Cross-validation of a model for predicting functional status and length of stay in patients with stroke.

OBJECTIVE In a study published in 2002, it was observed that a variable composed by the Functional Independence Measure (FIM) and the trunk control test at admission predicted 66.5% of the FIM at discharge in stroke patients. The objective was to confirm the reproducibility of this predictive model. METHODS Retrospective study of 245 hemiparetic stroke inpatients of the rehabilitation department. The main variables studied were: trunk control test FIM at admission and compound variable (FIM+trunk control test) as independent variables and FIM at discharge and inpatient rehabilitation length of stay as dependent variables. RESULTS Correlation between the compound variable and the length of stay was statistically significant (r=0.59), as was its correlation with the total FIM at discharge (r=0.82). The regression analysis predicted 34.3% of the length of stay variability and 66.4% of the total FIM at discharge variability. CONCLUSION The compound variable is a reliable tool because of its reproducibility in predicting the functional level at hospital discharge in hemiparetic patients.