Infant mortality and subsequent risk of stillbirth: a retrospective cohort study

Authors


Prof H Salihu, Department of Epidemiology & Biostatistics, College of Public Health, University of South Florida, 13201 Bruce B. Downs Blvd, MDC56, Tampa, FL 33612, USA. Email hsalihu@health.usf.edu

Abstract

Please cite this paper as: August E, Salihu H, Weldeselasse H, Biroscak B, Mbah A, Alio A. Infant mortality and subsequent risk of stillbirth: a retrospective cohort study. BJOG 2011;118:1636–1645.

Objective  To examine the association between infant mortality in a first pregnancy and risk for stillbirth in a second pregnancy.

Design  Population-based, retrospective cohort study.

Setting  Maternally linked cohort data files for the state of Missouri.

Population  Women who had two singleton pregnancies in Missouri during the period 1989–2005 (n = 320 350).

Methods  Women whose first pregnancy resulted in infant death were compared with those whose infant from the first pregnancy survived the first year of life. The Kaplan–Meier product limit estimator was employed to compare probabilities for stillbirth in the second pregnancy between both groups of women. Adjusted hazard ratios (AHRs) and 95% confidence intervals (95% CIs) were generated to assess the association between infant mortality in the first pregnancy and stillbirth in the second pregnancy.

Main outcome measures  Exposure was defined as infant mortality in the first pregnancy, and the outcome was defined as stillbirth in the second pregnancy.

Results  Women with prior infant deaths were about three times as likely to experience stillbirth in their subsequent pregnancy (AHR 2.91; 95% CI 2.02–4.18). White women with a previous infant death were nearly twice as likely to experience a subsequent stillbirth, compared with white women with a surviving infant (AHR 1.96; 95% CI 1.13–3.39). Black women with a previous infant death were more than four times as likely to experience subsequent stillbirth, compared with black women with a surviving infant (AHR 4.28; 95% CI 2.61–6.99).

Conclusions  Previous infant mortality results in an elevated risk for subsequent stillbirth, with the most profound increase observed among black women. Interconception care should consider prior childbearing experiences to avert subsequent fetal loss.

Introduction

Despite reports of an estimated 3.2–3.3 million stillbirths annually worldwide,1–4 stillbirth is relatively absent from global health policy agendas, including the United Nations Millennium Development Goals.5,6 Inconsistent definitions for stillbirth abound, which, along with other data limitations, particularly in developing countries (e.g. fragmented vital records systems, a prevalence of home births, and a lack of political will),2,3,6,7 can hinder efforts to describe, understand, and subsequently prevent the occurrence of stillbirth. Not surprisingly, most interventions to prevent stillbirth also require additional research to improve the quality of evidence.8,9

Many newborn infants that survive the neonatal period are not out of the proverbial woods. Infant mortality, defined as the death of a child within the first year of life,10 accounts for an estimated 5.75 million deaths worldwide, or 42 infant deaths per 1000 live births.11 The United Nations Millennium Development Goals call for a reduction by two-thirds of the mortality rate of children under the age of 5 years by the year 2015;5 however, it excludes an explicit goal for infant mortality. Preventive efforts must be directed at preterm delivery and low birthweight, which are the main precursors of infant mortality.12 Moreover, racial/ethnic disparities in infant mortality highlight the need to consider more distal factors, such as secondary effects of low socio-economic status, suboptimal social support, and unhealthy neighborhoods.13

Just as the first year of life can be a precarious time period for many infants, pregnancy following a fetal loss can carry with it a heightened risk for subsequent adverse events, such as preterm delivery,14,15 low birthweight,16 and stillbirth recurrence.17–19 A related association that has largely gone unexamined to date is the subsequent risk of stillbirth after the death of a previous infant. Despite the relatively frequent occurrence of both stillbirth and infant mortality in the USA, little is known regarding the association between the two. A recent study conducted in one US county attempted to examine the association between previous infant death and subsequent fetal mortality, but had an insufficient sample size.20 To address this gap in the literature, we used a population-based data set to examine the association between infant mortality and stillbirth during a subsequent pregnancy. Furthermore, we investigated whether racial disparities exist within this association.

