Determining minimum number of valid days for accurate estimation of sedentary behaviour and awake-time movement behaviours using the ActivPAL3 in nursing home residents

INTRODUCTION

Scarce evidence is available about the minimum number of valid days wearing the activPAL3 to obtain a precise estimate of sedentary behaviour (SB) and awake-time movement behaviours (ATMB) in nursing home (NH) residents. The study aimed to determine the minimum number of valid days required for accurately estimate SB and ATMB using the activPAL3 device in NH residents. It also investigated how the starting point of a day (the 24-h period) impacted reliability.

METHODS

Participants wore an activPAL3 for 7 consecutive days. The data was classified in two-time blocks (00:00 Ante Meridiem (AM)-00:00 AM midnight vs 12:00 Post Meridiam (PM) -12:00 PM midday) and the sample was stratified into two groups according to their capacity to stand and walk, to examine if timing of sampling or physical functioning affected minimum wear time. SB, ATMB, sociodemographic, and health-related variables were collected. Sensitivity of the time-blocks were tested through the dispersion frequencies and differences between blocks through Kolmogorov-Smirnov test for normality; parametric variables through two-related means T-test and Wilcoxon test for non-parametric data. Reliability was assessed with the Cronbach's Alpha and the intra-class correlation coefficient (ICC), using a one-factor model estimating the reliability for each measurement day loading in the same latent factor.

RESULTS

Ninety-five NH residents (81.1% women; age = 85.8 ± 7.2 years) were included. The midnight block had higher reliability, sensitivity and no statistically significant differences between days were found. At least three consecutive days of monitoring were necessary to achieve a reliability of ICC ≥ 0.8 for those NH residents able to stand and walk and six days for those unable.

CONCLUSIONS

NH residents who are able to stand and walk require a minimum of three consecutive days wearing the device, while those who are unable require at least six days due to their highly homogenous daily routines and sensitivity to PA events. Regardless of the activPAL3 recording start time, data processing should reference the midnight time block.

Agreement of the activPAL3 and activPAL for characterising posture and stepping in adults and children

Characterisation of physical activity and sedentary behaviour under free-living conditions is becoming increasingly important in light of growing evidence for the health implications of these behaviours. The integrity of long-term follow-up and the ability to compare outcomes between studies is critically dependent on the agreement of outcomes from successive generations of monitors. This study evaluated the agreement of the activPAL and second generation activPAL3 devices. Agreement was assessed in both adults (median 27.6y IQR 22.6) (n=20) and young people (median 12.0y IQR 4.1) (n=8) during standardised and daily living (ADL) test activities. During standardised activities; sedentary duration, upright duration, stepping duration and overall number of steps were all detected within small limits of agreement (≤5%). However, the activPAL characterised more steps during jogging than the activPAL3 (adults +8.36%, young people +6.80%). Also during ADL differences arose due to different posture characterisation in young people and lower step detection in the activPAL than the activPAL3 (adults -20.58%, young people -11.43%). Second-by-second posture analysis demonstrated high levels (>90%) of agreement for all activities between monitors. However, sensitivity (68.7%) and positive predictive value (78.8%) for adult stepping demonstrated disagreement between monitor interpretation of movement patterns during ADL. Agreement between monitor outcomes for standardised activities provides confidence that these outcomes can be considered almost equivalent. However, for characterisation of jogging and smaller movements during ADL, it is likely that significant differences between monitor outcomes will arise.