Purpose: To internally validate the EAGLE (Economic Assessment of Glycemic control and Long-term Effects) model version 2.0 according to current standards of validation practice. The EAGLE model simulates long-term diabetes-related complications and related costs for type 1 and type 2 diabetes. Published results from large interventional studies (DCCT, WESDR, and UKPDS) serve as the basis for EAGLE calculations.

Methods: First-order validation identified inconsistencies in results and corrected programming errors. Second-order validation included the following steps: 1) Simulation sets were created in EAGLE based on baseline data from the studies used to build the model; 2) Simulations were run. The results obtained with EAGLE were compared with published event rates; 3) Risk equations were refined if a deviation of >10% was observed between model-derived and published results. Patient numbers and iterations were systematically changed until a final run was performed with 50,000 patients and 100 iterations.

Results: Fulfilling a criterion for validity, the cumulative incidence per 1000 patient-years and incidence rates for all events simulated with EAGLE fell within the range of ±10%. The difference between published data and model results ranged from 0% to 9% for all patient populations after possible refinements. For example, in type 2 diabetes, EAGLE successfully predicted the end-stage renal disease and fatal event rates reported in UKPDS (deviation = 0%). The rates of severe hypoglycemia differed by 1%. The EAGLE event rates for proliferative and nonproliferative retinopathy corresponded well with event rates derived from the WESDR publications (deviation = 3% and 4%, respectively).

The use of data from different sources (main sources: UKPDS, DCCT, WESDR) in the risk equations may explain some of the deviations between EAGLE and the published results. Without access to original study data only published data could be used. In addition, it was not always possible to create exactly the same patient population as in the publications because of missing data.

Conclusions: The EAGLE model consistently predicts event rates reported by UKPDS, WESDR, and DCCT, for both type 1 and type 2 diabetes patients. The EAGLE model is a valid and robust tool for the analysis of the long-term diabetes-related complications and related costs in type 1 and type 2 diabetes.