Hostname: page-component-7b7567c8c-fvjsw Total loading time: 0 Render date: 2026-11-15T10:26:33.107Z Has data issue: false hasContentIssue false
  • English
  • Français

Adverse selection and moral hazard in the provision of clinical trial ancillary care

Published online by Cambridge University Press:  02 April 2012

Michael R. Richards and
Lorens A. Helmchen
Michael R. Richards*
Affiliation:
Doctoral Student, Division of Health Policy and Administration, School of Public Health, Yale University, New Haven, CT, USA
Lorens A. Helmchen
Affiliation:
Associate Professor, Department of Health Administration and Policy, College of Health and Human Services, George Mason University, Fairfax, VA, USA
*
*Correspondence to: Michael R. Richards, Division of Health Policy and Administration, School of Public Health, Yale University, 60 College Street, New Haven, CT 06520, USA. Email: michael.richards@yale.edu
Rights & Permissions [Opens in a new window]

Abstract

As more and more clinical trials are conducted in developing countries, concerns arise about non-trial medical care available to study participants. Recent work argues for ancillary care – medical care not part of the clinical trial per se – to be formally incorporated into these studies. Although the provision of ancillary care is often justified on ethical grounds, a number of crucial implementation issues remain unresolved, including its scope, duration and financing. Drawing on lessons from health insurance benefit design, we highlight two overlooked challenges for ancillary care adoption – adverse selection and moral hazard – and offer recommendations that could attenuate their consequences. Specifically, adverse selection and moral hazard could be reduced by offering a choice between ancillary medical care and monetary compensation or rewarding low ancillary care utilization. Alternatively, researchers’ financial risk due to ancillary care could be shifted to a third-party insurer. Recognizing participants’ behavioral responses to prospective offers of ancillary medical care would allow funders and research teams to forecast the demand for ancillary care more accurately and to prepare for its provision more adequately.

Information

Type
Perspective
Information
Health Economics, Policy and Law , Volume 8 , Issue 2 , April 2013, pp. 225 - 234
Copyright
Copyright © Cambridge University Press 2012

Introduction

As medical research is expanding in developing countries, its effects on the host communities and participants have become a more prominent point of discussion. Expressing a growing consensus, Shapiro and Benatar assert that “research participants and their communities should be left better off after a trial, and not merely no worse off” (Shapiro and Benatar, Reference Shapiro and Benatar2005). One proposed mechanism for providing benefits to participants is ancillary care – medical care for research subjects unrelated to the study aims and not required as a consequence of study participation (Belsky and Richardson, Reference Belsky and Richardson2004; Richardson and Belsky, Reference Richardson and Belsky2004; Brownsword, Reference Brownsword2007; Dickert et al., Reference Dickert, DeRiemer, Duffy, Garcia-Garcia, Mutabingwa, Sina, Tindana and Lie2007; Richardson, Reference Richardson2008; Dickert and Wendler, Reference Dickert and Wendler2009; Hyder and Merritt, Reference Hyder and Merritt2009; Merritt et al., Reference Merritt, Taylor and Mullany2010). Prior work on ancillary care argues that it is an ethical obligation implicit within the constructs of human subject research (Belsky and Richardson, Reference Belsky and Richardson2004; Brownsword, Reference Brownsword2007; Richardson, Reference Richardson2008). The obligations to provide ancillary care have been variously justified through the lenses of welfare, rescue, justice and entrustment (Benatar and Singer, Reference Benatar and Singer2000; Richardson, Reference Richardson2008; Hooper, Reference Hooper2010; Merritt et al., Reference Merritt, Taylor and Mullany2010), and such obligations can arise in interventional trials, as well as in observational and epidemiological studies (Richardson, Reference Richardson2008).

