Could providing financial incentives to research participants be ultimately self-defeating?

Abstract

Controversy over providing financial incentives to research participants has a long history and remains an issue of contention in both current discussions about research ethics and for institutional review bodies/human research ethics committees which are charged with the responsibility of deciding whether such incentives fall within ethical guidelines. The arguments both for and against financial incentives have been well aired in the literature. A point of agreement for many is that inducement in the form of financial incentive is permissible when the risk of harm to the individual is negligible in terms of degree and probability of occurrence. In the absence of harm to the individual, encouraging more people to participate in research would appear to be a good thing in so far as it will lead to statistically more robust research outcomes, which can then be translated into better healthcare and other practice. Whilst, on the face of it, this position seems highly defensible, I will explore the possibility that it is counterproductive – that is, providing individuals with financial incentives to become research participants may have the unintended outcome of reducing participation rates in some areas of research. In exploring this idea I will draw on empirical findings from the literature on crowding-out – the hypothesis that providing monetary incentives to people can backfire by overall reducing intrinsic motivation, in this instance intrinsic motivation to behave altruistically or undertake civic duties.

Introduction

Controversy over providing financial incentives to individuals to participate in research has a long history and remains an issue of contention in both current discussions about research ethics and for institutional review bodies/human research ethics committees (IRBs/HRECs) which are charged with the responsibility of deciding whether such incentives fall within ethical guidelines. Close to a third of researchers (29%) participating in a recent study indicated that the issue of reimbursement had been a point of discussion between them and an IRB/HREC (Fry et al., 2005: 545). The arguments both for and against financial incentives have been well aired in the literature and can be briefly summarized as follows. Foremost amongst the concerns of those who regard financial incentives as unethical are that the practice is coercive and hence undermines autonomous decision-making and compromises the scientific integrity of research owing to the disproportionate recruitment of participants from a particular, typically lower, socio-economic background (Macklin, 1989; McNeill, 1997). In contrast, those in favour of paying research participants argue that payment rightly recognizes the contribution individuals make to research outcomes and that without such financial recognition the number of participants opting to be involved in research would be insufficient to achieve statistically robust results; payment is viewed as a necessary means of increasing recruitment (Dunn and Gordon, 2005; Phillips, 2011; Roche et al., 2013; Stones and McMillan, 2010; Wilkinson and Moore; 1997).

Of note is that exponents on both sides of the debate tend to qualify their position – proponents drawing the line at some high risk trials and opponents arguing that, whilst in general payment is not permissible, there are some situations where we should not be too concerned. This middle ground is nicely captured by McNeill (1997), who argues that whilst paying participants to undertake risky research violates our social obligation to establish basic equitable safeguards for members of society, financial inducement is justified

when there is no known, and very little likelihood of, harm. The case against inducement is much more persuasive where there is a risk of harm to the subjects. There is something repugnant about offering money to relatively poor people, impecunious students, travellers and others, to take part in research which, by its nature, exposes them to risks of harm. The poor in our societies already have higher risks of poor health and other adverse life events. Inducements to take part in experimentation should not be allowed when it adds to those risks. (McNeill, 1997: 395)

A point of agreement for many is therefore that inducement in the form of financial incentive is permissible when the risk of harm to the individual is negligible in terms of degree and probability of occurrence. In the absence of harm to the individual, encouraging more people to participate in research would appear to be a good thing in so far as it will lead to statistically more robust research outcomes, which can then be translated into better healthcare and other practices. Whilst on the face of it this position seems highly defensible, I will argue that, at least for some kinds of research, we have reason to be less sanguine about providing financial incentives for reasons other than those given to date.

If it transpires that there is reason to doubt the long-term effectiveness of a financial inducement strategy, then independent of any ethical concerns there will exist a prudential reason not to encourage a system of paying research participants. Hence, even if a research study meets Grant and Sugarman’s (2004) conditions for ethical incentives, researchers may have reason to forgo the inclusion of an incentive in their study protocol. In what follows I will explore the possibility that facilitating a widespread practice of consistently providing individuals with financial incentives to become research participants may have the unintended outcome of overall reducing participation rates in some areas of research dependent on a particular, reasonably broad, group of participants. In exploring this idea I will draw on empirical findings in the literature on crowding-out – the hypothesis that providing monetary incentives to people can backfire by overall reducing intrinsic motivation, in this instance intrinsic motivation to behave altruistically by becoming a research participant.

To this end, I begin with a brief overview of the practices amongst IRBs/HRECs regarding approving research that offers financial inducements to individuals. The following sections look at some recent accounts of participants’ motivations for enrolling in research studies and examine key examples from the literature on crowding-out. The article concludes by applying the findings from the literature on crowding-out to the specific issue of providing individuals with financial inducements to participate in research.

