Bulletin of the World Health Organization

Recurrence of adverse perinatal outcomes in developing countries

Fengxiu Ouyang a, Jun Zhang a, Ana Pilar Betrán b, Zujing Yang c, João Paulo Souza b & Mario Merialdi b

a. MOE-Shanghai Key Laboratory of Children’s Environmental Health, Xinhua Hospital, Shanghai Jiao Tong University School of Medicine and School of Public Health, Shanghai, 200092, China.
b. Department of Reproductive Health and Research, World Health Organization, Geneva, Switzerland.
c. Xinhua Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China.

Correspondence to Jun Zhang (e-mail: junjimzhang@gmail.com)

(Submitted: 31 July 2012 – Revised version received: 04 February 2013 – Accepted: 06 February 2013 – Published online: 27 March 2013.)

Bulletin of the World Health Organization 2013;91:357-367. doi: http://dx.doi.org/10.2471/BLT.12.111021

Introduction

Although stillbirths and neonatal deaths are tragic events to the affected mothers and their families, they remain common pregnancy outcomes in developing countries.13 Almost all (97–99%) of the estimated 3 to 4 million stillbirths and 3 million neonatal deaths that occur each year globally occur in low- and middle-income countries.3,4 The causes of stillbirth and neonatal death are generally inseparable.5 The main risk factors for stillbirth include intrapartum complications, maternal infection in pregnancy, maternal disorders (such as hypertension and diabetes), fetal growth restriction and congenital abnormalities.1 The additional risk factors for neonatal death include preterm birth, low birth weight, intrapartum complications and neonatal infection.3,6

Although the women who have suffered stillbirths or seen their neonates die often still want children, they are naturally anxious about the recurrence of adverse outcomes during or after subsequent pregnancies. Most of the data available on the recurrence of perinatal deaths have come from a few developed countries that have efficient systems for the registration of perinatal deaths, and the data on the recurrence of stillbirth appear to be inconsistent.7,8 Very little is known about the risks of such recurrence in developing countries, partly because the recording of perinatal deaths in such countries – which often have no reliable maternity and neonatal databank linked to unique personal identification numbers – is generally poor and difficult.2,9,10 The main aim of the present study is to fill this knowledge gap by examining the associations between adverse outcomes in the first and second pregnancies among women in developing countries. The study of recurrence risk can help identify the causes of adverse pregnancy outcomes – including, possibly, intrinsic and unavoidable risk factors relating to the mother – and guide the counselling of women of childbearing age.10

We used survey data collected in 23 developing countries in Africa, Asia and Latin America as part of the Global Survey on Maternal and Prenatal Health of the World Health Organization (WHO). Our main aim was to examine how – and if – stillbirth or neonatal death in the first pregnancy of a woman in a developing country affected the risk of stillbirth, early neonatal death, admission of the infant to a neonatal intensive care unit (ICU), having a low-birth-weight child and/or the preterm delivery of a child that was small for gestational age in that woman’s second pregnancy. A secondary aim was to see if any associations that we detected differed according to the continent in which the woman lived. Since parity has an effect on pregnancy outcome,11 we analysed only the data for women whose second pregnancies and deliveries were included in the Global Survey.

Methods

Study population

The study population and data collection methods used in this survey are described elsewhere.12,13 The data analysed came from the 2004–2008 Global Survey (the data were collected in 2004–2005 in Africa and Latin America and in 2007–2008 in Asia). The Global Survey, which was multinational and facility-based, was originally designed to help develop a system for monitoring maternal and perinatal health worldwide.12 Stratified multistage cluster sampling was used to select study countries and was followed by random selection of those health facilities in the study countries that had each recorded more than 1000 births in the previous year. All women who had delivered at each of the 373 selected facilities during a specified period of 2 to 3 months were included in the survey and the corresponding mother–infant pairs were followed until the women were discharged or for a maximum of 7 days postpartum. In the time between delivery and discharge, trained data collectors transferred data on each mother–infant pair from the facility’s routine medical records onto a standardized abstraction form. We tried to fill any gaps in the subjects’ medical records via discussion with attending staff before the mothers were discharged.

