AbstractBackgroundClassifying preterm birth as spontaneous or indicated is critical both for clinical care and research, yet the accuracy of classification based on different data sources is unclear. We examined the accuracy of preterm birth classification as spontaneous or indicated based on birth certificate data.MethodsThis is a retrospective cohort study of 123 birth certificates from preterm births in Missouri. Correct classification of spontaneous or indicated preterm birth subtype was based on multi‐provider (RN, MFM Fellow, MFM attending) consensus after full medical record review. A categorisation algorithm based on clinical data available in the birth certificate was designed a priori and classification was performed by a single investigator according to the algorithm. Accuracy of birth certificate classification as spontaneous or indicated was compared to the consensus classification. Errors in misclassification were explored.ResultsClassification based on birth certificates was correct for 66% of preterm births. Most errors in classification by birth certificate occurred in classifying a birth as spontaneous when it was in fact indicated. The vast majority of errors occurred when preterm rupture of membranes (≥12 h) was checked on the birth certificate causing classification as spontaneous when there was a maternal or fetal indication for delivery.ConclusionsBirth certificate classification overestimated spontaneous preterm birth and underestimated indicated preterm birth compared to classification performed from medical record review. Revisions to birth certificate clinical data would allow more accurate population level surveillance of preterm birth subtypes.