Variations in graft and patient survival after kidney transplantation in Sweden: caveats in interpretation of center effects when benchmarking.

Benchmarking and comparisons between transplantation centers are becoming more common. A crude comparison indicated a 50% difference in patient survival between centers in Sweden. A 'task group' was formed to refute or confirm and learn from this observation. Patient survival and graft survival of 5 933 patients transplanted at three different transplantation centers in Sweden (Stockholm, Göteborg, and Malmö) were followed up until February 2007. Patient survival and graft survival were compared between the centers with and without consideration being given to important covariates such as time period, type of donation (living or deceased donor), gender, and age. A refined cohort of 2,956 adult patients that had been transplanted for the first time between 1991 and 2007 was assessed in more detail using Cox regression analysis. The difference in patient and transplant outcome observed in the crude comparison diminished considerably after adjustment for differences in case mix and time period of transplantation, and was neither evident nor significant after 1999. Patient survival and graft survival have improved considerably during the time period since 1991. The adjusted hazards ratio for mortality was 0.39 (95% CI 0.29-0.53) for patients who were transplanted after 1999 when compared with those transplanted between 1991 and 1994. Crude comparisons between results from transplantation centers may be severely confounded not only by case mix but also by differences in the proportion of patients transplanted during different time periods. Patient outcome and graft outcome have improved considerably since 1991, and after 1999 center effects were no longer apparent in Sweden.