Recently, some observers have advocated for mandating ancillary care's inclusion in all developing country research. In turn, two polar positions on this issue have emerged, ranging from the requirement that any and all opportunities to deliver ancillary care be seized to the position that no level of ancillary care should be required (Belsky and Richardson, Reference Belsky and Richardson2004; Richardson and Belsky, Reference Richardson and Belsky2004; Richardson, Reference Richardson2008). Yet, both positions have been promptly rejected by most of those devoted to the issue of ancillary care (Richardson, Reference Richardson2008). To eschew all ancillary care is seen as too callous to the needs of those enrolling as study subjects. Likewise, advocates of comprehensive ancillary care acknowledge that not all opportunities to deliver ancillary care can be reasonably pursued (Richardson, Reference Richardson2008). Unrestrained obligations to ancillary care could impose large costs on those conducting the research, and thus erode one comparative advantage of conducting clinical trials in developing countries (Belsky and Richardson, Reference Belsky and Richardson2004; Dickert et al., Reference Dickert, DeRiemer, Duffy, Garcia-Garcia, Mutabingwa, Sina, Tindana and Lie2007; Richardson, Reference Richardson2008; Dickert and Wendler, Reference Dickert and Wendler2009; Hooper, Reference Hooper2010). In addition, expansive obligations could damage study results by confounding the interventions’ treatment and side effects with the separate effects of simultaneously providing ancillary care, and thus be counterproductive to the underlying research aims (Belsky and Richardson, Reference Belsky and Richardson2004; Dickert et al., Reference Dickert, DeRiemer, Duffy, Garcia-Garcia, Mutabingwa, Sina, Tindana and Lie2007; Richardson, Reference Richardson2007, Reference Richardson2008; Hyder and Merritt, Reference Hyder and Merritt2009; Merritt et al., Reference Merritt, Taylor and Mullany2010).

That said, even eliminating these diametric views still leaves the optimal scope and duration of ancillary care indeterminate. Furthermore, much ambiguity surrounds ancillary care in terms of formally establishing a provision duty and any clear delineation of how to operationalize it consistently across studies (Richardson and Belsky, Reference Richardson and Belsky2004; Richardson, Reference Richardson2008; Hyder and Merritt, Reference Hyder and Merritt2009).

Drawing on insights from health insurance benefit design, we highlight two previously underappreciated facets of ancillary care adoption and develop plausible solutions to reduce the unintended and overlooked adverse consequences from mandating the provision of ancillary care in developing countries. By narrowly focusing on ancillary care actions of researchers and their financial sponsors, too little attention has been given to what behavioral responses ancillary care could elicit from study participants. Specifically, we caution that the universal adoption of ancillary care in developing countries may give rise to two well-known challenges present in health insurance markets: adverse selection and moral hazard. Unless they are acknowledged explicitly and managed proactively, these challenges could impair the projects’ scientific value, make ex ante cost projections underestimates and perhaps unknowable and potentially harm the very people that ancillary care intends to benefit – the study participants. Although the magnitude and implications of these behavioral responses will likely vary with the scope and duration of any given study, these principles apply across the gamut of trial types and financing models.

Adverse selection

Formal adoption of ancillary care would entail detailed descriptions of expected services and accompanying budgets within a given protocol submission and an international consensus on uniform requirements to cover all protocols (Belsky and Richardson, Reference Belsky and Richardson2004; Richardson and Belsky, Reference Richardson and Belsky2004; Brownsword, Reference Brownsword2007; Richardson, Reference Richardson2008). As Dickert states, “[g]uidelines suggesting what level of ancillary care provision is acceptable would aid investigators, review boards, and sponsors, and eventually different categories of research may come to be associated with particular levels of ancillary care commitment” (Dickert and Wendler, Reference Dickert and Wendler2009). In addition, many feel that ancillary care plans should be initially expressed to host communities and developed with their input (Shapiro and Benatar, Reference Shapiro and Benatar2005; Richardson, Reference Richardson2008; Barsdorf et al., Reference Barsdorf, Maman, Kass and Slack2010). In fact, recent work has shown that many prospective study participants from the host communities already have expectations of ancillary care benefits (Barsdorf et al., Reference Barsdorf, Maman, Kass and Slack2010).

However, a belief that some degree of ancillary care will be provided for all study participants might motivate individuals to participate in a given study simply out of the desire to receive treatment for other existing illnesses. As other researchers point out, an overcommitment to ancillary care might lead to an unethical outcome for participation, where research participants ignore or incompletely consider the risks and reasons for a particular study and focus only on the appealing prospect of ancillary care (Shaffer et al., Reference Shaffer, Yebei, Ballidawa, Sidle, Greene, Meslin, Kimaiyo and Tierney2006; Richardson, Reference Richardson2008; Hooper, Reference Hooper2010). This undesired scenario is also an indirect acknowledgment of the first behavioral challenge for ancillary care – selection.