An overview of IRB/HREC practices regarding payment

A thorough examination of the practices of all jurisdictions regarding providing financial incentives to research participants is beyond the scope of this article and unnecessary. Rather, I will briefly examine three different jurisdictions for which there is reasonable data, namely the United States, Australia and Ireland, with the view to showing that currently there does not exist a universal practice of consistently providing financial inducements for research participation.

Looking at the practices of 32 geographically diverse institutions engaged in the development, conduct and review of biomedical research within the United States, Dickert et al. (2002) found that the majority of institutions reported that participants were paid in at least some clinical studies (Dickert et al., 2002: 369). However, looking more closely at six of the institutions for which they had more detailed data, Dickert et al. (2002) found that only 23 percent of approved studies offered payment to participants. Similar results were found by Weise et al. (2002) with regard to paediatric research conducted in the United States. Weise et al. (2002) found that, of 128 institutions, ‘66% (84) reported that their IRB had approved at least one pediatric research protocol that offered payment for participation, 10.9% (14) reported not having approved protocols that offer payment, and 22.7% (29) reported never having had pediatric protocols that offer payment submitted to them’ (Weise et al., 2002: 578). Hence, as with Dickert et al.’s (2002) study, the majority of IRBs surveyed reported approving some studies that offered financial incentives; however, not all, or even the majority, of studies approved by a given IRB included financial inducements. Of further interest, Grady et al. (2005) found that approvals granted by a given United States IBC/HREC for research studies commensurable in terms of risk and participant involvement varied as to whether financial incentives were offered (Grady et al., 2005). Contrary to the commonly held view, these results illustrate a highly varied practice around payment of research participants within the United States.

Researchers examining payment practices within Australia found similarly varied results. Fry et al. (2005) distributed questionnaires to 89 individuals across 70 diverse organizations: universities, hospitals, research centres, pharmaceutical companies and market research organizations. Of the completed questionnaires that were returned, Fry et al. (2005) found that

Reimbursement via any means was standard practice for around one third (34%) of the respondents, and not standard for just over half (54%) … Examination of organisation type, research target group, and research method revealed no clear pattern to payment practices. (Fry et al., 2005: 543)

Hence, as with the United States, there is no evidence of a clear universal convention regarding payment to research participants in Australia.

Whilst providing research participants with financial incentives appears to be less acceptable in Ireland than in the United States and Australia, a high degree of variability regarding this practice can be seen to exist amongst Irish IRBs/HRECs. Roche et al. (2013) sent a survey to 89 research ethics committees in Ireland, of which replies were received from 50. Almost half of the respondents reported that they had never approved research that offered payment to participants, and 20 percent stated that they had refused approval of a research study on the grounds that it involved payment to participants.

Looking more closely at those studies that offered payment, the reason for doing so and the type of payment is also highly variable. In Australia, Fry et al. (2005) found that the most common reason for offering payment was that the ethics committee recommended it (24%), followed by perceived difficulty of recruiting the target group (21%), and whether a significant amount of participant time and effort was required (19%) (Fry et al., 2005: 544). Fry et al. (2005) found that the most common form of reimbursement was money (64%) and the least common form was reimbursement for travel costs (4%). Interestingly, Roche et al. (2013) found almost the complete opposite in Ireland. Travel was the most commonly reported form of reimbursement (20%), ‘while money was the least common form (6%) after “other” forms’ (Roche et al., 2013: 592).

What the above brief survey of the empirical findings pertaining to these jurisdictions illustrates is that there is no uniform practice or convention, either within these jurisdictions or between them, regarding the provision of financial incentives to research participants. Adding to this varied practice is the different kinds of incentives offered. Draper et al. (2009) distinguish between outright payment and reimbursement for expenses incurred, such as travel, and inducement.

This lack of uniform practice or convention is worth underscoring because it suggests that given there is currently a lack of clear convention regarding the provision of financial incentives to research participants, it is unclear what the effect would be if payment to research participants became the convention in the future, a concern I will explore in the following sections. In the next section I want to very briefly examine individuals’ motivations for participating in research.

