Background: The federal government wants outcomes of hospital care to be made publicly available. League tables based on single clinical indicators are misleading, largely because of their inability to take case-complexity into account.
Aim: To demonstrate the application of a graphical tool (the risk-adjusted funnel plot) to the comparison of clinical outcomes across hospitals; and its advantages over league tables.
Methods: We looked at publicly available data on intact lower genital tract (ILGT), for all hospitals in New South Wales at which more than 200 births occurred in 2005. The 'excess' percentage of women at each hospital with an ILGT following a vaginal birth, was calculated after adjustment for instrumental assistance, the use of epidural analgesia/anaesthesia, the use of induction/augmentation, and the number of births per annum.
Results: In 2005, ILGT ranged from 13.1 to 55.8%. A plot of ILGT against vaginal births per annum (a funnel plot) revealed huge heterogeneity among hospitals, and an inverse association with the number of births per annum. A residual funnel plot, constructed from the differences between observed and expected ILGT identified four hospitals (three public and one private) with consistently better ILGT than expected - and four public hospitals with ILGT consistently worse than expected. Some of these hospitals were not located at the extremes of the league table.
Conclusion: The risk-adjusted funnel plot is a useful graphical tool which may overcome the shortcomings of league tables. We need to become more sophisticated in our use of clinical indicators for comparing hospital performances.