Methods

We used the Missouri maternally linked cohort data files for the period 1989–2005 (inclusive), with a total population of 1 035 547 births. This data set links siblings to their biological mothers using unique identifiers. The methods and algorithm used in linking birth certificate data with sibling relationships and the process of validation have been described in detail previously.21 The Missouri vital record system is a reliable record system that has been adopted as the gold standard to validate US national data sets that involve matching and linking procedures.22 The Missouri maternally linked cohort data set contains information on both live births and fetal deaths for each sibling, and provides a platform for a longitudinal study of birth outcomes for each pregnancy.

The main exposure of interest in the present study was infant mortality (defined as the death of the infant within the period from day 0 to 364) associated with the first pregnancy. We also considered neonatal death (death of a newborn within the first 28 days of life) and post-neonatal death (death of the infant within the period from day 29 to 364). Our primary outcome was stillbirth (in utero fetal death occurring after at least 20 weeks of gestation) associated with the second pregnancy. As our objective was to examine the association between infant mortality and stillbirth in the first and second pregnancies, respectively, we selected two consecutive births from the same mother with a gestational age range of 20–44 weeks. After applying further exclusion criteria (Figure 1), we retained a total of 640 700 births (320 350 from each pregnancy) for this analysis.

Figure 1.

 Flow diagram of exclusion criteria for the study.

We compared mothers with previous infant death versus those who had infants who survived their first year with respect to the following maternal characteristics at the time of the second delivery: race, age, marital status, educational level, cigarette smoking during pregnancy, body mass index, interpregnancy interval, and adequacy of prenatal care. Maternal race was grouped into three categories: white, black, and other. Maternal age was dichotomized as women who were of advanced age (i.e. 35 years old or older) or were younger than 35 years old at the time of the second delivery. Maternal marital status was grouped as either married or unmarried, with all women who were single, divorced, or widowed being classified as unmarried. Maternal educational level was categorized as women with at least a high school diploma or its equivalent (≥12 years of education) and those without a high school diploma (<12 years of education). Pre-pregnancy obesity was determined based on body mass index (BMI), which was calculated as weight (kg)/height (m2) using height, as measured at the first prenatal visit of the second pregnancy, and pre-pregnancy weight, as reported at the first prenatal visit. Obese mothers were defined as those with a pre-pregnancy BMI of ≥30. Adequacy of prenatal care was assessed using the Revised-Graduated Index (R-GINDEX) algorithm. This index assesses the adequacy of care based on the trimester when prenatal care began, the number of visits, and the gestational age of the infant at birth.23 In the present study, inadequate prenatal care use refers to women who had prenatal care but the level was considered to be suboptimal (i.e. fewer prenatal care visits relative to the length of pregnancy) or had no prenatal care at all. The interpregnancy interval was calculated as the number of days between the date of birth of the child in the first pregnancy and the first day of the last menstrual period (LMP) of the second pregnancy.

We also performed crude frequency comparisons for the presence of common obstetric and medical complications during the second pregnancy. These complications included: anaemia (defined as <30 cl/l haematocrit and/or <12 g/dl haemoglobin); insulin-dependent diabetes mellitus (defined as the absolute deficiency of insulin secretion); other types of diabetes mellitus (defined as either gestational diabetes or adult-onset diabetes); chronic hypertension (defined as an increase in systolic or diastolic pre-existing blood pressure to a level of ≥140/90 mmHg, respectively, prior to 20 weeks of gestation); pre-eclampsia (defined as pregnancy-induced high blood pressure at ≥140/90 mmHg and excess protein in the urine after 20 weeks of gestation); eclampsia (defined as pregnancy-induced hypertension associated with convulsions); abruptio placenta (defined as the premature separation of normally implanted placenta from the uterus); placenta praevia (defined as an attachment of the placenta to the uterine wall close to or covering the cervix); and renal disease. We also constructed a composite variable indicating the presence of at least one of these conditions. Moreover, we compared the presence or absence of congenital anomalies during the second pregnancy by constructing a composite variable. The anomalies considered for the analysis were anencephalus, spina bifida/meningocele, hydrocephalus, microcephalus, heart malformation, rectal atresia/stenosis, tracheoesophageal fistula/esophageal atresia, omphalocele/gastroschisis, malformed genitalia, renal agenesis, cleft lip/palate, polydactyly/syndactyly/adactyly/clubfoot, diaphragmatic hernia/trisomy 21, and other chromosomal anomalies.

