RESEARCH ARTICLE

Measuring Physical Activity in Young People with Cerebral Palsy: Validity and Reliability of the ActivPAL™ Monitor Theofani Bania* Department of Physiotherapy, La Trobe University, Melbourne, Australia

Abstract Background and purpose. We determined the criterion validity and the retest reliability of the ΑctivPAL™ monitor in young people with diplegic cerebral palsy (CP). Methods. Activity monitor data were compared with the criterion of video recording for 10 participants. For the retest reliability, activity monitor data were collected from 24 participants on two occasions. Participants had to have diplegic CP and be between 14 and 22 years of age. They also had to be of Gross Motor Function Classification System level II or III. Outcomes were time spent in standing, number of steps (physical activity) and time spent in sitting (sedentary behaviour). Results. For criterion validity, coefficients of determination were all high (r2 ≥ 0.96), and limits of group agreement were relatively narrow, but limits of agreement for individuals were narrow only for number of steps (≥5.5%). Relative reliability was high for number of steps (intraclass correlation coefficient = 0.87) and moderate for time spent in sitting and lying, and time spent in standing (intraclass correlation coefficients = 0.60–0.66). For groups, changes of up to 7% could be due to measurement error with 95% confidence, but for individuals, changes as high as 68% could be due to measurement error. Discussion. The results support the criterion validity and the retest reliability of the ActivPAL™ to measure physical activity and sedentary behaviour in groups of young people with diplegic CP but not in individuals. Copyright © 2014 John Wiley & Sons, Ltd. Received 23 May 2013; Revised 5 December 2013; Accepted 13 February 2014 Keywords accelerometry; cerebral palsy; reproducibility of results; validity *Correspondence Theofani Bania, PT, MSc, PhD. Department of Physiotherapy, TEI of Western Greece, Prousiotissis 8, 30 100, Greece. Email: [email protected]

Published online 14 March 2014 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/pri.1584

Introduction People with cerebral palsy (CP) often have difficulty moving, and this puts them at risk of the health problems of reduced physical activity and increased sedentary behaviour (Durstine et al., 2000). This is especially true for those with diplegic CP and of Gross Motor Function Classification System (GMFCS) levels II and III whose walking ability, and therefore, physical

186

activity levels tend to decrease as they move from adolescence to adulthood (Jahnsen et al., 2004; Opheim et al., 2009). Being able to measure physical activity and sedentary behaviour in this group of people with CP accurately and reliably is important so the effect of interventions aimed at promoting their physical activity can be evaluated. Accelerometer-based activity monitors enable measurement of physical activity and sedentary behaviour over

Physiother. Res. Int. 22 (2014) 186–192 © 2014 John Wiley & Sons, Ltd.

T. Bania

prolonged periods. However, the gait impairments associated with people with diplegic CP such as walking slowly (Abel and Damiano, 1996; Eek et al., 2011) may affect the validity of activity monitors. In particular, activity monitors may not be able to accurately detect walking in these people because of cut-off thresholds at low cadences (and therefore speeds) (ActivPAL™ operating guide, 2006; Kanoun, 2009). Some preliminary evidence supporting the reliability of an activity monitor to measure physical activity in people with CP has been provided (McAloon, 2007), but this study only collected data for up to 10 minutes and did not assess physical activity over a number of days. A study that collected data from another activity monitor, an uptimer, over 6 days found high evidence of its retest reliability in people with CP (Pirpiris & Graham, 2004). Uptimers, however, measure total time spent in standing and stepping and do not measure important components of physical activity such as time spent in stepping separately from time spent in standing, number of steps or sedentary behaviour. Therefore, there is no adequate evidence of retest reliability of accelerometer-based activity monitors for measuring physical activity and sedentary behaviour in people with CP. Considering the preceding text, our first aim was to determine the criterion validity of an accelerometerbased activity monitor (ActivPAL™) in adolescents and young adults with diplegic CP using video observation as the criterion measure. Our second aim was to determine the retest reliability of this activity monitor in adolescents and young adults with diplegic CP over a clinically significant period.

Measuring PA in People with CP

Human Ethics and Hospital Human Ethics Committees. Written informed consent was obtained from all participants. For the criterion validity study, each participant was recorded with a hand held video camera for 12 minutes while wearing an accelerometer-based uniaxial activity monitor (ActivPAL™, PAL Technology, Glasgow) (Figure 1). During this 12-minute period, the participant was asked to stand up and sit down twice (during two 3-minute periods), as well as to walk around an oval circuit (a 6-minute walk). An observer, blinded to the activity monitor output, watched the videos and counted the number of steps and time spent in sitting and standing. The average time taken to attach the activity monitor to the participant’s thigh and the average time taken to remove the activity monitor from the participant’s thigh to download data in the computer were subtracted from the total time spent in standing and total time spent in sitting recorded by the activity monitor. Additionally, to confirm the stability of the criterion measure, retest reliability of the video data was evaluated using 10 video recordings assessed by the same observer on two occasions, 1-month apart. Intraclass correlation coefficients (ICCs) were all greater than 0.94, whereas limits of agreement were narrow for all outcomes measured (less than 0.09%).

