The study team defined six elements that must be specified to describe a trial outcome completely. A complete outcome definition comprises: outcome domain, specific measurement, specific metric, cutoff (when applicable), variable type, and time point. The authors reported that when all six elements are provided prospectively, readers can judge whether reported outcomes match planned outcomes and can better compare results across trials and reports.
Short, clear definitions matter. An isolated label such as “pain” is insufficient. The framework requires naming the construct, stating which instrument or laboratory measure will be used, describing how values are assigned to individuals, specifying any threshold used to group participants, declaring whether each individual’s value is continuous or categorical, and listing the assessment time(s).
The authors built on a 2011 framework and expanded it through a structured, iterative process. They reviewed existing literature and produced preliminary guidance. Pilot testing followed. The pilot sample consisted of a random selection of 225 randomized controlled trials from ClinicalTrials.gov; across those trials, the team examined 3,424 distinct outcomes.
Two raters used the guidance to attempt to define each outcome. When outcomes could not be classified, the authors modified the guidance and repeated testing. The process continued until the guidance required no further changes. Multiple drafts and group discussions among authors produced the finalized document. The sources indicate the guidance was then used in subsequent studies, where agreement between independent human raters—and between humans and an automated system—was reported as good.
The guidance sets out naming conventions for both subjective and objective measures. For subjective outcomes (for example, patient-reported symptoms), the authors described how to name scales and instruments precisely, including which version or subscale should be used. For objective outcomes (for example, laboratory values or blood pressure), the guidance requires specifying the exact test, measurement site, or device when relevant.
The manuscript expanded categories for the specific metric element beyond prior work. These categories help clarify whether the reported value will be a change from baseline, a single observed value, a maximum or minimum, a time-to-event, or another metric. The guidance therefore aims to reduce ambiguity that might otherwise allow multiple post hoc analytic choices.
The authors replaced the older “method of aggregation” item with two elements that describe individual-level values: variable type and cutoff. This change aligns the guidance with the estimand framework by focusing on the values observed in each participant rather than on how those values are summarized across a group.
The guidance defines variable type simply: whether the value for each individual is continuous or categorical. The cutoff element is expanded into three sub-elements in the paper (the source documents these sub-elements); these sub-elements describe the threshold, how it is applied, and any rationale or provenance for the threshold. The authors argue that separating these concepts improves clarity about what exactly the trial intends to measure at the participant level and prevents conflation between individual-level classification and population summaries.
Authors recommended that trialists and sponsors apply this six-element framework when registering trial outcomes on registries, when drafting protocols, and when reporting results in journals. The framework is intended to make it easier for external reviewers to determine whether reported outcomes were prespecified and to help systematic reviewers decide whether trials measured comparable outcomes.
The authors cautioned that specifying all six elements does not eliminate all forms of biased reporting. Complete definitions reduce ambiguity but do not prevent other selective analytic choices or reporting behaviors. The paper also notes that better tools and systems are needed to help researchers prepare registrations and protocols; complete outcome definitions are a practical step, not a standalone solution.
Limitations explicitly reported by the authors include the iterative reliance on pilot testing and rater judgement during development, and the fact that implementation in diverse trial settings will require operational work. The manuscript documents that subsequent applications of the guidance showed good rater agreement, but it does not claim that the guidance alone will prevent selective reporting in every context.
The article ends on those practical limits and on the next task the authors identified: broader implementation and integration with registration systems and protocol templates to improve consistency in outcome definitions across trials.