Outcome of hematopoietic stem cell transplants in an observational versus a solicited data base. quality control study of the ebmt

Many scientific observations and analyses have been generated from data collected retrospectively at centralized registries. Outcome results of such observational data bases are frequently criticized to rely on incomplete information from incomplete patient populations. EBMT decided to address this question by evaluating in depth the hematopoietic stem cell transplants performed in 1993, making use of its instruments: the annual activity survey, its standardized data collection and the implementation of an audit program. All teams were put under pressure to report all additional transplants performed in 1993. A total of 2995 transplants reported spontaneously by the teams (group A) was compared with 3870 transplants retrieved on solicitation (group B). Patient and disease characteristics as well as outcome concerning transplant related mortality (TRM), relapse incidence (RI), disease free survival (DFS) and overall survival (OS) at 3 years were analyzed. Both group were very similar. More autologous transplants were reported in group A (65%) than in group B (57%) (p<0.0001). More allografted patients in group A were transplanted early in the course of their disease (59% vs 39%, p<0.0001). However, no difference in any outcome measure was observed between the two groups with the exception of patients autografted for breast cancer where group A had a higher RI (74% vs 54%, p=0.0005) and a lower DFS (24% vs 44%, p=0.0005). In conclusion, no major difference was found when spontaneously reported and solicited data were compared. These findings support the use of observational data bases such as of EBMT to provide representative and reliable information.