By Isabel M. Estrada-Portales, Ph.D., M.S.
Our trusted clinical treatment research methods have served us well, but many of them, at least in their current form, are in need of significant upgrade and transformation. These changes are needed not just because of technological advances; the previously inconceivable richness and accessibility of data these new technologies produce also necessitates a shift in clinical research methods.
In “A new era of clinical research methods in a data-rich environment,” a chapter of the recently published book Oncology Informatics, William Riley, Ph.D., Director of the Office of Behavioral and Social Sciences Research (OBSSR), argues that clinical treatment research in the biomedical and behavioral sciences evolved from a data-poor environment in which each research project starts de novo from a tabula rasa. Given the conditions created by this new data-rich environment, that blank slate would, instead, be replaced by a continuous flow of data on the variables of interest. As Riley explains,
“This would allow naturalistic studies of the covariation of these variables over time and the ability to model how a given outcome might be produced from changes in the mechanisms targeted by a potential treatment. Participants would be identified and invited, not recruited. The treatment would be delivered or “inserted” into this relentless flow of data from the individual or system and the impact of the treatment observed via deviations from the outcomes predicted by the wealth of predictor data available. By leveraging temporally dense data and within subject designs, rapid determinations of the effects of the treatment can be made and the intervention improved and retested iteratively. Any additional data collection and treatment manipulations from a trial would be incorporated into the data stream for others to utilize.”
If you are immersed in traditional clinical research methodologies, this data-rich research environment may seem like a pie-in-the-sky concept, but, Riley posits, advances in medical informatics, big data analytics, and mobile and wearable technologies have laid the groundwork for it.
As a whole, “Oncology Informatics,” edited by Brad Hesse, David Ahern, and Ellen Beckjord:
- Presents a pragmatic perspective for practitioners and allied health care professionals for how to implement health information technology solutions in ways that will minimize disruption while optimizing practice goals;
- Proposes evidence-based guidelines for designers on how to create system interfaces that are easy to use, efficacious, and timesaving;
- Offers insight for researchers into the ways in which informatics tools in oncology can be utilized to shorten the distance between discovery and practice.
Riley’s chapter, in turn, problematizes and promises. It problematizes our current research methods and our attachments to them, which include a tendency to confer those methods a much higher trustworthiness than may be warranted. The chapter then suggests new, more accurate approaches to improve upon those old methodological friends with new technologies. Needless to say, most if not all of these new technologies and tools that may assist in clinical research would not exist, nor would they be as successful, were it not for the existence of those old methodologies, to which the future of research owes more than a debt of gratitude.
Riley discusses at length the randomized controlled trial (RCT). Although considered the gold standard of biomedical and behavioral research, RCTs are costly and time-consuming to conduct. RTCs are also very slow, as it takes 17 years for 14 percent of the evidence from a clinical trial to be implemented in practice (Balas and Boren, 2000).
“Similar to the astronomical concept of light-years,” writes Riley, “when an RCT is published, we are looking back in time and considering the evaluation of technologies and approaches that existed nearly a decade ago.”
Are we dissing RCTs altogether? Not at all. Not yet anyway. According to Riley, until we have developed a fully-functional data-rich environment for biomedical and behavioral research, RCTs will remain the primary methodology for assessing the effects of treatments on a variety of clinical outcomes. But he cautions, “we should less easily accept the RCT as the default design for all clinical research questions, consider its weaknesses, and be more open to alternative designs that in some cases are better suited to the question of interest.”
Riley’s chapter offers a tour de force for how other sciences, such as meteorology, seismology, and cosmology, have dealt with and hugely benefited from a transition to data-rich environment methodologies. From those experiences, Riley envisions a data-rich biomedical and behavioral research enterprise anchored in a comprehensive health research data infrastructure.
Among other methods that would benefit from this qualitative improvement, Riley discusses single case studies, which are difficult to perform in a data-poor environment because they require intensive longitudinal data—data that is now made possible by the availability of remote measurement and sensor technologies.
With the benefit of great computational simulations, single case studies become a much more reliable and cost effective method to test and adopt new treatments; while our most common reluctance to them—their lack of generalizability—becomes less of a concern. And, of course, Riley concludes, “single case studies are consistent with a precision medicine approach and can be used to optimize treatments for specific subgroups of patients.”
Riley warns of the danger of underutilizing this new wealth of resources and data if we, for instance, merely use them to speed up the current timeline for RCTs, thereby limiting their potential by our limiting framework.
As OBSSR is crafting its strategic plan, with input from the research community at large, the import and urgency of improved, timely, proven, and accessible research methodologies is shaping up to be a major scientific priority for the upcoming years. This chapter is timely in that it envisions and frames roads yet to be traveled (but hardly unthinkable anymore) by biomedical and behavioral researchers.
Read the book
Balas EA, Boren SA. Managing clinical knowledge for health care improvement: yearbook of medical informatics. Stuttgart: Schattauer; 2000.