Pragmatic Clinical Trials: The Ethics of Conducting Research in the Real World

Motto:

“Research is highly structured, like classical music, whereas the nature of clinical practice is to be flexible like improvisational jazz.“ (Paul J Martin)¹

Introduction.

The double blind randomized controlled trial (RCT) is the gold standard trial design to demonstrate a medical intervention’s efficacy and typically, two such RCTs are required by the US Food and Drug Administration to authorize investigational drugs in the United States. Such efficacy trials are usually conducted in near-ideal circumstances, following detailed protocols, with randomization and blinding to minimize bias, observing strict inclusion and exclusion criteria that yield relatively homogeneous subject groups, and undergoing stringent monitoring that requires complex and costly study infrastructure. Although RCTs remain the gold standard for assessing the effectiveness and safety of interventions, they trade off high internal validity for generalizability to the real world. Therefore, their utility to everyday clinical practice, beyond gaining regulatory approval, has been questioned.

Pragmatic Clinical Trials

Recognizing the limitations of explanatory RCTs, Schwartz and Lellouch^{2, 3} proposed the concept of the pragmatic clinical trial (PCT) in the 1960s. They made the distinction between explanatory trials (which aim to further knowledge as to how and why) and pragmatic/practical trials (which aim to inform healthcare decisions within routine practice). Most trials have elements of pragmatism to them, and the degree to which pragmatism defines the trial is a result of deliberate study design considerations. Pragmatic trials often are conducted in clinical settings that are, or closely mimic, real-world settings (eg, a community clinic as opposed to a clinical-trials unit at an academic medical center), where the subjects can be identified, and the outcomes ascertained. They typically have broad eligibility criteria, resulting in heterogeneous cohorts that reflect actual patient populations, typically compare the performance of two or more interventions that might be available, and evaluate patient-centered outcomes. By minimizing the number of exclusion criteria, selection bias is reduced as most patients seen at the clinical site with the condition of interest could be participants of the study. Pragmatic trials usually require larger sample sizes than RCTs due to heterogeneous cohorts and smaller expected effect sizes, but may not require random allocation of treatments at the individual level; instead, clinical sites might be the unit of randomization (cluster randomization) to allow for comparisons of both patient-level outcomes while providing an unbiased assessment of the impact of various treatments (or treatment algorithms) on the clinical site. Pragmatic trials frequently make broad use of electronic medical records, potentially using automated data extraction, and measure a variety of outcomes important to patients, health care administrators, and researchers alike. These features result in pragmatic trials offering lower internal validity but higher generalizability than RCTs.⁴

In practice, purely pragmatic (or purely explanatory) trials are the exception, and it is more useful to consider all RCTs on a spectrum of pragmatism. A multi-axis instrument like the PRECIS tool⁵ (Figure 1) can be used to qualitatively rank studies over several study design characteristics. Maximizing the study design characteristics results in a highly pragmatic clinical trial (Figure 1, red contour). Explanatory trials, on the contrary, would rank lower across pragmatic design criteria (Figure 1, blue contour). When these domains are shown on a radar chart, the tool allows one to quickly visualize and compare different studies on the degree to which the pragmatic design elements have been prioritized in the study design. It is critical to note, there is no “right or wrong” design decisions here; all trials will have some elements of pragmatism and very few trials can maximize all domains as shown in Figure 1.

Figure 1.

Radar chart representation of a highly pragmatic trial (red contour) as opposed to a mostly explanatory trial (blue contour). The “E” at the center of the hub represents the explanatory end of the pragmatic-explanatory continuum, while points on the circumference of the circle are at the pragmatic end. Adapted from BMJ⁵ with permission.

Pragmatic clinical trials are increasingly popular in medical research (Figure 2) and this is explained by factors beyond the limitations of explanatory clinical trials. There is a growing demand from a variety of end-users such as learning health care systems, or even for regulatory approval which can use real world evidence in some cases.⁶ Reflecting this need and feeding the increase in the number of trials is the increased availability of funding. For example, the Patient Centered Outcomes Research Institute (PCORI) has built a nationwide infrastructure to conduct PCTs and as of December 2021, was supporting 46 large pragmatic clinical studies⁷, while the National Institues of Health have established the Pragmatic Trials Collaboratory, a comprehensive and regularly updated repository of knowledge on the best practices to conduct PCTs and report their results.⁸

Figure 2.

Number of articles indexed in Ovid Medline with the phrases “pragmatic” and “trial” in their titles, as a fraction of the number of all articles with “trial” in their titles.

