**Background:** The COVID-19 pandemic led to social restrictions that negatively altered health behaviors, particularly physical activity. Physical inactivity and unhealthy eating may have contributed to weight gain and affected glycemic levels. Glycated hemoglobin (HbA1c) is the gold standard for assessing glycemic control over the preceding 120 days. This study aimed to evaluate serum HbA1c levels and their association with physical inactivity during leisure time, hypothesizing that overweight may mediate this association.
**Methods:** This cross-sectional study used data from the COVID-Inconfidentes project, a population-based household survey conducted between October and December 2020 in Ouro Preto and Mariana, Minas Gerais, Brazil. Stratified, multistage probability cluster sampling was used. Eligible participants were permanent residents aged ≥18 years. The final sample included 1,685 individuals. HbA1c was measured by immunoturbidimetry (COBAS INTEGRA 400 plus). Normal HbA1c was defined as ≤6.4%; high levels as ≥6.5%. Physical activity during leisure time was self-reported and classified per WHO guidelines: active (≥150 min/week moderate or ≥75 min/week vigorous) or inactive. Overweight was defined by BMI (≥25 kg/m² if <60 years; ≥28 kg/m² if ≥60 years). Covariates included sex, age, race/ethnicity, family income, education, tobacco use, and reported morbidities. A directed acyclic graph (DAG) was used to select confounders. Mediation analysis used the Karlson–Holm–Breen (KHB) method. Analyses accounted for complex sampling design using Stata 15.1.
**Key Results:** The sample was predominantly women (52.4%), aged 35–59 years (45.8%), married (52.9%), brown race/ethnicity (48.1%), with <9 years schooling (69.1%), income ≤2 minimum wages (41.2%), and overweight (56.5%). Mean HbA1c was 5.68% (95% CI: 5.58–5.77). Normal HbA1c was found in 92.9% (95% CI: 90.6–94.7), while 7.1% (95% CI: 5.3–9.4) had glycemic changes. Physical inactivity during leisure time was reported by 69.2% (95% CI: 64.2–73.7). In multivariate logistic regression adjusted for sex, age, race/ethnicity, family income, and reported morbidity, physically inactive individuals were 2.62 times more likely to have high HbA1c (OR: 2.62, 95% CI: 1.31–5.24). Mediation analysis showed that physically inactive individuals had 2.62 times greater odds of high HbA1c (OR: 2.62, 95% CI: 1.29–5.33), and 26.87% of this effect was mediated by overweight (OR: 1.30, 95% CI: 1.06–1.57).
**Clinical Implications:** Approximately 70% of the study population did not meet minimum physical activity guidelines during the pandemic. Physical inactivity was strongly associated with glycemic changes, and over a quarter of this effect was explained by overweight. These findings suggest that promoting regular physical activity is essential for glycemic control, acting both through weight management and independent metabolic mechanisms (e.g., GLUT-4 translocation, improved insulin sensitivity). The results are particularly relevant for public health strategies during periods of restricted mobility, as physical inactivity during the pandemic may have lasting impacts on diabetes incidence. Limitations include self-reported physical activity (recall bias), lack of dietary data, and the cross-sectional design precluding causal inference. Strengths include robust probability sampling, face-to-face interviews, and objective HbA1c measurement.