Although a variety of social, demographic and medical characteristics known at pregnancy confirmation are linked to stillbirth, they have relatively little clinical value for predicting early in the pregnancy which women will have stillbirths, according to a population-based study.1 The strongest predictors of stillbirth were pregnancy-related: having had a prior stillbirth (odds ratio, 5.9) and having a multifetal pregnancy (4.6). Women who were nulliparious, had diabetes, were black or were not living with a partner also had an elevated risk of stillbirth. However, these and other factors accounted for only 19% of the variance in birth outcomes.
To identify early, potentially modifiable risk factors for stillbirth (fetal death at 20 weeks’ gestation or later), as well as to explore possible causes of racial disparities in stillbirth, researchers examined pregnancies among women who gave birth at 59 hospitals in parts of Georgia, Massachusetts, Rhode Island, Texas and Utah between March 2006 and September 2008. All nonincarcerated women aged 13 or older who had a stillbirth (including any stillbirth in a multifetal pregnancy) were invited to enroll, as were a representative sample of those with live births. Births were classified as stillbirths if the fetus had Apgar scores of 0 after one and five minutes and had no visible signs of life. Because clinical estimates of gestational age are often imprecise, fetal deaths occurring at 18 or 19 weeks’ gestation were eligible for inclusion.
The investigators obtained information on maternal, fetal and pregnancy characteristics from in-hospital interviews with enrolled women, medical records, analysis of biological samples, pathological examination of the placenta and, in cases of stillbirth, postmortem examination of the fetus. Multivariate logistic regression analyses were conducted to identify characteristics present at the start of pregnancy that were associated with stillbirth; additional analyses were performed to identify associations with stillbirths at or after 24 weeks’ gestation, as well as stillbirths in low-risk pregnancies (i.e., singleton pregnancies of at least 24 weeks’ duration that did not involve a fetus with major congenital anomalies or fetal death during delivery).
About two-thirds of eligible women enrolled in the study; after exclusion of enrolled women who did not complete an interview or have adequate chart information, the analytic sample consisted of 614 pregnancies ending in stillbirth and 1,816 ending in live birth. Most women in the two groups were aged 20–34 (70–76%), were white (33–45%) or Hispanic (36% each), and were married (49–61%) or cohabiting (24–26%). In the three months prior to pregnancy, the majority of women had not smoked (81–87%) or consumed alcohol (58–60%). More than a fourth were pregnant for the first time (27–29%). Roughly half (47–58%) of participants were overweight, obese or morbidly obese, and 2–6% had diabetes. While nearly three in 10 had ever used illegal drugs (28–29%), only small proportions had been addicted (2–5%).
In multivariate analyses, the odds of stillbirth were greater among women aged 40 or older than among those aged 20–34 (odds ratio, 2.4); among blacks than among whites (2.1); among unmarried, noncohabiting women than among married ones (1.6); among women who had smoked in the months before pregnancy than among nonsmokers (1.6); and among women who had ever been addicted to illegal drugs than among those who had never used such substances (2.1). The risk of stillbirth was also elevated among women who had a multifetal pregnancy (4.6); were overweight, obese or morbidly obese, rather than normal-weight (1.4–1.7); had blood type AB, rather than type O (2.0); or had diabetes (2.5). Women who had had a previous stillbirth were more likely than those who had had only live births to have a stillbirth (5.9), as were nulliparous women, regardless of pregnancy history (2.0–3.1). The generalized R2 for the multivariate model was 0.19, indicating that the included characteristics explained little of the burden of stillbirth.
In analyses that excluded deliveries before 24 weeks’ gestation, stillbirth remained strongly associated with nulliparity and history of stillbirth; noncohabitation, smoking and diabetes also remained predictors of stillbirth, but race did not. Analyses focusing on low-risk pregnancies found that characteristics associated with stillbirth in this population were nulliparity, prior stillbirth, noncohabitation, being overweight or obese, and diabetes. R2 values were again low (0.14–0.16).
In a separate study of the same sample,2 researchers attempted to determine the likely causes of death of stillborn infants; they obtained permission to conduct complete postmortem examination of 512 cases. Half of the stillbirths had occurred before the 28th week of gestation, and a third before the 24th week.
The investigators identified a probable cause of death in 61% of cases, and at least one possible cause in 76%. The most common causes were obstetric complications (29% of cases, including all intrapartum deaths) and placental abnormalities (24%); less common causes were genetic or structural defects of the fetus (14%), infection (13%), umbilical cord abnormalities (10%), hypertensive disorders (9%) and maternal medical complications (8%). Blacks were more likely than white and Hispanic women to have stillbirths associated with obstetric complications (44% vs. 22–25%) and infections (25% vs. 7–9%), while cord abnormalities occurred more often among whites and Hispanics (13% each) than among blacks and other women (4–5%). Causes of death did not differ by race in births after 23 weeks’ gestation.
Although the new analysis was the largest population-based study to date of predictors and causes of stillbirth, it lacked the statistical power to identify rare risk factors, the investigators note; moreover, the findings may have been affected by nonparticipation and missing data. Many characteristics associated with stillbirth, such as race and pregnancy history, are not modifiable, but some may provide avenues for clinical intervention or research. The researchers single out maternal overweight as the modifiable characteristic with the greatest potential population-level impact on preventing stillbirth. The elevated stillbirth risk among unmarried, noncohabiting women suggests that screening for relationship status may help identify women "who would benefit from social support during pregnancy." Finally, because the racial disparity in stillbirth risk was no longer apparent when analyses were restricted to deliveries after 23 weeks’ gestation, the investigators note that research "focusing on the pathophysiology of early preterm birth" may yield findings that can help "reduce racial disparity in stillbirth."—P. Doskoch