Objective: To develop prognostic indexes with which to establish the likelihood of individuals achieving specific grades of physical independence by the conclusion of inpatient rehabilitation.
Design: Logistic regression with prospective validation.
Setting: Five hundred sixty inpatient rehabilitation facilities.
Participants: Records of 218,290 adults discharged in 1995 were used to establish the grades and the indexes predicting those grades. There were 259,806 1997 discharges included in the validation.
Interventions: Not applicable.
Main outcome measures: Six physical independence grades reflecting the most likely profiles of performance across the 13 motor FIM items.
Results: After severity adjustment, patients 65 years of age or younger, compared with those 84 years of age or older, had odds ratios of reaching higher grades ranging from 1.5 (95% confidence interval [CI], 1.4-1.7) to 7.5 (95% CI, 4.3-13.1). Admission to rehabilitation within 2 weeks of disability was associated with more favorable prognoses. Areas under the receiver operating characteristic curve ranged from.80 to.94 for the indexes, with minimal shrinkage on prospective validation.
Conclusion: The models have sufficient reliability to establish from admission information the likelihood that a patient will achieve a specific grade of physical independence by the time of discharge from rehabilitation. The capacity to quantify prognosis has clinical, policy, and research applications.