To examine the impact of potential bias on our results, the data were analysed using two strategies. First, we excluded cases with missing information from multivariable analysis. Second, we conducted multiple imputation for the missing data, inserting plausible values using Rubin’s method,24,25 and analysed the data. Both of these approaches yielded similar results.

Statistical analysis

Stillbirth rates were computed by dividing the number of stillbirths by the sum of live births and stillbirths, and multiplying by 1000. The chi-square test was used to assess differences between the two groups in sociodemographic characteristics and maternal pregnancy complications in the second pregnancy for categorical variables. The Student’s t test was used for the comparison of parametric continuous data (e.g. birthweight), and the Mann–Whitney–Wilcoxon test was applied when the variables were continuous but non-parametric (e.g. interpregnancy interval). Stillbirth among mothers with a prior infant death (exposed group) was compared with stillbirth among mothers whose infants survived infancy (unexposed group) by using the Kaplan–Meier product-limit estimator, which calculates the probability of stillbirth for each group. To determine whether the difference in the survival rate between the two groups was statistically significant, we applied the Wilcoxon statistic, instead of the more common Mantel log-rank test because preliminary analysis showed a preponderance of stillbirth in earlier gestations (data not shown). We used the Cox proportional hazards regression model to derive adjusted hazard ratios (AHRs) after testing for non-violation of the proportionality assumption in each case. We confirmed this by plotting the log-negative-log of the Kaplan–Meier estimates of the survival function versus the log of time. The resulting curves were parallel.26 AHRs were derived from loading all variables considered to be potential confounding factors into the model.

All hypothesis tests were two-tailed, with a type-1 error rate fixed at 5%. sas 9.2 (SAS Institute, Cary, NC, USA) was used to perform all analyses. This study was approved by the Institutional Review Board at the University of South Florida prior to initiation.

Results

A total of 320 350 women had first and second consecutive singleton pregnancies in the study period. Of these, 2483 women (0.78%) had experienced infant death in the first pregnancy (exposed group), whereas the remaining 317 867 women had an infant who survived the first year of life from the first pregnancy (unexposed group). Within the study population, 1347 cases of stillbirth occurred during the second pregnancy, representing a stillbirth rate of 4.2 per 1000.

A comparison of selected maternal demographic characteristics and obstetric complications in the second pregnancy by exposure status is presented in Table 1. Women with infant death in the first pregnancy were more likely to be black, obese, and to have a lower level of education. These women also tended to smoke during pregnancy and to have a higher level of adequate prenatal care. On the other hand, women in the unexposed group were more likely to be white and married. There was no difference in terms of maternal age between the two groups. On average, mothers with a prior infant death experienced their second pregnancy much earlier than their counterparts with surviving infants (mean interpregnancy interval ± standard deviation [SD]: 616 ± 708 versus 934 ± 761 days; < 0.01). When comparing birthweights of infants associated with the second pregnancy, infants born to mothers with prior infant death were, on average, 293 g smaller than those born to mothers whose previous infant survived their first year of life (mean birthweight: 3106 ± 761 versus 3399 ± 559 g; P < 0.01).

Table 1.   Distribution of selected maternal sociodemographic characteristics in the second pregnancy between women with infant mortality in the first pregnancy and women whose infants survived the first year of life (Missouri: 1989–2005)*
CharacteristicsInfant death (= 2483) n (%)No infant death (= 317 867) n (%)P
  1. *Significant values are set in bold font (P < 0.05 considered as significant).

Race<0.01
White1863 (75.03)267 326 (84.10)
Black582 (23.44)44 289 (13.93)
Other38 (1.53)6252 (1.97)
Maternal age0.70
<35 years2216 (89.25)282 923 (89.01)
≥35 years267 (10.75)34 944 (10.99)
Maternal educational level<0.01
<12 years562 (22.63)51 433 (16.18)
≥12 years1899 (76.48)264 515 (83.22)
Missing/unknown22 (0.89)1919 (0.60)
Marital status<0.01
Unmarried915 (36.85)82 269 (25.88)
Married1567 (63.11)235 435 (74.07)
Missing/unknown1 (0.04)163 (0.05)
Pre-pregnancy obesity<0.01
BMI < 301870 (75.31)251 969 (79.27)
BMI ≥ 30 (obese)523 (21.06)57 348 (18.04)
Missing/unknown90 (3.62)8550 (2.69)
Maternal prenatal smoking<0.01
Yes606 (24.41)61 978 (19.50)
No1870 (75.31)255 244 (80.30)
Missing/unknown7 (0.28)645 (0.20)
Prenatal care<0.01
Adequate1577 (63.51)167 446 (52.68)
Inadequate766 (30.85)138 114 (43.45)
Missing/unknown140 (5.64)12 307 (3.87)
Mean interpregnancy interval (in days)616 ± 708 days934 ± 761 days<0.01

