Psychosocial Stressors as a Determinant of Maternal Cardiovascular Health During Pregnancy (2024)

Associated Data

Supplementary Materials

Abstract

Maternal psychosocial stress may be a risk factor for poor cardiovascular health (CVH) during pregnancy. We aimed to identify classes of psychosocial stressors among pregnant women and to evaluate their cross-sectional association with CVH. We performed a secondary-analysis of women from the Nulliparous Pregnancy Outcomes Study: Monitoring Mothers-to-be (nuMoM2b) cohort (2010–2013). Latent class analysis was used to identify distinct classes of exposure to psychosocial stressors based on psychological (stress, anxiety, resilience, depression) and sociocultural indicators (social support, economic stress, discrimination). Optimal and suboptimal CVH were defined based on the presence of 0 to 1 and ≥2 risk factors (hypertension, diabetes, smoking, obesity, inadequate physical activity), respectively based on the American Heart Association’s Life’s Essential 8. We used logistic regression to evaluate the association between psychosocial classes and CVH. We included 8,491 women and identified five classes reflective of gradations of psychosocial stress. In unadjusted models, women in the most disadvantaged psychosocial stressor class were approximately 3 times more likely to have suboptimal CVH compared to those in the most advantaged class (OR 2.98, 95% CI: 2.54, 3.51). Adjusting for demographics minimally attenuated the risk (aOR 2.09, 95% CI: 1.76, 2.48). We observed variation across psychosocial stressor landscapes among women in the nuMoM2b cohort. Women in the most disadvantaged psychosocial class had a greater risk of suboptimal CVH which was only partially explained by differences in demographic characteristics. In conclusion, our findings highlight the association of maternal psychosocial stressors with CVH during pregnancy.

Keywords: maternal cardiovascular health, psychosocial stressors

Pregnancy represents a unique period of a woman’s life course with increased cardiometabolic demands and may act as a “stress test” predicting future cardiovascular events¹. Compared to non-pregnant women of the same age, pregnant women have elevated levels of cardiovascular risk factors². This is especially concerning given the potential for intergenerational transmission of poor cardiovascular health (CVH) to children and the association of poor CVH with adverse pregnancy outcomes³. Furthermore, there are racial and socioeconomic disparities in adverse pregnancy outcomes that must be addressed⁴. Therefore, examining CVH during pregnancy is critical for the identification of factors that may improve the health of children as well as women across the life course. While previous evidence indicates that psychosocial stress is a risk factor for poor CVH⁵, there are few studies exploring its association with CVH during pregnancy. Additionally, prior studies have relied on single measures of psychosocial stress failing to capture the multidimensional nature of stressful events experienced during pregnancy^6,7. In order to develop and conceptualize targeted prevention strategies, it is important to consider multiple indicators of psychosocial stress and to use a prevention-focused definition of CVH. Thus, the objective of this study was to identify classes of psychosocial stressors among pregnant women and to evaluate their association with suboptimal CVH during pregnancy.

Methods

Study participants were selected from the Nulliparous Pregnancy Outcomes Study: Monitoring Mothers-to-be (nuMoM2b) cohort, a prospective pregnancy cohort recruited between 2010 and 2013 at 8 academic medical centers across the United States⁸. Mothers were recruited prior to 14 weeks of gestation. Participants were eligible to enroll in the study if they were over 13 years of age, had a singleton pregnancy with no prior pregnancies lasting more than 20 weeks of gestation, and planned to deliver at a nuMoM2b Network Hospital. Participants completed 4 study visits; visit 1 was at 6 to 13 weeks of gestation; visit 2 was at 16 to 21 weeks; visit 3 was at 22 to 29 weeks; and visit 4 was at the time of delivery. Participants completed standardized questionnaires during each study visit, and additional information was abstracted from their medical records following delivery.

For the present study, we restricted our sample to individuals who consented to the release of their data into the National Institute of Child Health and Human Development Centralized Data and Specimen Hub database (n = 9,298). We further excluded participants missing indicators for our main outcome of CVH (n = 798). Our final analytical sample was 8,491 participants (Appendix Figure 1).

Each site’s institutional review board approved the nuMoM2b protocol (ClinicalTrials.gov identifier: NCT01322529), and we received approval from the Johns Hopkins Bloomberg School of Public Health IRB for this secondary analysis.

