maternal mortality at the end of a decade: signs of progress?

Maternal mortality at the end of a decade:signs of progress?Carla AbouZahr1 & Tessa Wardlaw2

Abstract Maternal mortality is an important measure of women’s health and indicative of the performance ofhealth care systems. Several international conferences, most recently the Millennium Summit in 2000, haveincluded the goal of reducing maternal mortality. However, monitoring progress towards the goal has proved to beproblematic because maternal mortality is difficult to measure, especially in developing countries with weak healthinformation and vital registration systems. This has led to interest in using alternative indicators for monitoringprogress. This article examines recent trends in two indicators associated with maternal mortality: the percentage ofbirths assisted by a skilled health care worker and rates of caesarean delivery. Globally, modest improvements incoverage of skilled care at delivery have occurred, with an average annual increase of 1.7% over the period 1989–99. Progress has been greatest in Asia, the Middle East and North Africa, with annual increases of over 2%. In sub-Saharan Africa, on the other hand, coverage has stagnated. In general, caesarean delivery rates were stable overthe 1990s. Countries where rates of caesarean deliveries were the lowest — and where the needs were greatest —showed the least change. This analysis leads us to conclude that whereas there may be grounds for optimismregarding trends in maternal mortality in parts of North Africa, Latin America, Asia, and the Middle East, thesituation in sub-Saharan Africa remains disquieting.

Keywords Maternal mortality/trends; Process assessment (Health care)/trends; Midwifery/statistics; Cesareansection/statistics; Vital statistics; Registries (source MeSH ).

Mots cles Mortalite maternelle/orientations; Evaluation methodes sante/orientations; Profession sage-femme/statistique; Cesarienne/statistique; Statistique demographique; Registre (source: INSERM).

Palabras clave Mortalidad materna/tendencias; Evaluacion de proceso (Atencion de salud)/tendencias;Tocologıa/estadıstica; Cesarea/estadıstica; Estadısticas vitales; Sistema de registros (fuente: BIREME ).

Bulletin of the World Health Organization, 2001, 79: 561–568.

Voir page 567 le resume en francais. En la pagina 568 figura un resumen en espanol.

Introduction

During the 1990s, a number of international confer-ences set goals for a reduction of maternal mortality. In1999, during appraisal of the implementation of theCairo Programme of Action, a reduction in maternalmortality was reiterated to be a high priority andcountries agreed to strengthen information systemsfurther to permit regular monitoring of maternalmortality (1). The reduction in maternal mortality wasadopted as an International Development Goal by theUnited Nations (UN), the Organisation for EconomicCooperation and Development, the InternationalMonetary Fund, and the World Bank (2, 3) andendorsed by 149 heads of state at the MillenniumSummit in 2000 (4). This unprecedented global

consensus creates additional stimulus for accuratemonitoring of progress in the attainment of this goal inindividual countries and across the world.

Goal setting presupposes the ability to monitorprogress; unfortunately, in the case of maternalmortality, this has proved to be a greater challengethan might have been anticipated. With the 1990snow behind us, and with several years of country dataupon which to draw, this is an opportune moment toassess what progress has been made. In this paper, wepresent what is currently known about trends inmaternal mortality in developing countries. We alsopresent data on trends in two indicators related tomaternal mortality: the percentage of pregnantwomen who have the help of a skilled attendantduring labour and childbirth and the percentage ofdeliveries carried out by caesarean section.

The maternal mortality ratio: aproblematic indicator to measure

Ascertaining progress in the efforts to reduce maternalmortality is problematic because maternal mortality is

1 Coordinator, Policy, Planning, and Evaluation, Office of the ExecutiveDirector, Family and Reproductive Health, World Health Organization,1211 Geneva 27, Switzerland. Correspondence should be addressedto this author.2 Senior Project Officer, Statistics and Monitoring, UNICEF, New York,USA.

Ref. No. 00-0498

561Bulletin of the World Health Organization, 2001, 79 (6) # World Health Organization 2001

difficult to measure (5). The definition in the tenthrevision of the International statistical classification of

diseases and related health problems includes deaths due toboth direct obstetric causes and to conditionsaggravated by pregnancy or delivery (6). In countrieswith statistically developed systems, the conventionalsource of information about maternal mortality is thecivil registration system, which records deaths bycause as well as recording live births on a continuousbasis. Even in such settings, however, maternal deathsare invariably found to be underrecorded in officialstatistics as a result of misclassification of cause (7–11).In countries that are less statistically developed, errorsin both numbers of deaths and attribution of causemay result in biased measures of maternal mortalityderived from vital registration.