Methods To be included in the studies, participants had to have diplegic CP and be aged 14–22 years. They also had to be of GMFCS level II or III (able to walk independently but may require the use of a handheld mobility device outdoors) (Palisano et al., 2007). Participants of the reliability study comprised a sub-set of participants (those assigned to the control group) of a larger intervention (strength training) trial. Volunteers were excluded if they had participated in a strengthening programme in the previous 6 months, had single event multi-level orthopaedic surgery in the previous 2 years or had contractures of more than 20° at the hip or knees. The studies were approved by the University Physiother. Res. Int. 22 (2014) 186–192 © 2014 John Wiley & Sons, Ltd.

Figure 1. The ActivPAL™ monitor placed mid-thigh on the front aspect of the thigh

187

Measuring PA in People with CP

For the retest reliability study, the activity monitor was attached to the participant’s thigh for both test and retest assessments (Figure 1). A 12-week interval was implemented between the test and retest measurements. A 12-week period is considered clinically significant for interventions such as strength training in people with CP (Dodd et al., 2003). Participants were instructed to wear the activity monitor for 7 days at both measurements. Participants were also advised to wear it at all times except during bathing and swimming. They were also encouraged to do their normal activities during the recording period. A daily log was also given to participants in which to note their activities during the 7-day period. Participants posted the activity monitor and the activity log back after the recording period. The following outcomes, representing physical activity, were measured: time spent in standing and number of steps. For the validity study, time spent in sitting was also measured (representing sedentary behaviour), whereas for the reliability study, time spent in sitting and lying (sedentary behaviour) was also measured. The sample size of the criterion validity study was calculated with a significance level of 0.05, a power of 0.80 and expected correlation coefficient of 0.95 (Howell, 1992). Ten participants were required, and the first 10 people eligible for inclusion in a larger intervention trial entered this study. For data analysis, coefficients of determination (r2) were calculated. An r2 ≥ 0.90 was considered as very high, an r2 = 0.80–0.89 as high, an r2 = 0.60–0.79 as moderate and an r2 ≤ 0.59 as low. Group limits of agreement were represented by the 95% confidence intervals (CIs) derived from paired t-tests, whereas limits of agreement for an individual were derived from Bland and Altman’s limits of agreement (Bland & Altman, 1999). The upper and lower band of the confidence limit was compared with the mean of the criterion measure, with a 10% difference designated as being acceptable. The sample size of the reliability study was estimated with a significance level of 0.05 and a power of 0.80 (Walter et al., 1998). It was assumed that the minimal expected level for reliability was 0.70 and the highest 0.90 (Pirpiris & Graham, 2004; Mackey et al., 2008). It was estimated that 26 participants would be required. Data were included for analysis if at least two full days of data for both test and retest measurements were available. As participants may have been engaged in 188

T. Bania

various activities during different days, to adequately assess physical activity, it was necessary to measure activity over multiple days (Baranowski et al., 2008). A sensitivity analysis, therefore, was also performed including only participants with at least six full days of recorded data. An ICC (model 2,1) was calculated to represent relative reliability. An ICC ≥ 0.75 was considered high reliability, an ICC = 0.75–0.50 moderate reliability and an ICC < 0.50 low reliability (Portney & Watkins, 2008). Retest reliability was also calculated in terms of units of measurement representing absolute reliability. A 95% CI for the group mean score was derived from the difference between means of paired scores (Altman, 1990): 95% CI ðmeanÞ ¼ Md ±

t0:975  SDdiff √Ν

where Md is the mean difference of retest minus test scores and SDdiff is the standard deviation of the difference between retest and test scores. For the interpretation of individual scores, otherwise known as the limits of agreement, N = 1 was substituted into the aforementioned equation (Bland & Altman, 1999). For the purposes of this study, reliability was determined to be adequate if the activity monitor’s measurement error expressed in the units of measurement would be smaller than changes usually observed in response to physical exercise programme interventions in published trials of people with CP.

Results/findings Four women and six men with spastic diplegic CP (18.6 ± 2.7 years; five of GMFCS level II and five of GMFCS level III) were included in the validity study. All coefficients of determination were very high (r2 ≥ 0.96) (Table 1). No systematic difference between the activity monitor and video observation was found for time spent in standing. There was a systematic difference for time spent in sitting and number of steps, but mean difference was small (Table 1). Limits of agreement for groups were narrow except for those for time spent in sitting, which were relatively wide (up to 20%). Limits of agreement for an individual were narrow for the number of steps (

Measuring physical activity in young people with cerebral palsy: validity and reliability of the ActivPAL™ monitor.

We determined the criterion validity and the retest reliability of the ΑctivPAL™ monitor in young people with diplegic cerebral palsy (CP)...
121KB Sizes 0 Downloads 3 Views