Core Outcome Sets: Check Scope Before Adopting the Minimum List

Picture a protocol team locking the endpoint section of a phase III trial. A search of the COMET database returns two published core outcome sets that name the trial's condition: one older and broadly scoped, the other newer but developed for a different intervention type. Adopting the newer set because it is recent could import outcomes chosen for, say, surgical trials into a trial of a systemic drug. Defaulting to the older set could mean inheriting a list agreed before current treatments existed, or one agreed without patients in the room.

Neither choice should be made by citation alone. This guide treats adoption as a documented scope-and-evidence decision with five steps: search the COMET database; match each candidate's scope to the trial; appraise how the set was developed (COS-STAD), how completely it was reported (COS-STAR) and whether it is still current; adopt the outcome list while selecting instruments separately; and record the fit and rationale in the protocol and trial registry. The worksheet and worked example later in the article are synthetic decision aids assembled from published methods guidance, not validated tools.

The COMET (Core Outcome Measures in Effectiveness Trials) Initiative describes a core outcome set as an agreed, standardized collection of outcomes that should be measured and reported, as a minimum, in all trials for a specific clinical area. The COMET Handbook, version 1.0 (Williamson et al., Trials 2017), grounds the idea in two familiar problems: trials in the same condition measure different outcomes, and outcome-reporting bias makes the results harder to compare and combine.

Two properties matter when the set meets an endpoint section. The first is that a COS is a floor, not a ceiling. The Handbook says that using a COS does not restrict trialists to the COS alone; it provides the minimum to include, and they remain free to add outcomes that are particularly pertinent to their study. Core outcomes are expected to be collected and reported alongside whatever the trial's own question requires, whether that is a primary efficacy endpoint, a safety assessment or an exploratory measure.

Inside the protocol, core outcomes still have to be placed in the endpoint hierarchy. Depending on the trial's objective, a core outcome may be the primary endpoint, a secondary endpoint under multiplicity control or a supportive measure. Whatever the role, the endpoint definition must still name the concept, the instrument, the timepoints and the estimand strategy for intercurrent events, the executable wording covered in our guide to ICH M11 protocol endpoint definitions.

Avoid overstating the mandate. WHO's 2024 Guidance for best practices for clinical trials (section 2.1.7) says that use of standardized core outcome sets should be considered for all trials, so that results can be compared, contrasted and combined. It adds that when a trial is intended to achieve marketing authorization or change policy, it is often helpful to discuss the choice of outcomes with regulators or policymakers. That is a recommendation, not a requirement. Documented regulator endorsement of a specific set is rare in this literature; the best-known case is rheumatoid arthritis, whose core set was endorsed by the FDA in 1996 and the EMA in 1998 (Hughes et al., J Clin Epidemiol 2021). Elsewhere, outcomes named in FDA and EMA guidance only partly overlap with published sets, as the final section shows.

The COMET database is a free, public, searchable repository of COS studies, kept current through regular systematic review updates, citation alerts and notifications from developers. In a 2022 primer, COMET authors reported nearly 500 published COS studies and nearly as many ongoing ones (Kirkham and Williamson, BMJ Med 2022). COMET's annual systematic review updates had counted 364 published COS by the end of 2018, and a further 267 were registered as in development by March 2020 (as summarized by Hughes et al.). The search page itself warns that the records returned by any search may increase daily, so date the search in the decision record and repeat it before the protocol is finalized.

Run the search while the endpoint strategy is still open, not on the day of protocol freeze, and use the structured fields rather than a single keyword. The advanced search offers:

• Health area: a disease category (for example, Cancer) and a disease name. Search the category as well as the specific condition, so that sets developed for a broader or narrower parent condition are not missed.

• Target population: minimum and maximum age, and sex. Age limits are a common source of mismatch; a set agreed for adults does not automatically fit a pediatric or adolescent trial.

