Assessing Physical Activity Levels In Advanced Prostate Cancer Survivors: Comparing Non-gravitational And Gravitational Cut-points

BACKGROUND: The literature on physical activity (PA) in advanced prostate cancer (PC) is mixed and methodologically heterogeneous, with much of it using non-gravitational cut points (e.g., average counts per minute; cpm). However, these cut points are influenced by study-specific factors such as sampling rate and epoch length, reducing comparability across studies. Therefore, using gravitational cut-points, unaffected by these limitations, is necessary. PURPOSE: The aim of the study is to compare non-gravitational and gravitational units when assessing PA levels in advanced PC survivors. METHODS: Advanced PC survivors (N = 24, mean age = 73.3 ± 7.7 years) wore an ActiGraph GT3X + on their non-dominant wrist for five days. Data were processed using Freedson (1998) (ActiLife; non-gravitational units; cpm), and Hildebrand (2014) cut points (GGIR; gravitational units; mg) and compared using Wilcoxon signed-rank (W) and interclass correlation coefficients (ICC) analysis. RESULTS: There were differences across the non-gravitational and gravitational cut points methods for all activity levels: inactivity (MD = −117 min, p < 0.001), light PA (MD = 421 min, p < 0.001), moderate PA (MD = 8.47 min, p = 0.003) and vigorous PA (MD = −0.617 min, p < 0.001). Agreement was poor for inactivity (ICC = 0.193, 95% CI [-0.026, 0.404]), light PA (ICC = 0.013, 95% CI [-0.012, 0.055]), and vigorous PA (ICC = 0.116, 95% CI [-0.136, 0.354]), but high for moderate PA (ICC = 0.845, 95% CI [0.728, 0.910]). CONCLUSIONS: The two methods showed high agreement for moderate PA based on ICC, but discrepancies and poor agreement were found for other intensities. Gravitational cut points appear advantageous, as they are unaffected by sampling rate and epoch length, enhancing comparability across studies. Future research in PC survivors may choose to focus on non-cut-point metrics such as Euclidean Norm Minus One (ENMO) and Intensity Gradient (IG), as these are not populations specific.