Validating the UK prospective diabetes study outcome model 2 using data of 94,946 Israeli patients with type 2 diabetes

Abstract

Aims: To externally validate the United Kingdom Prospective Diabetes Study (UKPDS) Outcome Model 2 (OM2) in contemporary Israeli patient populations.

Methods: De-identified patient data on demographics, time-varying risk factors, and clinical events of newly diagnosed type 2 diabetes patients were extracted from the Maccabi Healthcare Services (MHS) diabetes registry over years 2000-2013. Depending on the baseline risk, patients were categorized into low-risk and intermediate-risk groups. In addition to assessing discriminatory performance, the predicted and observed 15-year cumulative incidences of diabetes complications and death were compared among all patients and for the two risk-groups.

Results: The discriminatory capability of OM2 was moderate to good, C-statistic ranging 0.71-0.95. The model overpredicted the risk for MI, blindness and death (Predicted/Observed events (P/O: 1.32-2.31)), and underpredicted the risk of IHD (P/O: 0.5). In patients with a low baseline risk, overpredictions were even more pronounced. OM2 performed well in predicting renal failure and ulcer risk in patients with a low risk but predicted well the risk of death, stroke, CHF, and amputation in patients with an intermediate risk.

Conclusion: OM2 demonstrated good to moderate discrimination capability for predicting diabetes complications and mortality risks in Israeli diabetes population. The prediction performance differed between patients with different baseline risks.

Keywords: Diabetes complications; External validation; Prediction; Risk equations; Type 2 diabetes.