Analysis of neonatal mortality risk factors in Brazil: a systematic review and meta-analysis of observational studies 夽

(1)

JPediatr(RioJ).2019;95(5):519---530

www.jped.com.br

REVIEW ARTICLE

Analysis of neonatal mortality risk factors in Brazil: a systematic review and meta-analysis of observational studies ^夽

Felipe Camilo Santiago Veloso

^a

, Liliana de Meira Lins Kassar

^b

,

Michelle Jacintha Cavalcante Oliveira

^b^,^c

, Telmo Henrique Barbosa de Lima

^a^,^c

, Nassib Bezerra Bueno

^d

, Ricardo Queiroz Gurgel

^e

, Samir Buainain Kassar

^c^,∗

aUniversidadeEstadualdeCiênciasdaSaúdedeAlagoas(UNCISAL),CursodeMedicina,Maceió,AL,Brazil

bUniversidadeFederaldeAlagoas(UFAL),FaculdadedeMedicina,Maceió,AL,Brazil

cCentroUniversitárioTiradentes,CursodeMedicina,Maceió,AL,Brazil

dUniversidadeFederaldeAlagoas(UFAL),FaculdadedeNutric¸ão,Maceió,AL,Brazil

eUniversidadeFederaldeSergipe(UFS),FaculdadedeMedicina,Aracaju,SE,Brazil

Received19October2018;accepted18December2018 Availableonline24April2019

KEYWORDS Neonatalmortality;

Riskfactors;

Brazil;

Reviewsystematic;

Meta-analysis

Abstract

Objective: Toidentify,usingasystematicreviewandmeta-analysisofobservationalstudies, whichriskfactorsaresigniﬁcantlyassociatedwithneonatalmortalityinBrazil,andtobuilda comprehensivenationalanalysisonneonatalmortality.

Sources: Thisreviewincludedobservationalstudiesonneonatalmortality,performedbetween 2000and2018inBraziliancities.TheMEDLINE,Elsevier,Cochrane,LILACS,SciELO,andOpen- Greydatabaseswereused.Forthequalitativeanalysis,theNewcastle-OttawaScalewasused.

Forthequantitativeanalysis,thenaturallogarithmsoftheriskmeasuresandtheirconﬁdence intervalswereused,aswellastheDerSimonianandLairdmethodasarandomeffectsmodel, andtheMantel---Haenszelmodelforheterogeneityestimation.Aconﬁdencelevelof95%was considered.

Summaryoffindings: Thequalitativeanalysisresultedinsixstudiesoflowandfourstudiesof intermediate-lowbiasrisk.Thefollowingexposurefactorsweresignificant:absenceofpartner, maternal age≥35 years, malegender, multiple gestation, inadequateand absentprenatal care, presence ofcomplicationsduring pregnancy,congenital malformation intheassessed pregnancy,Apgar<7atthefifthminute,lowandverylowbirthweight,gestationalage≤37 weeks,andcaesareandelivery.

夽 Pleasecitethisarticleas:VelosoFC,KassarLM,OliveiraMJ,LimaTH,BuenoNB,GurgelRQ,etal.Analysisofneonatalmortalityrisk factorsinBrazil:asystematicreviewandmeta-analysisofobservationalstudies.JPediatr(RioJ).2019;95:519---30.

StudyconductedatUniversidadeEstadualdeCiênciasdaSaúdedeAlagoas,Maceió,AL,Brazil.

∗Correspondingauthor.

E-mail:[email protected](S.B.Kassar).

https://doi.org/10.1016/j.jped.2018.12.014

(2)

520 VelosoFCetal.

Conclusion: Themostsigniﬁcantriskfactorspresentedinthisstudyaremodiﬁable,allowing aimingatarealreductioninneonataldeaths,whichremainhighinthecountry.

4.0/).

PALAVRAS-CHAVE Mortalidadeneonatal;

Fatoresderisco;

Brasil;

Revisãosistemática;

Metanálise

AnálisedosfatoresderisconamortalidadeneonatalnoBrasil:umarevisão sistemáticaemetanálisedeestudosobservacionais

Resumo

Objetivo: Identiﬁcar,atravésdeumarevisãosistemáticaedametanálisedeestudosobserva- cionais,quaisfatoresderiscoassociam-sesigniﬁcativamentecomamortalidadeneonatalno Brasileconstruirumaanálisenacionalabrangentesobreamortalidadeneonatal.

Fontes: Foramavaliadososestudosobservacionaissobremortalidadeneonatalrealizadosentre osanos2000e2018emcidadesbrasileiras.Usaram-seasbasesMEDLINE,Elsevier,Cochrane, LILACS,SciELOeOpenGrey.Paraaanálisequalitativa,foiutilizadaaEscalaNewcastle-Ottawa.

Paraa análisequantitativa, foramutilizadosos logaritmosnaturais dasmedidas de riscoe deseus intervalosde confiança, ométodode DerSimonianeLairdcomo modelode efeitos aleatórios,eomodelodeMantel-Haenszelparaestimativadaheterogeneidade.Considerou-se níveldeconfiançade95%.

Resumodosachados: Aanálisequalitativaresultouemseisestudosdebaixoequatroestudos deintermediário-baixoriscodeviés.Foramsigniﬁcativososseguintesfatores deexposic¸ão:

ausênciadecompanheiro,idadematerna≥35anos,sexomasculino,gestaçãomúltipla,pré- natalinadequadoeausente,presençadeintercorrênciasduranteagestação,demalformação congênitanagestaçãoemestudo,Apgar<7noquintominuto,baixoemuitobaixopesoao nascer,idadegestacional≤37semanasepartocesariano.

Conclusão: Osfatoresderiscomaissignificativosapresentadosnesteestudosãomodificáveis,o quepossibilitaalmejarumareduçãorealdasmortesneonatais,queaindapermanecemelevadas nopaís.

0/).

