Applying Allostatic Load to Perinatal Outcomes Research

Abstract

Adverse perinatal outcomes such as preterm birth and low birth weight are significant public health concerns and contribute to neonatal morbidity and mortality. Maternal chronic stress (e.g., child maltreatment, posttraumatic stress disorder, depression) is an established predictor of adverse perinatal outcomes. However, the biological mechanisms by which maternal chronic stress affects adverse perinatal outcomes are less understood. Allostatic load (AL) refers to the cumulative dysregulations of multiple physiological systems responsive to multiple social-ecological levels of chronic stress. It is a promising conceptualization of the mechanism for stress effects on health. Little research has applied the AL theory to perinatal outcomes research to understand the complex pathophysiologic mechanisms for the stress-related adverse perinatal outcomes. Additionally, the optimal AL scoring method and the validity of pregnancy AL are less clear. Thus, the dissertation project had 3 aims: 1) to propose a theoretical model to situate AL in a role between maternal chronic stress and adverse perinatal outcomes; 2) to explore the optimal AL scoring method; and 3) to assess the gestational pattern of the AL summary score by the optimal scoring method and to test the validity of the pregnancy AL summary score for predicting a prior adverse birth outcome (as a proxy for adverse birth outcome subsequently on the current pregnancy). We used theory synthesis to construct a theoretical model to understand how maternal chronic stress contributes to adverse perinatal outcomes based on the AL theory. To address the second aim, women of reproductive age from the National Health and Nutrition Examination Survey (NHANES) data were included for analysis. We constructed AL summary scores using 5 scoring methods including the count-based, Z-Score, logistic regression, factor analysis, and grade of membership method and validated each score. We found the ALI score by the logistic regression method had the best predictive performances with regard to general health status, diabetes, and hypertension, but differences among the 5 summary scores were minor. When the outcome information is known or consistent across different contexts, the logistic regression method is optimal for use; otherwise we recommended the count-based method. To address the third aim, pregnant women from the NHANES data were included for analysis. The ALI score at each gestational month was not different from the average ALI score (M=2.35, SE=0.03, N=4319) in the non-pregnant population, suggesting that measuring AL at any gestational time point would reflect women's true physiological functions as long as gestational age is considered when scoring AL. We also found poor predictive performance of the ALI score for predicting prior adverse birth outcomes, which suggested that the AL summary measure is not sufficiently sensitive to use as a single predictor for the risk of adverse birth outcomes. This dissertation project may lay theoretical and methodological underpinnings for future research to understand the etiologic contribution of maternal chronic stress to adverse perinatal outcomes. Empirical research on maternal chronic stress, AL, and perinatal outcomes would assist in identifying women at risk for adverse perinatal outcomes and developing and evaluating effective interventions to mitigate stress-related adverse perinatal outcomes.