Does Medical Diagnosis Data Clarify the Relation Between Prenatal Care and Delivery Outcomes?

Studies indicating that prenatal care has minimal impact sparked additional work to assess whether this is a correct conclusion or a statistical artifact. Recent work highlights the importance of including medical diagnoses as regression variables, developing inclusive measures for health outcomes, and using providergenerated measures of prenatal care utilization. We explore these issues using administrative data provided by a tertiary care hospital and a prenatal care clinic. This data includes provider-generated utilization data, coded medical diagnoses, and variable direct costs for delivery-episode hospital care. Because all patients in this dataset obtained care from one set of providers in one urban area, the data set also minimizes heterogeneity due to variations in provider practice patterns and community services. We find that prenatal care exerts a significant beneficial impact on infant outcomes, and variable direct cost provides an inclusive outcomes measure when appropriate exogenous control variables are available. Estimated per-visit cost savings range from zero for patients with no diagnoses, to $10 for patients with one diagnosis, and $22 for patients with multiple diagnoses. The differences among these per-visit savings are significant, suggesting suboptimal allocation of clinic resources. Prenatal care utilization does not, however, increase with the number of diagnoses.