Functional Decline Over 1-year Follow-up in a Multicenter Cohort of Polypathological Patients: A New Approach to Functional Prognostication

Abstract

Little is known about the fitness of the available tools in predicting functional decline of polypathological patients (PPs). Our objective was to assess accuracy of the Triage Risk Screening Tool (TRST), the Variable Indicative of Placement risk (VIP) and to develop a specific functional prognostic index adjusted to this population in a multicenter cohort of hospital-based PP. Prospective 12-month follow-up study of PPs from 36 hospitals. Functional decline was defined as loss of ≥20 points on Barthel’s index (BI). Accuracy of TRST/VIP was assessed by calibration/discrimination tests. Development of the new score was performed by dividing into a derivation cohort (constructing the index by logistic regression), and a validation cohort (in which calibration/discrimination of the index were tested). Nine hundred and fifty-eight patients from the 1632 included survived during follow-up. Basal/12-month BI was 85/70, respectively. Mean fall in BI score was 11.7 ± 24 points [353 (36.8%) fell by ≥20 points]. The activities for daily living that declined most frequently were toilet use, grooming, dressing and bathing. TRST/VIP fitted well but their discrimination power was poor (area under the curve = 0.49 and 0.46, respectively). A simplified PROFUNCTION index was derived containing seven items (≥85 years, neurological condition, osteoarticular disease, III–IV functional class of dyspnea, ≥4 polypathology categories, basal BI < 60, and social problems). Functional decline risk ranged from 21% to 24% in the lowest risk group (0 items) to 38–46% in the highest (4–7 items). Calibration as well as discrimination power (area under the curve = 0.56–0.59) of this simplified index were good. We developed and validated a new functional prognostic index specifically focused on these patients with better discrimination power than other tools available.