Prognostic models for physical capacity at discharge and 1 year postdischarge from rehabilitation in persons with spinal cord injury

Objective: To develop prognostic models for physical capacity at discharge and 1 year after discharge from inpatient rehabilitation in persons with spinal cord injury (SCI). Design: Inception cohort; data collected at start of rehabilitation (n=104), at discharge (n=81), and 1 year later (n=74). Setting: Eight Dutch rehabilitation centers. Participants: Patients with SCI at initial rehabilitation. Interventions: Not applicable. Main Outcome Measures: Physical capacity determined by endurance capacity (peak oxygen uptake [Vo(2)peak, in L/min] and power output [POpeak, in watts]) during a maximal exercise test, arm muscle strength, and respiratory function. Multiple regression models, either with or without prior outcome, evaluated subject, lifestyle, and lesion-related predictors. Results: Only start Vo(2)peak contributed to the prediction of discharge Vo(2)peak (R-2=.51). Discharge Vo(2)peak contributed to its prediction 1 year later (R-2=.75). Start POpeak, sex, age, and level of lesion contributed to discharge POpeak (R-2=.73). Discharge POpeak, hours of employment before injury, and level of lesion contributed to POpeak 1 year later (R-2=.81). Models without prior outcome explained less variance. Education, employment, body mass index, not smoking, and conservative stabilization of the spine positively contributed to endurance capacity. Muscle strength was well predicted (R-2 range,.68-.84). Without prior outcome, respiratory function was poorly predicted. Conclusions: Because prior outcome contributed to an accurate prediction, the early assessment of physical capacity is important in establishing prognoses. Although their accuracy warrants caution in their application, models could complement clinical expertise when informing patients about expected physical outcome and identifying those at risk of low physical capacity.