Assessing the Quality of Birth Data in South Carolina: Comparing Birth Certificate Data to Abstracted Medical Records

BACKGROUND:  Birth certificate data are often used for public health monitoring and research.  Though the limitations of birth certificate data have been widely discussed, each revision of the US standard birth certificate attempts to improve the accuracy of data collection.  However, the validity of many of the items revised or added in the 2003 birth certificate revision has not been satisfactorily assessed.  The objective of this study is to assess the validity of selected items from the 2003 US standard birth certificate, as implemented in South Carolina, and determine whether validity varies by hospital.

METHODS:  Overall, 600 birth occurring between October 2010 and March 2011 were selected through stratified systematic sampling of births in four hospitals with a mix of characteristics.  Access to medical records was possible through a partnership with the South Carolina Hospital Association.  Sensitivity (for dichotomous variables) or exact agreement (for continuous variables) was calculated for selected data items using the medical record as the gold standard.

RESULTS:  Variables with sensitivity or exact agreement greater than 90% include: infant living at time of report, NICU admission, route and method of delivery-vaginal/spontaneous, route and method of delivery-cesarean, infant breastfed at discharge, birthweight, and obstetric estimate of gestation.  Variables with sensitivity or exact agreement less than 50% include: hypertension-prepregnancy, hypertension-gestational, previous preterm births, and total number of prenatal care visits.  The validity of fetal presentation and route and method of delivery items was consistent across hospitals, while the validity of Medicaid payment for delivery, induction of labor, and total number of prenatal care visits was inconsistent across hospitals.

CONCLUSIONS: Data for some items collected in South Carolina using the 2003 US standard birth certificate, such as route and method of delivery, had consistently high validity across all hospitals.  However, some items, such as Medicaid payment for delivery, were highly valid in some hospitals and invalid in others.  Other items, such as previous preterm births, had low validity in all hospitals.  Consistent education of personnel entering birth certificate data, on both proper definitions and procedures and on the importance of accurate data collection, may be as important as improved data collection instruments for the collection of valid data.