Predicting length of stay in the rehabilitation of stroke patients.

Stroke welcomes Letters to the Editor and will publish them, if suitable, as space permits. They should not exceed 1,000 words (excluding references) and may be subject to editing or abridgment. Please submit letters in duplicate, typed double-spaced. Include a fax number for the corresponding author. To the Editor: The anticipation of a patient's length of stay has obvious ramifications for the selection of an appropriate level of rehabilitation service and for discharge planning. The findings of Galski and associates, 1 therefore, have great importance. The intent of this letter is to compare the results of their research in a comprehensive rehabilitation setting with the results of a similar retrospective analysis in our exempt, hospital-based acute rehabilitation unit. The correlations that Galski and coworkers reported between length of stay and admission Functional Independence Measure (FIM) scores were-.65 for occupational therapy items,-.59 for physical therapy items, and-.63 for psychology items. Using multiple regression they found that only admission occupational therapy and psychology FIM scores made significant individual contributions to the prediction of length of stay. Our Table shows the results of the analysis of the data from 42 patients with stroke admitted consecutively to our unit. The lengths of stay ranged from 5 to 76 (mean, 21.6) days. The Spearman correlations of the self-care, mobility, and locomotion category FIM scores, and of the total FIM score with length of stay, are similar in magnitude to the correlations reported by Galski and colleagues. 1 Of particular note in our analysis is the correlation between the bed-to-chair transfer FIM score and length of stay. The correlation (.811) is higher than that of any other FIM score (individual, categorical, or total) with length of stay. Of all the other individual FIM scores, only one, social interaction (cognition), was found by multiple regression to make a significant additional contribution to the prediction of length of stay. Together the bed-to-chair transfer and social interaction FIM scores predicted 69.7% of the variance in length of stay. Both our analysis and that of Galski and associates leave a great deal of variance in length of stay unexplained. Nevertheless, the explanation of greater than 50% of the variance in length of stay is quite remarkable given the myriad of environmental, social, and financial factors that might conceivably contribute to length of stay among patients with stroke.