A Modular Approach to PROM Implementation – Reflections from ISPOR Europe 2022

Patient-reported outcome measures (PROMs) that measure health-related quality of life (HRQOL) in clinical trials are often composed of several different domains. As an example, the 36-Item Short-Form Health Survey (SF-36, QualityMetric, Johnston, RI)¹, a very popular instrument for HRQOL evaluation, is composed of eight domains: physical functioning, physical role, bodily pain, general health, vitality, social functioning, emotional role, and mental health. Similarly, the European Organization for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire – Core Questionnaire(QLQ-C30) contains five functional domains, nine symptom scales, and global health status², while the Health Assessment Questionnaire Disability Index (HAQ-DI) contains eight disability domains, as well as pain and a global health assessment³. The current industry thinking and standard practice around PROM completion in clinical trials is to administer these questionnaires in their entirety – for example, a participant would complete all 36 items of the SF-36, not just the 10 items corresponding to the physical functioning domain.

However, depending on the stage of the disease and the therapies under investigation, not all measure domains may be sensitive to detecting change or be meaningful to clinical trial participants. For instance, the Quality of Life Questionnaire Prostate Cancer Module (QLQ-PR25)⁴ is the indication-specific measure developed by the EORTC for prostate cancer and includes domains on sexual activity and sexual function, among others. For a castration-resistant prostate cancer population participating in a clinical trial, these domains may not be relevant or may even be distracting, affecting responses to other domains. In fact, scores on these domains are often non-evaluable or found to be dramatically low due to ongoing castration therapy^{5, 6}. This suggests that administration of the QLQ-PR25 in its entirety with the same HRQOL domains and questions, without considering indication-specific factors, may be problematic.

 This naturally leads to the question – should PROM strategy shift from the administration of questionnaires in their entirety (off-the-shelf) to the selection and administration of only the relevant core domains from existing validated assessments to clinical trial patients? This is an attractive proposition that has the potential to increase the clinical relevance of investigated constructs while decreasing participation burdens in clinical trials by reducing the number of questions to which patients are asked to respond. To implement such an approach, clinical trial sponsors need to be assured that the measurement properties of a validated instrument’s sub-domain are unchanged when the sub-domain items are implemented independently, compared to when they are implemented as part of the full instrument as originally intended.

A panel explored the possibility of following such a modular approach during the ISPOR Europe 2022 conference during the session on ‘The possibility of administering selected domains of existing PRO questionnaires to reduce patient burden and increase clinical relevance.” The session was moderated by Julia Braverman, Director of Worldwide Health Economics and Outcomes Research at Bristol Myers Squibb (BMS), and with presenting speakers Paul Kluetz, Lauren Podger, and Andrew Lloyd.

Considering the recent FDA draft guidance on patient-reported outcomes (PROs) in oncology clinical trials⁷, oncology PROM strategy may be well suited to a modular approach. The draft guidance recommends PROMs are selected to allow focused, independent measurement of five specific core domains: disease-related symptoms, symptomatic adverse events, the overall impact of side effects, physical function, and role function. Moreover, the frequency of administration of each core domain should vary – with physical function and adverse events identified as requiring more frequent assessment, especially during the early stages of treatment. Both the specificity and measurement frequency of these core domains lead us to consider the independent application of sub-scales within existing validated measures. For example, the FDA draft guidance identifies the physical function subscale of the EORTC QLQ-C30 as an example of a suitable measure for physical function. The alternative approach of using a number of complete instruments concurrently leads to increased completion burden, overlapping domains that result in repetition of similar questions, and inclusion of items that are distal to the effect of treatment and should be avoided. This was echoed by Paul Kluetz, Deputy Director of the Oncology Centre of Excellence at the FDA, who argued in favor of a modular PROM strategy and for using a combination of validated off-the-shelf PROMs and item libraries.

Evidence of unaffected measurement properties

As discussed above, one of the main potential concerns with administering a subset of domains from pre-existing, validated PROMs is the psychometric effect this may have. For measures that have been developed and validated to be modular with independency in domains and individual domain scoring, modular sampling should not impact their measurement properties. The EORTC QLQ-C30 is a good example of a modular measure, as it has been developed via factor analysis to create independence between domains⁸. As such, the fatigue domain (QLQ-FA12⁹) can be administered as an independent measure without creating concerns with regards to the psychometric properties of this new measure. In this case, the order of items and scoring method should be consistent with the original measure – but due to the independence of modules, the measurement properties of the sub-scale are conserved when implemented independently. However, for PROMs developed and validated with inter-domain associations, selecting only a subset of items or domains to administer has the theoretical potential to significantly alter the original measurement properties of the instrument, making modular sampling challenging. To help explore this, Lauren Podger, Scientist at OPEN Health, presented a simulation of the effect of selecting only one domain from a measure with inter-domain associations on measurement precision. For a hypothetical measure that included two domains with an inter-domain correlation of r=0.57, and with both domains containing ten items each, she showed that administering only one domain does not appear to affect measurement precision. This suggests that even if there are inter-domain associations, measurement precision may be unaffected by administering only a subset of the domains in a pre-existing questionnaire.

