeofdelivery and referral has the potential to further reduce neonatal Normalvaginaldelivery 219(84.6) deaths. CaesareanSection(C/S) 35(13.5) Inviewofthelimitedaccesstotrainedhealthcarepro- Unknown 5(1.9) viders that slum residents have [1] and the important Receivedpostnatalcare role that community member’s play in ensuring timely and appropriate medical referral [21], it is important to Yes 151(58.8) understandcommunityperceptionsofcausesofneonatal No 98(38.1) deaths. Specifically researchers have related caregiver/ Khatunetal.BMCPublicHealth2012,12:84 Page6of9 http://www.biomedcentral.com/1471-2458/12/84 Low birth weight 3% Cot death Others 4% 7% Neonatal jaundice 4% Hypothermia 6% Birth asphyxia 42% Birth trauma 7% Neonatal sepsis/ pneumonia 27% Figure3Maincausesofneonataldeaths(2008-9). guardian perceptions about the causes and severity of maternal deaths over the last decade [2]. PPH and diseasestothe seekingoftimelyandappropriatehealth- eclampsia continued to be a leading causes of maternal care[22-24].Ourdatashowthatone-fifthofrespondent deaths similarly indicated by two existing national sur- believe that the supernatural (the evil eyes/spiritual air) veys [2,25], which points to the need for continued causesnewborndeaths.Thisdemonstratesthatattempts attention in these problems. bytheMANOSHIprogramtoeducatemothersinslums The MANOSHI program has used a two pronged about neonatal complications and appropriate referral approachtoreducematernaldeathsduetobirthcompli- have not adequately succeeded and that many more cations:a)mothersareencouragedtocometotheBRAC innovationsandeffortsareneededtochangecommunity delivery centers where trained Urban Birth Attendants perceptionsandimprovehealthseekingbehavior. (Trained TBAs) provide misoprostol to reduce hemor- rhage,identifymaternalcomplicationsduringdeliveryand Maternal deaths refer appropriately; andb) mothersand otherhousehold Maternal deaths remain a problem in Bangladesh membershavebeenprovidedwitheducationaboutdanger despite recent surveys pointing to a 40% reduction in signs of pregnancy and delivery so that they can directly seekappropriatecarewiththehelpofCHWsifcomplica- tions arise during home births. As 19% of the deliveries Table 2Layperception regardingcauses ofdeaths in takeplaceattheBRACdeliverycenterand49%takeplace neonatal period at home [16] with no CHW present, educating commu- Layperceptionofcauseofneonataldeaths N(%) nities effectively about danger signs is crucial. It is also Supernaturalcauses(Evileye/spirit,Algabatas) 60(23.1) necessary to educate the communities and pregnant Prematurebirth 23(8.8) women to seek delivery services from skilled health per- Cold 22(8.5) sonneltopreventbothmaternalandearlyneonataldeaths. It is important to note that none of the family members Lowbirthweight 16(6.2) described supernatural causes for maternal deaths [26], Pneumonia 14(5.4) indicatingthatthecommunityhasreceivedeffectiveedu- Wrongprocessofdelivery 13(5.0) cation.Inaddition,therewasa57%reductioninthepro- Othercauses* 112(43.1) portion of home delivery from 2007 to 2009, and a *Othercausesincludedfever,jaundice,twincouldnotlivealone,Godknows, subsequent increaseinfacility-baseddeliveriesindicating vomiting,unabletoprovidetreatment,lackofgoodqualityoftreatmentat hospital,experienceofviolenceduringpregnancybymother thattheMANOSHIprogramfoundwaystogetpeopleto Khatunetal.BMCPublicHealth2012,12:84 Page7of9 http://www.biomedcentral.com/1471-2458/12/84 Table 3Reproductive and socio-demographic anddelayincareseeking.Inthefuture,MNCHprograms characteristics ofdeceased women in the studyarea could address this misconception by using a simple low during 2008-2009 costdeliverymat[28]thatcanhelplaypeopletorecognize Socio-demographiccharacteristics N(%)(n=38) excessive bleeding and could encourage them to seek Ageatdeath immediateemergencyobstetriccare. 