Contribution of hospital characteristics to the volume-outcome relationship: dislocation and infection following total hip replacement surgery.

OBJECTIVE Mortality and complication rates after total hip replacement (THR) are inversely associated with the volume of THRs performed at hospitals and by individual surgeons. It is not clear, however, why a higher volume of such procedures is associated with better outcomes. We evaluated the contribution of hospital structural characteristics to the volume-outcome relationship in THR by examining the rates and predictors of postoperative complications. METHODS We analyzed data pertaining to 5,211 Medicare patients who underwent primary THR in 1995 or 1996 at 167 hospitals in Colorado, Pennsylvania, and Ohio. Data were derived from several sources, including Medicare Part A and Part B claims, the American Board of Medical Specialties, a hospital survey regarding institution-specific characteristics and structural aspects of the care setting, and the American Hospital Association 1995 Annual Survey. Multivariate models were constructed to determine whether hospital structure or surgeon-associated factors may underlie the relationship between volume of THRs and the occurrence of perioperative orthopedic adverse events, defined as deep wound infection or hip dislocation within 90 days of surgery. RESULTS Of the patients studied, 2.6% experienced an orthopedic adverse event after THR. Sixty-nine percent fewer events occurred in hospitals where >100 THRs in Medicare patients were performed annually, compared with hospitals where <or=25 THRs were performed. In univariate analyses, several hospital-level factors were associated with a reduced (approximately 50%) risk of adverse events, including private (versus public) ownership, membership in the Council of Teaching Hospitals, presence of any residency training program, availability of a dedicated orthopedic nursing unit, and existence of operating rooms with laminar flow exhaust systems. However, the only hospital-level factor associated with adverse events in multivariate models was the use of laminar flow exhaust systems. When surgeon volume was added to the models, it was the strongest predictor of adverse events, with hospital volume and hospital-level factors having no appreciable association with adverse events. CONCLUSION Hospital-level factors were not independent predictors of the association between hospital volume and orthopedic adverse events. The volume of THRs performed by individual surgeons is the most important determinant of orthopedic complications and should be considered in efforts to improve THR outcomes.