Forgetting the learning curve for a moment: how much performance is <TOGGLE>un</TOGGLE>related to own experience?

Volume-outcome relationships are of clear importance for most participants in the health-care industry; research and appropriate policy implications are of critical importance. In this letter we critique the prevailing 'learning-by-doing' view in cardiac surgery. We illustrate the very wide disparity in empirical findings on volume-outcome relationships there, in the context of broader open issues in 'learning curves' in general. Potential complementary mechanisms, e.g. 'social learning by knowledge spillovers' are introduced; these cast into doubt the prevailing policy recommendations of simple regionalization and volume smoothing. Copyright © 2008 John Wiley & Sons, Ltd.