Patient-focused endpoints in advanced cancer: criterion-based validation of accelerometer-based activity monitoring.

BACKGROUND & AIMS: Objective assessment of daily physical activity (PA) by body-worn accelerometers offers potential as a novel endpoint in the clinical management of advanced cancer patients. This study aimed to assess criterion-based validity of an accelerometer-based activity monitoring system (AM-system), ActivPAL™, using two different methods. METHODS: Advanced cancer in patients and outpatients (Karnofsky Performance Status (KPS) 40-100). ActivPAL™ measurements were validated against (i) observations and (ii) energy expenditure (EE) measured by 2-week doubly-labelled water (DLW) protocol. RESULTS: Absolute errors for mean time spent in different body positions (<0.1%) and number of transfers (0%) were low. Step count error was significantly higher in patients with KPS 40-60 (non-self caring) compared to KPS 70-100 (self-caring) (33 vs. 24%, p = 0.006). Post-hoc mathematical analysis demonstrated that absolute errors for the mean energy expenditure of activity (EEA) (1.4%) and mean total EE (0.4%) were low, but agreement was also low. CONCLUSIONS: AM-systems provide valid estimates of body positions and transfers, but not step count, especially in non-self caring patients. ActivPAL™ can derive estimates of EE but there is considerable variability in results, which is consistent, in part, with the inaccuracy in step count. Further studies are required to assess the validity of different endpoints derived from AM-systems in advanced cancer patients.