Incentives and Payments in Pragmatic Clinical Trials: Scientific, Ethical, and Policy Considerations

Abstract

Pragmatic clinical trials (PCTs) are increasingly used to generate knowledge about real-world clinical interventions. However, they involve some distinctive ethical and regulatory challenges. In this paper we examine a set of issues related to incentives and other payments to patients in PCTs. Although many of the ethical concerns related to incentives and payments in explanatory trials pertain to PCTs, the pragmatic features may introduce additional challenges. These include those related to the risk of incentives and payments undermining the scientific validity and social value of PCTs, the sources of data used in PCTs, and when the PCTs are conducted under waivers of consent. Based on our examination of these matters, we offer some preliminary recommendations regarding incentives and payments in PCTs, recognizing that additional data and experiences are needed to refine them.

Introduction

Pragmatic clinical trials (PCTs) are becoming increasingly prevalent, promising to generate data to inform a variety of stakeholders and at less cost than explanatory trials.^1,2 A pragmatic clinical trial attempts to study “real world” care, typically of interventions that are commonly used in clinical practice, as opposed to the idealized setting common to traditional explanatory trials.³ At the same time, PCTs pose distinctive scientific, logistical, ethical, and regulatory issues.^4,5 Despite a growing literature about PCTs, the use of incentives and other payments to patients and others included in PCTs has received limited attention. In this paper, we examine some scientific, ethical, and policy issues related to them in PCTs.

Incentives and other payments are commonly used in explanatory trials to improve adherence and to recognize participants’ contributions,^6-10 yet it does not necessarily follow that they are appropriate and feasible in PCTs. First, in contrast to explanatory clinical trials, PCTs seek to evaluate the “real world” effects of an intervention in the context of actual clinical practices. However, incentives can perturb standard clinical practices, thereby undermining the pragmatic aspects of a trial. Yet forgoing incentives risks decreased adherence to data collection (in cases where data beyond that provided by usual clinical care are needed) and might not recognize the contributions of those included in a PCT. In addition, those included in PCTs are sometimes unaware of their inclusion, because they are involved simply by being a patient or clinician at a particular site (i.e., everyone meeting specified criteria at a site is included). These features of PCTs suggest that existing conceptual frameworks and practices regarding incentives and other payments may not always be appropriate for PCTs.

It is difficult to describe accurately the range of practices regarding payments in PCTs because there is a paucity of systematic information about PCT practices. However, 9 of the 21 ongoing and completed demonstration projects within the National Institutes of Health’s Health Care Systems Research Collaboratory² incorporate some kind of incentive or payment [Table 1]. This suggests that incentives and other payments is a relevant concern for at least some PCTs.

Table 1:

NIH Collaboratory Demonstration Projects with incentives and other payments

PCT	Payment	Description
BackInAction (NCT pending)	Patients	Paid at timepoints; $15 at baseline, scaling up to $30 at 12 months, plus $5 for each of the 9 monthly assessments
BeatPain Utah (NCT pending)	Patients	For completing follow-up surveys
FM TIPS (NCT04683042)	Patients & institutions	Patients receive free TENS unit and $200; clinics receive $200 per subject enrolled and bonuses for completing training
GGC4H (NCT04040153)	Patients	For completing a study survey (increased to $20 from $10)
OPTIMUM (NCT04129450)	Patients	For responding to PRO assessments; loaning smartphones/tablets
PRIM-ER (NCT03424109)	Clinicians	Gift cards and CME credit for clinicians who complete study training
STOP-CRC (NCT01742065)	Institutions	Approx. $20,000 per year
TiME (NCT02019225)	Patients & institutions	Piloted an incentive program for adherence targets
TSOS (NCT02655354)	Patients	For participation in baseline and follow-up interviews, ranging from $20 to $50

Although those included in PCTs can include patients, clinicians, and/or institutions, and thereby be the potential recipients of incentives and payments, in this paper we focus on included patients, leaving important issues related to clinicians and institutions for another analysis. Because patients or their data may be unknowingly included in a PCT, we tend to use the term “subject” rather than “participant”¹¹. We begin by articulating the standard conceptual framework for ethical incentives and payments in conventional clinical trials, and then examine each rationale for payment separately in the PCT context. We conclude by offering some policy implications.

