OBJECTIVES

Rates of renal failure are substantially higher in patients with type 2 diabetes (T2D) than the general population. Given the considerable cost and quality of life implications associated with renal replacement therapies, accurate modeling of renal failure is critical in the evaluation of the cost-effectiveness of novel T2D interventions. The aim of the present study was to compare renal failure projections from the widely-used United Kingdom Prospective Diabetes Outcomes Model 2 (UKPDS OM2) and a novel estimated glomerular filtration rate (eGFR) threshold-based model.

METHODS

The PRIME T2D Model, a high-performance web-based simulation model of T2D, was used to project renal failure incidence over patient lifetimes using the UKPDS OM2 exponential renal failure risk equation, and a novel renal failure model based on a 15 ml/min/1.73 m² eGFR renal failure threshold combined with an eGFR decline model based on a prospective study of 1,682 patients with T2D. Baseline characteristics were taken from the UKPDS, with a mean baseline eGFR of 77.5 ml/min/1.73 m² (standard deviation 15ml/min/1.73 m²).

RESULTS

Over a 50 year time horizon, the cumulative incidence of renal failure was 17.8% with the UKPDS OM2 versus 13.4% with the eGFR threshold-based model. Mean time to onset of renal failure differed substantially between the models with areas under the curves (AUCs) of 6.6 years with the UKPDS OM2 model versus 1.4 years with the eGFR threshold-based model, arising from much later renal failure onset with the eGFR threshold-based model.

CONCLUSIONS

Renal failure models not based exclusively on eGFR can make renal failure incidence projections that are incongruous with the underlying risk factor trajectories, which can have important implications for quality-of-life and survival estimation. The PRIME T2D Model supports multiple renal failure modeling options, including a flexible eGFR threshold-based model that enforces congruity between eGFR and renal failure incidence.