In reply: We agree with Arnold that, ideally, expert witnesses should attempt to assess management decisions before acquainting themselves with the outcome and allegations in negligence cases. In practice, we suspect this is rarely done. In any case, the mere seeking of an expert opinion conveys the information that there has been an adverse outcome and, as we noted, there is evidence that, even if experts attempt to guard against it, hindsight bias is unavoidable in such circumstances. The central problem is that the expert is, as it were, looking back down one fork in the pathway of events, whereas the treating doctor was looking forwards at many possible and often uncertain forks.1

Hickie's statement "in my experience the unfortunate outcome can usually be predicted within reading the first few paragraphs of the brief" epitomises the very problem we address. Such retrospective snap judgements are characteristic of hindsight bias and are often accompanied by the telltale phrase, known to be a marker for hindsight bias,1 "it should have been obvious".

We are unable to understand Hickie's statement that our "views . . . might have been slightly biased on the basis of . . . cases . . . reviewed as Chairmen of the Australian Cases Committee of the Medical Defence Union". The Committee contained representatives from the major specialties, including two consultant physicians, and the cases ranged over all specialties and subspecialties. Experts from subspecialties, including cardiology, were co-opted when appropriate.

We agree with Hickie that clinical practice guidelines are useful. We did not recommend that they be excluded, but we did draw attention to their difficulties and limitations.

We acknowledge the admirable work done by the American College of Cardiology in developing an impressive range of guidelines, but our view remains unaltered that they are costly in terms of time and effort to produce, cannot cover all clinical contingencies, and have limitations when applied to negligence cases. The guidelines for heart failure referred to by Hickie took more than three years to prepare, involved numerous committee members and no fewer than 26 reviewers, and were not subsequently updated for six years. Relatively few guidelines have been modified for Australian use and some are obviously deficient. For example, the current National Health and Medical Research Council (NHMRC) guidelines relating to the common problem of chest pain2 are six years old, and have been criticised on the grounds that they have not been rigorously tested to ensure clinical usefulness and do not include appropriate management strategies for patients with non-cardiac chest pain.3

We adhere to our view that these problems make it likely that clinical practice guidelines will have a limited role in negligence cases.