The incidence of pregnancy complications during the second pregnancy was 6.7% (21 448) in the study population. Overall, pregnancy complications were almost twice as frequent among mothers who had experienced an infant death during their first pregnancy, as compared with those whose infants survived their first year of life (10.91 versus 6.66%; < 0.01). Using crude rates, mothers with a prior infant death were nearly twice as likely to experience insulin-dependent diabetes, chronic hypertension, and abruptio placenta in their second pregnancy, and were 1.6 times as likely to present with pre-eclampsia and other types of diabetes. Comparable rates of anaemia, eclampsia, placenta praevia, renal disease, and congenital anomalies in the infant were observed in both groups (Table 2).

Table 2.   Distribution of selected complications in the second pregnancy between women with infant mortality in the first pregnancy and women whose infants survived the first year of life (Missouri: 1989–2005)*
Pregnancy complicationsInfant death (= 2483) n (%)No infant death (n = 317 867) n (%)P
  1. *Significant values are set in bold font (P < 0.05 considered as significant).

Anaemia26 (1.05)4591 (1.44)0.15
Insulin-dependent diabetes mellitus44 (1.77)2437 (0.77)<0.01
Other types of diabetes95 (3.83)7278 (2.29)<0.01
Chronic hypertension55 (2.22)2957 (0.93)<0.01
Pre-eclampsia123 (4.95)9359 (2.94)<0.01
Eclampsia2 (0.08)186 (0.06)0.53
Abruptio placenta41 (1.65)2412 (0.70)<0.01
Placenta praevia10 (0.40)1358 (0.43)0.68
Renal disease8 (0.32)780 (0.25)0.44
Congenital anomalies of the infant35 (1.41)3974 (1.25)0.48
Overall complications271 (10.91)21 177 (6.66)<0.01

A comparison between mothers who had experienced prior infant death and those whose infants had survived their first year of life relative to the cumulative probability of survival before 44 weeks of gestation is displayed by the Kaplan–Meier curves in Figure 2, which shows the rate of stillbirth to be significantly higher among mothers with prior infant mortality (< 0.01).

Figure 2.

 Kaplan–Meier curves comparing the probability of stillbirth between infants born to mothers with a prior infant death and those born to mothers who had not experienced a prior infant death (Missouri: 1989–2005).

The AHRs, adjusted for maternal sociodemographic characteristics, among women who had previously experienced an infant death are presented in Table 3. Compared with white women, black women were more than twice as likely to experience stillbirth (AHR 2.06; 95% CI 1.78–2.39). Single women had a 34% elevated likelihood of stillbirth, as compared with married women (AHR 1.34; 95% CI 1.16–1.55). Furthermore, women with pregnancy complications had a nearly 3.5-fold increased risk for stillbirth (AHR 3.48; 95% CI 2.99–3.98). Smoking during pregnancy and inadequate prenatal care were both associated with approximately a 70% heightened risk for stillbirth (smoking, AHR 1.27, 95% CI 1.50–1.95; prenatal care, AHR 1.68, 95% CI 1.50–1.89). Maternal age, educational level, and pre-pregnancy obesity did not show elevated risks for stillbirth.

Table 3.   Adjusted hazard ratios (AHRs) for stillbirth among women who experienced a prior infant death*
 Stillbirth after infant death
AHR (95% CI)
  1. *Significant values are in bold font.

Race
WhiteRef.
Black2.06 (1.78–2.39)
Maternal age
<35 yearsRef.
≥35 years1.17 (0.98–1.39)
Maternal level of education
≥12 yearsRef.
<12 years0.99 (0.85–1.17)
Marital status
MarriedRef.
Unmarried1.34 (1.16–1.55)
Pre-pregnancy obesity
BMI < 30Ref.
BMI  30 (obese)1.05 (0.91–1.20)
Maternal prenatal smoking
Yes1.71 (1.50–1.95)
NoRef.
Prenatal care
AdequateRef.
Inadequate1.68 (1.50–1.89)
Pregnancy complications
Yes3.48 (2.99–3.98)
NoRef.