Introduction

Neonataldeathisdeﬁnedasthedeathofanewbornbefore completing28 daysoflife.^1,2 In Brazil,the neonatalmor- talityrate in 2016 consistedof eightneonatal deathsper 1000livebirths---asigniﬁcantreductionwhencomparedto 1990,whenitwas26neonataldeathsper1000livebirths.^3,4 In 2011,the country reached goal four of the Millennium DevelopmentGoals,acommitmentofthemembergovern- mentsoftheUnitedNationstoreduceinfantmortality.³The Brazilianneonatalmortalityrate,despitethisreduction,is stillhighandunevenlydistributedregionally,withthedif- ferencebetweentheNorth/NortheastandSouth/Southeast regionsbeingthemostdiscrepant.⁵

The reduction of thisrate is a goaltobe pursuedand areality thatcan beachieved,sincepreventablefactors, suchaslow birth weight,prematurity,neonatal asphyxia, and low-quality prenatal care are the main reasons for this high value.^6---8 However, the maternal causes consti- tutea complementaryfactor of equalimportance,witha signiﬁcantcontributiontoneonatalmortalityindeveloping countries.^5,8Nevertheless,thereisnosystematicevaluation ofallthesefactorsinBrazilatthistime.

A systematic review with meta-analysis constitutes a real possibility to assess the impact of risk factors on

neonatal mortality in the Brazilian territory, aiding clini- caldecision-makingandtheprofessionals’epidemiological understandingof thesituation,especiallybypublichealth managers inBrazil.⁹ The studyaimstoidentify,througha systematicreviewandmeta-analysisofobservationalstudies, which risk factors are signiﬁcantly associated with neonatalmortalityinBrazil,andtobuildacomprehensive nationalanalysisonthesubject.

Methods

A PROSPERO protocol (https://www.crd.york.ac.uk/

prospero/)waspublishedunderNo.CRD42018108716.

Studyselection,dataextraction,andriskofbiasanalysis wereindependentlyperformedbytwoauthors.Thediffer- ences were resolved by consensus between the authors;

if there was noconsensus, a third authorwould end the impasse.

Eligibilitycriteria

The inclusion criteria consisted of the following items:

observationalstudiesonneonatalmortality;studiescarried outbetween2000and2018;studiescarriedoutinBrazilian

(3)

Neonatalmortality:systematicreview 521 cities; studies that usedadjusted risk measures, together

withtheirconﬁdenceintervals;studiesthatemployedone ormoreofthefollowingexposurefactors:maternalschooling and/or marital status of the mother and/or maternal age(years)and/or gender of thenewbornand/or typeof delivery and/or previous stillbirth and/or type of gestation and/or adequacy of prenatal care and/or maternal complications and/or birth weight (g) and/or gestational age(weeks)and/orcongenitalmalformationand/orApgar atthe5thminuteoflife.

Databases

The main databases used for this study

were MEDLINE, accessed via PubMed

(https://www.ncbi.nlm.nih.gov/pubmed/) and Bireme (http://bvsalud.org/). The secondary database was Else- vier (https://www.sciencedirect.com/). The regional databases were LILACS (http://bvsalud.org/) and SciELO (http://www.scielo.org/php/index.php). The grey litera- turedatabasewasOPENGREY(http://www.opengrey.eu/).

Searchstrategy

The search strategy was deﬁned based on the PECOS strategy (Patients | Exposure | Comparison | Out- come | Type of study). Therefore, two base strategies were constructed, one in the English language, using MeSH terms (https://www.ncbi.nlm.nih.gov/mesh), and another in the Portuguese language, using DeCS terms (http://decs.bvs.br/),asfollows:

1. (earlyneonatalmortalityORneonatalmortalityORperi- natalmortalityORinfantmortality)AND(riskfactorsOR socioeconomicfactorsORlowbirthweightORperinatal care)AND(BrazilORBrasil)AND(casecontrolstudiesOR cohortstudies);

2. (mortalidade neonatal OR mortalidade perinatal OR mortalidade infantil) AND(fatores de risco OR fatores socioeconômicos OR baixo peso ao nascer) AND(caso- controleORcoorte).

Studyselection

Study selection consisted of three stages: title phase, abstractphase,andfull-articlephase,whichstarted after thesearchesinallthedatabases.Inadditiontothisselec- tion,theinclusionofthearticleswascomplementedbythe searchforhelpfromspecialistsintheareauponﬁndingstud- iesrelevanttothisreview.Duplicateswereremoved with thehelpofMendeleysoftware(Mendeley,version1.17.1,1 Elsevier---NY,USA).

Dataextraction

Aspreadsheet wascreated usingMicrosoft Excelsoftware (Microsoft,Excelsoftware,version2016,WA,USA)contain- ingthefollowingelements:nameoftheauthorsofthestudy;

city and state of thestudy; year of study; type ofstudy;

totalnumberofstudysubjects;adjustedriskmeasures,and conﬁdenceintervalsoftheincludedexposurefactors.

Factorsofexposureandoutcome

The assessed exposure factors were: ‘‘maternal schooling,’’‘‘mother’smaritalstatus,’’‘‘maternalage(years),’’

‘‘newborn’sgender,’’‘‘previousstillbirth,’’‘‘typeofpregnancy,’’‘‘prenatalcareadequacy,’’‘‘complicationsduring the pregnancy,’’ ‘‘birth weight,’’ ‘‘congenital malformation,’’ ‘‘Apgar at the 5th minute,’’ ‘‘gestational age (weeks),’’and‘‘typeofdelivery.’’Theoutcomewasneona- talmortality.

Riskofbiasinindividualstudies

TheriskofbiaswasbasedontheNewcastle-Ottawa Scale (NOS)¹⁰toassessthequalityofnon-randomizedstudiesina meta-analysis.TheNOSevaluatesthebiasregardingthree aspects:the selection of study groups; the comparability ofgroups;thedeterminationofexposurefor case---control studies,ortheoutcomeofinterestforcohortstudies.Amax- imumoffourstarsisgiventotheﬁrstaspect;tothesecond, twostars;and,ﬁnally,tothethird,threestars.

Analysisofdata

Thedataanalysiswasbasedonthequantitativestudyofthe variables.Stata12software(StataCorp---CollegeStation--- TX,USA)wasusedforthisinvestigation.Theanalyzeddata werethe adjusted risk measures in each exposure factor analyzed,inadditiontotheirconfidenceintervals.Thenat- urallogarithmsoftheriskmeasuresandoftheirconfidence intervals,theDerSimonian andLaird methodasa random effectsmodel,andtheMantel---Haenszelmodelforestima- tionofheterogeneitywereused.Aconfidencelevelof95%

wasconsidered.

