Maternal Mortality Estimates In Latin America And The Caribbean

9m ago
1.18 MB
8 Pages
Last View : 2m ago
Last Download : 7m ago
Upload by : Milena Petrie

Maternal Mortality Estimates inLatin America and the Caribbean:A Brief OverviewMarch 20121. IntroductionFigure 1. Maternal Deaths in Latin America in 2008Of the one thousand women aroundthe world who die every day fromcomplications of pregnancy andchildbirth, 99% are from developingcountries. Most of them are poorand have limited or no access tocomprehensive, skilled pregnancyand childbirth care. Almost all ofthese deaths are preventable. Bysigning the Millennium Declaration,in particular the fifth MillenniumDevelopment Goal (MDG 5: improvematernal health), countries committed themselves to preventing theseneedless deaths. Keeping this promise is a matter of basic human rights.EstimatedRegistered10 0009 0009,0758 0007,8647 0006 0005 0004 00085 deathsper 100,000live births3 0005,67074 deathsper 100,000live births2 0001 0000However, in order to target the most affectedpopulations with life-saving policies andprograms, governments need accurateinformation about maternal mortality intheir countries. Unfortunately, measuringmaternal mortality is complicated by the factthat nations with the least developed healthsystem infrastructure also tend to lack reliablemechanisms for identifying, registering, andcounting maternal deaths. National statisticson maternal mortality, therefore, vary considerably from maternal death estimates: Officialgovernment figures for 2008 showed a totalof 5,670 maternal deaths in Latin America,substantially fewer than the 9,075 estimatedby the United Nations Maternal MortalityEstimation Inter-agency Group (MMEIG)1 orthe 7,864 estimated by the Institute for HealthMetrics and Evaluation (IHME)2 for the sameperiod (see Figure 1).MMEIGIHME58 deathsper 100,000live birthsVital RegistrationIn this paper we will explain and compare thedifferent approaches to measuring maternalmortality, focusing on examples from LatinAmerica and the Caribbean (LAC).2. How is maternal mortalitymeasured?Maternal mortality is commonly measured asthe number of maternal deaths in a population divided by the number of live births(normally deaths per 100,000 live births).The ratio – called the maternal mortality ratio(MMR) – highlights the risk of maternal deathrelative to childbirth. MMEIG is comprised of four agencies: the World HealthOrganization, the United Nations Children’s Fund, the UnitedNations Population Fund and the World Bank. Their annualestimates make up the official data used by the United Nationsto measure progress towards MDG 5.2 IHME, based at the University of Washington in Seattle, releasesannual country estimates since

The most common source of maternal mortalitydata is the national vital registration system,tracked by ministries of health or nationalstatistical agencies.Box 1. WHO definitionThe Pan AmericanHealth Organizationof Maternal Death(PAHO) also usesThe death of a womanthese statistics forwhile pregnant or withincross-national42 days of termination ofcomparisons in itsa pregnancy, irrespectiveannual report on theof the duration and site ofHealth Situation inpregnancy, from any causethe Americas. Forrelated to or aggravatedthis reason, in thisby the pregnancy or itspaper we refer tomanagement but not fromstatistics publishedaccidental or incidentalby PAHO as “ statistics.”There are two issues of concern when usinginformation from vital registration systems.First, since some national vital registrationsystems lack full coverage and completeness,many deaths are not registered. This leads toundercounting. The second factor is misclassification of maternal deaths (see Box 1). Evenwhen vital registration systems are complete, alarge fraction of maternal deaths are incorrectly reported as non-maternal in the registrationsystem: even in many developed countries, asmany as one-third or even one-half of maternal deaths are misclassified.Some countries complement the information from vital registration systems with otherdata sources, such as household surveys (e.g.Demographic and Health Surveys – DHS), census data, the “Sisterhood Method” (estimates ofmaternal mortality based on information aboutsurvival of adult sisters) and Reproductive AgeMortality Studies (RAMOS). However, currentprocedures for country data reporting result inimportant inconsistencies in the types of statistics countries report to PAHO each year. Forexample, a country might report figures basedon vital statistics one year and a RAMOS-typestudy the next, or might adjust reported statisticsfor misclassification one year but not the next.PAHO does not routinely collect information on2the methods used by the ministries of health incalculating maternal mortality.3. Estimating the MMRIn seeking to develop accurate MMR estimates,social scientists make a series of statisticaladjustments to the “measured MMR,” asreported by national registration systems. Theseinclude adjusting for misclassification of maternal deaths, for under-registration of deathsof women aged 15 to 44, and for underregistration of births. As the latter two generallyoffset each other, the main difference between“measured MMR” and “estimated MMR” is dueto the misclassification adjustment factor, whichcan range from 1.1 (10% of maternal deathsmisclassified) to 2.0 (50% of maternal deathsmisclassified). Despite similarities, the methodologies employed by MMEIG and IHME – twointernational groups that have provided annualestimates for every country since 1990 – differin significant ways (see Table 1).The MMEIG methodologyThe MMEIG methodology divides countriesinto three groups according to the type ofinformation available (see Box 2). For Group Acountries, MMEIGestimates areBox 2. MMEIGbased on nationalclassification of countriesvital registrationGroup A: Countries withdata. Estimation ofgood vital registration dataMMR for Group B(half of the LAC countries).countries is a twostep process. First,Group B: Countries withthe proportion ofother types of data sourcesfemale deaths that(half of the LAC countries).are maternal isGroup C: Countries with noestimated throughadequate data sources (noa model in whichLAC countries).three predictingfactors are usedas a measure of risk exposure: GDP per capitaas a measure of economic development;proportion of live births attended by a skilledbirth assistant as a measure of health care;and the general fertility rate (live births per

