Comparison Of Physical Activity Scale For Individuals With Physical Disabilities And Accelerometry In Arthritic Individuals

Abstract

PURPOSE: To compare physical activity (PA) estimates from the PA Scale for Individuals with Physical Disabilities (PASIPD) to accelerometer data in individuals with arthritis. METHODS: Adults aged 64-88y, a sub-sample recruited for a larger protocol, and self-reported with arthritis were included in this analysis. Subjects completed a seven-day monitoring period that included wearing a thigh worn activPAL (AP) accelerometer during all waking hours and completing a wear-time log. At the end of the monitoring phase individuals completed the PASIPD. Accelerometer data was processed with PALstudio (v8.9.1.24) and raw data was manipulated using the activPAL processing package in RStudio (1.2.1335) to calculate hours in stepping, light (1.5-2.99 METs) PA (LPA), and moderate-vigorous (>3.0 METs) PA (MVPA). Items were taken from the PASIPD to calculate hours of LPA and MVPA, and to derive a total activity score. Spearman correlations comparing AP stepping and total PASIPD, AP LPA and PASIPD LPA, and AP MVPA and PAISPD MVPA were computed. Wilcoxon Signed Rank tests were computed for differences between AP and PASIPD LPA and AP and PASIPD MVPA. RESULTS: Twenty-seven subjects (16 male, 7 female) ([mean±SD] age 75.8±6.2 yrs; height 168.4±9.5 cm; mass 83.8±17.6 kg) were analyzed. AP Stepping was significantly correlated with total PASIPD score with a Spearman’s rho of .425; p=.014. AP LPA and PASIPD LPA, and AP MVPA and PASIPD MVPA were significantly correlated, Spearman’s rho of .436; p=.012, and .435; p=.012, respectively. On average, PASIPD underestimated AP LPA by 1.38 hours (p=.024) and overestimated AP MVPA by 1.34 hours (p<.0001). CONCLUSION: Differences between PASIPD and AP measures of LPA and MVPA were apparent, but the PASIPD was moderately correlated to PA levels in this sample of arthritis individuals. Future work on examining the precision and accuracy of PA surveys in heterogeneous populations with varying disease and disability is warranted. This work was partially supported by NIH 1R21HD080828 and NIH 1R01CA215318