Heterogeneity was considered important when I² was

>50%.If necessary,the sensitivity analysiswas performed byremovingthestudiesonebyonetoinvestigatetheorigin ofthe divergence.The evaluation ofpublicationbias was performedusingafunnelplot.

Results

Studyselection

The searchin thedatabases identiﬁed10,274 articles.Of these,61wereduplicates,resultingin10,203articles.After readingallthetitles,98articleswerechosentobeanalyzed intheabstractphase.Attheendofthisphase,17articles wereselected,ofwhich tenarticles wereincluded inthe study3,6,11---18(Fig.1).Theoverallcharacteristicsoftheten studiesincludedintheanalysisareavailableinTable1and thecharacteristicsrelevanttothequantitativeanalysisare showninTable2.

(4)

522 VelosoFCetal.

Articles identified through searches in the databases

(n =10274)

Additional articles identified through other sources

(n =0)

Excluded articles (n =10151)

Selection phase: Abstract (n =45) 1- Did not have neonatal mortality

as the outcome (n =14) 2 - Not a case-control or cohort study

(n =1)

3 - Did not occur in any Brazilian city (n =1)

4- Not performed between 2000 and 2018 (n =14) 5- No analysis of risk factors

(n =6) Excluded articles

(n =42)

1- Did not use adjusted risk measures (n =18) 2- Exclusive study on infant

mortality (n =7) 3- Not a case-control or

cohort study (n =1) 4- Analysis in several countries

(n =1)

5- Exclusive study on perinatal mortality (n =2)

7- Did not study neonatal mortality as an outcome (n =1) 6- Not performed between 2000

and 2018 (n =2)

8- Exclusive study on fetal mortality (n =1) 9- Analyzed neonatal mortality

in a conditional manner (n =7)

Articles after the removal of duplicates (n =10203)

IdentificationScreeningEligibilityInclusion

Article screening (n =10203)

Selection phase: Title (n= 10106)

Selection phase: full articles

Full articles evaluated after the screening

(n =52)

Studies included in the qualitative analysis

(n =10)

Studies included in the quantitative analysis

(meta-analysis) (n =10)

1 - Excluded for having been published before 2000 (n =813) 2 - Excluded because they were not

related to the study in general (n =9293)

Figure1 Flowchartofstudyselection.

Biasriskassessmentofincludedstudies

The ten studies included in the present analysis were qualitatively assessed using the Newcastle-Ottawa Scale.

Four studies^3,13,16,17 obtained a high classiﬁcation, i.e., theyshowlowriskofbias.However,sixstudies6,11,12,14,15,18

obtainedan intermediate-highclassiﬁcation,thusshowing anintermediate-lowrisk(Table3).

Analysisofdataoftheincludedstudies

Fig.2shows theoverallassociation betweentheanalyzed factorsandtheassociation withneonatalmortality,repre- sentedintheimagebytheﬁnaladjustedoddsratios(faOR) andtheirrespectiveconﬁdenceintervals.Theresultofeach exposurefactorcanbeseenbelow,separately.

Maternalschooling

Maternal schooling could only be analyzed in two ways:

completeelementaryschool(8---11yearsof schooling)and incomplete elementary school (4---7 years of schooling).

Complete elementary school did not show a signiﬁcant association with neonatal mortality (faOR=1.853; 95%

CI: [0.417---8.241]; p=0.418; I²=87.4%; p=0.005). The same was observed for incomplete elementary school (faOR=1.727, 95% CI: [0.466---6.400], p=0.414, I²=84.8%, p=0.010).

Mother’smaritalstatus

The pregnancy of mothers who did not have a partner showed a signiﬁcantly higher chance of neonatal mortality(faOR=2.236,95%CI:[1.630---3.068],p<0.001,I²=0.0%, p=0.344).

Maternalage

The pregnancy of mothers aged 35 years or older showed a signiﬁcantly higher chance of neonatal mortality(faOR=1.568,95%CI:[1.141---2.154],p=0.006,I²=0.0%, p=0.882).

(5)

Neonatalmortality:systematicreview 523 Table1 Overallcharacteristicsofincludedstudies.

Analyzedstudies Year City/State Studytype NumberofsubjectsAnalyzedvariables I Lanskyetal.³ 2014 Severalcitiesin

theﬁveBrazilian macro-regions

Cohort 24,329 Maternalschooling|

Maternalmaritalstatus| Maternalage|Newborn’s gender|Previousstillbirth

II Limaetal.¹⁷ 2012 Serra/ES Cohort 32,275 Maternalschooling|

Maternalage|Typeof pregnancy|Adequacyof prenatalcare|Birthweight

|Typeofdelivery III Zaninietal.¹³ 2011 35

micro-regions/RS

Birthweight|Congenital malformation|Apgarinthe 5thminute|Typeof delivery|Gestationalage V Almeidaand

Barros¹²

2004 Campinas/SP Case---control 351 Birthweight|Gestational age

VI Nascimento etal.¹¹

2012 Fortaleza/CE Case---control 396 Newborn’sgender|

Adequacyofprenatalcare| Birthweight|Gestational age

VII Kassaretal.⁶ 2013 Maceió/AL Case---control 408 Adequacyofprenatalcare| Birthweight

VIII Schoepsetal.¹⁴ 2007 SãoPaulo/SP Case---control 459 Maternalmaritalstatus| Adequacyofprenatalcare| Complicationsduring pregnancy|Birthweight| Gestationalage

IX Paulucciand Nascimento¹⁵

2007 Taubaté/SP Case---control 392 Birthweight|Congenital malformation

X Mendesetal.¹⁸ 2006 CaxiasdoSul/RS Case---control 354 Adequacyofprenatalcare| Birthweight|Typeof delivery|Gestationalage

Newborn’sgender

Malenewbornshadasigniﬁcantlyhigherchanceofneona- talmortality(faOR=1.591,95%CI:[1.193---2.121],p=0.002, I²=0.0%,p=0.362).

Previousstillbirth

Thepregnancyofmotherswhohadapreviousstillbirthdid notshow a signiﬁcant association withneonatalmortality (faOR=2.090, 95% CI:[0.762---5.733], p=0.152, I²=90.6%, p=0.001).

