Self-reported home exercise adherence: a validity and reliability study using concealed accelerometers

Background: Accurate measurement of adherence to prescribed exercise programs is essential. Diaries and self-report rating scales are commonly used, yet little evidence exists to demonstrate their validity and reliability.Objectives: To examine the concurrent validity of adherence to home strengthening exercises measured by (1) exercise diaries and (2) a self-report rating scale, compared to adherence measured using an accelerometer concealed in an ankle cuff weight. Test-retest reliability of the self-report rating scale was also assessed.Methods: In this clinical measurement study, 54 adults aged 45 years or older with self-reported chronic knee pain were prescribed a home quadriceps-strengthening program. Over 12 weeks, participants completed paper exercise diaries and, at appointments every 2 weeks, rated their adherence on an 11-point numeric rating scale. A triaxial accelerometer was concealed in the ankle cuff weight used for exercises. Self-reported adherence rating scale data over each 2-week period were analyzed using descriptive statistics, the Wilcoxon signed-rank test, and a Bland-Altman plot to assess agreement, Spearman correlations for validity, and intraclass correlation coefficients for test-retest reliability.Results: Exercise adherence was significantly overestimated in diaries during the 12 weeks (diary median, 220 exercises; accelerometer, 176; P<.001) and was moderately correlated with accelerometer data (r = 0.52; 95% confidence interval: 0.26, 0.69). A Bland-Altman plot indicated large between-participant variability in agreement between these measures. Self-reported adherence showed poor to fair correlations with accelerometer data (mean r = 0.23–0.39), and less than acceptable reliability (intraclass correlation coefficient = 0.79; lower 1-sided 95% confidence limit, 0.68).Conclusion: Exercise diaries showed questionable validity and variable levels of agreement compared with accelerometer-measured exercise completion. A self-reported adherence rating scale had limited validity and less than acceptable test-retest reliability