A COSMIN systematic review of generic patient-reported outcome measures in Switzerland

Patient-reported outcome measures (PROMs), defined as “a report coming directly from patients about how they feel or function without interpretation by healthcare professionals or anyone else” continue to figure prominently in quality-based paradigm shifts of national health systems [1]. In Switzerland, PROMs’ growth is further exhibited by several governmental and professional recommendations toward improving healthcare quality [2‐5]. In contrast to Switzerland’s rapidly evolving usage of disease-specific PROMs, however, there is no coherent guidance for usage of generic PROMs [2, 6, 7]. Extrapolating, the literature’s systematic reviews of generic PROMs have largely focused on specific diseases/conditions or patient segments [8], whereas only one study that we are aware of specifically focus geographically for country-specific recommendations [1]. Steinbeck et al. [9] highlighted that, until 2021, projects were initiated from either voluntary bottom-up provider or regional level, mainly focusing on orthopedics and cancer care because of their feasibility. Efforts have been paid to study beyond individual level such as a pilot project from Canton of Wallis, where they collect PRO-data across settings from hospitals, care homes and disease areas. An outstanding example is from the University of Basel, where they have successfully implemented 15 outcome standard sets in areas like orthopedics, cancer care, and chronic diseases. However, sofar, there is no uniform PRO-collection framework and no standard PROM sets that prohibits sustainable implementation with high quality of data. In extreme cases, lack of country-specific generic PROMs has resulted in Switzerland utilizing neighboring or linguistically similar countries’ reference values for comparative effectiveness purposes [10]. Similar to other developed nations’ health systems, given Switzerland’s aging population (20% > age 65) and concomitant multimorbidity-related costs (33%/chronic condition), formulating actionable recommendations for generic PROMs holds great import [11, 12]. Furthermore, the intersection of Switzerland’s advanced, private Bismarckian health-insurance model and its heterogeneous linguistic composition presents a challenge that may be concomitant only to its value-potential for establishing generic-PROM recommendations [13]. Finally, conducting a precedential systematic review within a multi-linguistic country may serve substantive interest to health researchers from other multi-linguistic countries (e.g., Belgium, Canada, India) aiming to render their own PROM recommendations. Fortunately, the COnsensus-based Standards for the selection of health Measurement INstruments (COSMIN) methodology facilitates the process of generating recommendations for PROM usage [14]. Generally, the COSMIN system enables researchers to: (1) Assess methodological quality of PROM development, content validity, and psychometric studies, (2) Evaluate PROMs’s measurement properties (content and psychometric), (3) Grade the overall quality of evidence, and (4) Synthesize all aspects to formulate recommendations. For the current systematic review, therefore, the aim is to identify all previously used generic PROMs that measure health-related quality of life for adults living in Switzerland and evaluate their measurement properties using the COSMIN methodology, summarize evidence, compare and formulate recommendations (recommendable, needing further validation, or unrecommendable) for future usage by clinicians and researchers.

This COSMIN systematic review provides precedential recommendations for generic-PROMs usage in Switzerland. The generic PROM in this systematic reviews concerned only HR-QOL tool, that measure at least three aspects: physical, mental and social health.

In our study, we found that different health sectors have been utilizing and benefiting from generic PROMs primarily from Clinical Medicine extending to public health, social medicine and alternative medicine. Our research has also shown that Switzerland has a tendency of increasingly utilizing generic PROM. Many generic PROMs have been implemented but through years, there is a convergence towards SF-36, EQ-5D, WHOQOL-BREF. SF-36, EQ-5D-3L and WHOQOL-BREF were first developed and published between 1990 and 2000 where PROMIS initiative was first developed in 2004 with PROMIS-29 first published in 2 010 [26, 70‐72]. The observed time shift in our study could be a result of the know-how transfer, implementation process, translation and cross-culture and other psychometrics validation.

Although diverse stakeholder interest for a generic PROM in Switzerland has been explored, empirical evidence-based recommendations were absent [23]. Although SF-36, EQ-5D have been long introduced and validated in Switzerland, some back before 2010 [39, 47, 49, 73], no up-to-date systematic review has been conducted according to the present-day standard to standardize generic PROM usage in Switzerland, to suggest recommendations for the national generic PROM program. The current study identified five generic instruments previously used in Switzerland. Our findings are elaborated below according to the following COSMIN evaluative criteria for focusing future research efforts: (1) Content validity, (2) Internal structure, and (3) Additional measurement properties [20].

First, regarding development and content validity, it is notable that the vast majority of generic PROMs exhibited low to moderate quality of content validity that rely on studies from other countries [21, 22, 28]. An exception is the WHOQOL-BREF, which exhibited moderate content validity [26], due to the global cross-cultural adaption effort with large focus on qualitative method in 1998. Encouragingly, qualitative research has begun systematically examining patient-reported experiences, and these data may be utilized to yield insights into further generic PROM development and content validation efforts [74]. EQ-5D has showed in Rohr et al. [28] to have insufficiency in content validity in a group of German geriatric patients with physical limitation. In Switzerland, where population is aging rapidly with more than 60% are German speakers, for population wise implementation, it is recommended that EQ-5D should be adapted and tested again before use.

Second, regarding internal structure, SF-36 and WORQ exhibited the highest evidence, whereas WHOQOL-BREF exhibited the lowest evidence. Auspicious for future research in Switzerland is the measurement property of cross-cultural validity. Of all our studies, only two examined measurement invariance and the evidence for structural validity for all PROMs except for WORQ was borrowed from studies outside of Switzerland. This will be critical to future linguistic-/cultural-validation efforts, and should be guided by modern psychometrics, such as item response theory. We are currently undertaking such research in a Swiss-nationally representative cross-sectional survey. Interestingly, although the use of official-language translations were prominent, there was a disproportionate absence of English-based generic PROMs. Switzerland has about 27% non-Swiss residents and English is the most widely spoken non-official language in Switzerland, with 45% of the population using it regularly [75]. This absence of English-based generic PROMs could inhibit the collection and interpretation of QOL of a large subgroup of Swiss population and could not reflect their perspective. Hence, this could be critical for researching migrant, urban, or high-system usage populations [11].

Third, regarding additional measurement properties, evidence varied considerably. For example, across all PROMS, construct validity as reported most often, whereas reliability and especially measurement error were reported least often as shown in Table 7. Furthermore, SF-36 exhibited highest quality evidence, whereas WHOQOL-BREF exhibited lowest quality of evidence. Stephan et al. [42‐44] proved PROMIS short forms for pain and function to have sufficiency with high quality. This provides partial confidence for further validation of the full version of PROMIS-29. In constructive critique, we observed partial conflation with properties of reliability, measurement error, and responsiveness, and hope that these properties may be better delineated in impending updates to the COSMIN criteria.

Following the COSMIN guidelines v1.0 together with up-to-date evidence collected from Roser et al. [47, 49], the SF-36 is recommendable for further use for population scale measurement for HR-QoL.

Given the high-quality evidence of insufficient measurement error and moderate-quality evidence of inadequate comprehensiveness, the EQ-5D is recommended to be revised and adapted and currently is not recommended for continued use for population scale measurement.

The WHOQOL-BREF requires further validation of its internal consistency before it can be recommended. While the PROMIS-29 shows promising results for a subset of items, supported by its modern IRT-based model, additional evidence is needed to validate the full 29-item version. WORQ is reviewed here in this study because it was developed in Switzerland, shows to have potential and requires further validation. However, it is worth to point out again that it is a work-related QoL.

As with all validation evidence, this review should be updated to reflect the growing empirical information made available.