Researchers must consider several factors when using patient-reported outcome (PRO) endpoints to evaluate treatment benefit for drugs and medical devices. Significant attention is often paid to substantiating the evidence for a PRO instrument’s content validity and appropriate psychometric performance (i.e., reliability, construct/criterion validity, and sensitivity to change), however clearly communicating the interpretation of PRO endpoints is equally important.
With the 2009 PRO Guidance for Industry, the FDA has indicated that assessment of treatment benefit using scores derived from PRO instruments must be evaluated not solely based on statistical significance, but with a clear understanding of the amount and type of change that is clinically meaningful to patients. Several quantitative methods have been used to establish score thresholds associated with meaningful change on PROs including anchor-based approaches, distribution-based methods, cumulative distribution functions, and benchmarking. Recently, attention has also been paid to the qualitative research methods that can also be employed to assess how patients view treatment-related change.
An example of how we apply qualitative research methods is a case study in which we used post-trial qualitative interviews with phase 2 study participants to debrief PRO instruments and assess the meaning patients placed on changes at the individual symptom level. Those patient-derived estimates were then used to model meaningful change thresholds for the overall PRO score alongside anchor-based and other quantitative methods.
There are 4 key considerations to be aware of for future work on this topic:
1) Triangulation of qualitative data from patient interviews alongside anchor-based methods and cumulative distribution function (CDF) approaches can be quite valuable.
In our case study, we were able to demonstrate consistency in the meaningful change thresholds estimated through qualitative methods and anchor-based approaches. This strengthened our study sponsor’s proposed analytic approach for their upcoming phase 3 program, as we were able to use valuable first-hand accounts of the actual patient experience to provide a comprehensive context for the threshold estimates.
2) Interviewing clinical trial participants offers potential scientific advantages over other approaches but requires careful planning and execution.
In many cases, qualitative exploration of patient-perceived meaningful change occurs in separate stand-alone non-interventional cognitive interview studies. While these studies are often designed to enroll a sample of patients like those who will ultimately use the PRO instrument(s) in the clinical trial program, the non-interventional nature means that patients are often responding to hypothetical questions about changes in their symptoms, rather than reflecting on their actual experience with symptoms.
In contrast, we sampled patients who had recently participated in the phase 2 clinical trial program, which offered two distinct advantages: first, it capitalized on the patients’ actual lived experience with treatment-related symptom change and could gather information and specific examples from the patient’s experiences to provide context for their interpretation of the PRO items; second, the participants’ familiarity with the PRO instrument from the trial allowed them to have greater insight into the underlying response scale and how they applied it in real-life settings. Taken together, we believe these features led to richer qualitative data and more precise estimates from patients about the definition of meaningful change.
There are, however, definite logistical challenges to consider when attempting to gather qualitative interview data from clinical trial participants. Exit interview studies are often added after the trial itself is already underway, which creates a need for additional site- and IRB-related study activities. Embedding qualitative interviews into the overall clinical trial design itself offers some clear logistical efficiencies but requires a significant amount of advanced planning. It also requires close consideration of the timing of the interview data collection against the other data collection interactions in the study, so as not to create biases or otherwise compromise the overall study measurement. For these reasons, it is important to work with a skilled and experienced Clinical Outcomes Assessment partner to provide guidance on the optimal integration of qualitative data collection in trial settings.
3) Consider bidirectional change in the outcome of interest when evaluating meaningful change using qualitative methods.
To be truly useful as endpoints, PRO instruments need to reflect both improvement and worsening of patients’ condition. Exploring these aspects with patients directly via qualitative interviews allows for a more nuanced understanding of the relationship between PRO scores and meaningful change and allows researchers to account for cases in which patients express a lower threshold for symptom worsening to be considered meaningful than the researchers might have expected.
4) Start working to understand meaningful change as early as possible in the clinical development pathway.
Starting early is important to determine and establish thresholds for clinically meaningful change in PRO scores and opening a dialogue with the FDA to ensure that appropriate evidence is gathered to substantiate the PRO analytic plan. Beginning this process early, in the pre-trial design phase, provides greater opportunity to leverage mixed qualitative and quantitative designs and navigate the complicated nature of integrating qualitative data collection with clinical trial processes.
The use of qualitative interview methods to better understand the optimal interpretation of PRO-based endpoints is an emerging approach within the HEOR field. While the data gathered using these methods can provide important context on patient-experienced change to aid the development of PRO-related analytic plans, the complex nature of the approaches requires careful planning and execution to be effective.
Contact OPEN Health for more information on Patient Reported Outcomes.