Measuring Additional Generalised Impacts on Cardiovascular outcomes (MAGIC) model for diabetes

Research question To develop and validate a new economic model to inform current type 2 diabetes (T2D) health policy using patient-level cardiovascular outcomes trials (CVOT) data.

Background Existing economic models cannot accurately predict long-term health outcomes for populations prescribed novel glucose-lowering drugs such as sodium-glucose cotransporter-2 inhibitors (SGLT2i) and glucagon-like peptide-1 receptor agonists (GLP-1 RA). Such models operate through conventional risk factors (i.e.

HbA1c) which cannot fully quantify the cardiovascular benefits of these drugs. This presents a challenge in assessing the economic value of new diabetes interventions (e.g. education programmes).

A new model has the potential to improve outcome predictions for economic evaluations by incorporating drug-mediated effects through a standardised modelling framework.

Aims and objectives This research aims to develop and validate a new economic simulation model capable of simulating health outcomes and costs of modern T2D patients over a lifetime horizon. The objectives are to: Harmonised individual patient-level data from the CVOTs (0-6 months).

Estimate parametric time-to-event and risk factor progression equations using the harmonised trial data to develop an economic model (6-21 months). Use external data to validate and calibrate the economic model (18-30 months). Apply the model to evaluate clinically relevant case studies (24-32 months).

Methods Trials will be selected for analysis based on the completeness and real-world generalizability of the data.

I will harmonise the trial data to predict the time to cardiovascular and kidney complications, adverse events, and death.

Parametric time-to-event models will be estimated for each event, including death, based on time-invariant (e.g. ethnicity) and time-varying (e.g. cholesterol) clinical risk factors, and time-varying complication history. Time-varying risk factor progression will be estimated using autoregressive panel models.

A patient-level microsimulation economic model will be developed using these equations to predict the incidence of complication events and death informed by patient characteristics. Appropriate model fit, clinical guidance, and external data will inform lifetime model extrapolation.

Event-related healthcare costs will be calculated using Clinical Practice Research Datalink data, and health utility will be derived from EQ-5D trial data.

I will validate the model using CVOT data not included in the model development by comparing the predicted and observed event incidence.

The model will be applied to two relevant cost-effectiveness scenarios: (1) intensification to metformin and an SGLT2i for patients without existing cardiovascular disease and/or chronic kidney disease; (2) targeted intensification to a GLP-1 RA with a lifestyle intervention for people with obesity.

My research will be guided by a leading team of supervisors and mentors, active PPI engagement, and close collaboration with NICE stakeholders to ensure the model's relevance to UK policymaking.

Impact and dissemination A new economic model will improve the confidence and accuracy of economic evaluations of diabetes interventions. This will ensure patients receive effective and cost-effective treatments.

A licensed software version model will be developed and made freely available to public organisations, academic institutions, and charity stakeholders.

I will disseminate the findings in clinical and health economic journals, and at conferences including Diabetes UK and the Mount Hood Diabetes Challenge to increase stakeholder awareness.