Predicting the change in incidence of T2DM with change in BMI The correlation between BMI and the annual incidence of T2DM was calculated based on two large American prospective epidemiological studies, the Nurses Health Study and the Health Professionals Follow-up Study.36 For men with BMI=35 kg / m2, the annual incidence of diabetes was estimated as 1.4% and for men with BMI=25 kg / m2 as 0.04%. For women, the equivalent figures were 0.61% and 0.13%. Based on these data, it was estimated that a 10% reduction in BMI should reduce the annual incidence of diabetes by approximately 30%. As the change in the annual incidence of T2DM in our model is directly linked to a change in weight loss, the difference in annual incidence of T2DM between the two arms will disappear when patients regain their initial weight. The temporary reduction in the annual incidence will effect the cumulative incidence of T2DM because it reduces / delays the number of patients who develop diabetes. We assumed that the differences in cumulative incidence of T2DM would last for 10 y. Therefore a time horizon of 11 y was used for effects on diabetes costs in the model.
Sustained weight loss As weight loss affects both the utility gained and incidence of T2DM, it is important to be clear about the sustainability of weight loss following the 12-month treatment period. For the base-case analysis, we assumed that patients achieved their maximum weight loss at 12 months and regained their original weight at a uniform rate over the following 3 y (ie in line with NICE recommendation)