Opportunities and Pitfalls in Drug Safety Studies after Marketing Approval An Evaluation with a Focus on Older Patients

In the next decades, the number of older people will rise substantially in Germany. Likewise, drug treatment and polypharmacy will also augment due to the high prevalence of multimorbidity in this subpopulation. Since older people are often excluded from randomized controlled trials prior to drug approval, pharmacoepidemiological safety studies (PSS) based on spontaneous reporting systems and electronic healthcare databases often represent the only opportunity to investigate the safety of drugs in this population. However, these studies have specific methodological challenges related to the clinical characteristics of older patients and the nature of the data sources. Thus, the overall aim of this thesis is to (1) critically assess methodological challenges of PSS based on spontaneous reporting systems and electronic healthcare databases with a focus on older patients and (2) to define further areas of research to enhance their quality. In this context, disproportionality analyses based on spontaneous reports, cohort studies, nested case control studies, and case only designs are introduced as study designs for PSS. Confounding (e.g., by frailty), outcome and exposure misclassification as well as time related biases in PSS are illustrated as selected methodological challenges. These challenges are then discussed in the context of my research articles with a focus on older people, and opportunities to address these challenges are presented. More specifically, the role of spontaneous reporting systems in the detection of adverse drug reactions in older people is critically assessed. Afterwards, drug utilization studies as well as the application of high dimensional propensity score methods and case only designs are discussed as options to overcome the specific problem of confounding by indication and unmeasured confounding in PSS among older people. Further, a detailed review of the patienta s profile is recommended to increase the specificity of the outcome case algorithms in administrative claims databases. Moreover, it is highlighted that sensitivity analyses in drug utilization and safety studies are particularly important in the case of a as neededa treatment among older patients and if information on the prescribed daily dose is missing. Finally, it is highlighted how time related biases can be prevented in cohort and nested case control studies using a time dependent analysis and risk set sampling, respectively. In the conclusion, future research perspectives with regard to PSS in older patients are pointed out as, for instance, the use of semi automated drug safety monitoring based on electronic healthcare databases, the availability of additional medical information in the context of the German a e healtha legislation or the need for external validation studies to study the impact of outcome and drug exposure misclassification.