Despite recognized benefits of regular physical activity, little is known about the physical activity behavior of patients after THA. We therefore examined the measurement properties of the SQUASH as a tool to provide more insight into this behavior, as this short, self-reported physical activity questionnaire has been shown to be a fairly reliable and valid tool to assess the physical activity behavior of the general Dutch adult population [
10].
The Spearman correlation for overall reliability of the SQUASH in our study was 0.57. As, to our knowledge, this is the first study to assess the measurement properties of a physical activity questionnaire in patients after THA, we are unable to compare our results to studies in a similar population. However, this overall reliability of the SQUASH is almost identical to the reliability of 0.58 found in the study by Wendel-Vos [
10], assessing the measurement properties of the SQUASH in a population of 50 healthy adults (mean age 44 ± 6 yr). Although our study design was largely identical to that of Wendel-Vos, if we are to compare our results to those of the original study into the reliability and validity of the SQUASH it must be stated that the Wendel-Vos study differed from ours in that participants first completed the SQUASH for a second time before the Actigraph™ readings were performed. This was done to prevent a possible influence on reliability due to an increased awareness about physical activity, which might occur when the Actigraph™ is worn between the two measurements of the SQUASH. Reliability is also consistent with the reliability of other physical activity questionnaires, validated by means of an accelerometer in adult populations. In a review of seven physical activity questionnaires, validated with accelerometers in adults, reliability varied between 0.34 and 0.89 [
22]. Also, a study into the reliability and validity of the International Physical Activity Questionnaire (IPAQ), which is comparable to the SQUASH but was developed for cross-national monitoring of physical activity, showed a Spearman correlation coefficient for the short forms of the IPAQ ranging from 0.25 to 0.88, with a pooled reliability of 0.76 [
23]. The reliability found in our study is thus comparable with reliabilities found in other physical activity questionnaires. Furthermore, Bland and Altman analysis showed no systematic bias on total activity scores between test and retest.
The total activity score on the SQUASH correlated significantly with the mean activity counts per minutes measured by the Actigraph™ (
r
Spearman = 0.67). The total minutes of activity as assessed by the SQUASH and the Actigraph™ also correlated significantly (
r
Spearman = 0.56). Hence the SQUASH can explain 31% of the total variation in physical activity. When comparing the tertiles of activity scores with the tertiles of activity counts, exact agreement was 67%, which is fair to good. The weighted kappa was 0.50, representing fair agreement. The relative validity of the SQUASH in our study is higher than that found in the study by Wendel-Vos, showing a Spearman correlation coefficient between total activity score and accelerometer readings of 0.45. Comparison of the tertiles of the activity score with tertiles of the activity counts in their study showed an exact agreement of 46% and a weighted kappa of 0.30, which are lower values than those found in our study. In the Sallis review of seven physical activity questionnaires [
22], validity correlations ranged from 0.14 to 0.53. The IPAQ short forms showed validity ranging from -0.12 to 0.57, with a pooled Spearman correlation coefficient of 0.33 [
23]. It can therefore be concluded that the validity found in our study lies in the upper range of validity found in other questionnaires validated with an accelerometer in adult populations. However, consideration should be given to the sizeable systematic bias between the scores on the SQUASH and the Actigraph™ readings. This systematic bias may be the result of overestimating physical activity level by the SQUASH, as people tend to overestimate their physical activity level [
22]. At the same time the Actigraph™ may have underestimated physical activity level. The Actigraph™ is a uniaxial accelerometer for vertical movement and is relatively insensitive to physical activities that require little vertical movement. When positioned on the waist activities such as cycling or activities involving large upper-body movement may be underestimated. Additionally, the accelerometer is not waterproof and therefore cannot be worn during activities such as swimming. Since in our study 21% of patients reported swimming and 77% cycling as part of leisure-time activities, this will have led to an underestimation of physical activity by the Actigraph™. The systematic bias may also reflect true variations in participants' physical activity levels. Since the SQUASH asks patients to recall physical activity during an average week in the past months, this timeframe was not identical to the period of time used to acquire the accelerometer data. Furthermore, to estimate the energy expenditure spent in physical activity the activity counts as obtained by the Actigraph™ have to be transformed into MET values. To do this, regression equations have been developed from studies under laboratory as well as field conditions [
20,
24,
25]. In line with the study of Wendel-Vos we used the Freedson equation to transform activity counts into MET values. As this regression equation was developed under laboratory conditions, it may not be valid under the "field conditions" of our study as it particularly has been shown to underestimate moderate-intensity activity [
26]. Additionally, the regression equation was developed in adults and may not be appropriate for older adults. However, to our knowledge, there are no regression equations specifically for older adults. This may be another factor accounting for the differences found in our study in terms of time spent in the different intensity categories between the physical activity questionnaire and the accelerometer.