• Open Access

Feedback of individual genetic results to research participants: in favor of a qualified disclosure policy

Authors

  • Annelien L. Bredenoord,

    Corresponding author
    1. Department of Medical Ethics, University Medical Center Utrecht, Division Julius Center, Utrecht, The Netherlands
    • Department of Medical Ethics, University Medical Center Utrecht, Division Julius Center, Huispostnummer Stratenum 6.131, P.O. Box 85500, 3508 GA Utrecht, The Netherlands
    Search for more papers by this author
  • N. Charlotte Onland-Moret,

    1. Department of Epidemiology, University Medical Center Utrecht, Division Julius Center, Utrecht, The Netherlands
    Search for more papers by this author
  • Johannes J.M. Van Delden

    1. Department of Medical Ethics, University Medical Center Utrecht, Division Julius Center, Utrecht, The Netherlands
    Search for more papers by this author

  • Communicated by Richard G.H. Cotton

Abstract

This article discusses whether and when researchers have a moral obligation to feedback individual genetic research results. This unsettled debate has rapidly gained in urgency in view of the emergence of biobanks and the advances in next-generation sequencing technology, which has the potential to generate unequalled amounts of genetic data. This implies that the generation of many known and unknown genetic variants in individual participants of genetics/genomics research as intentionally or collaterally obtained byproducts is unavoidable. As we conclude that valid reasons exist to adopt a duty to return genetic research results, a qualified disclosure policy is proposed. This policy contains a standard default package, possibly supplemented with (one or more of) three additional packages. Whereas the default package, containing life-saving information of immediate clinical utility, should be offered routinely and mandatory to all research participants, offering (one of) the three additional packages is context-specific. Such a qualified disclosure policy in our opinion best balances the potential benefits of disclosure with the potential risks for research participants and the harms of unduly hindering biomedical research. We appeal to the genetics community to make a joint effort to further refine the packages and set thresholds for result selection. Hum Mutat 32:1–7, 2011. © 2011 Wiley-Liss, Inc.

Introduction

Suppose your research team is setting up a genome-wide association study (GWAS) investigating genetic variation that may contribute to the occurrence of a particular cardiovascular disorder, say stroke. Your study indeed identifies susceptibility loci associated with an elevated risk of having an ischemic stroke. Will you return those results to participants carrying these risk factors? Or suppose your team deploys whole exome- or whole genome sequencing (WGS) and identifies several unknown and known variations, among which a rare variant that gives a large increase in the risk of having an ischemic stroke. Should you contact all research participants in whom one of these variations are found? Suppose, as a final example, you are the manager of a novel biobank targeted at pharmacogenomic studies. It is your task to design a policy for disclosure of genetic findings to biobank donors. What will your policy be and on what grounds?

In recent years, a debate evolved regarding the question whether researchers have a duty to return individual genetic research results to research participants [Bredenoord et al., 2011; Knoppers et al., 2006]. This unsettled debate has rapidly gained in urgency in view of the emergence of biobanks and the advances in next-generation sequencing technology, which has the potential to generate both in quantity and significance unequalled amounts of genetic data [Kaye et al., 2010]. This implies that the generation of many known and unknown genetic variants in individual participants of genetics/genomics research as intentionally or collaterally obtained byproducts is unavoidable, including genetic variants that were outside the focus of the study.

Interestingly, the extreme positions of full disclosure and no disclosure whatsoever are seldom defended in this debate [Bredenoord et al., 2011]. Full disclosure is often viewed as nonsensical at best, as it could imply disclosure of all raw sequencing data [Knoppers et al., 2006]. No disclosure ignores the widely recognized duty to warn when clear and present danger can be avoided. This imperative to rescue identifiable individuals when facing avoidable death is also known as the rule of rescue, which is “the duty to give reasonable emergency assistance to persons in grave and immediate peril when this assistance could be given easily, without risking harm to oneself and without neglecting one's other duties” [McIntyre, 1994, p. 158]. Similarly, if researchers have life-saving genetic information of immediate clinical utility about a research participant then a strong case can be made for disclosing these results. This can be grounded in the principle of beneficence (i.e., doing good for the sake of others), but also in empathy. If feedback of genetic research results prevents the death of a research participant, then the benefits of disclosure (saving a life) will generally speaking outweigh the costs (particularly the creation of an infrastructure for disclosure).

