Interrupted time series analysis to assess changes in prescription filling around conception and implications for exposure misclassification

摘要

Purpose: Medication exposures in pregnancy are often defined by one or more prescription fills. Harmful effects could be underestimated if rapid discontinuation of use after pregnancy recognition is common. We used conception, a critical biological period, as an intervention in a novel application of interrupted time series analysis (ITSA). Methods: Among 645 049 pregnancies from the Medical Birth Registry (2005-2015) linked to the Norwegian Prescription Database, we modeled the total number of prescription fills in the 12 weeks before and after estimated conception date with ITSA. We examined psychostimulants, antidepressants, antipsychotics, and antiepileptics (AEDs; separated by use for epilepsy or other indications). We used relative measures (%) to compare model coefficients. We also compared number of pregnancies defined as exposed when the earliest fill considered was 30 days before the last menstrual period (LMP -30 days), LMP, or estimated conception date (LMP +14 days). Results: We observed a sudden decline in prescription fills from 2 weeks after conception and decreasing fills thereafter for psychostimulants, antidepressants, AEDs for other indications, and antipsychotics excluding incident users. Fills for AEDs for epilepsy did not fall after conception. Only 77% of pregnancies with fills for psychostimulants from LMP and 58% with fills from LMP -30 days had fills from conception. Similar figures for AEDs for epilepsy were 99% and 96%. Conclusions: This application shows that ITSA can help researchers understand rapid changes in patient behavior around conception that have consequences for exposure misclassification in pregnancy drug safety studies. ITSA results can help pharmacoepidemiologists guide study exposure definitions.