Typeofpregnancy

Amultiple pregnancyshowedasigniﬁcantlyhigherchance ofneonatalmortality(faOR=3.361,95%CI:[1.612---7.009], p=0.001,I²=48.7%,p=0.163).

Adequacyofprenatalcare

Itwasonlypossibletoanalyzetheinadequacyandabsence ofprenatalcare.Inadequateprenatalcareshowedasigni- ﬁcantlyhigherchanceofneonatalmortality(faOR=2.182, 95% CI:[1.761---2.703], p<0.001, I²=0.0%, p=0.771). The same was observed with the absence of prenatal care (faOR=6.799,95%CI:[1.534---30.147],p=0.012,I²=78.8%, p=0.030).

(6)

524VelosoFCetal.

Table2 Characteristicsofthestudiesincludedinthequantitativeanalysis.

Analyzedvariables Factors Analyzedstudies AdjustedOR 95%CI

Maternalschooling CompleteElementarySchool(8---11years) (3)

(17)

4.24 0.92

[1.61---11.16]

[0.59---1.41]

IncompleteElementarySchool(4---7years) (3) (17)

3.60 0.94

[1.43---9.07]

[0.60---1.47]

Maternalmaritalstatus Motherwithoutapartner (3)

(14)

2.49 1.80

[1.69---3.66]

[1.00---3.00]

Maternalage Maternalage≥35years (3)

(17)

1.62 1.54

[0.95---2.78]

[1.04---2.29]

Newborn’sgender Malegender (3)

(11)

1.49 2.09

[1.08---2.05]

[1.09---4.03]

Previousstillbirth Presenceofpreviousstillbirth (3)

(13)

3.62 1.29

[2.05---6.41]

[1.01---1.65]

Typeofpregnancy Multiplepregnancy (3)

(17)

4.79 2.26

[2.37---9.68]

[1.03---4.96]

Adequacyof prenatalcare

Inadequateprenatalcare (3)

(11) (6) (14) (16) (18)

2.84 2.03 2.49 2.00 3.04 1.45

[1.44---5.62]

[1.03---3.99]

[1.14---5.40]

[2.00---3.50]

[1.61---5.97]

[0.44---4.77]

Absenceofprenatalcare (14)

(17)

16.10 3.48

[4.70---55.40]

[1.86---6.49]

Complicationsduring pregnancy

Presenceofcomplicationsduringpregnancy (3) (14)

6.07 8.20

[3.85---9.55]

[5.00---13.50]

Birthweight Birthweightbetween1500and2499g (3)

(17) (13)

5.19 2.36 5.46

[2.44---11.04]

[1.46---3.83]

[4.07---7.31]

Birthweightlessthan1500g (3)

(17) (13)

32.27 5.49 41.15

[12.65---82.35]

[2.52---11.96]

[29.25---57.87]

Birthweightlessthan2500g (12)

(11) (6) (14) (15) (16) (18)

3.84 14.75 2.57 17.30 22.80 9.46 34.60

[1.18---12.50]

[5.26---41.35]

[1.16---5.72]

[8.40---35.60]

[7.52---69.12]

[4.06---23.60]

[3.84---311.50]

Congenitalmalformation Presenceofcongenitalmalformation (3)

(13) (15) (16)

16.55 14.79 12.91 6.12

[6.47---42.38]

[11.18---19.59]

[2.13---78.38]

[2.14---15.67]

(7)

Neonatalmortality:systematicreview525

Table2(Continued)

Analyzedvariables Factors Analyzedstudies AdjustedOR 95%CI

Apgaratthe5thminute PresenceofApgarscore<7atthe5thminute (3) (13) (16)

15.79 11.39 19.08

[6.54---38.14]

[8.51---15.25]

[8.78---42.92]

Gestationalage Gestationalage<37weeks (12)

(11) (14) (13) (16) (18)

5.37 3.41 8.80 1.84 6.09 43.46

[1.83---17.98]

[1.09---4.03]

[4.30---17.80]

[1.42---2.39]

[2.58---15.23]

[10.26---83.97]

Typeofdelivery Caesareandelivery (17)

(13) (18)

1.52 0.80 2.36

[1.17---1.97]

[0.68---0.93]

[0.70---7.61]

(8)

526 VelosoFCetal.

Table3 Biasriskassessmentofincludedstudies.

Analyzedstudies Selection Comparability Outcome Total Classiﬁcation

Cohort

(3) *** * *** 7*/9* Highquality

(17) *** * *** 7*/9* Highquality

(13) *** * *** 7*/9* Highquality

(16) *** * *** 7*/9* Highquality

Case---control

(12) *** * ** 6*/9* Intermediate-highquality

(15) *** * *** 7*/9* Highquality

(18) *** * *** 7*/9* Highquality

Mother finished elementary school (8 to 11 years) Exposure factors

Mother did not finish elementary school (4 to 7 years) Mother without a partner

Maternal age ≥ 35 years Male newborn

Previous stillbirth in prior gestation Multiple pregnancy

Inadequate prenatal care Absence of prenatal care

Presence of complications during the pregnancy Birth weight between 1,500 and 2,499 grams Birth weight < 1,500 grams

Birth weight < 2,500 grams

Apgar score < 7 at the 5th minute Gestational age < 37 weeks

1.23(0.69, 2.21) 5.74 (2.59, 12.72) 12.40 (9.55, 16.10) 14.03 (10.85, 18.13) 9.88 (4.92, 19.83) 19.75 (5.56, 70.22) 4.04 (2.26, 7.24) 6.96 (4.98, 9.73) 6.80 (1.53, 30.15) 2.18 (1.76, 2.70) 3.36 (1.61, 7.01) 2.09 (0.76, 5.73) 1.59 (1.19, 2.12) 1.57 (1.14, 2.15) 2.24 (1.63, 3.07) 1.73 (0.47, 6.40) 1.85 (0.42, 8.24) OR (95% CI)

Caesarean delivery

Presence of congenital malformation in the assessed pregnancy

Figure2 Overallviewoffactorsassociatedwithneonatalmortality.Abbreviations:OR,oddsratio;CI,conﬁdenceinterval.