In view of the above, it is reasonable to assume a strong moral duty to return life-saving genetic information, an obligation that cannot easily be overruled by other moral considerations—except, for example, when the participant has expressed a prior wish not to be informed [Bovenberg et al., 2009]. Less consensual, however, is the question of whether researchers should disclose genetic research results short of life-saving data. The duty to return life-saving information emerges in situations where death is imminent, the risk of occurring is substantial and disclosure of genetic results averts death. This may, for example, be the case when a genetic study reveals the presence of an APC mutation, which will lead to familial adenomatous polyposis (FAP, a lethal variant of colon cancer) in 95% of all persons with mutations in that gene when not screened and treated. But do researchers still have a duty to return results when death is less imminent, the risks less substantial and preventive or actionable options are suboptimal or not (yet) available?

Autonomy as a Possible Ground for Disclosure

A first possible ground put forward to support disclosure is that researchers, out of respect for the participant's autonomy, should return genetic data if participants wish to receive these (e.g., [Fernandez and Weijer, 2006]). Whether showing respect for a participant's autonomy indeed supports disclosure depends on how one understands autonomy.

Autonomy, in a “negative” or thin interpretation, is most commonly understood as the individual's right to make one's own decisions without interference or coercion from others [Berlin, 1969]. From this perspective, researchers can deploy a very restrictive disclosure policy (i.e., only returning life-saving data) and at the same time respect a participant's autonomy as long as participants are well informed about the restrictive disclosure policy, have an adequate understanding, and no coercion or undue influence has occurred.

In a positive account, autonomy entails the ability to take control of one's life and to live according to one's values and beliefs [Berlin, 1969]. This understanding of autonomy is associated with concepts such as authenticity, self-expression, and self-governance. From this perspective, the researcher owes more to the participant. To show respect for people's autonomy does in some contexts not only include noninterference, it may also entail maintaining or fostering people's capacities for autonomy [Feinberg, 1987]. Many autonomous choices could not occur without the cooperation of other people in making options available [Beauchamp and Childress, 2009]. People may attribute existential value to receiving genetic results, as disclosure of genetic information can result in medical or lifestyle changes, or it can affect the participant's life decisions or strategies for coping [Ormond et al., 2010]. Knowledge about genetic risk may also guide reproductive decisions, not only to prevent the transmission of a genetic mutation but also when the genetic mutation affects fertility, for example, by influencing the timing of menopause [Broekmans et al., 2007]. Or people just want to receive information about their genetic constitution, including small and trivial risks, because they want to avoid anticipated regret [Bovenberg et al., 2009]. If we interpret the duty to respect autonomy in a positive account, then this indeed forms a ground to support disclosure, also of genetic results other than and beyond life-saving data. After all, disclosure may stimulate people's ability to take control of their life and realize or adjust their life plans. There are, however, two potential objections to deploying this positive interpretation of respect for autonomy as a ground for disclosure.

Too Demanding for Participants

First, one could counter that it is too demanding for participants. For a decision to qualify as autonomous a person should have a sufficient degree of understanding, a sufficient degree of noninterference, and the decision should be reasonably in line with one's personal values and beliefs [Beauchamp and Childress, 2009]. This means that participants should understand what range of possible findings the study may generate and subsequently decide whether and if so what kind of genetic results they would like to have feedback on. In addition, once they get feedback they should be capable of absorbing this information. One could argue that this is too demanding, even more so because most genetic information is probabilistic (i.e., it present risks, not certainties), pleiotropic (i.e., the ability of a single gene to influence multiple traits or conditions) and variants can even have opposite effects (e.g., an elevated risk for cardiovascular disease but a decreased risk for cancer). An additional complication is that new (unclassified) genetic variants may be reclassified or gain (or lose) significance. These factors challenge a proper understanding and the possibility to make a reasonable selection.

