The trouble with likelihood ratios.

ity scale for clinicians is symmetrically skewed in both directions. In primary care, elective surgery, and oncology, clinicians work at the extremes of the probability scale. The time has come to apply the Feynman-Tufte principle to clinical logic, to offer a visual representation of Bayesian logic. Earlier attempts with natural logarithmics do not allow a simple graphic. We teach a scale of log10 odds from –4 to 4, indicating the corresponding probabilities, and adding rounded log10 likelihood ratios to the pretest probability. In doing so, clinicians can apply Bayesian logic without formal calculations (figure).