Inpatient hospital utilization among veterans with traumatic spinal cord injury.

OBJECTIVE: To describe the pattern of inpatient hospital utilization, up to 15 years after injury, among a cohort of veterans with service-connected traumatic spinal cord injury (SCI). PATIENTS: A cohort of 1,250 male veterans, with traumatic SCI occurring between 1970 and 1986, who visited the VA within 1 year of injury, was assembled from VA administrative files; diagnosis was verified by examining hospital discharge summaries. DESIGN: Computerized record linkage among Department of Veterans Affairs (VA) administrative files was used to determine patterns of inpatient hospital utilization. MAIN OUTCOME MEASURE: Pattern of inpatient admissions and length of stay (LOS). RESULTS: Patients were typically white males injured in their mid-twenties. The initial VA hospitalization began approximately 6 weeks after injury and lasted 4 to 7 months, depending on injury level and completeness. Subsequent hospitalizations usually lasted approximately 10 days, but 22% of stays exceeded 1 months. Most hospitalizations took place in specialized SCI Centers. Comparing the 1980s with the 1970s, patients in the 1980s entered VA facilities sooner after injury, were more likely to visit SCI Centers, and had shorter initial stays. Rates for the incidence of rehospitalization decreased rapidly in years 2-5 after injury and declined less rapidly thereafter. Occupancy rates and proportion rehospitalized followed similar patterns. The incidence rate for persons with complete quadriplegia was approximately twice that of patients with incomplete paraplegia. Between 1970 and 1991, both the rehospitalization incidence rate and LOS decreased by approximately 20%. Only 10% of patients accounted for 46% of the total LOS. LOS during the first five years was predictive of later LOS. CONCLUSIONS: The pattern of rehospitalization in VA facilities was generally consistent with that of the Model Systems. Efforts toward preventing rehospitalization should target persons with previous high utilization.

Duke Scholars

Author Gregory P. Samsa Biostatistics & Bioinformatics, Division of Biostatistics