We do not deny that these are serious impediments for making meaningful decisions. This is not, however, the first time genetics faces such complex decision making. With the introduction of genetic screening programs debate about the feasibility of obtaining valid informed consent emerged as well. To balance between too much and too little information the concept of generic consent was introduced: a general consent providing sufficient information to make informed decisions but that avoids information overload [Elias and Annas, 1994; Health Council, 2010]. In addition, in daily life people constantly face complex decisions based on incomplete or ambiguous information. If we entrust people to make their own often far-reaching decisions in, for example, education, partner, mortgage and being an organ donor, we can reasonably expect them to make choices regarding whether they want to receive information about their genetic constitution.

Too Demanding for Researchers

Second, one could also object that deploying this positive interpretation of respect for autonomy as a ground for disclosure is too demanding for researchers. If taken seriously, the researcher (or another professional) should make efforts to help people understand the genetic information and subsequently to adequately select and communicate any genetic findings. The practicalities in returning results may impose untenable burdens on researchers and the existing research infrastructure. After all, giving feedback does require a careful administration, counselling services, possibly retesting in a clinical laboratory, and so forth. To what extent is it reasonable to ask researchers to meet those demands? This question can be situated in the so-called demandingness debate. In this debate, in which the limits of moral duties are discussed, it is acknowledged that some moral duties ask so much of an agent as to become unreasonable and implausible [van den Hoven, 2006].

Instead of disqualifying the positive conception of autonomy, we favor an approach in which the researcher (or another professional) makes efforts to help people understand the genetic information and subsequently to help articulating their preferences regarding disclosure. Obviously, those efforts should be in reasonable proportion to any possible benefits of feedback. In the second part of this article we discuss the scope and limits of an appropriate disclosure policy.

Beneficence as a Possible Ground for Disclosure

A second possible argument to support disclosure is found in the principle of beneficence. There are many genetic results conceivable above and beyond life-saving data that may promote the well-being of research participants, for example, because it improves their health (or avoids deterioration), their quality of life, or because they may find results of reproductive or recreational interest. Although beneficence could indeed be a valid ground for disclosure, we identify three potential objections.

Disclosure may be Harmful

First, one could argue that disclosing genetic research results other than directly life-saving information is not necessarily beneficial and often even harmful, as genetic information can have several negative consequences. Knowledge of being at risk may have adverse psychological and social consequences. GWAS and WGS studies may, for example, reveal information about psychiatric disorders or behavioral traits [Ormond et al., 2010]. In addition, genetic information may undermine someone's capacity to obtain insurance. Further, risk alleles may be discovered that have familial significance, implying that these possible negative effects may not only affect the participant herself but her family members and offspring as well. One could therefore argue that researchers, by disclosing genetic information, potentially harm their participants and their family members.

Genetic data indeed have a dual character as they can be both a benefit and a risk, causing grief, anxiety, and concerns [Wolf et al., 2008]. Medical information is often disturbing, but this is usually not considered a reason not to tell the patient about her medical status—and even considered unduly paternalistic. Similarly, the mere fact that genetic research results may have negative consequences cannot be a sufficient reason to withhold this information. Obviously, it does require that people are sufficiently aware what type of study they participate in and what kind of genetic information this may generate. In addition, if not already available, jurisdictions may have to draw guidelines to ensure sufficient access to insurance. Although all this implies that measures should be taken to prepare and educate participants and society, it does not constitute an overriding objection against disclosure. A caveat, however, should be made here for family members of participants. Whereas a participant can consent whether to participate, family members do not even have to be aware of a study. Although in clinical practice and other types of clinical research potential harm to family members is not considered a decisive reason to withhold information, this is a topic that needs further attention in the context of genetics/genomics research.

Conflation of Research and Clinical Care

Second, many have argued that disclosure would conflate the distinction between research and clinical care [Bredenoord et al., 2011; Cho, 2008]. To blur the distinction between research and clinical care would have the potential to lead to a therapeutic misconception, which means that research participants fail to recognize that the goal of research is not primarily therapeutic but to generate knowledge through answering a scientific question [Applebaum and Litz, 2008].

