Leibniz Institute for Prevention Research and Epidemiology - BIPS, Germany
Background: Fetal anomaly is an important outcome in studies on the safety of medication during pregnancy. Safety studies focusing only on anomalies in live births may underestimate risks and might be prone to selection bias. Therefore, it is important to also consider termination of pregnancy due to fetal anomaly (TOPFA). Direct information on the reason for termination is often lacking in pharmacoepidemiological databases. Claims data contain several care and diagnostic codes related to (suspected) anomalies (“TOPFA indicators”) which might be useful to identify TOPFA.
Objectives: To assess the frequency and plausibility of TOPFA indicators in induced abortions and to develop and validate an algorithm to identify TOPFA in German claims data.
Methods: All pregnancies in the German Pharmacoepidemiological Research Database (GePaRD) between 2004 and 2022 ending in an induced abortion were included and the frequency of individual TOPFA indicators and their combinations was assessed. Plausibility was examined based on the timing and sequence of their coding as well as their combinations. The validity of potential algorithms was assessed by review of all available information on the mother and the pregnancy, especially other diagnoses, other diagnostic procedures and counseling codes.
Results: Overall, we included 94,978 induced abortions. In 14,588 (15%) of them at least one TOPFA indicator was coded. In pregnancies with at least one indicator, 38% had at least one code for an abnormal screening result (O28), 36% had at least one code for care due to (suspected) anomaly (O35) and 73% had at least one code for additional sonographic diagnostics due to suspected anomaly. The frequency of coding peaked shortly after the two screening periods, i.e. after 12-14 and 20-23 completed weeks. After patient profile review, TOPFA was plausible in 97% of pregnancies with at least one abnormal screening result and in 90% of pregnancies with no coded abnormal screening result, but at least two different other TOPFA indicators. In the final algorithm, at least one code for an abnormal screening result (O28) or two different other TOPFA indicators are required. Q-Codes for congenital malformations were coded in 65% of reviewed TOPFA, thereof in 40% only Q90.9 “Congenital malformation, unspecified”.
Conclusions: Care and diagnostic codes allow the identification of TOPFA in (German) claims data. The type of anomaly is often not identifiable, as TOPFA indicators are too unspecific and Q codes for specific anomalies were only seen in 39% of reviewed TOPFA.
