Author response

We thank Dr Matthew Large for his helpful comments. We wished to respond only by clarifying that the figures in Table 2 were from a hypothetical population, based on a hypothetical risk assessment tool with certain sensitivity and specificity values. The purpose was to illustrate that, even in risk assessments with unrealistic accuracy levels, the positive predictive value (PPV) was still low, as it was greatly influenced by the base rate. Any misleading odds ratios arising from the table was not intentional and arose (perhaps ironically) by chance.