Actually, people only suffer from the therapeutic misconception when they mistakenly believe that the research project they are about to enter will benefit them directly. This could indeed be the case in GWAS and other genetic association studies, as these are likely to find common genetic variations that are associated with a particular phenotype with low or modest effect sizes. Furthermore, these genetic variations will often only be proxy markers for a yet unknown functional variant. In addition, genetic epidemiology studies usually deploy less accurate or sensitive methods than would be the case when used for clinical aims, resulting in a lower analytic validity. All this implies that the results will usually not be very useful at the individual level [Janssens and van Duijn, 2010]. By contrast, in studies that deploy WGS methods the chances may be greater that new variants are found that give rise to an increased disease risk [Ashley et al., 2010]. In addition, these studies will also produce information on the presence of mutations that are already known to be involved in diseases. Hence, whether people rightly think the research project could benefit them also depends on the type and aim of the genetic study and the methods used. Much genetics/genomics research will initially not generate findings of immediate relevance. Contrarily, it will often generate variations of unknown significance, or variations only modestly or poorly associated with a particular phenotype, or variations in need of replication. Single gene mutations will be much more predictive, but the probability of finding known mutations will be relatively low [Janssens and van Duijn, 2010]. Nevertheless, if researchers would truly believe that eventually (also on the long term) no information relevant for health would accrue from the research one could seriously wonder why they expose participants to the risks of participation or use public resources for it [Manolio, 2006].

The blurring between research and clinical care, by the way, does not necessarily have to be negative if appropriately recognized and may increasingly occur in the context of biobanks and WGS, where cutting edge science and clinical care are intertwined. Exome-based sequencing methods are, for example, increasingly used to investigate sporadic, unexplained mental retardation in affected individuals [Vissers et al., 2010]. This generates both diagnostically and scientifically relevant information. WGS methods are emerging in the clinical context as well [e.g., Ashley et al., 2010]. It has been suggested that personal genome studies may be the only way to explore some unresolved questions in human biology and medicine [Lupski, 2010]. Nevertheless, in case a study is highly unlikely to generate beneficial findings and a participant decides to enrol on the mistaken assumption that the research is likely to benefit her directly, measures should be taken to mitigate the therapeutic misconception [Applebaum and Litz, 2008]. A proper consent procedure could add to a rudimental understanding of the nature and goal of science. In addition, researchers should explain that the methods used for research purposes are usually less accurate than would be the case when used for clinical purposes, resulting in a lower analytic validity. Further, some distance can be created if team members other than the treating physician take the informed consent. Although the therapeutic misconception is a persistent phenomenon that may not be precluded entirely, we do not consider this an a priori argument against disclosure.

The Reasonable Limits of Beneficence

Third, whereas beneficence may be appropriate as a guiding value for clinicians, one could question whether it is up to researchers to benefit individual participants. Their primary task is to conduct research and to protect participants against undue risk. This negative duty of nonmaleficence is different from any positive duty to promote the well-being of study participants [Melzer et al., 2006].

If there would be such a positive duty at all this cannot be without qualifications, as this would require unreasonable and limitless efforts to improve the lives of study participants. Researchers should first and foremost not expose their participants to undue risk (outside the risks inherently related to being a research participant). We agree with Jonas' [1969] statement that “averting a disaster carries greater weight than promoting a good.” Researchers' protective duties toward participants are generally clear and strong. As the risks in much genetics/genomics research are primarily informational, this means that researchers should protect the data and privacy of research participants and take due care that no sensitive information will be disclosed unasked or inadvertently to participants, their family members, or third parties. Researcher's duties to promote study participant's best interests, however, require much more elucidation and demarcation [Litton and Miller, 2010]. The question here, again, is what we can reasonably ask from researchers—which will be discussed later.

Engagement, Understanding and Reciprocity

A third possible argument to support disclosure contends that an active involvement of potential research participants in decisions about disclosure of individual genetic results may educate them about the nature and scope of biomedical research and eventually also democratize research oversight [Sharp and Foster, 2006]. It has been suggested that it may in a broader sense even educate the general public about biomedical research—including the complexity, ambiguity, and occasional meaninglessness of many genetic findings [Fernandez and Weijer, 2006; Sharp and Foster, 2006]. This could potentially also reduce the therapeutic misconception, as people may have an increased awareness of the goals and nature of scientific research.

