Paying for Patients

Last summer, Alan Farsin - his name has been changed at his request - checked into a medical clinic for five days to receive a total of two injections of an anticoagulant drug. Each injection was followed by a series of blood draws, at first taken every ten minutes but gradually tapering down to once every three or so hours. To keep other factors known and controlled, he was allowed to eat only foods provided by the clinic at specific times, and his water intake was similarly regulated.

The healthy 30-year-old English major at a Los Angeles university earned $1,500 - exactly as much as his new computer cost. "It's a form of income," he says. The study - the fifth in which he's taken part - compared the pharmacokinetics of enoxaparin sodium, a generic drug that Amphastar Pharmaceuticals makes, to Lovenox, Sanofi-Aventis' brand-name version.

The anticoagulant trial wasn't Farsin's first. About two-and-a-half years ago, a friend convinced him to enroll in a trial for West Coast Clinical Trials, a contract research organization (CRO) that recruits patients for industry-funded studies. Since then, the company has called him many times with offers to participate in other studies, and he's earned $5,000 to $6,000. "It's easy to have it become your profession, because they keep up with you," he says. In early January, he received a call inviting him to enroll in a 12-day trial, for which he'd be paid $2,500. The money is tempting, he says, but whether or not he signs on will depend on the schedule at a city college where he works as a tutor.

Farsin is one of about 2.5 million people annually who participate in research studies involving human subjects. Somewhere between 33% and 40% of the 50,000 industry-funded human studies underway pay their subjects, guesses Kenneth Getz, who founded Thomson CenterWatch and now works with the Tufts Center for the Study of Drug Development. For the approximately 30,000 publicly-funded studies, no one knows for sure.

Those payments cost companies and funders tens, and perhaps hundreds, of millions of dollars per year, and that number is getting bigger. "There's a growing acknowledgement, especially in the US, that you have to pay participants," says Getz. "Increasingly, I'm seeing more government-funded studies even in the behavioral arena that are paying well."

Still, neither the currently used guidelines for human experimentation - including the Nuremberg Code and the Declaration of Helsinki - nor any US bodies such as the Health and Human Services Office of Human Research Protection or the Food and Drug Administration, offer any guiding principles on paying subjects beyond stating that the payment not be coercive and not exert undue influence.

So how do researchers decide whether - and, more importantly, perhaps, how much - to pay participants? Those questions began puzzling Christine Grady about a decade ago. Grady, a nurse with a doctorate in bioethics who heads the Section on Human Subjects Research at the National Institutes of Health, had been reading a string of grant proposals that had a curious logic: Investigators said they'd pay healthy volunteers to undergo a medical procedure, but they would offer patients in the same trial no compensation at all. The logic was that patients already have a reason for participating in a trial - to help researchers find a treatment for their condition, or potentially to receive an experimental medicine - while healthy subjects do not.

To Grady, that was flawed logic. If the procedure is for the research, she says, and not for the individual, all participants should be paid, or not paid, the same amount. Also, other protocols had no problems with paying patients.

So she began to investigate. Grady looked in the literature, and it became clear that no one knows for sure the number of studies that pay their subjects. When she reviewed the protocols and consent forms of 467 research studies, she found the opposite of what her grant proposals had suggested: More than half of studies paid patient-subjects, compared to just a quarter that paid only healthy volunteers.¹ The wide range of payments, from $5 to $2,000 with a median of $155, also surprised her. "Everyone is worried so much about payment," says Grady, "but the amounts were pretty modest."

The only thing that seemed clear was that "there doesn't seem to be a lot of clarity," she says.

Ethicists worry about payment because of two theoretical concerns. First, money may tempt people to agree to take risks they might not take otherwise. "If money clouds your judgment, it's doing something that we would all consider unacceptable," says Scott Halpern, a senior fellow in the University of Pennsylvania's department of bioethics. Second, money may be a stronger incentive for the poor than for the wealthy, putting the disadvantaged at even higher risk.

The healthy 30-year-old English major at a Los Angeles university earned as exactly as much as his new computer cost

Halpern and his colleagues decided to test those theoretical concerns. They asked 126 patients with mild to moderate hypertension how they felt about participating in a trial for antihypertensive drugs. Regardless of whether subjects were promised $100, $1,000, or $2,000 for participating, their perception of risk did not change.² "The big take-home," says Halpern, who is also a pulmonologist, "is that [payment] doesn't turn out to be nearly as problematic as reasonable people have assumed it to be."

If the poor are vulnerable, drug addicts are even more so. With addicts, the assumption is that giving subjects money is a terrible idea, because they'll spend it on drugs. For a long time, clinicians working with substance abusers have argued that giving cash payments is just bad clinical practice. Instead, studies have traditionally offered compensation in gift cards or vouchers, or even paid part of subjects' rent or utility bills. (When Josephine Ensign, of the University of Washington in Seattle, asked homeless adolescents, some of whom are drug addicts, how they felt about being paid for health surveys, most said they'd rather be paid in store gift cards rather than in cash.) Others say that stance is patronizing, pointing out that gift cards can be sold for cash anyway. Increasingly, paying cash to substance abuse subjects is becoming almost routine.

