The Florida Investigation of Primary Late Preterm and Cesarean Delivery: The accuracy of the birth certificate and hospital discharge records

Heather B. Clayton, William M. Sappenfield, Elizabeth Gulitz, Charles S. Mahan, Donna J. Petersen, Kara M. Stanley, Hamisu M. Salihu

(1) Assess the accuracy of public health data sources used to investigate primary late preterm cesarean delivery (PLPCD) and (2) compare differences in data accuracy by hospital PLPCD rate classification. This analysis uses data from the Florida Investigation of Late Preterm and Cesarean Delivery (FILPCD), an investigation of singleton, PLPCD's that occurred from 2006 to 2007 in hospitals classified with either a low or high PLPCD rate (high rate 39.4-58.3 %, low rate 11.9-25.1 %). Three data sources were validated with maternal medical records: birth certificates, hospital discharge data, and combined birth certificate and hospital discharge data. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and kappa values were calculated. A summary measure of kappa values was compared by hospital PLPCD rate classification using the paired sample Wilcoxon signed rank test. Large variations in accuracy of data elements were found by hospital PLPCD rate classification, with low PLPCD rate hospitals demonstrating higher overall data accuracy. The summary measure of agreement was significantly higher for low PLPCD rate hospitals compared to high PLPCD rate hospitals (0.60 vs. 0.50, p < 0.01). Accurate estimates of CD and late preterm birth are vital for public health practitioners and policy makers who seek to address the growing concern over recent increases in CD and late preterm birth. Understanding the potential for systematic differences in reporting accuracy by hospital PLPCD rate is important to data quality improvement efforts.