The effects of changing exercise levels on weight and age-related weight gain

Abstract

Objective: To determine prospectively whether physical activity can prevent age-related weight gain and whether changing levels of activity affect body weight.

Design/subjects: The study consisted of 8,080 male and 4,871 female runners who completed two questionnaires an average (+/-standard deviation (s.d.)) of 3.20+/-2.30 and 2.59+/-2.17 years apart, respectively, as part of the National Runners' Health Study.

Results: Changes in running distance were inversely related to changes in men's and women's body mass indices (BMIs) (slope+/-standard error (s.e.): -0.015+/-0.001 and -0.009+/-0.001 kg/m(2) per Deltakm/week, respectively), waist circumferences (-0.030+/-0.002 and -0.022+/-0.005 cm per Deltakm/week, respectively) and percent changes in body weight (-0.062+/-0.003 and -0.041+/-0.003% per Deltakm/week, respectively, all P<0.0001). The regression slopes were significantly steeper (more negative) in men than women for DeltaBMI and Delta%body weight (P<0.0001). A longer history of running diminished the impact of changing running distance on men's weights. When adjusted for Deltakm/week, years of aging in men and years of aging in women were associated with increases of 0.066+/-0.005 and 0.056+/-0.006 kg/m(2) in BMI, respectively, increases of 0.294+/-0.019 and 0.279+/-0.028% in Delta%body weight, respectively, and increases of 0.203+/-0.016 and 0.271+/-0.033 cm in waist circumference, respectively (all P<0.0001). These regression slopes suggest that vigorous exercise may need to increase 4.4 km/week annually in men and 6.2 km/week annually in women to compensate for the expected gain in weight associated with aging (2.7 and 3.9 km/week annually when correct for the attenuation due to measurement error).

Conclusions: Age-related weight gain occurs even among the most active individuals when exercise is constant. Theoretically, vigorous exercise must increase significantly with age to compensate for the expected gain in weight associated with aging.

Figure 1

Total mean changes (±s.e. represented by bars) in BMI, %body weight and waist circumference by change in weekly running distance in male and female runners after adjustment for Δ age and mean age. Significance levels are coded *P<0.05; †P<0.01; ‡P<0.001 and §P<0.0001. The trend for an inverse relationship between Δkm/week and changes in BMI, Δ%body weight and waist circumference were all significant at P<0.0001.

Figure 2

Total change in BMI, %body weight and waist circumference per Δkm/week in male runners by the number of years run at 19 or more kilometers per week after adjustment for Δ age and mean age. Significance levels are coded *P<0.05; †P<0.01; ‡P<0.001 and §P<0.0001. The trend for an inverse relationship between the slopes and the number of years run were all significant at P<0.0001.

Figure 3

Change in BMI and waist circumference per Δkm/week in male runners by the their percentage below greatest lifetime weight on the first survey. Slopes all significantly different from zero at P<0.0001.

Figure 4

Annual increases in BMI, Δ%body weight and waist circumference in men and women who remained within ±8 km/week of their baseline running distance by average running distance. Bars represent ± one s.e. Significance levels are coded *P<0.05; †P<0.01; ‡P<0.001 and §P<0.0001.