"I'm trying to walk a tightrope by giving the most I can give to make recruiting easy and not so much that I'm being coercive." - Steven Zeisel

David Festinger, the deputy director of law and ethics at the Treatment Research Institute at the University of Pennsylvania, says that, like Grady, he was sure of something that hadn't been tested. A clinical psychologist who also runs trials on drug abuse treatments, Festinger says he thought that giving cash was bad, but he couldn't find any literature to support the idea. "That's the widely held belief," he says. "It makes sense; that's why it's so widely held. But the fact of the matter is, we don't know. It's an empirical question."

Festinger recruited 350 clients in a drug abuse outpatient clinic to complete a basic demographic questionnaire and to return six months later for follow-up. He paid subjects $10, $40, or $70, either in cash or in gift certificates. Initially they were paid most of the sum they were promised, and were asked to return in three to seven days for a drug test and a final survey assessing their level of satisfaction with the study. Festinger found that the higher the payment, the better the subject retention, and cash worked better than vouchers. He also found that drug use did not increase concomitantly with payment.³ He has since repeated the study with payments of as much as $160 and obtained the same result.

Festinger acknowledges some weakness in the study. First is its size: To convincingly demonstrate the hypothesis, he would need at least 800 and perhaps as many as 4,000 subjects. Also, subjects were drug-tested three to seven days after payment, but a week is plenty of time for a drug such as cocaine to clear the system, at least for the lab tests that were used. "I think one of the reasons it hasn't been researched is, just simply, how would you do that?" says Frances Clark-Patterson, director of Behavioral Health Services at the Metro Public Health Department in Nashville, Tenn. "Are they going to report that they went and used drugs? Are you going to follow them?"

Still, Festinger argues that the findings underscore the need for more empirical work to test assumptions about payment. "The policy should be borne out of science," he says. "Most of these decisions are borne out of debate, assumption, or conjecture, or they're influenced by one or two observations or media reports. Often you'll hear clinicians or researchers tell you, 'right outside my window I could see him buying crack.' And it's probably true, but it shouldn't influence policy, because that's just one person."

The study's conclusion, it would seem, is that it's acceptable to pay drug addicts to take part in studies. The data made Festinger comfortable with increasing payment when he had trouble with recruiting subjects for a particular study. "We're struggling to come up with interventions, but we're plagued with terrible follow-ups," he says. "If [the follow-up is] too low, you don't have any defensible data." The Institutional Review Board (IRB), however, refused to let him to increase payments. So he presented his research on payment and drug use. The IRB was convinced, and reversed its decision.

IRBs should be at least one way to standardize payments, but even within a given institution, practices can vary. Steven Zeisel studies the role of the essential nutrient choline at the University of North Carolina, Chapel Hill. He is currently enrolling subjects in an NIDDK-funded study in which subjects will be admitted to the hospital for 62 days and fed an artificial diet. The study pays participants around $3,500. "I'm trying to walk a tightrope by giving the most I can give to make recruiting easy and not so much that I'm being coercive," he says. "If I tried to give much more, the IRB might come and say that it's not quite appropriate."

The IRB is much more sensitive on payments for child studies. "We have to justify payments much more to parents," says Zeisel. One study underway in his lab tests the effect of choline in pregnancy on infants' cognition. Mothers take choline supplements during pregnancy and lactation, make four visits to the clinic during and after their pregnancy, bring their infants in for tests four times; they receive $750. His lab is running the same study in Kingston, Jamaica, and initially proposed paying parents $150 (an amount adjusted for relative income rates), but the IRB allowed only reimbursement for parking and gas. "They felt that in Jamaica mothers would be more susceptible, because of their economic status, to being pushed to volunteer their kids."

For Zeisel, who has run clinical research studies since the 1980s, money for research participants is a line item he budgets into his grants. (He receives public funding only.) "I usually budget for $75 per day for whatever subjects do, and then I'd add a little for, say, having to come back for blood draws," he says. "If we didn't give a payment, I would not be able to recruit subjects. So there's no question it's an inducement."

But the idea of inducements often makes IRBs uncomfortable. "An investigator [who] came and said, 'We'd like to pay subjects because otherwise we don't know if we'd get enough enrollment'" would raise a red flag for the IRB, says Dean Gallant, assistant dean for research policy at the School of Arts and Sciences at Harvard University and executive officer of the IRB there. "We'd want to know why - is it because the stuff they're asking them to do is not something you'd ask an ordinary person to do?" The IRB's concern in monitoring payment is not only justified, he says, but also keeps payments consistent between studies so as not to drive up costs for other researchers. Although he estimates that his IRB raises concerns over payment in just 5% of all studies reviewed, he could imagine objecting to a researcher who proposed to pay $10 for a 15-minute survey when most other researchers paid just $5.

