Abstract
Background The assessment of physical activity is of concern in patients after total hip arthroplasty (THA). However, so far, no questionnaire has demonstrated adequate reproducibility and validity for assessing physical activity in these patients.
Objective The aim of this study was to evaluate the reproducibility and validity of the Physical Activity Scale for the Elderly (PASE) questionnaire in patients after THA.
Design This was a measurement study.
Methods Fifty patients who had undergone THA (25 women, 25 men), with an average age of 68 years, were evaluated. Of these patients, 25 were assessed between 2 and 7 months after surgery (THAearly), and another 25 were assessed between 7 and 12 months after surgery (THAlate). Reproducibility of the PASE questionnaire was evaluated by administering the questionnaire on 2 different occasions. Construct validity of the PASE questionnaire was assessed by comparing the physical activity level reported by patients with that objectively recorded by a body-mounted accelerometer. Reproducibility was investigated with intraclass correlation coefficients (ICC [2,1]) for reliability and standard errors of measurement (SEM) for agreement. Validity was investigated with Pearson correlation coefficients (r).
Results The ICC (2,1) for the PASE total score was .77 (95% confidence interval [95% CI]=.63, .86); the SEM was 23.0% (95% CI=19.2, 28.7). Validity correlation for the PASE total score was .38 (95% CI=.12, .60). No significant differences were found between THAearly and THAlate groups for reliability, agreement, and validity outcomes.
Limitations Reproducibility of the PASE questionnaire may have been underestimated because the physical activity of patients was compared between 2 consecutive but different weeks. Reliability and validity analyses were underpowered.
Conclusions Further study with a larger sample size is necessary to obtain precise reliability and validity estimates. Nevertheless, inadequate agreement calls into question the PASE questionnaire's ability to assess the physical activity level of patients after THA surgery.
Total hip arthroplasty (THA) is the most effective and successful intervention to treat patients with end-stage hip osteoarthritis.1 In the last decade, the number of THAs has increased in developed countries by an average of 25% and will increase again in the next decade.2 Following THA, patients typically experience a reduction of symptoms (ie, reduced hip pain and reduced stiffness) and an improvement in their physical function (ie, the ability to perform daily and sport activities).3 The physical function of patients who have undergone THA has been shown to constantly improve in the first 7 months after surgery, whereupon it tends to stabilize to normal or almost normal levels.4 Accordingly, patients usually have some functional disability after THA surgery, at least in the first 7 months, which could result in low levels of physical activity.5 In turn, low amounts of physical activity in these older patients represent a potential risk for the development of chronic diseases such as cardiovascular pathologies and obesity.6 In addition, an adequate level of physical activity is the key to enhance the recovery of lower limb muscle function in patients after THA, which has been shown to be compromised until at least 1 year after the operation.7 For these reasons, the assessment of physical activity is of particular concern in patients after THA so as to distinguish between patients who are more active and those who are less active (for discriminative purposes) or to monitor their level of physical activity over time (for evaluative purposes).
Questionnaires are the most widely used tools to assess the level of physical activity in clinical settings because they are practical and cost-effective.8 To date, however, no questionnaire has demonstrated adequate reproducibility and validity in order to be used for assessing physical activity in patients after THA.9 The Physical Activity Scale for the Elderly (PASE) is a brief (time to completion is 5–10 minutes) and easily administered questionnaire (contains categorical response options) developed for evaluating physical activity participation of healthy older adults.10 It has already been used for both discriminative11,12 and evaluative13 purposes to assess the physical activity level of patients with hip and knee osteoarthritis. The PASE questionnaire inquires about the type (intensity) and amount (frequency and duration) of physical activity participation in recreational, household, and work-related settings. It particularly focuses on activities commonly engaged in by older adults (eg, light-intensity recreational and household activities)14–18 and does not focus on sport and moderate- to high-intensity recreational activities. So far, the PASE questionnaire has demonstrated controversial reproducibility and validity results in healthy older adults, moderate validity in patients with knee pain and functional disability,19 and poor validity and moderate reproducibility in patients with hip osteoarthritis.20 Considering the need for a valid physical activity questionnaire in patients after THA, the brief and easy administration of the PASE questionnaire, and the questionnaire's positive ratings obtained in some older adult populations,9,21 it is worthwhile to evaluate its measurement properties in patients after THA.
