Assessing the quality of medical and health data from the 2003 birth certificate revision: results from two states.

Objectives: A primary goal of the 2003 revision of the U.S. Standard Certificate of Live Birth was to improve data quality, in part by improving data sources, definitions, and instructions. This report evaluates the quality of selected medical and health data from the 2003 revision of the birth certificate by comparing birth certificate data with information abstracted from hospital medical records.
Methods: A random sample of records for 600 births that occurred in 2010-2011 in State A, and a convenience sample of 495 births that occurred in State B in 2009 were reviewed. Birth certificate and hospital medical record data were compared for these categories: pregnancy history, prenatal care, gestational age, birthweight, pregnancy risk factors, obstetric procedures, onset of labor, source of payment, characteristics of labor and delivery, fetal presentation, method of delivery, abnormal conditions of the newborn, infant living, and infant breastfed. Levels of missing data, exact agreement, kappa scores, sensitivity, and false discovery rates are presented, where applicable.
Results: Exact agreement or sensitivity, was high for a number of items for both states (e.g., number of cesarean deliveries, cephalic presentation, cesarean delivery, and birthweight within 500 grams), but exact agreement or sensitivity was low or extremely low for both states for several items (e.g., total number of prenatal visits, previous preterm birth, meconium staining, and fetal intolerance of labor) (Figure 1). Levels of agreement or sensitivity for most items (e.g., prenatal care beginning in first trimester and source of payment-private insurance) were substantial or moderate. Data quality varied by state, and often, varied widely by hospital.