Digital Health and Clinical Trials
To date, significant uncertainty surrounds the adoption of digital health tools into clinical development programs and trial protocols.
The increasing ease with which digital health, mHealth, and information technology (IT) can be developed and the wide availability of smartphones, tablets, and wearable devices are having a growing impact on the pharmaceutical industry. Nontraditional design-oriented groups are helping to shift the focus to the user experience. Within companies, multiple functions that include clinical research and development, medical affairs, diagnostics and biomarkers, IT, outcomes research, health economics, and patient engagement are developing strategies for the evaluation and use of digital health tools.
However, much uncertainty surrounds the adoption of these tools into clinical development programs and trial protocols. To date, despite the profusion of potential outcome measures, there are few well-characterized trial end points. In the conservative realm of the industry-sponsored clinical trial, the uptake of novel technology is slow and there is often an expectation that it is the newcomer who must adapt. This raises the question, What are the potential opportunities and challenges for companies in the emerging era of digital health?
Increased Patient Engagement
With long and complex trial designs, issues with low patient engagement that lead to missed assessments and visits, as well as withdrawal from trials, are a major concern for regulators and the industry. The concern is so great that both the FDA and the European Medicines Agency stress the need to prevent missing data in guidance/guidelines, with the FDA also initiating a National Research Council Panel published as The Prevention and Treatment of Missing Data in Clinical Trials in 2010. Remote and passive data collection offers new avenues to facilitate trial participation and reduce burden for participants.
Perhaps as exciting is the influence that a user-focused design of digital health tools could have in boosting levels of engagement. The possibility to shift from dry, onerous data reporting to tasks that are interesting and potentially even pleasurable to complete—or that provide valuable feedback—opens a new horizon.
One group working in this space is uMotif (umotif.com), which takes such a design-driven approach to the collection of patient data on individuals’ own mobile phones. This effort to create simple and compelling digital health tools has been applied to drive adoption and retention of participants in studies such as 100 for Parkinson’s (100forparkinsons.com/the-project),
to learn about health over time, and Cloudy with a Chance of Pain, (cloudywithachanceofpain.com), to investigate a potential association between weather and chronic pain.
An issue that still needs to be addressed is the potential impact of greater engagement with and awareness of the data collected. It remains to be seen, for example, whether traditional patient-reported outcomes might be adapted into more engaging formats without changing the way that patients complete the data or how the use of actigraphy to measure parameters such as step count or sleep might change behavior. Many are familiar with the initial surge in activity that happens after purchasing a new Fitbit, iWatch, Jawbone, or other fitness tracker. Such challenges are surmountable, and changes in format and potential sources of bias have long been addressed in clinical trial methodology. Yet the conservative pharmaceutical industry will expect such questions to be thoroughly addressed before the wide adoption of these techniques.
Increased Patient Centricity
Many aspects of patient centricity may be enhanced by making the participation in clinical trials simpler, more engaging, and less burdensome for patients. It is critical that this is not conflated with patient relevance. Patient relevance (ie, the importance that patients attach to a given outcome and ensure that trials measure what matters) cannot be guaranteed by engagement and participation. Such patient relevance has yet to be directly established by first asking patients what matters to them, defining these concepts, and generating measures that address their concerns. The lure of digital health technologies and the possibility to generate more sensitive, more engaging, and more ecologically valid (in their “dense” measurement of patient experience vs infrequent in-clinic visits) outcomes do not automatically mean we measure what is patient relevant. Furthermore, the way in which digital health technology may capture certain concepts such as activity and sleep parameters needs careful scrutiny. Additionally, care should be taken to avoid overinterpreting these data. For example, the development of tools that directly measure patient-relevant sleep quality or step-count patient-relevant social behavior may take us beyond what these types of data are truly capable of showing.
Digital Biomarkers, Surrogates, and eCOA: Statistics Versus Concepts
In addition to fluid imaging and other biological parameters, behavioral measures provide another rich source of potential diagnostic, prognostic, theragnostic, and surrogate clinical trial outcomes. Digital health vastly increases the potential for the identification of novel biomarkers through the huge number of additional parameters that might be generated. Data-driven approaches can then be used to select optimal characteristics from multiple parameters. This may then identify outcomes with increased diagnostic accuracy, sensitivity to detect treatment effects, etc. The potential for the development of digital biomarkers and surrogates also raises the possibility for novel measures of treatment benefit or electronic clinical outcome assessments (eCOAs). However, caution is needed with eCOAs regarding the use of data-driven approaches in which the measurement concept may be a secondary consideration to the use of clinical and/or patient insights where the measurement concept is the primary consideration.
Statistical parameters—such as the accuracy or sensitivity of a tool or its association with important concepts—do not confer any direct meaning of the outcome measure itself for patients or clinicians. In this sense, digital health/eCOA is only a data collection platform. New end points intended to measure treatment benefit should have a clear conceptual basis and start with the end in mind (ie, What do we intend to measure, why is it important, and to whom?). Several precompetitive initiatives are now seeking to provide guidance that will assist in this process. The Clinical Trials Transformation Initiative published recommendations and released a set of tools from its Mobile Clinical Trial initiative for developing its webinar Novel Endpoints Generated by Mobile Technology for Use in Clinical Trials.
These recommendations stress the need to include the patient’s voice along with the voices of clinical experts at the outset. This is in line with the FDA’s Roadmap to Patient-Focused Outcome Measurement in Clinical Trials, which starts the end point development process with an understanding of the disease and a conceptualization of treatment benefit. Furthermore, the FDA cautions that the device selection should come after the selection of measurement concept.
Digital health is positioned to transform the conduct of clinical trials and the experience of trial participants. The removal of barriers to trial participation, increased patient centricity, and inclusiveness represent much-needed improvements that have the potential to substantially increase our understanding of disease and how potential therapies work. Nevertheless, a technology-driven approach to the development and selection of measures risks a loss of focus on measuring what matters to patients and clinicians. Starting with the end in mind and the measurement concept instead of the measurement device is critical to ensuring patient centricity in the fullest sense.