• Nature or type of intervention: the intervention scope recorded for the set. Treat a blank or unknown value as a reason to read the full paper, not as evidence that the set covers every intervention.

• Method(s) and stakeholders involved: the development methods (for example, Delphi process, focus groups, systematic review) and the participant groups (for example, patients, clinical experts, pharmaceutical industry representatives). These fields give an early read on the COS-STAD appraisal described below.

• Study type, publication year and published or unpublished status: restrict to sets intended for clinical trials or clinical research, and include unpublished records to see registered work in progress.

An unpublished, registered study is a signal rather than an outcome list. If one is under way for your condition, record its registration and expected timing in the decision record. Adoption itself should rest on a published set and its documented methods. If a relevant set is published during trial conduct, the record should say how the team will respond, for example by reporting any additional core outcomes the trial already collects.

A matching disease name is not a matching scope. COMET's search guidance says that, in deciding the relevance of a COS to your study, you should consider its scope: the health condition, the target population, and whether it is for any intervention or specific types of intervention. The COMET Handbook notes that these are the first three elements of a PICO question, and that a partial overlap between a COS and a trial may still help a trialist decide which outcomes to measure.

Scope is also often under-reported. Of the 227 COS published to the end of 2014, 53% did not specify whether they were intended for all interventions or a particular intervention type; 7% were for any intervention and 40% for a specific type (COMET Handbook). An unstated intervention scope is a question to resolve from the full paper, not permission to assume broad applicability.

Compare each candidate with the trial on four dimensions. The first three come from COMET's scope definition; the fourth, setting, is one of COS-STAD's scope standards.

1. Health condition and stage: Check whether the set covers all stages of the condition or a subset. Williamson et al. (Trials 2012) give the example of a colorectal cancer COS that may be developed for all patients or only for those with metastatic disease. A set built for one stage may leave out outcomes that matter at another.

2. Target population: Compare age limits and any subgroup restrictions with your eligibility criteria. In pediatric or cognitively impaired populations, a set may assume a reporter, such as the patient rather than an observer, that your trial cannot use.

3. Intervention type: A set may be developed for all treatments or for one intervention; Williamson et al. contrast all interventions for morbid obesity with bariatric surgery alone. Outcomes chosen for a procedure can include procedure-specific complications that have no meaning in a drug trial.

4. Setting: COS-STAD asks developers to state the setting the set is meant for, such as research, routine care or both. A set designed for routine care or clinical audit may carry different feasibility assumptions from one designed for effectiveness trials.

Once scope fits, appraise development. COS-STAD (Kirkham et al., PLoS Med 2017) sets out 11 minimum standards for developing a COS, agreed through a two-round Delphi survey of COS developers, COS users, medical journal editors and patient representatives. Its authors state that the standards are meant to help users assess whether a COS should be adopted, and that anyone appraising a published set still needs their own judgment about whether its scope applies to their purpose.

Patient involvement is not a formality. Williamson et al. (2012) recount that the original OMERACT rheumatoid arthritis core set was established without direct patient input; when patients first attended an OMERACT meeting in 2002, fatigue and sleep were identified as missing from it. Kirkham and Williamson (2022) report that 85% of COS studies now involve public representatives, compared with 17% in 2014, so older sets are the ones most likely to fall short here. A set agreed without patients is not automatically unusable, but the gap belongs in the decision record and may justify adding patient-important outcomes, informed by the trial's own patient input before endpoint selection.

Expect uneven quality. When Gargon et al. (J Clin Epidemiol 2019) assessed 49 cancer COS against COS-STAD, using 12 criteria to represent the 11 standards, none met all 12 (range 4 to 11, median 6). Every set met the four scope standards, but only 8 (16%) met all three stakeholder standards and 2 (4%) met all four consensus-process standards. The authors noted that poor reporting often made it hard to tell whether a standard had been met, especially the a priori requirements. An adopting team will therefore often record “not reported” rather than “not met”, a distinction worth keeping in the worksheet.