Our exposure was a composite measure of maternal psychosocial stress that encompassed psychological and sociocultural measures during pregnancy. The psychological measures captured stress, anxiety, resilience, and depression, while the sociocultural measures captured social support, payment concern, and experiences of racial discrimination (Appendix Table 1). The measures were based on validated scales, and we categorized each indicator to reflect an increased burden of maternal psychosocial stress (e.g., high stress, moderate stress, and low stress).^9–14.

Our primary outcome was suboptimal maternal CVH measured early in pregnancy based on the American Heart Association’s Life’s Essential 8 targets, excluding diet, sleep, and hyperlipidemia¹⁵. We excluded diet and sleep from our primary analysis due to the large amount of missing data for these measures (n = 1,756), although they are included in sensitivity analyses described below. We excluded markers of pre-pregnancy hyperlipidemia because these were not publicly available. Thus, our composite variable of CVH (range: 0–5) was comprised of 5 binary indicators: chronic hypertension, pre-gestational diabetes, smoking, physical activity, and pre-pregnancy obesity. At visit 1 (median gestational age = 12.1 weeks), participants self-reported chronic hypertension and diabetes prior to pregnancy, and study staff validated these self-reported measures through chart abstraction. Smoking was a self-reported measure of smoking anytime in the 3 months prior to pregnancy. The duration, frequency, and intensity of physical activity was assessed in the 4 weeks prior to visit 1. We categorized physical activity as inadequate if participants reported < 150 minutes of moderate/intense activity per week¹⁶. Pre-pregnancy body mass index (BMI) ≥ 30 kg/m² indicated obesity. We created a binary variable that captured suboptimal (≥ 2 risk factors) vs. optimal (0–1 risk factors) CVH¹⁷.

As a sensitivity analysis among the sample with non-missing diet and sleep data, we created a second CVH variable (range: 0–7) that included additional indicators for sleep and diet. The amount of sleep in the past week was assessed at visit 1, and we categorized < 7 hours of sleep as inadequate¹⁸. Diet quality in the 3 months prior to pregnancy was obtained from a modified version of the Block 2005 Food Frequency Questionnaire, and this was used to calculate the Healthy Eating Index 2010. We categorized inadequate diet quality as a Healthy Eating Index < 51¹⁹. The binary variable of suboptimal vs. optimal CVH was calculated in the same manner with the inclusion of the indicators for sleep and diet.

Demographic covariates included maternal age (13–19, 20–24, 25–29, 30–34, ≥ 35 years), self-reported race/ethnicity (non-Hispanic white, non-Hispanic Black, Hispanic, and other), highest level of maternal education (less than high school, high school or GED, some college or associate or technical degree, or completed college), insurance payer (private/HMO/self-pay or public), and nativity (born in the United States or born outside the United States). We also included the presence of a significant other or partner measured at the first study visit (yes or no).

We performed latent class analysis, a person-oriented approach, to identify psychosocial stressor classes within the entire cohort using the indicators for stress, anxiety, resilience, depression, social support, payment concern, and experiences of racial discrimination²⁰. First, we fit the model with a 1 class solution followed by models with increasing number of classes. We performed 50 repetitions per model to account for missing indicator data²¹. We used established statistical measures of fit to inform our final model selection, and we assigned participants to the class with the highest posterior probability²².

We compared demographic characteristics across the latent classes using analysis of variance for continuous variables and chi-square analysis for categorical variables. In our primary analysis, logistic regression was used to examine the association between latent class membership and a binary composite indicator of suboptimal CVH based on indicators of hypertension, pre-gestational diabetes, smoking, physical activity, and obesity. In sensitivity analyses on a smaller sample with no missing data on diet and sleep, we estimated an additional logistic regression model to test the association between latent class membership and a composite measure of suboptimal CVH that also included sleep and diet. We also examined the association of latent class membership with each individual component of CVH. Logistic regression models were (1) unadjusted and (2) adjusted for maternal age, race/ethnicity, education, insurance, nativity, and significant other and are reported as odds ratios (OR) and 95% confidence intervals (CI). Values for nativity and health insurance were missing for 1% of individuals and were imputed using regression.

Analyses were performed using R version 4.1.1 (R Foundation for Statistical Computing, Vienna, Austria). We used the ‘poLCA’ package to perform the latent class analysis²³.