Vital registration systems in most developingcountries are absent or inadequate and it is necessaryto estimate maternal mortality through householdsurveys. But even where overall levels of maternalmortality are high, maternal deaths are relatively rareevents, limiting the applicability of sample surveymethods. The most commonly used measure, thematernal mortality ratio (MMR), expresses maternaldeaths per 100 000 live births, yet MMRs rarelyexceed 1000, or one per 100 live births. Even withlarge sample sizes, the standard errors of the MMRwill inevitably be large. Large-scale surveys areextremely expensive and thus not feasible whereresources are limited (10–12). The sisterhoodmethod is a survey-based measurement techniquethat substantially reduces sample size requirementsbecause it obtains information by interviewingrespondents about the survival of all their adultsisters. Although sample size requirements arereduced, the problem of wide confidence intervalsremains and therefore the technique is not appro-priate for monitoring. Furthermore, the methodprovides a retrospective rather than a currentestimate (13). Because of such problems, the use ofthe ratio as the sole indicator of maternal mortality isno longer considered appropriate in settings wheresurvey techniques have to be used.

Measuring maternal mortality usingcivil registration

Where data on levels of maternal mortality areavailable on a regular basis from routine vitalstatistics, they can be used for monitoring trends.In practice, however, only a small number ofcountries, accounting for under a quarter of theworld’s births (and less than 7% of births if China isexcluded), have data on trends in maternal mortalityderived from the civil registration system. All of thesecountries have relatively low levels of maternalmortality, at under 100 per 100 000 live births sincethe early 1980s. Furthermore, there are a number ofissues that must be borne in mind when using thesedata. Although ratios derived from routine vitalstatistics are not subject to huge margins in standard

errors in the same way as those derived using surveymethodologies, all the evidence indicates that there issignificant underreporting and misclassification ofmaternal deaths. The latter is of particular signifi-cance, as registration systems are known to failroutinely to identify correctly a proportion ofmaternal deaths (14). The deaths most oftenmisclassified in routine reporting include thoseassociated with abortion (especially where it is illegal);early pregnancy deaths (resulting from ectopic ormolar pregnancy), where the pregnancy may havebeen unknown to the woman or her family; indirectmaternal deaths (e.g. due to malaria, anaemia,tuberculosis, hepatitis, or cardiovascular disease);and deaths that occur some time after the end ofpregnancy, especially where the death occurs in anon-obstetric hospital ward, for example, in anintensive care or other specialized unit (9, 15–17).

Trends in Asia and Latin America

Annex Fig. A and Annex Fig. B (available on our website: http://www.who.int/bulletin) show maternalmortality trends derived from civil registration data inselected developing countries in Asia and LatinAmerica, respectively (countries classified by the UNStatistics Offices as having complete registration, i.e.90% or more of adult deaths recorded). Annex Fig. Aalso includes trend data for China, where Reproduc-tive Age Mortality Study (RAMOS) data have beenused to estimate maternal mortality since 1989.

Malaysia, Sri Lanka, and Thailand also havedata on maternal mortality derived from routinereporting coupled, in recent years, with variousmethods of triangulation (i.e. using different infor-mation sources to cross-check for completeness) inorder to reduce misclassification of maternal deathsand improve the completeness of reporting. We havenot included data for these three countries becauseimproved case ascertainment in the more recent datarenders any trend analysis problematic. Apparentincreases in levels of maternal mortality during the1990s, notably in Malaysia and Thailand, are thoughtto reflect improved data collection as a result of thetriangulation rather than a real increase in the rates.Trends from vital registration data are not shown forMauritius, the only African country classified by theUN as having complete registration, because theactual numbers are very small.

For the countries included in this analysis, wehave assumed that the overall proportion of maternaldeaths missed in routine reporting has remainedrelatively stable, given the absence of efforts toimprove case finding. However, there is no evidenceto support this assumption and it is conceivable thatthe quality of vital registration has either improved ordeteriorated over the years analysed. These trend datacannot be assumed to capture fully the true levels ofmaternal mortality because of the problems ofunderreporting and misclassification. In Argentinaand Mexico, for example, small-scale studies in major

562 Bulletin of the World Health Organization, 2001, 79 (6)

Policy and Practice

urban areas have concluded that officially reportedlevels of maternal mortality may be underreported byas much as 50% (18).

Bearing all these caveats in mind, significantreductions in registered maternal mortality haveoccurred in the countries analysed. However, most ofthe decline took place prior to 1990, the baseline datefrom which progress is measured here. OnlyArgentina, Chile, China, Costa Rica, and Uzbekistanwere able to demonstrate sustained reductions inmaternal mortality over the 1990s. Elsewhere, thereappears to have been a relative stagnation in maternalmortality ratios since 1990. It is important to stressthat Argentina, Chile, China, Costa Rica, andUzbekistan cannot be considered representative oftrends in developing countries as a whole. All havelevels of maternal mortality that are low fordeveloping countries, at less than 100 maternaldeaths per 100 000 live births.

