Journal of Aging and Physical Activity, 2015, 23, 391 -394 http://dx.doi.org/10.1123/japa.2013-0051 © 2015 Human Kinetics, Inc.
ORIGINAL RESEARCH
An Examination of the Reliability and Factor Structure of the Physical Activity Scale for Individuals With Physical Disabilities (PASIPD) Among Individuals Living With Parkinson’s Disease J. Jimenez-Pardo, J.D. Holmes, M.E. Jenkins, and A.M. Johnson Physical activity is generally thought to be beneficial to individuals with Parkinson’s disease (PD). There is, however, limited information regarding current rates of physical activity among individuals with PD, possibly due to a lack of well-validated measurement tools. In the current study we sampled 63 individuals (31 women) living with PD between the ages of 52 and 87 (M = 70.97 years, SD = 7.53), and evaluated the amount of physical activity in which they engaged over a 7-day period using a modified form of the Physical Activity Scale for Individuals with Physical Disabilities (PASIPD). The PASIPD was demonstrated to be a reliable measure within this population, with three theoretically defensible factors: (1) housework and home-based outdoor activities; (2) recreational and fitness activities; and (3) occupational activities. These results suggest that the PASIPD may be useful for monitoring physical activity involvement among individuals with PD, particularly within large-scale questionnaire-based studies. Keywords: Parkinson’s disease, physical activity, measurement, validation
The positive impact of physical activity has been documented for several sequelae of Parkinson’s disease (PD), including gait impairment, loss of balance, poor movement speed and coordination, and decreased strength (e.g., Goodwin, Richards, Taylor, Taylor, & Campbell, 2008). In addition, animal models of PD have suggested that benefits of physical activity may include increased neural plasticity (Smith & Zigmond, 2003) and a slowed progression of symptoms (Faherty, Raviie Shepherd, Herasimtschuk, & Smeyne, 2005). Interestingly, there are few references within the literature as to the actual amount of physical activity in which individuals with PD engage. Ford et al. (2010) reported that individuals with PD had approximately 322 min of step activity per day, but were inactive for approximately 77% of the day. This level of step activity was confirmed by Cavanaugh et al. (2012), who further determined that this level of activity remained stable over the course of a one-year period. Perhaps the most comprehensive study of physical activity in PD was, however, conducted by van Nimwegen et al. (2011), who found that individuals with PD engaged in approximately 111 min of physical activity (broadly conceived) in an average day. Van Nimwegen et al. (2011) employed the Longitudinal Aging Study of Amsterdam Physical Activity Questionnaire (LAPAQ), a physical activity questionnaire that was originally designed to be an interview-based assessment. Although the LAPAQ has been validated within the general population, we could find no evidence that it has been specifically validated for use among individuals with PD. In fact, it is important to note that there are no physical activity assessment instruments that have been specifically evaluated for use Jimenez-Pardo, Holmes, Jenkins, and Johnson are with Health and Rehabilitation Sciences, Faculty of Health Sciences, The University of Western Ontario, London, Ontario, Canada. Holmes is also with the School of Occupational Therapy, Faculty of Health Sciences, The University of Western Ontario, London, Ontario, Canada. Jenkins is also with Clinical Neurological Sciences, Schulich School of Medicine and Dentistry, The University of Western Ontario, London, Ontario, Canada. Johnson is also with the School of Health Studies, Faculty of Health Sciences, The University of Western Ontario, London, Ontario, Canada. Address author correspondence to A.M. Johnson at [email protected].
among individuals with PD. This dearth of appropriate measurement tools is not surprising given the unique considerations that must be taken into account when measuring physical activity within a population that includes older adults and individuals with chronic (and progressive) motoric dysfunction. Measurement instruments should be able to accurately assess (and differentiate between) lowintensity and moderate-intensity activity, as these are the activity levels that are likely to predominate within this population (van Nimwegen et al., 2011). Furthermore, instruments should assess alternative forms of physical activity (e.g., household, yard, and caregiving activities), as these forms of physical activity may be more common than organized recreation (Warms, 2006). Although not specifically designed for assessing the physical activity of individuals with PD, the Physical Activity Scale for Individuals with Physical Disability (PASIPD) is a measure of physical activity that takes into account the previously mentioned recommendations. It has a demonstrated ability to differentiate: individuals with excellent health from those with poor health; younger participants from older participants; moderately and extremely active individuals from inactive individuals; and individuals receiving attendant care from individuals that are not receiving attendant care (Washburn, Zhu, McAuley, Frogley, & Figoni, 2002). The PASIPD has also demonstrated (van der Ploeg et al., 2007) test-retest reliability (r = .77) and criterion validity (r = .30 with an accelerometer) comparable to well-established selfreport physical activity measures used within the general population (Sallis & Saelens, 2000) and within populations experiencing chronic neurological conditions, such as brain injury (Tweedy & Trost, 2005). Thus, the PASIPD not only takes into account the challenges of assessing physical activity within an older adult and/or disabled population, but has also been demonstrated to be psychometrically sound for use with individuals with various physical disabilities (van der Ploeg et al., 2007; Washburn et al., 2002). Even though the PASIPD was validated within a diverse sample of individuals with physical disabilities, individuals with progressive neurological dysfunction (such as PD) were not included within the validation sample. Given the physically debilitating symptoms of PD, it is likely that this scale could be used to assess physical activity within this population. The purpose of this study is to estimate 391
392 Jimenez-Pardo et al.
discriminative validity and to assess the internal consistency and factor structure of the PASIPD when administered to a sample of individuals with PD.
numeric question scored as an integer value); and (6) Did anyone help you to answer this survey? (‘yes’ or ‘no’).
