Maternal stressful life events during pregnancy and childhood asthma and wheeze

Abstract

Background:

Studies have linked prenatal maternal psychosocial stress to childhood wheeze/asthma but have rarely investigated factors that may mitigate risks.

Objective:

To investigate associations between prenatal stress and childhood wheeze/asthma, examining factors that may modify stress effects.

Methods:

Participants included 2056 mother-child dyads from ECHO PATHWAYS, a consortium of three prospective pregnancy cohorts (CANDLE, TIDES, and GAPPS) from six cities. Maternal stressful life events experienced during pregnancy (PSLEs) were reported using the Pregnancy Risk Assessment Monitoring System SLE questionnaire. Parents reported child wheeze/asthma outcomes at age 4-6 years using standardized questionnaires. We defined outcomes as ever asthma, current wheeze, current asthma, and strict asthma. We used modified Poisson regression with robust standard errors to estimate risk ratios (RR) and 95% confidence intervals (95%CI) per one-unit increase in PSLE, adjusting for confounders. We examined effect modification by child sex, maternal history of asthma, maternal childhood traumatic life events, neighborhood-level resources, and breastfeeding.

Results:

Overall, we observed significantly elevated risk for current wheeze with increasing PSLE (RR:1.09 (95%CI:1.03,1.14)), but not for other outcomes. We observed significant effect modification by child sex for strict asthma (p-interaction=0.03), where risks were elevated in boys (RR:1.10 (95%CI:1.02,1.19)) but not girls. For all other outcomes, risks were significantly elevated in boys and not girls, although there was not statistically significant evidence of effect modification. We observed no evidence of effect modification by other factors (p-interactions>0.05).

Conclusion:

Risk of adverse childhood respiratory outcomes is higher with increasing maternal PSLEs, particularly in boys.

Introduction

Asthma is a common chronic respiratory condition that usually begins in early childhood and is associated with substantial social and economic burden.¹ Risk factors for asthma are complex and include both genetics as well as numerous environmental exposures during prenatal and early postnatal development that likely act through diverse mechansims.^2–4 In addition to maternal prenatal exposures to certain pollutants and nutrients,^5–9 maternal psychosocial stress during pregnancy has emerged as a potential contributor to asthma development.¹⁰ Respiratory and immune system development begin in utero and are guided, in part, by neural and endocrine system interactions that may be susceptible to stress responses during pregnancy.^11,12 Stress induces activation of the hypothalamic-pituitary-adrenal (HPA) axis, and consequently, promotes cortisol production. Leading hypotheses suggest that, during pregnancy, excess maternal cortisol can influence placental function as well as fetal HPA axis programming, leading to a T-helper 2 dominant immune response, consistent with asthma and atopic disease phenotypes.^12,13 Maternal cortisol levels have been linked to childhood respiratory outcomes, particularly when accompanied by other asthma risk factors.^14,15

Stress during pregnancy can be measured in multiple ways, ranging from perceived distress (i.e., sense of being overwhelmed, sad, or other stress correlates) to more objective measures such as exposure to stressful or negative life events. Stressful life events (SLEs), including major events related to finances, relationships, health, or legal problems, are common among pregnant women. Recent studies using U.S. national survey data spanning years 2016 to 2019 estimate that over 40% of women experience 1 or 2 SLEs in the year prior to childbirth, and over 25% experience 3 or more.^16–18 Prior literature links exposure to SLEs during pregnancy with adverse outcomes including preterm birth, child mental health problems, and maternal depression.^19–22 As such, SLE screening is increasingly recognized as an important part of comprehensive perinatal care, supporting the health of both women and infants.^23–25

