Effect of centre on outcome of bone-marrow transplantation for acute myeloid leukaemia

Background There is increasing pressure for the recognition and replication of good clinical practice. We undertook a study to assess the variability in outcome of allogeneic bone-marrow transplantation among major European centres. Methods We studied 13 centres, including 522 patients (aged 16-55 years), which had undertaken more than 30 bone-marrow transplantations between Jan 1, 1987, and Dec 31, 1995, for acute myeloid leukaemia in first complete remission. We undertook a (global) multivariate analysis of all factors known previously to influence outcome and a stratified analysis that initially defined, by multivariate analysis, significant variables in this study and then used a proportional-hazard model including centres. Findings The overall results at 3 years were 57% (95% CI 53-61) for leukaemia-free survival (LFS), 23% (19-27) for relapse incidence (RI), and 26% (22-30) for treatment-related mortality (TRM) with a range for centres of 36-75%, 10-37%, and 8-54%, respectively. Both methods of analysis showed the centre effect to be highly significant for LFS and TRM, but not for RI. Variables associated with a significantly poor outcome were age over 43 years (p=0.01), time from diagnosis to first complete remission longer than 65 days (p=0.02), and centre (p=0.013) for LFS, and age over 43 years (p=0.023), time from first complete remission to transplantation of longer than 93 days (p=0.03), and centre (p=0.001) for TRM. Moreover, different centres had different prognostic criteria for good-risk or bad-risk patients indicating that risk factors do not have the same impact in each individual centre. Interpretation The outcome of bone-marrow transplantation for acute myeloid leukaemia in first complete remission is influenced by the centre in which the procedure is done, even with adjustment for known prognostic risk factors. Significant prognostic factors vary among centres, which means that the relative risk is not the same in each individual centre. However, centres may treat populations with different risks of as yet unidentified prognostic factors. Experience may partly account for the difference in outcome among centres.