Long-term prediction of functional outcome after stroke using single items of the Barthel Index at discharge from rehabilitation centre

Abstract

Purpose: To determine the prognostic value of single items of the Barthel Index (BI) at discharge from rehabilitation, in predicting independence in personal activities of daily living (ADL) (BI score ≥95/100) at five years after stroke. Method: People with stroke were recruited consecutively from four European rehabilitation centres. BI was assessed on discharge and at five years after stroke. Stepwise multivariate logistic regression analysis was used to determine independent predictors of BI score ≥95/100 at five years after stroke. Thereupon, percentage chance of reaching BI ≥ 95/100 at five years after stroke was calculated. Results: Data were available for 153 patients. Independence in dressing (odds ratio (OR) = 5.22, 95% confidence interval (CI) = 1.85–14.76, p = 0.002) and bathing (OR = 8.10, 95% CI = 3.40–19.32, p < 0.0001) were independent predictors. Independence in both items resulted in 74.1% (57.6–85.8) chance of reaching BI ≥ 95/100 at five years after stroke. Dependence in both items resulted in 6.3% (5.1–7.9) chance. Independence in bathing, but dependence in dressing resulted in 35.4% (30.7–40.4) chance whereas the opposite resulted in 26.1% (20.7–32.3) chance. Conclusion: Simple assessment of dressing and bathing on discharge from rehabilitation enables therapeutic staff to predict prognosis for long-term independence in personal ADL. This method can be used for early identification of persons with stroke who need intensive follow-up.

Implications for Rehabilitation

• (In)dependence for dressing and bathing at discharge from a rehabilitation centre are significant factors in the prediction of (in)dependence in personal ADL at five years after stroke.

• This predictive tool can be used for targeting inpatient stroke rehabilitation and early identification of those patients who need intensive follow-up.

• Activities of daily living
• prognosis
• stroke

Acknowledgements

The authors wish to thank all patients for their participation in the CERISE study and its follow-up and R. Buyl, PhD (Department of Biostatistics and Medical Informatics, Vrije Universiteit Brussel) for the statistical advice.