BACKGROUND: Most studies indicate an inverse relationship between physical activity (PA) and body mass index (BMI). However, the impact of obesity on this relationship is unclear.
OBJECTIVE: To scrutinize the BMI/PA relationship by analysing multiple categories of PA from a sample with a wide BMI range.
DESIGN: PA was measured with accelerometry for 7 consecutive days during free-living conditions in 85 severely obese outpatients (mean BMI 42.7 kg/m(2) (s.d. 6.1); age 43.0 year (12.6)) and 193 control subjects (24.0 kg/m(2) (3.5); 41.6 year (13.0)). Six categories of PA were calculated from the accelerometer data (min/day of sedentary time, min/day of light PA, min/day of moderate PA, min/day of vigorous PA, activity counts/day and steps/day). Participants were stratified in obese and non-obese subgroups (BMI=30 kg/m(2) as cutoff). Associations between BMI and PA were examined in the total sample, and in subgroups. The impact of sex and age on the BMI/PA association was tested.
RESULTS: In the total sample, the association between BMI and PA was significant in all PA categories except for time spent sedentary (P=0.68). However, in subgroup analyses, the association between BMI and PA in non-obese was only significant for activity counts/day (r=-0.16, P<0.05) and vigorous intensity PA (r=-0.15, P=0.05). After adjustment for age, vigorous PA remained significantly associated with BMI in the non-obese (r=-0.17, P<0.05). In obese individuals, significant associations between BMI and PA were found for all six PA categories (age adjusted), sedentary time (r=0.26, P=0.05), light PA (r=-0.30, P<0.01), moderate PA (r=-0.35, P<0.01), vigorous PA (r=-0.39, P<0.001), activity counts/day (r=-0.50, P<0.001) and steps/day (r=-0.54, P<0.001).
CONCLUSION: The association between PA and BMI was weak in non-obese individuals. In contrast, BMI was highly significantly associated with PA in obese individuals. Longitudinal studies are needed to tease out the direction of association between PA and BMI across BMI categories, as the cross-sectional associations seem to be dependent on obesity status.
2007. Vol. 31, no 4, p. 663-8