Leisure time and occupational physical activity in relation to obesity and insulin resistance: a population-based study from the Skaraborg Project in Sweden.

Abstract: The objective was to study obesity and insulin resistance in relation to leisure time physical activity (LTPA) and occupational physical activity (OPA) in a Swedish population, with particular focus on sex differences. Using a cross-sectional design, waist circumference, body mass index (BMI), glucose/insulin metabolism, blood pressure, heart rate, self-reported education, smoking, alcohol consumption, LTPA, and OPA were assessed in 1745 men and women (30-74 years) randomly chosen from 2 municipalities in southwestern Sweden. In both men and women, LTPA was inversely associated with BMI, waist circumference, and the homeostasis model assessment of insulin resistance (HOMA-IR), respectively. These associations remained statistically significant after adjustments for age, OPA, education, alcohol consumption, smoking, and study area, and also for BMI in the analyses concerning waist circumference and HOMA-IR. A statistically significant interaction term (P = .030), adjusted for multiple confounders, revealed a stronger association between LTPA and HOMA-IR in women compared with men. Occupational physical activity was positively associated with BMI (P < .001), waist circumference (P < .001), and HOMA-IR (P = .001), however, only in women. These associations remained when adjusting for multiple confounders. The sex differences were confirmed by statistically significant interaction terms between sex and OPA in association with BMI, waist circumference, and HOMA-IR, respectively. The observed sex differences regarding the strength of the association between LTPA and insulin resistance, and the positive association between OPA and obesity and insulin resistance found solely in women, warrant further investigation. Although exploration of the metabolic effects of OPA appears to be needed, thorough measurement of potential confounders is also vital to understand contextual effects. [Copyright &y& Elsevier]