A Core Outcome Set for Multimorbidity Research (COSmm)

PURPOSE We aimed to develop a consensus-based set of core outcomes specifically for studies in multimorbidity.

METHODS We undertook a consensus study following the COS-STAR (Core Outcome Set-STAndards for Reporting) guidelines for the design and reporting of core outcome sets. A Delphi panel of experts completed a web-based survey with 2 rounds. Panelists were presented with a range of outcomes that had been identified in previous workshops and a related systematic review. They indicated their level of agreement on whether each outcome should be included in the core set using a 5-point Likert scale, and outcomes reaching a prespecified consensus
level were included.

RESULTS Of 30 individuals invited to be panelists, 26 from 13 countries agreed. All 26 completed both rounds of the survey. The Delphi panel reached consensus on 17 outcomes for inclusion in a core outcome set for multimorbidity (COSmm). The highest-ranked outcomes were health-related quality of life, mental health outcomes, and mortality. Other outcomes were grouped into overarching themes of patient-reported impacts and behaviors (treatment burden, self-rated health, self-management behavior, self-efficacy, adherence); physical activity and function (activities of daily living, physical function, physical activity); consultation related (communication, shared decision making, prioritization); and health systems (health care use, costs, quality of health care).

CONCLUSIONS This consensus study involved a wide range of international experts who identified a large number of outcomes for multimorbidity intervention studies. Our results suggest that quality of life, mental health outcomes, and mortality should be regarded as essential core outcomes. Researchers should, however, also consider the full range of outcomes when designing studies to capture important domains in multimorbidity depending on individual study aims and interventions.


Susan M Smith (PI), Emma Wallace. HRB Centre for Primary Care Research, Department of General Practice, RCSI, Dublin
Liz Bayliss
Chris Salibury
Mamime Sasseville
Martin Fortin


Journal: Annals of Family Medicine
Issue: 16
Pages: 132 - 138
Year: 2018
DOI: 10.1370/afm.2178

Further Study Information

Current Stage: Not Applicable
Date: February 2016 - November 2016
Funding source(s): Administrative and running costs will be sourced through the Irish Health Research Board Centre for Primary Care Research

Health Area

Disease Category: Other

Disease Name: Multimorbidity

Target Population

Age Range: 18 - 100

Sex: Either

Nature of Intervention: Any

Stakeholders Involved

- Clinical experts
- Consumers (patients)
- Guideline developers
- Policy makers
- Researchers

Study Type

- COS for clinical trials or clinical research


- Delphi process
- Semi structured discussion

We will use a Delphi survey approach to develop the core outcome set.
The initial outcome set will be developed by the author group. This will be based on work undertaken over the last number of years that has included several meetings and workshops between the author group and other experts in multimorbidity research. A provisional list of core outcomes and related outcome measures has been identified and published as part of a general guideline to conducting multimorbidity research:
Table 2 of Framework paper

This list was further refined at a workshop at NAPCRG 2015: “Think-tank on outcomes for patient-centered intervention for persons with multimorbidity”

The authors will identify a panel of international experts with broad stakeholder representation outlined below.
We anticipate having at least 20 individuals with representation from each of these groupings. Ethical approval will be sought through RCSI Dublin and expert panel members will be approached inviting them to participate in the study and to return consent sheets.

The consensus process will involve two rounds of web-based surveys. The survey containing the draft core outcomes and related outcome measures (referred to together as the core outcome set) will be piloted (to test usability) by members of staff in the departments of the author panel and modified accordingly. This pilot study will incorporate use of the proposed cut-points in the likert scale describ For the first round survey, panel members will receive a link (via e-mail) to a questionnaire using an online survey software tool (SurveyGizmo®). The aims of the study will be explained to the panel members in the email. The panel will be presented with the core outcome set and references to each of the proposed outcome measures where appropriate and will also be given information on terms of use and related costs. Panellists will be asked to indicate their level of agreement for inclusion of each outcome using a 5-point Likert scale (where 1 is strongly disagree and 5 is strongly agree) and will be asked to provide comments as necessary. Using this scale, the median response and the interquartile range (IQR) will be calculated. An upper quartile ranking of =2, indicates general disagreement and the outcome will be rejected. A lower quartile of =4, will indicate general agreement with the outcome and it will be adopted. Outcomes with IQR that includes 3 will be modified based on panel feedback and will be included in the Round 2 survey. The pilot study will also test the feasibility of these proposed cut-points in the likert scale. Panellists will also be invited to contribute new outcomes during the first round process though it will be stressed that only validated outcome measures will be included in the core outcome set. We will also limit the inclusion of new outcomes for round 2 of the process, to outcomes that have been suggested by at least two of the panel members to avoid an excessively long list of outcomes.

Outcomes for which there is a lack of agreement will be reviewed by the author group, who will consider feedback from the panel and any additional outcomes suggested. A second round survey will take place including remaining outcomes and using the same approach. Depending on the levels of agreement regarding outcomes, it may become necessary to employ a ranking sytem for outcomes within each category if there are too many outcomes to be included in the final core outcome set.

Linked Studies

    No related studies

Related Links

    No related links