15-19y 3(7.9) Hepatic failure due to viral hepatitis was found to be one of the major underlying causes for maternal death 20-24y 11(28.9) in our study. During the study period, a hepatitis E out- 25-29y 9(23.7) break was observed in the slums of Dhaka [29]. Viral 30-35y 6(15.8) hepatitis increased risk of death by up to 20% during 35-39y 7(18.4) pregnancy [30]. Fecal contamination of drinking water is >40y 2(5.3) the most important cause of hepatitis E epidemic [31] Monthlyhouseholdincome and, therefore, improvements both in the supplying of <5000Tk.(US$66) 23(60.5) safe water and in sanitation in the slums need to be ≥5000Tk.(US$67andmore) 15(39.5) addressed. The MANOSHI program focused on clean Education delivery and appropriate referral but limited its ability to be effective by not addressing detection of, and appro- None 16(42.1) priate referral for, infection such as viral hepatitis during 1-5years 10(26.3) pregnancy. For future reductions of maternal mortality 6+years 9(23.7) it may be important for MNCH programs to identify Unknown 3(7.9) prevalent infections based on location and context and Occupation to at least have the CHWs identify and refer illness Housewife 30(78.9) from these infections during pregnancy. MNCH pro- Garmentsworker 4(10.5) grams could also provide information about disease epi- Others 4(10.5) demics to other public health departments for other Receivedante-natalcare (non-health specific) interventions to reduce disease pre- Yes 31(81.6) valence among neonates and mothers. No 4(10.5) Strengths of the study Unknown 3(7.9) Our study provides data on the causes of neonatal and ANCtakenfrom(n=31) maternal death for urban Bangladeshi slum population; BRAC(CHWs) 16(51.6) this data has not been previously available. With a vital MBBSdoctors 6(19.4) event registration system lacking, we used verbal autop- Unknown 9(29.0) sies to identify the causes of death and ICD-10 codes to Placeofdelivery(n=29) unify the identification of different health conditions. Home 9(31.0) We conducted this research in MANOSHI program BRACdeliverycentre 3(10.3) area and we were notified quickly about the deaths of mothers and neonates through existing program MIS. Governmenthospital 12(41.4) OtherNGO/privatehospital 5(17.2) Limitations of the study Modeofdelivery(n=28) We relied on community reportsofsigns and symptoms Normalvaginaldelivery 21(75) ofillnesstoclassifymedicalcausesofdeath.Thestudydid Caesareansection(C/S) 5(17.9) notcapturealldeathsinthestudyareaassomematernal Unknown 2(7.1) and newborn deaths were missed due to migration after Receivedpostnatalcare(n=28) death and others may not have been identified due to Yes 15(49.5) interpersonal sensitivities. It is also likely early maternal No 12(31.6) deaths due to abortion or ectopic pregnancy were also Unknown 11(28.9) missed. Further, the study was conducted in selected slums and the findings may not be representative of all slums. seek care from trained providers [16]. Researchers have reported that in rural Bangladeshi community members Conclusions and recommendations believed that hemorrhage after delivery is normal [27], We found birth asphyxia and sepsis to be the major whichcouldleadtonon-recognitionofexcessivebleeding causes of neonatal death. In the context of existing Khatunetal.BMCPublicHealth2012,12:84 Page8of9 http://www.biomedcentral.com/1471-2458/12/84 Indirect causes , 21% ( hepatic failure 11%, others 10%) PPH, 37% Unknown, 3% Complication of C/S, 3% Puerperal sepsis, 3% APH, 5% Retained placenta, 3% Eclampsia , 16% Abortion related complications, 11% Figure4Maincausesofmaternaldeaths(2008-9). MNCH program such as MANOSHI, it is possible that MANOSHI program has emphasized clean delivery and ensuring clean delivery may be necessary but not suffi- appropriate referral,leadingtoeducatingthe community cient to reduce neonatal mortality. With health systems andencouragingpeopletoaccesshealthcare facilities for a comprehensive package of services to ensure detection, maternal delivery complications. However, for creating a management and referral for neonatal complication is comprehensivehealthcareforreducingmaternalmortality, needed within the mechanism providing safe delivery. further innovation will be needed to increase access to Innovative approaches to improve access to essential healthcare during pregnancy and appropriate and timely health services for neonatal complications should be referral. tried within the existing health systems to ensure sus- Finally, the process of delivery should be made safe for tainability. Further, much work is needed to educate the mother and neonate. To that end MNCH programs community members about neonatal danger signs and focused on safe delivery only will not achieve their full to engage them in effective care-seeking behavior. potential. A complete package of services from preg- FormaternaldeathswefoundthatPPHisstillthemajor nancy through the neonatal period must be provided for causeofdeaths.Ourstudyshowsthatviralhepatitisisan the urban slums or for other such resource poor-set- important cause of deaths for urban slums. The tings. Considering the serious lack of infrastructure and projected growth of slum areas in Bangladesh and other countries, striving for a complete service package and Table 4Perception ofkey respondent regardingcauses innovative modes of delivery is crucial for achieving mil- ofmaternal deaths lennium development goals. Layperceptionofcauseofmaternaldeaths N(%) Bleedingafterdelivery 7(18.9) Acknowledgements Jaundice 6(15.8) ThisresearchactivitywasfundedbytheBillandMelindaGatesFoundation, Complicationofdelivery 4(10.5) USAandBRAC,Bangladesh.ICDDR,Backnowledgeswithgratitudethe commitmentoftheBillandMelindaGatesFoundationandBRACtoits Retainedplacenta 3(7.9) researchefforts.Wewanttoextendourappreciationtoalltherespondents Convulsion 3(7.9) fromthevariouscommunitiesfortheirwillfulcontributionsandsincere commitmenttowardsfulfillingthisresearchendeavor.Wesincerely Othercauses* 15(39.4) appreciatetheeffortsofthecauseofdeathstudyteamandMANOSHI *Othercausesincludedtransfusedbloodwasbad,difficultbreathing, programstafffortheircooperationandsupport.Wealsoacknowledge Caesariansection(C/S)donebyunqualifieddoctor,abortion Khatunetal.BMCPublicHealth2012,12:84 Page9of9 http://www.biomedcentral.com/1471-2458/12/84 MahbubElahiChowdhuryandGeorgeSmithfortheircriticalreviewand 