Multidimensional Results Reporting to Participants in Genomic Studies: Getting It Right

See allHide authors and affiliations

Science Translational Medicine  23 Jun 2010:
Vol. 2, Issue 37, pp. 37cm19
DOI: 10.1126/scitranslmed.3000809


Recent surveys about participation in cohort studies reconfirm that participants value and desire the return of research results to a degree that is out of step with the restrictive recommendations of various ethics advisory groups, which have historically limited disclosure based on clinician value judgments and the severity and treatability of the disease in question, among other factors. Rather than framing the current inconclusive ethics discussion as a standstill among competing ethical principles and their potential applicability, we introduce a new element, communicability (that is, those properties of a message that will determine how likely it is that its informational intent will be grasped by the study participant), as the subject of empirical research to align participants’ goals with beneficent and responsible results reporting. Structural changes in research design, combined with governance changes in assessing impact, allow us to move beyond a binary construction of report/do not report and to create a structure in which the communicability of the message and the participants’ preferences are variables in a function that affects results reporting. Here we illustrate this structure and its principles.

Genomic studies that aim to link sequence variations with disease risk pose an ethical quandary: Should individual research results be revealed to participants? This issue is intensely debated, given that associations between sequence variations and risk can be misinterpreted and do not necessarily imply causality. Furthermore, risk factors might be identified for conditions that cannot be treated, and our knowledge of triggering environmental cofactors is often rudimentary at best. Should participants in research studies know what we know, or think we know, about their genes? Several values of medical and research ethics are invoked: (i) beneficence—the idea that a practitioner should promote the best interests of patients and research participants at the same time as they contribute to the social good; (ii) respect for persons, and its better-known component, autonomy—here casting participants as thinking persons with an inherent right to choose what not to know; and (iii) nonmaleficence—the idea that it is better to do nothing than to do something that might cause harm. Yet, whereas academic ethics has reached an elegantly worked-out stalemate, recent papers demonstrate that it is critically important to innovate ethical methods for reporting research results to participants in genomic studies. First, Kaufman et al. show that potential biobanking participants (who would contribute DNA samples to large repositories) overwhelmingly seek research results and, almost regardless of how such results are precisely defined, in a manner that conflicts with the restrictive disclosure standards long proposed by advisory groups (1). The National Bioethics Advisory Commission (2), a working group of the National Heart, Lung, and Blood Institute (3), and a multidisciplinary group convened by the Centers for Disease Control and Prevention (4), motivated to protect participants from the dangers of misinterpretation, have all concluded that even validated research results should be released only when results portend substantial health risks (for example, mortality or significant reproductive morbidity) and the condition is treatable. Yet 9 in 10 surveyed participants wanted to know all their individual results, “even if there was nothing [they] could do about them.” Three out of four participants would be less likely to participate if results were not provided. Indeed, providing results was the single greatest contributor to increasing survey participants’ willingness to donate to a biobank; a hypothetical fourfold increase in compensation was required to generate a similar level of increased willingness.

A key open question, being pursued by the same research group as well as others, is what these research participants mean by “results.” From our experience, it is something different than a typed sequence of DNA base pairs; rather, it is raw results plus X, where X represents hopes as varied as a functional need for personal security or a broad faith in the power of science to reveal our own evocative place within a vivid new facet of our shared human life, such as the migration routes of countless clans of Paleolithic African ancestors. The search for “results” is always personal, a word we need to pause over and deeply reflect on. If we want science to be taken as “personally” by a populace we both serve and depend on, are we right to cabin it within the dimensions of treatability and severity, the porthole slammed shut at only hypothetical storms, and the door barred with justifications that, let us be frank, are those that our grandparents heard in far different eras?

These results echo findings in diverse studies [for example, (5)], all of which deserve careful review. Participants believe that researchers are obligated to provide useful personal results. They challenge genetic researchers—who are keenly alert to loose misinterpretations in this evolving field—to rethink whether different standards might better harmonize with participants’ interests, and whether the desire to shield participants from even results from data validated with U.S. federal regulatory standards for clinical laboratory testing [the Clinical Laboratory Improvement Amendments (CLIA)] might yield to a new standard that better reconciles competing concerns. The logistical challenges of returning results to participants can seem daunting, as can the uncertainties inherent in developing and testing any new standard precise enough to apply. Nonetheless, various research ethicists have cogently argued that researchers have an obligation to return results (6, 7) and that participant preferences concerning factors such as treatability and severity should play an important role (8). Others are opposed, arguing that nonmaleficence, applied against remote but serious harms—such as the potential for suicide in response to a more horrifying form of the genetic death sentence we all face—is the more weighty principle here, and that providing results would both depend on and foster the misconception that clinical research is about clinical care (referred to here as a therapeutic misconception) (9, 10). Indeed, the genomic research context often provides little infrastructure or staff support for “treating” participants.

