The goal of this FIRST Award is to gain greater understanding concerning an important, yet understudied, group of violence victims, namely, pregnant women who have been physically abused by their male partners (husbands/boyfriends). More specifically, this study examines: (1) how patterns of partner violence before and during pregnancy are related to patterns of partner violence after infant delivery; (2) how partner violence affects women's mental health (including symptoms of post-traumatic stress, depression, alcohol abuse, and illicit drug abuse); and (3) whether infants of mothers who are victims of partner violence are at increased health risk (with a focus on the infant's risk of being physically abused), and whether the potential relationship between partner violence and infant health are modified by the mother's mental health status. Three groups of pregnant women will be recruited for study from a large prenatal clinic of a Health Department: (1) 50 women who were victims of partner violence both before, but not during, pregnancy; (2) 50 women who were victims of partner violence both before and during pregnancy; (3) a matched comparison group of 100 women who were not victims of partner violence before or during pregnancy. These women will be followed until their infants are approximately one year of age. Study data will be gathered from both reviews of the study participants' health and social service records, as well as study interviews conducted with the women at four time points (when the women are 6 months pregnant, 1, 6, and 12 months after infant delivery). Incorporated into the study interview will be several assessment instruments including the Conflict Tactics Scales 2, a Trauma Assessment, the Post-traumatic Stress Disorder Symptom Scale Interview, the Center for Epidemiologic Studies Depression Scale, the short form of the Michigan Alcohol Screening Test, the Drug Abuse Screening Test, and Child Abuse Potential Inventory. State-of-the-art multivariate statistical procedures, specifically, generalized estimating equations analysis, will be used to address each of the study questions. These newly developed procedures are appropriate for the analysis of longitudinal data (both continuous and categorical outcomes) and will allow for the inclusion of potentially important time-dependent covariates in the study models.
