Validity of a multi-context sitting questionnaire across demographically diverse population groups: AusDiab3

Background: Sitting time questionnaires have largely been validated in small convenience samples. The validity of this multi-context sitting questionnaire against an accurate measure of sitting time is reported in a large demographically diverse sample allowing assessment of validity in varied demographic subgroups. Methods: A subgroup of participants of the third wave of the Australian Diabetes, Obesity, and Lifestyle (AusDiab3) study wore activPAL3 (TM) monitors (7 days, 24 hours/day protocol) and reported their sitting time for work, travel, television viewing, leisure computer use and "other" purposes, on weekdays and weekend days (n = 700, age 36-89 years, 45 % men). Correlations (Pearson's r; Spearman's.) of the self-report measures (the composite total, contextual measures and items) with monitor-assessed sitting time were assessed in the whole sample and separately in socio-demographic subgroups. Agreement was assessed using Bland-Altman plots. Results: The composite total had a correlation with monitor-assessed sitting time of r = 0.46 (95 % confidence interval [CI]: 0.40, 0.52); this correlation did not vary significantly between demographic subgroups (all > 0.4). The contextual measure most strongly correlated with monitor-assessed sitting time was work (rho = 0.25, 95 % CI: 0.17, 0.31), followed by television viewing (rho = 0.16, 95 % CI: 0.09, 0.24). Agreement of the composite total with monitored sitting time was poor, with a positive bias (B = 0.53, SE 0.04, p < 0.001) and wide limits of agreement (+/- 4.32 h). Conclusions: This multi-context questionnaire provides a total sitting time measure that ranks participants well for the purposes of assessing health associations but has limited accuracy relative to activPAL-assessed sitting time. Findings did not differ in demographic subgroups.