Statistical Analysis
Methods
Downloaded by University of California on 09/17/16, Volume 23, Article Number 3
Participants A package containing the PASIPD, a letter of information, and an engraved pen (included as a token of appreciation for participants) was mailed to 120 individuals with PD. All prospective participants were drawn from the practice of a single movement disorder specialist, and were sampled nonpurposively, except for the following inclusion/exclusion criteria: (1) confirmed diagnosis of PD; (2) not currently hospitalized; and (3) not in a residential care, or longterm-care, facility. Sixty-three individuals (31 women) with PD between the ages of 52 and 87 (M = 70.97, SD = 7.53) returned the survey package, for an overall return rate of 52.5%. An institutional ethics review board approved the procedures, questionnaire, and consenting practices.
Instrumentation The PASIPD consists of 13 items that document the number of days per week and hours per day of participation in leisure activities, household activities, occupational activities, and inactivity over the past seven days (Washburn et al., 2002). Factor analysis and group differentiation supported its construct validity within a sample of 372 individuals with physical disabilities (e.g., vision impairment, paraplegia, spinal cord injury) between the ages of 35.8 and 63.6 (145 women; M = 48.4, SD = 12.6). Washburn et al. (2002) identified five factors within the PASIPD: (a) home, lawn, garden repair; (b) housework; (c) light exercise through sport and recreation; (d) vigorous exercise through sport and recreation; and (e) occupational activity. The scoring process is based on intensity values known as the metabolic equivalent of the task (MET); one MET is the ratio of the energy expenditure of an activity over the energy cost of the resting metabolic rate, which is approximately equivalent to expending 1 kcal per kilogram body weight per hour (Ainsworth et al., 2011). The scale’s total score is obtained by multiplying the average hours per day of each item by the MET value associated with the intensity of the activity and adding these values across items 2 through 13. The maximum possible score is 199.5 MET hours per day (Washburn et al., 2002). Activities that included the use of a wheelchair were removed from items 2, 4, and 5 in the original PASIPD template, as these activities lack generality within the population of individuals with PD. These modifications to the scale items do not affect the scoring of the questionnaire as they represent changes to the examples used within each of the items, rather than a deletion of actual content from the scale. Six additional items were included with the PASIPD for the purpose of evaluating test properties: (1) I am satisfied with my current physical activity (scaled on a five-point scale from ‘strongly disagree’ to ‘strongly agree’); (2) The amount of physical activity in which I have engaged over the past 7 days is typical of my usual activity (scaled on a five-point scale from ‘strongly disagree’ to ‘strongly agree’); (3) To what extent has Parkinson’s disease affected your level of physical activity? (scaled on a five-point scale from ‘greatly decreased’ to ‘greatly increased’); (4) This survey allowed me to properly describe my current levels of physical activity? (‘yes’ or ‘no’); (5) How many hours in total (i.e., for all activities) per week do you usually devote to physical activities? (an open-ended
The factor structure within the PASIPD was tested using a principal components analysis, and interpretation was facilitated by a varimax rotation (with Kaiser normalization). The number of extracted factors was determined through the application of a Monte Carlo parallel analysis (Zwick & Velicer, 1986). This method involves generating a set of random correlation matrices (in this study, we included 100 runs within the set) based upon the number of variables in the measure and the number of participants in the sample. Several principal components analyses are then performed, using these randomly generated correlation matrices. Finally, the average of the eigenvalues (from the unrotated factor loading matrix) derived from these principal component analyses are compared with the eigenvalues found within the experimental data. A factor is extracted if its corresponding experimental eigenvalue is higher than the one generated by the parallel analysis (Ledesma & Valero-Mora, 2007). Given the small sample size, the interpretability of the final factor solution was tested through the calculation of correlations between each item and the total unit-weighted composite score, and these correlations were reported along with the eigenvalues (experimental and Monte Carlo generated), the percentage of variance accounted for by each factor, and the factor loadings for each item within the solution. In addition, Cronbach’s alpha was computed for each factor, again within a unit-weighted variable comprised of the principally loading items within each factor. In addition to the principal components analysis, the additional self-report items that were included with the PASIPD were tabulated, and descriptives were reported as appropriate to the question form. Pearson product–moment correlations were presented for the Likert-style items, and presented in the context of an estimation of convergent validity.
Results Principal Components Analysis The Monte Carlo simulations within the parallel analysis suggested a four-factor solution, owing to the fact that only the first four factors obtained higher eigenvalues than the ones generated by the Monte Carlo parallel analysis. The four-factor solution demonstrated signs of over-extraction, however, in that several of the items were split between factors three and four. On the other hand, the three-factor solution was considerably closer to simple structure, demonstrating theoretically-sound factors (with distinct item loadings for each factor) and high Cronbach’s α coefficients. The three factors identified in this solution were: (1) housework and home outdoor activities (items 2, 7, 8, 9, 10, 11); (2) recreational and fitness activities (items 3, 4, 5, 6); and (3) occupational activities (items 12, 13). These factors accounted for 51.5% of the variance. Item 2 (walking with assistance) was, however, problematic within the factor solution and did not fit with the theoretical definition of the factor on which it loaded. Similarly, the Cronbach α coefficient improved considerably when this item was deleted (from .488 to .721 in the three-factor solution). To test the factor structure without this item, the principal components analysis was repeated after excluding item 2 from the variable list. This analysis also approached simple structure, and the factor solution accounted for 54.75%. The final three-factor solution (excluding item 2) is presented in Table 1.
