Jamie Daw

Jamie Daw

Assistant Professor, Department of Health Policy & Management, Columbia Mailman School of Public Health

Dissertation Title:  "Insurance Coverage for Pregnant Women: Assessing Patterns, Policy Impacts and Methods for Evaluation"This dissertation includes two applied studies on health insurance coverage for pregnant women in the United States and one methodological contribution to the field of health services research.

Chapter one describes patterns of insurance coverage for pregnant women using national longitudinal survey data from 2005 to 2013. First, we estimate rates of insurance coverage, insurance transitions and insurance lapses in the twelve calendar months before and the six calendar months after childbirth. We find that the period surrounding childbirth is characterized by frequent gaps and changes in coverage that may compromise timely access to recommended care and the continuity and quality of care. Second, we use logistic regression models to identify risk factors associated with insurance lapses before and after pregnancy. We find that risk factors associated with insurance loss after delivery include not speaking English at home, having Medicaid coverage at delivery, living in the South and having a family income of 100 to 185 percent of the poverty level. The findings of this study emphasize the need to develop policies to improve continuity of insurance coverage for women of reproductive-age, particularly low-income women who are eligible for pregnancy-related Medicaid.

Chapter two estimates the association between the Affordable Care Act’s dependent coverage provision, which allowed young adults to enroll in their parent’s plan until age 26, and payment for birth, prenatal care use, and infant birth outcomes among unmarried and married women. Drawing on birth certificate data for nearly 3 million births, we use difference-in-differences analysis to compare outcomes among eligible women (ages 24-25) to a control group of slightly older women (ages 27-28) before and after the implementation of the provision. We find that the policy was associated with a decline in Medicaid payment and a 20% increase in private payment for delivery among unmarried women. We also find an association between the policy and modest improvements in early prenatal care, cesarean delivery and preterm birth among unmarried women. We do not find an association between the policy and payment for birth, adequate prenatal care or birth outcomes among married women, nor do we any find changes in low birthweight or NICU admission. The findings of this study suggest that the ACA’s dependent coverage provision shifted unmarried pregnant women from Medicaid to their parent’s private plans and that this shift was associated with neutral to small positive changes in prenatal care and birth outcomes.

Chapter three uses a Monte Carlo simulation to estimate the bias that can be introduced by applying matching to difference-in-differences. Our results show that matching can have an important impact on estimated intervention effects, particularly when matching on pre-period levels of the outcome itself or on time-varying covariates with low serial correlation. We find that the bias introduced by regression-to-the mean increases with pre-period differences between the treatment and control group and with decreasing serial correlation in the matching covariates. The findings of this study suggest that researchers should exercise caution when matching on pre-period variables in study designs that estimate effects based on changes over time. Based on our results, we provide guidance for selecting matching variables in difference-in-difference analysis.


Graduation Year