Glycosylated hemoglobin level, race/ethnicity, and cognition in midlife and early old age.

Empirical evidence linking racial/ethnic differences in glycosylated hemoglobin levels (HbA1c) to cognitive function in midlife and early old age is limited. We use biomarker data from the Health and Retirement Study (HRS, 2006-2014), on adults 50-64 years at baseline (57-73 years by 2014), and fit multinomial logistic regression models to assess the association between baseline HbA1c, cognitive function (using Langa-Weir classifications) and mortality across 8-years. Additionally, we test for modification effects by race/ethnicity. In age- and sex-adjusted models high HbA1c level was associated with lower baseline cognition and higher relative risk ratios (RRR; vs. normal cognition) for cognitive impairment no dementia (CIND; RRR= 2.3; 95%CI=[1.38;3.84]; p<0.01), and dementia (RRR= 4.00; 95%CI=[1.76;9.10]; p<0.01). Adjusting for sociodemographic, behavioral risk factors, and other health conditions explained the higher RRR for CIND and attenuated the RRR for dementia by approximately 30%. HbA1c levels were not linked to the slope of cognitive decline, and we found no evidence of modification effects for HbA1c by race/ethnicity. Targeting interventions for glycemic control in the critical midlife period can protect baseline cognition and buffer against downstream development of cognitive impairment. This can yield important public health benefits and reductions in burdens associated with cognitive impairment, particularly among race/ethnic minorities who are at higher risk for metabolic diseases.