JAPA Vol. 23, No. 3, 2015
Using the PASIPD for Individuals With PD 393
Table 1 Factor Loading Matrix for PASIPD Three-Factor Solution** Factor 1: Housework and Home Outdoors Activities
Item
Factor 3: Occupational Activities
Home repairs
0.72
–
–
Lawn and yard work
0.77
–
–
Outdoor gardening
0.76
–
–
Light housework
0.52
–
–
Heavy housework
0.64
–
–
–
0.81
–
Light sport and recreation Moderate sport and recreation
–
0.66
–
Strenuous sport and recreation
–
0.46
–
Muscle strength and endurance training
–
0.67
–
Care for another person
–
–
0.74
Work for pay/volunteer Downloaded by University of California on 09/17/16, Volume 23, Article Number 3
Factor 2: Recreational and Fitness Activities
Eigenvalue
–
–
0.74
3.01
1.53
1.49
Parallel analysis eigenvalue
1.79
1.53
1.35
% variance
27.35
13.88
13.52
Cumulative % variance
27.35
41.23
54.75
.72
.60
.49
Cronbach’s α*
* Cronbach’s alpha was calculated within the unit-weighted composite created from the items most indicative of the factor. **Excluding item 2 (Walking with assistance).
Congruence of the PASIPD With Additional Self-Report Items Seventy-nine percent of participants reported that their physical activity over the preceding seven days was typical of their usual activity, suggesting that the population considers the activities listed by the PASIPD to be representative of their typical activities. Furthermore, 74.6% of subjects completed the survey without any assistance, suggesting that individuals with PD are physically and cognitively capable of completing the PASIPD. Similarly, 85.7% of participants reported that this survey allowed them to properly describe their physical activity, which provides some evidence of content validity. A significant negative correlation (r = –.39) was found between the total PASIPD score and the extent to which the individual perceived PD to have affected his or her physical activity level, and a significant positive correlation (r = .50) was found between the total PASIPD score and the self-report estimate of total number of hours per week spent in physical activity. In addition, a significant negative correlation (r = –.30) was found between the frequency of stationary activities (e.g., reading, playing cards or board games, watching TV, playing computer games, or doing handicrafts) and the total PASIPD score. Taken together, these correlations provide evidence of convergent validity.
Discussion To the best of our knowledge, there are no tools for the measurement of physical activity that have been validated among individuals with PD. Given the nature of this disease, special considerations have to be taken into account when assessing physical activity within this population. The PASIPD was originally designed for individuals with physical disabilities in response to the need for a practical method of evaluating the physical activity of individuals
with limited physical capacity (Washburn et al., 2002). This survey thus includes daily activities such as housework, home outdoor activities, and occupational activities, in addition to formal physical activity training. The simplicity of the measure (13 items) and its consideration of physical activities more relevant to populations with limited mobility made this survey feasible for the assessment of physical activity among individuals with PD. Our analysis supported the extraction of three factors within a principal components analysis. Examination of the variables constituting these three factors (housework and home-based outdoor activities, recreational and fitness activities, and occupational activities) suggested that the solution is theoretically sound, providing a good explanation of the construct of physical activity within the population of individuals with PD. The solution obtained in the current study varies from the original solution suggested by Washburn et al. (2002), which presented a five-factor solution (home repair and lawn and garden work, housework, vigorous sports and recreation, moderate sport and recreation, and occupation and transportation). The current study focused, however, on obtaining a parsimonious solution in which at least 50% of the variance was explained by an interpretable model. Given that the measure captures intensity through the scaling of the measure (i.e., it is scaled on METs), the identified factor structure grouped items in dimensions that pertain more to the type of activity engaged than to the intensity levels. When dealing with individuals with chronic conditions and/or seniors, the focus of physical activity assessment changes, as these individuals are more likely to engage in lowintensity activities (Warms, 2006). Given the nature of the present population, therefore, a model that provides insight into the types of activities in which individuals are more likely to engage may be more useful to program planners and health practitioners aiming to increase physical activity engagement. Evidence of convergent validity was provided by the moderate correlation found between the PASIPD score and the self-reported
JAPA Vol. 23, No. 3, 2015
394 Jimenez-Pardo et al.
extent to which PD has affected level of physical activity, the total number of hours per week spent in physical activity, and rate of engagement in stationary activities. Further, some evidence of content validity was presented by the high percentage (85.7%) of individuals who reported that the survey allowed them to properly describe their physical activity. The survey was reported to be easily comprehensible, which is likely to reduce bias (Wilcox, Ainsworth, Shumaker, Ockene, & Riekert, 2009).