Participants’ motivations for enrolling in research

In an Australian study, Fry and Dwyer (2001) explored the motivations of injecting drug users to participate in research studies throughout the Melbourne metropolitan area in the State of Victoria. A total of 154 current injecting drug users were recruited via notices stating that help was needed for a study on illicit drug use and that participants would be reimbursed AUS$20 for their time. The majority of participants (70%) were unemployed, 29 percent had trade or technical qualifications, and 14 percent university qualifications. Amongst other questions, participants were asked, ‘What are your main reasons for participating in research projects?’ (Fry and Dwyer, 2001: 1320). Fry and Dwyer (2001) found that for 42 percent of participants, benefit or gain to self – economic gain, personal satisfaction, seeking information/assistance – was the reason given for study participation. However, 37 percent of respondents cited reasons for research participation ‘in which the primary beneficiaries of participation were other individuals or groups (i.e. citizenship, drug user activism, altruism)’ (Fry and Dwyer 2001: 1323). The remaining 21 percent of participants cited both other and self-regarding motives for participation. Hence, for 58 percent of respondents, other-regarding motivations were of reported consideration when deciding whether to enrol in a study. A point worth underscoring is that in this study individuals were paid to participate. Hence, if the empirical data on crowding-out, which I discuss in the following section, is taken into account the results may be artificially skewed towards self-regarding motives. In other words, the offer of payment may have had the effect of reducing the percentage of enrolled participants who primarily held other regarding motives for research participation; this point will become clearer in the following sections.

Examining the views of pregnant women who had chosen to participate in an H1N1 vaccine trial at Duke University in the United States, researchers Lyerly, Namey, Gray, Swany and Faden concluded that ‘altruism was a common motivation for the women in our study who choose to participate in H1N1 vaccine trials rather than receive the vaccine in the clinical setting, although the women were also motivated by recognition that doing the latter would not be risk-free’ (Lyerly et al., 2012: 7).

In a Canadian study, Russell et al. (2000) explored the views of unpaid healthy volunteers regarding paying research participants in a study in southern Alberta. Participants came from a pool of individuals who were firstly approached to participate in a vaccine study; both those who consented and those who refused participation in the vaccine study were approached for the study investigating views about payment for research participation. Data were collected via self-administered questionnaires. The majority of respondents disagreed with the suggestion that participants should be paid for their time. Many of the comments provided by respondents were similar to those aired in the literature by opponents of a payment system – that is, concerns about the coerciveness of payment and the introduction of study bias. Of note, as illustrated by the following two responses from participants, civic duty motives were clearly articulated as reasons for participating in research:

People should volunteer because they want to help. Not because they will receive a fee.

Providing the study has very limited/or minimal risks, I feel it is our duty to participate in the advancement of knowledge without any monetary incentives. (Russell et al., 2000: 128)

In some contrast to the above discussed studies, a Brazilian study undertaken by Nappo et al. (2013) found very little evidence of other-regarding motives in reasons given for research participation. Nappo et al. (2013) interviewed individuals who had participated in at least one Phase 1 or Phase 111 clinical trial in one of three research centres in Brazil. Participants took part in an interview of approximately 40 minutes’ duration after which they were paid R$20. Nappo et al. (2013) found that the primary motivation for participants with some kind of pathology was therapeutic, with potential benefit from the new medication being the most cited reason for participating in the research. For healthy participants the primary motivator was financial gain. Only a minority of participants mentioned altruistic reasons as a motivator and only ever as a secondary motivator. Of interest is that, according to Nappo et al. (2013), in Brazil ‘the general perception is that the risks to study subjects are greater than the benefits, and the term “human guinea pig” is commonly used to described clinical study participation’ (Nappo et al., 2013: 2). The effect of such a widespread sceptical view of research participation is difficult to determine. However, it is not unreasonable to speculate that such a widespread cultural attitude will make it less likely that individuals will see research participation as something which falls into the category of civic duty, and hence the crowding-out effect (discussed below) will not be as relevant in the context of research for these participants.

The data regarding motivation for research participation is varied and interpretation complicated by the fact that the studies took place in countries with very different healthcare systems, levels of wealth, and the fact that some participants were paid to participate whereas others were volunteers. Nonetheless, what is clear is that motives for participating in research are mixed and certainly include for a notable number of participants motives that are clearly other-regarding or civic in nature. What the effect would be on these other-regarding motives, and hence research participation, if payment was to become the norm is a question I will explore in the following sections.

Crowding out

In his book The Gift Relationship: From Human Blood to Social Policy, Richard Titmuss (1970) introduced the hypothesis that paying people to perform a task that they would have otherwise performed for free has the overall effect of reducing people’s willingness to perform the act. As evidence for his hypothesis, Titmuss (1970) compared blood donation and transfusion services at that time in the United States with those in England and Wales. For the most part, blood donation services in the United States were market-based: donors were either paid to give blood or required to do so in exchange for the use of blood products by themselves or relatives. In contrast, the system in England and Wales was based on a gift model: donors gave their blood to strangers for free.

