Accuracy of the Australian national subacute and nonacute patient classification in predicting rehabilitation length of stay for stroke survivors who are ≥ 65 years of age and have lateropulsion
Topics in Stroke Rehabilitation
Taylor & Francis
School of Medical and Health Sciences
Department of Health/Raine Foundation Clinician Research Fellowship (Raine Medical Research Foundation)
Charlies Foundation for Research
Australian Government Research Training Program Scholarship
Lateropulsion is a common impairment after stroke. Regardless of stroke severity, functional recovery is slower in people with lateropulsion, resulting in requirement for longer rehabilitation duration. In Australia, inpatient rehabilitation funding is determined via the Australian National Sub-Acute and Non-Acute Patient Classification (AN-SNAP). AN-SNAP class is determined using age, diagnosis, weighted Functional Independence Measure (FIM) motor score, and FIM cognitive score.
To explore accuracy of the AN-SNAP to predict length of stay (LOS) for people with poststroke lateropulsion.
A retrospective database audit was undertaken. AN-SNAP predicted LOS for each participant was calculated based on 2019 calendar year national benchmarks. A multivariable linear regression model estimated mean differences in reported LOS and AN-SNAP predicted LOS after adjusting for lateropulsion severity (Four Point Pusher Score). A separate logistic regression model assessed whether FIM change during admission was associated with reported LOS exceeding AN-SNAP predicted LOS.
Data were available from 1126 admissions. Reported LOS exceeding AN-SNAP predicted LOS was associated with greater lateropulsion severity on admission. Where AN-SNAP predicted LOS was longer, those with no lateropulsion on admission showed shorter reported than predicted LOS. Greater improvement in FIM during rehabilitation was associated with increased odds of reported LOS exceeding AN-SNAP predicted LOS (OR 1.02, 95%CI 1.01–1.03, p < .001).
Inclusion of a measure of poststroke lateropulsion in the AN-SNAP classification model would result in more accurate LOS predictions to inform funding. Costs of longer rehabilitation LOS may be countered by optimized long-term physical function, reducing requirement for ongoing care.