15. LawnJE,ManandharA,HawsRA,DarmstadtGL:Reducingonemillion editorialsupport. childdeathsfrombirthasphyxia-asurveyofhealthsystemsgapsand priorities.HealthResPolicySyst2007,5:4. Authordetails 16. AlamN,UddinA,RahmanM,SumiNS,AhmedAM,StreatfieldPK:Manoshi: 1ICDDR,B,68ShaheedTajuddinAhmedSarani,Mohakhali,Dhaka1212, communityhealthsolutionsinBangladesh;midlinesurvey,Dhakaurban Bangladesh.2BRAC,BRACCentre,75Mohakhali,Dhaka1212,Bangladesh. slums.Dhaka:InternationalCentreforDiarrhoealDiseaseResearch, Bangladesh;2009,(ICDDR,Bscientificreportno.113). Authors’contributions 17. BangAT,BangRA,StollBJ,BaituleSB,ReddyHM,DeshmukhMD:Ishome- FKandSRanalyzedthedataanddraftedthepaper;FK,MIandMSS baseddiagnosisandtreatmentofneonatalsepsisfeasibleandeffective? assignedcauseofdeathandprovidedreviewsatdifferentstages;MAAand SevenyearsofinterventionintheGadchirolifieldtrial(1996to2003). ACMdesignedandimplementedthestudyandprovidedintellectualinput JPerinatol2005,25(Suppl1):S62-S71. intotheanalysis;NCandMSprovidedtechnicalexpertise,inputin 18. BaquiAH,El-ArifeenS,DarmstadtGL,AhmedS,WilliamsEK,SerajiHR, interpretingtheresults;andABprovidedoverallsupervisionfordesign, MannanI,RahmanSM,ShahR,SahaSK,SyedU,WinchPJ,LefevreA, analysisanddraftingofthemanuscript. SantoshamM,BlackRE,ProjahnmoStudyGroup:Effectofcommunity- basednewborn-careinterventionpackageimplementedthroughtwo Authors’information service-deliverystrategiesinSylhetdistrict,Bangladesh:acluster- FatemaKhatunAssistantScientist,CentreforEquityandHealthSystems, randomisedcontrolledtrial.Lancet2008,371(9628):1936-1944. ICDDR,B,68,ShaheedTajuddinAhmedSarani,Mohakhali,Dhaka-1212, 19. ChowdhuryME,AkhterHH,ChongsuvivatwongV,GeaterAF:Neonatal Bangladesh;Email:http://[email protected],Fax:+880-2-8826050 mortalityinruralBangladesh:anexploratorystudy.JHealthPopulNutr 2005,23(1):16-24. Competinginterests 20. BaquiAH,AhmedS,ElArifeenS,DarmstadtGL,RosecransAM,MannanI, Theauthorsdeclarethattheyhavenocompetinginterests. RahmanSM,BegumN,MahmudAB,SerajiHR,WilliamsEK,WinchPJ, SantoshamM,BlackRE,Projahnmo1StudyGroup:Effectoftimingoffirst Received:7August2011 Accepted:26January2012 postnatalcarehomevisitonneonatalmortalityinBangladesh:a Published:26January2012 observationalcohortstudy.BMJ2009,339:b2826. 21. ChoudhuryN,NeeloyAA,RashidSF,etal:Maternal,NewbornandChild References HealthPracticesanExploratorystudyofKorailSluminDhaka.Dhaka: 1. NationalInstituteofPopulationResearchandTraining:Bangladesh InternationalCentreforDiarrhoealDiseaseResearch,Bangladesh;2009, demographicandhealthsurvey2007.Dhaka:NationalInstituteof (Manoshiworkingpaperseriesnumber3). PopulationResearchandTraining;2009. 22. DarmstadtGL,KumarV,YadavR,ShearerJC,BaquiAH,AwasthiS,SinghJV, 2. NationalInstituteofPopulationResearchandTraining:Bangladesh MehrotraH,SrivastavaK,GuptaA,SharmaA,WinchPJ,SantoshamM, maternalmortalityandhealthcaresurvey2010.Dhaka:NationalInstitute SakshamStudyGroup:Communityperceptionsofbirthweightinrural ofPopulationResearchandTraining;2011. UttarPradesh,India:implicationsforcareoflow-birth-weightinfants.J 3. NationalInstituteofPopulationResearchandTraining:Bangladeshurban Perinatol2008,28(Suppl2):S53-S60. healthsurvey.Dhaka:NationalInstituteofPopulationResearchand 23. ThaddeusS,NangaliaR,VivioD:Perceptionsmatter:barrierstotreatmentof Training;2006. postpartumhemorrhage.JMidwiferyWomensHealth2004,49(4):293-297. 