While the association between prenatal maternal stress and child/offspring asthma and wheeze has been supported across multiple studies²⁶ and characterized in several meta-analyses,^10,27,28 the association between SLEs, specifically, and asthma and wheeze is inconsistent. For instance, Flanigan et al.,²⁷ estimated an elevated pooled risk for childhood asthma of 1.13 (95% confidence interval (CI): 1.03, 1.24) across 10 studies measuring maternal stress (any versus none) as anxiety, depression, bereavement, or negative life events. However, of the four asthma studies that measured negative life events specifically (sample sizes ranging from 68 to 1193),^29–32 pooled risk estimates were null for childhood asthma (1.08 (95% CI: 0.87, 1.33)) and borderline for childhood wheeze (1.23 (95% CI: 1.00, 1.52)).^27,31–34 When interpreting these findings, it is also important to consider possible heterogeneous effects of pregnancy stress on offspring health, as numerous individual, family, and environmental factors may exacerbate or buffer the effects of stress exposure on children’s respiratory conditions. For example, Lee et al. observed significant associations between prenatal negative life events and childhood asthma in boys, but not girls.²⁹ Hartwig et al., observed associations between negative life events and childhood asthma only in children of women without a history of asthma.³⁰ Data gaps persist for potentially modifiable factors, including positive health behaviors such as breastfeeding or environmental factors such as neighborhood resources and conditions, thus limiting our understanding of susceptible subgroups or possible interventions. The ECHO-PATHWAYS consortium was funded to harmonize data from three large pregnancy cohorts and support the examination of important hypotheses around developmental origins of disease and related effect modifiers, such as child biological sex. Using the rich data of this consortium, we investigated the association between maternal SLEs during pregnancy and childhood asthma and wheeze outcomes under the a priori hypothesis that the risk of asthma and wheeze increases with increasing maternal SLEs. We capitalize on the consortium’s large, multi-region sample to examine potential effect modifiers of this association including individual and environmental factors that may modify or buffer the adverse effects of maternal prenatal stress, including child sex, maternal history of asthma, maternal history of exposure to childhood trauma, postnatal neighborhood resources and conditions, and exclusive breastfeeding duration. Accordingly, we hypothesize that the association between maternal stress and childhood asthma and wheeze will be strongest among male children,^26,29 in children without maternal history of asthma,³⁰ and in children of women who experienced childhood trauma;³⁵ we hypothesize that the association between maternal stress and childhood asthma and wheeze will decrease with increasing neighborhood resources^36,37 and breastfeeding duration.³⁸

Methods

Study Population

We studied women and children enrolled in the NIH ECHO-PATHWAYS consortium, which is comprised of three prospective cohorts: the Conditions Affecting Neurocognitive Development and Learning in Early Childhood (CANDLE) study, The Infant Development and Environment Study (TIDES) and a subset of the Global Alliance to Prevent Prematurity and Stillbirth study (GAPPS-PW).³⁹ The CANDLE study included eligible women from Shelby County (Memphis), TN, enrolled during the second trimester of pregnancy (16 – 28 weeks). Women were eligible for inclusion if they were between 16 and 40 years of age, able to speak English and had a low medical risk, singleton pregnancy.⁴⁰ A total of 1503 women enrolled in the study and 1157 mother-child dyads participated in the 4 year follow-up study visit.^41–44 TIDES recruited women from four sites: University of Minnesota Medical Center (Minneapolis, Minnesota); University of California—San Francisco Medical Center (San Francisco, California); University of Rochester Medical Center (Rochester, New York); and University of Washington (Seattle, Washington). Eligible women were in their first trimester (<13 weeks) of a non-medically threatened, singleton pregnancy at the time of recruitment, aged ≥18 years and able to read and write English. TIDES includes 749 live births (2010-2012), 544 of whom participated in follow-up at approximately age 4 years. Finally, the GAPPS-PW study includes women originally enrolled in the parent GAPPS study, a cohort established in 2007 to collect biospecimens during pregnancy. Eligible women were ≥18 years of age, able to speak English, and had plans to deliver at the hospital of recruitment. The PATHWAYS follow-up study (GAPPS-PW) enrolled women and their children if they were part of a GAPPS Washington study site (Seattle, WA or Yakima, WA), gave consent for future contact, provided at least one urine sample during pregnancy and had a GAPPS child 4-8 years of age at recontact. Recruitment into GAPPS-PW is ongoing.

