Long-term glycemic variability and risk of adverse health outcomes in patients with diabetes: A systematic review and meta-analysis of cohort studies

Abstract

Aims: To quantify associations of different metrics of long-term glycemic variability (GV) with multiple adverse diabetes-related outcomes.

Methods: We searched PubMed and Embase from database inception to 23 August 2021. GV was based on measurements of HbA1c or fasting plasma glucose (FPG) and calculated by standard deviation (SD), coefficient of variance (CV) or other metrics. Outcomes included mortality, cardiovascular disease (CVD), renal disease, peripheral neuropathy, retinopathy, dementia and cancer. Random-effects meta-analyses were adopted to pool the relative risks (RRs).

Results: Seventy-five articles with 2,051,701 participants were included. When comparing top with bottom quartiles, HbA1c variabilities were associated with all-cause mortality (RR_CV = 1.63, 95 % CI 1.37-1.92; RR_SD = 1.87, 1.55-2.26), CVD (RR_CV = 1.38, 1.07-1.78; RR_SD = 1.34, 1.12-1.59), renal disease (RR_CV = 1.43, 1.18-1.74; RR_SD = 1.44, 1.24-1.67), and peripheral neuropathy (RR_CV = 1.84, 1.40-2.43; RR_SD = 1.98, 1.51-2.61), but not retinopathy. FPG variabilities were associated with all-cause mortality (RR_CV = 1.59, 1.43-1.78; RR_SD = 1.67, 1.26-2.20), renal disease (RR_CV = 1.77, 1.32-2.38), and retinopathy (RR_CV = 1.92, 1.10-3.35), but not CVD and peripheral neuropathy. Associations of GV with Alzheimer's disease (RR_HbA1c-CV = 1.38, 1.13-1.70; RR_FPG-CV = 1.32, 1.07-1.63) and cancer (RR_HbA1c-SD = 2.19, 1.52-3.17; RR_FPG-CV = 3.64, 2.21-5.98) were each found significant in one study.

Conclusions: Long-term GV was associated with multiple adverse diabetes-related outcomes, while the strength of associations varied. The findings support the use of long-term GV for diabetes management in clinical practice.

Keywords: Death; Diabetes mellitus; Glycemic variability; Meta-analysis; Vascular complications.