Table 1. Comparison of MMEIG and IHME maternal mortality estimation methodologiesMMEIGIHMEData source for births andpopulationUnited Nations PopulationDivisionUnited Nations PopulationDivisionData source for mortalityrates among reproductiveage womenWHO Life TablesIHME Life TablesData source for proportionmaternal deathsMainly vital registration data.Some other national sources.Mainly vital registration data.Some other national and subnational sources.Misclassificationadjustment for vitalregistration dataDeaths increased by a factorof 1.5 for most countries.Usually held constant overestimation period (1990-2008).Deaths from other sources areincreased by a factor of 1.1.Deaths increased using analgorithm that redistributesdeaths from causes (“garbagecodes”) assumed to erroneouslycontain maternal deaths. Variesby country and over time. Onaverage, the adjustment factor is1.4 – but large variation aroundthis value.Regression approachA linear regression model isused only for countries thatlack good vital registrationsystem (10 of 20 Latin Americancountries).A linear regression model isused for all countries.Dependent variable inregressionLog of the proportion of deathsthat are maternal amongreproductive age women.Log of the age-specificmaternal death rate.Main predictor variablesGDP per capitaSkilled attendants at birthGeneral fertility rateGDP per capitaFemale education by ageNeonatal mortality rateTotal fertility rateHIV seroprevalence rateCountry and regionaleffectsModeled using variable interceptterms for country and region.Modeled using a temporalspatial locally weightedregression in which observationsfrom other time periods andneighbor countries influence thecountry-specific estimate.Treatment of AIDSMaternal deaths attributed toAIDS are estimated separatelyfrom the regression model.Maternal deaths attributedto AIDS are calculated withinthe regression model usingHIV seroprevalence rate as apredictor variable.3

