A Prediction Model for Identifying Type 2 Diabetics at Highest Risk of Cardiorenal Outcomes

Abstract

Background Type 2 diabetes mellitus (T2DM) affects around 10% of the US adult population and associates with high levels of cardiovascular (CV) and renal morbidity and mortality. Recent randomized clinical trials in T2DM have shown cardiorenal benefits with novel antidiabetic therapies. Real world utilization of these therapies may be maximally cost-effective when targeted to those at highest risk of cardiorenal outcomes. Accordingly, this study created a risk prediction model for a composite cardiorenal endpoint of CV death, hospitalization for heart failure (HHF), and end stage renal disease (ESRD) in T2DM patients and proposes a quantitative risk threshold for labeling patients as being at high risk of this outcome. Methods This retrospective cohort study included patients with pre-existing or newly diagnosed T2DM drawn from the electronic medical record (EMR) of a single, large integrated health care system. Candidate predictors for prediction model development were determined from EMR documentation in the two years prior to a baseline office visit. A time-to-first-event variable was created for the composite endpoint of CV death, HHF, and ESRD, and significant predictors determined using Cox regression. “High risk” patients were defined as the 10% highest risk patients according to prediction model output. Results Among 57,646 T2DM patients, 8505 (15%) experienced an event over a mean follow-up of 6.9 years. CV death was the most frequent outcome (53% of first events), followed by HHF (25%) and ESRD (22%). The strongest predictors of the composite outcome were age, blood urea nitrogen, coronary artery disease, HF, and atrial fibrillation. The final proposed 19-predictor model had a c-statistic of 0.792. The 10% highest risk patients (“high risk”) according to prediction model variables had estimated event rates >5%, >15%, and >30% at 1, 3, and 5 years after the baseline date. Conclusions T2DM patients have a high rate of CV and renal outcomes, and recent clinical trial evidence shows novel antidiabetic therapies reduce these outcomes. Patients at highest risk for these outcomes are expected to receive the greatest event rate reduction from these novel therapies.