Lynne M Boddy1, Robert J Noonan2, Alex V Rowlands3, Liezel Hurter4, Zoe R Knowles4, Stuart J Fairclough2. 1. Physical Activity Exchange, Research Institute for Sport and Exercise Sciences, Liverpool John Moores University, UK. Electronic address: L.M.Boddy@ljmu.ac.uk. 2. Department of Sport and Physical Activity, Edge Hill University, UK. 3. Diabetes Research Centre, University of Leicester, Leicester General Hospital, UK; NIHR Leicester Biomedical Research Centre, UK; Alliance for Research in Exercise, Nutrition and Activity (ARENA), Sansom Institute for Health Research, Division of Health Sciences, University of South Australia, Australia. 4. Physical Activity Exchange, Research Institute for Sport and Exercise Sciences, Liverpool John Moores University, UK.
Abstract
OBJECTIVES: To examine the backward comparability of a range of wrist-worn accelerometer estimates of sedentary time (ST) with ActiGraph 100countmin-1 waist ST estimates. DESIGN: Cross-sectional, secondary data analysis METHODS: One hundred and eight 10-11-year-old children (65 girls) wore an ActiGraph GT3X+ accelerometer (AG) on their waist and a GENEActiv accelerometer (GA) on their non-dominant wrist for seven days. GA ST data were classified using a range of thresholds from 23 to 56mg ST estimates were compared to AG ST 100countmin-1 data. Agreement between the AG and GA thresholds was examined using Cronbach's alpha, intraclass correlation coefficients (ICC), limits of agreement (LOA), Kappa values, percent agreement, mean absolute percent error (MAPE) and equivalency analysis. RESULTS: Mean AG total ST was 492.4min over the measurement period. Kappa values ranged from 0.31 to 0.39. Percent agreement ranged from 68 to 69.9%. Cronbach's alpha values ranged from 0.88 to 0.93. ICCs ranged from 0.59 to 0.86. LOA were wide for all comparisons. Only the 34mg threshold produced estimates that were equivalent at the group level to the AG ST 100countmin-1 data though sensitivity and specificity values of ∼64% and ∼74% respectively were observed. CONCLUSIONS: Wrist-based estimates of ST generated using the 34mg threshold are comparable with those derived from the AG waist mounted 100countmin-1 threshold at the group level. The 34mg threshold could be applied to allow group-level comparisons of ST with evidence generated using the ActiGraph 100countmin-1 method though it is important to consider the observed sensitivity and specificity results when interpreting findings.
OBJECTIVES: To examine the backward comparability of a range of wrist-worn accelerometer estimates of sedentary time (ST) with ActiGraph 100countmin-1 waist ST estimates. DESIGN: Cross-sectional, secondary data analysis METHODS: One hundred and eight 10-11-year-old children (65 girls) wore an ActiGraph GT3X+ accelerometer (AG) on their waist and a GENEActiv accelerometer (GA) on their non-dominant wrist for seven days. GA ST data were classified using a range of thresholds from 23 to 56mg ST estimates were compared to AG ST 100countmin-1 data. Agreement between the AG and GA thresholds was examined using Cronbach's alpha, intraclass correlation coefficients (ICC), limits of agreement (LOA), Kappa values, percent agreement, mean absolute percent error (MAPE) and equivalency analysis. RESULTS: Mean AG total ST was 492.4min over the measurement period. Kappa values ranged from 0.31 to 0.39. Percent agreement ranged from 68 to 69.9%. Cronbach's alpha values ranged from 0.88 to 0.93. ICCs ranged from 0.59 to 0.86. LOA were wide for all comparisons. Only the 34mg threshold produced estimates that were equivalent at the group level to the AG ST 100countmin-1 data though sensitivity and specificity values of ∼64% and ∼74% respectively were observed. CONCLUSIONS: Wrist-based estimates of ST generated using the 34mg threshold are comparable with those derived from the AG waist mounted 100countmin-1 threshold at the group level. The 34mg threshold could be applied to allow group-level comparisons of ST with evidence generated using the ActiGraph 100countmin-1 method though it is important to consider the observed sensitivity and specificity results when interpreting findings.