Abstract
Background
Increasing physical activity (PA) and reducing sedentary behaviour (SB) have been associated with healthy ageing, but their effects when adjusted for reverse causation and selection bias remain unclear.
Methods
A deficits accumulation (DA) index based on the number and severity of 51 health deficits (0–100%) was calculated at baseline and three biannual follow-up visits, in a representative cohort of 3 228 community-dwelling older adults in Spain. Average differences in DA index by previous recreational PA, household PA, mentally-active SB and passive SB were estimated using marginal structural models with inverse probability of exposure and censoring weights.
Results
Compared with participants with previous recreational PA of 10–19.9 metabolic equivalent hours/week (MET-hours/week), average differences in DA index (95% confidence intervals) were 0.19 (−1.09, 1.48), 0.69 (−0.23, 1.61), −0.66 (−1.34, 0.02), −0.87 (−1.59, −0.13) and −0.55 (−1.37, 0.28) for 0, 0.1–9.9, 20–29.9, 30–39.9 and ≥40 MET-hours/week, respectively (P for trend = 0.006). Household PA showed no effect on subsequent DA after adjusting for reverse causation. Women, but not men, who spent 7–14.9, 15–20.9 and ≥21 h/week on mentally active SB had DA decreases of 0.09 (−1.00, 1.19), 1.08 (−0.28, 2.45) and 2.17 (0.58, 3.75), respectively, compared with 1–6.9 h/week (P for trend = 0.005); whereas women who spent 3–3.9, 4–4.9 and ≥5 h/day on passive SB showed DA increases of 0.41 (−0.52, 1.35), 1.35 (0.13, 2.57) and 2.13 (0.78, 3.47), respectively, compared with 2–2.9 h/day (P for trend = 0.001).
Conclusions
The proposed methodology allows estimation of the causal effects of PA and SB on ageing, by simulating a random assignment in which all subjects have the same probability of exposure.