Real World Pharmacoepidemiology of Drugs used in Diabetes: Harnessing e-heath records and artificial intelligence

We aim to examine the real-world efficacy and safety of drugs commonly used in diabetes in the total population of Scotland with diabetes. Many unanswered questions about drug efficacy and safety in diabetes are not resolved by clinical trials.

We will use the Diabetes Research Platform that we have created in Scotland containing 597,000 linked e-health records (EHRs) from all people with diabetes in Scotland.

Specific objectives are: To examine the real-world efficacy and safety of newer drugs used in type 2 diabetes GLP-1 receptor agonists: atrial fibrillation, microvascular complications, certain cancers, acute pancreatitis, gall bladder disease, suicide attempts DPP-4 inhibitors: acute pancreatitis, pancreatic cancer, microvascular complications To examine the safety of gabapentinoids in people with type 1 and people with type 2 diabetes (outcomes: cardiovascular and thrombotic events, pneumonia) In doing so we will create a reusable analysis pipeline facilitating rapid evaluation of many exposures and outcomes.

It will include artificial intelligence and Bayesian approaches to analysis.

Our database schema is based on the Observational Medical Outcomes Partnership (OMOP) Common Data Model so this pipeline will be portable to other platforms that use OMOP, including those being developed in the NHS England Secure Data Environment.