The aim of this study was to evaluate the (1) reproducibility (reliability and agreement) and (2) construct validity of the PASE questionnaire relative to accelerometry-based physical activity in patients at 2 different time points in the first year following THA (early [2–7 months] and late [7–12 months] postoperative). A final aim was to compare the reproducibility and construct validity between patients early and late postoperatively who have undergone THA. It was hypothesized that the PASE questionnaire would demonstrate high reliability (our null and alternate hypotheses were intraclass correlation coefficient (ICC [2,1]) <.70 and ICC [2,1] ≥.70, respectively)22 and high agreement (our null and alternate hypotheses were standard error of measurement [SEM] ≥10% and SEM <10%, respectively).23 It also was hypothesized that the PASE questionnaire would demonstrate good construct validity as (1) high correlations to the time spent in light-intensity physical activities,22 (2) high correlations to the total level of physical activity (our null and alternate hypotheses were r<.50 and r≥.50, respectively),22 and (3) higher correlations to the time spent in light-intensity activities than in moderate- to high-intensity physical activities (P<.05). In addition, it was hypothesized that the PASE questionnaire would show higher reliability, agreement, and construct validity in patients with THA evaluated late postoperatively compared with those evaluated early postoperatively (P<.05).
Method
Participants
A series of 50 consecutive patients who had undergone primary THA (25 women, 25 men) and volunteered to participate were retrospectively recruited in the study. A sample size of 50 patients was considered adequate according to the guidelines proposed by Terwee et al24 for the evaluation of measurement properties of health status questionnaires. All of the patients had been operated on in the same orthopedic clinic between November 2010 and December 2011 by 2 experienced hip surgeons. Patients with total arthroplasty of other lower limb joints and symptoms or signs referable to other overt cardiorespiratory, orthopedic, neurological, or general diseases that could have negatively influenced physical activity evaluation with questionnaires or accelerometers were not included in the study. The characteristics and sociodemographic data of the evaluated patients are presented in Table 1. Patients were separated into 2 groups based on the time elapsed between surgery and physical activity assessment. The first group included 25 patients (13 women, 12 men) evaluated between 2 and less than 7 months after surgery (THAearly). The second group included 25 patients (12 women, 13 men) evaluated between 7 and 12 months after surgery (THAlate). In the former group of patients, physical function is usually improving but still impaired, whereas in the latter group, it is usually recovered to normal or almost normal levels.4 All of the patients provided written informed consent before participating in the study.
Participants' Characteristics and Sociodemographic Dataa
Study Design and Procedure
The original English version of the PASE questionnaire was first cross-culturally adapted so as to be used for evaluating the physical activity level of German-speaking patients. The reproducibility of the PASE questionnaire was evaluated by comparing the PASE scores of patients during 2 consecutive weeks (mean number of days between test sessions 1 and 2=7.1, SD=0.5). This reproducibility study design was chosen to prevent recall and to ensure that clinical changes had not occurred between the 2 test sessions.24 Thus, participants were explicitly asked to maintain their usual physical activity habits during this time period. The construct validity of the PASE questionnaire was evaluated by comparing the PASE scores with body-mounted accelerometry measures.22 Three hypotheses about convergent relations between the PASE scores and accelerometry measures were formulated so as to evaluate the construct validity of the PASE questionnaire.25 Specifically, correlations ≥.50 were expected between (1) PASE scores and the time spent in light-intensity physical activities and (2) PASE scores and a measure of total physical activity.22 In addition, higher correlations were expected between PASE scores and the time spent in light-intensity physical activities compared with moderate- to high-intensity physical activities. The PASE questionnaire was specifically developed for measuring light recreational, household, and work-related activities commonly engaged in by older people, who are usually underrepresented in age-neutral physical activity questionnaires.10 The time spent in light-intensity physical activity measured with accelerometers was considered the most appropriate indicator of our reference construct. Construct validity was arbitrarily defined as good if all 3 hypotheses were confirmed, moderate if 2 of the 3 hypotheses were confirmed, and poor if only 1 of the 3 hypotheses was confirmed.24,25 The PASE questionnaire was filled out by patients at test sessions 1 and 2. Participants were instructed to wear an accelerometer between test sessions 1 and 2.