Recently, Wolf and colleagues argued persuasively for an ethical and perhaps legal obligation to return incidental clinical findings arising from research (11) that are not related to the primary purpose of the study [consistent with our 2007 proposal, now funded by the National Institutes of Health (NIH), for reporting results to de-identified participants (12)]. In Wolf et al.s view, the obligation arises precisely from participants’ incurable therapeutic misconceptions; a need to maximize participants’ benefits and minimize their risks; clinical obligations to participants; participant autonomy; and recent legal trends. Their argument applies as well to non-incidental findings. Indeed, for genome-wide studies, the incidental/non-incidental distinction loses force: any new finding is both within the scope of a generally framed study and in a sense is unanticipated. From a participant perspective, the desire for information about genetic risks does not depend on investigators’ verbal formulation of research aims and methods. Triangulating their insight—with the insight that standards should address participants’ personal stake in what a study may mean for them and the certainty that data will always surpass anecdotes concerning benefit and harm—we feel compelled to call for a new approach.

Research results, in the raw sense, do need validation and critical evaluation. Laboratories returning results should be CLIA-certified or the equivalent. The implications of such results need clinical assessment; overinterpretations of incidental probabilities may be excluded by research methods (13), but the clinical pertinence of a finding to a particular participant may well require a clinician’s knowledge of the patient’s history, family, and environment if the finding can be truthfully said to be more than interesting and suggestive at this stage of our knowledge. This problem is not simply a conflict between researchers’ obligation to avoid some mental harm and giving weight to participant choice to face that possible harm. That framing of the problem would assume that personal harm or benefit, as the researcher or ethicist conceives it, is the same as personal harm or benefit as the participant sees it (14) and tends to treat autonomy and beneficence or nonmaleficence as distinct competing principles in an uncertain hierarchy, as indeed they have traditionally been treated. But the principles are really intertwined, because personal views and values shape the benefits of even painful knowledge for a participant; respect for persons, as the broader principle, will acknowledge the contribution of our shared connections to our very identity; autonomous choice reaches its fullest freedom in a society rich in commitments to pluralist social benefit; and autonomous choices, to be ethical, must heed the needs of our social and interdependent natures (15). Thus, the Kaufman survey results are not merely reflections of a self-obsessed autonomy; they tell us that we have misgauged a key component of participants’ benefit and harm. If we can find a way to inform participants of results more sensitively, without causing harm, we will have enhanced the beneficence of our research and the autonomy interests of participants who wish to know more. We will have deepened our respect for them as persons. It’s no wonder that they tell us they will participate more in research.

The new element we suggest is a two-pronged mechanism to evaluate and improve how results are communicated, based on understanding participants better. Specifically, we need to understand the factors that affect benefit and harm, including what participants think “results” will tell them, and depending on that, the factors that will affect their resilience in the rare but real “bad news case.” Second, we replace fixed limitations based on the severity or treatability of a condition with respect for participant preferences. Is that respect absolute? No. The two components are variables in an equation interacting with a third, “communicability,” representing the tailored result of existing methods, and techniques yet to be developed, to create a message that is maximally understandable, maximally beneficial, and minimally harmful in a participant’s situation. A feedback loop from participants triggers both a safety net and refined approaches; this is a system that will need to learn, as researchers and participants both learn from it and create it together. We thus combine the calls of cited scholars to move toward an approach that respects participant preferences, with the recent consensus statement of leading ethicists that changes in the structure and governance of whole-genome research are essential for its progress (16). The key new point is that research designs should provide methods for querying participants not just about their disclosure preferences, but also about information relevant to appropriate communicability. Such information might include the literacy of the participant; social supports; risk tolerance; what the participant means by results, which is likely to embed a presumed inferential overlay quite distinct from a mathematical gene-phenotype probability correlation or risk relative to a given population; as well as the comprehensibility, appropriateness, and clarity of alternative possible messages describing the results. Much of this information can be gathered in initial and periodic surveys, or, as in the Boston Children’s Hospital’s design, drawn directly from the electronic medical record. Additional information and the correction of interpretive misjudgments can be obtained by querying participants in the period immediately after the nature of the results, in some form, has been given to them. Assessing communicability is analytically and practically essential to address the significance of the results and implement participant-specific preferences. Indeed, we believe that from an ethical perspective, participant preference and results significance are so intertwined that they actually cannot be understood without reference to each other or without effective communication between researcher and participant. Implementing that view produces an unusually plastic and sensitive mode of analysis.

