In Dr. Buchanan's excellent review (1), he presents a novel approach to analysis of diabetes prevention trial outcomes based on “absolute differences” in the percentage of intervention versus control subjects who develop diabetes. In his analysis, rising absolute differences in the conversion from pre-diabetes to diabetes reflect “prevention,” while stable absolute differences reflect “delay/masking.” I believe that readers will find this difficult to understand, and I suggest an alternative that may be more intuitive. If the percentage of patients with pre-diabetes who convert to diabetes is plotted over time, an intervention that prevented diabetes in 100% of subjects would have no conversions to diabetes and a slope of zero, which is intuitively obvious and easy to teach. Accordingly, the success of an intervention in preventing diabetes can be expressed as the ratio of the conversions to diabetes with intervention versus control, such as the following: percentage of effectiveness = [1 − (conversion with intervention/conversions with control)] × 100. With this analysis, for example, the Diabetes Prevention Study and the Diabetes Prevention Program would have shortterm percentages of effectiveness that were relatively constant and averaged 58%. In contrast, metformin and acarbose provided percentages of effectiveness that were not constant and decreased progressively over time, which is the intuitive expectation for delay/masking. This analysis also has the merit of providing values that correspond to analytical hazard ratios. For examples, see Fig. 1.
Model examples of interventions aimed to prevent progression from pre-diabetes to diabetes. Shown are plots for percent conversion to diabetes and effectiveness for diabetes prevention with interventions that are 100% effective and 50% effective and an intervention that delays (masks) but does not prevent progression to diabetes.
Model examples of interventions aimed to prevent progression from pre-diabetes to diabetes. Shown are plots for percent conversion to diabetes and effectiveness for diabetes prevention with interventions that are 100% effective and 50% effective and an intervention that delays (masks) but does not prevent progression to diabetes.
Article Information
This work was supported in part by grants DK066204 and RR00039.