Trends in Eastern Europe

Annex Fig. C (available on our web site: http://www.who.int/bulletin) shows trends in maternalmortality in some countries of Eastern Europe whichare thought to have relatively efficient systems of vitalregistration. The apparent transient increases in somecountries (e.g. Latvia) may be the result of randomfluctuations due to the small numbers involved ormay be due to improved case reporting. In Romania,the dramatic fall in maternal mortality observed in1989–90 reflects the liberalization of the lawregarding availability of safe abortion. Prior to 1989,strongly pronatalistic policies, lack of reliable contra-ception, prohibition of abortion, and economicdifficulties for ordinary people combined to resultin extremely high levels of abortion-related mortality(19). Although the situation appears to havestabilized somewhat since 1990, there are markeddifferences between levels of maternal mortality inthese countries of Eastern Europe and levels incountries of the European Union (data not shown).

Process indicators

In countries where maternal mortality is measuredusing household surveys, the margins of uncertaintyare such that it is not possible to draw firmconclusions about the directions of trends. Forassessing progress in these countries, processindicators are needed for regular monitoring. Goodprocess indicators are closely correlated with theoutcome of interest, but have a number ofadvantages over outcome indicators: they are gen-erally easier and cheaper to collect and are moresensitive to change, and are thus very useful forregular and short-term monitoring. Experience hasshown that reducing maternal mortality dependscritically on the availability and use of obstetric carefor managing complications. Unfortunately, data on

this remain extremely limited at present. We aretherefore proposing two other indicators for whichdata are more widely available, namely the percentageof deliveries assisted by a skilled attendant and thosecarried out by caesarean section.

Assistance of a skilled attendant at deliveryThe percentage of births assisted by a skilledattendant is one potential process indicator and thereis evidence of its strong association with levels ofmaternal mortality (20, 21). There is a clear clinicaljustification for the presence of a skilled attendant atdelivery, as this may reduce both the incidence ofcomplications and case fatality (22). However,correlation does not necessarily imply causation.Unequivocal epidemiological evidence for the impactof skilled attendants at delivery on reducing maternalmortality is lacking. The evidence in favour of a causallink is, for the moment, largely circumstantial.Moreover, the apparent simplicity of the indicatortends to obscure the complex reality of care duringchildbirth. There are large differences in the skills,equipment, and supplies needed for the appropriatecare and management of normal compared withcomplicated deliveries. Graham, Bell, & Bulloughhave shown that the correlation between maternalmortality reduction and assistance of a skilledattendant appears to be stronger when the attendantsare doctors rather than nurses or midwives (23), afinding that could be a reflection of better access toessential obstetric care services. Although this issue istoo complex to address here, it is important to bear inmind and caution is needed in interpreting the data inthis paper.

Despite these concerns, the skilled-attendantindicator has a number of advantages for monitoringpurposes, including the fact that data are widelyavailable for developing countries up to and includingthe year 2000. Currently, the source of the informa-tion is generally the Demographic and HealthSurveys (DHS), PAPCHILD, or ReproductiveHealth Surveys. These use a standardized methodol-ogy and sampling framework and have strict criteriaregarding the maintenance of data quality, thusenhancing the comparability of data between coun-tries and over time. The DHS, PAPCHILD, andReproductive Health Surveys subdivide birth atten-dants into doctors, nurses, midwives, or nurse-midwives, and trained traditional birth attendants.The latter initially acquire their ability either bydelivering babies themselves or through apprentice-ship to other traditional birth attendants, undergosubsequent extensive training, and are then inte-grated in the formal health care system. Theremaining category comprises relatives, untrainedtraditional birth attendants, and no assistance. Forthese calculations, only doctors and nurses, mid-wives, and nurse-midwives who have the necessarymidwifery skills are classified as skilled attendants.

Problems of definition. Like any crude mea-sure, however, this indicator may hide important

563Bulletin of the World Health Organization, 2001, 79 (6)

Trends in maternal mortality in the 1990s

differences (e.g. between regions or socioeconomicgroups). Moreover, while efforts are being made tostandardize definitions of skilled attendants, there isdoubt about the comparability of some of the resultsbetween countries and within a country at differenttime periods. One source of potential confusion is thediffering interpretations of ‘‘skilled attendant’’. Theprecise competencies of health care providers withvarious titles can be clarified only at the local level. Thisissue is particularly complex in Latin America wherethere appear to be discrepancies in the definition of thehealth workers classed as skilled attendants (e.g.parteras, comadronas, and dais). Moreover, the definitionshave, in some cases, changed from one DHS toanother, particularly in the definition of parteras. Inorder to avoid these problems, for countries in LatinAmerica and the Caribbean, we have used theproportion of births taking place in a health carefacility, for which the definitions are consistent fromone survey to the next, rather than the less clearlydefined term skilled attendant. This should not betaken to imply any recommendation as to the place ofdelivery; the decision is based on statistical considera-tions alone. For all other countries, we have followedthe survey definition but have only included in ouranalysis only attendants described as doctors, nurses,midwives or nurse-midwives.

