Comparison of single- and dual-monitor approaches to differentiate sitting from lying in free-living conditions

High levels of sedentary time have been detrimentally linked to health outcomes. Differentiating sitting from lying may help to further understand the mechanisms associated with these health impacts. This study compares the inter-method agreement between the "single-monitor" method (thigh-worn activPAL^3TM ) and a more robustly validated "dual-monitor" method (trunk and thigh-worn activPAL^3TM ) in their classifications of sitting and lying under free-living conditions. Thirty-five participants (20-50 years) wore two activity monitors (thigh and trunk) for 24 hours. Total time spent lying and sitting was calculated for both methods, and agreement was determined using ICC and Bland-Altman methods. As there was no gold standard, further data were collected from five participants during structured activities that were designed to challenge classification, to better understand any disagreement between the methods. ICCs were 0.81 for sitting time and 0.64 for lying time. The single-monitor method detected less lying time than the dual-monitor method, with a mean difference of -25 minutes (95% agreement limits: -172 to 221 minutes), including three cases with extreme disagreement (mostly in daytime lying classification). The additional data collection suggested a major source of disagreement was failure of the single-monitor method to identify lying that involved no rotation around the longitudinal axis. In conclusion, there was some agreement between the single- and dual-monitor estimates of lying time under free-living conditions, but measures were not interchangeable. The main disagreement was in how the methods classified daytime lying and lying tasks involving no lateral movement. Both methods yield promise for measuring time in bed.