Biostatistics Seminar Series: Jesse Berlin, ScD

Title: Canagliflozin Versus Other Antihyperglycaemic Agents on the Risk of Below-knee Amputation for Patients With Type 2 Diabetes: Results of the OBSERVE-4D Study, a Real World Analysis of >700,000 US PatientsAbstract: We (Janssen Pharmaceuticals) recently published a very large study of the risk of below-knee amputation for patients with type 2 diabetes, which compared incidence rates across multiple anti-hyperglycemia drugs and drug classes.  Interaction with a regulatory agency led to the addition of numerous combinations of study design and analysis features. We performed 192 analyses using open-source tools across the anonymized health care records of 714,582 adults, using seven different medications or classes of medications, for two health outcomes of interest, and across four different databases, generating 10,752 results.  Each result was fully-adjusted using large-scale propensity scores and could have been a publication unto itself.  I'll first present some of the methodological developments that allowed us to complete these analyses in a timely manner, mostly focusing on the Observational Health Data Sciences and Informatics (OHDSI) collaboration.  I'll then present some detail on the study and its findings. One message from this study is that there is still considerable uncertainty (not always recognized) in the epidemiology community as to the "correct" design to address particular questions.