DIABETES AS A PREDICTOR OF MORTALITY IN A COHORT OF BLIND SUBJECTS

Background, There is only a little information about survival in newly registered blind subjects. Methods. A closed cohort of blind subjects (n = 2680, 1803 of them women), newly registered between 1990 and 199 3 in the district of Wurttemberg-Hohenzollern, Germany, was observed f or up to 48 months. Mortality was compared to that of the general popu lation. Predictors of mortality within that cohort were identified by Cox proportional hazards regression analysis. Results. Before 1 Februa ry 1994, 582 of the subjects had died. Diabetes had been diagnosed in 772 of the subjects, 226 of them died. The overall incidence rate of d eath was 12 179 per 100 000 per year. The probability of survival afte r 47 months was 0.64 (95% confidence interval (CI) :0.59-0.70) in the non-diabetic, and 0.46 (95% CI :0.37-0.55) in the diabetic subjects. P redictors of mortality in the regression model were age (risk ratio [R R] per year of age 1.047), sex (RR for men 1.247) and diabetes (RR whe n blindness was unrelated to diabetes: 1.448, RR when diabetes was the only cause of blindness: 2.253). Compared with the entire population, mortality was considerably increased in the blind cohort (comparative mortality figure [CMF] 4.79), particularly in individuals with diabet es (CMF = 6.55). The relative risks decreased with increasing age. Con clusions. Overall mortality in this cohort was high, even higher than in previous studies on the mortality of the blind. Diabetes increased the risk of death. In addition, the cause of blindness in diabetic ind ividuals was a major predictor of mortality.