Physical activity: Potential confounder in alcohol epidemiologic studies?

Background: Alcohol consumption (alcohol) and physical activity (PA) are associated with many of the same health outcomes. If alcohol is related to PA, confounding may occur. That is, observed associations between alcohol and outcomes could be partially due to PA.

Methods: Participants were 14,397 men and 19,929 women age ≥20 years from the National Epidemiologic Survey on Alcohol and Related Conditions Wave 2. Self-reported drinking status was classified as never, former, or current. Within current drinkers, average daily drinking level was determined. Moderate drinking was defined as ≤2 drinks in men, ≤1 drink in women; heavier as ≥ 3 drinks in men, ≥ 2 drinks in women (2005 US Dietary Guidelines). Self-reported PA level during leisure, work, and daily life, was categorized as inactive, low, or medium/high (2008 Physical Activity Guidelines for Americans). Weighted, gender-specific distributions of PA by alcohol were calculated as predictive margins adjusted for age, demographics, and lifestyle from multinomial logistic regressions performed using SUDAAN.

Results: Overall, 66% of participants were current drinkers and 21% were physically inactive. Current drinkers were significantly less likely to be physically inactive than never or former drinkers (men: 16%, 23%, 22%, respectively; women: 21%, 30%, 27%). Within current drinkers, inactive proportions significantly decreased as alcohol increased from <1 to ≥ 3 drinks in men (17% to 15%) and <1 to ≥ 2 drinks in women (22% to 18%).

Conclusions: PA differed by drinking status and drinking level. Confounding by PA should be considered in epidemiologic studies of alcohol and health outcomes.