“…2), one can imagine that knowledge of the type of test may influence subsequent management decisions of a physician, despite a prespecified link between test results and treatment. For example, an obstetrician may be reassured more with the results of a magnetic-resonance pelvimetry in breech presentation at term compared with manual pelvimetry, which could influence subsequent decisions, such as to perform an emergency caesarean section [21]. One could choose a design that bases randomization on test results, to overcome this problem.…”