Informed Consent

The informed consent process is the central mechanism for obtaining permission from patients or participants, which prioritizes respect and autonomy for the research participants; ensures alignment of patients’ values while protecting their welfare interests; provides transparency; ensures the integrity of the medical process; and satisfies regulatory requirements.⁹

It is recognized that the consent process has important practical limitations, in part due to framing and cognitive biases among the subjects.⁹ Despite these limitations, there is broad ethical and public support for obtaining patient consent for participation in clinical research. In one survey, members of the public favored prior written consent over notification after enrollment¹⁰, yet they also preferred a brief verbal consent process over written approaches to obtaining consent.¹¹ In another study, an overwhelming majority (73–80%) of patients with myocardial infarction preferred to be asked for permission prior to enrollment in a trial¹² although it can be difficult to gain full information and make an autonomous decision about joining a trial while in the throes of a life-threatening illness.⁹ Most reviews of public attitudes toward informed consent reinforce the perspective that the consent process should rarely be waived even if allowances are made for practical alterations to enhance its efficacy.¹³

Because pragmatic trials often compare alternatives which are otherwise commonly used in routine care, they can be hard to classify into the “Minimal Risk” vs “Greater than Minimal Risk” dichotomy mandated by current Federal regulations. In fact, pragmatic trials often have features that could be called, only half-facetiously, “minimally greater than minimal risk.” This evokes the insight from pragmatist philosophers that humans often use dichotomous classification to understands natural phenomena that are continuous; in other words, “we think by digital concepts about an analogue reality.”¹⁴

Truog and colleagues¹⁵ have highlighted the paradox that physicians are given a wide latitude to prescribe innovative treatments as long as their stated goal is to benefit the given patient, yet once the goal is changed to generating systematic knowledge from the intervention, substantial hurdles must be overcome as regulated research. They argued that specific written informed consent should not be mandatory for all randomized trials provided that (1) all treatments can be offered outside of the trial; (2 there is minimal additional risk or certainly no risk greater than that encountered in daily life; (3) clinical equipoise exists; (4) a “reasonable person” would not have a preference between the alternatives; (5) subjects are informed that they are participating in a clinical research study, and (6) are given the option to seek more information or care elsewhere.¹⁵ Since the publication of Troug and colleagues’ initial paper, interpretations of the Common Rule have clarified that the distinction of minimal vs. above minimal risk lies in the research-related risks, not the underlying medical condition or medical care risks that the participants would be expected to be exposed during usual care.⁸

Some authors have gone further, suggesting that blending clinical research into clinical care raises questions of mandatory participation in clinical research as a precondition to accessing health care. Faden and Kass challenge the distinction between the ethics of medical care and that of research and use the concept of “learning activities” to cover both, and propose that patients have an affirmative obligation, rooted in the common purpose of improving care and reciprocity, to participate in learning activities.¹⁶ Although through a different reasoning, Steel¹⁷ also advocates for compulsory participation in research in learning health systems, but he also advised against a minimal risk limit.

In contrast, Kim and Miller propose an “integrated consent model” in which no consent would be needed under certain circumstances¹⁸ such as for many pragmatic trials. Largent and Joffe¹⁹ advocate separating trials into “prescribed” and “invited” clinical trials. Prescribed clinical trials compare interventions viewed as optimal medical care and involve minor net risk.²⁰ Physicians prescribe enrollment and patients are enrolled through a less formal consent process^{19, 21} similar to that proposed by Kim and Miller.¹⁸ Invited medical trials on the other hand have greater uncertainty about the benefits of the alternatives, involve more than minor net risk,²⁰ and would require more conventional forms of patient consent and institutional ethics oversight.¹⁹ Other mechanisms include passive enrollment in low-risk pragmatic trials, combined with notification of subjects after the enrolment and with mechanisms in place to allow withdrawal for those patients who prefer that course of action.

Research introduces competing interests (financial rewards, career advancement, reputation) that separate it from routine clinical care, where the provider pursues the best interest of the patient. The ethical framework within which PCT are to be conducted continues to evolve through its lively literature^22–25. As the conversation continues about the PCTs place on the spectrum between the standard informed consent and the streamlined (or waived) consent with mandatory participation in clinical research, institutional review boards remain the independent mechanisms to prevent the erosion of the subject protection safeguards. It appears reasonable to consider Largent and Joffe’s prescribed clinical trial model combined with Kim and Miller’s integrated consent paradigm as a practical way to integrate pragmatic trials into the everyday activities of academic medical centers that embrace the ideology of a learning health care system. This approach supports the adoption of practical strategies that reduce the burdens on pragmatic trialists.

Conclusion

Pragmatic clinical trials are becoming increasingly popular as they deliver real world evidence sought by researchers and regulators alike, and are uniquely suited to power learning health care systems as they seek to improve the quality of routine care and encourage more physicians to be involved in research. Pragmatic trials embedded in routine care require novel paradigms of informed consent and the solicitation of patient support. Respectfully acknowledging the history of abuse under the guise of research, that prompted the current legislative framework of human subject protection in the first place, makes us recognize that embedding research in clinical care requires specific safeguards to meet the obligation of respect for persons. Societal trust in the research enterprise can only be maintained as long as participants’ desire for transparency and retention of decisional autonomy, two primary aims of current informed-consent approaches are meaningfully observed in the context of PCTs.

Abbreviations:

RCT

randomized controlled trial

PCT

pragmatic clinical trial