COS-STAR (Kirkham et al., PLoS Med 2016) is an 18-item checklist describing what a COS study report should contain, covering the background, scope, methods, results, conclusions and limitations. It applies to studies deciding which outcomes to include, not to later work on how those outcomes are defined or measured. Its stated aim is reporting that is transparent and complete enough for COS users to judge whether a recommended set is relevant to their work. For an adopting team, the most useful items are the ones that let the scope and consensus checks above be completed at all.

Currency is the last appraisal check. Treatment options, diagnostic criteria and assessment technology all change after a set is agreed. A COS agreed before a new class of therapy entered practice may not include outcomes that matter for that class, and a set framed around site-administered paper assessments may not anticipate remote or electronic collection. The COMET Handbook notes that periodic review would let developers confirm a set's continuing validity, check that included outcomes are still relevant and important, and consider whether excluded outcomes should be added. It also acknowledges uncertainty about how often that review should happen and what should trigger it.

There is therefore no evidence-based age at which a COS expires. Instead of applying a fixed cutoff, record the publication year, any update or linked follow-on study visible in the COMET record or the literature, and the answers to three questions:

• Has the treatment landscape changed? Does the trial test an intervention class whose benefits or harms the original panel could not have considered?

• Have diagnostic or staging definitions changed? If classification criteria were revised, does the set's population still correspond to your eligibility criteria?

• Has the assessment context changed? Were the outcomes framed around assessment modes the trial will not use, such as clinic-only administration in a trial with remote or electronic collection?

A yes to any of these is a reason to supplement or justify, not automatically to reject. The core outcomes remain the agreed minimum unless the record explains why one of them does not apply to this trial.

After a set has passed the scope, development and currency checks, keep two decisions apart: the outcomes the set requires (what to measure) and the instruments that will measure them (how). Merging the two can lead a team to reject an otherwise suitable set because a questionnaire named alongside it does not suit the trial, or to adopt an instrument without examining it. Lack of validated measures is one of the barriers to COS uptake reported in the uptake literature (Hughes et al.).

For the how, the COSMIN/COMET consensus guideline (Prinsen et al., Trials 2016) sets out four steps:

1. Conceptual considerations: define the construct (the outcome or domain) to be measured and the target population.

2. Finding existing instruments: aim to identify all existing outcome measurement instruments, using systematic reviews, literature searches and, optionally, other sources.

3. Quality assessment: evaluate the instruments' measurement properties, with evidence in the target population, and their feasibility.

4. Selection: the guideline recommends selecting only one instrument for each outcome, with minimum requirements that include high-quality evidence of good content validity and good internal consistency (where applicable), and feasibility.

The guideline is written for COS developers, so a trial team applies it by extension. Two passages speak directly to trialists. Where running every step is not feasible for a single trial, trialists can use the instruments already included in a COS. And when a trial's primary outcome is not a core outcome, the COS still needs to be measured; the same recommendations can be used to choose the instrument for the primary outcome.

Instrument choice then meets trial-specific questions: which COA type, whether PRO, ClinRO, ObsRO or PerfO, fits each concept; what evidence shows the instrument is fit for purpose in your context of use; and whether the full questionnaire load is sustainable for participants, as discussed in our guide to deciding which PRO questionnaires a trial needs. Two further checks sit outside the COS decision and remain mandatory: confirming that each instrument is licensed, with validated translations, for every country and mode of administration in the trial; and a statistical review of how the added core outcomes enter the analysis plan, including their place in any multiplicity strategy.

The worksheet below turns the preceding checks into one record per candidate set. It is a synthetic decision aid assembled from COMET, COS-STAD, COS-STAR and COSMIN/COMET guidance. It has not been validated and does not score sets. Where possible, its sources point to fields the COMET database already exposes.