Downloaded by University of California on 09/17/16, Volume 23, Article Number 3
Limitations The model presented in this study is a parsimonious representation of the physical activities included within the PASIPD, but it is important to note that this reduced factor solution explains only 54.75% of participant variability on the measure. Thus, although this depiction of the typical physical activities of individuals with PD is an important initial step in the measurement of physical activity in this population, the intrinsic variability of this population may require larger-scale studies. Along similar lines, it is possible that our sample demonstrated some self-selection bias. In other words, it is possible that respondents were more physically active (and cognitively intact) than nonrespondents. This concern is somewhat mollified by the relatively high response rate (> 50%), but ideally researchers would have some clinical data available on nonrespondents. Finally, as the current study made no attempt at validating the PASIPD against objective assessments of physical activity such as accelerometers or other motion sensors, this may prove to be an important area for future research.
Conclusion The current results suggest that the PASIPD is an appropriate measure for assessing the physical activity of individuals with PD, and the easily-comprehensible format of the measure should facilitate its use in large-scale questionnaire-based studies. Hopefully this facilitates the quantification of physical activity participation within this population, as this may be an important first step in the identification of methods for improving the quantity of the physical activity engaged in by individuals with PD.
References Ainsworth, B.E., Haskell, W.L., Herrmann, S.D., Meckes, N., Bassett, D.R., Jr., Tudor-Locke, C., . . . Leon, A.S. (2011). 2011 Compendium of Physical Activities: a second update of codes and MET values. Medicine and Science in Sports and Exercise, 43(8), 1575–1581. PubMed doi:10.1249/MSS.0b013e31821ece12 Cavanaugh, J.T., Ellis, T.D., Earhart, G.M., Ford, M.P., Foreman, K.B., & Dibble, L.E. (2012). Capturing ambulatory activity decline in
Parkinson’s disease. Journal of Neurologic Physical Therapy; JNPT, 36(2), 51–57. PubMed doi:10.1097/NPT.0b013e318254ba7a Faherty, C.J., Raviie Shepherd, K., Herasimtschuk, A., & Smeyne, R.J. (2005). Environmental enrichment in adulthood eliminates neuronal death in experimental Parkinsonism. Brain Research. Molecular Brain Research, 134(1), 170–179. PubMed doi:10.1016/j.molbrainres.2004.08.008 Ford, M.P., Malone, L.A., Walker, H.C., Nyikos, I., Yelisetty, R., & Bickel, C.S. (2010). Step activity in persons with Parkinson’s disease. Journal of Physical Activity and Health, 7(6), 724–729. PubMed Goodwin, V.A., Richards, S.H., Taylor, R.S., Taylor, A.H., & Campbell, J.L. (2008). The effectiveness of exercise interventions for people with Parkinson’s disease: a systematic review and meta-analysis. Movement Disorders, 23(5), 631–640. PubMed doi:10.1002/mds.21922 Ledesma, R.D., & Valero-Mora, P. (2007). Determining the number of factors to retain in EFA: an easy-to-use computer program for carrying out parallel analysis. Practical Assessment, Research & Evaluation, 12(2), 1–11. Sallis, J.F., & Saelens, B.E. (2000). Assessment of physical activity by selfreport: status, limitations, and future directions. Research Quarterly for Exercise and Sport, 71(2 Suppl.) S1–S14. PubMed doi:10.1080/ 02701367.2000.11082780 Smith, A.D., & Zigmond, M.J. (2003). Can the brain be protected through exercise? Lessons from an animal model of parkinsonism. Experimental Neurology, 184(1), 31–39. PubMed doi:10.1016/j.expneurol.2003.08.017 Tweedy, S.M., & Trost, S.G. (2005). Validity of accelerometry for measurement of activity in people with brain injury. Medicine and Science in Sports and Exercise, 37(9), 1474–1480. PubMed doi:10.1249/01. mss.0000177584.43330.ae van der Ploeg, H.P., Streppel, K.R., van der Beek, A.J., van der Woude, L.H., Vollenbroek-Hutten, M., & van Mechelen, W. (2007). The Physical Activity Scale for Individuals with Physical Disabilities: test-retest reliability and comparison with an accelerometer. Journal of Physical Activity and Health, 4(1), 96–100. PubMed van Nimwegen, M., Speelman, A.D., Hofman-van Rossum, E.J., Overeem, S., Deeg, D.J., Borm, G.F., . . . Munneke, M. (2011). Physical inactivity in Parkinson’s disease. Journal of Neurology, 258(12), 2214–2221. PubMed doi:10.1007/s00415-011-6097-7 Warms, C. (2006). Physical activity measurement in persons with chronic and disabling conditions: methods, strategies, and issues. Family & Community Health, 29(1 Suppl), 78S–88S. PubMed doi:10.1097/00003727-200601001-00012 Washburn, R.A., Zhu, W., McAuley, E., Frogley, M., & Figoni, S.F. (2002). The physical activity scale for individuals with physical disabilities: Development and evaluation. Archives of Physical Medicine and Rehabilitation, 83(2), 193–200. PubMed doi:10.1053/apmr.2002.27467 Wilcox, S., Ainsworth, B.E., Shumaker, S.A., Ockene, J.K., & Riekert, K.A. (2009). The measurement of physical activity The Handbook of Health Behavior Change (3rd ed., pp. 327–346). New York, NY: Springer Publishing Co. Zwick, W., & Velicer, W.F. (1986). Comparison of five rules for determining the number of components to retain. Psychological Bulletin, 99(3), 432–442. doi:10.1037/0033-2909.99.3.432
JAPA Vol. 23, No. 3, 2015
ORIGINAL RESEARCH