In general, Titmuss (1970) found that the system in England and Wales was efficient and on the whole met the demand for blood and blood products. In stark contrast, the various commercial centres in the United States were plagued by both shortages and wastage. Focusing on a centre in New York City, Titmuss (1970) found that over time the proportion of paid donors increased: 15 percent in 1952, 42 percent in 1956 and 55 percent in 1966 (Titmuss, 1970: 96). A 1966 report found that surgery was often postponed in New York owing to a lack of blood even though wastage was estimated to be between 10,000 and 30,000 units a year (Titmuss, 1970: 63–64). Titmuss (1970) concluded that the growth of commercial blood banks ‘had the effect of making the contribution of the voluntary donor an even less significant one in the United States’ and moreover did not ameliorate the inefficiencies in the system (Titmuss, 1970: 95). In contrast, Titmuss’s analysis of the voluntary service in England and Wales during the same period found that it responded well to changes in demand for blood and blood products from medicine and surgery (Titmuss, 1970: 44).

Titmuss’s hypothesis that introducing monetary incentives has the overall effect of reducing people’s willingness to perform acts previously undertaken for altruistic motivations has gained support from social psychology. In a study surveying people’s attitudes to having a nuclear waste facility in their neighbourhood, Frey and Oberholzer-Gee (1997) found that respondents were less likely to accept the facility being built in their area if they were offered monetary compensation. Similarly, Gneezy and Rustichini (2000) found that high school students in Israel collected less money for charity if they were offered a small financial incentive than if an incentive was not offered.

More recently, Mellstrom and Johannesson (2008) set out to empirically test Titmuss’s original hypothesis that offering financial incentives to blood donors was counterproductive to ensuring adequate supply. Mellstrom and Johannesson (2008) conducted their study in Sweden where individuals complete a health examination prior to donating blood. If the health assessment is favourable, individuals receive an appointment time to donate blood. Mellstrom and Johannesson (2008) recruited participants who had not had the required health assessment into one of three study arms: no offer of compensation to complete the health assessment; offer of about $7(US) to complete the assessment; and offer of a choice between $7(US) or an equivalent amount donated to the Swedish Children’s Cancer Foundation charity.

Overall, Mellstrom and Johannesson (2008) found that whilst a crowding-out effect was seen for women it was not found for men, suggesting a significant gender effect. Of interest is that the crowding-out effect for women was counteracted when a donation to charity was offered. Mellstrom and Johannesson’s results are summarized as follows:

For the overall sample the supply of blood donors decreases from 43% to 33% when a payment is introduced, consistent with a crowding-out effect, but the effect is not statistically significant … However, the results differ markedly between men and women. For men there is no significant difference among the three experimental groups. For women there is a significant crowding out effect; the supply of blood donors decreases from 52% to 30% when a payment is introduced. There is also a significant effect of allowing subjects to donate the payment to charity, which fully counteracts the crowding-out effect. (Mellstrom and Johannesson, 2008: 847)

The gender effect found by Mellstrom and Johannesson (2008) is striking and warrants careful consideration by researchers, who are often pressed to meet recruitment targets. For those researchers whose studies can utilize either men or women, the consequences of Mellstrom and Johannesson’s (2008) findings will be of less concern than for those researchers engaged in research areas such as obstetrics and gynaecology that require female participants. However, it is not only researchers working in the area of women’s health who will find offering incentives counterproductive to the aim of maximizing recruitment but presumably also researchers working in areas which typically rely on women to provide proxy consent; an obvious example is the broad category of paediatric research. Mellstrom and Johannesson’s (2008) findings also potentially raise concerns for funding bodies such as the US National Institutes of Health (NIH) and regulating bodies such as the Australian National Health and Medical Research Council (NHMRC). Both of these organizations have views on the requirement of gender equity in recruitment. The NIH has long required studies involving humans to include both men and women and the NHMRC presses for equitable gender representation in research, with the inclusion of a specific question about gender in the section pertaining to recruitment in the National Ethics Application Form (NEAF) – a form now required by most public health HRECs.

Whilst further empirical data confirming the gendered crowding-out effect from the provision of financial inducements in research is undoubtedly required, Mellstrom and Johannesson’s (2008) findings should at least make researchers proceed with caution before adopting a universal policy of offering participants financial incentives. Of course, a policy that differentially offered inducements in response to the gender effect would raise concerns of equity and fairness. Furthermore, were such a discriminatory policy to be adopted it is likely that even fewer women would agree to become research participants. The issue needs further consideration but will have to await further empirical data.