Table 4 provides adjusted estimates for the association between the death of a first child in infancy and stillbirth during the subsequent pregnancy. Model 1 provides hazard ratios adjusted for sociodemographic variables in the second pregnancy, whereas model 2 provides estimates with further adjustment for pregnancy complications. In model 1, mothers with a previous infant death or neonatal death were more than three times as likely to experience stillbirth during the second pregnancy (infant death, AHR 3.09; 95% CI 2.15–4.45; neonatal death, AHR 3.51; 95% CI 2.34–5.27). Adjusting for pregnancy complications in model 2 attenuated the overall risk (among women with prior infant death) of stillbirth in the second pregnancy by 18% (AHR 2.91; 95% CI 2.02–4.18), and the risk of stillbirth following neonatal death by 24% (AHR 3.27; 95% CI 2.18–4.91). Further analysis stratified by race showed that black mothers who previously had an infant death in the first pregnancy were more than four times as likely to experience stillbirth in their subsequent pregnancy, as compared with black mothers whose first pregnancy did not result in infant death. For white mothers who previously experienced an infant death, the risk of subsequent stillbirth was about two-fold as great, as compared with white mothers whose first pregnancy did not result in infant death.

Table 4.   Risk of stillbirth among women with a history of infant, neonatal, or post-neonatal death in the first pregnancy (Missouri: 1989–2005)*
 Model 1**Model 2***
Stillbirth after infant deathStillbirth after neonatal deathStillbirth after post-neonatal deathStillbirth after infant deathStillbirth after neonatal deathStillbirth after post-neonatal death
AHR (95% CI)AHR (95% CI)AHR (95% CI)AHR (95% CI)AHR (95% CI)AHR (95% CI)
  1. *Significant values are set in bold font.

  2. **Adjusted for sociodemographic variables.

  3. ***Adjusted for model 1 plus pregnancy complications.

Women with previous infant death3.09 (2.15–4.45)3.51 (2.34–5.27)2.11 (0.94–4.70)2.91 (2.02–4.18)3.27 (2.18–4.91)2.02 (0.90–4.50)
Women with no previous infant deathRefRefRefRefRefRef
White women with previous infant death2.08 (1.20–3.60)2.45 (1.35–4.44)1.16 (0.29–4.67)1.96 (1.13–3.39)2.28 (1.26–4.14)1.11 (0.28–4.43)
White women with no previous infant deathRefRefRefRefRefRef
Black women with previous infant death4.60 (2.82–7.52)5.18 (2.96–9.06)3.34 (1.25–8.91)4.28 (2.61–6.99)4.84 (2.49–8.47)3.07 (1.40–8.24)
Black women with no previous infant deathRefRefRefRefRefRef

We further compared the risk of stillbirth among women with and without a history of infant death, neonatal death, and post-neonatal death by race, using white mothers without a history of infant death as the referent category (Table 5). The results showed that black women with a history of infant, neonatal, or post-neonatal death have the most elevated rates of stillbirth in a subsequent pregnancy, as compared with white mothers with no prior history of infant, neonatal, and post-neonatal death. Black women with a prior infant death have an eight-fold increased likelihood of stillbirth (AHR 8.41; 95% CI 5.14–13.76). More specifically, black women with a history of neonatal death were more than nine times as likely to experience stillbirth (AHR 9.46; 95% CI 5.42–16.49), whereas those with post-neonatal death were more than six times as likely to experience stillbirth (AHR 6.21; 95% CI 2.30–16.78). Interestingly, black women without previous infant, neonatal, or post-neonatal deaths experienced similar rates of stillbirth (AHR 2.07; 95% CI 1.77–2.41) as white women with a history of infant death (AHR 2.00; 95% CI 1.55–3.46) and neonatal death (AHR 2.33; 95% CI 1.29–4.23).

Table 5.   Risk of stillbirth among women with and without a history of infant, neonatal, or post-neonatal death in the first pregnancy by race (Missouri: 1989–2005)*,**
 Stillbirth after infant deathStillbirth after neonatal deathStillbirth after post-neonatal death
AHR (95% CI)AHR (95% CI)AHR (95% CI)
  1. *Significant values are set in bold font.