For this analysis, we included mother-child dyads who participated in 4-6 year child visits (child age 7.5 years or younger) and delivered at ≥32 estimated gestational weeks because infants born prior to 32 weeks gestation are at increased risk for serious complications and lung diseases of prematurity, such as bronchopulmonary dysplasia.^38,45 All adult participants provided informed consent for themselves and their children. This current study was approved by the Institutional Review Board of the primary institution.

Stressful Life Events during Pregnancy

Our primary exposure of interest was maternal exposure to stressful life events during pregnancy (PSLE). Women reported PSLEs using a questionnaire that was adapted from the Centers for Disease Control and Prevention Pregnancy Risk Assessment Monitoring System (PRAMS) survey. The questionnaire characterized 14 possible PSLEs related to illness, death, relationship problems, housing difficulties, legal issues and financial problems experienced during pregnancy (eSupplement).^46,47 Affirmative responses were summed into the PSLE score (possible range: 0-14). Women reported PSLEs retrospectively at child age 8, 6, or 4-6 years for CANDLE, TIDES, and GAPPS-PW, respectively.²¹ Recall of stressful life events, particularly those associated with trauma, has been shown to be relatively stable over the span of years.^48,49

Childhood Respiratory Outcomes

We assessed childhood wheeze and asthma between ages 4-6 years using the International Study of Asthma and Allergies in Childhood (ISAAC) questionnaire⁵⁰, a widely used and validated instrument for assessing asthma in childhood,^51–53 and additional questions regarding asthma-specific medication use and diagnoses, as previously described.³⁸ We defined current wheeze as parent report of a child experiencing any wheeze or whistling in the chest in the last 12 months; ever asthma as parent report of a child ever having asthma; and current asthma as report of any two of the following: 1) current wheeze, 2) ever asthma, or 3) asthma-specific medication use in the previous 12 months (CANDLE, GAPPS-PW) (yes/no) or 24-months (TIDES, based on free text reporting). We defined strict asthma as a child having ever asthma and either current wheeze or asthma-specific medication use (or both) to identify children with asthma who were also having current symptoms. Asthma diagnosis alone is a highly specific, but only modestly sensitive metric for identifying asthma disease; characterization of symptoms improves the identification of disease.⁵⁴ Therefore, we use a combination of symptoms and diagnosis to characterize our outcomes.

Additional Variables

Variables used as covariates and effect modifiers were collected from participant surveys, completed at both prenatal and postnatal visits, and were harmonized by the ECHO-PATHWAYS consortium. These variables included maternal age (years), race (White, Black, Asian, Multiple and Other [including Native Hawaiian/Pacific Islander, American Indian/Alaskan Native, and Other]), ethnicity (Hispanic, Non-Hispanic), education level (less than high school, high school/GED, college degree or technical school, and graduate or professional degree), parity (0, ≥1), smoking during pregnancy (yes/no); pre-pregnancy body mass index (BMI) (kg/m³), maternal history of asthma (yes/no), duration of exclusive breastfeeding (none to <2 months; 2-4 months; 5-6 months, or > 6 months) and household income at follow-up (adjusted for region and inflation). We characterized self-reported maternal exposure to childhood trauma using three items from the Traumatic Life Events Questionnaire (TLEQ), specifically assessing whether physical, sexual or verbal abuse was experienced before adulthood (range: 0-3).⁵⁵ Child characteristics assessed at delivery include child sex (male, female) and gestational age at birth (days). As a measure of postnatal neighborhood conditions and resources, we linked geocoded childhood residence indicators (census tract) from birth to age 4 (weighting scores by the proportion of time spent at each address to account for residential mobility) to the indicators of the Childhood Opportunity Index (COI). The COI is a publicly-available index that provides an overall “opportunity” score ranging from 0 to 100, based on 29 objective indicators covering education, health and environment, and social and economic domains using data derived from nationally representative surveys, estimated for 2010 and 2015.⁵⁶ Higher scores indicate more neighborhood opportunity. We used 2010 COI data for children who reached age 4 before 2013 and 2015 COI data for children who reached age 4 after 2013.

