Figure 3.

Antibiotic treatment for clinical pneumonia. Hypothetical results are shown for A = antibiotic for gram-positive, B = antibiotic for gram-negative, C = antibiotic for gram-positive that is more effective than A. In the sequential study, A bests B in the first trial (when 80% of the subjects are gram positive), and B bests C in the second trial (when 10% of the subjects are gram positive). However, as shown with the three-arm trial, it would be incorrect to make the transitive inference that when 10% of the subjects are gram-positive, A is better than C.