PASE Questionnaire
The cross-cultural adaptation process was performed according to the guidelines of the American Academy of Orthopaedic Surgeons Outcomes Committee.26 It comprised 5 steps. Step 1 included forward translation from English to German by 1 informed translator (T1) and 1 uninformed translator (T2), both native German speakers and fluent in English. Step 2 comprised synthesis of T1 and T2 into 1 version (T12), under the supervision of 1 methodologist. In step 3, the T12 version was translated from German back to English by 2 uninformed native English speakers (BT1 and BT2) who were fluent in German. Step 4 comprised a consensus meeting of all people involved in the translation process to establish the prefinal German version. Step 5 involved testing the prefinal German version in 15 consecutive patients who had undergone THA or total knee arthroplasty to examine the accuracy of wording and ease of understanding.
The PASE questionnaire was developed to assess physical activity in a population of healthy older subjects.10 It consists of 12 questions inquiring about participation in recreational activities (5 questions), household activities (6 questions), and paid or unpaid work-related activities (1 question) during the previous week. The frequency of participation in recreational activities (ie, outside walking; light-, moderate-, and high-intensity recreational and sport activities; and strength training) is recorded as “never,” “seldom” (1–2 days per week), “sometimes” (3–4 days per week), or “often” (5–7 days per week). The participation duration is recorded as “less than 1 hour,” “1 to 2 hours,” “2 to 4 hours,” or “more than 4 hours.” Participation in household activities (ie, light housework, heavy housework or chores, home repairs, lawn work or yard care, outdoor gardening, caring for another person) is recorded as “yes” or “no,” without any concern for frequency and duration. Participation in paid or unpaid work-related activities, which requires some physical activity, is recorded in hours per week. The questionnaire was administered by face-to-face interview.10 One trained interviewer first explained to the participants how the questionnaire is structured, going through all the questions. Afterward, participants filled out the questionnaire under the supervision of the interviewer. Each question score was calculated by multiplying the time spent (hours per week for recreational and work-related activities) or participation (“yes” or “no” for household activities) in the respective activity by an empirically derived question weight.10 The recreational, household, and work-related activity PASE subscale scores were calculated by summing the scores of the respective questions. The PASE total score was calculated by summing all question scores.
Accelerometry
Physical activity was monitored using GT3X ActiGraph accelerometers (ActiGraph, Pensacola, Florida).27 These are small and lightweight devices (4.6 × 3.3 × 1.5 cm and 19 g) that reliably measure accelerations ranging between −6 g to +6 g in 3 individual orthogonal planes (vertical, anteroposterior, and mediolateral). The acceleration signal was sampled at 100 Hz and digitized by a 12-bit analog-to-digital converter. The digital outcome was band-pass filtered with a frequency range of 0.25 to 2.5 Hz. The accelerometer outcome is defined as activity counts. Activity counts are the sum of acceleration amount and frequency in a defined time interval referred to as an epoch and represent the activity intensity in that epoch. Participants were instructed by the same experienced investigator to wear the accelerometer from getting up in the morning until going to bed in the evening, except during water activities (ie, bathing, water gymnastics, and swimming). The device was worn on a belt at the right hip waistline.28 Participants were instructed to complete a standardized daily log (time sheet) to record when the device was put on and removed.
Data were exported from the device in 60-second epochs using ActiLife 5 analysis software (ActiGraph, Pensacola, Florida). A valid day consisted of at least 10 hours (not necessarily continuous) of wear time recordings.29 Nonwear time was defined as the sum of time intervals of at least 60 minutes of zero counts that contained no more than 2 minutes of counts between 0 and 100.29 Individual patient files had to have at least 5 valid days so that their accelerometry data could be used for analysis.30
The primary measure for construct validity was the time (in minutes) spent by participants in light-intensity activities. Secondary measures were the time spent by participants in moderate- to high-intensity activities and the counts per minute of wear time, defined as total physical activity. The activities were classified as light and moderate to high using (1) the metabolic equivalent task (MET) classifications proposed by the US Department of Health and Human Services31 and (2) the regression equation proposed by Sasaki et al32 (METs=0.000863 [counts] + 0.668876) to convert counts into METs. For patients younger than 65 years, activities with a MET between 2.5 and 4.4 and activities with a MET ≥4.5 were classified as light-intensity activities and moderate- to high-intensity activities, respectively. For patients aged 65 years or older, activities with a MET between 2.0 and 3.5 and activities with a MET ≥3.6 were classified as light-intensity activities and moderate- to high-intensity activities, respectively.
