Abstract
PURPOSE: To examine the relationship of self-reported television (TV) viewing time with accelerometer-derived total sedentary time and to determine whether it differs by subgroup. METHODS: Using data for adults (≥20 yr) from the 2003-2004 and 2005-2006 nationally representative US National Health and Nutrition Examination Surveys (NHANES; n=5738), linear regression models examined the associations of categories of self-reported TV viewing time (<1, 1, 2, 3, 4, and >5 h·d) with accelerometer-derived sedentary time (<100 counts per minute; h·d). Spearman ρ assessed the correlation between participants' rankings on the two measures. Analyses were stratified by gender, age, race/ethnicity, and, in the 2003-2004 NHANES cycle, by work status among working-aged adults (20-65 yr, n=2069). RESULTS: TV viewing time was significantly associated with sedentary time, with positive associations for all gender, age, race/ethnicity groups, and for those not working or working part-time, but not for those in full-time work. However, correlations between rankings of the measures were only "fair" overall (ρ=0.22) and were similar for all gender and racial/ethnic groups and for those of mid- and older age but not for those of younger age (20-39 yr, ρ=0.05). In the working-aged subgroup, there was also a fair correlation between the measures for those not working (ρ=0.22) but no significant correlation for those in part-time (ρ=0.14) or full-time work (ρ=0.03). CONCLUSIONS: Associations of TV viewing time with accelerometer-derived total sedentary time were statistically significant, but correlations were of only fair magnitude, and the strength of the relationship was not consistent across all population subgroups. These findings suggest that TV viewing time has an influence on overall sedentary time at a population level; however, measurement of sedentary time in other domains is also important.