Apart from problems of definitions, there areother issues related to methods of data collection onskilled attendants at delivery. Concerns include theextent to which respondents can accurately report thelevels of skills of the attendant and potential biasintroduced by the fact that most household surveysreport on live births, thus missing many adversehealth outcomes, such as still births or miscarriages.

Despite such limitations, the indicator providessome useful information for comparing countries andstudying recent trends. Annex Table A (available onour web site: http://www.who.int/bulletin) showsdata available in January 2001 on assistance of skilledattendants at delivery for developing countries overthe period 1989–99. Current estimates indicate that

globally, skilled attendants assist only around 56% ofbirths (WHO and UNICEF databases). The lowestlevels in developing countries are in South Asia (29%)and sub-Saharan Africa (37%). The highest levels arein Latin America and the Caribbean (83%) and theCentral and Eastern Europe/Commonwealth ofIndependent States regions (94%).

Trend data are not available for all developingcountries. The trend information presented hererepresents only the subset of developing countriesthat have a minimum of two data points derived fromsources using similar estimation methods, generallyDHS surveys. Overall, these countries account for76% of live births in developing countries, althoughthis figure varies considerably by region. Given theimprecision of the data collection methodologiesused to generate this indicator, it would beinappropriate to read too much into the apparentchanges in coverage of care. Table 1 shows the annualrate of change in this indicator for major regionalgroupings. Because data are available for differentyears and cover a different time period for eachcountry, we have adjusted trends to a common 10-year period: 1989–99. The observed rate of changewas used to project data for the end points in 1989and 1999. The regional averages are weighted by thenumbers of live births.

Trends in coverage of care (assistance of

skilled attendant) at delivery. The evidence showsthat, in general, only modest improvements incoverage of care at delivery have occurred, with anaverage annual increase of 1.7% over the period1989–99. In sub-Saharan Africa, there was no overallchange over the period, despite improvements insome countries, notably Togo and Niger. On theother hand, countries of Asia, the Middle East, andNorth Africa show significant improvements, withaverage annual increases of 2.2% (Asia) and 2.5%(Middle East and North Africa).

Figs 1–3 show the evolution of skilledattendant data for selected developing countriesfrom the early 1980s until the late 1990s. For Latin

Table 1. Trends in percentage of deliveries assisted by skilled attendants for 53 countries, 1989–99(data available up to April 2001)

Region No. of countries % of births % of deliveries assisted Averagewith trend in the region by skilled attendants annual rate

dataa covered by of changeb

the data 1989 1999 1989–99

Sub-Saharan Africa 17 59 44 44 0.1Middle East and North Africa 9 56 49 63 2.5Asia 7 89 39 48 2.2Latin America and the Caribbean 18 74 74 81 0.9

Total 53c 76d 45c 52c 1.7c

a Data published up to April 2001.b Weighted average of individual country data. Regional averages were weighted by the numbers of live births.c Includes two countries from Central and Eastern Europe and the Commonwealth of Independent States.d Data for developing countries only.


Policy and Practice

America and the Caribbean, as explained earlier, thetrends in the percentage of births occurring in healthcare facilities is shown (Fig. 4). In Brazil, Jamaica,Jordan, Kuwait, Oman, Panama, and South Africa,there are relatively high levels of coverage withmodest but positive growth. In contrast, in Bangla-desh, Burkina Faso, Haiti, and Mali very low levels ofcoverage coincide with little change or even a decline.Niger and Yemen, though starting from a lowbaseline, have shown some improvements in cover-age. Significant improvements can be observed inBolivia, Egypt, Indonesia, Morocco, and Togo —countries characterized in recent years by a deter-mined and high-level commitment to addressmaternal mortality. In Egypt, for example, coverageat delivery increased from 35% to 61% over theperiod 1988–99. In Indonesia, the percentage ofdeliveries assisted by a skilled attendant increasedfrom 36% in 1997 to 56% in 1999. In Morocco,coverage increased from 24% in 1984 to 40% in1995. Significant increases in the percentage of birthsin health care facilities were observed in othercountries such as Argentina, Ecuador, Honduras,and Mexico.