Differential self-selection into a participant pool is a common challenge for health insurance benefit design. It does not require individuals to have complete knowledge of their current or future health state, rather only more knowledge or suspicion than an observing party. In this way, if a research group promises to provide some level of ancillary care, it may attract prospective participants who are disproportionately predisposed to accessing the full menu of available ancillary care services. Although initial health inventories are commonplace in contemporary research, there will always be a multitude of constraints on the amount of pre-participation health assessment and screening that can be undertaken. Thus, even if ancillary care options are truncated and specific, some level of adverse selection will remain because of the informational advantage the study participants have vis-à-vis the research teams. Here we highlight two direct implications.

Biased projections of ancillary care demands and costs

If individuals with preexisting or emerging conditions unrelated to the disease under study are more likely to enroll in a research study, then the ancillary care they will require may exceed pre-study projections. This situation is analogous to health insurance markets where the already ill or most likely to become ill are also the most likely to buy health insurance and drive up future costs for a given insurer (Cutler and Reber, Reference Cutler and Reber1998; Simon, Reference Simon2005), who in turn raises premiums, and thus makes the purchase of insurance even less appealing to the low-risk individuals. The consequent ‘death spiral’ destroys the insurance market for all but the highest risk groups. In the context of ancillary care, far more participants could require ancillary care than population-wide, epidemiologic metrics of the disease burden would otherwise suggest if the research study's ancillary care obligation is particularly attractive to the afflicted individuals.

This is in stark contrast with some popular recommendations calling for research sponsors to readily provide additional financial support for ancillary care and to relax restrictions on care not deemed ‘study-related’ (Richardson, Reference Richardson2008). Indeed, some claim that “ancillary care needs can be readily foreseen and planned for” (Richardson, Reference Richardson2008). However, in the presence of adverse selection among study participants, greater ancillary care demands would likely translate to greater spending, and the resulting cost overruns might force the research team to prematurely terminate a current study and deter others from undertaking similar studies thereafter. Moreover, longer and more involved trials could in turn face greater exposure to and variance in the risks and budget consequences of adverse selection, which may unduly constrain some of the scientific community's most ambitious work.

Biased inference from study results

Adverse selection could also compromise the external validity of the study results. Disease profiles or behavior patterns that were overrepresented or underrepresented in the study population as a result of self-selection, but are unobservable to the research evaluators and unrelated to observable participant attributes, could interact with the study intervention and consequently confound its estimated effects. Although random assignment of participants to the intervention and control groups would still ensure the study's internal validity, self-selection into the trial by participants could render the study no longer representative of the larger population, and thus erode the study's external validity.

Hooper reminds us that “…researchers will often be working with people who simply have no one else to turn to” (Hooper, Reference Hooper2010). Although this fact can be used as a justification for ancillary care provision, clinical trial planners and funders will also need to consider that such situations can encourage adverse selection within developing country research, which may generate substantive consequences.

Moral hazard

Universal implementation of ancillary care could also lead to moral hazard. In health insurance markets, there is a risk that the partial protection from all future costs related to an individual's medical care will discourage caution and personal investment in health by the enrollee. Likewise, the promise of ancillary care may lead individuals to relax their efforts to protect themselves against disease, not dissimilar from studies that have found that promises and perceptions of Highly Active Antiretroviral Therapy for Human Immunodeficiency Virus can tip individuals toward riskier sexual practices (Huebner and Gerend, Reference Huebner and Gerend2001; Suarez et al., Reference Suarez, Kelly, Pinkerton, Stevenson, Hayat, Smith and Ertl2001; Eisele et al., Reference Eisele, Mathews, Chopra, Brown, Silvestre, Daries and Kendall2008). Thus, by explicitly shielding these individuals from the cost of illness, ancillary care may inadvertently encourage risky health behaviors. If expectations of ancillary care overwhelm the study participants’ prior health vigilance, then the benefits of ancillary care may be offset by the emergence or exacerbation of new and existing health conditions. Furthermore, most ancillary care is unlikely to be offered indefinitely; therefore, any health conditions brought on by moral hazard and not resolved during ancillary care's availability may persist beyond the research study's lifespan.

