Glycated hemoglobin (HbA1c) is an important surrogate measure of glycemic control in patients with diabetes and is a key risk factor for many diabetes-related complications. As a result, HbA1c plays an important role in many long-term health economic models. The aim of the present analysis was to evaluate the importance of realistically simulating HbA1c progression over time in patients with type 1 diabetes in a health economic model.

The PRIME Diabetes Model, a long-term, externally audited and validated, patient-level simulation model of type 1 diabetes was used to model long-term clinical and cost outcomes. Scenarios were based on either a linear assumption for HbA1c progression, or a target-driven HbA1c model, capturing covariance, developed from patient-level data from the Diabetes Control and Complications Trial (DCCT). Parameters significantly covarying with baseline and subsequent HbA1c were incorporated into covariance matrices in the target-driven model. The model used age and recent severe hypoglycemic episodes to derive patient-specific HbA1c targets. Costs were reported in 2016 pounds sterling.

Simulating HbA1c progression based on patient-level data was shown to affect the projected cumulative incidence of diabetes-related complications, life expectancy, quality-adjusted life expectancy and the cost of complications versus the standard linear approach. Quality-adjusted life expectancy was 0.18 QALYs higher with simulated HbA1c progression. The reduction in diabetes-related complications projected with simulated HbA1c progression decreased overall direct costs per patient by GBP 3,062 over patient lifetimes.

Long-term projections using the PRIME Diabetes Model indicate that realistically simulating the progression of important risk factors, such as HbA1c, over time can influence the outcomes of a health economic analysis compared with standard linear assumptions. Simulating risk factor progression, informed by analysis of patient-level data, may directly influence the outcomes of economic evaluations in diabetes and should be taken into consideration by modelers and decision-makers.