For the Global Survey, data were collected on 290 610 births and 287 035 pregnancies. Study facilities were scattered across 24 countries in Africa (Algeria, Angola, Congo, Kenya, Niger, Nigeria and Uganda), Asia (Cambodia, China, India, Japan, Nepal, Philippines, Sri Lanka, Thailand and Viet Nam) and Latin America (Argentina, Brazil, Cuba, Ecuador, Mexico, Nicaragua, Paraguay and Peru). For the present analysis, all but one of the study countries (Japan) was categorized as “developing” and only complete data on singleton second pregnancies in the developing countries were included (Fig. 1).

Fig. 1. Flowchart showing the selection of study participants
Fig. 1. <b>Flowchart showing the selection of study participants</b>
a Global Survey on Maternal and Perinatal Health.

Outcomes of the first pregnancy

For the first pregnancies, the main outcomes investigated were stillbirth (i.e. fetal death during or after week 22 of gestation), neonatal death (i.e. death of a liveborn infant at an age of < 28 days) and neonatal survival (i.e. infant living ≥ 28 days postpartum). The secondary outcome investigated – only for infants who lived for at least 28 days postpartum – was birth weight, which was categorized as very low, low and normal when < 1500, 1500–2499 and ≥ 2500 g, respectively.

Outcomes of the second pregnancy

For the second pregnancies, the main outcomes investigated were stillbirth, early neonatal death (i.e. death of a liveborn infant during the first 7 days of life) and admission of the infant to a neonatal ICU (or a similarly high level of care). The secondary outcomes investigated were birth-weight category (very low, low or normal) and the preterm delivery of an infant that was small for gestational age. There were 745 individuals with missing birth weights or gestational ages and 283 “outliers” who had birth weights that fell below percentile 0.001 or above percentile 99.999. This left 60 752 infants with valid gestational ages and birth weights for inclusion in our analyses (Fig. 1).

Small size for gestational age was defined as a birth weight below the 10th percentile of the continent-specific reference values for the same week of gestation. The reference values were determined using the singleton birth weights recorded in the Global Survey and a percentile calculator.13 For week 40, the mean (and standard deviation, SD) singleton birth weights recorded in Africa (n = 24 221), Asia (n = 28 936) and Latin America (n = 26 122) were 3252.4 (480.0), 3099.1 (455.8) and 3346.2 (435.2) g, respectively.

Definitions of major covariates

All the major covariates that we assessed were variables examined for the second pregnancy. They included maternal sociodemographic factors, medical conditions and obstetric complications (see next section). Gestational age, in completed weeks, was based on the best obstetric estimate. Hypertensive disorders were categorized as chronic or as gestational hypertension, pre-eclampsia or eclampsia.

Statistical analysis

Statistical significance was investigated using mixed-model analysis-of-variance F-tests for the continuous variables and χ2 tests for the categorical data (Table 1). We then used multinomial regressions to evaluate odds ratios (ORs) for very low and low birth weights (compared with normal birth weights; Table 2) and for stillbirth and early neonatal death (compared with neonatal survival; Table 3) in the second pregnancy, according to the outcome of the first pregnancy. Logistic regressions were used to evaluate the associations between previous pregnancy outcomes and the preterm delivery of an infant who was small for gestational age or admission of the infant to a neonatal ICU (Table 4).

We included covariates that were of a priori interest (Model 1) or that are known risk factors for adverse pregnancy outcomes (Model 2) in all of the logistic and multinomial regression models. Covariates in Model 1 included sociodemographic factors that may influence the outcomes of second pregnancies, such as maternal age (categorized as < 18, 18–34 or ≥ 35 years), marital status, years of education (categorized as < 7 or ≥ 7 or unknown), country of residence (22 indicator variables were generated for 23 countries), and the level of the health facility in which the second pregnancy was managed (categorized as primary, secondary, tertiary or “other”). In Model 2 we used the same covariates plus several maternal medical conditions and obstetric factors (pre-existing diabetes, hypertensive disorders, mode of delivery, type of onset of labour, pre-labour rupture of membranes, fetal presentation at delivery, urinary infection and antenatal antibiotic treatment). Since small size for gestational age and preterm birth are major risk factors for fetal and/or neonatal death,14, 15 we investigated a third model (Model 3) that included small size for gestational age and preterm status (categorized as a gestational age at birth of < 34, 34–36 or ≥ 37 weeks), as well as all of the covariates included in Model 2 (Table 3).