Figure 1 summarizes the basic design of such a mechanism. It includes certain principles that ethics bodies have regarded as pivotal, including the significance and validity of the finding. The difference is that, rather than treating these questions as solely determinative, it requires some mechanism for assessing the communicability of an accurate picture to the participant to interplay with participant preferences and circumstances, to understand actual communicative effect. What emerges is not a one-dimensional analysis in which significance and scientists’ or ethicists’ abstract concepts of benefit are determinative, but a three-dimensional analysis, in which participant preference lies along the x axis, significance along the y axis, and communicability along the z axis. There are many ways to implement the principles in Fig. 1, but all require multidisciplinary engagement and participant opportunities to change preferences, given experience and the dialogued education inherent in this approach.

Fig. 1. Multidimensional results reporting.

Variables that we believe should inform the reporting of research results to study participants are indicated along three axes. Participant preference is indicated on the x axis, significance on the y axis, and communicability on the z axis. The labels of the values and their ordering are meant to be illustrative rather than normative, although in general, the further a given participant’s characteristics are from the origin in all dimensions, the more appropriate it is to communicate results. The challenge lies in situations in which characteristics in one or more dimensions are far from the origin and close to it in the other(s).


The reality is that we know little about what it takes to make a message communicable, because previous standards have prevented us from trying to find out. But we do know that many patients and family members, particularly in the context of serious disease, emphatically wish for research results to be communicated to them, even though their preferences for the mechanism and quality of that communication vary widely (17). As a practical and ethical matter, translational genomics cannot progress without participants’ active engagement. We believe that support for learning about what makes a message communicable, new as this concept is, is of the first importance. Our assessments of communicability will depend on three things: (i) the information we have solicited from participants, and whether it is predictive of their understanding and helpful in identifying and categorizing various communication methods; (ii) our apt creation of effective communication methods; and (iii) the degree to which we appropriately incorporate clinical professionals in the process. Information in the first category might include recent traumas or current health, family history with comparable conditions, and the availability of clinician support for a known condition. In studies that are continuously enriched from ongoing electronic and personally controlled medical records, such as that at Boston Children’s Hospital, information pertinent to communicability is potentially abundant. Whereas study structure would embed query mechanisms, governance might add independent review of participant and clinician feedback, with authority to propose modifications to the research team and communicate as necessary with the institutional review board. In our NIH-funded model (12), we are exploring all of these avenues, through initial participant pool surveys, a multidisciplinary Informed Cohort Oversight Board, and post-message feedback from participants. We are also involving local genetic counselors in results receipt and interpretation, and we challenge ourselves to match their commitment to sensitivity, recognizing that there will probably always be cases in which face-to-face communication is essential. Especially until we understand how we can best predict the best time and way for participants to hear bad news, we will be cautious and suggestive rather than definitive in messaging. But our hope is that through this work and the innovative work of others, we will come to understand and refine the elements essential to effective communicability, and that increases in knowledge, through appropriate public engagement, will be matched by an increasingly sophisticated and altruistic populace.

We thus respect participants’ ability to process appropriately communicated, complex, contingent information, rather than assuming that they will be unable to understand genomic complexities that are actually less intellectually taxing than some other issues they might encounter in clinical care. We also grapple with communicating more complex, probabilistic information. But we recognize the dangers of moving from theory to practice and the knowledge gap that we must transcend to do good and avoid harm. Both “sides” in the ethics debate are well intentioned: taken together, they remind us of our obligation to provide results but our responsibility to do so in an informed, knowledge-based, and skillful manner. If so informed, multidimensional factors provide greater sensitivity and ethical maneuverability than a one-dimensional approach that relies solely on abstract conceptualizations of significance. This blended approach, which incorporates autonomy, beneficence, and communicability, does not make consent and disclosure less challenging. But it increases ethical sensitivity and reconciles beneficent motives with respect for participants as communicative, autonomous individuals. The examples shown in Table 1 illustrate several of the model’s workings and challenges.

Table 1. Hypothetical situations addressed using the multidimensional results reporting model.

View this table:


  • Citation: I. S. Kohane, P. L. Taylor, Multidimensional results reporting to participants in genomic studies: Getting it right. Sci. Transl. Med. 2, 37cm19 (2010).

References and Notes

  1. Dedication: This commentary is dedicated to Gary Fleisher, Chief of Medicine, Chair of Pediatrics, and now Physician-in-Chief at Children’s Hospital Boston, in appreciation of his ability to listen; his extraordinary support of new ventures and endless questions; his active support for collaborative, multidisciplinary researchers; and his ability to discern new fields of inquiry, from the very first, when all others are in doubt. Funding: This work was funded in part by a grant from the National Library of Medicine (1RC1LM010470-01) and a grant from NIH (1RC1HG005491-01). Competing interests: The authors have no competing interests to declare.
View Abstract

Navigate This Article