  2. **Model adjusted for sociodemographic variables.

White women with no previous infant death, neonatal death, or post-neonatal deathRef.Ref.Ref.
White women with previous infant death, neonatal death, or post-neonatal death2.00 (1.55–3.46)2.33 (1.29–4.23)1.13 (0.28–4.56)
Black women with previous infant death, neonatal death, or post-neonatal death8.41 (5.14–13.76)9.46 (5.42–16.49)6.21 (2.30–16.78)
Black women with no previous infant death, neonatal death, or post-neonatal death2.07 (1.77–2.41)2.10 (1.80–2.45)2.07 (1.77–2.41)

Discussion and conclusion

In this population-based, retrospective cohort study, we found that a history of infant mortality was associated with a nearly three-fold increased risk of subsequent stillbirth, which was elevated further when the previous delivery resulted in neonatal death. This heightened risk remained present after an adjustment for sociodemographic and pregnancy-related covariates. Within our model of association, multiple covariates were included, based on biological plausibility and a review of the literature, which may have affected the resulting association observed between prior experiences of infant mortality and the ensuing likelihood of stillbirth. Therefore, it is possible that this association reflects an indirect relationship.

Our findings provide further support of the black–white disparity in the occurrence of stillbirth. Black women with prior infant, neonatal, or post-neonatal deaths were more than nine times as likely to experience stillbirth, compared with white women with no such history. Black women who previously experienced infant death had a two-fold greater likelihood of stillbirth, as compared with white women who similarly experienced infant death. Furthermore, an association between post-neonatal mortality in the first pregnancy and stillbirth in the second pregnancy was only observed among black mothers. In a similar pattern, a study by Willinger, Ko and Reddy27 found that black women have a 2.2-fold increased risk of stillbirth compared with white women. Although no known studies have specifically examined previous infant loss and subsequent stillbirth, previous research that studied the association of a history of stillbirth with its recurrence showed heightened rates among black women compared with white women.28

Overall, we found a stillbirth rate of 4.2 per 1000 births in our study population. This rate is approximately one-third lower than the US fetal mortality rate of 6.22 fetal deaths of 20 weeks of gestation or more per 1000 live births and fetal deaths.29,30 Furthermore, the fetal mortality rate in our study falls below the previously reported rate for Missouri (5.6 per 1000).30 Within our study, the stillbirth rate only reflects fetal deaths at 20 weeks of gestation or more during the second pregnancy within sibling pairs. Therefore, given the high reliability of the Missouri maternally linked cohort data files,22 it is possible that the application of our exclusion criteria reduced the number of cases of stillbirth included in our analysis and the subsequent calculation of the stillbirth rate (restricted only to two consecutive pregnancies).

Although global attention towards stillbirth has increased notably in recent years,2,6 the present paper represents an important contribution to an under-researched area: namely, the association between infant mortality in a primary pregnancy and a subsequent stillbirth. The relative deficiency of existing data regarding stillbirth partially explains the limitations of research findings in this area. Only about half of US stillbirths are reported to have a determined cause of death on record.18 Not surprisingly then, global-level reports are often forced to rely on household surveys and modelled estimates;2,6 even stillbirth data within high-income countries have been labelled as deficient.1,2,8

Intrapartum stillbirths are estimated to represent one-third of stillbirths globally,2,31 which would suggest the importance of appropriate care in the hours leading up to delivery. However, as evidenced by the current study, the causal continuum of stillbirth may extend beyond the delivery period and include more distal factors, such as a prior history of infant loss. To date, relatively few US studies have focused on the causal and risk factors associated with stillbirth.20,32 A notable exception is the recently published case–control study by Kothari et al.,20 which attempted to identify the maternal characteristics associated with fetal loss, as well as infant mortality. Prenatal and delivery data were abstracted from hospital records on a number of characteristics, including previous pregnancy outcomes. However, the authors performed multivariate analyses using a combined feto-infant mortality population (n = 66). Previous fetal or infant death approached significance as a predictor for subsequent feto-infant mortality.20 However, it is difficult to compare their findings with those presented here given the differences in the classification of exposure and outcome.

