A methodology for predicting lengths of stay for spinal cord injury patients

Although the focus of Medicare's prospective payment system is on acute care hospitalization, other health care delivery systems are likely to be included in the future. In this paper, we report on a model for predicting length of stay (LOS) for spinal cord injury patients requiring hospitalization and rehabilitation services. Using multiple regression on data on 2,169 patients from the National Spinal Cord Injury Statistical Center data base, we predicted LOS for seven injury severity groups, and then validated our results with a second sample of 505 patients. Our predictive model explained a robust 45% of the variance in LOS and adjusted hospital charges. We conclude that our model would be useful in projecting LOS for patients whose injuries or illnesses require extensive rehabilitation services after acute care.