Among the women who had first pregnancies that ended in neonate survival, we also investigated if very low or low birth weight in the first pregnancy was associated with stillbirth, preterm birth of a small-for-gestational age infant and/or admission of the infant to a neonatal ICU in the second pregnancy, with adjustment for the same covariates considered in Model 1 or 2 (Table 5). Small sample sizes precluded an investigation of the association between neonatal death in the second pregnancy and low birth weight in the same pregnancy.

Generalized estimating equations were applied to all logistic and multinomial regression models and independence correlation structures were used to adjust for the variable correlation within each health facility. All of the data analysis was performed using version 9.1 of the SAS software package (SAS Institute, Cary, United States of America). A P-value of < 0.05 was considered indicative of a statistically significant difference or association.

Results

The present study was based on the data for 61 780 women who had been included in WHO’s Global Survey on Maternal and Prenatal Health as they experienced their second pregnancies. The mean age of these women at their second deliveries was 26.1 years (SD: 5.0 years). They lived in Africa (25.4%), Asia (40.7%) or Latin America (33.8%). In their first pregnancies, there had been 1261 stillbirths and 1052 neonatal deaths (Table 1). In their second pregnancies, there had been 910 stillbirths and 398 early neonatal deaths.

The women could be split into three groups according to the outcomes of their first pregnancies (i.e. whether the first pregnancies had ended in stillbirth, neonatal death or neonatal survival). Although these three groups of women showed similar frequencies of antenatal antibiotic treatment, diabetes and pre-labour rupture of membranes in their second pregnancies (P > 0.05 for each), they differed significantly in terms of most of the maternal demographic variables, maternal medical conditions and obstetric factors that had been recorded for the second pregnancies and also in terms of the outcomes of their second pregnancies (P ≤ 0.05 for each; Table 1).

In Model 2 – after adjusting for possible confounders and in comparison with the women whose first pregnancies had ended in neonate survival (Table 2) – the women who had previously suffered a stillbirth were found to be significantly more likely to have had a second pregnancy ending in the delivery of an infant with very low birth weight (OR: 2.52; 95% confidence interval, CI: 1.68–3.76) – but not more likely to have delivered an infant with low birth weight (OR: 1.22; 95% CI: 0.99–1.52). In the same model, and again in comparison with the women whose first pregnancies had ended in neonate survival (Table 2), the women who had previously suffered a neonatal death were found to be more likely to have had second pregnancies ending with the delivery of an infant with low birth weight (OR: 1.60; 95% CI: 1.31–1.96) or very low birth weight (OR: 2.78; 95% CI: 1.69–4.56). Of the infants delivered at the end of the second pregnancies, those born to women who had previously suffered a stillbirth or neonatal death were at increased risk of admission to an intensive-care unit (OR: 1.64 and 1.68, respectively; P < 0.001 for each; Table 4).

In their second pregnancies, in comparison with the women whose first pregnancies had ended in neonate survival, the women who had previously suffered a stillbirth were more likely to have had a stillbirth (4.6% versus 1.4%) and, similarly, the women who had previously suffered a neonatal death were more likely to have had an infant that died as a young neonate (2.4% versus 0.6%) (Table 1 and Table 3). The results from Model 1 indicate that, after adjustment for demographic variables and the level of the health facility where the delivery occurred – and in comparison with the women whose first pregnancies had ended in neonate survival – the odds of stillbirth in the second pregnancy were 2.79-fold higher (95% CI: 1.97–3.94) for the women who had already suffered a stillbirth and 1.81-fold higher (95% CI: 1.31–2.51) for the women who had previously suffered a neonatal death (Table 3). In comparison with the women whose first pregnancies had ended in neonate survival, the odds of early neonatal death in the second pregnancy were 3.21-fold higher (95% CI: 2.03–5.08) for the women who had already suffered a neonatal death but were not significantly higher (OR: 1.61; 95% CI: 0.90–2.89) for the women who had previously suffered a stillbirth (Table 3). These associations were found to be slightly weaker after the additional adjustments made – for known risk factors for stillbirth or neonatal death – in Model 2 (Table 3). In Model 3, the additional adjustments for small size for gestational age and preterm status further reduced the odds of the second pregnancy of a woman who had previously suffered a stillbirth ending in a second stillbirth – from an OR of 2.35 (95% CI: 1.65–3.37) to one of 1.91 (95% CI: 1.25–2.92) – and the odds of a woman who had previously suffered a neonatal death giving birth to a second child who died as a young neonate – from an OR of 2.82 (95% CI: 1.76–4.52) to an OR of 2.19 (95% CI: 1.29–3.73). However, the addition of small size for gestational age and preterm status to Model 3 left no significant association between neonatal death after the first pregnancy and stillbirth in the second pregnancy (Table 3).