Countries with already high rates of coverage atdelivery (e.g. Costa Rica, Cuba, Jordan, and Kuwait)appear to be making continuous progress. Countrieswith intermediate levels of coverage (e.g. Guinea,South Africa, Togo, Tunisia, and Zimbabwe) showongoing, slow improvements. On the other hand,other countries (e.g. Burkina Faso, Cameroon, Kenya,Madagascar, Mali, United Republic of Tanzania, andZambia) appear to have lost ground. Of the 17 sub-Saharan countries for which trend data are available,only six (Ghana, Guinea, Niger, Nigeria, Senegal, andTogo) have significantly increased levels of coveragesince 1988. While acknowledging the uncertaintiesinherent in the data, the apparent stagnation incoverage shown by these results should alert nationalauthorities to a potential problem.

Rate of caesarean deliveryAnother process indicator, which was originallyproposed for monitoring access to health care services,is the proportion of all deliveries carried out bycaesarean section (24). The rationale for using thisindicator as a proxy for maternal mortality is based onthe premise that in the absence of surgical interven-tions such as caesarean delivery or hysterectomy, manywomen with serious obstetric complications (ob-structed labour, eclampsia, or intractable haemorrhage)will die. It is important to be clear that increasing therate of caesarean deliveries is not the goal; rather, theindicator can be used as a proxy for the extent to whichhealth care facilities provide — and women are able togain access to — this essential element of obstetriccare. This indicator is controversial because caesareandelivery is sometimes overused or used for inappropri-ate indications, i.e. for reasons not related to preservingthe life and health of mother or infant. The procedurecan be convenient and lucrative for physicians but



carries risks for the woman, particularly whenconducted in less than optimal conditions. It alsoimposes additional costs for the woman and her family.UNICEF, WHO, and UNFPA guidelines recommendthat, as a general rule, a minimum of 5% of deliveriesare likely to require a caesarean section in order topreserve the life and health of mother or infant. Rateshigher than 15% indicate inappropriate use of theprocedure. It goes without saying that a study of therates by themselves tells us nothing about whether thewomen who really needed a caesarean delivery actuallyhad one. Such information can only be obtainedthrough a detailed study of the indications forcaesarean delivery in individual health care facilities.

The numbers of caesarean deliveries can beobtained from examination of health facility log-books, but the information is very rarely collated atthe national level. However, in recent years, DHSsurveys have included questions about mode ofdelivery for women who had a live birth in thepreceding five years. For a few countries, trend dataon caesarean delivery are now available (data fromDHS surveys, various years).

The relationship between rates of caesareandelivery and levels of maternal mortality is not simple.The maternal mortality estimates were developed byWHO and UNICEF using a range of predictorvariables, but not including caesarean delivery rates.As shown in Annex Fig. D (available on our web site:http://www.who.int/bulletin), very high levels ofmaternal mortality (over 500 maternal deaths per100 000 live births) are generally associated with lowlevels of caesarean delivery (5% or below). Allcountries with caesarean delivery rates below 2%have levels of maternal mortality above 500 per100 000 live births. However, low levels of maternalmortality coexist with a wide range of caesareandelivery rates. In several countries with maternalmortality ratios below 100 per 100 000 live births,caesarean delivery rates were around 3–7%. It ispossible that these low rates of caesarean delivery are

associated with high levels of normal vaginal delivery,but it is impossible to ascertain this from the availableDHS data.

In general, caesarean delivery rates were higherin countries of the Americas than in Africa and Asia.Particularly low levels of caesarean delivery — lessthan 5% of all deliveries — were found in Cameroon,Niger, Madagascar, and Zambia. Trends in caesareandelivery rates tended to be stable over the 1990s,although in Ecuador there appears to have been arather substantial decline (Annex Fig. E, available onour web site: http://www.who.int/bulletin). Incontrast, in Bolivia, the Dominican Republic,Guatemala, and Jordan, rates appear to haveincreased. For the other countries, the changes inlevels were very small. What is particularly note-worthy is that it is precisely in those countries whererates of caesarean deliveries were the lowest — andpresumably where the needs were greatest — thatthere was the least apparent change.

Conclusions

We have attempted to assess progress in thereduction of maternal mortality over the 1990s usingboth direct and proxy indicators. Where civilregistration exists, temporal trends can be analysedfor signs of progress, though care is required inevaluating the data because of the well-documentedproblems of incomplete registration and inaccuracyin cause-of-death attribution.

Disappointing rates of reduction in maternalmortalityOnly a few developing countries, accounting for 24%of live births worldwide, have complete vitalregistration, according to the UN. In these countries,while there has been a decline in maternal mortality,most of the decrease took place during the 1970s andearly 1980s. Since then, progress appears to haveslowed despite the increased interest in the issuegenerated around the Safe Motherhood Initiative,which started in 1987 (19).

