Am J Epidemiol 2013 Aug 19;178(3):474-83. Epub 2013 Jun 19.
Harvard School of Public Health, Building 2, 665 Huntington Avenue, Boston, MA 02115, USA.
The isotemporal substitution model (ISM) was previously developed as a methodology to study the time-substitution effects of 1 type of activity for another in a data setting with continuous outcomes. To demonstrate the application of ISM with a dichotomous outcome, we prospectively examined the associations of different activities with various activity displacements with depression risk among 32,900 US women from the Nurses' Health Study who were free from depressive symptoms at baseline (in 1996). During a 10-year follow-up, 5,730 incident depression cases were documented. Results from the ISMs indicated that for each physical activity, differences in magnitude of effects of each activity type were observed, dependent on the activity being displaced/substituted. Notably, an isotemporal substitution gradient was found for television watching, in which its association with depression risk varied by its substitution for slow-, average-, or brisk-paced walking in a gradient toward high depression risk when television watching replaced a faster walking pace (relative risk = 1.18, 95% confidence interval: 1.05, 1.31). Conversely, no association with depression was found for replacement of television watching with 60 minutes/day of slow walking, whereas a lower depression risk (relative risk = 0.85, 95% confidence interval: 0.76, 0.95) was found when 60 minutes/day of brisk walking replaced 60 minutes/day of television watching. Thus, the ISM could offer a more meaningful alternative to the standard nonsubstitution models to support public health recommendations.