Is Real-Time ELSI Realistic?

Abstract

Background

A growing literature has raised—skeptically—the question of whether cutting-edge scientific research can identify and address broader ethical and policy considerations in real time. In genomics, the question is: Can ELSI contribute to genomics in real time, or will it be relegated to its historical role of after-the-fact outsider critique? We address this question against the background of a genomic screening project where we participated as embedded, real-time ELSI researchers and observers, from its initial design through its conclusion.

Methods

As part of the ELSI study design, the project included an ongoing reflexive ethnography in which the authors studied the process of its design and implementation. The authors were true participant observers, serving as members of various task-oriented groups while recording meetings and other events for ongoing qualitative analysis. We also conducted and analyzed interviews of multiple participants at the conclusion of the project.

Results

Our real-time ELSI initiative had a mixed record of successes and challenges. If we take define success as ELSI researchers having had an opportunity to participate fully in the project and to make the ELSI perspective heard, then our assessment is largely positive. If, however, we define successes as instances where real-time ELSI contributions changed the direction of the genomic or public health aspects of the GeneScreen project or, after careful deliberation, confirmed the appropriateness of the status quo, then we can identify only a few examples. While we had a seat at the table, we were, for the most part, tolerated guests.

Conclusions

We conclude that there are significant barriers to real-time ELSI influence. The difficulty does not reside in any intended exclusion of an ELSI perspective, but in factors endemic to genomic research, including knowledge disparities, epistemological biases, and the pressures of time and money.

Background

The birth of the Human Genome Project (HGP) engendered great enthusiasm for the potential impact that sequencing the human genome could have on science and medicine. However, concerns arose that a better understanding of the human genome could have serious unintended social consequences, including unwanted disclosures of sensitive information, misuse of such information, and genomic discrimination (Greely 1998). These issues were particularly compelling against the background of multiple notorious eugenics projects in the U.S. and around the world (Greely 1998). Public and professional concern about the potential social consequences of the HGP led the project’s leadership in 1990 to allocate five percent of its budget to fund the Ethical, Legal and Social Implications (ELSI) Research Program within what is currently known as the National Human Genome Research Institute (NHGRI). The program would fund projects designed to examine ELSI issues in genetics and genomics research (Marshall 1996).

There is an extensive literature on the value of ELSI projects, going back to the beginning of ELSI research. With respect to their relative success, the most generous reading of the literature is that the results have been mixed. (We discuss below the problematic issue of what constitutes an ELSI project’s “success.”) One particularly strident critic concluded in 2001 that “‘ELSI has failed’ because, although the ELSI programme produced thousands of papers, ‘virtually nothing has been accomplished’” (Penders, Horstman, and Vos 2008, 711, quoting Kitcher 2001, 189). Consistent with this critique, several early reports of ELSI projects within the HGP described dismal results. According to these reviews, ELSI was “doomed to fail from the beginning”; it purportedly suffered from a suppression of dialogue by the funder, the National Institutes of Health (NIH), and its difficulties led to the public resignation of the chair of the ELSI Working Group (Andrews 1999, 199; Fisher 2005, 323; Annas 2002, vii). James Watson is quoted as describing his goals for this working group as follows: “I wanted a group that would talk and talk and never get anything done” (Annas 2002, vii). The more recent literature has continued to report a myriad of problems ranging from personality disputes (Beckwith and Beckwith 2009, 191-210) to disciplinary conflicts or barriers (Bell 2016; Beckwith and Beckwith 2009, 191-210), to the risk of rubber-stamping when ELSI is funded by the scientific project into which it is integrated (Burke et al. 2015, 15; Pullman and Etchegary 2015).