Statistical Analysis

We summarized maternal and child descriptive statistics using means (+/− standard deviations), medians (interquartile range [IQR]) and proportions, as appropriate. We used modified Poisson regression with robust standard errors (SE) to estimate relative risks (RR) and 95% confidence intervals (95%CI) for each respiratory outcome (current wheeze, ever asthma, current asthma and strict asthma) per one-unit (i.e., event) increase in PSLE, adjusting for potential confounders. Relative risks with 95% CIs that excluded the null (1.0) were considered statistically significant. Robust standard errors are needed when using a Poisson generalized linear model to estimate risk ratios from binary outcome data and provide robustness to deviation from log-linearity in the mean model. We imputed missing data using multiple imputation by chained equations (MICE package 3.14.0, R) and present pooled results from the 10 imputed datasets. The proportion of missingness was highest for PSLEs (15.6%) and COI (6.6%); missingness was less than 4% for all other covariates (footnote, Table 1). Missingness for PSLE information was due, in part, to loss of participants between outcome assessment at 4-6 years and PSLE assessment at later follow-up. We adjusted all models for maternal age, race, ethnicity, education, history of asthma, parity, and prepregnancy BMI; child sex and age at assessment; and study site. We included maternal race as a confounder on the basis that it is a proxy for the impact of racist practices and structural inequality, acknowledging that it is political and social, not biological, construct.⁵⁷ In a series of sensitivity analyses, we additionally adjusted for gestational age, prenatal smoking, duration of exclusive breastfeeding and household-size-adjusted income, one at a time, under the premise that these factors may be the result of PSLEs and thus on the causal pathway between PSLEs and child outcomes. We explored possible effect modification for covariates selected a priori based on their hypothesized role as modifiers (child sex, maternal history of asthma) or important asthma risk (or protective) factors in prior studies (maternal childhood trauma score, breastfeeding duration and postnatal COI score).^{35–38,42,58,59} Here, we calculated interaction p-values based on robust SEs and estimated relative risks within strata of each covariate using product terms added to fully adjusted models. Finally, to assess the robustness of our pooled findings to the influence of any one cohort, we performed a series of sensitivity analyses in which each cohort was omitted from the analysis (i.e., leave-one-out). We used R version 4.1.3 for all analyses; we considered statistical tests significant at the level of 0.05, but draw conclusions based on the totality of significant and not significant results. As such, we did not adopt a formal adjustment for multiple comparisons.

Table 1.

Descriptive characteristics of eligible dyads in ECHO-Pathways Consortium (born 2006–2015)