women aged 15-49). Second, this proportionis applied to United Nations (UN) estimatesof the total number of deaths of reproductiveage women divided by the UN estimates of thetotal number of births. For most countries, thedata entered in the model are adjusted by 1.5,a median value to correct for misclassification.As the model does not include AIDS-relateddeaths, an independent estimate of indirectmaternal deaths from AIDS is added to theratio. MMEIG provides free and open access toits research data. The results are reproducibleand can be independently confirmed.The IHME methodologyThe IHME model is applied to all countriesregardless of the quality of their vital registration systems. The predictor variables of thismodel are different: GDP per capita, femaleeducation by age, neonatal mortality rate,total fertility rate and HIV seroprevalence rate.Through the last factor, maternal AIDS deathsare modeled directly in the regression. Anotherimportant difference with the MMEIG model isthe approach to adjusting for misclassification.The IHME reassigns deaths as maternal fromcauses in the International Classification ofDiseases (ICD) assumed to erroneously contain maternal deaths (“garbage codes”). Theseadjustments vary by country and over timewithin a given country. The average value ofthe adjustment factor is 1.4, a bit lower thanthe 1.5 used by MMEIG. The IHME group hasnot released its data and methods, so it is notpossible to independently confirm or reproduce its results.4. Comparing levels andtrends in maternal mortalityin LAC countriesFrom the description of the methodologicalapproaches to measuring or estimating MMR,we can compare the levels and trends in thevarious figures for countries in LAC.3 Figure 2shows data for each country over the period1990-2008. The black line in each figurerepresents the MMEIG point estimates whilethe grey area represents a range of uncertainty4surrounding the estimates. Similarly, the redline represents the IHME point estimates, whilethe pink area around it represents the range ofuncertainty around the estimates. The blue linerepresents the national statistics reported by theministries of health to PAHO.Comparison of MMEIG and IHME MMRsTwo of the most probable sources of differencein the MMR estimates between the groups are1) their treatment of AIDS-related deaths and2) their adjustment of vital registration datafor misclassification. The two groups differsharply, for example, in their estimate of theMMR for Haiti – the country with the highestlevel of AIDS-related mortality in the region.We suspect that this difference is caused by thedifferent treatment of AIDS within the model.For most of the other countries, a major causeof differences in estimates lies in the different adjustments to vital registration data formisclassification (MMEIG adjusts the maternaldeath estimates by 50% for most countries,whereas the IHME group, on average, addsonly 40% more deaths).For some countries, like Bolivia and Brazil,there is fairly close agreement between theMMEIG and IHME point estimates in both leveland trend over time. In other countries, likeEcuador and El Salvador, the MMEIG pointestimates exceed those of IHME, but bothshow similar declines in the ratios over time. Incountries like Mexico, both the level and trendin estimates are different.Comparison of national statistics withMMEIG and IHME MMRsIn countries such as Argentina, Chile, andNicaragua, the national statistics (blue lines inFigure 2) as reported by PAHO lie below boththe IHME and MMEIG estimates. In other countries like Brazil, Guatemala, and the DominicanRepublic, the country data lie above both theIHME and MMEIG estimates. And in some3 This analysis takes into account information available in September 2011. However, both IHME and MMEIG have releasednew estimates in 2012. These new estimates will be taken intoaccount when we do further analysis and conduct the casestudies recommended as part of this report.

Figure 2. MMR national statistics and estimates for selected LAC countries, 1990-20085

cases, like Mexico, Colombia, and Venezuela,national statistics lie in between: with MMEIGestimates below and IHME estimates abovethem. Understanding these differences wouldrequire analysis on a country-by-country basis,using information on the individual components (proportion of maternal deaths, totaldeaths among women in reproductive age andnumber of births) used in the calculations bythe ministries of health and IHME.Progress towards MDG 5 and MMEIGand IHME estimatesAlthough IHME estimates are generally lowerthan those from MMEIG, MMR trends tend tobe similar. So, to a large extent, the measurement of progress toward MDG 5 does notvary much between the two methods. Bothshow that countries like Brazil, Chile, Peru,and Bolivia have made great strides in reducing maternal mortality – by more than 50%since 1990; whereas countries like Mexico,Argentina, and Costa Rica have made lessprogress – achieving reductions of less than30% since 1990. However, there are somenotable exceptions to the general concordancebetween the sets of estimates. For example,in the case of Nicaragua, MMEIG estimatesshow substantial progress in reducing maternalmortality – with a decline of nearly 50% – butIHME estimates show no progress at all. Whilethe two groups are in close agreement in theirestimates of the MMR in 2008 (of around 100deaths per 100,000 live births), they differsubstantially in the estimates for 1990.5. Conclusions andrecommendationsAccurate measurement of maternal mortalityis difficult due to problems in data measurement. Complex models such as those used byMMEIG and IHME are designed to provide thebest possible MMR estimates for a large groupof countries. Of necessity, they are based ona number of assumptions made because oflimited information. Therefore, the resultingestimates have a large range of uncertainty.6Below are a number of recommendationsregarding the interpretation of the multipledata sources available for MMR:4 To understand the causes of differencesbetween estimates, one needs to look at inputand intermediate data used in the calculations. Thus, GTR strongly recommends thatdata are made available on a country-bycountry basis. For IHME, this means publiclyreleasing the country-specific data on whichtheir estimates are based. For PAHO, itmeans collecting the necessary metadata onthe methods and the data sources used byeach country in calculating the MMR. Latin American and Caribbean DemographicCentre (CELADE) and PAHO could play animportant role in sharing knowledge aboutdata sources as well as differences in methods and calculations among countries in theregion. A concrete step would be to undertake a case study for two or three countriesto explain the sources of differences betweenMMEIG, IHME, and national statistics.Possible countries could be Guatemala andEl Salvador (as RAMOS studies have recentlybeen completed there) and Mexico (becausenational statistics were closer to MMEIGestimates in the past but are closer to IHMEestimates in the present). Taking actions ofthis sort will promote reconciliation amongthe figures – which will be critical to measureachievements towards the MDGs in 2015. Both MMEIG and IHME have each independently assembled a data set of observationson maternal mortality. It would be very usefulto combine these data sources and presentthem online as a central repository of maternal mortality data. MMR should be used with care – especiallywhen the overall counts are low, as will usually be the case among countries with smallpopulations. The plausibility of MMR shouldbe assessed, by comparing it to other indicators such as infant and child mortality, fertility,education, and access to health care.