As an example, if the scope of ancillary care within a particular study included the treatment of sexually transmitted infections (STIs), ostensibly this service could then be expected to improve public health for the participants and surrounding community by reducing the number of infectious individuals. In contrast, the expectation of treatment might induce study participants to dampen their efforts to protect themselves against STIs, such as using condoms less. In turn, the prevalence of STIs may actually increase following the introduction of STI-related ancillary care. Worse, the prevalence of STIs might increase not only among study participants, who would still have access to treatment, but also among those who did not participate in the study and consequently might not have adequate access to treatment. In this way, moral hazard could further burden the study's financial resources and potentially exhaust treatment supplies prematurely because of an unforeseen increase in the number of cases, while concurrently fostering negative health spillovers on the proximate community. The greater demand for related ancillary care would also exacerbate the cost-increasing effects of adverse selection and further complicate research in low-income international settings.

Neglecting to account for moral hazard could actually reduce the research participants’ respective health status in the medium and long term once the project's ancillary care ended and, similar to adverse selection, lead to an underestimate of project costs. In light of the fragile and occasionally absent public health infrastructure in many developing countries, it will be important to recognize the possibility that ancillary care-related moral hazard may crowd out of private prevention efforts and, in the case of communicable disease spillovers, raise the demand for medical care from non-participants. Although any negative health spillover effects might suggest expanding ancillary care coverage beyond study participants, these benefits would have to be carefully weighed against the added costs of doing so.

Adverse selection causing adverse selection

Even in the absence of any adverse selection or moral hazard effects on study participants, an ancillary care requirement will raise the costs of conducting clinical research in developing countries. As a result, “…placing society's general obligations on researchers will discourage research in underserved populations, making the population that the researchers intend to help even worse off” (Dickert and Wendler, Reference Dickert and Wendler2009). Moreover, if the universal adoption of ancillary care does lead to adverse selection and moral hazard behaviors among study participants, then research teams should be even more likely to respond by carrying out their studies among healthier and wealthier populations where the expected cost of providing ancillary care would be less. In this way, an ancillary care mandate could engender adverse selection among research teams, whereby the teams shun the most medically needy populations and leave them with less benefits than what they would have otherwise received under current study practices.

Overcoming adverse selection and moral hazard

Likely, the challenge of choosing the appropriate level and scope of ancillary care within developing country clinical trials will only grow in complexity (Richardson, Reference Richardson2008). Here, we outline three approaches that aim to minimize the potential impact of adverse selection and moral hazard on study cost projections and participant health.

Provide two options for participant benefits: ancillary care or cash

If healthier individuals value ancillary care less than the less healthy, then research teams should consider offering compensation on other margins for study participation. At the time of enrollment, instead of the default right to ancillary care, study participants could be allowed to opt for highly fungible cash payments that could be used for non-medical consumption, such as food, clothing or household supplies. This option, not unlike the common practice of monetary payments for survey participation, would draw in healthier study participants, who expected to use little ancillary care, and subsequently reduce adverse selection, while preserving the ancillary care option for the participants who valued it highly. Such a broadening of reward options for study participants would draw in a more representative sample of the general population, and thereby strengthen the study's external validity. Similar to the offer of ancillary care, a cash payment might also be construed as ‘undue inducement’ for participation, which reinforces the need for proper and comprehensive education for prospective participants before study initiation.

Reward less ancillary care utilization

To reduce the risk of moral hazard, study participants could be offered a reward, such as a cash payment or vouchers, if they used less ancillary care than projected. The reward would be limited to diseases that study participants could be expected to prevent on their own and would need to be well defined and disclosed to the participant during enrollment. In addition, and unlike an ancillary care exclusive benefit, this design would draw in study participants with low expectations for ancillary care needs and commensurately high expectations for claiming the reward, thereby mitigating adverse selection. At the same time, it would not prevent those requiring or preferring ancillary care from obtaining it. In addition, it would reduce the moral hazard element in the ‘use it or lose it’ nature of the ancillary care exclusive benefit design by giving individuals a tangible incentive to preserve their health status as best they can, while still having a health care safety net available.

