Abstract

OBJECTIVE: The prognostic value of long-term glycemic variability is incompletely understood. We evaluated the influence of visit-to-visit variability (VVV) of fasting blood glucose (FBG) on incident cardiovascular disease (CVD) and mortality. RESEARCH DESIGN AND METHODS: We conducted a prospective cohort analysis including 4,982 participants in the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT) who attended the baseline, 24-month, and 48-month visits. VVV of FBG was defined as the SD or variability independent of the mean (VIM) across FBG measurements obtained at the three visits. Participants free of CVD during the first 48 months of the study were followed for incident CVD (coronary heart disease [CHD], stroke, and heart failure [HF]) and all-cause mortality. RESULTS: Over a median follow-up of 5 years, there were 305 CVD events (189 CHD, 45 stroke, and 81 HF) and 154 deaths. The adjusted hazard ratio (HR) comparing participants in the highest versus lowest quartile of SD of FBG (≥26.4 vs. <5.5 mg/dL) was 1.43 (95% CI 0.93-2.19) for CVD and 2.22 (95% CI 1.22-4.04) for all-cause mortality. HR for VIM was 1.17 (95% CI 0.84-1.62) for CVD and 1.89 (95% CI 1.21-2.93) for all-cause mortality. Among individuals without diabetes, the highest quartile of SD of FBG (HR 2.67 [95% CI 0.14-6.25]) or VIM (HR 2.50 [95% CI 1.40-4.46]) conferred a higher risk of death. CONCLUSIONS: Greater VVV of FBG is associated with increased mortality risk. Our data highlight the importance of achieving normal and consistent glycemic levels for improving clinical outcomes.