While IRBs are extremely sensitive to payments that might seem excessive, rarely if ever will an IRB call out a trialist for paying too little. "If [payment] is absent or on the low end, no one raises an eyebrow, even though it has consequences," says Carolyn Szetela, a researcher and bioethicist who serves as the IRB chair at Meharry Medical College in Nashville, Tenn. Studies that can't recruit enough subjects to yield interpretable results waste funds, she says.

Guidelines might help IRBs navigate these ethical issues, but the vast majority of institutions lack such guidelines. One exception is the NIH Clinical Center, which sets dollar amounts for paying research participants in intramural NIH studies in units of time and inconvenience: the guideline suggest $40 per day for inpatients or and an hourly rate for outpatients, and $10 per inconvenience unit. But it doesn't tell researchers how many units an MRI might be worth, compared to, say, a spinal tap.

Even when institutions do have such guidelines, IRB members may not be aware of them. A survey of 36 IRB members at Virginia Commonwealth University found that 75% of respondents either didn't know whether their university had a written compensation policy or incorrectly believed that it didn't.⁴

As she was combing the literature, Grady found tremendous variation in payments. Subjects undergoing an MRI, for example, might be paid different sums, even within a single institution. Some of that variation may be intentional. Kevin Irwin of Yale University, who conducts intervention studies at with substance abusers, pays about $25 per hour for to subjects to complete questionnaires in urban areas. Recently, Irwin and a colleague began a study of suburban drug users, which asked then to come in for an hour and included a blood draw. Irwin settled on paying each subject $50 after doing a pilot study in which he asked potential subjects how much they thought was fair. "About 100%" said no to $25, he says, but most responded with a number between $40 and $50.

"About 100%" said no to $25, says Kevin Irwin of Yale University, but most responded with a number between $40 and $50.

Variation is likely to increase, as industry-sponsored trials such as the one in which Farsin participated are increasingly conducted at multiple sites. Payment seems to vary significantly more than would be predicted by different regions' economic climate. It's hard to gauge the effect of these variations, however, as a number of companies contacted for this article - including Merck, Biogen Idec, Schering Plough and Genentech - would not provide any information on their payment practices.

One oncology and infectious diseases clinical trial director said she pays a host medical center anywhere from $8,000 to $18,000 for each subject that completes the study. How much of that fee goes to the subjects varies.

Researchers and ethicists are beginning to grapple with these questions in a more rigorous way. At a September 2007 conference on evidence-based research ethics, sponsored by Meharry Medical College, attendees agreed it was time to apply the same standards of empirical evidence to ethical questions as to medical ones. Just as some advocates are trying to make sure that medical care becomes more evidence-based, ethicists say similar efforts are needed in clinical research.

This, of course, requires study. The obvious way to test the effects of payment is to randomize subjects to payment versus non-payment. But because of the "funny myths" around payment, says Grady, such an experiment would create a Catch-22: If it's OK to pay, how could you justify paying some people but not others? And if it's not OK to pay, wouldn't you be unduly influencing those being paid? "I'm not certain that it wouldn't be approved but that's what I would worry about," says Grady. "But I don't think it's impossible, and it's been half on my mind for a long time."

So far, most payment studies are designed to test hypothetical situations, and you can't always assume that how subjects respond will translate to what people would actually do when put to the test, says Halpern. "That's an assumption that can be made more or less valid, based on things in the research design." For example, he reviewed a study that had surveyed people who had at some time been incapacitated, in order to identify ways in which they were comfortable granting consent. The problem: Researchers didn't give subjects the option of refusing participation entirely.

While studies of payment practices go on, there's already a lot of untapped information. The NIH Clinical Center's Healthy Volunteers Program keeps a database of how much subjects have received in different studies. That data could be pulled to create a range for how researchers have ascribed "inconvenience units" to specific procedures. "Even if the formula was subject to revision," Grady says, "it might still be useful as a standard.

Joan Seiber, editor in chief of the Journal of Empirical Research on Human Research Ethics (JERHRE), says another place to start is just doing key word searches in databases of published literature. In a lot of cases, payment data gets cut from a final manuscript, but not always. "There's a whole gold mine of knowledge that no one knows how to access," she says. To fill in the gaps, JERHRE is initiating an IRB "self-study," in which IRBs will gather data on questions like how much different studies pay, and what kinds of cases raise red flags for IRBs. (See details at www.csueastbay.edu)

Even private study data may be easier to find than most researchers assume. Getz notes that at least one company, Fast Track Systems, compiles grant information for industry-funded studies. "They have a pretty good handle on the portion that's allocated to the patient as compensation," Getz says. "It would really be a matter of looking at the data." Whether they would be willing to release it to researchers, however, is unclear.

Elizabeth Ripley, a nephrologist at the Virginia Commonwealth University and Vice Chairperson of the university's IRB, recently began an NIH-funded study that asks 800 researchers and IRB members across the country to rate different procedures for the kind of payment they believe they should get, to begin to see how investigators set such values. "Really, one of my questions is, 'Can I model how people are thinking about payment?'" she says, "or is it so up in the air that we just make this stuff up?"