Complicationsduringpregnancy

The presence of complications during pregnancy showed a signiﬁcantly higher chance of neonatal mortality (faOR=6.961, 95% CI: [4.979---9.733], p<0.001, I²=0.0%, p=0.381).

Birthweight

Itwasonlypossible toanalyzebirthweightbetween1500 and2499g(faOR=4.044, 95%CI:[2.260---7.238],p<0.001, I²=77.0%,p=0.013),lowerthan1500g(faOR=19.752;95%

CI: [5.556---70.223]; p<0.001; I²=90.7%; p<0.001), and lower than 2500g (faOR=9.876; 95% CI: [4.920---19.826];

p<0.001;I²=69.4%;p=0.003).Afterthesensitivityanaly- sis,birthweightbetween1500and2499g(faOR=5.424,95%

CI:[4.128---7.126],p<0.001,I²=0.0%,p=0.902),lowerthan 1500g (faOR=39.995, 95% CI: [29.026---55.109], p<0.001, I²=0.0%, p=0.633), andlowerthan 2500g (faOR=15.519, 95% CI: [9.997---24,092]; p<0.001, I²=0.0%, p=0.684)

continuedtoshowasigniﬁcantlyhigherchanceofneonatal mortality after removal of one study in the ﬁrst and the second described situations,¹⁷ and after the removal of anothertwointhethirdsituation.^6,12

Congenitalmalformation

Newbornswithacongenitalmalformationshowedasigniﬁ- cantlyhigherchance ofneonatalmortality(faOR=14.025, 95%CI:[10.849---18.132],p<0.001,I²=0.0%,p=0.402).

Apgarscoreatthe5thminute

ThepresenceofanApgarscorelessthan7atthe5thminute showedasigniﬁcantlyhigherchance ofneonatalmortality (faOR=12.397,95%CI:[9.545---16.102],p<0.001,I²=0.0%, p=0.417).

(9)

Neonatalmortality:systematicreview 527

1.55 (1.20, 2.00) 7.13 (4.33, 11.74) 12.40 (9.55, 16.10) 14.03 (10.85, 18.13) 15.52 (10.00, 24.09) 39.99 (29.03, 55.11) 5.42 (4.13, 7.13) 6.96 (4.98, 9.73) 6.80 (1.53, 30.15) 2.18 (1.76, 2.70) 3.36 (1.61, 7.01) 2.09 (0.76, 5.73) 1.59 (1.19, 2.12) 1.57 (1.14, 2.15) 2.24 (1.63, 3.07) 1.73 (0.47, 6.40) 1.85 (0.42, 8.24) OR (95% CI)

Mother finished elementary school (8 to 11 years) Exposure factors

Mother did not finish elementary school (4 to 7 years) Mother without a partner

Maternal age ≥ 35 years Male newborn

Previous stillbirth in prior gestation Multiple pregnancy

Inadequate prenatal care Absence of prenatal care

Presence of complications during the pregnancy Birth weight between 1,500 and 2,499 grams Birth weight < 1,500 grams

Birth weight < 2,500 grams

Presence of congenital malformation in the assessed pregnancy Apgar score < 7 at the 5th minute

Gestational age < 37 weeks Caesarean delivery

Figure3 Finalassociationofthefactorsassociatedwithneonatalmortalityaftertheheterogeneityanalysis.Abbreviations:OR, oddsratio;CI,conﬁdenceinterval.

Gestationalage

A gestational age <37 weeks showed a signiﬁcantly higherchanceof neonatalmortality(faOR=5.738,95%CI:

[2.589---12.716], p<0.001, I²=86.9%, p<0.001). After the sensitivity analysis, it continued to show a signiﬁcantly higher chance of neonatal mortalityafter the removalof threestudies^11,13,18 (faOR=7.130; 95% CI:[4.329---11.742];

p<0.001;I²=0.0%;p=0.706).

Typeofdelivery

Caesarean delivery did not show a signiﬁcant association with neonatal mortality (faOR=1.233, 95% CI:

[0.688---2.210],p=0.483,I²=89.7%,p<0.001).Whenexplor- ingheterogeneity,therewasasigniﬁcantlygreaterchance of neonatal mortality after a study was removed¹⁷ (faOR=1.551, 95% CI: 1.202---2.000, p=0.001, I²=0.0%, p=0.480).

It wasnotpossible toperform asensitivity analysison thefollowingfactors:completeelementaryschool,incompleteelementaryschool,presenceofpreviousstillbirth,and absenceofprenatalcare,sincetheseanalyzesconsistedof onlytwostudies.Fig.3shows theﬁnaloverallassociation afterthesensitivityanalysis.

Regarding the publication bias, it was not possible to evaluateitdue tothe numberofstudiesincluded ineach analysis.Table4showsaﬁnalsummary-tableoftheresults obtainedinthequantitativeanalysis.

Discussion

The study showed that factors related to the mother (absence of partner, age≥35 years, multiple pregnancy, absence or inadequacy of prenatal care, presence of complicationsduringgestation,andCaesareandelivery)and to the newborn (male gender, congenital malformation,

perinatalasphyxia,lowbirthweight,andprematurity)are risksassociatedtoneonatalmortalityinBrazil.

Themain factortobeconsidered inthiscontextislow birth weight and, more precisely, very low birth weight.

