Accuracy of Assisted Reproductive Technology Information on Birth Certificates: Florida and Massachusetts, 2004-06

Background Assisted Reproductive Technology (ART) includes fertility procedures where both egg and sperm are handled in the lab. ART use has increased considerably in recent years, accounting for 47 090 livebirths in the US in 2010. ART increases the probability of multiple gestation births, which are at higher risks than singletons for adverse outcomes. Additionally, ART is associated with a greater risk of complications during pregnancy, labour, and delivery, and increased risk of adverse perinatal outcomes in singleton births. Methods We merged Florida and Massachusetts birth records from 2004-06 with the National ART Surveillance System (NASS) and using NASS as the gold standard, calculated sensitivity, specificity, and positive predictive value (PPV) of ART reporting on the birth certificates by maternal, infant, and hospital characteristics. We fit random-effects logistic regression models to evaluate simultaneously the association of ART reporting with these predictors while accounting for correlation among births occurring in the same hospital. Results Sensitivity of ART reporting on the birth certificate was 28.9% in Florida and 41.4% in Massachusetts. Specificity was >99% in both states. PPV was 45.5% in Florida and 54.6% in Massachusetts. The odds of ART reporting varied by state and by several maternal and delivery characteristics including age, parity, history of fetal loss, plurality, race/Hispanic ethnicity, delivery payment source, pre-existing conditions, and complications during pregnancy or labour and delivery. Conclusions There was significant under-reporting of ART procedures on the birth certificates. Using data on ART births identified only from birth certificates yields a biased sample of the population of ART births.