Spinal cord injury "outliers": An analysis of etiology, outcomes, and length of stay

Spinal cord injury (SCI) requires significant and expensive medical intervention, including prolonged hospitalization and intense in-patient treatment and rehabilitation. Development of predictive models for lengths of stay in spinal cord injury patients provides a method for early prediction of patients that will require greater care, incur greater costs, and need more intensive medical and rehabilitative services. Early identification of SCI patients at high risk of extended lengths of stay will also allow physicians to treat those patients more aggressively, and permit families, as well as sponsors, to estimate the costs of long-term care. Results of a forward-conditional stepwise multiple logistic regression indicate that the model including age at injury, number of days to rehabilitation admission, number of pressure ulcers, number of medical complications, level of injury, and sponsor of initial hospitalization significantly (chi(2) = 220.063, p < 0.001) predicts outliers in terms of rehabilitation length of stay. Overall, the percentage of persons who were correctly classified by the multivariate model was 96.92% (chi(2) = 66.85,p < 0.001). The correct prediction rate for outliers was 46% and nonoutliers 97%. This model provides a tool that can be used by health providers, sponsors and patients to aid in the identification of individuals with SCI that may require extended lengths of stay. Today, unfortunately, our health care system is burdened by the heavy influence of economics rather than functional outcome. We maintain that implementation of this model will help to improve care of the SCI patient by early identification of those in need of more extensive resources and improve the economic efficiency needed to provide maximal functional outcome.