Increase in the use of medical careat deliveries, but still not worldwideWe have also analysed two process indicators: thepercentage of births assisted by a skilled attendantand the rates of caesarean delivery. Taken together,they appear to show that while progress has beengenerally slow, most parts of the world havesucceeded in increasing the use of medical care atdeliveries. The exception is sub-Saharan Africa as awhole, where, despite progress in a few countries, ingeneral coverage has stagnated or, in severalcountries, actually declined.

Currently available data do not permit us toestablish clear and unequivocal links between trendsin the percentage of deliveries assisted by a skilledattendant and maternal mortality. Like any other


Policy and Practice

indicator, this represents a snapshot of a complexreality. Much more detailed information is needed,both with regard to how the indicator is measured,and also, perhaps more importantly, precisely what itmeans in different settings with regard to the level ofcare available to women with normal and compli-cated deliveries.

If the association between maternal mortalityand the two indicators holds, however, the datacurrently available suggest that there may be groundsfor optimism regarding trends in maternal mortalityin parts of North Africa, Latin America, Asia, and theMiddle East. In contrast, the situation in large parts ofsub-Saharan Africa is disquieting, with unchanged ordeclining levels of coverage of skilled care for womenduring the crucial period of childbirth. This is aparticular cause for concern in settings with highprevalence of HIV/AIDS, where the need for skilledcare during labour and childbirth is critical.

Implications for the futureIn 1999, at the Special Session of the UN GeneralAssembly to mark five years after the InternationalConference on Population and Development held inCairo, Egypt, it was agreed that all countries shouldstrive to ensure that 80% of deliveries are assisted by askilled attendant by 2005. Based on current trends,only countries of Latin America and the Caribbean willattain this goal (Fig. 5). Countries in the Middle Eastand North Africa will not reach the goal until around2010 and Asian countries as a whole will fall short ofthe goal even in 2015. In sub-Saharan Africa, there is nooverall sign of progress towards the 2015 goal.

If the validity of using process indicators tomonitor trends in maternal mortality is accepted, wemust conclude that while there have been modestimprovements in Asia, the Middle East, and NorthAfrica it is likely that levels of maternal mortality insub-Saharan African countries have remained un-changed or have even deteriorated. n

AcknowledgementsThe authors wish to thank Nyein Nyein Lwin andElisabeth Aahman for their assistance in compilingthese data.

Conflicts of interest: none declared.

Resume

La mortalite maternelle a la fin d’une decennie : des signes de progres ?La mortalite maternelle est une mesure importante de lasante des femmes et donne une indication de laperformance des systemes de sante. Plusieurs conferen-ces internationales, dont le Sommet du Millenaire en2000, ont inscrit a leur ordre du jour l’objectif de lareduction de la mortalite maternelle. Cependant, lasurveillance des progres realises s’avere problematiquecar la mortalite maternelle est difficile a mesurer,notamment dans les pays en developpement ou lessystemes d’information sanitaire et d’enregistrement desdonnees demographiques sont insuffisants. C’est pour-quoi l’on s’est interesse a l’utilisation d’indicateurs deremplacement. L’article examine les tendances recentesde deux indicateurs associes a la mortalite maternelle, lepourcentage d’accouchements realises par un agent desante qualifie et la proportion d’accouchements parcesarienne. A l’echelle mondiale, on note une augmen-

tation modeste de la proportion d’accouchementsrealises par un personnel qualifie, avec une augmenta-tion annuelle moyenne de 1,7 % pendant la periode1989-1999. Les progres les plus importants ont eteobserves en Asie, au Moyen-Orient et en Afrique duNord, avec une augmentation annuelle de plus de 2 %.En Afrique subsaharienne, par contre, la couverture n’aguere evolue. En general, la proportion d’accouchementspar cesarienne est restee stable au cours des annees 90.Les pays dans lesquels les taux de cesariennes etaient lesplus faibles – et ou les besoins etaient les plus grands –sont ceux ou la situation a le moins change. Cetteanalyse nous amene a conclure que si les tendances de lamortalite maternelle autorisent un certain optimismedans certaines parties de l’Afrique du Nord, del’Amerique latine, de l’Asie et du Moyen-Orient, lasituation en Afrique subsaharienne reste preoccupante.



