Abstract 5013: Estimating the Reduction in Cardiovascular Disease Mortality Risk Due to Long Term Physical Activity in the Framingham Heart Study
Background: Physical activity is known to prevent chronic disease, with long term physical activity considered to be more preventive of CVD death than short term exposure. However, prospective research studies typically measure physical activity only at a single initial time point and implicitly assume it remains constant over the duration of the study, even though it is well known that there is substantial variation over a person’s life time.
Objective: To identify the reduction in CVD mortality risk associated with long term physical activity in adulthood using contemporary statistical methods.
Methods: Marginal structural models (MSM) are a class of models that allow for the unbiased estimation of the effect of cumulative exposure on an outcome in the presence of time dependent confounding by factors that are both consequences of previous exposure and predictive of future exposure. Standard methods, such as Cox regression, produce biased estimates in these situations whether or not the time dependent confounders are adjusted for. We applied MSMs to estimate the effect of long term exposure to physical activity on CVD death controlling for time dependent confounding.
Results: The estimated hazard ratio for CVD death for continuous physical activity compared to continuous inactivity was 0.710 [95% CI: 0.575, 0.877]. However, the estimated ratio for recent physical activity compared to no recent activity, is 0.727 [95% CI: 0.633, 0.836].
Conclusions: Using measures of physical activity spanning 40 years from the Framingham Heart Study, the effect of continuous physical activity on CVD mortality differed little to that estimated using short term recent physical activity. This suggests that recency, rather than duration, is the key component in the preventive effect. Further investigation using other cohorts and applying similar methodology is needed to validate this finding.