The purpose of this thesis is to develop a statewide, hospital profiling methodology for the analysis of variation of maternity length of stay. The study will examine whether hospital-to-hospital variability in maternity length of stay in Arizona is related to inter-hospital differences in patient, physician, and hospital characteristics. By using a three-level, hierarchical linear model and a broad number of relevant characteristics, the goal is to build upon existing provider profiling methodologies, health service research models, and maternity care studies. The study aims include: to estimate the infra-hospital variability in maternity length of stay for vaginal delivers; to determine what factors explain inter-hospital variability in maternity length of stay for vaginal deliveries, controlling for patient physician, and hospital characteristics; to estimate the intra-hospital variability in maternity length of stay for cesarean deliveries; to determine what factors explain inter-hospital variability in maternity length of stay for cesarean deliveries, controlling for patient, physician, and hospital characteristics; to be able to develop statewide hospital profiling indicators. To examine these aims, a study will be conducted using secondary data sources for 1996: Arizona hospital discharge summary data, Arizona birth certificate data, Arizona physician data, and American Hospital Association data. Developing hospital profiles using hierarchical linear modeling should build upon current performance indicator models. The proposed methodology distinguishes intra-hospital as well as inter-hospital variation attributed to patient casemix, treating physicians, and hospital of admission. The benefits from understanding variation in maternity resource utilization might lead to public health interventions and policy-making initiatives directed at differences in patient, physician, and hospital characteristics. Moreover, once demonstrated, this model may be applied to other area of healthcare, diagnosis groups, and outcomes of interest.