Comparing payments in explanatory trials and PCTs

The three standard types of incentives or other payments to research participants are reimbursement, wages, and incentives.^7,12 PCTs do not add new ethical justifications for incentives and other payments. If a payment or incentive is ethically appropriate in a PCT, it will be for reasons that would make it appropriate in an explanatory trial. However, PCTs may introduce exceptions to explanatory trial paradigms as described above. (See Table 2.) Although studies can be more or less pragmatic based on their differences from traditional explanatory trials,³ PCTs are designed primarily to inform decision makers “regarding the comparative balance of benefits, burdens, and risks of a biomedical or behavioral health intevention”.⁵ We keep this in mind as we examine the three main types of incentives and payments.

Table 2:

Considerations and Questions for Different Types of Payments in PCTs

Rationale	Considerations for PCTs	Unanswered Questions
Reimbursement	Payments for out-of-pocket expenses may undermine the informativeness of the study for real-world practice, since these kinds of payments typically do not occur in standard care.	Should payments be part of the intervention? Should researchers reimburse for what looks to the patient like standard care?
Wages
for time and burden	Some PCTs intentionally require additional time and burden. Extra patient surveys or lengthier treatments may justify compensating for the additional time that participation requires.	Can the study be designed to accommodate different refusal and non-adherence rates due to different times or burdens for the various interventions?
for use of private data	Some PCTs use only patient data and do not interact with the patients directly at all. Even so, it may be appropriate to compensate patients for access to private information about them.	Will offering payment to a patient for access to their data be feasible? Will offering payment lead patients to distrust the clinicians or researchers? How much should researchers pay for use of the data? What kinds of use should be paid for?
Incentives	Incentives for participation or adherence will usually jeopardize the real-world informativeness of the study, since these kinds of payments are not common in standard care. Because there is little or no risk when the PCT operates under a waiver of informed consent, the threat of undue influence is largely absent.	Could there be alternative benefits, such as lower copays for broad consent to research?

Reimbursement

In traditional explanatory trials it is ordinarily permissible, and even ethically appropriate, to reimburse research participants for their incremental out-of-pocket costs incurred as a result of participating. Indeed, there is a strong presumption in favor of doing so^13-16 based on the argument that patients should not have to bear extra financial burdens related to research participation. In addition, failing to reimburse participants may result in a sample who can afford to participate, thereby raising questions about fair access to the study while also limiting the generalizability of the results.¹⁷ Consequently, though reimbursement makes research more expensive, it removes some economic barriers to participation and can improve its scientific value.

The risk of offering reimbursements in PCTs is that they might compromise the ability of the PCT to produce findings that will help predict what the effect of the intervention(s) would be in the real world in which people are not reimbursed. That is, the reimbursements effectively become a part of the intervention(s) that would not likely be present if the intervention(s) were implemented outside of the trial.

Wage payments

In some explanatory trials, participants receive “wage payments”.^7,12 In this context, "wage payment" is traditionally compensation for time and effort, and the amount paid generally scales with the amount of time and effort, as it might in an ordinary job. This does not mean that those enrolled are paid their normal hourly wage, nor that research participation is akin to other kinds of jobs, with additional ramifications of an employer-employee relationship. Rather, it refers to market-like compensation, where participants exchange something they have a right to control for money. While it is beyond the scope of this article to comprehensively elaborate the necessary arguments, by extension, this model could pertain to another category of wage payments to other things that a person might rightfully “own”, such as personal data or biological specimens. Regardless, the wage payment model seems the best fit among existing payment models to capture these sorts of concerns and clearly contrasts with reimbursement payments to offset out-of-pocket expenses described earlier. When considering the appropriateness of wage payments in PCTs it is important to consider not only payment for time and effort, but also for the release of clinical data for research. We address these in turn.

Wage payments for time and burden.

Many PCTs involve little subject involvement beyond the use of their electronic health record data, so there would be little or no extra time or burden for which to offer a wage payment. However, some PCTs include patient-reported outcome measures to create study endpoints with which interventions are compared.¹⁸ These assessments often take time to complete beyond that needed for usual clinical care. Alternatively, some PCTs involve different kinds of procedures than usual care, some of which may take longer to complete than others. For example, a study may compare the pain-management effectiveness of acupuncture, guided relaxation, and usual care,¹⁹ with acupuncture taking longer than relaxation, and both taking much longer than simply using a medication. Thus, in PCTs there may be situations in which additional data collection activities or participation in more time-consuming or burdensome interventions might justify a wage payment.

However, as with reimbursements, payments for extra time may affect how well the results can inform stakeholders’ choices in usual care. In usual care, some patients refuse to initiate treatment they judge to be too time consuming or burdensome. Moreover, patients will sometimes be less adherent to time-consuming or burdensome interventions. However, sometimes effective interventions do take more time and impose more burdens. If an investigator expects differences in time or burden between the interventions being studied, and attempts to equalize them through payments, then the findings may not reflect the overall treatment effect expected in normal care without payment.

