Risk equations to predict life expectancy of people with Type2 diabetes mellitus following major complications: A study from Western Australia

Abstract

Aims To develop a model for predicting life expectancy following major diabetes-related complications and to summarize these results by age and gender in the form of a simple table. Methods Equations for forecasting mortality were derived using an observational cohort of 12792 patients who had one of the following complications of diabetes: myocardial infarction, stroke, heart failure, amputation or renal failure, recorded in administrative health and mortality data from the state of Western Australia between 1990 and 1999. Logistic regression was used to estimate mortality within the first month post-event and a Gompertz proportional hazards model was used to estimate survival over the patients' remaining lifetime. After examining the internal validity over a 5-year period, these equations were used to estimate remaining life expectancy by age and sex following specific complications. Results Of the complications examined, renal failure had most impact on life expectancy at all ages, followed by heart failure; the best prognosis was following stroke, myocardial infarction and amputation. For a 60-year-old male, life expectancy immediately post-event ranged from 10.1years (95% CI 9.4-10.8years) for stroke to 4.3years (95% CI 3.1-6.1years) for renal failure. Life expectancies for women at 60 and 70years of age were significantly lower than men following myocardial infarction and significantly higher than men following heart failure and amputation at 70 and 80years of age. Conclusion The model allows estimation of both survival probability and life expectancy post-event for men and women of any age. The summary table may provide a useful and simple reference for clinicians and diabetes specialists.