Data Analysis
Descriptive and sociodemographic data are presented as mean (SD) or number (%), and reproducibility and validity results are presented with the corresponding 95% confidence interval (95% CI) value. Normal distribution of data was assessed with Shapiro-Wilk tests. If the data were not normally distributed, they were log transformed before statistical analyses. Reproducibility was assessed as reliability (ie, the extent to which patients can be distinguished from each other despite measurement error) and agreement (ie, the extent to which scores on repeated measurements are close to each other). Reliability was evaluated using ICC (2,1) (2-way analysis of variance, random model, consistency, single measure). An ICC (2,1) ≥.70 was considered as acceptable according to recommendations for physical activity questionnaires.22 Hypothesis testing for reliability was performed using F tests.33 Agreement was evaluated using percent SEM (%SEM=100 × [SEM/PASE score grand mean], SEM=√mean square error) and percent smallest detectable change (%SDC=1.96 × √2 × %SEM).24,34 Standard error of measurement <10% and SDC <28% were considered as low (ie, high agreement).23 Hypothesis testing for agreement was performed by observing if the confidence intervals contained the agreement benchmarks (10% and 28% for SEM and SDC, respectively).35 Percent changes in the mean were examined with paired t tests. Construct validity was evaluated using Pearson correlation coefficients. A correlation coefficient ≥.50 was considered as adequate for physical activity comparisons between questionnaires and accelerometry according to recommendations for physical activity questionnaires.22 Hypothesis testing for validity was performed using z tests (differences between independent correlation coefficients) and Hotelling-William t tests (differences between dependent correlation coefficients). Between-group differences were investigated with unpaired z tests (reliability and validity) or unpaired t tests (agreement). Statistical analyses were performed with PASW Statistics 18.0 (SPSS Inc, Chicago, Illinois). The significance level was set at P<.05.
Role of the Funding Source
The study was funded by the Schulthess Clinic Research Fund (internal funding).
Results
Descriptive data for the PASE questionnaire are presented in Table 2. Participants in the THAlate group demonstrated higher household activity PASE subscores compared with participants in the THAearly group (P=.03).
Physical Activity Scale for the Elderly (PASE) Questionnaire Descriptive Dataa
Reproducibility
The PASE total score showed overall nonsignificant changes in the mean that ranged between 10.1% and 12.5% (P>.05) (Tab. 3). No significant difference was observed for the change in the mean between the THAearly and THAlate groups (P>.05). The %SEM and %SDC were overall significantly larger than 10% and 28%, respectively (P<.05). No significant difference was observed for %SEM and %SDC between the THAearly and THAlate groups (P>.05). Given that the lower limit of the 95% CI for ICC (2,1) was overall less than .70, we could not reject the null hypothesis. No significant difference was observed for ICC (2,1) between the THAearly and THAlate groups (P>.05).
Physical Activity Scale for the Elderly (PASE) Questionnaire Reproducibility Resultsa
Validity
Accelerometry descriptive data are presented in Table 4. No significant difference was found between the THAearly and THAlate groups (P>.05).
Accelerometry Descriptive Dataa
When considering both groups of patients, the lower limit of the 95% CI for the correlation coefficient between the PASE total score and the time spent in light-intensity physical activities was less than .50; therefore, we could not reject the null hypothesis (Tab. 5). The correlation coefficient between PASE total score and the total physical activity level was significantly lower than .50 (r=.27, P=.03). The correlation coefficient between PASE total score and the time spent in light-intensity physical activities was not significantly higher than the correlation coefficient between PASE total score and the time spent in moderate- to high-intensity physical activities (r=.38 versus r=.26, P>.05).