Wage payments for data.

While the release of clinical data for research has received little attention in the literature regarding incentives and payments in traditional explanatory trials, the nature of many PCTs highlights why more attention is warranted. We classify this type of payment as a wage payment because as mentioned earlier the rationale for it most closely resembles payments in this category in explanatory trials (in contrast to reimbursements or incentives). Specifically, those involved in a trial are offered payment in exchange for something they have the right to control. In explanatory trials, as discussed above, this rationale has focused on participants’ time and burden, which is not a common a feature of PCTs. In contrast, PCTs tend to use patients’ clinical data for research purposes. That is, PCTs may harvest data from patients’ electronic health records or other sources that patients ordinarily regard as private. Researchers may judge it ethically appropriate to pay subjects for access to and use of this information.

To date, we know relatively little about whether and how individuals “price” their personal data. In addition, there is little evidence of a consensus price for anything associated with payments and incentives to research participants.^20-23 Nevertheless, recent research suggests that an appreciable fraction of U.S. consumers would not sell their data for any amount of payment, which suggests that they also would be unwilling to accept payment for the use of their data in PCTs. When people are asked about biometric data in particular, less than half claim to be willing to “sell” personal data.^24-26 Of course, consumers trade their personal data all the time, but they typically think they are getting something personally valuable in exchange, such as free software, access to a social network, genealogical data, etc. Sometimes participating in a trial may seem valuable to someone. However, some of the potential benefits to the individual subject may be unavailable when a PCT involves only a release of personal data rather than, for example, the collateral benefits associated with developing close relationships with research staff in an explanatory trial.²⁷

However, the act of offering a payment in these circumstances may have unexpected effects, even if the payment itself is justified. When the research involves only data analysis, often the researcher is not the patient’s clinician. Thus, an offer of payment for data, particularly when it comes from a third party, may seem confusing to a patient who is not even aware of the study. Other studies have shown that patients do not always respond positively to being told that they were included in a study without their consent, even if their inclusion was ethically justifiable and posed no risk to their care.^28-30 Though subjects may not refuse a small payment for their data (recognizing that in large PCTs the marginal value of any one subject’s data diminishes), the benefits to them and to the study may be greatly outweighed by the logistical burden of administering the payments and the corresponding opportunity costs of doing so.

Incentives

Incentives are used in explanatory trials with the intent to change participant behaviors, such as adherence to a study intervention or attending study visits. The predominant ethical concerns about incentives center around the ways in which they can distort decision-making about participation.^31-37 Offers of large amounts of cash or its equivalents may prompt potential participants to accept risks or burdens that they otherwise would not tolerate, and when researchers pay in such instances they treat research participants as commodities.^38,39 In some cases, a generous offer may inadvertently encourage research participants to lie about their eligibility for the study or to hide adverse effects of an intervention.^31,37,40,41

These distortions of decision-making may imply “undue influence” which undermines informed consent. What counts as undue influence has been a topic of much controversy. Recent literature has complained that the focus on undue influence of incentives has swamped other relevant ethical concerns, such as questions of justice and fairness.^35,41 Fairness in research can implicate several aspects of a study,¹⁷ but incentives can help with some aspects of fairness, most notably by improving the representativeness of the study. (Reimbursements and wage-like payments can promote fairness as well.)

Research regulations in the United States forbid paying participants for the purpose of accepting risk.^13,42,43 Generally, PCTs involve treatments that are considered standard of care and have typically already been approved by regulatory authorities, so there is no incremental risk to usual care for those included in them. In addition, PCTs are often conducted under a waiver of consent, which depends upon a determination of minimal risk. Thus, the concerns about incentives playing a role in patients accepting risk seem inapt in the PCT context; and if there is no explicit consent process for being included in the PCT, it cannot be adversely affected. Of note, the risks associated with the clinical interventions would otherwise be managed within existing regulatory and professional norms.⁴⁴

Nevertheless, a participation incentive will usually undermine the pragmatic validity of a PCT. Incentives for treatment completion or ongoing adherence to the treatment plan will usually not occur in standard practice. While there are emerging approaches among some health care plans to encourage participation in preventative health care interventions, clinicians do not typically pay their patients to take their medications, to keep their scheduled visits, or to do anything else related to their treatment. Changing this practice for the sake of a PCT directly contradicts its pragmatic aims. One exception could be experimental interventions related to behavioral economics, where the incentive itself is part of what is being studied.