In their article ‘Inducements revisited’, Wilkinson and Moore (1999) are sceptical about the overall detrimental effect of offering research participants inducements. They consider the argument that if inducements were offered, ‘soon most potential subjects would want to be paid’ (Wilkinson and Moore, 1999, 119), which might lead to the commercialization of research, which would tend to favour well-funded researchers and disfavour those less well funded. Wilkinson and Moore (1999) ask why this would necessarily be a concern, and they may well be correct. However, Mellstrom and Johannesson’s study suggests a different concern regarding the crowding-out effect. The concern is not that allowing inducements will lead to the commercialization of research, but rather that those individuals who are motivated by civic duty will simply stop participating in research or direct their charitable efforts elsewhere. If this effect manifested in a particular group, as opposed to random individuals, then researchers dependent on that group would experience additional challenges in meeting recruitment targets. Women motivated by civic duty to become research participants such as demonstrated in the HINI trial (Lyerly et al., 2012) would not enrol in the study, perhaps choosing other forms of altruism. Given the large amount of unmet social need that would benefit from charitable interest, it could be asked why this would matter, and this may well be a valid point. However, the focus of this article is not on the very laudable concern of raising net social benefit but rather on the issue of whether researchers should offer inducements if their motivation is to optimize recruitment in their particular study. Given the difficulties researchers often face with recruitment this is a very real concern.

Interestingly there have been other studies that have provided some evidence that financial incentives do not reduce research participation. For example, Slomaka, McCurdy, Ratliff, Timpson and Williams (2007) elicited the views of 37 economically disadvantaged African American crack cocaine users regarding the importance of offering financial incentives for participation in research. Slomaka et al. (2007) concluded that participants ‘in our study viewed payment for participating in research as essential to attract individuals to a study’ (Slomaka, 2007: 1408). For this group of participants, money from research studies provided a relatively safe, legal and honest avenue of income:

As one interviewee said, ‘it’s not drugs that was the first thing went in my mind. Closest thing on my mind was getting a hamburger or something’. One woman explained that necessities such as food for her children and gasoline for their car would take priority over drug use. Another explained that her decision to participate in a clinical trial for $1500 was motivated by the need to pay household bills out of concern for her children. During the latter clinical trial, this interviewee described another woman who had also enrolled in the drug study. According to the participant, the woman was employed, but had enrolled in the clinical trial to obtain extra cash for a rental deposit. (Slomaka et al., 2007: 1407)

It is important to note that the crowding-out effect is not absolute, extinguishing all other intrinsic and extrinsic motivators such as a need to provide food for one’s children or money for the rent. Rather, the crowding-out effect is one factor amongst many that researchers engaged in some areas of research should consider when seeking to maximize recruitment.

In another study designed to assess, amongst other things, the effect of monetary payment on participants’ purported willingness to participate in research, Bentley and Thacker found that monetary payment had a significant effect on willingness to participate in research, with ‘lower levels of payment leading to lower willingness ratings’ (Bentley and Thacker, 2004: 296). In Bentley and Thacker’s (2004) study the highest level of hypothetical payment was US$1800 and the lowest US$350. It is worth emphasizing that the majority of actual studies do not pay participants even the lower level of incentives hypothetically offered in Bentley and Thacker’s (2004) study. However, as Bentley and Thacker (2004) acknowledge, a limitation of their study is that it was hypothetical in nature – a limitation which also applies to Slomaka et al.’s (2007) study. It is for this reason that Mellstrom and Johannesson’s (2008) blood donation study is of particular interest; it tracks actual behaviour rather than behaviour that participants predict they would manifest in a given situation.

Conclusion

Empirical findings such as those from Mellstrom and Johannesson’s (2008) study suggest that researchers working in some areas may have a very pragmatic reason to be wary of offering financial incentives to research participants. Hence, researchers working in areas such as obstetrics, where participants necessarily have to be women, should proceed with caution before offering financial incentives if their aim in doing so is to boost participation rates. However, the lessons from Mellstrom and Johannesson’s (2008) study are potentially wider. Research that relies on women – not as participants but as persons who provide proxy consent or facilitate participants attending study appointments – may also be subject to the crowding-out effect. Paediatric trials, such as vaccination studies, where it is more likely to be mothers than fathers who consent to their children being in the study and regularly accompany them to study visits, are examples of a wider area of concern. What the long-term effect of creating a universal expectation of payment will be on research participation and smaller investigator-led studies which lack the financial resources of well cashed-up sponsored trials is unclear. Certainly, more research of a non-hypothetical nature is warranted.