Identification of prenatal opioid exposure within health administrative databases

Background — Health administrative data offer a vital source of data on maternal prenatal opioid exposure (POE). The impact of different methods to estimate POE, especially combining maternal and newborn records, is not known.

Methods — This population-based cross-sectional study included 454 746 hospital births with linked administrative data in Ontario, Canada, in 2014–2017. POE ascertainment included 3 sources: (1) prenatal opioid prescriptions, (2) maternal opioid-related hospital records, and (3) newborn hospital records with neonatal abstinence syndrome (NAS). Positive percent agreement was calculated comparing cases identified by source, and a comprehensive method was developed combining all 3 sources. We replicated common definitions of POE and NAS from existing literature and compared both number of cases ascertained and maternal socio-demographics and medical history using the comprehensive method.

Results — Using all 3 data sources, there were 9624 cases with POE (21.2 per 1000 births). Among these, positive percent agreement (95% confidence interval) was 79.0% (78.2–79.8) for prenatal opioid prescriptions, 19.0% (18.2–19.8) for maternal opioid-related hospital records, and 44.7% (43.7–45.7) for newborn NAS. Compared with other definitions, our comprehensive method identified up to 523% additional cases. Contrasting ascertainment with maternal opioid-related hospital records, newborn NAS, and prenatal opioid prescriptions respective rates of maternal low income were 57%, 48%, and 39%; mental health hospitalization history was 33%, 28%, and 17%; and infant discharge to social services was 8%, 13%, and 5%.

Conclusions — Combining prenatal opioid prescriptions and maternal and newborn opioid-related hospital codes improves identification of a broader population of mothers and infants with POE.