	Total N = 2056	CANDLE N = 1122	GAPPS N =402	TIDES N = 532
Maternal Characteristics
Age, years; mean (SD)1	28.4 (5.9)	26.3 (5.5)	30.7 (5.5)	31.0 (5.3)
Race, n (%)1
Asian	52 (3)	9 (<1)	11 (3)	32 (6)
Black	764 (38)	700 (62)	9 (2)	55 (11)
White	1053 (52)	355 (32)	318 (82)	380 (72)
Other	57 (3)	3 (<1)	14 (4)	40 (8)
Multiple	107 (5)	55 (5)	34 (9)	18 (3)
Hispanic Ethnicity, n (%)1	126 (6)	21 (2)	59 (15)	46 (9)
Education at enroll, n (%)1
< High School	172 (8)	125 (11)	15 (4)	32 (6)
High School/GED	710 (35)	524 (47)	103 (27)	83 (16)
College or Technical	672 (33)	338 (30)	169 (43)	165 (31)
Grad/Professional	484 (24)	134 (12)	102 (26)	248 (47)
Adjusted Income, $1000s, median (IQR)1,2	57.0 (23.9 -86.0)	31.8 (13.6 – 74.2)	86.0 (49.1 - 122)	111 (56.0 – 155)
Maternal Asthma, n (%)1	335 (18)	198 (18)	73 (18)	84 (17)
Firstborn, n (%)1	852 (42)	446 (40)	129 (32)	277 (53)
Prenatal Smoking, n (%)1	142 (7)	102 (9)	12 (3)	28 (5)
Pre-pregnancy BMI, median (IQR)1	25 (22 – 31)	26.0(22 – 32)	25 (22 – 30)	24 (21 – 28)
Exclusive Breastfeeding, n (%)1
None - < 2 months	816 (41)	600 (54)	106 (27)	110 (23)
2–4 months	370 (19)	227 (20)	74 (19)	69 (15)
5–6 months	432 (22)	160 (14)	112 (28)	160 (34)
More than 6	366 (18)	127 (11)	104 (26)	135 (28)
Maternal childhood traumatic events, n (%) 1
0	1274 (64)	689 (63)	261 (65)	324 (67)
1	452 (23)	279 (25)	75 (19)	98 (20)
2	188 (9)	98 (9)	46 (11)	44 (9)
3	71 (4)	34 (3)	20 (5)	17 (4)
Pregnancy Stressful Life Events (PSLE)1
Median (IQR)	1 (0 – 2)	1 (0 – 3)	1 (0 – 2)	1 (0 – 2)
Count (n (%))
0	660 (38)	303 (36)	148 (37)	209 (43)
1	416 (24)	189 (22)	105 (26)	122 (25)
2	269 (16)	124 (15)	71 (18)	74 (15)
3	166 (10)	91 (11)	32 (8)	43 (9)
4+	224 (13)	135 (16)	46 (11)	43 (9)
Child Characteristics
Male Sex, n (%)	1023 (50)	560 (50)	208 (52)	255 (48)
Gestational age at birth, mean weeks (SD)	39.1 (1.62)	39.0 (1.42)	38.8 (1.96)	39.3 (1.72)
Age at Assessment (years), mean (SD)	4.7 (0.7)	4.4 (0.6)	5.6 (0.7)	4.5 (0.3)
Child Opportunity Index, median (IQR)1	48.2 (11.0 – 77.2)	20.2 (6.0 – 56.0)	64.8 (48.8 – 75.4)	79.0 (47.9 – 92.0)
Current Wheeze 1	309 (15)	212 (19)	46 (12)	51 (10)
Ever Asthma 1	232 (11)	162 (15)	30 (8)	40 (8)
Current Asthma 1	239 (12)	177 (16)	33 (8)	29 (5)
Strict Asthma 1	183 (9)	132 (12)	22 (6)	29 (6)

¹missing: maternal age n = 16 (0.8%); maternal race n = 23 (1.1%); maternal ethnicity n = 9 (0.4%); maternal education n = 18 (0.9%); maternal asthma n = 59 (2.9%); smoking n = 10 (0.5%); firstborn n = 10 (0.5%); pre-pregnancy BMI n = 48 (2.3%); breastfeeding duration n = 72 (3.5%); early traumatic events n = 71 (3.5%); pregnancy stressful life events (PSLE) n = 321 (15.6%); child opportunity index n = 136 (6.6%); current wheeze n = 14 (0.7%); ever asthma n = 20 (1%); strict asthma n = 41 (2%).

²Region and inflation-adjusted household income. SD = standard deviation; IQR = interquartile range

Results

This study included 2,056 mother-child dyads (CANDLE: 1,122 [55%]; GAPPS-PW: 402 [19%]; and TIDES: 532 [26%]; Table 1). Across all cohorts, women averaged 28.4 years (+/− 5.9) of age at enrollment; 52% identified as White, 38% identified as Black, 3% identified as Asian, 3% identified as Native Hawaiian/Pacific Islander, American Indian/Native American or Other (combined as “Other”) and 5% identified as multiple races. Most women were non-Hispanic (94%). For 43% of women, the highest level of educational attainment was high school degree or less. Comparing across cohorts, women participating in CANDLE were younger and more likely to be Black than those in GAPPS or TIDES; TIDES included a higher proportion of women with graduate/professional level education and women who were multiparous. Children were a mean of 4.7 years (+/− 0.7) at assessment; half of children were male. The majority of children in the combined cohort were exclusively breastfed for two months or more (59%). The median COI score was 48.2 (IQR: 11.0, 77.2), with cohort-specific medians ranging from 20.2 (IQR: 6.0, 56.0) in CANDLE to 79.0 (IQR: 47.9, 92.0) in TIDES.

