Physical activity and the association with sporadic ALS
AlbertoAscherio, Harvard School of Public Health, Depat. of Nutrition, HSPH, 665 Huntington Ave, Boston, MA 02115[email protected]
Submitted April 16, 2005
Veldink et al [1] reported a strong association between leisure time physical activity and earlier age at ALS onset and propose that this may reflect a possible adverse effect of physical activity. However, this association was based on analyses conducted among ALS cases only, and is thus potentially confounded by age (or, more specifically, birth cohort).
For example, it could be explained if younger individuals in the source population were more physically active than older individuals, as could be expected from secular trends in leisure time physical activity. It would be interesting to see results appropriately adjusted for age. A simple and direct way to do this would be to report: 1) the odds ratio (OR) for physical activity stratified by age; and 2) the results of a test for interaction between age and physical activity. The latter could be done by adding to the multivariate logistic regression model (with disease status as the dependent variable), an interaction term between physical activity (below or above the median), and age (continuous).
Although the reporting of the relation between exposures of interest and age at onset in clinical series is common, it is potentially misleading and should be discouraged.
References
1. Veldink et al. Physical activity and the association with sporadic ALS. Neurology, 2005; 64:241-5.
For example, it could be explained if younger individuals in the source population were more physically active than older individuals, as could be expected from secular trends in leisure time physical activity. It would be interesting to see results appropriately adjusted for age. A simple and direct way to do this would be to report: 1) the odds ratio (OR) for physical activity stratified by age; and 2) the results of a test for interaction between age and physical activity. The latter could be done by adding to the multivariate logistic regression model (with disease status as the dependent variable), an interaction term between physical activity (below or above the median), and age (continuous). Although the reporting of the relation between exposures of interest and age at onset in clinical series is common, it is potentially misleading and should be discouraged.
References
1. Veldink et al. Physical activity and the association with sporadic ALS. Neurology, 2005; 64:241-5.