1520-P: Developing New Risk Equations to Predict Diabetes-Related Complications and Mortality in U.S. Adults with Type 2 Diabetes

Risk equations to predict diabetes-related complications and mortality in most type 2 diabetes simulation models were based on data from the UK Prospective Diabetes Study. Two recently completed U.S. diabetes trials, the Action to Control Cardiovascular Risk in Diabetes and the Action for Health in Diabetes and their follow-up studies with \u003e15,000 participants and \u003e7 years follow-up, provide a unique source for developing new risk equations. We estimated parametric hazard models for cardiovascular, neuropathy, nephropathy, vision, and hypoglycemia complications and mortality in persons with type 2 diabetes using longitudinal data from the trials. We included demographic variables, time-varying risk factors, and histories of complications. Variable selection was based on backwards stepwise regression and expert opinion. We estimated 17 complication and 3 mortality equations. HbA1c was selected in all but one complication/mortality equation, with hazard ratios ranging from 1.09 for stage 3 chronic kidney disease to 1.85 for amputation per 1% increase. Other risk factors were selected less frequently. For mortality, smoking was the risk factor with the largest hazard ratio. At least one lagged complication was selected for every complication/mortality equation. Previous congestive heart failure was associated with increased risk for 13 complications, with hazard ratios as high as 2.6 for amputation. Other lagged complications selected for over half of the complications included microalbuminuria, previous revascularization, macroalbuminuria, and previous stroke. Congestive heart failure, myocardial infarction, and stroke increased mortality in persons with current or previous cardiovascular disease. Kidney-related complications also increased mortality. Our equations can be used in type 2 diabetes simulation models to estimate costs, outcomes, and cost-effectiveness of interventions for type 2 diabetes, particularly in U.S.-based models. Disclosure T.J. Hoerger: Research Support; Self; Centers for Disease Control and Prevention, Centers for Medicare and Medicaid Services. M. Kaufmann: None. S. Neuwahl: None. H. Shao: None. H. Chen: None. M. Laxy: None. Y.J. Cheng: None. S.R. Benoit: None. A.M. Anderson: None. T. Craven: None. P. Zhang: None. Funding Centers for Disease Control and Prevention (200-2016-92270)