Although results from such simulations are promising, operationalizing a shift from an ‘off-the-shelf’ approach to a modular PROM selection strategy will involve discussion and coordination across multiple stakeholders. Importantly, before selecting specific domains from pre-existing measures to use in a clinical trial, it is essential to ensure copyright holders are in agreement, and to determine the measurement properties of the new instrument resulting from this domain sampling. Additionally, as stressed by Andrew Lloyd, Director at Acaster Lloyd Consulting, it is also important to consider the post-approval landscape: HTA decision makers generally require generic PROM data, typically in the form of the EQ-5D¹¹, and the domain specificity and selectivity of a modular approach may not be aligned. Nonetheless, over the past few years, developers of widely used measures have already driven a shift to a more modular approach to PROM selection with the creation of item libraries, such as those developed by EORTC, Patient Reported Outcomes Measurement Information System (PROMIS), and Quality of Life in Neurological Disorders (Neuro-QoL). In the future, we anticipate a more flexible and modular approach to gain popularity, especially in fields such as oncology where regulators are aligned with this thinking.

Signant’s eCOA experts’ recommendations when considering a modular approach to PROMs:

1. Consider the indication: A modular PROM approach may be more applicable to some indications or therapeutic areas than others. As discussed, clinical trials in oncology can benefit from a modular PROM strategy as this can aid the focused selection of items corresponding to core domains. A modular approach may be less suitable for indications relying heavily on home-grown or ‘de novo’ diaries, such as epilepsy and migraine.
2. Consider burden and item redundancy: Explore proposed measures to consider overlap between items and domains and aim to reduce duplication of the same measurement constructs across measures. Consider the proposed schedule of assessments in terms of completion burden for patients.
3. Consider which domains need measuring and when: Construct the schedule of events based on the optimal measurement of each sub-domain of interest. Based on the expected profile of each, some subdomains may be subject to change over different time periods than others, and this may lead to the optimal approach being to split the sub-domains out for separate measurement.
4. Explore modularizing existing validated measures: Consider utilizing existing subscales that include the items of the domain you are interested in administering. Have early discussions with instrument owners on the suitability of the approach and the scoring of the sub-domain, and pre-existing validation evidence to support the validity and reliability of the approach. 
5. Ensure consistency with the original instrument: Ensure the order of items in the module are consistent with the original instrument, and module scoring mirrors the sub-domain scoring of the original instrument.

References

1. Saris-Baglama, R. N., Dewey, C. J., Chisholm, G. B., Plumb, E., Kosinski, M. A., Bjorner, J. B., & Ware Jr, J. E. (2010). QualityMetric health outcomes™ scoring software 4.0: installation guide. Lincoln (RI): QualityMetric Incorporated.
2. Scott, N. W., Fayers, P. M., Aaronson, N. K., Bottomley, A., de Graeff, A., Groenvold, M., Gundy, C., Koller, M., Petersen, M. A., & Sprangers, M. A. G. (2008). EORTC QLQ-C30 reference values manual. Brussels: EORTC.
3. Bruce, B., & Fries, J. F. (2003). The Stanford health assessment questionnaire: dimensions and practical applications. Health and quality of life outcomes, 1(1), 1-6.
4. van Andel, G., Bottomley, A., Fosså, S. D., Efficace, F., Coens, C., Guerif, S., Kynaston, H., Gontero, P., Thalmann, G., Akdas, A., D’Haese, S., & Aaronson, N. K. (2008). An international field study of the EORTC QLQ-PR25: a questionnaire for assessing the health-related quality of life of patients with prostate cancer. European Journal of Cancer, 44(16):2418-24
5. Westdorp, H., Creemers, J. H., van Oort, I. M., Mehra, N., Hins-de Bree, S. M., Figdor, C. G., Witjes, J. A., Schreibelt, G., de Vries, J. M., Gerritsen, W. R., & Ottevanger, P. B. (2020). High health-related quality of life during dendritic cell vaccination therapy in patients with castration-resistant prostate cancer. Frontiers in oncology, 10, 536700.
6. Nielsen, T. K., Højgaard, M., Andersen, J. T., Jørgensen, N. R., Zerahn, B., Kristensen, B., Henriksen, T., Lykkesfeldt, J., Mikines, K. J., & Poulsen, H. E. (2017). Weekly ascorbic acid infusion in castration-resistant prostate cancer patients: A single-arm phase II trial. Translational andrology and urology, 6(3), 517.
7. US Food and Drug Administration. (2021). Core patient-reported outcomes in cancer clinical trials: draft guidance for industry. FDA-2020-D-2303.
8. Aaronson, N. K., Ahmedzai, S., Bergman, B., Bullinger, M., Cull, A., Duez, N. J., Filiberti, A., Flechtner, H., Fleishman, S. B., … & Takeda, F. (1993). The European Organization for Research and Treatment of Cancer QLQ-C30: a quality-of-life instrument for use in international clinical trials in oncology. JNCI: Journal of the National Cancer Institute, 85(5), 365-376.
9. Weis, J., Tomaszewski, K. A., Hammerlid, E., Ignacio Arraras, J., Conroy, T., Lanceley, A., … & Bottomley, A. (2017). International psychometric validation of an EORTC quality of life module measuring cancer related fatigue (EORTC QLQ-FA12). JNCI: Journal of the National Cancer Institute, 109(5).
10. The EuroQol Group (1990). EuroQol-a new facility for the measurement of health-related quality of life. Health Policy 16(3):199-208

Alexandra Kalpadakis-Smith, PhD

Associate Director, Digital Health Sciences

Bill Byrom, Ph.D.

VP Product Intelligence and Positioning