In Model 2, after adjustment for possible confounders, the women whose infants were alive 28 days after their first pregnancies were found to be at increased risk of the preterm birth of an infant who was small for gestational age, infant admission to an ICU and stillbirth in their second pregnancies if their first child had a low or very low birth weight, than if that child had a normal birth weight (Table 5).

The associations that we detected appeared to be largely unaffected by the women’s continent of residence (Table 6 and Table 7).

Discussion

In this study, women with an adverse outcome (i.e. stillbirth, neonatal death, infant with low birth weight and/or infant with very low birth weight) in their first pregnancy were found to be at increased risk of an adverse outcome (i.e. stillbirth, neonatal death, neonate requiring admission to an ICU, preterm birth of an infant that was small for gestational age, an infant with low birth weight or very low birth weight) in their second pregnancies, indicating that some of the adverse outcomes have causes in common. Of note, after adjusting for small size for gestational age, preterm status and other covariates, women who had had a first pregnancy that ended in a stillbirth still appeared to be at increased risk of stillbirth at the end of their second pregnancy, while women who had seen their first children die as neonates were at increased risk of seeing a live birth followed by early neonatal death in their second pregnancies.

In earlier investigations, perinatal deaths have been associated with an increased risk of low-birth-weight infants in subsequent pregnancies.16, 17 Low birth weights probably reflect suboptimal fetal environments and/or short pregnancies.18 Very low birth weights are generally preterm infants.10 Even among the women whose first children survived for at least 28 days postpartum, there was a strong link between very low birth weights in the first pregnancy and the risk of stillbirth in the second pregnancy. These findings indicate that the causes of at least some perinatal deaths are related to the causes of very low birth weights.

In this study, recurrence associations for stillbirth and neonatal death were seen even after adjustment for the known risk factors for perinatal death and potential confounders. Stillbirths in the first two pregnancies may therefore have common biological causes beyond any known risk factors for stillbirth that may develop during the second pregnancy. This may also be true for neonatal deaths.

Although there have been a few previous population-based studies on stillbirth recurrence in the second pregnancy, the findings were inconsistent.7,19 In a study in the Grampian region of Scotland, in the United Kingdom of Great Britain and Northern Ireland, no association was detected between stillbirth in one pregnancy and the risk of stillbirth in the subsequent pregnancy.7 In an analysis of data collected across Scotland, however, women who had previously suffered a stillbirth were found to be almost twice as likely to have a pregnancy ending in stillbirth (OR: 1.94; 99% CI: 1.29–2.92) than women who had no history of stillbirths.8 The present results indicate that – in developing countries at least – stillbirths in first pregnancies are often predictors of the same adverse outcome in second pregnancies. The strength of this association in our study population was similar to that reported in a study of national data from Sweden19 but weaker than that reported in a study based in the state of Missouri, in the USA (OR: 5.8; 95% CI: 3.7–9.0).20

Our findings appear to be biologically plausible. Stillbirths have diverse etiologies.19,2123 Although we adjusted in our analysis for many maternal and fetal factors, there were still a few factors that we did not take into account. For example, we did not consider the impact of genetic abnormalities and birth defects, which are known to recur through multiple pregnancies.24 In addition, >  20% of stillbirths are currently “unexplained” by maternal, obstetric, placental or fetal factors.25,26 Although there is no available evidence to support this possibility, a previous history of stillbirth may still be a useful predictor of even these “unexplained” stillbirths.

