Background Adequate physical activity levels have been associated with wide-ranging beneficial health outcomes. It is therefore important to adequately measure physical activity. This is often done by the SQUASH (Short Questionnaire to Assess Health-enhancing physical activity), a patient-assessed instrument. However, this is a subjective method of which data on validity in large, middle-aged population groups are lacking.
Objectives To assess 1) the construct validity of the SQUASH, and 2) participant characteristics associated with difference between SQUASH and ActiHeart outcome.
Methods Cross-sectional data from the population-based Netherlands Epidemiology of Obesity (NEO) study were used, of which participants were 45 to 65 years of age. Physical activity (in Metabolic Equivalent of Task (MET) hours) was assessed using the SQUASH on one week and in a random subset using the ActiHeart accelerometer for approximately 96 hours continuously, extrapolated to one week.
Convergent validity of the SQUASH was assessed by calculating Spearman’s rank correlation between the SQUASH and the ActiHeart. Extreme group validity was assessed based on the difference in SQUASH activity between 1) 25% of participants with highest and lowest activity according to ActiHeart 2) participants with and without any comorbidity, and 3) participants with BMI< 25 kg/m2 versus participants with BMI> 30 kg/m2. Discriminative validity was assessed by a lack of Spearman’s rank correlation between SQUASH physical activity and participant height (correlation <0.10).
Participant characteristics associated with the difference between SQUASH and ActiHeart outcome were assessed by comparing quintiles of SQUASH minus ActiHeart outcome with a reference quintile (=quintile with lowest difference between SQUASH and ActiHeart).
Results SQUASH data were available for 6,550 participants (mean age 56 years, 56% women, 24% having any comorbidity) of which 875 had ActiHeart data. Median (interquartile range) physical activity was 118 (76;154) MET hours/week according to the SQUASH and 75 (58;99) according to the ActiHeart (mean difference 47 (standard deviation 63)). The Bland-Altman plot of the difference between SQUASH and ActiHeart measurement per patient is shown in figure 1.
Convergent validity was weak (ρ = 0.20, p < 0.01). Participants with the 25% highest ActiHeart scores had 38 MET hours/week more SQUASH activity (95% confidence interval (95%CI) 27;50) than the lowest 25%. Participants with any comorbidity had -13 (95%CI -17;-9) MET hours per week of SQUASH compared with those without any comorbidity, participants with BMI > 30 had -13 (-17;-9) MET hours per week compared with BMI< 25 kg/m2. Discriminative validity was present (ρ = 0.01, p = 0.41). Cut-offs of the quintiles of participants according to SQUASH minus ActiHeart physical activity (range: -61 to 411) were: -6, 27, 61 and 91 MET hours/week (reference quintile: -6;27). Men were overrepresented in the lowest quintile of SQUASH minus ActiHeart activity (-161 to 6 MET hours/week, odds ratio (OR) for being male compared with the reference quintile: 2.35 (95%CI 1.52;3.64)). Men were also overrepresented in the second-highest quintile (61 to 91 MET hours/week, OR 1.89 (95%CI 1.23;2.94)) and highest quintile (91 to 411 MET hours/week, OR 2.47 (95%CI 1.61;3.84)).
Conclusion In this study to investigate the convergent validity of the SQUASH, there were differences between MET hours assessed by SQUASH and ActiHeart, with weak correlation between these instruments; SQUASH activity was on average higher than ActiHeart. However, discriminative and extreme group validity of the SQUASH was present. Men appeared more prone for a difference between SQUASH and ActiHeart outcome than women. These results indicate that construct validity of the SQUASH is sub ideal, although an alternative explanation could be that SQUASH and ActiHeart measure different domains or have different underlying methods. This underscores the importance of further study on adequate physical activity measurement.