Evidence based diagnosis: does the language reflect the theory?

We introduced an error into figure 1 of this analysis and comment article by Matt T Bianchi and Brian M Alexander (BMJ 2006;333:442-5, 26 Aug). When we redrew the figure that the authors had supplied, we omitted a zero from the three likelihood ratios at the lower end of the scale; the correct values should be 0.001, 0.002, and 0.005.