Assessing the real-world effectiveness and safety of type 2 diabetes treatments for important patient subgroups not represented in clinical trials

Diabetes

Summary

This project validates and refines a novel statistical framework to assess the safety and effectiveness of type 2 diabetes treatments in real-world data.
The work aims to improve treatment decision-making and clinical care by providing high-quality, causal evidence especially for patient groups often excluded from clinical trials.

How are we doing it?

To build credibility beyond proof-of-concept work, the framework will be benchmarked against established causal evidence from published clinical trials. Target trials will be emulated using data from the Clinical Practice Research Datalink, a UK primary care database with records from around 1.6 million people with type 2 diabetes. The analysis will mirror trial characteristics such as eligibility criteria, follow-up, and outcome definitions. Comparing published and replicated results will assess the framework’s validity in routine care.

To evaluate added value, the framework will be compared with established observational methods, including propensity score approaches recommended by the NICE real-world evidence framework. This will highlight where the new method improves on standard approaches in addressing unmeasured confounding and data limitations.

The framework will also be enhanced with methods for sensitivity analyses and bias quantification, including negative control outcomes and diagnostic plots. All project code will be openly shared to support reproducibility and uptake in type 2 diabetes research.

What happens next?

Over this two-year fellowship, I will follow a series of structured work packages, starting with benchmarking my framework against published clinical trial results and refining it through comparisons with established methods and sensitivity analyses. I will collaborate with public and patient involvement groups, especially those representing people with type 2 diabetes who are often excluded from clinical trials. Their lived experience will provide valuable insight into treatment challenges, informing both current research and future questions. To strengthen my expertise, I will attend the 2025 Causal Inference and Machine Learning for Health Research Summer School at the Harvard T.H. Chan School of Public Health, focusing on advanced trial emulation methods.

Funding

NIHR Exeter BRC.

People Involved

Project Lead: Laura Güdemann
BRC Theme Leads: Prof Inês Barroso, Prof Andrew Hattersley
Line Manager: Prof Angus Jones
Other collaborators: Prof Beverley Shields, Prof John Dennis