An Examination of the Reliability and Factor Structure of the Physical Activity Scale for Individuals With Physical Disabilities (PASIPD) Among Individuals Living With Parkinson’s Disease J. Jimenez-Pardo, J.D. Holmes, M.E. Jenkins, and A.M. Johnson Physical activity is generally thought to be beneficial to individuals with Parkinson’s disease (PD). There is, however, limited information regarding current rates of physical activity among individuals with PD, possibly due to a lack of well-validated measurement tools. In the current study we sampled 63 individuals (31 women) living with PD between the ages of 52 and 87 (M = 70.97 years, SD = 7.53), and evaluated the amount of physical activity in which they engaged over a 7-day period using a modified form of the Physical Activity Scale for Individuals with Physical Disabilities (PASIPD). The PASIPD was demonstrated to be a reliable measure within this population, with three theoretically defensible factors: (1) housework and home-based outdoor activities; (2) recreational and fitness activities; and (3) occupational activities. These results suggest that the PASIPD may be useful for monitoring physical activity involvement among individuals with PD, particularly within large-scale questionnaire-based studies. Keywords: Parkinson’s disease, physical activity, measurement, validation
The positive impact of physical activity has been documented for several sequelae of Parkinson’s disease (PD), including gait impairment, loss of balance, poor movement speed and coordination, and decreased strength (e.g., Goodwin, Richards, Taylor, Taylor, & Campbell, 2008). In addition, animal models of PD have suggested that benefits of physical activity may include increased neural plasticity (Smith & Zigmond, 2003) and a slowed progression of symptoms (Faherty, Raviie Shepherd, Herasimtschuk, & Smeyne, 2005). Interestingly, there are few references within the literature as to the actual amount of physical activity in which individuals with PD engage. Ford et al. (2010) reported that individuals with PD had approximately 322 min of step activity per day, but were inactive for approximately 77% of the day. This level of step activity was confirmed by Cavanaugh et al. (2012), who further determined that this level of activity remained stable over the course of a one-year period. Perhaps the most comprehensive study of physical activity in PD was, however, conducted by van Nimwegen et al. (2011), who found that individuals with PD engaged in approximately 111 min of physical activity (broadly conceived) in an average day. Van Nimwegen et al. (2011) employed the Longitudinal Aging Study of Amsterdam Physical Activity Questionnaire (LAPAQ), a physical activity questionnaire that was originally designed to be an interview-based assessment. Although the LAPAQ has been validated within the general population, we could find no evidence that it has been specifically validated for use among individuals with PD. In fact, it is important to note that there are no physical activity assessment instruments that have been specifically evaluated for use Jimenez-Pardo, Holmes, Jenkins, and Johnson are with Health and Rehabilitation Sciences, Faculty of Health Sciences, The University of Western Ontario, London, Ontario, Canada. Holmes is also with the School of Occupational Therapy, Faculty of Health Sciences, The University of Western Ontario, London, Ontario, Canada. Jenkins is also with Clinical Neurological Sciences, Schulich School of Medicine and Dentistry, The University of Western Ontario, London, Ontario, Canada. Johnson is also with the School of Health Studies, Faculty of Health Sciences, The University of Western Ontario, London, Ontario, Canada. Address author correspondence to A.M. Johnson at [email protected].
among individuals with PD. This dearth of appropriate measurement tools is not surprising given the unique considerations that must be taken into account when measuring physical activity within a population that includes older adults and individuals with chronic (and progressive) motoric dysfunction. Measurement instruments should be able to accurately assess (and differentiate between) lowintensity and moderate-intensity activity, as these are the activity levels that are likely to predominate within this population (van Nimwegen et al., 2011). Furthermore, instruments should assess alternative forms of physical activity (e.g., household, yard, and caregiving activities), as these forms of physical activity may be more common than organized recreation (Warms, 2006). Although not specifically designed for assessing the physical activity of individuals with PD, the Physical Activity Scale for Individuals with Physical Disability (PASIPD) is a measure of physical activity that takes into account the previously mentioned recommendations. It has a demonstrated ability to differentiate: individuals with excellent health from those with poor health; younger participants from older participants; moderately and extremely active individuals from inactive individuals; and individuals receiving attendant care from individuals that are not receiving attendant care (Washburn, Zhu, McAuley, Frogley, & Figoni, 2002). The PASIPD has also demonstrated (van der Ploeg et al., 2007) test-retest reliability (r = .77) and criterion validity (r = .30 with an accelerometer) comparable to well-established selfreport physical activity measures used within the general population (Sallis & Saelens, 2000) and within populations experiencing chronic neurological conditions, such as brain injury (Tweedy & Trost, 2005). Thus, the PASIPD not only takes into account the challenges of assessing physical activity within an older adult and/or disabled population, but has also been demonstrated to be psychometrically sound for use with individuals with various physical disabilities (van der Ploeg et al., 2007; Washburn et al., 2002). Even though the PASIPD was validated within a diverse sample of individuals with physical disabilities, individuals with progressive neurological dysfunction (such as PD) were not included within the validation sample. Given the physically debilitating symptoms of PD, it is likely that this scale could be used to assess physical activity within this population. The purpose of this study is to estimate 391
392 Jimenez-Pardo et al.
discriminative validity and to assess the internal consistency and factor structure of the PASIPD when administered to a sample of individuals with PD.
numeric question scored as an integer value); and (6) Did anyone help you to answer this survey? (‘yes’ or ‘no’).