Results

The study population comprised 8,491 participants (29.5% between the agesZ of 25 to 29 years, 16.5% Hispanic, 12.9% non-Hispanic Black, 51.0% completed college or more). Among all participants, 2,423 (28.5%) had suboptimal CVH, with inadequate physical activity (n = 5,861, 69.0%), obesity (n = 1,880, 22.1%), and smoking (n = 1,520, 17.9%) acting as the greatest drivers of CVH risk factors. Only 219 (2.6%) of the sample reported chronic hypertension and 131 (1.5%) reported pre-gestational diabetes (Table 1).

We identified 5 classes that represented the experiences of psychosocial stressors within the entire cohort (Appendix Table 2). For interpretability, we ordered our latent classes from lowest to highest stressor burden. For example, Class 1 (n = 1,567, 18.5%) represented the group exposed to the least psychosocial stressors, while Class 5 (n = 1,524, 17.9%) represented the group exposed to the highest burden of psychosocial stressors based on the estimated response probabilities for our psychological and sociocultural indicators (Table 2). The proportion of participants assigned to each class ranged from 3,433 (40.4%) in Class 2 to 919 (10.8%) in Class 4.

The demographic characteristics by class membership are displayed in Table 1. Compared to participants in classes with less psychosocial stressors, participants in classes with more psychosocial stressors were more likely to be younger, identify as non-Hispanic Black or Hispanic, have less education, receive government sponsored health insurance, and report not having a significant other (all p-values < 0.001). Additionally, there were no differences in the timing of visit 1 (overall mean [SD]: 12.1 [1.5] weeks’ gestation) and visit 2 (overall mean [SD]: 19.0 [1.6] weeks’ gestation) between the classes.

With increasing exposure to psychosocial stressors, there was an increased proportion of individuals with suboptimal CVH within each class. Class 1 had the lowest prevalence of suboptimal CVH (n = 310, 19.8%), while Class 5 exhibited the highest prevalence of suboptimal CVH (n = 646, 42.4%) (Table 1). The proportions of both chronic hypertension and diabetes were not different across psychosocial stressor classes (p = 0.67 and 0.12, respectively). However, the distributions of obesity, smoking and physical activity demonstrated significant differences across the psychosocial stressor classes (p < 0.001 with all risk factors).

In our primary model results, compared to individuals in Class 1, the unadjusted ORs for suboptimal CVH compared to optimal CVH were 1.33 (95% CI: 1.15, 1.54) for Class 2, 1.98 (95% CI: 1.66, 2.37) for Class 3, 1.74 (95% CI: 1.44, 2.10) for Class 4, and 2.98 (2.54, 3.51) for Class 5 (Appendix Table 3). Adjusting for demographic characteristics slightly attenuated these associations. The multivariable-adjusted ORs for suboptimal CVH compared to optimal CVH were 1.26 (95% CI: 1.09, 1.48) for Class 2, 1.59 (95% CI: 1.31, 1.92) for Class 3, 1.50 (95% CI: 1.23, 1.83) for Class 4, and 2.09 (95% CI: 1.76, 2.48) for Class 5 (Table 3). We also found that the odds of suboptimal CVH were different between every class except for Class 3 and Class 4 (Appendix Table 4); however, these results should be interpreted with caution as they were not adjusted for multiple comparisons.

In our sensitivity analysis, we restricted the sample to the 6,735 individuals with non-missing values for both diet and sleep, with 19.4% (n = 1,394) and 24.8% (n = 1,851) reporting poor diet and inadequate sleep, respectively (Appendix Table 5). Our effect estimates were consistent with our primary model and showed a slightly increased magnitude. Compared to individuals in Class 1, the multivariable-adjusted ORs for suboptimal CVH compared to optimal CVH were 1.29 (95% CI: 1.09, 1.53) for Class 2, 1.67 (95% CI: 1.35, 2.08) for Class 3, 1.56 (95% CI: 1.25, 1.95) for Class 4, and 2.20 (95% CI: 1.81, 2.68) for Class 5 (Appendix Table 6). Similar to our primary model results, we again found that all the classes except for Class 3 and Class 4 differed in terms of their risk of suboptimal compared to optimal CVH. When we examined the individual components of CVH, the associations for chronic hypertension and diabetes were null but were significant for obesity, physical activity, smoking, diet, and sleep (Appendix Table 7).

Discussion

Within a well-characterized pregnancy cohort of nulliparous women, we observed 5 classes of psychosocial stressors that reflect the multidimensional components of psychological and sociocultural experiences. Women in the most disadvantaged psychosocial class had 2-fold greater odds of suboptimal CVH, which was only partially explained by differences in demographic characteristics. Additionally, our findings were robust to a sensitivity analysis that included additional measures of CVH. These findings highlight the burden of psychological and sociocultural experiences on CVH during pregnancy and may have important implications for maternal and fetal outcomes.

