Fig. 3.

Effect of confirming and excluding power of clinical arguments on the post-test probability of TB, shown on a base-10 logarithmic scale and the corresponding probability scale. Ln: lymphadenopathy; pos: positive; neg: negative. The figure shows two Y-axis: one axis shows the log10 odds scale and the second axis shows the corresponding probabilities. Very strong, strong, good, or weak confirming power allows to advance, respectively, 2, 1.5, 1, 0.5 or 0 steps upward on the log₁₀ odds scale, and thus results in a higher post-test probability. Similarly, the excluding power of an argument allows to regress downward and thus results in a lower post-test probability. Patient A presented without lymphadenopathy, had a positive CXR, and a positive Xpert. The pre-test probability (prevalence) was 17%. When converted to an odds, this equals 0.2. The log₁₀ of that odds is −0.7. The absence of lymphadenopathy has an excluding power of 1, thus has no effect on the probability. Signs of TB on a CXR has weak confirming power: move 0.5 step upward: −0.7 + 0.5 = −0.2). A positive Xpert has very strong power: −0.2 + 2 = 1.8. After converting 1.8 log₁₀ odds to odds, then to a probability, this patient has about 98% post-test probability of TB. The use of the log₁₀ odds scale has been published elsewhere [13]. Patient B presented with lymphadenopathy, had a negative CXR, and a negative Xpert. In this patient the post-test probability of TB is about 2%