These analyses included subjects with estimated dates of delivery (EDD) from January 2006 to December 2009 who participated in NBDPS; questions regarding stress and social support were not asked for earlier study years. Detailed study methods have been published.²³ In brief, eight states included liveborn, stillborn (fetal deaths ≥20 weeks gestation), and prenatally diagnosed and electively terminated cases (AR, CA, GA, IA, NC, NY, TX, UT) and one state included only liveborn and stillborn cases (MA).

Cases included infants or fetuses with neural tube defects (NTDs, including anencephaly, spina bifida, encephalocele and craniorachischisis); orofacial clefts (cleft lip with or without cleft palate, CLP, or cleft palate alone, CP); or the conotruncal heart defects tetralogy of Fallot (TOF) or d-transposition of the great arteries (dTGA), as confirmed by clinical, surgical, or autopsy reports. Conotruncal heart defects required verification by echocardiography, cardiac catheterization, surgical report, or autopsy. Cases resulting from known single gene or chromosomal abnormalities (syndromic cases) were ineligible, given their presumed genetic determinants. Infants whose clefts were believed to be secondary to another defect (e.g., holoprosencephaly) were ineligible for the study.

Each participating center randomly selected approximately 100 liveborn controls without birth defects per study year from birth certificates (AR, GA, IA, MA, NC, UT) or birth hospitals (CA, NY, TX) to represent the population from which cases were derived.

Maternal interviews were conducted using a standardized, computer-based questionnaire, primarily by telephone, in English or Spanish, no earlier than six weeks after the infant’s EDD and no later than 24 months after the EDD. Interviews were conducted with mothers of 2532 cases (63% participation rate) and 3351 controls (60% participation rate) from the 2006–2009 cohort. Median time from actual date of delivery to interview was 10.5 months for cases (interquartile range from 6.9 to 15.0 months) and 7.8 months for controls (interquartile range 5.3 to 12.2 months).

Mothers were asked yes/no questions about whether they experienced five life events during the three months before or first three months of pregnancy (relationship difficulties; legal/financial problems; violence/crime; illness/injury; or a relative’s death). In a separate section of the interview, they reported their employment and residence history during that time period; we considered a change in employment or residence as two additional stressful life events, bringing the total to seven. Women were asked three questions about social support (whether they could count on someone for emotional support, financial help, and help with daily tasks), all applicable to the same periconceptional time period as the stress questions. (Exact questions are in Supplementary Table 1.) Covariates were maternal race-ethnicity; age, education; prepregnancy body mass index; smoking or alcohol drinking in the month before or first trimester of pregnancy; and intake of folic acid-containing vitamin/mineral supplements. Complete data were available for 2974 controls and 2244 cases (552 NTDs, 413 CP, 797 CLP, 189 dTGA, 311 TOF; 18 cases had multiple phenotypes).

Stress and social support questions were examined individually and summed to create two indices. Multivariable logistic regression analyses were conducted to estimate odds ratios and 95 percent confidence intervals reflecting the association of each stressful life event and social support question, as well as the indices, with each birth defect. Results for anencephaly and spina bifida groups were similar so they are presented together. We examined the stress index in categorical and continuous (ordinal) form. We also examined stress and social support in combination, dichotomizing the stress index as 0–3 versus 4–7 and the social support index as 0–2 versus 3 to reflect ‘high’ or ‘low’ stress or social support. Analyses were adjusted for the covariates described above.

Most mothers of controls were non-Hispanic white (58%), between the ages of 25 and 34 years (55%), and had more than a high school education (63%); 21% were obese; and 17% smoked, 38% drank alcohol, and 89% took folic acid-containing supplements periconceptionally (Table 1). Each stressful life event was reported by 7–26% of control mothers, and 60% of all control mothers reported yes to at least one of the stressful life event questions. A ‘yes’ response was reported by 85–87% of control mothers for each of the social support questions, and 75% reported ‘yes’ to all three. Percentages were similar for case mothers.