Women reported a median of 1 PSLE, with 38% reporting 0 events, 24% reporting one event, 16% reporting two events, and 22% reporting three or more. The most commonly reported PSLEs included moving to a new address (28%), arguing with a partner more than usual (19%), illness in a close family member (16%), and difficulty paying bills (16%) during pregnancy. Parents reported current wheeze in 15% of children, ever asthma in 11%, current asthma in 12% and strict asthma in 9%.

We estimated a modest increase in risk with increasing number of reported PSLEs across all outcomes (Table 2), with a statistically significant increase in risk of current wheeze (adjusted(a) RR: 1.08 [95%CI = 1.02, 1.13]). Sensitivity analyses that incorporated additional adjustment for gestational age, prenatal smoking, duration of exclusive breastfeeding and income did not change our results appreciably (not shown). Further, results were largely unchanged in leave-one-cohort-out sensitivity analyses (eFigure 1). Results were somewhat attenuated when we limited analysis to subjects with complete data (n = 1529), although they remain in the positive direction consistent with increased risk (eTable 1).

Table 2:

Risk ratios and 95% confidence intervals for the association between stressful life events during pregnancy (PSLE) and childhood respiratory outcomes (n = 2056)

	Risk Ratio (95% Confidence Interval)
	Unadjusted	Adjusted 1
Current Wheeze	1.14 (1.08, 1.19)	1.08 (1.03, 1.13)
Ever Asthma	1.13 (1.06, 1.20)	1.06 (1.00, 1.13)
Current Asthma	1.11 (1.05, 1.18)	1.04 (0.98, 1.11)
Strict Asthma	1.12 (1.04, 1.20)	1.04 (0.97, 1.12)

¹Adjusted for maternal age, race, ethnicity, education, history of asthma, parity, and pre-pregnancy body mass index (BMI); child sex and age at assessment; and study site.

The association between each respiratory outcome and maternal PSLEs was stronger for male children than female children (Figure 1). Modification by sex was significant for strict asthma (aRR_male = 1.10 [95%CI = 1.02, 1.19]; aRR_female = 0.92 [95%CI = 0.80, 1.07]; p-interaction = 0.03) and approached significance for current asthma (aRR_male = 1.09 [95%CI = 1.01, 1.17]; aRR_female = 0.95 [95%CI = 0.85, 1.07]; p-interaction = 0.07). For all outcomes, associations in males were significantly elevated. We did not observe any difference in association between PSLE and respiratory outcomes when assessing effect modification by maternal history of asthma, maternal childhood traumatic life events, duration of exclusive breastfeeding, or postnatal COI (eTable 2 and eFigure 2).

Figure 1.

Risk of respiratory outcomes per stressful life event experienced during pregnancy, by child sex.

* P-interaction for strict asthma = 0.03. P-interaction values for current wheeze, ever asthma and current asthma are 0.15, 0.34 and 0.07, respectively. Risk ratios are adjusted for maternal age, race, ethnicity, education, history of asthma, parity, and pre-pregnancy BMI; child sex and age at assessment; and study site

Discussion

In this large, multi-cohort observational study, we observed significantly elevated risk of childhood current wheeze with increasing number of SLEs experienced by women during pregnancy, corresponding to an 8% increase in risk per additional PSLE. Current wheeze is a symptoms-based outcome that may include not only children with diagnosed uncontrolled asthma, but also those with undiagnosed asthma (due to mild illness or limited clinical care), or transient wheeze. We observed a similar, but not statistically significant, association between maternal PSLEs and children ever having asthma. These findings are broadly consistent with prior literature demonstrating maternal stress during pregnancy as a risk factor for adverse childhood respiratory outcomes, and negative or stressful life events, specifically, as risk factors for childhood wheeze.^{10,11,27,28,60} While we did not observe an overall association between PSLEs and the outcomes current asthma and strict asthma, we did observe interactions between PSLEs and child sex for these outcomes, such that the associations between PSLEs and current and strict asthma were evident among boys but not girls. We did not observe evidence of effect modification by any of the other factors we examined, including maternal history of asthma, maternal childhood traumatic life events, duration of exclusive breastfeeding, or postnatal neighborhood resources.