Particularly in the context of biobank research, commentators have expressed ideas and ideals about “scientific citizenship.” This notion aims to offer participants the opportunity to be actively involved in biomedical research if they wish to, including the opportunity to receive relevant genetic data [Arnason, 2009]. Also, the research community seems to favor a more direct engagement of participants and the general public [Khoury et al., 2010]. A more active and participatory approach towards research participants is in line with the recognition that clinical research is a collaborative, social enterprise that does not occur in isolation [Emanuel and Grady, 2007]. Although research participants may have altruistic or other idealistic motivations to participate in research, they may expect a form of compensation by means of a more active involvement in the research and/or feedback on their personal genetic data as well.

Translational genomics cannot progress without people being more engaged with and informed about biomedical research [Kohane and Taylor, 2010]. Offering feedback on particular genetic findings may form an impetus to gain a more profound understanding of genetics research, as people may be triggered to learn more about the type of studies and the possible findings. Although not a decisive argument to disclose genetic research results, we perceive this as a favourable side effect of disclosure.

The Duty to Disclose: Scope and Limits

Above we argued that the principles of autonomy and beneficence provide justification for disclosure of genetic research results in addition to life-saving data of immediate clinical relevance. Moreover, disclosure may have the favorable side effect that people may be more engaged with biomedical research. In sum, valid reasons exist to state a duty to disclose individual genetic research results. However, to acknowledge there is such a duty does not imply that researchers should incur any effort to give feedback. The duty to disclose genetic research results could be perceived as a prima facie moral duty, that is, a duty that must be fulfilled unless it conflicts with an equal or stronger duty. To what extent, then, should we expect researchers to make efforts to enable feedback? What efforts can we still consider reasonable in view of competing duties and values?

In order to answer those questions, it is important to determine which results should be eligible for disclosure and who constitutes the proper person to make this selection.

Participants Make a Selection

A first option would be that participants decide themselves what kind of genetic data they would like to have feedback on. This would imply that participants indicate at enrolment whether they want to be notified when results are available and if so, which. An advantage of this option is that it acknowledges the normative component of result selection. The determination of the usefulness or value of genetic results is not merely a scientific judgment; it requires normative judgment as well. In addition, this option maximizes choice for participants as they may select any potential result they might find useful or interesting.

Serious disadvantages can be identified as well. First, as notified earlier, most people will have difficulties with making a reasonable selection out of the wide array of possible genetic findings.

The quantity, significance, and ambiguity of genetic data make any choice highly complex. Although this does not constitute an argument against disclosure, it does imply that people should be offered assistance in making a selection. One could rebut that participants should organize their own assistance, but even in that case researchers cannot be put out of action completely. After all, any type of feedback would imply efforts of researchers to translate raw sequencing data into meaningful results. The distinction made between sequencing of the genome and the subsequent analysis of those data is relevant here [Health Council, 2010]. For instance, WGS results in raw sequencing data, which needs to be processed into intelligible information before the information can be interpreted. Moreover, and this is the second disadvantage, this option would constitute a significant burden on research infrastructure. To enable such an unrestrictive disclosure is resource intensive and expensive. This would imply that the resources used to inform participants detract from the actual research itself. The current trend toward increasingly larger scale studies and WGS studies would only reinforce this argument, as feedback on all useful results would be extraordinarily costly in time and money [Ossorio, 2006].

To conclude, leaving the decision about which results should be eligible for disclosure completely to the participant is neither feasible nor desirable.

Researchers Make a Selection

A second option would be to leave it up to researchers to make a selection of findings eligible for disclosure. An argument in favor of this option is that researchers are better (although still incomplete) capable of understanding the nature of the study and the potential results this may generate. In addition, they can involve clinical geneticists in the design of a study in order to discuss the findings that could possibly accrue from the study. Further, researchers have the ability to estimate the potential burden a particular disclosure policy will make on their laboratory.

