1272 a Pilot Routine Electronic Health Record Functional Tracking Score for Older Patients in Hospital

Abstract

Introduction Approximately one-third of older patients leave hospital with a new functional impairment. Tracking rehabilitation progress following acute illness could improve recognition and understanding of hospital-acquired disability. However, traditional mobility and functional scores include measures that are not part of routine rehabilitation therapy, adding a time burden for staff to report. Capturing data already recorded in routine electronic records could provide an efficient patient tracking measure of rehabilitation success. Methods A scoping literature review appraised existing scores of mobility and functional status. Analysis of 15 admissions through the Royal Infirmary of Edinburgh identified mobility and functional domains which were reliably recorded in free text electronic health records. A pilot score was drafted, comprising admission and discharge scores (0-30 points), medical progress (0-10), physiotherapy and occupational therapy tracking (0-30). Higher scores indicate greater functional independence. Expert feedback was obtained through focus group discussion with physiotherapists and occupational therapists. The approach was tested in a fresh set of six case studies. Two independent scorers applied the scoring schema and agreement was assessed using Cohen’s weighted-kappa coefficient. Results The selected electronic health records contained 438 medical, 352 nursing and 183 therapist entries. Existing measures such as the Barthel Index were not recorded for any patient. Focus group discussion identified value in the overall approach and informed item-weighting. The pilot functional score allowed visualisation of rehabilitation trajectories over the course of each admission. Excellent inter-rater reliability was demonstrated for the medical (Cohen’s Kappa 0.99, 95% confidence interval [CI] 0.96–1.00) and physiotherapy (Kappa 0.96, 95% CI 0.93–0.99) components. Conclusions A functional tracking score generated from routine health records proved feasible and reproducible in this pilot. Future development should assess validity, reliability and prognostic power in larger populations, exploring automation using natural language processing. Development of graphic visualisations may aid communication within multidisciplinary teams.