In their review of potential determinants of stillbirth, Silver et al.,32 noted that the ‘loss of a child’ may be a stressor that increases the occurrence of subsequent stillbirth. To the best of our knowledge, our study is the first to provide empirical evidence of an association between previous infant loss and subsequent stillbirth. Similarly, in their analysis using the same data set as used in the present study, Salihu et al.33 found that a history of perinatal mortality, stillbirth, or neonatal mortality was associated with a four- to five-fold increase in the risk of recurrence of perinatal mortality. However, the present investigation expands the knowledge base by broadening the exposure of interest beyond the perinatal period to include overall infant mortality, which underscores the uniqueness of the findings in this paper.

There are some potential limitations to note regarding the present study. First, as we used an existing data set, we were restricted in considering the role of variables potentially important in the association between infant mortality and subsequent stillbirth. However, the Missouri maternally linked cohort data files originate from a vital record system that is considered to be reliable, and has a documented validation process that has been described in previous studies.21,22 Second, we made an a priori decision to dichotomise all covariates included in multivariable analyses. Previous authors have cautioned against dichotomising continuous predictors in regression analyses.34,35 It is possible that our decision to dichotomise may have influenced our findings in an unintended manner. However, it is unlikely that this a priori decision impacted our study results, as a sensitivity analysis that excluded women with missing information for these variables yielded similar results (data not shown). Finally, an important shortcoming in our analysis is that we did not examine the accompanying causes of stillbirth and infant death. Although the lack of such data is a worldwide research hurdle,2,32 our investigation should be appropriately framed as an examination of risk markers rather than causal factors.

Despite these limitations, this study has significant strengths. As this study used a population-based data set, the resulting sample is large enough to offer substantial power for the detection of group variances in the analysis. Moreover, because of the large sample size, the results are minimally affected by selection biases and are reasonably generalisable. The subanalysis conducted by race/ethnicity yields important information on the black–white disparity regarding adverse birth outcomes. Given the persistent black–white disparity in infant mortality rates,36 the observation that black women who experienced infant mortality had an elevated risk for subsequent stillbirth reinforces the need for added public health efforts to reduce infant mortality rates within this subpopulation.

Our findings have potential implications for a number of stakeholder groups, including researchers, clinicians, and public health professionals. Following the death of a child, mothers and couples have questions regarding whether they can attempt another pregnancy (and when).37 Even for those without a history of loss, interconception care is becoming widely recognised as an important part of a comprehensive strategy to improve feto-infant outcomes.38–40 Our finding that women with a history of infant loss experience higher risk for subsequent stillbirth could serve as an important criterion in determining and targeting appropriate interconception strategies to improve future fetal survival. Clinicians should note the potential risk for subsequent stillbirth following infant mortality when conducting such discussions with patients. A survey of 804 US obstetricians revealed that about two-thirds of respondents endorsed a short interpregnancy interval (of <6 months) for mothers who experienced a perinatal death.41 This is a concern, as short birth intervals have been associated with adverse birth outcomes worldwide.42–46

Public health professionals also have an important role in efforts to prevent the occurrence (and recurrence) of stillbirth. For example, public health planners should note the disparities observed here between white and black mothers. Analyses stratified by race showed that black mothers who previously had experienced an infant death – regardless of whether it occurred during the neonatal or post-neonatal period – had an increased risk of stillbirth in their subsequent pregnancy. However, in order for public health professionals to plan interventions to address this disparity, there is also a need for researchers to strengthen the evidence base. Most critically, there is a need to elucidate the causal path linking infant mortality to subsequent stillbirth, and the findings in this paper should provide the impetus for such future research endeavours.

Disclosure of interests

None of the authors of this manuscript have conflicts of interest with this research.

Contribution to authorship

EMA conceived the idea, developed the study objectives, and led the preparation of the article, including writing and editing. HMS provided guidance regarding the study question, and helped with the writing up and editing of the article. HEW conducted the statistical analyses and helped with the writing up and editing of the article. BJB contributed to the writing up and editing of the article. AKM helped with the analysis, as well as the writing up and editing of the article. APA provided expert and intellectual guidance on the article.

Details of ethics approval

Prior to starting work, this study was approved by the Institutional Review Board at the University of South Florida (IRB approval letter reference: FWA # 00001669 of 24 June 2009).

Funding

This study was funded by a grant from the Flight Attendant Medical Research Institute (FAMRI 024008; principal investigator Hamisu M. Salihu, MD, PhD). The funding agency did not play any role in any aspect of the study.

Acknowledgements

None.

Ancillary