Our results are similar to those of prior studies that highlight the heterogeneity of psychosocial stressors during pregnancy. An analysis of 7,740 pregnant women enrolled in the Amsterdam Born Children and their Development study noted five clusters of women with different intensities of psychosocial stress²⁴. Findings from the Healthy Start Study (n = 1,410) based in Denver, Colorado²⁵ and the Healthy Pregnancy, Healthy Baby Study (n = 1,852) in Durham County, North Carolina both identified three stressor domains among pregnant women⁶. In a latent profile analysis, Grobman et al. used psychometric indicators to identify 4 groups of women among participants in the nuMoM2b cohort²⁶. The differences between our findings and these studies may be due to the geographical locations of the cohort, statistical methods to quantify psychosocial stress, and the availability of psychosocial indicators within the cohort.

The distribution of cardiovascular risk factors we observed in this cohort is comparable to prior studies. The overall prevalence of suboptimal CVH in this cohort of nulliparous pregnant women was 28.5%. A study of pregnant women from the National Health Interview Survey identified the prevalence of suboptimal CVH at 33.6%¹⁷, while an analysis of pregnant women participating in the National Health and Nutrition Examination Survey from 1999–2014 also found a similar prevalence of low CVH at 34.8%². The slightly lower prevalence of suboptimal CVH in the nuMoM2b cohort may be due to the fact that nulliparous women are likely younger and healthier than multiparous women¹.

Behavioral factors comprised the greatest drivers of suboptimal CVH within the cohort, including inadequate physical activity, obesity, and smoking. These risk factors also contribute to adverse pregnancy outcomes, including preterm birth and preeclampsia, and are likely a result of distal drivers of health (e.g., economic, structural, and sociocultural factors)²⁷. Previous studies have shown that lower socioeconomic status and social support are associated with behavioral risk factors of cardiovascular disease^28,29. In our study, low social support and high payment concern were most prevalent in the classes with the highest burden of social stresses, which also had higher rates of suboptimal CVH. To our knowledge, studies have yet to specifically explore the impact of psychological stressor measures such as stress, anxiety, resilience, and depression on maternal CVH. Our study suggests that these factors are heavily intertwined with social determinants of health and that psychosocial stressors may be on the pathway between social determinants of health and poor maternal CVH.

There are several limitations to note. First, we were unable to include a measure of pre-pregnancy hyperlipidemia in the construct of CVH. Results from the National Health and Nutrition Examination Survey from 1999 to 2014 demonstrate that among pregnant individuals, 38.9% had ideal cholesterol (< 200 mg/dL untreated), 29.9% had intermediate cholesterol (200–239 mg/dL or treated to this level), and 31.3% had poor cholesterol (≥ 240 mL/dL)². However, these findings may be unreliable during early pregnancy due to physiologic changes that lead to an expected rise in cholesterol³⁰. Second, the indicators for physical activity, smoking, diet, and sleep were all based on participant self-report, which may be susceptible to social desirability bias and recall error. Third, the American Heart Association provides recommendations on metrics to measure and quantify CVH. However, since this is a secondary analysis of a prospective pregnancy cohort, we are limited by the specific measurement instruments chosen at the start of the study and used previously validated metrics to ascertain suboptimal CVH¹⁷.

Our study has several strengths. First, this is one of the first studies to incorporate the updated Life’s Essential 8 CVH indicator of sleep to study CVH of pregnant women. This acknowledges the association of inadequate sleep duration and quality with poor CVH¹⁵. Second, we used measurements of psychosocial stressors from early in pregnancy based on validated scales. Last, we used a diverse pregnancy cohort that enabled the identification of 5 class of psychosocial stressors.

Our findings highlight the importance of viewing psychosocial stress within a larger context of factors, including social determinants that influence preconception and maternal health. We also demonstrate the association of psychosocial stressors with CVH prior to and during early pregnancy. Although this relationship may be bidirectional, findings from this study may help in the development of screening tools to implement in the clinical setting during early pregnancy. We also identify factors that are not routinely measured in the clinical setting, such as payment concern, that may be important to incorporate. Future research should focus on both the measurement of psychosocial stressors among pregnant women as well as ways to support positive psychosocial wellbeing to encourage sustainable, healthy behaviors.

Supplementary Material

Acknowledgements

Footnotes

References