Background: Given the increased prevalence of cancer survivors in the United States, it is imperative to define risk factors for potential reductions in total and cause-specific mortality. Physical activity (PA) represents a promising target for intervention.
Design: We prospectively evaluated PA from questionnaires before and after cancer diagnosis with total and cause-specific mortality among 13,297 subjects diagnosed with invasive cancer combined from the Physicians’ Health Study (PHS) (n=6328), Physicians’ Health Study II (PHS II) (n=912), and Women's Health Study (WHS) (n=6057). WHS and PHS participants were free of baseline cancer; PHS II participants reported no active cancer at baseline. We ascertained PA before and after an incident cancer diagnosis based on reports on repeated follow-up questionnaires. Death was ascertained by medical records and death certificates. Cox regression estimated combined hazard ratios (HRs) of mortality by PA adjusted for age, randomized treatments, BMI, and other lifestyle/demographic factors. We evaluated the interaction between PA before and after cancer diagnosis by comparing PA ≤1 versus ≥2 times/wk.
Results: The mean follow-up after cancer diagnosis was 8.0, 7.5, and 5.2 y for WHS, PHS, and PHS II, respectively, during which there were 5623 deaths (WHS, 2164; PHS, 3269; PHS II; 190). Higher PA before cancer diagnosis was associated with significantly lower mortality. Compared with PA ≤ once/wk, the HRs (95% CIs) associated with PA 2-4 and >4 times/wk were 0.87 (0.82-0.93) and 0.88 (0.82-0.94) for total mortality; 0.77 (0.63-0.95) and 0.79 (0.62-0.997) for CVD mortality, and 0.90 (0.83-0.98) and 0.90 (0.83-0.98) for cancer mortality. Higher PA after cancer diagnosis was associated with significantly lower total and cancer mortality and non-significantly lower CVD mortality, with HRs (95% CIs) of 0.65 (0.58-0.72) and 0.66 (0.59-0.73) for total mortality; 0.78 (0.59-1.03) and 0.82 (0.61-1.10) for CVD mortality, and 0.66 (0.57-0.77) and 0.64 (0.55-0.74) for cancer mortality. There was a significant interaction of PA before and after cancer diagnosis for total (pint=0.02) and cancer (pint=0.007) mortality, but not CVD mortality (pint=0.38).
Conclusions: Greater PA both before and after cancer diagnosis were significantly associated with lower total and cancer mortality. Higher PA before cancer diagnosis was also associated with lower CVD mortality. PA may be an important target for lower mortality after cancer diagnosis.