The plausibility of a sex-specific response to prenatal stress is supported by well-established differences in male and female fetal development and placental function.^61,62 It is theorized that male and female placentas incorporate different strategies to adapt to maternal stressors,⁶³ and multiple studies have commented on the importance of reporting sex-specific findings.^64–66 Accordingly, several prior epidemiologic studies have reported sex-specific associations between maternal stress and asthma and related outcomes. However, the directions of associations found in these prior studies are inconsistent. Similar to our study, in a study of maternal stress and severe infections during childhood, infection-related hospitalizations including upper and lower respiratory infections were shown to be associated with prenatal exposure to SLEs in boys, but not girls.⁵⁸ A prospective U.S. study of prenatal and postnatal exposure to negative life events and childhood asthma by age 6 reported that boys had increased vulnerability to prenatal stress, whereas girls were more vulnerable to postnatal or cumulative (pre-and postnatal) stress.²⁹ A similar finding was reported in a study of pregnancies in Mexico, where prenatal negative life events were associated with ever wheeze in boys and postnatal negative life events were associated with ever and current wheeze in girls by age 4.³⁴ In a large registry-based study of maternal bereavement, boys whose mothers experienced the death of a close relative, particularly during the second trimester, had an increased risk of asthma at age 1-4 years, while no association was observed in girls.⁶⁷ In contrast, other studies have reported no modification by sex in the association between prenatal stress and childhood asthma or wheeze,³⁴ or a lack of any overall or sex-stratified association.⁶⁸ Other studies, including a small observational study of 68 children at age 12,³¹ as well as studies in prenatally stressed adult mice,^69,70 suggest maternal stress effects on lung development in females but not males. A potential driver of these opposing findings may be onset of puberty; at younger ages, asthma is more common among boys than girls, but shifts to becoming more prevalent among girls post-pruberty.⁷¹ Our study population is pre-pubertal so additional longitudinal follow up is needed to further examine whether reproductive life-stage influences observed associations.

We did not observe evidence of modification by the other potential modifiers examined in this study. In contrast to Hartwig et al.,³⁰ PSLEs were not more strongly associated with outcomes in children without maternal asthma history. Likewise, there was no difference in association by maternal history of exposure to the types of childhood traumatic events measured in this study. Prior work in ECHO-PATHWAYS reported a strong protective association between the duration of exclusive breastfeeding and asthma and wheeze outcomes.³⁸ In this analysis, however, we did not observe evidence that breastfeeding duration buffers PSLE associations. Similarly, ample evidence suggests that the quality of neighborhood resources and conditions, as measured by social, economic and environmental indicators, can impact asthma prevalence rates and severity.^72–75 In this study, COI score did not modify the association between PSLEs and study outcomes. In general, COI scores are strongly predictive of child asthma,⁵⁶ but are heavily weighted towards the social and economic domain (e.g., poverty rate, foreclosure rate), and less so towards the health and environment (e.g., proximity to parks, healthy food retailers) or education domains that might be more impactful in buffering maternal stress effects on offspring health; further, the COI does not capture neighborhood social factors such as cohesion or social capital that may be protective against stress effects. Given the robust protective nature of both breastfeeding and neighborhood resources on child health, interventions around both of these factors are still warranted to improve childhood asthma outcomes, as are additional studies to better understand whether they may mitigate stress effects in other populations and study designs.

