In Reply: Dr Sheehe makes the valuable point that if you are using the chance that a patient will have a good outcome as the way of expressing the results of a clinical trial, you must include death in the composite of those bad outcomes or the results will make no sense. However, our observation that the reader may be subject to a “bait and switch” strategy may still hold if the majority of the events in the composite bad outcome are less severe. For example, if the composite end point includes death, clinical worsening, or tumor enlargement on imaging, showing that there is improvement in the complement of this bad outcome may be misleading if most of the improvement comes from a reduced risk of tumor enlargement.
Tomlinson G, Detsky AS. Composite End Points in Clinical Trials—Reply. JAMA. 2010;303(17):1698. doi:10.1001/jama.2010.523