Initiating Evaluation of Composite Outcome Measures for Psoriatic Arthritis: 2022 Updates From the GRAPPA-OMERACT Working Group

Why do we need composite outcome measures for PsA?

Composite outcome measures allow the combination of outcomes measuring several domains of similar significance to clinicians and patients to generate a single score to give an estimated net clinical benefit of an intervention. Typically, the US Food and Drug Administration (FDA) defines “composite event endpoints” as the occurrence of any of the events from a prespecified list.³ On the contrary, composite outcome measures have been commonly used for measuring the concept of disease activity in rheumatology, and are recognized by the European Medicines Agency guideline.⁴ The potential benefits of using composite outcome measures include the potential to reduce the sample size and the duration of follow-up in clinical trials, thus avoiding statistical adjustment for multiple testing. Composite outcome measures also reduce the risk of underestimating disease through the measurement of multiple domains, as they incorporate patient and clinician perspectives and enhance face validity of the outcome measure.⁵

Recently, OMERACT has set forth a 4-step framework for the evaluation of composite outcome measures,⁵ including choosing the domains to be combined, selecting high-quality instruments for the domains, weighing the domains in the composite, and finally putting the composite outcome measures through the OMERACT filter 2.2 to comprehensively appraise an outcome measure’s validity of truth, discrimination, and feasibility.² Composite outcome measures were further subclassified by the OMERACT filter 2.2 into composite outcome domain and multioutcome domain measures, which can be conceptualized as categorical and continuous composite outcome measures, respectively.

Several existing composite outcome measures have been used in PsA clinical trials and longitudinal studies, yet consensus on which measure to use in different settings has not been reached.^6,7 Although there are emerging data supporting their psychometric properties,^8-11 none of the composite outcome measures have undergone comprehensive evaluation using the OMERACT filter. As OMERACT initiates new methodology guidance on evaluation of composites,^2,5 the use of composite outcome measures in PsA is being revisited.

The Composite Outcome Measures working group

A working group of 16 persons, including 11 rheumatologists, 1 dermatologist, 3 patient research partners (PRPs), and 1 methodologist was set up. The goal of the project is to develop recommendations on composite outcome measures for PsA to be used in clinical trials and longitudinal studies. The working group opted to evaluate existing composite outcome measures rather than developing a new instrument. The group may consider the latter if none fulfill the measurement requirements. To succeed, each of the candidate composite outcome measures should be evaluated in a specified population, for use in a well-defined context with an intended purpose.² There could be different composite outcome measures appropriate for different settings.

The candidate composite outcome measures

The working group selected 10 candidate composite outcome measures and carefully defined the population and context of use (Table 2; Supplementary Material, available with the online version of this article). Notably, none of the existing composite outcome measures encompass all components of the core domain set (Table 3). Some examples of composite outcome measures stratified according to domains, scoring, and weighting were illustrated during the GRAPPA annual meeting. The working group acknowledged the Psoriatic Arthritis Impact of Disease (PsAID)¹² as a composite outcome that measures the impact of PsA on multiple aspects of patients’ lives. As the 12-item PsAID (PsAID12) has been endorsed by both GRAPPA and OMERACT as a measure of the health-related quality of life domain,¹³ the working group decided not to include the PsAID in the present project.

The working group then conducted a preliminary Delphi exercise¹⁴ in June 2022. For each composite outcome measure, participants rated (1) the agreement on the defined purpose of further evaluation and (2) the priority to be evaluated using the OMERACT filter on a scale of 1 to 9, with 1 to 3 as not important, 4 to 6 as important but not critical, and 7 to 9 as critically important. A similar but more succinct Delphi exercise among broader GRAPPA stakeholders was conducted subsequently. Overall, 149 members responded (77.4% rheumatologists, 15.1% dermatologists, 2.7% PRPs, and 4.8% others). Following the Delphi exercise within the working group, ACR response criteria,¹⁵ minimal disease activity (MDA),¹⁶ and Disease Activity in PsA (DAPSA)¹⁷ received a consensus rating as critically important to move forward; Psoriatic Arthritis Disease Activity Score (PASDAS),¹⁸ Composite Psoriatic Disease Activity Index (CPDAI),¹⁹ and 3 visual analog scale (VAS) or 4VAS²⁰ were important but not critical; and Disease Activity Score in 28 joints (DAS28),²¹ Psoriatic Arthritis Responder Criteria (PsARC),²² and Routine Assessment of Patient Index Data 3 (RAPID 3)²³ were rated low priority/not important to proceed with further evaluation. In contrast, in the Delphi exercise for GRAPPA stakeholders, only MDA received consensus rating as critically important (Table 2).

Patient perspective

It is important for patients to have a composite outcome measure that provides a reliable indicator of how they are doing. However, no existing composite outcome measure accounts for all domains in the core domain set that both patients and clinicians recognized as essential to include in all PsA clinical trials.¹ There are some additional points that would be important from the patient perspective. First, the composite outcome measures should be comprehensive, measuring as many domains as possible that are important to patients. Second, the measures should be disease specific. There are numerous composite outcome measures developed for other conditions that are still used in clinical trials for PsA and may not represent a match to the domains relevant to patients with PsA. Although a change toward using PsA-specific composite outcome measures may not be immediate, the conversation toward such a change should be continued. Third, composite outcome measures developed with patient participation should be encouraged. Some of the important domains to include were fatigue and skin disease activity.

In the question-and-answer session during the annual GRAPPA meeting in July 2022, PRPs once again echoed the importance of the comprehensiveness of composite outcome measures. It has been recognized that, at some point in time, patients may experience flares in some domains while experiencing improvement in other domains. Therefore, it may be useful to evaluate the changes in different domains in response to treatment to help select the best domains to be combined in the composite outcome measures. This is especially important for composite outcome measures used as responder criteria in trials.