Policy implications

There are several policy implications related to the use of incentives and other payments in PCTs. First, when PCTs are conducted under a waiver of consent, patients, or even their clinical teams, may not know they are being studied or what about the clinical care is different than it might have been otherwise. In these circumstances, it would be impractical or impossible to pay those included, yet having a different means of respecting their contributions seems important. Research participants are often willing to participate in a study for its own sake (up to a point), or at least for reasons other than their own financial or health interests.^38,45 Patients’ moral concerns are not limited to their immediate self-interest. Indeed, many appear to regard the possibility of improved future care, even for others, as a sufficient reason to participate in some explanatory trials.²⁸ Consequently, researchers, sponsors, and healthcare systems would do well to be able to carefully explain how PCTs are ultimately good for society and for clinical care at particular institutions. Furthermore, researchers may pursue non-monetary ways to accommodate subjects’ interests and to express appreciation for their contributions. Various options include disclosing results, reporting on study and implementation progress, facilitating entry to patient support communities, or even giving some way for participants to know that they were part of the trial so that they could mention it to their friends and family. None of these alternatives is free of ethical pitfalls, but they do not involve direct monetary payments.

Second, given the lack of a broad social consensus about personal data, the rapid changes in what is technologically feasible, and uncertainty about what subjects may expect, it will be difficult to establish a defensible payment policy regarding data use in PCTs. If the aim is to compensate subjects proportionate to the value of the data they supply, it may be impossible to even approximate the appropriate level of payment. Compensating for risks associated with these data is similarly fraught. As mentioned earlier, regulations governing research prohibit payment for accepting risks associated with research. However, it is unclear whether this prohibition extends to the use of personal data for research. Genetic data, for example, has narrow and well-defined privacy risks at the individual level, but when combined with many other people’s genetic information the nature of the risk can change. Furthermore, the value of the data accrues mostly to the controller rather than the supplier. In the software world, the tradeoff is easier to see (even if it still ends up being somewhat exploitative), as individuals enjoy “free” software in exchange for having their personal data collected and used. The tradeoff for health data is less obvious. A social consensus on the appropriate uses of personal data seems unlikely in the near future, and the history of biomedical ethics would recommend caution about assuming broad moral license on these matters. Further deliberation of multiple stakeholders regarding these issues is warranted.

Third, there are practical hurdles. Often PCTs are very large, involving thousands of participants. Paying every patient at a site, for example, could become very expensive and difficult to administer. This added administrative and financial cost may make a study less feasible, even if the basic design is sound and promises useful information about real-world outcomes. If we rule out trial designs that are very large because they cost too much, then we will not study interventions that could have large aggregate effects, but small individual effects. Suppose, for example, that a PCT could find that a particular treatment method was better than standard care, but only very slightly so. Yet because so many people get treated in this way, even a tiny improvement in each individual case could make a large difference at a systemwide or national (or larger) scale. To detect the difference, the trial needs to be quite large. Rules that demand paying each participant may effectively exclude research on these kinds of large-sample, small-effect studies, though the prospect of comparing interventions in this way is part of the rationale for PCTs. Foregoing payments in these kinds of large-sample, small effect studies may at first glance seem unfair to those included in them because they must bear distinctive uncompensated burdens. Yet some degree of inequality can arguably be justified by significant improvements in the quality of care for everyone. On this line of reasoning, in certain circumstances the benefits of improved health care for the society can outweigh some degree of unequal burdens for those included in this kind of PCT.

Fourth, much of the analysis in this paper assumes that the policy distinction between research and care remains in force. Other mechanisms for providing payments to patients included in PCTs might be possible in settings where this distinction is relaxed. For example, in a research-integrated health system patients could be offered lower copay costs in exchange for their participation in research on their care,⁴⁶ thereby recognizing patients for their contributions while avoiding some of the hurdles that incentives and other payments in PCTs present.

Conclusion

PCTs offer new ways to conduct research about medical care, but their objectives and design can make it hard to apply certain long-standing principles and guidelines for ethical conduct. Research incentives and other payments are no exception. Standard ethical concerns about undue influence and fairness remain, but in considering the use of incentives and payments, due attention needs to be directed at whether payments are compatible with the social value of a study.

More research needs to be done on several aspects of research payment practices in PCTs. First, additional research on the actual use and amounts of payments in PCTs would be welcome. Second, as studies that use data from electronic health records are becoming more common, as are public concerns about data privacy, it would be helpful to learn whether and what type of compensation might be deemed appropriate for access to personal data in PCTs, given the importance of this kind of research for various stakeholders. Such findings would be useful in informing conceptual work related to the use of payments in this setting, the deliberations of investigators, sponsors, and oversight bodies such as Institutional Review Boards, as well as policy on this point.