Moreover, a reward, monetary or otherwise, for using less ancillary care might be preferable to other commonly used means of reducing moral hazard, such as outright exclusion of certain ancillary care services at the study's outset. Indeed, a potential downside is that treatment refusal could prematurely rule out some ancillary care services for individuals with preexisting conditions or those who acquired the illness for reasons separate from a moral hazard response. Thus, it might be preferable to incentivize low utilization of services rather than risk denying care unfairly.

Purchase insurance for project cost overruns

Research teams faced with the risk of cost overruns might be willing to purchase insurance that would shift the risk of expenditures above an agreed-upon ceiling to a third party. In effect, research teams would be buying health insurance for the study participants that covered a well-defined set of medical services unrelated to the clinical trial per se. In this way, researchers would be able to include ancillary care in their protocols without taking on any budgetary risks. This mechanism would also shift the burden of pre-study reconnaissance and calculations to the third-party insurer, allowing the research teams to concentrate on their core tasks, missions and expertise. It would also shift the adverse selection and moral hazard challenges to the third-party insurer, who would have the financial incentive to creatively tackle both problems. In addition, if a sufficient number of insurers are available, competition among them should reveal to research teams and external funders the true costs of providing ancillary care to a given population. Indeed, fostering an ancillary care insurance market might have some beneficial spillover effects on the local population if it encouraged development of more and better local medical services and mechanisms to manage the financial risks of poor health.

However, once ancillary care cost differentials among various populations became known, some research teams might refrain from choosing high-cost populations for their trials. To attract research teams and their insurers to these populations, public subsidies or philanthropic donations could be considered.

Conclusion

Promising benefits to prospective research participants is often a pragmatic part of study recruitment (Richardson and Belsky, Reference Richardson and Belsky2004). Yet, as clinical trials are increasingly conducted in developing countries, the provision of ancillary care is moving from an ad hoc consideration to codification within standards of research ethics and best practices. However, the discussion on how to universally implement such a standard without rendering studies prohibitively costly or jeopardizing scientific validity remains incomplete.

Just as it is important to determine the conduct of researchers and their sponsors in the context of ancillary care, it is equally important to anticipate any behavioral responses by study participants. We draw on insights from health insurance markets to identify adverse selection and moral hazard as possible and important overlooked consequences of ancillary care adoption. Redesigning how ancillary care benefits are structured and delivered might assist in minimizing adverse selection and moral hazard, which will be beneficial for scientists and study participants alike.

Ultimately, researchers and funders will have to choose the duration and scope of the ancillary care they wish to provide – or have it mandated upon them. The twin challenges of adverse selection and moral hazard are likely to become more acute as the level of generosity within a given ancillary care regime rises. However, the tradeoff between ancillary care generosity and risks of behavioral changes can be constructively managed to leave all parties better off. Indeed, the more researchers and funders proactively prepare, the more they can provide.

Although shifts in development and health improvements within developing countries may influence the nature and challenges of ancillary care, ancillary care can likewise affect population health and development through careful consideration and creative planning. In these ways, ancillary care that is even modest in scope could produce lasting benefits that go beyond a one-off ‘payment’ for study participation.

Acknowledgment

We are grateful to Henry Richardson for presenting his ongoing work on ancillary care at the Yale University and to Thomas Pogge for organizing the global health policy and ethics seminar at the Yale University and for encouraging this work.