Physical Activity Scale for the Elderly (PASE) Questionnaire Validity Resultsa
In the THAearly group, the correlation coefficient between the PASE total score and the time spent in light-intensity physical activities was significantly lower than .50 (r=.16, P=.03). The correlation coefficient between the PASE total score and the total physical activity level was significantly lower than .50 (r=.14, P=.02). The correlation coefficient between PASE total score and the time spent in light-intensity physical activities was not significantly higher than the correlation coefficient between the PASE total score and the time spent in moderate- to high-intensity physical activities (r=.16 versus r=.24, P>.05).
When considering participants in the THAlate group, the 95% CI for the correlation coefficient between the PASE total score and the time spent in light-intensity physical activities included the null value; therefore, we could not reject the null hypothesis. Also, the 95% CI for the correlation coefficient between the PASE total score and the total physical activity level included the null value; therefore, we could not reject the null hypothesis. The correlation coefficient between PASE total score and the time spent in light-intensity physical activities was not significantly higher than the correlation coefficient between the PASE total score and the time spent in moderate- to high-intensity physical activities (r=.55 versus r=.28, P>.05). No significant difference was found between the THAearly and THAlate groups for all correlation coefficients.
Discussion
The present study demonstrated that there is uncertainty of the reliability and validity of the PASE questionnaire for assessing the physical activity level of patients after THA surgery. In contrast, the PASE questionnaire clearly demonstrated inadequate agreement. No differences were found between the THAearly and THAlate groups for reliability, agreement, and validity outcomes.
This is the first study that has investigated the reproducibility and validity of the PASE questionnaire in patients following THA. The need for a questionnaire properly assessing the physical activity level in this patient population is of particular concern for researchers and preventive medicine physicians. An increasing number of individuals will undergo THA in the next decade,2 and low levels of physical activity represent a potential risk factor for the development of chronic diseases in this population of older patients.6
A limitation of the present study is the design used to test the reproducibility of the questionnaire. The physical activity level of our patients was assessed with the PASE questionnaire during 2 consecutive but different weeks. Measuring the physical activity level twice but referring to the same week would have been methodologically more appropriate. However, a time period of less than 1 week between the 2 questionnaire administrations would have been improper because of an increased risk of recall bias.24 In the attempt to minimize the negative effects of physical activity level variability on reproducibility results, participants were explicitly asked to maintain their usual physical activity habits in the time period between the 2 questionnaire administrations. Another limitation of this study is the small size of our sample. Indeed, retrospective analyses (not presented) based on our results showed that a sample size of 50 patients who had undergone THA resulted in underpowered comparisons.36 Accordingly, further research should be conducted to investigate the reproducibility and validity of the PASE questionnaire in a larger sample of patients who have undergone THA.
The PASE total scores of our patients with THA were similar to those previously reported for healthy populations of older adults.10,15,17,18 In the present study, a general but nonsignificant tendency to report lower PASE total scores at the second questionnaire administration compared with the first questionnaire administration was observed. It is thought that because patients are encouraged by surgeons and physical therapists to exercise after surgery, they tend to overestimate their level of physical activity during the first questionnaire administration in order to gain social approval by the interviewer.37,38 Our patients likely answered the questions more precisely and truthfully during the second questionnaire administration because they knew that the reported physical activity level would be compared with that objectively recorded with accelerometry.
We cannot determine with certainty the population value for ICC (2,1), according to our results. Indeed, although it appears that the reliability coefficients support the alternate hypothesis (ICC [2,1] ≥.70),22 we cannot exclude that ICC (2,1) was lower than .70 because the lower limits of the 95% CIs for ICC (2,1) were less than the null value. Therefore, we cannot state if the PASE questionnaire has acceptable reliability for assessing the physical activity level of patients after THA surgery. Accordingly, it cannot be definitely demonstrated if the PASE questionnaire has an adequate ability to distinguish between different levels of physical activity within patients after THA, nor can it be definitely demonstrated if the PASE questionnaire can be used for studies with discriminative purposes, where patients have to be allocated into different groups according to their level of physical activity.
