Smoking Status and HbA1c in U.S. Adults Without Diabetes: A Cross-Sectional National Health and Nutrition Examination Survey (NHANES) Analysis

Background: Cigarette smoking is a recognized risk factor for type 2 diabetes, and prior work has reported higher glycated hemoglobin (HbA1c) levels among smokers. Whether this reflects a direct metabolic effect of smoking or residual confounding by central adiposity, systemic inflammation, and lipid dysregulation remains unclear, and whether the association persists after cessation in adults without diabetes has not been fully characterized. Aim: We examined associations between smoking and HbA1c among United States (U.S.) adults, and whether these associations vary by diabetes status. Methods: We analyzed National Health and Nutrition Examination Survey (NHANES) data from 2015-2018 for adults aged ≥20 years. Smoking was assessed by self-report and serum cotinine. Survey-weighted multivariable linear regression was used to evaluate the association between smoking and HbA1c in the full population (N=9,214) and in adults without diabetes (N=7,328), adjusting for demographics, blood pressure, waist circumference, lipids, and high-sensitivity C-reactive protein (hsCRP). Results: After adjustment for cardiometabolic covariates, there was no statistically significant association between smoking and HbA1c in the full population (former: β=0.029%, p=0.30; current: β=0.053%, p=0.13). Among adults without diabetes, former smoking was not associated with HbA1c, whereas current smoking remained significantly associated (former: β=−0.001%, p=0.923; current: β=0.067%, p<0.001). These findings were similar when cotinine was used as the exposure measure, with active smoking (≥3.0 ng/mL) associated with higher HbA1c among non-diabetic adults (p<0.001), but not in the full population. Conclusions: Among adults without diabetes, current smoking was statistically significantly associated with higher HbA1c, whereas former smoking was not. The formal smoking × diabetes interaction test did not reach statistical significance, so the subgroup pattern should be interpreted as hypothesis-generating. The absence of a significant association in former smokers is compatible with attenuation following cessation. These findings may inform cessation counseling and diabetes screening and warrant replication in larger, adequately powered studies.