Statistical Analysis
Methods
Downloaded by University of California on 09/17/16, Volume 23, Article Number 3
Participants A package containing the PASIPD, a letter of information, and an engraved pen (included as a token of appreciation for participants) was mailed to 120 individuals with PD. All prospective participants were drawn from the practice of a single movement disorder specialist, and were sampled nonpurposively, except for the following inclusion/exclusion criteria: (1) confirmed diagnosis of PD; (2) not currently hospitalized; and (3) not in a residential care, or longterm-care, facility. Sixty-three individuals (31 women) with PD between the ages of 52 and 87 (M = 70.97, SD = 7.53) returned the survey package, for an overall return rate of 52.5%. An institutional ethics review board approved the procedures, questionnaire, and consenting practices.
Instrumentation The PASIPD consists of 13 items that document the number of days per week and hours per day of participation in leisure activities, household activities, occupational activities, and inactivity over the past seven days (Washburn et al., 2002). Factor analysis and group differentiation supported its construct validity within a sample of 372 individuals with physical disabilities (e.g., vision impairment, paraplegia, spinal cord injury) between the ages of 35.8 and 63.6 (145 women; M = 48.4, SD = 12.6). Washburn et al. (2002) identified five factors within the PASIPD: (a) home, lawn, garden repair; (b) housework; (c) light exercise through sport and recreation; (d) vigorous exercise through sport and recreation; and (e) occupational activity. The scoring process is based on intensity values known as the metabolic equivalent of the task (MET); one MET is the ratio of the energy expenditure of an activity over the energy cost of the resting metabolic rate, which is approximately equivalent to expending 1 kcal per kilogram body weight per hour (Ainsworth et al., 2011). The scale’s total score is obtained by multiplying the average hours per day of each item by the MET value associated with the intensity of the activity and adding these values across items 2 through 13. The maximum possible score is 199.5 MET hours per day (Washburn et al., 2002). Activities that included the use of a wheelchair were removed from items 2, 4, and 5 in the original PASIPD template, as these activities lack generality within the population of individuals with PD. These modifications to the scale items do not affect the scoring of the questionnaire as they represent changes to the examples used within each of the items, rather than a deletion of actual content from the scale. Six additional items were included with the PASIPD for the purpose of evaluating test properties: (1) I am satisfied with my current physical activity (scaled on a five-point scale from ‘strongly disagree’ to ‘strongly agree’); (2) The amount of physical activity in which I have engaged over the past 7 days is typical of my usual activity (scaled on a five-point scale from ‘strongly disagree’ to ‘strongly agree’); (3) To what extent has Parkinson’s disease affected your level of physical activity? (scaled on a five-point scale from ‘greatly decreased’ to ‘greatly increased’); (4) This survey allowed me to properly describe my current levels of physical activity? (‘yes’ or ‘no’); (5) How many hours in total (i.e., for all activities) per week do you usually devote to physical activities? (an open-ended
The factor structure within the PASIPD was tested using a principal components analysis, and interpretation was facilitated by a varimax rotation (with Kaiser normalization). The number of extracted factors was determined through the application of a Monte Carlo parallel analysis (Zwick & Velicer, 1986). This method involves generating a set of random correlation matrices (in this study, we included 100 runs within the set) based upon the number of variables in the measure and the number of participants in the sample. Several principal components analyses are then performed, using these randomly generated correlation matrices. Finally, the average of the eigenvalues (from the unrotated factor loading matrix) derived from these principal component analyses are compared with the eigenvalues found within the experimental data. A factor is extracted if its corresponding experimental eigenvalue is higher than the one generated by the parallel analysis (Ledesma & Valero-Mora, 2007). Given the small sample size, the interpretability of the final factor solution was tested through the calculation of correlations between each item and the total unit-weighted composite score, and these correlations were reported along with the eigenvalues (experimental and Monte Carlo generated), the percentage of variance accounted for by each factor, and the factor loadings for each item within the solution. In addition, Cronbach’s alpha was computed for each factor, again within a unit-weighted variable comprised of the principally loading items within each factor. In addition to the principal components analysis, the additional self-report items that were included with the PASIPD were tabulated, and descriptives were reported as appropriate to the question form. Pearson product–moment correlations were presented for the Likert-style items, and presented in the context of an estimation of convergent validity.
Results Principal Components Analysis The Monte Carlo simulations within the parallel analysis suggested a four-factor solution, owing to the fact that only the first four factors obtained higher eigenvalues than the ones generated by the Monte Carlo parallel analysis. The four-factor solution demonstrated signs of over-extraction, however, in that several of the items were split between factors three and four. On the other hand, the three-factor solution was considerably closer to simple structure, demonstrating theoretically-sound factors (with distinct item loadings for each factor) and high Cronbach’s α coefficients. The three factors identified in this solution were: (1) housework and home outdoor activities (items 2, 7, 8, 9, 10, 11); (2) recreational and fitness activities (items 3, 4, 5, 6); and (3) occupational activities (items 12, 13). These factors accounted for 51.5% of the variance. Item 2 (walking with assistance) was, however, problematic within the factor solution and did not fit with the theoretical definition of the factor on which it loaded. Similarly, the Cronbach α coefficient improved considerably when this item was deleted (from .488 to .721 in the three-factor solution). To test the factor structure without this item, the principal components analysis was repeated after excluding item 2 from the variable list. This analysis also approached simple structure, and the factor solution accounted for 54.75%. The final three-factor solution (excluding item 2) is presented in Table 1.