Inaprospectivecohortstudy,inwhich neonatalmortality wasanalyzed in very low birth weight preterm infants in northeasternBrazil,it wasobserved that 76% diedwithin the ﬁrst 24h of life.¹⁹ This fact was also observed in anotherprospectivecohort carriedoutin Brazil,basedon theanalysisofbirthsin20publicuniversitycentres,which resulted in a death rate of 30%,²⁰ thus showing that low birthweight is adeterminant factor in theoccurrence of neonatalmortalityinBrazil,representingitsmainisolated predictor.3,6,11---19,21---23Lowbirthweightisconsideredoneof the main risk factors for neonatal mortality in numerous locationsand services, asobserved in the studies carried outin Mexico,²⁴ Jordan,²⁵ Trinidad&Tobago,²⁶ theUnited States,²⁷ andChina.²⁸

Similarlytolow birth weight,otherfactors aremodifi- able,suchasperinatalasphyxia,prematurity,complications during pregnancy, prenatal care, caesarean delivery, older maternal age, and absence of a partner at home.3,6,11,14,17,19,21,22,29---33 Inthiscontext,itis worthhigh- lightingtheprenatalcare,which,whenproperlyperformed, contributesto the reduction of other modifiable factors, mainly prematurity and low birth weight.3,6,11,14,16---18 This importancecanbe verifiedin a cross-sectionalstudy car- riedoutinamunicipalityintheSoutheastregionofBrazil, whichshowedthat26.8%ofthedeathscouldhavebeenpre- ventedifanadequateprenatalcarehadbeenperformed.³⁴ Asimilarexample canbefound inanothercross-sectional studyconductedinseveralmunicipalitiesinastate inthe BrazilianSoutheastregion,whichshowedthat47%ofdeaths couldhavebeenpreventedthroughtheimplementationof earlyprevention,diagnosis,andtreatmentactions.³⁵These ratesindicate problemsin thecare giventothepregnant

(10)

528VelosoFCetal.

Table4 Summaryoftheresultsobtainedinthequantitativeanalysis.

FACTORS Studiesincludedintheanalysis Totalnumberofanalyzedsubjects FinaladjustedOR Final(95%)CI p I² p

Completeelementaryschool(8---11years) (3)(17) 56,604 1.853 [0.417---8.241] 0.005 87.4% 0.418

Incompleteelementaryschool(4---7years) (3)(17) 56,604 1.727 [0.466---6.400] 0.010 84.8% 0.414

Motherwithoutpartner^b (3)(14) 24,788 2.236 [1.630---3.068] 0.344 0.0% <0.001

Maternalage≥35years^b (3)(17) 56,604 1.568 [1.141---2.154] 0.882 0.0% 0.006

Malegender^b (3)(11) 24,725 1.591 [1.193---2.121] 0.362 0.0% 0.002

Presenceofpreviousstillbirth (3)(13) 162,736 2.090 [0.762---5.733] 0.001 90.6% 0.152

Multiplepregnancy^b (3)(17) 56,604 3.361 [1.612---7.009] 0.163 48.7% 0.001

Inadequateprenatalcare^b (3)(11)(6)(14)(16)(18) 41,825 2.182 [1.761---2.703] 0.771 0.0% <0.001

Absenceofprenatalcare^b (14)(17) 32,734 6.799 [1.534---30.147] 0.030 78.8% 0.012

Presenceofcomplicationsduringpregnancy^b (3)(14) 24,788 6.961 [4.979---9.733] 0.381 0.0% <0.001

Birthweightbetween1500and2499g (3)(17)^a(13) 195,011 5.424 [4.128---7.126] 0.902 0.0% <0.001

Birthweight<1500g^b (3)(17)^a(13) 195,011 39.995 [29.026---55.109] 0.603 0.0% <0.001

Birthweight<2500g^b (12)^a(3)(6)^a(14)(15)(16)(18) 18,239 15.519 [9.997---24.092] 0.684 0.0% <0.001

Presenceofcongenitalmalformation^b (3)(13)(15)(16) 179,007 14.025 [10.849---18.132] 0.402 0.0% <0.001

PresenceofApgarscore<7atthe5thminute^b (3)(13)(16) 178,615 12.397 [9.545---16.102] 0.417 0.0% <0.001 Gestationalage<37weeks^b (12)(11)â(14)(13)â(16)(18)â 155,846 7.130 [4.329---11.742] 0.706 0.0% <0.001

Caesareandelivery^b (17)^a(13)(18) 171,036 1.551 [1.202---2.000] 0.480 0.0% 0.001

a Indicatesstudiesthatwereasourceofheterogeneityineachanalysis.

b Indicatessigniﬁcantfactor(s).

(11)

Neonatalmortality:systematicreview 529 woman, due tothe lack of professionalqualiﬁcation or a

deﬁcientstructure.21,32,33,36---39

As with prenatalcare, caesarean delivery,if correctly indicated,reducestheoccurrenceofprematurity,lowbirth weight,andperinatalasphyxia.3,6,7,11---18Oneofthebigprob- lemsistheincorrectindicationandexcessivenumberofthis typeofdelivery.The rateofcaesareandeliveriesin Brazil is56%,wellabovetherecommendedrate,whichis15%.⁴⁰ Inamulticenter,observational,hospital-basedstudycarried outintheNortheastregionofBrazil,itwasobservedthat 13.7%ofneonataldeathswererelatedtocaesareandelivery inhigh-riskneonatalunits.⁴¹

Congenital malformation, multiple pregnancy, and the newborn’s malegender werefound tobedeterminantsof neonatalmortalityinBrazilinthepresent study,butthey arenotmodiﬁable factors.3,6,7,11---18,32 Of these,congenital malformationisthe mostimportantfactor, beingthesec- ond most decisivefactor for neonatalmortality in Brazil.

Across-sectionalobservationalstudycarriedoutinseveral municipalities of astate in the BrazilianSoutheast region showed that 42.8% of infant deaths were caused by this factor.³⁵ Thesametrendcanbeobserved inaprospective cohort carried out in a Mexican city, which showed in a neonatalmortalityrateof32%.²⁴

The knowledge of risk factors is essential for the prevention of neonatal mortality in Brazil, since health professionals and managers, when aware of these con- ditions, can ﬁnd ways to prevent clinical and structural complications.21,32,33,36---39Thus,preventionconsistsoftrain- ingprofessionalsandprovidingadequate structurefor the birth, as well as offering an individualized care to the pregnancies, paying individualized attention to the pregnancies, while taking intoaccount thespeciﬁcity of each condition.32,33,38,39Thisindividualizationhasbecomeworld- wide trend since 2015,⁴² which has resulted in a 5.4%

reduction in global neonatal deaths between 2015 and 2017.⁴³

The present study consisted of six articles with low biasriskandfourarticleswithintermediate-lowbiasrisk, thus increasingthedegree of reliabilityof thisstudy.The reason for the intermediate classiﬁcation is due to the non-randomizationof theinterviewersinthreestudies,in additiontotheinequalitybetweenthelossesofthecaseand controlgroupsin anotherstudy.The formercanbesolved by performingrandomizationandensuring blinddata col- lection,whileequalityoflossesbetweengroupswouldsolve thesecondimpasse.

