NDPH researchers have helped revise a long-term forecasting tool so that it more accurately reflects disease progression in individuals with type 2 diabetes.
Computer simulations are often used by healthcare providers to make long-term predictions about outcomes for individuals with a disease. These models are usually based on associations between biochemical and/or physical risk factors and disease outcomes, which themselves are calculated from large-scale, observational studies. These are overlaid by the predicted effects of the treatments and interventions that the individual receives.
For type 2 diabetes, one of the most widely used simulation models in the UK is the United Kingdom Prospective Diabetes Study Outcomes Model, Version 2 (UKPDS-OM2). It is mainly used to predict the occurrence of major diabetes-related complications (such as heart disease, stroke, amputation, kidney failure and blindness) and death in patients diagnosed with type 2 diabetes. However, it has also been applied to evaluate the cost-effectiveness of diabetes interventions, and in resource planning for healthcare services.
A key challenge with the UKPDS-OM2, and indeed most health-related computer simulations, is developing equations to accurately reflect how risk factors change for type 2 diabetes patients, whose metabolism may deteriorate over time, or improve if a treatment starts to take effect. Until now, the equations in the UKPDS-OM2 model have assumed that risk factors stay constant over time.
To address this, researchers from NDPH’s Health Economics Research Centre (HERC) led a study to calculate a revised set of equations for the UKPDS-OM2. These take into account how the magnitude of risk factors may change over time, including as a result of effective diabetes treatments. The results have been published today in Diabetic Medicine.
The new equations are based on data from the original UK Prospective Diabetes Study: a clinical trial evaluating different glycaemic and blood pressure management regimens for individuals newly diagnosed with type 2 diabetes. A wide range of anti-diabetes therapies were investigated by the study, including metformin, sulphonylureas and insulin. The study data included 24 years of follow-up data from over 5,100 participants, with up to 65,252 person-years of data for each clinical outcome.
The research team used dynamic statistical modelling techniques to develop new equations for 13 clinical risk factors (including systolic blood pressure, LDL- and HDL-cholesterol, BMI, and heart rate). The results showed that the predictive power of the new risk factor equations was significantly improved compared with when these risk factors were assumed to remain constant. For instance, the ability to accurately predict the incidence of stroke improved by 4.4 times.
Dr José Leal, who led the study, said: ‘These revised equations give modellers and the wider research community a useful additional tool when trying to simulate the long-term effects of type 2 diabetes and its therapies. Potentially, our approach of using dynamic risk factors may also be useful to those developing simulation models for other diseases, such as chronic kidney failure.’