Development of algorithms to determine the onset of pregnancy and delivery date using health care administrative data in a university hospital in Japan

Purpose: To develop and assess algorithms to determine the onset of pregnancy and delivery date using health administrative data from a university hospital in Japan.

Methods: All women who were hospitalized in the maternity ward and had at least one pregnancy that ended with a delivery during the period of January 2014 and December 2015 were included in this study. The true delivery date was obtained from the electronic medical records and was used as a gold standard. The onset of pregnancy was calculated by subtracting the gestational age at birth from the delivery date based on the electronic medical records and was also used as a gold standard. The administrative data-based algorithms to identify (1) the onset of pregnancy estimated from the gestational age recorded as part of a diagnosis during a specific visit and (2) the delivery date estimated using the delivery-related diagnosis, procedure, or prescription were compared with the gold-standard data.

Results: Of the 1705 women included in this study, the onset of pregnancy was determined in 1704 subjects with 1582 (92.8%) within ± 7 days from the gold-standard date of pregnancy onset. The delivery date was determined in 1654 subjects, and 1594 (96.4%) were within ± 7 days before the true delivery date using the algorithm of "selected" diagnosis and a surgical procedure followed by some other delivery-related data.

Conclusions: The algorithms developed in this study are expected to accelerate future studies for real-world exposure and quantify drug safety during pregnancy using Japanese health care administrative databases.