Resumen

La mortalidad materna al termino de una decada: ¿hay indicios de progreso?La mortalidad materna es un indicador de la salud de lamujer y una variable indicativa del desempeno de lossistemas de atencion de salud. Varias conferenciasinternacionales, la mas reciente de las cuales es laCumbre del Milenio celebrada en 2000, han incluidoentre sus objetivos la reduccion de la mortalidadmaterna. Sin embargo, el seguimiento de los progresoshacia esa meta ha sido problematico, pues la mortalidadmaterna es difıcil de medir, especialmente en los paısesen desarrollo, caracterizados por la precariedad de sussistemas de informacion sanitaria y de registro civil. Estoha despertado el interes por el uso de indicadoresalternativos para vigilar los progresos. En este artıculo seexaminan las tendencias recientes en lo que atane a dosindicadores asociados a la mortalidad materna: elporcentaje de partos atendidos por un agente de saluddebidamente capacitado para ello, y las tasas de parto

por cesarea. A nivel mundial se ha registrado una levemejora de la cobertura de atencion cualificada al parto,con un aumento promedio del 1,7% anual durante elperiodo 1989–1999. Los progresos han sido mayores enAsia, Oriente Medio y Africa septentrional, con aumentosanuales de mas del 2%. En el Africa subsahariana, encambio, la cobertura se ha estancado. En general, lastasas de parto por cesarea se mantuvieron establesdurante los anos noventa. Los paıses con las tasas masbajas de parto por cesarea — y con las mayoresnecesidades – fueron los que registraron la menorvariacion. Este analisis nos lleva a concluir que, aunquequiza hay motivos para ser optimistas respecto a lastendencias de la mortalidad materna en zonas del nortede Africa y en America Latina, Asia y Oriente Medio, lasituacion en el Africa subsahariana sigue siendopreocupante.

References

1. Report of the Ad Hoc Committee of the Whole of the Twenty-firstSpecial Session of the General Assembly, A/S-21/5/1Add1.New York, United Nations, 1999.

2. Shaping the 21st century: the contribution of developmentcooperation. Paris, Organisation for Economic Cooperationand Development, 1996.

3. A better world for all: progress towards the internationaldevelopment goals. Washington, DC, International MonetaryFund; Paris, Organisation for Economic Cooperation andDevelopment; New York, United Nations; and Washington, DC,World Bank, 2000.

4. The millennium declarations, Resolution A/RES/55/2. New York,United Nations, 2000.

5. Campbell OMA, Graham WJ. Measuring maternal mortalityand morbidity: levels and trends. London, London School ofHygiene and Tropical Medicine, 1990.

6. International statistical classification of diseases and relatedhealth problems. Tenth revision. Geneva, World healthOrganization, 1992.

7. Bouvier-Colle MH et al. Reasons for the underreporting ofmaternal mortality in France, as indicated by a survey of alldeaths among women of childbearing age. International Journalof Epidemiology, 1991, 20: 717–721.

8. Hibbard BM et al. Confidential enquiries into maternal deathsin the United Kingdom, 1988–1990. London, Her Majesty’sStationery Office, 1994.

9. Smith JC et al. An assessment of the incidence of maternalmortality in the United States. American Journal of Public Health,1984, 74 (8): 780–783.

10. Thonneau P et al. Risk factors for maternal mortality: results ofa case-control study conducted in Conakry (Guinea). InternationalJournal of Gynaecology and Obstetrics, 1992, 39: 87–92.

11. Child survival project. Cairo, Ministry of Health, 1994.12. Agoestina T, Soejoenoes A. Technical report on the study

of maternal and perinatal mortality, Central Java Province.Jakarta, Ministry of Health, 1989.

13. Graham W, Brass W, Snow RW. Indirect estimation of maternalmortality: the sisterhood method. Studies in Family Planning,1989, 20: 125–135.

14. Atrash H, Alexander S, Berg C. Maternal mortality indeveloped countries: not just a concern of the past. Obstetricsand Gynecology, 1995, 86 (4, Part 2): 700–705.

15. Turnbull A et al. Report on confidential enquiries into maternaldeaths in England and Wales 1982–84. London, Her Majesty’sStationery Office, 1989.

16. Maternal mortality surveillance — Puerto Rico 1989. Mortalityand Morbidity Weekly Report, 1991, 40: 521–513.

17. Hibbard BM et al. Report on confidential enquiries into maternaldeaths in the United Kingdom 1991–1993. London, Her Majesty’sStationery Office, 1996.

18. WHO/UNICEF/UNFPA. Americas Region consultation onmaternal mortality estimates, Washington, DC, 20–22 April 1998[report]. Geneva, World Health Organization, 1998(unpublished document WHO/RHT/98.27).

19. Royston E, Armstrong S, eds. Preventing maternal deaths.Geneva, World Health Organization, 1989.

20. Reduction of maternal mortality: a joint WHO/UNFPA/ UNICEF/World Bank statement. Geneva, World Health Organization,1999.

21. De Browere V, Tonglet R, Van Lerberghe W. Strategiesfor reducing maternal mortality in developing countries: what canwe learn from the history of the industrialised West? TropicalMedicine and International Health, 1998, 3 (10): 771–782.

