Glycemic variability and dementia risk in a diverse cohort of older adults with type 2 diabetes

Abstract

AbstractBackgroundType 2 diabetes is associated with a two‐fold increase in dementia risk. Although glycemic variability is becoming increasingly recognised as contributing to end‐organ damage, less is known about its contribution to dementia risk. We aimed to study the association between HbA1c variability and dementia risk in a large, diverse sample of people in Northern California.MethodOver 209,000 members of a fully integrated health care delivery system with type 2 diabetes aged ≥ 50 years were followed from 1996 to 2018 to obtain repeated measures of HbA1c, dementia diagnoses, and comorbidities. Variability was measured using standard deviation (SD), coefficient of variation, and average real variability. Cox proportional hazards models estimated the association between HbA1c variability in the first 3 years of cohort entry and incident dementia at least 2 years after the end of the exposure period adjusting for age (as timescale) and baseline health conditions.ResultThere were a total of 209,703 people in the cohort (mean baseline age 61.0 years (SD = 9.2), 47.5% women) (Table1). Approximately 49.6% of the sample were non‐Latino White. The mean number of HbA1c values available for each participant was 5.8 (SD = 2.3). The mean HbA1c for each participant in the first 3 years was 7.4% (SD = 1.33). Greater HbA1c variability was associated with greater hazard of dementia when measured using SD (aHR = 1.12 [95%CI: 1.10, 1.14]), coefficient of variation (aHR = 2.17 [95%CI: 1.85, 2.55]) (Figure1), or average real variability (aHR = 1.06 [95%CI: 1.06, 1.07]). When stratified by mean HbA1c, we found that ‐ relative to the lowest quintile of variation ‐ increasing variability of SD of HbA1c was associated with greater hazard of dementia in HbA1c categories of <6% (p = 0.0004) and 6 ‐8% (p<0.0001) but not in those with a mean HbA1c ≥8% (p = 0.42).ConclusionGreater HbA1c variability is associated with greater dementia hazard in those with good glycemic control. The absence of association in less well controlled (HbA1c ≥8%) patients may stem from greater variability including lower risk periods of optimal glycemic control. These results support current clinical recommendations to minimize glycemic variability and extend known benefits to include lowering of dementia risk.