We must not forget that national data hidemajor differences within a country, both acrossregions and social groups. In order to redressinequalities and truly make progress towardsMDG 5, it is important to develop policies onthe basis of the needs of specific populations.In LAC, it is a priority to develop and strengthen national registration systems so that theycan accurately measure and track maternalmortality, including among vulnerable groups.Until accurate registration systems are in placein every country, it is important to rememberthat models may give the illusion that maternalmortality is actually being measured, when it isnot. Estimates can be a useful guide, but theydo not replace reliable national vital registration and surveillance systems.Even with their limitations due to underestimation and misregistration of maternal deaths,national registration systems are still amongthe best sources for understanding the causesof maternal mortality in each country, and theyshould be utilized. To obtain more consistentand reliable information, the way forwardin the LAC region is to strengthen both thevital registration systems and the institutionalregistration systems in clinics and hospitals.This would permit a cross-validation betweenthe two systems. Significant resources must beallocated to developing national capacities interms of measuring maternal mortality andstrengthening basic data collection systems,and to supporting studies that go beyond ratesand ratios to understand the causes of maternal mortality and the impact of interventions.Without such efforts, we are left with a paucityof data and extreme difficulty in measuringmaternal mortality. As the statistician John Tukeyonce said: “The combination of some data andan aching desire for an answer does not ensurethat a reasonable answer can be extracted froma given body of data.” We must do better.4 These recommendations were adapted from Stupp et al.,2011 and from AbouZahr, 2011. In addition, CELADE addedsome region-specific considerations.7

ReferencesAbouZahr, Carla. (2010). Making sense of maternalmortality estimates. University of Queensland Schoolof Population Health. Health Information SystemsKnowledge Hub Working Paper Series, Number 11,November 2010.Hogan, Margaret C, Kyle J. Foreman, Mohsen Naghavi,Stephanie Y Ahn,Mengru Wang, Susanna M. Makela,Alan D. Lopez, Rafael Lozano, and Christopher J.Murray. (2010). “Maternal Mortality for 181 countries,1980-2008: a systematic analysis of progress towardsMillenium Development Goal 5.” The Lancet, Vol 375,May 8, 2010.Naghavi, Mohsen, Susanna Makela, Kyle Foreman,Janaki O’Brien, Farshard Pourmalek, and Rafael Lozano(2010), “Algorithms for enhancing public health utility ofnational cause-of-death data.” Public Health Metrics, Vol8, No. 9, 2010.Stupp, Paul, Florina Serbanescu, Dan Williams, andStephen McCracken. (2011). Comparison of recentmaternal mortality estimates with national maternalmortality surveillance data: Findings from three countries.Presentation at Global Health Metrics and EvaluationConference, March 14-16, 2011.Wilmoth, John, Sarah Zureick, Nobuko Mizoguchi, MieInoue, and Mikkel Oestergaard (2010). Levels and trendson maternal mortality in the world: The development ofnew estimates by the United Nations. Technical reportsubmitted to WHO, UNICEF, UNFPA, and the WorldBank. September 2010.Wilmoth, John (2011). New Estimates of MaternalMortality: How much can we believe them? Presentationat Global Health Metrics and Evaluation Conference,March 14-16, 2011.World Health Organization (2010), Trends in MaternalMortality: 1990 to 2008: estimates developed by WHO,UNICEF, UNFPA, and the World Bank. Geneva: WorldHealth Organization.Data SourcesData for the UN Inter-agency Group estimates were takenfrom “Trends in maternal mortality, 1990-2008, Data andprogrammes used for computing the maternal mortalityestimates, accessed August 2011 at health/mortality/maternal/en/index.html.Data for the IHME estimates were taken from “MaternalMortality for 181 countries, 1980-2008”, accessedAugust 2011 1980-2008-systematic-analysis-progress.Data for the country statistics as reported to PAHOwere taken from “PAHO Country Health IndicatorProfile – Infant and Maternal Mortality” accessedSeptember 2011 at: ortality.asp.AcknowledgementsThis document was prepared by the Latin American andCaribbean Demographic Centre - Population Divisionof the Economic Commission for Latin America and theCaribbean jointly with the Latin American Centre forPerinatology, Women’s and Reproductive Health (CLAPWRH) of the Pan American Health Organization – WorldHealth Organization (PAHO-WHO) at the request of GTRwith a view to harmonize estimates of maternal mortalityin the region. It was coordinated by Magda Ruiz, RegionalDemography and Population Information Adviser and wasdeveloped by Tim Miller of the same Division. Mariekevan Dijk, independent consultant, revised and synthesizedthe document. We appreciate the technical contributionsof Dr. Alma Virginia Camacho, United Nations PopulationFund, Dr. Peg Marshall, United States Agency forInternational Development, Dr. Isabella Danel, Centersfor Disease Control and Prevention, Dr. Bremen de Mucio,CLAP-WRH/PAHO-WHO, Dr. Ricardo Fescina, Directorof CLAP-WRH/PAHO-WHO, and Dr. Douglas Jarquin,Latin American Federation of Societies of Obstetricsand Gynecology, among others. The English version ofthis document was edited by Adam Deixel and AriadnaCapasso and designed by Virginia Taddoni, of FamilyCare International.Production of this document was possible thanks to agrant from the Latin American Centre for Perinatology,Women’s and Reproductive Health of the Pan AmericanHealth Organization – World Health Organizationthrough a grant from the United States Agency forInternational GTR, 2012.