The cohortstudies were able toequally represent the population,inadditiontomatchingthecontrolgroupinan equivalentmanner. Oneof the studies³ useda structured interview,differingfromtheotherthree,^13,16,17whichused datafromofficialdatabases.Regardingthefinalevaluation, allofthemusedlinkageanalysiswiththeofficialdatabases, inadditiontoestablishingasufficientfollow-upperiodfor theanalysis.Duringfollow-up,oneofthestudies¹⁷ showed alossoflessthan10%oftheindividuals,differingfromthe otherstudies,whichdidnotshowanyloss.

The case---controlstudies, in turn, used linkage to the official database to adapt and represent cases, as well as to select and define controls. At the final evaluation, three studies^14,15,18 used linkage analysis with data from official databases, while the other three^6,11,12 performed

interviews without the blinding procedure. One of the studies¹⁴ showed an unequal loss between the case and controlgroups, butbecauseitwaslessthan10%,theﬁnal effect was not harmed. However, despite this statistical effect,theNewcastle-OttawaScaledoesnotscorethisfact inthecase---controlstudies.Inrelationtotheotherstudies, therateswerethesameinbothgroups.

Theoverallresultofthisstudyshowstheimportanceof theseriskfactorsforneonatalmortalityinBrazil,highlight- ing,inthiscontext,thepowerthatstructural,behavioural, andmedicalchangeshaveinreducingtheserates.

Themostsigniﬁcantriskfactorsshowninthisstudyare modiﬁable,makingitpossibletoachievearealreductionin neonataldeaths,whichremainhighinBrazil.

Funding

ThisworkwasfundedbytheConselhoNacionaldePesquisae DesenvolvimentoCientíﬁcoeTecnológico(No.129015/2017- 2).

Conﬂicts of interest

Theauthorsdeclarenoconﬂictsofinterest.

References

1.LehtonenL,GimenoA,Parra-LlorcaA,VentoM.Earlyneona- taldeath:achallengeworldwide.SeminFetalNeonatalMed.

2017;22:153---60.

2.BlencoweH,CalvertC,LawnJE,CousensS,CampbellOM.Mea- suringmaternal,foetalandneonatalmortality:challengesand solutions.BestPractResClinObstetGynaecol.2016;36:14---29.

3.LanskyS,LimaFricheAA,SilvaAA,CamposD,AzevedoBitten- courtSD,CarvalhoML,etal.BirthinBrazilsurvey:neonatal mortality,pregnancyandchildbirthqualityofcare.CadSaude Publica.2014;30:S192---207.

4.UnitedNationsInter-agencyGroupforChildMortalityEstima- tion.LevelsandTrendsinChildMortalityReport2017.UNICEF;

2017.

5.RodriguesNC,MonteiroDL,AlmeidaAS,BarrosMB,PereiraNeto A,O’DwyerG,etal.Temporalandspatialevolutionofmaternal andneonatalmortalityratesinBrazil,1997---2012.JPediatr(Rio J).2016;92:567---73.

6.KassarSB,MeloAM,CoutinhoSB,LimaMC,LiraPI. Determi- nantsofneonataldeathwithemphasisonhealthcareduring pregnancy,childbirthandreproductivehistory.JPediatr(Rio J).2013;89:269---77.

7.PedrosaLD,SarinhoSW,OrdonhaMR.Análisedaqualidadeda informac¸ãosobre causa básica de óbitos neonatais registra- dosnoSistemadeInformac¸õessobreMortalidade:umestudo paraMaceió, Alagoas,Brasil,2001---2002.CadSaudePublica.

2007;23:2385---95.

8.Almeida MC, Gomes CM, Nascimento LF. Spatial analysis of neonatalmortalityinthestateofSãoPaulo,2006---2010.Rev PaulPediatr.2014;32:374---80.

9.MoherD,ShamseerL,ClarkeM,GhersiD,LiberatiA,Petticrew M,etal.Preferredreportingitemsforsystematicreviewand meta-analysisprotocols(PRISMA-P)2015statement.Syst Rev.

2015:4:1.

10.Wells GA,Shea B, O’Cornell D, Peterson J, Welch V, Losos M, et al. The Newcastle-Ottawa Scale (NOS) for assess- ing the quality of nonrandomised studies in meta-analyses

(12)

530 VelosoFCetal.

[cited 2018 Sep 14]. Available from: http://www.ohri.ca/

programs/clinicalepidemiology/oxford.asp.

11.Nascimento RM, Leite ÁJ, Almeida NM, Almeida PC, Silva CF. Determinantes da mortalidade neonatal: estudo caso- controle em Fortaleza, Ceará, Brasil. Cad Saude Publica.

2012;28:559---72.

12.AlmeidaSD,BarrosMB.Atenc¸ãoàsaúdeemortalidadeneona- tal:estudocaso-controlerealizadoemCampinas,SP.RevBras Epidemiol.2004;7:22---35.

13.ZaniniRR,deMoraesAB,GiuglianiER,RiboldiJ.Determinantes contextuaisdamortalidadeneonatalnoRioGrandedoSulpor doismodelosdeanálise.RevSaudePublica.2011;45:79---89.

14.SchoepsD,Almeida MF, Alencar GP,Franc¸a I Jr, NovaesHM, SiqueiraAA,etal.Fatoresderiscoparamortalidadeneonatal precoce.RevSaudePublica.2007;41:1013---22.

15.PaulucciRS,NascimentoLF.MortalidadeneonatalemTaubaté:

umestudocaso-controle.RevPaulPediatr.2007;25:358---63.

16.GarciaLP,FernandesCM,TraebertJ.Riskfactorsforneonatal deathinthecapitalcitywiththelowestinfantmortalityrate inBrazil.JPediatr(RioJ).2019;95:194---200.

17.LimaEdeF,SousaAI,GriepRH,PrimoCC.Fatoresderiscopara mortalidadeneonatalnomunicípiodeSerra,EspíritoSanto.Rev BrasEnferm.2012;65:578---85.

18.MendesKG,OlintoMT,CostaJS.Case---controlstudyoninfant mortalityinSouthernBrazil.RevSaudePublica.2006;40:240---8.