Since they are at increased risk of an adverse outcome in their second pregnancies, women who have suffered an adverse outcome in their first pregnancies should be immediately identified as “at-risk”, offered educational materials/pamphlets – and/or be invited to attend educational courses – on the risks and prevention of adverse outcomes in pregnancy, and be encouraged to attend a hospital for any subsequent deliveries and, if appropriate and possible, be offered emergency obstetric care for subsequent pregnancies.5,27 It is also important to raise awareness at the community level of the risks and needs of such women. Home visits for women and children and the participation of women’s groups have been identified as useful, evidence-based, community strategies for reducing reproductive, maternal, neonatal and child mortality and for promoting reproductive health.6

In this study, controlling for maternal medical conditions and pregnancy complications weakened the association between stillbirth in the first pregnancy and the preterm delivery of an infant who was small for gestational age at the end of the second pregnancy (Table 4). Similarly, controlling for small size for gestational age and preterm status weakened the association between death of the neonate after the first pregnancy and stillbirth in the second pregnancy (Table 3). It therefore appears that maternal medical conditions, pregnancy complications, small size for gestational age and preterm (especially very preterm) delivery may play a substantial role in the occurrence of perinatal death.

The present study appears to be the first multinational investigation of the recurrence of adverse pregnancy outcomes in developing countries. Its strengths include the large sample size, the consideration of multiple, maternal, medical and obstetric factors, and its use of fairly detailed information on pregnancy outcomes and neonatal mortality. The data on second pregnancies that were used in the analysis were all collected over a period of just 2–3 months. This minimized the potentially confusing effects of any long-term trends that may have occurred in the variables that were recorded, such as decreases in stillbirth rates as the general levels of health care gradually improved.28

Our study also had several limitations. To be included in the study, a woman had to have delivered at a health facility that dealt with ≥ 1000 births per year. This inclusion criterion excluded data from very small hospitals and home births. Consequently, the findings may not be generalized to the countries and continents involved in this study, especially not to those countries where delivery in a health facility is the exception rather than the norm. We could not make any allowance for the effects of maternal smoking, the mother’s body mass index before she became pregnant or the interval between the mother’s first and second pregnancies, since no data on these variables were recorded in the Global Survey. However, smoking and pre-pregnancy body mass index are known to have an impact on two of the covariates that we did consider in some of our models: small size for gestational age and preterm status.11,29,30 In the present study, size for gestational age status may have been recorded inaccurately for antepartum stillbirths because, for the infants concerned, gestational age at delivery may have been longer than gestational age at death. Such inaccuracy may have led to the misclassification of some stillborns as small for gestational age. Finally, no account was taken of the duration of each woman’s hospitalization for her deliveries although, in countries where women who have been admitted for delivery are generally discharged a few days postpartum, many neonatal deaths may occur at home and never be listed in the woman’s medical records. In China, for example, around 17% of the infants born in urban hospitals were found to have died at home during their first 28 days of life.31 If, in the present study, mothers had been discharged before their infants had died aged <  7 days, the outcomes of their pregnancies would have been incorrectly recorded as neonate survival. However, the number of early neonatal deaths missed because of rapid discharge from Chinese hospitals should have been relatively small because about 63% of neonatal deaths in China occur within 2 days of birth.31

In conclusion, analyses of data collected as part of the WHO’s Global Survey on Maternal and Prenatal Health indicate that an adverse outcome in the first pregnancy of a woman living in the developing world is often a predictor of an adverse outcome in that woman’s second pregnancy. A woman whose first pregnancy had ended in a stillbirth or in the birth of an infant who died as a neonate was at increased risk of suffering exactly the same tragedy in her second pregnancy. The identification of such women before they become pregnant for a second time would allow them to be closely monitored for various fetal and maternal complications, with the aim of increasing the chances that the women would subsequently deliver healthy infants.


Acknowledgements

JZ and FO were supported by the National Natural Science Foundation of China (via grants 81273091 and 81102139, respectively). FO was also supported by the Shanghai Pujiang Program (via grant 12PJD026).

Competing interests:

None declared.

References

Share