Nonetheless, ELSI research continues, with reports of successful outcomes notwithstanding these difficulties (Hall et al. 2010, S34; Smith et al. 2016; McGuire et al. 2012, 11). One reported success is the promotion of increased dialogue, both between researchers across disciplines and also with community groups outside of the research realm—though ELSI studies have been criticized for merely engaging in dialogue without effectively influencing policy (Burke et al. 2015). The growth and prosperity of ELSI Centers of Excellence (CEERs) might be viewed as another success, as each CEER focuses its attention on specific issues associated with genomic research’s rapid growth. Others argue that the inclusion of ELSI requirements in NIH consortia grants that focus more directly on medical care, such as the Clinical Sequencing Evidence-Generating Research (CSER) consortium and the Electronic Medical Records and Genomics (eMERGE) Network, has had the positive effect of giving ELSI researchers access to diverse types of genomics research (Meriom et al. 2016, 33; Burke et al. 2015). Another indication of (indirect) success is that nanotechnology researchers report that their community has learned from the experience of ELSI genomics and is working to develop a more successful integration strategy (Viseu 2015).

Debate continues about what role ELSI should play in genomics research (Burke et al. 2015). One view is that its primary purpose is retrospective, to follow genomics research to study the aftermath. An alternative vision is that ELSI research should come first, illuminating the path that genomic research should take in order to avoid undesirable ethical, legal, and social consequences. Indeed, this was how much of early ELSI research outside the HGP began (McEwen et al. 2014, 493). In this latter view, one possible outcome of prospective ELSI research is to argue that particular genomic projects should not be undertaken at all. A third, middle-ground view is that ELSI research should be integrated into genomic projects, advising on unfolding consequences so that ongoing projects might be modified in real time or, in extreme cases, shut down.

The ELSI and genomic (and other scientific) research communities have had an especially uneasy relationship in projects that take this third approach. This unease has been most acute when ELSI researchers have been “embedded” within a project in both the spatial and temporal senses, being physically present as the scientific process unfolds, developing their own perspectives, and offering advice when they deem it relevant, whether solicited or not (Pullman & Etchegary 2015; McEwen et al. 2014; Marshall 1996). We call these “real-time ELSI” projects, and use the term to refer to a study that uses ethical, legal, or social scientific research to influence the basic scientific (and possibly clinical or public health) aspects of the study as it unfolds, from inception and design through analysis and publication.

In this paper we consider the issue of ELSI’s role in scientific research through the lens of our experience as ELSI researchers embedded in—indeed, built into the research design of—a joint genomics-ELSI study. Entitled GeneScreen, the study aimed to examine whether and how a genetic screening panel for healthy adults might be employed in routine patient care. The question of whether the screening should take place at all was a critical component of the study design but, as we shall explain, it was forgotten immediately as the project began. Whereas the ELSI participants saw the project as an opportunity to challenge and perhaps reshape first principles of public health screening in the novel genomic context, in the event GeneScreen proceeded ineluctably according to established norms and precedents.

Our focus here is in keeping with our real-time ELSI definition. We analyze the ways in which ELSI researchers, including the authors, attempted to encourage examination of the ELSI issues in order to help shape the genomic and public health aspects of the GeneScreen study, and the ensuing effects. We draw on these experiences to address the larger policy question of the value of embedded, real-time ELSI, and to offer suggestions for others involved in comparable projects.

GeneScreen commenced in mid-2013 with an overall aim of evaluating whether the public health-scale screening of the general population is a desirable goal. Toward this end, GeneScreen planned to screen 1,000 healthy adults in order to analyze the harms and benefits of such screening. The project was conceived of as the main research activity of the Center for Genomics and Society, an NHGRI-funded CEER. GeneScreen was innovative in that it was designed to be an ELSI study with an embedded genomics component, as opposed to a genomics (or other clinical) study with an ELSI component.

The project’s first major endeavor, selecting which genes to include on the screening panel, occupied most of the researchers’ time through early 2014. Briefly, the GeneScreen project used a four-point scoring metric to evaluate the clinical actionability of genes and associated medical conditions. This metric had been developed by some GeneScreen researchers as members of a CSER project, NCGENES, which enrolls patient-subjects. Most of the GeneScreen clinicians, but only two of its ELSI researchers (and none of the authors) participated in NCGENES. Ultimately, the GeneScreen panel consisted of 17 genes related to 11 conditions (Adams et al. 2016). At the conclusion of the gene selection process, efforts to design and manage implementation of the screening commenced, including discussions of who would be eligible to participate in GeneScreen.

