Accuracy of reporting maternal in-hospital diagnoses and intrapartum procedures in Washington State linked birth records

While the impact of maternal morbidities and intrapartum procedures is a common topic in perinatal outcomes research, the accuracy of the reporting of these variables in the large administrative databases (birth certificates, hospital discharges) often utilised for such research is largely unknown. We conducted this study to compare maternal diagnoses and procedures listed on birth certificates, hospital discharge data, and birth certificate and hospital discharge data combined, with those documented in a stratified random sample of hospital medical records of 4541 women delivering liveborn infants in Washington State in 2000. We found that birth certificate and hospital discharge data combined had substantially higher true positive fractions (TPF, proportion of women with a positive medical record assessment who were positive using the administrative databases) than did birth certificate data alone for labour induction (86% vs. 52%), cephalopelvic disproportion (83% vs. 35%), abruptio placentae (85% vs. 68%), and forceps-assisted delivery (89% vs. 55%). For procedures available only in hospital discharge data, TPFs were generally high: episiotomy (85%) and third and fourth degree vaginal lacerations (91%). Except for repeat caesarean section without labour (TPF, 81%), delivery procedures available only in birth certificate data had low TPFs, including augmentation (34%), repeat caesarean section with labour (61%), and vaginal birth after caesarean section (62%). Our data suggest that researchers conducting perinatal epidemiological studies should not rely solely on birth certificate data to detect maternal diagnoses and intrapartum procedures accurately.