Endpoints and the people missing from the chart

A trial result is more than a number attached to an outcome. It is an estimate produced by a plan for what to measure, when to measure it and how to handle events that interrupt the original treatment strategy. If some participants stop treatment or miss an assessment, the question does not disappear. The analysis must specify how those events relate to the effect the trial is trying to estimate.

Two distinctions make this easier to read. First, stopping treatment is not the same as leaving follow-up: a participant may discontinue the study medicine and still contribute later measurements. Second, an unobserved value is not an observed failure, success or absence of change. Statistical methods may estimate or account for it under stated assumptions, but the result should not be described as if every final observation were directly measured.

This worksheet connects the endpoint list to the people behind the chart. It introduces the estimand as the clearly defined treatment effect of interest, then asks how the analysis addresses discontinuation, other relevant events and missing data. You do not need to prefer one estimand in every situation. You need to know which question was chosen, whether the method fits it and whether the headline preserves that choice. The most useful reading note makes these assumptions visible rather than treating them as a technical footnote. That is a useful skill even when you leave the formal statistical details to specialists.