Drug induced adverse pregnancy outcomes: innovative event history analysis for non-continuously exposed pregnancies in the national German Embryotox patient database

• Applicants: Professor Dr. Jan Beyersmann; Professorin Dr. Ulrike Grömping, since 8/2017; Professor Dr. Christof Schaefer
• Subject Area: Epidemiology and Medical Biometry/Statistics ; Reproductive Medicine, Urology
• Term: from 2016 to 2021
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 288952608

Prenatal development is the most vulnerable phase in human life. Drug toxicity may result in congenital anatomical and functional defects with life-long impacts. With a prevalence of 15-20%, miscarriages are the most frequent adverse events in pregnancy apart from birth defects. Teratogenic agents are the only causes being primarily preventable by avoiding exposure.A major determinant of teratogenic effects is the exposure interval within the first trimester. Exposure before or after the sensitive time period is not expected to cause birth defects. This principle of vulnerable periods for distinct malformations has been precisely characterised in humans only for thalidomide. The rationale of vulnerable periods not only applies to birth defects but also to spontaneous abortions.Empirical evidence on adverse effects of medication in pregnancy is based on observational data. National Teratology Information Services (TIS) like the German Embryotox project provide unique sources to study pregnancy outcome in association with suspicious drug exposure. Embryotox is the largest European Drug in Pregnancy Database with detailed prospectively ascertained information on maternal medication during pregnancy and neonatal outcome.One key characteristic of pregnancy outcome data is its time scale gestational age. However, a pregnancy can only be studied after it has been recognised. The data are, therefore, left-truncated. Drug intake is in general a time-dependent process associated with possibly time-dependent risks. Finally, pregnancy outcome is characterized by competing risks, as a pregnancy may end in a spontaneous abortion but also in an induced abortion or a live-birth. To prevent biased estimates, left-truncation, medication dynamics, and competing risks must be accounted for when analyzing pregnancy outcome data.Statistical event history methodology is both theoretically and practically established and is started to be used with success on pregnancy outcome data. Analyzing time-dependent and time-specific effects in pregnancy outcome data and in the context of left truncation and competing risks, however, raises practical and methodological issues, still to be solved.Managing the risk of adverse drug reactions in pregnancy has a great impact on society. In Germany, around 900000 pregnancies are recognised annually, resulting in approximately 130000 miscarriages, more than 100000 elective terminations, and 660000 live births. A better estimation of teratogenic effects of medication would allow reducing both the number of elective terminations based on irrational overestimation of drug risks as well as the individual and societal burden of preventable congenital anomalies.

• DFG Programme: Research Grants
• Ehemaliger Antragsteller: Professor Dr. Reinhard Meister, until 7/2017