counting maternal deaths. National statistics on maternal mortality, therefore, vary consid-erably from maternal death estimates: Official government figures for 2008 showed a total of 5,670 maternal deaths in Latin America, substantially fewer than the 9,075 estimated by the United Nations Maternal Mortality

Related Documents:

situation of maternal mortality is still worrying. Sixteen out of the forty-eight countries in the region have very high maternal mortality ranging between 500 to 999 maternal deaths per 100,000 live births in 2017 (WHO et al.2019). The current estimates on maternal mortality show that the maternal mortality ratio, an

South Asia have the highest maternal mortality, contrib-uting up to 86% maternal mortality globally [1]. Tanzania is among the sub-Saharan countries with the highest maternal mortality. The maternal mortality ratio in Tanzania is as high as 556 maternal death in every 100,000 live births [2]. It has been reported that the most leading causes of

Overview of Maternal Mortality in Kentucky and Strategies for Change 2020 KHA Quality Conference Connie Gayle White, MD, MS, FACOG. Deputy Commissioner. Maternal Mortality 2018 - CDC. 2. Maternal Mortality Rate, United States. . reduce maternal morbidity. .

Latin Primer 1: Teacher's Edition Latin Primer 1: Flashcard Set Latin Primer 1: Audio Guide CD Latin Primer: Book 2, Martha Wilson (coming soon) Latin Primer 2: Student Edition Latin Primer 2: Teacher's Edition Latin Primer 2: Flashcard Set Latin Primer 2: Audio Guide CD Latin Primer: Book 3, Martha Wilson (coming soon) Latin Primer 3 .

9 OVERVIEW OF THE SITUATION OF MATERNAL . MORBIDITY AND MORTALITY: LATIN AMERICA AND THE CARIBBEAN 9ntroduction I 11he Context of Maternal Morbidity T . 22onclusions C 25 ADDITIONAL REFERENCES 25nnex A: Glossary A 27nnex B: The International Context for A the Reduction of Maternal Mortality and . Morbidity 30nnex C: Regional Trends in .

current system should continue to be monitored and evaluated through the Maternal . which includes an overview of the maternal mortality situation worldwide and in Botswana. It also briefly describes the demographic, socio-economic and health . maternal mortality ratios (MMR) ranged from 2 maternal deaths per 100,000 live

Maternal mortality reduction is one of the MDG that Ghana seeks to achieve since it affects the development of the nation . The [9] objective of the MDG 5 is to improve maternal health, and to minimize maternal mortality (Target 6) ratio by 75% by 2015. Subjecting women to poor maternal health situation is also considered as a violation of .

using artificial intelligence (AI)—and machine learning in particular—increasingly match or surpass human-level performance; news about the rapid pace of technological advancement abounds, and market capitalizations for technology firms are at all-time highs. Yet, measured productivity growth in the United States has declined by half over the past decade, and real income has stagnated .