A disadvantage, however, of this approach is that it completely puts participants out of action. Although it does not necessarily ignore the negative account of autonomy, it does ignore autonomy if taken into a positive account. One could argue it is not solely up to the researcher to assess whether learning a result is useful or valuable, but up to the participant as well [Lavieri and Garner, 2006]. After all, whether results may be useful requires a normative assessment of the value of the outcome, and this minimally requires input from research participants. In addition, researchers' interests do not necessarily have to be in line with the interests of participants.

To conclude, leaving the decision about which results should be eligible for disclosure completely in hands of researchers would not be an appropriate avenue either.

A Default with Additional Packages

A third option, then, would be an intermediary approach in which researchers in consultation with the institutional review board (research ethics committee), clinicians, and preferably participant representatives designate several “packages” for disclosure, or a “menu” of options [Rothstein, 2006]. This could be interpreted as a variant of the earlier mentioned generic consent: the packages can be compared with the “panel” of screening tests as proposed by Elias and Annas [1994]. We propose to offer a standard default package routinely, possibly added with (one or more of) three additional packages (Table 1).

Table 1. Qualified Disclosure Policy
PackageContentOpt-in/opt-outMoral justificationWhen to offer?
Default packageLife-saving data and data of immediate clinical utilityOpt-out systemBeneficence/Autonomy (positive account)Always
Additional package ♯1Data of potential or moderate clinical utilityOpt-in systemAutonomy (negative account)Context-specific
Additional package ♯2Data of reproductive significanceOpt-in systemAutonomy (negative account)Context-specific
Additional package ♯3Data of personal or recreational significanceOpt-in systemAutonomy (negative account)Context-specific

The standard default package minimally includes life-saving data and data of immediate clinical utility that entail a significant health problem. This will usually involve highly penetrant mutations coding for life-threatening disorders where preventive or therapeutic measures are available. Those results should be analytically valid, actionable, and accurate [Knoppers and Laberge, 2009]. An example concerns the earlier mentioned APC mutation involved in FAP. An opt-out procedure could be appropriate for the default, meaning that people will receive this information unless they have explicitly indicated they do not want to receive feedback. The moral justification of using an opt-out procedure is found in the principle of beneficence and the positive account of autonomy. Such a default based opt-out system is a variant of what Thaler and Sunstein [2008] coined liberal paternalism: liberalism in the sense that participants are still free to choose not to receive results, (soft) paternalism because we assume that people want to receive those results and, therefore, researchers should try to assist and influence the choices people make in order to make their lives healthier or better [Thaler and Sunstein, 2008]. Obviously, further discussion is needed about the precise content of this package.

The first additional package contains data of potential or moderate clinical utility. This package includes, for example, susceptibility loci associated with an elevated risk of having a disease, such as ischemic stroke, diabetes, or cancer or pharmacogenomic variants associated with adverse drug reactions. Although these data point out at less immediate threats, less severe disorders, or less actionable results (or preventive or therapeutic measures are controversial or suboptimal), the rationale behind disclosing these findings is that they may still be clinically or personally useful. However, compared with the default package the benefits of offering this package are usually lower. As any disclosure policy contains a trade-off between potential benefits (and risks) of disclosure for the participant versus the harms of hindering research, the proportionality of offering this package is context-specific [Beskow and Burke, 2010]. Relevant factors are the significance of the findings (including the clinical utility of the results), the possibilities of the research team to provide feedback and the structure of the healthcare system.

The second additional package contains data of reproductive significance. This could, for example, include information for women at risk of being in early menopause [Broekmans et al., 2007]. In addition, in view of the predicted frequency of recessive mutations in the population, participants in WGS studies are likely to learn they are a heterozygous carrier of more than one serious autosomal recessive disease (such as cystic fibrosis), which may have consequences for people in their reproductive age (and their family members) [Ormond et al., 2010]. However, only information about single gene mutations will probably have sufficient significance for considering reproductive options such as preimplantation genetic diagnosis, prenatal diagnosis, or refraining from having children. Also in view of the proportionality, this package should (currently) only include genetic information that entails intermediate or high reproductive risks.

The third additional package contains data of personal or recreational significance. The existence of Internet-based direct-to-consumer companies shows that people can be interested in receiving genetic information, even though many genetic associations are unreliable or poorly predictive. Many not yet replicated results of early GWAS studies will fit this category. Whether the tradeoff between the benefits of disclosure versus the risks of hindering biomedical research is still in balance here depends completely on the possibilities of the research team to provide feedback and the structure of the healthcare system.