JAPA Vol. 23, No. 3, 2015
Using the PASIPD for Individuals With PD 393
Table 1 Factor Loading Matrix for PASIPD Three-Factor Solution** Factor 1: Housework and Home Outdoors Activities
Item
Factor 3: Occupational Activities
Home repairs
0.72
–
–
Lawn and yard work
0.77
–
–
Outdoor gardening
0.76
–
–
Light housework
0.52
–
–
Heavy housework
0.64
–
–
–
0.81
–
Light sport and recreation Moderate sport and recreation
–
0.66
–
Strenuous sport and recreation
–
0.46
–
Muscle strength and endurance training
–
0.67
–
Care for another person
–
–
0.74
Work for pay/volunteer Downloaded by University of California on 09/17/16, Volume 23, Article Number 3
Factor 2: Recreational and Fitness Activities
Eigenvalue
–
–
0.74
3.01
1.53
1.49
Parallel analysis eigenvalue
1.79
1.53
1.35
% variance
27.35
13.88
13.52
Cumulative % variance
27.35
41.23
54.75
.72
.60
.49
Cronbach’s α*
* Cronbach’s alpha was calculated within the unit-weighted composite created from the items most indicative of the factor. **Excluding item 2 (Walking with assistance).
Congruence of the PASIPD With Additional Self-Report Items Seventy-nine percent of participants reported that their physical activity over the preceding seven days was typical of their usual activity, suggesting that the population considers the activities listed by the PASIPD to be representative of their typical activities. Furthermore, 74.6% of subjects completed the survey without any assistance, suggesting that individuals with PD are physically and cognitively capable of completing the PASIPD. Similarly, 85.7% of participants reported that this survey allowed them to properly describe their physical activity, which provides some evidence of content validity. A significant negative correlation (r = –.39) was found between the total PASIPD score and the extent to which the individual perceived PD to have affected his or her physical activity level, and a significant positive correlation (r = .50) was found between the total PASIPD score and the self-report estimate of total number of hours per week spent in physical activity. In addition, a significant negative correlation (r = –.30) was found between the frequency of stationary activities (e.g., reading, playing cards or board games, watching TV, playing computer games, or doing handicrafts) and the total PASIPD score. Taken together, these correlations provide evidence of convergent validity.
Discussion To the best of our knowledge, there are no tools for the measurement of physical activity that have been validated among individuals with PD. Given the nature of this disease, special considerations have to be taken into account when assessing physical activity within this population. The PASIPD was originally designed for individuals with physical disabilities in response to the need for a practical method of evaluating the physical activity of individuals
with limited physical capacity (Washburn et al., 2002). This survey thus includes daily activities such as housework, home outdoor activities, and occupational activities, in addition to formal physical activity training. The simplicity of the measure (13 items) and its consideration of physical activities more relevant to populations with limited mobility made this survey feasible for the assessment of physical activity among individuals with PD. Our analysis supported the extraction of three factors within a principal components analysis. Examination of the variables constituting these three factors (housework and home-based outdoor activities, recreational and fitness activities, and occupational activities) suggested that the solution is theoretically sound, providing a good explanation of the construct of physical activity within the population of individuals with PD. The solution obtained in the current study varies from the original solution suggested by Washburn et al. (2002), which presented a five-factor solution (home repair and lawn and garden work, housework, vigorous sports and recreation, moderate sport and recreation, and occupation and transportation). The current study focused, however, on obtaining a parsimonious solution in which at least 50% of the variance was explained by an interpretable model. Given that the measure captures intensity through the scaling of the measure (i.e., it is scaled on METs), the identified factor structure grouped items in dimensions that pertain more to the type of activity engaged than to the intensity levels. When dealing with individuals with chronic conditions and/or seniors, the focus of physical activity assessment changes, as these individuals are more likely to engage in lowintensity activities (Warms, 2006). Given the nature of the present population, therefore, a model that provides insight into the types of activities in which individuals are more likely to engage may be more useful to program planners and health practitioners aiming to increase physical activity engagement. Evidence of convergent validity was provided by the moderate correlation found between the PASIPD score and the self-reported
JAPA Vol. 23, No. 3, 2015
394 Jimenez-Pardo et al.
extent to which PD has affected level of physical activity, the total number of hours per week spent in physical activity, and rate of engagement in stationary activities. Further, some evidence of content validity was presented by the high percentage (85.7%) of individuals who reported that the survey allowed them to properly describe their physical activity. The survey was reported to be easily comprehensible, which is likely to reduce bias (Wilcox, Ainsworth, Shumaker, Ockene, & Riekert, 2009).
