Predictive value of a self-administered frailty screening questionnaire for the effectiveness of functional rehabilitation evaluated with the locomotor functional independence measure in a geriatric rehabilitation unit: a multicentre cohort study.

Background: Patient Reported Outcome Measures (PROMs) are questionnaires that collect health data directly from the patient, without any intervention from a third party. The aim of rehabilitation units is to restore function. Functional gain can be evaluated with classic scales, such as the locomotor subscale of the Functional Independence Measure. This study aimed to assess the accuracy of a new self-assessment questionnaire pertaining to physical, sensory and cognitive ability (abbreviated SEPCO) for the prediction of functional prognosis in older patients admitted to a rehabilitation unit.

Methods: In this multicentre observational study including patients admitted to 12 rehabilitation centres in France, all included patients completed the SEPCO on admission. Poor response to rehabilitation was defined as relative effectiveness < 40% on the evolution of the locomotor FIM subscale. Components of the questionnaire potentially associated with the outcome of rehabilitation were confirmed for inclusion upon expert review and summed to form an overall score. The final score had five components: the depression score of the HADS, the SOFRESC vision score, the SOFRESC balance score, the stress urinary incontinence subscale of the USP, and the EPICES socio-economic deprivation score. A logistic regression model adjusted for baseline characteristics assessed the performance of the SEPCO score to predict change in functional status, defined by the relative functional gain for the locomotion subscale of the Functional Independence Measure (FIM).

Results: A total of 153 patients (mean age 79.2 ± 8.1 years, 72.5% women) were included. By multivariate analysis, a 5-scale SEPCO score ≥ 1.1 predicted worse functional improvement with an odds ratio (OR) of 2.575, 95% Confidence Interval (CI) 1.081 to 6.133, p = 0.03. Sensitivity for this threshold was 67.4% (95% CI 52.0-80.5%), with a specificity of 58.8% (95% CI 46.2-70.6%). Having a SEPCO ≥ 1.1 almost doubled the probability of poor response to rehabilitation (from 27.3 to 52.5%).

Conclusion: The SEPCO score can predict poor functional gain from rehabilitation. Future studies should validate this score on an external cohort. The SEPCO could serve as a complement to the initial clinical evaluation performed by physicians, and assist physicians in setting each patient's rehabilitation goals.