Objectives: Cross-sectional data suggest a strong association between low levels of physical activity and obesity. The EPIC-PANACEA (European Prospective Investigation into Cancer-Physical Activity, Nutrition, Alcohol, Cessation of Smoking, Eating out of home And obesity) project was designed to investigate the associations between physical activity and body mass index (BMI) and waist circumference based on individual data collected across nine European countries.
Methods: In the European Prospective Investigation into Cancer and Nutrition (EPIC), 519 931 volunteers were recruited between 1992 and 2000, of whom 405 819 had data on main variables of interest. Height, body weight and waist circumference were measured using standardized procedures. Physical activity was assessed using a validated four-category index reflecting a self-reported usual activity during work and leisure time. The associations between physical activity and BMI and waist circumference were estimated using multilevel mixed effects linear regression models, adjusted for age, total energy intake, smoking status, alcohol consumption and educational level.
Results: A total of 125 629 men and 280 190 women with a mean age of 52.9 (s.d. 9.7) and 51.5 (s.d. 10.0) years, respectively were included. The mean BMI was 26.6 kg/m(2) (s.d. 3.6) in men and 25.0 kg/m(2) (s.d. 4.5) in women. Fifty percent of men and 30% of women were categorized as being active or moderately active. A one-category difference in the physical activity index was inversely associated with a difference of 0.18 kg/m(2) in the mean BMI (95% confidence interval, CI, 0.11, 0.24) and 1.04-cm (95% CI 0.82, 1.26) difference in waist circumference in men. The equivalent figures for women were 0.31 kg/m(2) (95% CI 0.23, 0.38) and 0.90 cm (95% CI 0.71, 1.08), respectively.
Conclusions: Physical activity is inversely associated with both BMI and waist circumference across nine European countries. Although we cannot interpret the association causally, our results were observed in a large and diverse cohort independently from many potential confounders.