Downloaded by University of California on 09/17/16, Volume 23, Article Number 3
Limitations The model presented in this study is a parsimonious representation of the physical activities included within the PASIPD, but it is important to note that this reduced factor solution explains only 54.75% of participant variability on the measure. Thus, although this depiction of the typical physical activities of individuals with PD is an important initial step in the measurement of physical activity in this population, the intrinsic variability of this population may require larger-scale studies. Along similar lines, it is possible that our sample demonstrated some self-selection bias. In other words, it is possible that respondents were more physically active (and cognitively intact) than nonrespondents. This concern is somewhat mollified by the relatively high response rate (> 50%), but ideally researchers would have some clinical data available on nonrespondents. Finally, as the current study made no attempt at validating the PASIPD against objective assessments of physical activity such as accelerometers or other motion sensors, this may prove to be an important area for future research.
Conclusion The current results suggest that the PASIPD is an appropriate measure for assessing the physical activity of individuals with PD, and the easily-comprehensible format of the measure should facilitate its use in large-scale questionnaire-based studies. Hopefully this facilitates the quantification of physical activity participation within this population, as this may be an important first step in the identification of methods for improving the quantity of the physical activity engaged in by individuals with PD.
References Ainsworth, B.E., Haskell, W.L., Herrmann, S.D., Meckes, N., Bassett, D.R., Jr., Tudor-Locke, C., . . . Leon, A.S. (2011). 2011 Compendium of Physical Activities: a second update of codes and MET values. Medicine and Science in Sports and Exercise, 43(8), 1575–1581. PubMed doi:10.1249/MSS.0b013e31821ece12 Cavanaugh, J.T., Ellis, T.D., Earhart, G.M., Ford, M.P., Foreman, K.B., & Dibble, L.E. (2012). Capturing ambulatory activity decline in
Parkinson’s disease. Journal of Neurologic Physical Therapy; JNPT, 36(2), 51–57. PubMed doi:10.1097/NPT.0b013e318254ba7a Faherty, C.J., Raviie Shepherd, K., Herasimtschuk, A., & Smeyne, R.J. (2005). Environmental enrichment in adulthood eliminates neuronal death in experimental Parkinsonism. Brain Research. Molecular Brain Research, 134(1), 170–179. PubMed doi:10.1016/j.molbrainres.2004.08.008 Ford, M.P., Malone, L.A., Walker, H.C., Nyikos, I., Yelisetty, R., & Bickel, C.S. (2010). Step activity in persons with Parkinson’s disease. Journal of Physical Activity and Health, 7(6), 724–729. PubMed Goodwin, V.A., Richards, S.H., Taylor, R.S., Taylor, A.H., & Campbell, J.L. (2008). The effectiveness of exercise interventions for people with Parkinson’s disease: a systematic review and meta-analysis. Movement Disorders, 23(5), 631–640. PubMed doi:10.1002/mds.21922 Ledesma, R.D., & Valero-Mora, P. (2007). Determining the number of factors to retain in EFA: an easy-to-use computer program for carrying out parallel analysis. Practical Assessment, Research & Evaluation, 12(2), 1–11. Sallis, J.F., & Saelens, B.E. (2000). Assessment of physical activity by selfreport: status, limitations, and future directions. Research Quarterly for Exercise and Sport, 71(2 Suppl.) S1–S14. PubMed doi:10.1080/ 02701367.2000.11082780 Smith, A.D., & Zigmond, M.J. (2003). Can the brain be protected through exercise? Lessons from an animal model of parkinsonism. Experimental Neurology, 184(1), 31–39. PubMed doi:10.1016/j.expneurol.2003.08.017 Tweedy, S.M., & Trost, S.G. (2005). Validity of accelerometry for measurement of activity in people with brain injury. Medicine and Science in Sports and Exercise, 37(9), 1474–1480. PubMed doi:10.1249/01. mss.0000177584.43330.ae van der Ploeg, H.P., Streppel, K.R., van der Beek, A.J., van der Woude, L.H., Vollenbroek-Hutten, M., & van Mechelen, W. (2007). The Physical Activity Scale for Individuals with Physical Disabilities: test-retest reliability and comparison with an accelerometer. Journal of Physical Activity and Health, 4(1), 96–100. PubMed van Nimwegen, M., Speelman, A.D., Hofman-van Rossum, E.J., Overeem, S., Deeg, D.J., Borm, G.F., . . . Munneke, M. (2011). Physical inactivity in Parkinson’s disease. Journal of Neurology, 258(12), 2214–2221. PubMed doi:10.1007/s00415-011-6097-7 Warms, C. (2006). Physical activity measurement in persons with chronic and disabling conditions: methods, strategies, and issues. Family & Community Health, 29(1 Suppl), 78S–88S. PubMed doi:10.1097/00003727-200601001-00012 Washburn, R.A., Zhu, W., McAuley, E., Frogley, M., & Figoni, S.F. (2002). The physical activity scale for individuals with physical disabilities: Development and evaluation. Archives of Physical Medicine and Rehabilitation, 83(2), 193–200. PubMed doi:10.1053/apmr.2002.27467 Wilcox, S., Ainsworth, B.E., Shumaker, S.A., Ockene, J.K., & Riekert, K.A. (2009). The measurement of physical activity The Handbook of Health Behavior Change (3rd ed., pp. 327–346). New York, NY: Springer Publishing Co. Zwick, W., & Velicer, W.F. (1986). Comparison of five rules for determining the number of components to retain. Psychological Bulletin, 99(3), 432–442. doi:10.1037/0033-2909.99.3.432
JAPA Vol. 23, No. 3, 2015
