Decomposing the effects of physical activity and cardiorespiratory fitness on mortality

Abstract

<p><h3 id="x-x-st0010">Background</h3> <p id="x-x-sp0025">Characterizing the effects of physical activity (PA) and cardiorespiratory fitness (CRF) on mortality is challenging because the causal relationship between PA, CRF, and other cardiovascular risk factors is unclear. <h3 id="x-x-st0015">Methods</h3> <p id="x-x-sp0030">To better understand the effects of PA and CRF on mortality, we re-analyzed data from 42,373 participants in the Aerobics Center Longitudinal Study (ACLS) using a modified version of VanderWeele's four-way causal effect decomposition method. The method was applied to decompose the causal effects of PA and CRF on median time to death into parts reflecting mediation, interaction, mediated interaction, and neither interaction nor mediation. <h3 id="x-x-st0020">Results</h3> <p id="x-x-sp0035">We found that 67% of the effect of PA on mortality was mediated by CRF, while the effect of CRF was not significantly mediated by PA. The effects of both PA and CRF were mediated to a small extent by hypertension and diabetes. There were no meaningful interactions. <h3 id="x-x-st0025">Conclusions</h3> <p id="x-x-sp0040">Our findings strengthen the evidence that the benefit on mortality from PA is largely mediated by its effect on CRF, and support efforts to increase longevity by encouraging PA</p>