Diabetes and early postoperative outcomes following lumbar fusion.
Browne JA., Cook C., Pietrobon R., Bethel MA., Richardson WJ.
STUDY DESIGN: Retrospective cohort study using data from the Nationwide Inpatient Sample administrative data from 1988 through 2003. OBJECTIVE: To examine perioperative morbidity and mortality for patients with and without diabetes mellitus following lumbar spinal fusion. SUMMARY OF BACKGROUND DATA: Diabetes has been associated with worse outcomes in a variety of orthopedic procedures including spinal surgery. There is limited evidence that diabetic patients have more complications following lumbar fusion with little published data to support this conclusion. METHODS: Data from 197,461 patients who underwent lumbar fusion were included. Over 11,000 patients (5.6%) with a postoperative diagnosis of diabetes mellitus were identified. Selected variables were used for comparison of patients with and without diabetes. Bivariate statistical analyses compared postoperative complication rates while multivariate statistics were used to determine likelihood of complications with diabetes. RESULTS: Bivariate analysis demonstrated that diabetes was significantly associated with postoperative infection, need for transfusion, pneumonia, in-hospital mortality, and nonroutine discharge (P <or= 0.001). Adjusted multivariate regression analyses, however, suggested no difference in mortality although infection, transfusion, and nonroutine discharge continued to be highly significant (P <or= 0.002). Significantly higher inflation adjusted total charges were also present with patients with diabetes as well as increased lengths of stay (P < 0.001). CONCLUSION: This nationally representative study of inpatients in the United States provides evidence that diabetes is associated with increased risk for postoperative complications, nonroutine discharge, increased total hospital charges, and length of stay following lumbar fusion. Prospective studies to determine causality as well as the potential impact of diabetes control on these variables have not yet been done.