Understanding differences in findings from pharmacoepidemiological studies: The case of antidepressant and benzodiazepine use and hip fracture

Abstract

Post-approval knowledge from observational pharmacoepidemiological studies has increased since the availability of electronic healthcare records. These studies have, however, sometimes reported variable and even contradictory results on the same medication use - adverse event associations. This thesis explores and discusses the variability in results from observational studies by taking the example of antidepressant and benzodiazepine use in relation to the risk of fractures. Harmonised study protocols have been applied to eight electronic healthcare databases from five European countries in the studies in this thesis. The studies in this thesis discuss variability in the results due to methodological choices specifically with respect to: The prevalence of antidepressant medication use and trends over time The prevalence of benzodiazepine and benzodiazepine like medication use and trends over time The incidence of hip and femur fractures The risk of hip and femur fracture among antidepressant users Concomitant use of antidepressants and benzodiazepines and the impact on the risk of hip and femur fractures This thesis discusses the contribution of harmonised study protocols in understanding the variability in the results across studies. In addition, the importance of a better characterisation of concomitant exposure of two medications is discussed when a common adverse event is examined.