Applied to the opening scenario, a completed comparison might look like this:

The same sequence, from search to registry entry, is summarized below.

flowchart TD
    A["Define trial condition, population, intervention and setting"] --> B["Search the COMET database and date the search"]
    B --> C{"Does the set's scope fit the trial?"}
    C -- "No" --> D["Record the reason; reuse only in-scope condition-level outcomes as trial-specific additions"]
    C -- "Fully or partly" --> E["Appraise development against COS-STAD"]
    E --> F["Check reporting completeness with COS-STAR"]
    F --> G["Check currency: date, updates, changes since publication"]
    G --> H{"Adopt?"}
    H -- "In full or in part" --> I["Adopt the outcome list as the minimum: what to measure"]
    H -- "No" --> D
    I --> J["Select instruments separately: how to measure"]
    J --> K["Write endpoint definitions and the COS rationale into the protocol"]
    D --> K
    K --> L["Cite the COS and list core plus additional outcomes in the registry entry"]

The completed record belongs in three places. In the protocol, the endpoint section, or a short rationale subsection beside it, should name the sets considered and their citations, summarize the scope comparison, state which core outcomes are included and in what role, and give reasons for any additions or omissions. In the trial registry, the COMET Handbook advises researchers to say whether they will use a relevant COS and, if so, to cite it and specify each core outcome and any additional outcomes; our guide to aligning ClinicalTrials.gov outcome measures with the protocol covers that wording. In the trial's essential records, keep the dated search output and the completed worksheet so the decision can be reconstructed later.

Some funders already ask for this. The NIHR Health Technology Assessment programme's guidance for full proposals, as quoted in the 2017 COMET Handbook, says that where established core outcomes exist they should be included among the trial's outcomes unless there is good reason to do otherwise, and directs applicants to the COMET website. Kirkham and Williamson (2022) likewise encourage researchers to search for a relevant set and to explain why one was not used, or was only partly used, when one was available. The resulting outcome list then feeds the protocol's critical-to-quality review under ICH E8(R1).

Uptake evidence explains why a documented rationale matters more than a citation. Hughes, Clarke and Williamson (J Clin Epidemiol 2021) found that only 17 of 337 COS (5%) had been assessed for uptake, covering 5 of the 31 disease categories in which COS have been developed, and no uptake study at all for cancer, the category with the most COS. Among randomized trials, uptake of the full set ranged from 0% (gout) to 82% (rheumatoid arthritis). In ankylosing spondylitis, one study found full-set uptake rose from 0% of trials before the set was published to 20% after.

Rheumatoid arthritis is the outlier, with full-set uptake of 60% to 82% across studies. Hughes et al. cite an earlier assessment by Kirkham and colleagues suggesting that endorsement by the FDA in 1996 and the EMA in 1998 may have contributed in part. The same review describes a contrasting case: uptake of the knee, hip and hand osteoarthritis COS was lower (45%), declined over time, and was accompanied by some inconsistency in recommendations across regulators. Endorsement is a plausible contributor in one condition, not a demonstrated rule. Barriers reported in the uptake literature include lack of awareness, lack of validated measures, lack of patient involvement and limited stakeholder involvement in development.

A survey of European Organisation for Research and Treatment of Cancer (EORTC) members, opened in November 2023 and reported by Lawlor et al. (Trials 2025), shows how far awareness lags. Of 3,733 members invited, 710 (19%) answered the COS questions; 93% were based in Europe and 93% were physicians. Half were unfamiliar with COS, 23% knew of COS but reported none relevant to their specialty, 18% knew of a relevant set but did not use it, and 10% used one. Among the 126 respondents who knew of a relevant set but did not use it, the reasons reported included:

• Volume of outcomes (37%): the number of outcomes a set asks the trial to collect.

• Not understanding how to use COS (18%).

• Lack of time (16%), raised in open-ended responses.

• Lack of infrastructure or capacity (11%), also from open-ended responses.

