Comparison between a standard manual and automated analysis ofaccelerometer data and the effect methodical decisions have onaccelerometer output

QP Physiology

Cuthill, Ruth ; Pitsiladis, Yannis

University:University of Glasgow

Department:Institute of Cardiovascular and Medical Sciences

关键词: IDEFICS - Accelerometry - MVPA - objective measurement;   

Others  :  http://theses.gla.ac.uk/3147/1/2012cuthillmsc.pdf

来源: University of Glasgow

【 摘 要 】

Background: The impact of accelerometer methodological decisions relating to the assessment of physical activity and sedentary time has not been conclusively determined inyoung children. With increasing numbers of large scale studies measuring physical activity, it is essential to have a validated method of analysis capable of analysing multiple files at any one time.Objectives: To describe and compare a standard method of analysis with an automated method of analysis of accelerometer data for use in large scale epidemiological studies. The automated approach also provides investigators with a powerful tool to effectively assess the effects of different decisions/choices on the classification of physical activity and sedentary behaviour by determining 1) the effects of epoch and cut-points on the assessment of physical activity and sedentary time, 2) how to define non wear time and, 3) accelerometer wear time required to achieve reliable accelerometer data in children.Design: The physical activity levels of 86 children aged 4-10 were measured as part of a larger European study. Children were recruited from centres at Ghent, Glasgow, Gothenburg and Zaragoza.Methods: Physical activity was assessed for 1 week in 86 children (41 female, 45 male; mean age 7±2 years) by uni-axial accelerometry. The epoch was set at 15 s and re-integrated to 30 s and 60 s. Time spent in sedentary and moderate and vigorous physical activity (MVPA) wasassessed using Pate, Puyau, Reilly and Sirard cut points. Non wear time of accelerometer was defined by removal by the 10-, 20-, 30- and 60-mins of consecutive zeros.Results: There was excellent agreement between the automated method of analysis and accelerometer outputs generated by the standard manual method of analysis. The Reilly cut3points (<1100 counts/min) indicated less sedentary time per day when comparing 15 s vs. 30 s and 15 s vs. 60 s epochs: 570±91 min vs. 579±93 min and 570±91 min vs. 579±94,respectively; P<0.05). Pate cut-points (>420 counts/15 s) reported more MVPA time per day compared to Sirard (890 counts/15 s) and Puyau cut-points (>3200 counts/min) using 15 s epoch: 88 (4-197) mins (median (range) vs. 18 (1-80) mins and 24 (1-100) mins, respectively; P<0.001). Compliance with guidelines of at least 60 mins MVPA was 83%, 77% and 72% for Pate cut-points using 15 s, 30 s and 60 s epoch, respectively but 0% for Sirard and Puyau cutpointsacross epochs. The number of days required to achieve 80% reliability for counts per minute (CPM), sedentary and MVPA time was 7.4 – 8.5 days.Conclusion: An automated method of analysis of accelerometer data has successfully compared with manual analysis and should be recommended for use in large scale epidemiological studies. Choice of epoch and cut-points significantly influenced the classification of sedentary and MVPA time and observed compliance to MVPA guidelines, emphasising the need to standardise accelerometer data reduction methods. In order to accurately measure and asses physical activity levels of a population, a uniform analysis must be generated to be able to compare physical activity across populations.