Time-Dependent Confounding in the Study of the Effects of Regular Physical Activity in Chronic Obstructive Pulmonary Disease: An Application of the Marginal Structural Model

PURPOSE: Results from longitudinal studies about the association between physical activity and chronic obstructive pulmonary disease (COPD) may have been biased because they did not properly adjust for time-dependent confounders. Marginal structural models (MSMs) have been proposed to address this type of confounding. We sought to assess the presence of time-dependent confounding in the association between physical activity and COPD development and course by comparing risk estimates between standard statistical methods and MSMs. METHODS: By using the population-based cohort Copenhagen City Heart Study, 6,568 subjects selected from the general population in 1976 were followed up until 2004 with three repeated examinations. RESULTS: Moderate to high compared with low physical activity was associated with a reduced risk of developing COPD both in the standard analysis (odds ratio [OR] 0.76, p = 0.007) and in the MSM analysis (OR 0.79, p = 0.025). In the subgroup with COPD (n = 2,226), high physical activity was associated with a reduced risk of COPD admissions during follow-up (standard, incidence rate ratio, 0.74; p = 0.096; MSM, 0.68, p = 0.044), and with a reduced risk of mortality (standard, hazard ratio 0.80, p = 0.001; MSM, 0.81, p = 0.008). CONCLUSION: These results support the previously reported associations between physical activity and reduced risk of COPD development, hospitalizations, and mortality, thereby suggesting they were not due to time-dependent confounding. Ann Epidemiol 2008;18:775-783. (C) 2008 Elsevier Inc. All rights reserved.