### Abstract

Original language | English |
---|---|

Publication status | Published - 2010 |

### Fingerprint

### Keywords

- outcome measures
- physical activity monitoring

### Cite this

*Finding optimum outcome measures for physical activity monitoring*.

}

**Finding optimum outcome measures for physical activity monitoring.** / Mandrychenko, O; Granat, M; Chastin, S.

Research output: Contribution to conference › Paper

TY - CONF

T1 - Finding optimum outcome measures for physical activity monitoring

AU - Mandrychenko, O

AU - Granat, M

AU - Chastin, S

PY - 2010

Y1 - 2010

N2 - Background: Physical activity (PA) monitoring provides long-term data records displayed using various parameters (e.g. step counts, number of posture transitions, energy expenditure, time spent sedentary, etc.). Generally, studies tend to focus on one or a small number of these parameters in isolation. Currently, it is not clear which of these parameter(s) can describe PA in the fullest way and provide the best outcome measure. There is therefore a need to develop a robust method for PA parameter selection. Purpose: This study proposes the use of Factor Analysis in order to objectively find the simplest combination of PA parameters that can describe PA more fully. Methods: One hundred twenty health participants (age: mean = 45.4, s = 15.6, BMI: mean = 25.1, s = 4.0) were recruited in the study. PA was measured continuously over seven days using a validated PA monitor (ActivPAL) (Grant et al., 2006: British Journal of Sports Medicine, 40, 9292-997). Daily average for six common PA parameters (step count, number of sit-to-stand transitions; time spent sitting, standing and walking) were computed. These parameters were entered in a Factor Analysis with oblique procrustes rotation (Costello Osborne, 2005: Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical Assessment, Research Evaluation, 10, 7). The stability of the loading vectors was assessed using a Monte Carlo technique comparing results for 100 subsamples of 60 participants. Results: Two factors are enough to describe 86% of the variance in this population. The loading vectors were found to be stable with values deviation mean = 0.07 and s = 0.14, indicating that these two factors are a robust representation of PA in this sample. There was no cross loading of the parameters upon the two factors. Discussion: The first factor describing 62% of the variance is heavily loaded by energy expenditure, step count and walking time hinting that it might represent ambulatory activities. The time spent sitting and in quiet standing are more dominant in the second factor which can therefore be interpreted as a measure of sedentary time. Factor analysis using averages of different time scale might lead to more independent factors. Conclusion: It is possible to reduce PA parameters to a stable and robust smaller set of outcomes that fully encapsulate the variance. However, this led to outcome measure that might be difficult to interpret.

AB - Background: Physical activity (PA) monitoring provides long-term data records displayed using various parameters (e.g. step counts, number of posture transitions, energy expenditure, time spent sedentary, etc.). Generally, studies tend to focus on one or a small number of these parameters in isolation. Currently, it is not clear which of these parameter(s) can describe PA in the fullest way and provide the best outcome measure. There is therefore a need to develop a robust method for PA parameter selection. Purpose: This study proposes the use of Factor Analysis in order to objectively find the simplest combination of PA parameters that can describe PA more fully. Methods: One hundred twenty health participants (age: mean = 45.4, s = 15.6, BMI: mean = 25.1, s = 4.0) were recruited in the study. PA was measured continuously over seven days using a validated PA monitor (ActivPAL) (Grant et al., 2006: British Journal of Sports Medicine, 40, 9292-997). Daily average for six common PA parameters (step count, number of sit-to-stand transitions; time spent sitting, standing and walking) were computed. These parameters were entered in a Factor Analysis with oblique procrustes rotation (Costello Osborne, 2005: Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical Assessment, Research Evaluation, 10, 7). The stability of the loading vectors was assessed using a Monte Carlo technique comparing results for 100 subsamples of 60 participants. Results: Two factors are enough to describe 86% of the variance in this population. The loading vectors were found to be stable with values deviation mean = 0.07 and s = 0.14, indicating that these two factors are a robust representation of PA in this sample. There was no cross loading of the parameters upon the two factors. Discussion: The first factor describing 62% of the variance is heavily loaded by energy expenditure, step count and walking time hinting that it might represent ambulatory activities. The time spent sitting and in quiet standing are more dominant in the second factor which can therefore be interpreted as a measure of sedentary time. Factor analysis using averages of different time scale might lead to more independent factors. Conclusion: It is possible to reduce PA parameters to a stable and robust smaller set of outcomes that fully encapsulate the variance. However, this led to outcome measure that might be difficult to interpret.

KW - outcome measures

KW - physical activity monitoring

M3 - Paper

ER -