Predicting Persistent Reduced Physical Performance in Adults 65+ in Emergency Departments: A Temporal Validation.

Abstract

Background and purpose: Targeted interventions to maintain physical performance in older adults are important. We aimed to validate an existing 4-item prediction model and, if necessary, develop a new model for early identification of adults aged 65+ with persistent reduced physical performance.

Methods: A temporal validation study on adults aged 65+ admitted to the emergency department for medical reasons and who performed ≤8 repetitions in the 30-second chair-stand test (30s-CST) within the first 48 hours of admission. The primary outcome was the number of 30s-CST repetitions (≤8 or >8) performed at the older adults' homes 3 to 4 weeks after admission.

Physical performance and self-reported data were analyzed using univariate and multivariate logistic regressions. The optimal model was chosen based on the area under the curve.

Results and discussion: The validation of a previously developed 4-item prediction model indicated that the model did not consistently apply to other populations of older adults. To develop a new model aimed at identifying acutely hospitalized older adults with persistent reduced physical performance, we combined 2 samples comprising 132 older adults with a 30s-CST score > 8 and 250 adults with a 30s-CST score ≤ 8 when measured 3 to 4 weeks after admission. The new model included 5 variables: age >80, female, self-reported difficulties in climbing a flight of stairs, less good/poor self-rated health, and a 30s-CST score ≤ 5. The model had an area under the curve of 84%. The model is expected to improve the identification of older adults with persistent reduced physical performance compared to health professionals' subjective assessments and/or adults' self-reported information.

Conclusions: The initial 4-item prediction did not consistently apply to other populations of older adults. Consequently, 2 samples were combined and a 5-item model was developed. Since the model involves only 5 items, it is easy to implement and provides health professionals an opportunity for targeted intervention on older adults during and after acute hospitalization. A validation study for the 5-item model is necessary.