Objective Metrics of Patient Activity: Use of Wearable Trackers and Patient Reported Outcomes in Predicting Unexpected Healthcare Events in Cancer Patients Undergoing Highly Emetogenic Chemotherapy.

6519 Background: Functional status and predictors that identify those cancer patients at risk for unplanned hospitalization can have broad implications for the health care system and clinical trials. We evaluated the feasibility of monitoring physical activity (PA) using wearable activity trackers in cancer patients on highly emetogenic chemo as well as potential correlations between PA and unplanned healthcare events (UHE) and ECOG scores. Methods: This study was conducted as a multi-institutional single arm observational clinical trial of 65 patients with solid tumors undergoing highly emetogenic chemo based on Hesketh classification. We measured PA by analyzing daytime hourly metabolic equivalents (1 MET = resting metabolic rate) from 10 AM - 7 PM over 60 days via Microsoft band 2. Patient reported outcome data was collected using smartphone apps. UHE were collected by review of medical records over the 60 days of band wear plus 90 days of clinical follow up. Results: Data was successfully captured from 41 of the 65 activity trackers. Patients were compliant with wearing the activity trackers for > 7 of 9 total hrs on 67.7% of study days. Only 9 out of 41 patients exhibited > 60 hours of non-sedentary activity, defined as > 1.5 METs, over the 60-day band wear period. Mean step counts/day were similar between higher and lower PA groups at 2564 steps/d and 2261 steps/d respectively. 9 patients with > 60 hrs of 1.5 METs had significantly fewer UHE compared to the 32 patients with < 60 hrs of 1.5 METs (p = 0.02). The physician reported ECOG scores had no correlation with PA or UHE. Conclusions: In solid tumor patients undergoing highly emetogenic chemo regimens, activity trackers are feasible and identify those patients with a profile of lower activity that predicts UHE. Incorporation of activity trackers has the potential to identify patients who are at need for interventions to prevent hospitalization and may also predict the subset of patients enrolled in clinical trials who are more likely to record serious adverse events.