19.CastroECde,LeiteÁJ,GuinsburgR.Mortalidadecom24horas devidaderecém-nascidospré-termodemuitobaixopesoda RegiãoNordestedoBrasil.RevPaulPediatr.2016;34:106---13.

20.GuinsburgR,DeAlmeidaMF,DeCastroJS,SilveiraRC,CaldasJP, FioriHH,etal.DeathorsurvivalwithmajormorbidityinVLBW infantsbornatBrazilianneonatalresearchnetworkcenters.J MaternNeonatalMed.2016;29:1005---9.

21.TeixeiraGA,CostaFM,MataMS,CarvalhoJB,SousaNL,Silva RA.Riskfactorsforneonatalmortalityinthelifeofﬁrstweek.

RevPesquiCuidFundam.2016;8:4036.

22.BorgesTS,VayegoSA.Fatoresderiscoparamortalidadeneona- talemummunicípionaregiãosul.CiêncSaúde.2015;8:7---14.

23.VargaP,BereczB,GasparicsÁ,DombiZ,VargaZ,JeagerJ,etal.

Morbidityand mortalitytrendsinvery---verylowbirthweight prematureinfantsinlightofrecentchangesinobstetriccare.

EurJObstetGynecolReprodBiol.2017;211:134---9.

24.ReyesJC.Neonatalmortalityandassociatedfactorsinnewborn infantsadmittedtoaNeonatalCareUnit.ArchArgentPediatr.

2018;116:42---8.

25.ObeidatN,KhaderY,BatiehaA,AbdelRazeqN,Al-SheyabN, KhassawnehM.NeonatalmortalityinJordan:secondaryanalysis ofJordanPopulationandFamilyHealthSurvey(JPFHS)data.J MaternNeonatalMed.2019;32:217---24.

26.CupenK,BarranA,SinghV,DialsinghI.Riskfactorsassociated withpretermneonatalmortality:acasestudyusingdatafrom Mt.Hope Women’sHospitalinTrinidad andTobago.Children (Basel).2017;4:E108.

27.RatnasiriAW,ParrySS,AriefVN,DeLacyIH,HallidayLA,DiLibero RJ,etal.Recenttrends,riskfactors,anddisparitiesinlowbirth weightinCalifornia,2005---2014:aretrospectivestudy.Matern HealNeonatolPerinatol.2018;4:15.

28.HonKL,LiuS,ChowJC,TsangKY,LamH,SoKW,etal.Mor- talityand morbidityofextremelylowbirthweightinfantsin HongKong,2010---2017:asingle-centrereview.HongKongMed J.2018;24:460---5.

29.AbdullahA,HortK,ButuY,SimpsonL.Riskfactorsassociated withneonataldeaths:amatchedcase---controlstudyinIndone- sia.GlobHealthAction.2016;9:30445.

30.KaboréR, MedaIB, KoulidiatiJE, MillogoT, KouandaS. Fac- torsassociatedwithveryearlyneonatalmortalityinBurkina Faso: a matched case---control study. Int J Gynecol Obstet.

2016;135:S93---7.

31.ParkJH,ChangYS,AhnSY,SungSI,ParkWS.Predictingmortal- ityinextremelylowbirthweightinfants:comparisonbetween gestationalage,birthweightApgarscore,CRIBIIscore,initial andlowestserumalbuminlevels.PLoSONE.2018;13:e0192232.

32.DemittoMD,Gravena AAF,MilleneC,AgnoloD,AntunesMB, PellosoSM.Gestac¸ãodealtoriscoefatoresassociadosaoóbito neonatal.RevEscEnfermUSP.2017;51:e03208.

33.MirandaMH,FernandesFE,CamposME.Determinantesassoci- adosàmortalidadeperinatalefatoresassociados.RevEnferm UFPEOnline.2017;11:1171---8.

34.VidalE,SilvaSM,TuonRA,ProbstLF,GondinhoBV,PereiraAC, etal.Factorsassociatedwithpreventableinfantdeath:amul- tiplelogisticregression.RevSaudePublica.2018;52:32.

35.DaSilvaPL,CostaAD,FariasHM,RochaLM,OliveiraMA,Dam- ascenoRF.Evitabilidadedamortalidadeinfantilnaregião de saúdedeJanaúba/MonteAzul,MinasGerais,Brasil.JHealBiol Sci.2017;6:35.

36.FilhoAC,SalesIM,AraújoAK, AlmeidaPD, RochaSS.Aspec- tos epidemiológicos da mortalidade neonatal em capital do NordestedoBrasil.RevCuid.2017;8:1767---76.

37.LiaoXP, Chipenda-DansokhoS, LewinA, AbdelouahabN,Wei SQ.AdvancedneonatalmedicineinChina:anationalbaseline database.PLoSOne.2017;12:1---15.

38.WilmotE,YotebiengM,NorrisA,NgaboF.Missedopportunities inneonataldeathsinRwanda:applyingthethreedelaysmodel ina cross-sectionalanalysisofneonatal death.Matern Child HealthJ.2017;21:1121---9.

39.MengeshaHG, Wuneh AD,Lerebo WT, Tekle TH. Survival of neonates and predictors of their mortality in Tigray region NorthernEthiopia: prospective cohortstudy. BMC Pregnancy Childbirth.2016;16:202.

40.Ribeiro LB. Nascer em Belo Horizonte: Cesarianas desnecessárias e prematuridade [dissertation]. Universi- dade Federal de MinasGerais, Escola de Enfermagem: Belo Horizonte;2016.

41.SilvaCF,LeiteAJ,AlmeidaNM,LeonAC,OloﬁnI.Factorsasso- ciatedwithneonataldeathinhigh-riskinfants:amulticenter studyinhigh-riskneonatalunitsinNortheastBrazil.CadSaude Publica.2014;30:355---68.

42.LawnJE,BlencoweH,OzaS,YouD,LeeAC,WaiswaP,etal.

Everynewborn:progress,priorities,andpotentialbeyondsur- vival.Lancet.2014;384:189---205.

43.UnitedNationsInter-agencyGroupforChildMortalityEstima- tion.LevelsandTrendsinChildMortalityReport2018.UNICEF;

2018.