At the conclusion of the gene selection process, efforts to design and manage the implementation of the screening commenced. Various groups dealt with such issues as participant eligibility, recruitment, consent, and website design (Butterfield et al. 2018; Cadigan et al. 2017; Waltz et al. 2017). The actual implementation of GeneScreen was plagued by technical problems, including unexpected difficulties in making the website operational and, in the fall of 2016, the failure of the unproven molecular inversion probes (MIPs) sequencing technology that GeneScreen had intended to use.

During the transition from gene selection to implementation, discussions had begun between GeneScreen’s leaders and researchers at Kaiser-Permanente (K-P) in California for a “scale-up” of GeneScreen using K-P’s vast genomic biobank. These developments were announced to the rest of the GeneScreen researchers in fall 2014. The intended collaboration with K-P was not funded and thus never materialized. GeneScreen had intended to recruit participants from multiple sites in North Carolina, but ultimately used two sites, recruiting patients in a general medicine clinic in North Carolina and participants in the K-P biobank (Cadigan et al. 2017). After the sequencing problems were finally resolved, positive and negative results were reported to participants in April 2017.

Methods

The entire GeneScreen project, including the research reported here, was approved by an IRB at the University of North Carolina at Chapel Hill. As part of the ELSI study design, GeneScreen included an ongoing reflexive ethnography in which the authors studied the process of designing and implementing GeneScreen. We did this in multiple ways. We observed and audio-recorded 11 meetings of two GeneScreen working groups (the “Gene Selection Committee” (GSC) and the “Eligibility Criteria Committee” (ECC)) that occurred from July 2013 through April 2014. In these meetings, a multidisciplinary group of GeneScreen researchers (including bioethicists, social scientists, genomic researchers, and clinicians) debated, respectively, which conditions and genes would be included on the targeted screening panel (Lázaro-Muñoz et al. 2017) and what the eligibility criteria for participants should be, such as age and insurance status. We were participant observers in those meetings, as well as five two-hour meetings of GeneScreen’s Community Advisory Board (CAB).

We also conducted interviews with nine GeneScreen researchers, seven of which occurred after the GSC and ECC meetings concluded, and two at the completion of the GeneScreen study. At least one author conducted each of the interviews, which lasted about 60 minutes and were audio-recorded and transcribed. In the interviews, we asked the researchers to reflect on the process of designing and implementing GeneScreen, including the challenges and benefits of having a multi-disciplinary research team. Additionally, we reviewed transcripts of the five CAB meetings, extensive email traffic among GeneScreen researchers throughout the project, documents created by various researchers during the course of the project, and the authors’ observation notes, which were taken at all meetings and events in which we participated.

As the project proceeded, we met regularly as a group to review these data sources to identify important issues that emerged in real time, and that subsequently became major topics as we interviewed the various GeneScreen researchers. We organize those issues into two categories: successes, on the one hand, and challenges, on the other—a problematic dichotomy that we discuss below. In many instances we use quotes from transcribed meetings and interviews to illustrate how the respective issues emerged and the ways the GeneScreen researchers have thought about them. In doing our analysis, we assigned each GeneScreen researcher to one of two categories, practicing clinician or ELSI researcher. We defined an ELSI researcher as someone whose primary focus of inquiry and practice is ethics, the law, or the social sciences. This would not, of course, necessarily preclude a clinician researcher from being an ELSI researcher, but among the GeneScreen team members, none of the practicing clinician researchers had a primary focus on ethics, the law, or the social sciences. All of the authors are among the ELSI researchers. The practicing clinician researchers were all medical geneticists and genetic counselors. We found this distinction between practicing clinician and ELSI researchers important because many of the debates that emerged during the course of the project saw GeneScreen researchers’ views divided along this axis.