Our study has several strengths. We implemented a harmonized, combined-cohort study design to yield a large sample size, allowing for the examination of many novel effect modifiers. Some of the modifiers we assessed, such as child sex, suggest biological processes that have been explored elsewhere in the literature.⁷⁶ Others, such as the Childhood Opportunity Index, provide insight into community level resources and supports available not only to children, but also their mothers by proxy, that are less well studied in this context. Further, our combined-cohort model allowed evaluation in a study population with wide geographic and sociodemographic diversity. This racially, ethnically, and socioeconomically diverse study population reported SLEs at a frequency similar to other published estimates,^16–18 thus enhancing the generalizability of our findings. We used this rich data source to carefully and thoroughly adjust for confounding. We avoided over-adjustment of our primary models by including potential mediators, such as gestational age and smoking, only in sensitivity analyses in contrast to other studies where smoking adjustment is common in primary models.

Our study also has several limitations. While we used a well validated questionnaire to assess asthma and wheeze outcomes, we did not perform clinical assessment or medical record review to validate these conditions. We note that our primary exposure was assessed retrospectively. We did not collect data on when during pregnancy the SLEs occurred, or the level of severity or perceived stress associated with that event. Perceived stress may be particularly important to address in future studies as a means to further explore hypotheses around increased cortisol and child asthma risk. Purposefully, we relied on yes/no responses for each event evaluated so as to reduce inaccurate recall, as implemented in other studies.^77,78 Prior studies have shown that reporting of traumatic events is accurate over time,⁴⁹ suggesting a low likelihood for exposure misclassification for certain stressors. Further, prior work assessing retrospective recall of prenatal events up to 6, 8, and 30 years after pregnancy^79–81 demonstrate that women have good recollection of prenatal events. If exposure misclassification due to recall is present, we expect that it is non-differential by outcome status, biasing our results towards the null. Parents did not know the study hypothesis for the current analysis when completing the PSLE survey. Although asthma was an outcome of interest for the studies, all cohorts collected extensive child health information on multiple outcomes, so differential recall specifically by child asthma is unlikely. We note a possible inverse scenario in which women who recall PLSEs may be more likely to report asthma or wheeze outcomes; such a scenario would over-estimate true associations between PLSEs and childhood asthma and wheeze. However, we expect that such a scenario is unlikely given that exposure and outcomes were assessed at separate time points, and that parental recall of recent asthma symptoms is a validated measure^50,53,82 that is unlikely to be influenced by the occurrence of a prenatal stressor, particularly when that stressor was not referenced during the asthma assessment. Further, the PSLE has been shown to be predictive of a broad range of objective child health measures,^21,83,84 evidence that associations with PSLE are not driven by factors or tendencies inherent to maternal self-report. We also note that timing of exposures during pregnancy may be important to consider in future studies. In general, few asthma studies have captured timing of stress events during pregnancy.^10,30,33 Finally, we based our conclusions on the totality of significant and not significant results and did not adjust for multiple comparisons. We acknowledge that were we to do so, some of our significant findings would likely no longer reach statistical significance. However, this does not change our interpretation of results which remain important evidence upon which future studies can build.

Conclusion

In this large combined-cohort analysis, we observed that maternal stressful life events experienced during pregnancy are associated with increased risk of childhood asthma and wheeze, particularly in boys. Postnatal factors such as breastfeeding and childhood neighborhood did not buffer this association in this combined cohort at age 4-6 years but should continue to be explored as protective factors in future work. Current pregnancy cohorts can also consider obtaining contemporaneous PSLE and perceived stress data to support additional studies on this topic. Additionally, future studies investigating effective ways to support pregnant women in the prevention and management of stress upon the occurrence of a stressful life event may be warranted, as they may provide a two-generation impact on health and wellbeing.

Abbreviations:

CANDLE

Conditions Affecting Neurocognitive Development and Learning in Early Childhood study

CI

confidence interval

COI

Childhood Opportunity Index

GAPPS-PW

Global Alliance to Prevent Prematurity and Stillbirth study - Pathways

ISAAC

International Study of Asthma and Allergies in Childhood

PRAMS

Prevention Pregnancy Risk Assessment Monitoring System

PSLE

Pregnancy Stressful Life Event, or Stressful Life Event experienced during Pregnancy

RR

Risk Ratio

SLE

Stressful Life Event

TIDES

The Infant Development and Environment Study