Most of the adjusted odds ratios for individual life events and the specific birth defects under study were approximately 1.0, with only a few being elevated and having 95% confidence intervals that did not include 1.0 (relationship difficulties (NTDs), legal problems (CLP or ToF)) (Supplementary Table 1). Associations with the stress index tended to be higher with higher scores, although few 95% confidence intervals excluded one (Table 2). A 4-point increase in the index was moderately associated with NTDs (OR 1.5, 95% CI 1.1, 2.0) and CLP (OR 1.3, 95% CI 1.0–1.7). Associations with social support questions and the social support index tended to be associated with reduced risk but most 95% CIs included one, with the exception of dTGA (Table 2).

Analyses examining the association of stress and social support in combination indicated that the largest odds ratios were for women with high stress and low support (versus low stress and high support), for all of the studied phenotypes except TOF (Supplementary Table 3). However, even these odds ratios were modest and only the confidence interval for CLP excluded one (OR 1.6, 95% CI 1.01, 2.4).

In this study, maternal periconceptional stressful life events, social support, and the two factors in combination were at most modestly, if at all, associated with risks of NTDs, orofacial clefts or conotruncal heart defects. There was some suggestion of increased risk for NTDs and CLP for mothers who reported stressful life events. Social support was associated with reduced risks, most notably for dTGA.

Previous studies have suggested that increased stress during pregnancy is associated with increased risks of birth defects.^1–14 These assessments of stress has tended to be less comprehensive, however, including for example a few specific life events,^{8,10,11,13,14} dichotomies of whether the mother was stressed or not (sometimes with uncertain methods of assessment),^2,4–7,9 or prevalence before versus after a natural disaster.^3,12 Most recently, a data linkage-based study from Denmark indicated higher risk of orofacial clefts among offspring born to women who experienced the death of a close relative during early pregnancy,¹¹ and another study indicated that the prevalence of clefts was higher after versus before Hurricane Katrina in the New Orleans area.¹² A previous study of California births, which included the same phenotypes as those included in the current study, included the most detailed assessment of stress to date, with data on 18 stressful life events. All of the studied phenotypes were associated with increased number of stressful life events.¹ For NTDs, these associations were particularly strong among women who did not take folic acid-containing supplements. The explanation for weaker associations in the current study is uncertain. One potential explanation is that the current study included fewer stressful life event questions than the previous California study (seven items versus 18). We did observe a tendency toward higher risks with higher number of stressful life events for most of the studied outcomes, but the results were imprecise.

We are aware of two other studies that examined social support and birth defects. One focused on social network size;²⁰ the other examined social networks and emotional support.¹³ Both studies focused on NTDs and reported reduced risks. Associations in our study were in the same direction but only confidence intervals for dTGA excluded one.

When examining potential influences in combination, our results suggested that women with highest stress and lowest support had the largest risk estimates for most of the studied phenotypes, but again associations were modest and imprecise. Suarez et al. reported that social support did not modify the association with stress in their study of NTDs.¹³ More detailed assessment of social support will be useful to confirm its potential role in birth defects etiology and in particular whether it can reduce the potential negative impact of stress.

Strengths of the current study include its population-based, multi-center design, careful case ascertainment, large size, examination of stress and social support in combination, and adjustment for several potential confounders. The general lack of association observed could be influenced by the relatively limited assessment of stress and social support. The assessment of stress included only seven questions, but this does represent more detail than most previous studies. In addition, the 6-month window of exposure does not allow separate investigation of acute versus chronic stressors. Responses to the questions about life events were limited to yes/no rather than having women rate the stressfulness of the events, in order to maximize ability to recall events objectively (job and address change were assessed via more detailed questions). Interviews were conducted with mothers of 63% of eligible cases and 60% of controls. We have no particular reason to believe that participation was related to periconceptional stress or social support, differentially for cases versus controls, but if it was, then selection bias may have affected our results. Similarly, we do not know whether mothers’ recall of stress and social support was related to case or control status, but if it was, then recall bias could have affected our results. We adjusted for several potential covariates, but other potential confounders were not evaluated. For example, we did not adjust for medications that may be used to cope with stress (e.g., SSRIs, benzodiazapines); however, given that the associations with stress that we observed are modest at best, and use of these medications tends to be rare during pregnancy, it is unlikely that adjustment for use of these medications would make a substantive difference.

The negative impact of stress on risks of birth defects has been studied much less than other reproductive outcomes such as preterm birth.²⁴ It is important to continue to strive toward a better understanding of stress as a risk factor for birth defects, perhaps through a more in-depth assessment of stress and social support than is currently included in the NBDPS.