The same paper counted 85 cancer-related COS in the COMET database, 40 (47%) of them specific to particular treatments, which is one more reason the intervention-scope check matters in oncology. The survey is a perception snapshot from one research network with a low response rate, not a prevalence estimate.

Regulatory alignment is also partial. Saldanha et al. (BMJ Med 2022) examined 98 COS and found relevant FDA or EMA guidance for 38 of them. Across 44 COS and EMA document pairs, a median of 70% of COS outcomes had a specific or general match in the guidance (46% specific matches); across 30 COS and FDA pairs, the median was 52% (26% specific). Where outcomes matched only generally, the regulatory outcome was usually the broader one; overlap tended to fall as the number of COS outcomes rose; and none of the core outcome sets referred to their corresponding regulatory documents. The comparison predates WHO's 2024 guidance.

Low and uneven uptake cuts both ways. A reviewer cannot assume that a set was skipped for a good reason, and a team cannot assume that adoption is expected. What a funder, ethics committee, journal or regulator can evaluate is the record: which sets were found, how their scope compared with the trial, what the appraisal showed, and why the final outcome list contains what it does. The record does not guarantee acceptance, but it makes the decision reviewable.

• COMET Initiative. Core Outcome Measures in Effectiveness Trials: home page. https://www.comet-initiative.org/

• COMET Initiative. Search the COMET database (advanced search). https://www.comet-initiative.org/Studies

• Williamson PR, Altman DG, Bagley H, et al. The COMET Handbook: version 1.0. Trials 2017;18(Suppl 3):280. https://link.springer.com/article/10.1186/s13063-017-1978-4

• Kirkham JJ, Davis K, Altman DG, et al. Core Outcome Set-STAndards for Development: the COS-STAD recommendations. PLoS Med 2017;14(11):e1002447. https://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.1002447

• Kirkham JJ, Gorst S, Altman DG, et al. Core Outcome Set-STAndards for Reporting: the COS-STAR statement. PLoS Med 2016;13(10):e1002148. https://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.1002148

• Prinsen CAC, Vohra S, Rose MR, et al. How to select outcome measurement instruments for outcomes included in a ‘Core Outcome Set’ — a practical guideline. Trials 2016;17:449. https://link.springer.com/article/10.1186/s13063-016-1555-2

• Hughes KL, Clarke M, Williamson PR. A systematic review finds Core Outcome Set uptake varies widely across different areas of health. J Clin Epidemiol 2021;129:114–123. https://pmc.ncbi.nlm.nih.gov/articles/PMC7815247/

• Lawlor A, MacLennan S, et al. Core outcome sets in cancer clinical trials: current status and future opportunities — an EORTC perspective. Trials 2025;26:129. https://pmc.ncbi.nlm.nih.gov/articles/PMC11983956/

• Gargon E, Gorst SL, Harman NL, et al. Improvement was needed in the standards of development for cancer core outcome sets. J Clin Epidemiol 2019;106:1–12. https://pubmed.ncbi.nlm.nih.gov/31009657/

• Saldanha IJ, Hughes KL, et al. Comparison of published core outcome sets with outcomes recommended in regulatory guidance from the US Food and Drug Administration and European Medicines Agency: cross sectional analysis. BMJ Med 2022;1(1):e000233. https://pmc.ncbi.nlm.nih.gov/articles/PMC9978677/

• World Health Organization. Guidance for best practices for clinical trials. 25 September 2024. https://www.who.int/publications/i/item/9789240097711

• Kirkham JJ, Williamson PR. Core outcome sets in medical research. BMJ Med 2022;1(1):e000284. https://pmc.ncbi.nlm.nih.gov/articles/PMC9951367/

• Williamson PR, Altman DG, Blazeby JM, et al. Developing core outcome sets for clinical trials: issues to consider. Trials 2012;13:132. https://pmc.ncbi.nlm.nih.gov/articles/PMC3472231/