4. ChowdhuryHR,ThompsonS,AliM,AlamN,YunusM,StreatfieldPK: 24. WinchPJ,AlamMA,AktherA,AfrozD,AliNA,EllisAA,BaquiAH, CausesofneonataldeathsinaruralsubdistrictofBangladesh: DarmstadtGL,ElArifeenS,SerajiMH,BangladeshPROJAHNMOStudy implicationsforintervention.JHealthPopulNutr2010,28(4):375-382. Group:Localunderstandingsofvulnerabilityandprotectionduringthe 5. UnitedNations:Worldurbanizationprospects:the2007revision.New neonatalperiodinSylhetDistrict,Bangladesh:aqualitativestudy.Lancet York:UnitedNations;2008. 2005,366(9484):478-485. 6. AfsanaK,ChowdhuryM,RhodeJ,AhmedF,RahmanH,ChowdhuryT, 25. NationalInstituteofPopulationResearchandTraining:Bangladesh ChowdhuryM,AliA,AhmedSM,Hashima-E-Nasreen,KoblinskyM, MaternalHealthServicesandMaternalMortalitySurvey2001.Dhaka: BhuiyaA,KairySN:Manoshi:aprogrammeforimprovingmaternal, NationalInstituteofPopulationResearchandTraining;2003. neonatalandchildhealthintheurbanslumsofBangladesh.Dhaka: 26. GoodburnEA,GaziR,ChowdhuryAMR:BeliefsandPracticesRegarding InternationalCentreforDiarrhoealDiseaseResearch,Bangladesh;2009, DeliveryandPostpartumMaternalMorbidityinRuralBangladesh.Studies (Manoshiworkingpaperseriesnumber1). inFamilyPlanning1995,26(1):22-32. 7. AhsanKZ,StreatfieldPK,AhmedSM:Manoshi:communityhealth 27. SibleyLM,HruschkaD,KalimN,KhanJ,PaulM,EdmondsJK,KoblinskyMA: solutionsinBangladesh;baselinesurvey.InDhakaurbanslums.Dhaka CulturaltheoriesofpostpartumbleedinginMatlab,Bangladesh: InternationalCentreforDiarrhoealDiseaseResearch,Bangladesh;2007, implicationsforcommunityhealthintervention.JHealthPopulNutr2009, (ICDDR,Bscientificreportno.104). 27(3):379-390. 8. WorldHealthOrganization:WHOtechnicalconsultationonverbal 28. QuaiyumM:Asimplemethodofmeasuringpostpartumbloodlossin autopsytools.Geneva:WHO;2005. Bangladesh.ProceedingsofInternationalCongressonEvidencebased 9. InternationalStatisticalClassificationofDiseasesandRelatedHealth interventionstopreventpostpartumhemorrhage:Translatingresearchinto Problems. WorldHealthOrganization;2007[http://www.who.int/ practiceGoa,India;2006,2006. classifications/icd/en/],[cited011;. 29. InternationalCentreforDiarrhoealDiseaseresearch,Bangladesh:Outbreak 10. WorldHealthOrganization:HandbookIMCIintegratedmanagementof ofhepatitisEinalowincomeurbancommunityinBangladesh.Health childhoodillnessGeneva:WorldHealthOrganization;2005. SciBull2007,7(3):14-19. 11. NationalInstituteofPopulationResearchandTraining:Bangladesh 30. HussainiSH,SkidmoreSJ,RichardsonP,SherrattLM,CooperBT,O’GradyJG: DemographicandHealthSurvey2004.Dhaka:NationalInstituteof SeverehepatitisEinfectionduringpregnancy.JViralHepat1997,4(1):51-54. PopulationResearchandTraining;2005. 31. KrawczynskiK,AggarwalR,KamiliS:HepatitisE.InfectDisClinNorthAm 12. ManandharSR,OjhaA,ManandharDS,ShresthaB,ShresthaD,SavilleN, 2000,14(3):669-687. CostelloAM,OsrinD:Causesofstillbirthsandneonataldeathsin Dhanushadistrict,Nepal:averbalautopsystudy.KathmanduUnivMedJ Pre-publicationhistory (KUMJ)2010,8(29):62-72. Thepre-publicationhistoryforthispapercanbeaccessedhere: 13. BryceJ,Boschi-PintoC,ShibuyaK,BlackRE,WHOChildHealth http://www.biomedcentral.com/1471-2458/12/84/prepub EpidemiologyReferenceGroup:WHOestimatesofthecausesofdeathin doi:10.1186/1471-2458-12-84 children.Lancet2005,365(9465):1147-1152. Citethisarticleas:Khatunetal.:Causesofneonatalandmaternal 14. EllisM,ManandharD:Progressinperinatalasphyxia.SeminNeonatol1999, deathsinDhakaslums:Implicationsforservicedelivery.BMCPublic 4(3):183-191. Health201212:84.