22. WHO Reproductive Health Library, No. 2. Geneva, World HealthOrganization, 1999 (computer disk).

23. Graham WJ, Bell JS, Bullough CHW. Can skilled attendanceat delivery reduce maternal mortality in developing countries?Paper presented at: European Commission Expert Meeting onSafer Motherhood, Brussels, Belgium, 27–28 November 2000.

24. UNICEF/WHO/UNFPA. Guidelines for monitoring the availabilityand use of obstetric services. New York, UNICEF, 1997.


Policy and Practice

Annex Table A. Percentage of births assisted by skilled attendants: data available for developingcountries, 1989–2000

Country, area, or territory % of births assisted by Yearskilled attendants

Sub-Saharan AfricaBenin 60 1996Botswana 87 1996Burkina Faso 27 1999Burundi 19 1994Cameroon 55 1998Cape Verde 54 1995Central African Republic 46 1995Chad 15 1997Comoros 52 1996Cote d’Ivoire 47 1999Equatorial Guinea 5 1994Eritrea 21 1995Ethiopia 10 2000Gambia 44 1990Ghana 44 1998Guinea 35 1999Guinea-Bissau 25 1995Kenya 44 1998Lesotho 50 1993Madagascar 47 1997Malawi 55 1992Mali 24 1996Mauritania 40 1991Mauritius 97 1994Mozambique 44 1997Namibia 68 1992Niger 18 1998Nigeria 42 1999Rwanda 26 1992Sao Tome and Principe 86 1989Senegal 47 1997South Africa 84 1998Swaziland 56 1994Togo 51 1998Uganda 38 1995United Republic of Tanzania 35 1999Zambia 47 1996Zimbabwe 73 1999

Middle East and North AfricaAlgeria 77 1992Bahrain 98 1995Cyprus 100 1998Egypt 61 2000Iran (Islamic Republic of) 86 1997Iraq 54 1989Jordan 97 1997Kuwait 98 1995Lebanon 89 1996Libyan Arab Jamahiriya 94 1995Morocco 40 1995Oman 91 1995Qatar 98 1996Saudi Arabia 91 1996Sudan 86 1993Syrian Arab Republic 76 1993Tunisia 81 1995




United Arab Emirates 99 1995West Bank and Gaza Strip 95 1996Yemen 22 1997

South AsiaBangladesh 13 2000Bhutan 15 1994India 43 1999Maldives 90 1994Nepal 9 1996Pakistan 18 1997Sri Lanka 94 1993

East Asia and PacificBrunei Darussalam 98 1994Cambodia 34 1998China 67 1998Cook Islands 99 1991Fiji 100 NAIndonesia 56 1999Kiribati 72 1994Lao People’s Democratic Republic 14 NAMalaysia 96 1998Micronesia (Federated States of) 90 1989Mongolia 93 1998Myanmar 56 1997Niue 99 1990Palau 99 1990Papua New Guinea 53 1996Philippines 56 1998Republic of Korea 98 1990Samoa 76 1990Singapore 100 NASolomon Islands 85 1994Tonga 92 1991Tuvalu 100 1990Vanuatu 87 1994Viet Nam 77 1997

Latin America and the CaribbeanAntigua and Barbados 100 1996Argentina 98 1998Bahamas 100 NAa

Barbados 100 1995Belize 77 1991Bolivia 59 1998Brazil 92 1996Chile 100 1998Colombia 85 1995Costa Rica 98 1998Cuba 100 1999Dominica 100 1999Dominican Republic 96 1996Ecuador 71 1999El Salvador 90 1998Grenada 99 1998Guatemala 41 1999Guyana 95 1997Haiti 21 1995Honduras 55 1996Jamaica 95 1997Mexico 86 1997


Policy and Practice


Nicaragua 65 1998Panama 90 1998Paraguay 71 1998Peru 56 1996Saint Kitts and Nevis 100 1996Saint Lucia 100 1997Saint Vincent and the Grenadines 96 1996Uruguay 100 1999Venezuela 95 1997

CEEb/CISc

Armenia 97 1999Azerbaijan 100 1998Georgia 95 1998Kazakhstan 98 1999Kyrgyzstan 98 1997Tajikistan 79 1996Turkey 81 1998Turkmenistan 96 1996Uzbekistan 98 1996

Source: Demographic and Health Surveys, PAPCHILD, Gulf Child Health Surveys, Reproductive Health Surveys, WHO and UNICEF databases.a NA = Year not available.b CEE = Central and Eastern Europe.c CIS = Confederation of Independent States.




Policy and Practice

maternal mortality at the end of a decade: signs of progress?

Documents