References

Barsdorf, N., Maman, S., Kass, N.Slack, C. (2010), ‘Access to treatment in HIV prevention trials: perspectives from a South African community’, Developing World Bioethics, 10(2): 78–87.CrossRefGoogle ScholarPubMed
Belsky, L.Richardson, H. S. (2004), ‘Medical researchers’ ancillary clinical care responsibilities’, British Medical Journal, 328(7454): 1494–1496.CrossRefGoogle ScholarPubMed
Benatar, S. R.Singer, P. A. (2000), ‘A new look at international research ethics’, British Medical Journal, 321(7264): 824–826.CrossRefGoogle Scholar
Brownsword, R. (2007), ‘The ancillary-care responsibilities of researchers: reasonable but not great expectations’, The Journal of Law Medicine and Ethics, 35(4): 679–691.CrossRefGoogle Scholar
Cutler, D. M.Reber, S. J. (1998), ‘Paying for health insurance: the trade-off between competition and adverse selection’, Quarterly Journal of Economics, 113(2): 433–466.CrossRefGoogle Scholar
Dickert, N.Wendler, D. (2009), ‘Ancillary care obligations of medical researchers’, Journal of the American Medical Association, 302(4): 424–428.CrossRefGoogle ScholarPubMed
Dickert, N., DeRiemer, K., Duffy, P. E., Garcia-Garcia, L., Mutabingwa, T. K., Sina, B. J., Tindana, P.Lie, R. (2007), ‘Ancillary-care responsibilities in observational research: two cases, two issues’, Lancet, 369(9564): 874–877.CrossRefGoogle ScholarPubMed
Eisele, T. P., Mathews, C., Chopra, M., Brown, L., Silvestre, E., Daries, V.Kendall, C. (2008), ‘High levels of risk behavior among people living with HIV initiating and waiting to start antiretroviral therapy in Cape Town South Africa’, AIDS and Behavior, 12(4): 570–577.CrossRefGoogle ScholarPubMed
Hooper, C. R. (2010), ‘Ancillary care duties: the demands of justice’, Journal of Medical Ethics, 36(11): 708–711.CrossRefGoogle ScholarPubMed
Huebner, D. M.Gerend, M. A. (2001), ‘The relation between beliefs about drug treatments for HIV and sexual risk behavior in gay and bisexual men’, Annals of Behavioral Medicine, 23(4): 304–312.CrossRefGoogle ScholarPubMed
Hyder, A. A.Merritt, M. W. (2009), ‘Ancillary care for public health research in developing countries’, Journal of the American Medical Association, 302(4): 429–431.CrossRefGoogle ScholarPubMed
Merritt, M. W., Taylor, H. A.Mullany, L. C. (2010), ‘Ancillary care in community-based public health intervention research’, American Journal of Public Health, 100(2): 211–216.CrossRefGoogle ScholarPubMed
Richardson, H. S. (2007), ‘Gradations of researchers’ obligation to provide ancillary care for HIV/AIDS in developing countries’, American Journal of Public Health, 97(11): 1956–1961.CrossRefGoogle ScholarPubMed
Richardson, H. S. (2008), ‘Incidental findings and ancillary-care obligations’, The Journal of Law, Medicine & Ethics, 36(2): 256–270.CrossRefGoogle ScholarPubMed
Richardson, H. S.Belsky, L. (2004), ‘The ancillary-care responsibilities of medical researchers. An ethical framework for thinking about the clinical care that researchers owe their subjects’, Hastings Center Report, 34(1): 25–33.CrossRefGoogle ScholarPubMed
Shaffer, D. N., Yebei, V. N., Ballidawa, J. B., Sidle, J. E., Greene, J. Y., Meslin, E. M., Kimaiyo, S. J.Tierney, W. M. (2006), ‘Equitable treatment for HIV/AIDS clinical trial participants: a focus group study of patients, clinician researchers, and administrators in Western Kenya’, Journal of Medical Ethics, 32(1): 55–60.CrossRefGoogle ScholarPubMed
Shapiro, K.Benatar, S. R. (2005), ‘HIV prevention research and global inequality: steps towards improved standards of care’, Journal of Medical Ethics, 31(1): 39–47.CrossRefGoogle ScholarPubMed
Simon, K. I. (2005), ‘Adverse selection in health insurance markets? Evidence from state small-group health insurance reforms’, Journal of Public Economics, 89(9–10): 1865–1877.CrossRefGoogle Scholar
Suarez, T. P., Kelly, J. A., Pinkerton, S. D., Stevenson, Y. L., Hayat, M., Smith, M. D.Ertl, T. (2001), ‘Influence of a partner's HIV serostatus, use of highly active antiretroviral therapy, and viral load on perceptions of sexual risk behavior in a community sample of men who have sex with men’, Journal of Acquired Immune Deficiency Syndromes, 28(5): 471–477.CrossRefGoogle Scholar