Results: Successes and Challenges

Successes

Success is a relative term, of course. If we take a broad, procedural view and define success as ELSI researchers having had an opportunity to participate fully in the project and to make the ELSI perspective heard, then our assessment is largely positive. If, however, we define successes narrowly and substantively, as instances where real-time ELSI contributions changed the direction of the genomic or public health aspects of the GeneScreen project or, after careful deliberation, confirmed the appropriateness of the status quo, then we can identify only a few examples. We take the broader view first.

Ability to participate and be heard.

The ELSI researchers had unfettered access to all aspects of GeneScreen’s planning stages, including importantly the ability to participate in meetings of the GSC and ECC groups. Additionally, the two anthropologist authors on this article were permitted to study all aspects of the process as participant observers and to audio-record all meetings. Moreover, ELSI researchers had abundant opportunity, at least in the early stages, to ask questions and state their views. At a minimum, the clinician researchers listened and responded to questions and comments, despite the time pressures that occurred (described in the “Challenges” section below).

Selecting the genes to screen.

The ELSI researchers’ efforts to contribute to the process of defining and applying the gene selection metric criteria provide a mixed example of success and challenge: while these efforts did relate to the specific substantive topic of gene selection, they demonstrated little more than the ELSI researchers’ general ability to participate. As described earlier, GeneScreen inherited a scoring metric from NCGENES that evaluated genes for their clinical actionability. The metric criteria for both projects were the severity of the condition, the likelihood of the condition, the effectiveness of the intervention, the acceptability of the intervention, and the knowledge base about the condition and the intervention.

Because of the metric’s history, the clinician researchers treated it—or at least its overall framework—as a given and not open for debate. The presence of the taken-for-granted metric was, from the ELSI researchers’ perspective, an important factor in taking off the table the fundamental question of whether the screening was appropriate. Instead, the GSC meetings began with a Powerpoint lecture by a clinician on the metric and a short list of candidate genes. The same clinician opened the next meeting by saying:

It would be unworkable and kind of silly to just think in abstract terms about the criteria and then try to do it. I think one should start to formulate some lists, figure out why it’s important, etc.

Statements such as these suggested to the ELSI researchers that we had already assumed away the question of whether we should screen—the ultimate question of the project, as we understood it—and were starting with the details of how we would do it.

During the meetings of the GSC, one challenge did emerge to the metric itself. The metric contained five criteria (i.e., severity of the health condition, likelihood that the condition would materialize given the pathogenic variant, effectiveness of available preventive interventions, acceptability of available interventions in terms of burdens placed on the individual, and knowledge or evidence available to score the other four criteria) (Lázaro-Muñoz et al. 2017). Each criterion received a score between 0 and 3. The acceptability criterion was scored as follows: 0 = Not acceptable, 1 = minimally acceptable, 2 = modestly acceptable, 3 = highly acceptable.

ELSI researchers and some clinicians questioned and debated the meaning of acceptability, asking in particular: “acceptable to whom?” The concern was that clinicians, often invoking grim patient care anecdotes, were relying solely on their own sense of what participants would deem acceptable. The ELSI efforts could be deemed successful in that they compelled the GSC to talk—and thus presumably to think—at greater length about acceptability than some might have been prepared to do. As one of the clinicians acknowledged during this discussion, “[w]e have tried very hard to ignore the fact that we’re clinicians when we come up with the acceptability scores… [but] we have no idea if we’re right compared to the lay person.”

In fact, the GSC decided to bring the discussion of acceptability to the CAB about halfway through the process of selecting genes. But the CAB’s input had little impact on the way acceptability was scored or the overall selection of genes, in large part because the CAB, too, had difficulty sorting out how to score acceptability. In fact, the CAB’s members were reluctant to offer scores for acceptability at all, preferring instead to talk about the subjectivity inherent in the concept. In the end, the clinicians’ collective judgments about acceptability prevailed.

Ultimately, the ELSI researchers extended the scope and length of the GSC’s deliberations in ways that, we hope, enhanced the understanding and sensitivities of all participants. But real-time ELSI participation had no obvious influence on the genes that were selected. Whether we should reasonably have expected anything else is taken up in the Discussion.