Whereas the default package makes use of an opt-out, opt-in procedures are more appropriate for the additional packages. Opt-in procedures are grounded primarily in the negative account of autonomy and the participant's right not to know. Whereas the benefits of offering the additional packages are not always crystal clear, the risks for research participants could be substantial. The protective account of autonomy therefore increases in importance here. Whereas the default package should be offered routinely and mandatory, the decision as to offer one or more (or all) additional packages is context-specific and should be decided case by case [Beskow and Burke, 2010]. One could, for example, consider to only offer the second additional package when many participants are in their reproductive age or to offer the third additional package in order to enroll more participants. However, in countries lacking basic healthcare provisions, the feasibility of disclosing genetic results may be diminished if not absent. Similarly, if a research team has very limited resources these should not be used for feedback if this would be at the costs of other important values. As a rule of thumb, one could therefore state that the less significant the finding, or the more costs involved with providing feedback, the looser the duty to return results. Actually, giving participants feedback on trivial or poorly predictive results will in many cases be unjustified when this consumes resources that were mentioned for general biomedical research. In the ideal world, researchers would offer disclosure of all data if participants wish so, but in reality, individual benefits have to be balanced against societal benefits.

Conclusion: A Qualified Disclosure Policy

Valid reasons exist to state a duty to return individual genetic research results, but important competing duties and values need attention as well. We therefore have argued in favor of a qualified disclosure policy. This policy contains a “standard” default package routinely and mandatory offered, including life-saving information and variants of immediate clinical utility. Whether (one of) the three additional packages should be offered is context-specific and should be decided case by case. Such a qualified disclosure policy in our opinion best balances the benefits and harms of disclosure. From a participant perspective, it acknowledges the importance of autonomy and beneficence and the normative component of result appraisal. By offering packages, it also acknowledges the difficulties people will have with unrestricted result selection. On the other hand, it acknowledges that the efforts to realize disclosure will be relatively high while the benefits of disclosure are expected to be low, which poses an unreasonable burden on research infrastructure. After all, the predictive ability of many variants predisposing for common complex disease is low. Single gene mutations will be much more predictive, but in the absence of a positive family history or early symptoms, the a priori probability of carrying a known mutation will be low. In addition, the unknown variants are difficult to interpret [Janssens and van Duijn, 2010].

We have offered the moral underpinning and general outline of a qualified disclosure policy. It is now up to the genetics community to undertake (with some urgency) the next steps, which include but are not limited to the following issues.

First, a joint effort should be initiated to allocate results into the packages. Instead of reinventing the wheel, standard packages can be proposed—that obviously need timely updates. The already known and confirmed variants can be allocated in a package beforehand. For new findings thresholds should be set by means of two criteria: the amount of replication and the strength of the effect (odds ratio). Whether results are actionable and communicable is relevant as well [Knoppers and Laberge, 2009; Kohane and Taylor, 2010]. The assessment of these criteria could be done by a newly established (inter)national advisory committee [Fabsitz et al., 2010], but working groups of genetics societies could serve for this purpose as well. In addition, empirical studies are necessary to shed light on what participants find useful information, how they perceive and use results of various levels of validity and utility [Biesecker et al., 2009], and how they like results to be communicated—this could be used as input when allocating results into one of the package. Second, further discussion is needed about how long the duty to return results last, which is relevant when, for example, unclassified genetic variants are reclassified or variants turn out to be more or less significant than previously thought. We are inclined to follow the proposal of Fabsitz et al. [2010] that it ends when the funding ends, but this implies that the duty to return in the context of biobanks could last decades. Third, interactive educational and communication tools should be developed to help people understand and interpret possible findings and to engage people with research. Novel developments in ICT and social media could relatively easy contribute to provide supportive decision-making strategies. This could mitigate the limitations of generic consent as those people requiring more specific and in-depth information on which to base their decision can approachable obtain this.

Acknowledgments

We thank the participants of our focus groups for their valuable input and contribution.

Ancillary