However, the PASE questionnaire demonstrated overall low agreement according to the benchmarks proposed by Flansbjer et al23 (SEM <10% and SDC <28%). The SDC represents the smallest change in the questionnaire score that can be interpreted as a “real change” above measurement error.24,34 For detecting meaningful changes in the level of physical activity over time or after an intervention, the SDC of the questionnaire should be smaller than the clinically important change. Unfortunately, the minimal change in the PASE total score that can be interpreted as clinically relevant is not known. However, the acceptability of the agreement results can be interpreted using the noise-to-signal ratio, which is expressed as an effect size (preintervention to postintervention difference divided by SEM).39,40 Based on previous studies, the PASE questionnaire showed preintervention to postintervention changes in total score lower than 20%.13,41 Tsonga et al41 showed an increase of 18% in the PASE total score for female patients between 3 and 6 months after total-knee arthroscopy, whereas Fernandes et al13 reported an increase of 8% in the PASE total score for a group of patients with hip osteoarthritis undergoing a 12-week supervised exercise program. Accordingly, the effect sizes detected by the PASE questionnaire range from .35 to .78 following the above-mentioned interventions. Interpreting the intervention effects as the signal, the noise (indicated by the agreement) is about 1.5 to 3 times larger than the signal. Overall, these findings suggest that the PASE questionnaire has a very limited ability to distinguish measurement error from real changes, and hence the questionnaire is not very suitable for studies with evaluative purposes, where the physical activity level of patients has to be monitored over time.
According to our results, we also cannot determine with certainty the population value for the validity correlation coefficients. Indeed, although it appears that the validity correlation coefficients do not support the alternate hypothesis (r≥.50),22 we cannot rule out this possibility given that the 95% CIs include the null value. Therefore, we cannot state if the PASE questionnaire has adequate construct validity for assessing the physical activity level of patients after THA surgery. Because the PASE questionnaire was specifically developed to focus on light-intensity recreational, household, and work-related activities engaged in by older individuals,10 we examined the construct validity of the questionnaire by assessing the association between the PASE total score and the minutes spent in light physical activities measured with accelerometry. Indeed, for the total sample and for the THAlate group, we observed a tendency for higher relationships between the PASE score and the time spent in light-intensity physical activities than the time spent in moderate- to high-intensity physical activities. In addition, when compared with patients in the THAearly group, patients in the THAlate group tended to show overall higher relationships between the PASE score and accelerometry outcomes. This finding may be partially explained by the fact that many recreational and sport activities reported by our patients in the PASE questionnaire early after THA are underestimated (eg, stationary biking, cross-training, upper limb muscle strength exercises)42 or even not measured (eg, water gymnastics, swimming) by accelerometers.
When compared with patients in the THAlate group, and even with healthy older individuals assessed in previous studies, patients in the THAearly group demonstrated higher participation in recreational and sport activities and lower participation in household and work-related activities.10,15,17,18 These observations are further reinforced by the fact that the PASE questionnaire was originally developed for healthy older people10 and that, in contrast to patients in the THAearly group, the participation of our patients in the THAlate group in recreational, household, and work-related activities is very similar to the participation previously reported in healthy older adults.10,15,17,18
In conclusion, further study with a larger sample size is necessary to obtain precise reliability and validity estimates. Nevertheless, inadequate agreement calls into question the PASE questionnaire's ability to assess the physical activity level of patients after THA surgery.
Footnotes
Dr Casartelli, Dr Impellizzeri, and Dr Maffiuletti provided concept/idea/research design. Dr Casartelli and Dr Maffiuletti provided writing and project management. Dr Casartelli provided data collection and participants. Dr Casartelli, Mr Bolszak, and Dr Maffiuletti provided data analysis. Dr Maffiuletti provided facilities/equipment and institutional liaisons. All authors provided consultation (including review of manuscript before submission).
The study protocol was conducted according to the Helsinki Declaration, and the protocol was approved by the Ethics Committee of the Canton of Zurich, Switzerland.
The study was funded by the Schulthess Clinic Research Fund (internal funding).
- Received November 15, 2013.
- Accepted August 11, 2014.
- © 2015 American Physical Therapy Association