Mortality and volume of cases in paediatric cardiac surgery: retrospective study based on routinely collected data

Abstract Objectives: To determine whether mortality between 1991 and 1995 in hospitals in England carrying out surgery for congenital heart disease in children was associated with the annual volume of cases and to estimate the extent to which an association could explain the apparent divergent mortality at Bristol Royal Infirmary. Design: Retrospective analysis of data from two sources, a register of returns by surgeons to their professional society and an administrative database. Setting: 12 hospitals in England carrying out surgery for congenital heart disease over the period April 1991 to March 1995. Main outcome measure: 30 day mortality. Results: For open heart operations in children under 1 year old, and in particular for arterial switches and repair of atrioventricular septal defect, there is strong and consistent evidence of an inverse association between mortality and volume of cases (not taking into account any data from Bristol). A hospital carrying out 120 open operations per year in 1991-5 on children aged under 1 year would be expected to have a mortality 25% lower than that in a hospital carrying out 40 operations. If the children in the hospitals had the same mix of operations, this reduction is 34%. Stratifying for types of operation or including the results from Bristol strengthens this association. It was also estimated that less than a fifth of the excess mortality at Bristol Royal Infirmary in open operations in children less than 1 year old was due to the hospital9s lower volume of surgery. Conclusions: Using appropriate methods, this study showed that mortality in paediatric cardiac surgery was inversely related to the volume of surgery. Considerable caution is needed in interpreting these results, and it does not necessarily follow that concentrating resources in fewer centres would reduce mortality. What is already known on this topic Mortality in children undergoing heart operations has been shown to be lower in hospitals with a high volume of such operations Studies showing a relation between volume of cases and mortality have a range of methodological inadequacies, in particular the choice of a threshold defining high and low volume after the analysis to increase the significance of the results What this study adds Disregarding data from Bristol, there is strong and consistent evidence that in England in 1991-5 hospitals performing a higher number of open heart operations in children aged under 1 year tended to have lower mortality This association explains only a small proportion (less than a fifth) of the excess mortality seen at the Bristol Royal Infirmary over this period