Background: We recently reported that long-term fasting plasma glucose (FPG) instability predicts all-cause mortality in elderly patients with non-insulin-dependent diabetes mellitus (NIDDM). The aim of the present study was to evaluate whether glucose instability, as represented by the coefficient of variation of FPG concentrations (CV-FPG) measured during a 3-year period, can predict specific causes of death in the subsequent 5 years.
Methods and results: Five hundred sixty-six elderly patients with NIDDM were followed up for 5 years to assess mortality and causes of death. All FPG determinations of the 3 years preceding the follow-up available in the clinical records were collected and analyzed. Patients were grouped in tertiles of mean FPG, CV-FPG, and the slope of FPG. These parameters of glucose control, as well as sex, age, duration of diabetes, insulin treatment, cigarette smoking, hypertension, and total cholesterol, were included in a multivariate analysis of mortality. During the follow-up, 63 men and 128 women died. Diabetes- and malignancy-related mortality were not independently associated with any parameter of glucose control, whereas cardiovascular-related mortality was independently associated with CV-FPG (P=.007) but not with the mean or the slope of FPG. In particular, the relative risk of cardiovascular mortality in subjects in tertile III versus tertile I of CV-FPG was 2.40 (95% CI, 1.28 to 4.53).
Conclusions: These results indicate that FPG instability is a predictor of cardiovascular-related mortality in elderly patients with NIDDM and suggest that glucose stability might be a goal in the management of these patients.