Including the uninsured.

In our interviews, one specific real-time ELSI contribution was mentioned by just about everyone, clinician and ELSI researchers alike: persuading the group to allow uninsured people to enroll in GeneScreen. The initial operating assumption had been that we should allow only participants with health insurance, because it would be unethical to inform people of a genetic condition for which they could not afford follow-up care. But in April 2014 one of the authors, a lawyer and bioethicist who had previously assisted patients with genetic conditions who were battling their insurance companies, made a presentation to all of the GeneScreen researchers. The presentation revealed that the insured/uninsured boundary was illusory in this context, since having health insurance would not guarantee coverage for the kind of preventive follow-up that would be appropriate here. The group’s operating assumption was discarded and we opened recruitment to otherwise qualified participants regardless of insurance status. Impressed with the presentation, the researchers recommended its presentation to the CAB, where it was clearly appreciated as well.

This persuasive presentation had some of the attributes of a scientific talk: 1) it was based on discrete pieces of evidence (e.g., statements of law and insurance policy clauses) rather than broader arguments, 2) evidence was displayed to and evaluated by those in attendance, and 3) the conclusion advocated for was a single yes/no change. The presenter’s extensive practical experience with patients and insurance companies was also highly valued by the GeneScreen team. In the end, the change was made, and the episode has entered the folklore of the project. Both clinician and ELSI researchers continue to praise the event as an example of an ideal ELSI contribution. As one clinician enthused:

Yeah. Yeah. Because I mean just talking with her I learned so much about this stuff. It’s like “I don’t – I mean how do I know about medical records and where they go and you know all that?” And so that’s been very interesting to learn some of that. Yeah. So something like that.

Challenges

For purposes of this discussion we define a challenge as an aspect of the project that inhibited real-time ELSI contributions. Some of the challenges we observed seemed to be structural and likely to occur in any ELSI-genomics collaboration, whereas others seemed more specific to our project.

Disciplinary gaps in knowledge.

Some saw ELSI researchers’ lack of genetic knowledge as a major barrier to true collaboration and thus to meaningful real-time ELSI. (This worry about lack of genetic knowledge extended to the CAB meetings where at least one clinician researcher was invited to every meeting in case the CAB members had questions about genetics.) Importantly, concern about the knowledge gap was not confined to clinicians, as ELSI researchers sometimes worried about it too. Clinician researchers, however, rarely seemed to worry about how much they did not know about ELSI or its components, knowledge that seemingly would have been relevant to many specific questions that arose.

The knowledge gap issue emerged early in the gene selection process. One of the clinician researchers who played a prominent role in that process described the problem in strikingly balanced terms during a meeting of the GSC:

We could spin our wheels forever. Right. And one of the reasons we could spin our wheels forever is that we have different domains of expertise, and we’re never gonna learn everything about each other’s domain of expertise.

But in a retrospective interview, another clinician researcher remarked about the GSC meetings:

It was very different from how I anticipated it being in that it was – it seemed to be more directed discussion from the “genetics experts” as opposed to a more collaborative effort. That being said, because of the time crunch it would have been difficult to educate the non-genetics professionals sufficiently on a broad array of genetic conditions to make it a more collaborative process. But I did sometimes feel that there were people who wanted to speak up that weren’t doing so because of the way the process was playing out.

In their interviews, the ELSI researchers varied in their self-assessments of the adequacy of their genetic knowledge. One commented, “I think [the ELSI researchers] did or could have easily had enough to contribute… in general I didn’t see a, you know, expressions of genetic ignorance or confusion about the genetics getting in the way.” Another ELSI researcher acknowledged, “I absolutely don’t have the adequate background knowledge.” But this same person then, without further prodding, went directly to the one-sided nature of the gap:

I feel like there is a disparity between the feeling of the humanities and social science folks that they really need to understand the science versus the science folks feeling that they really need to understand the humanities and social science nuances. So I feel like we spend a lot of time unpacking the really relative nuances of how to understand the genetic issues and very little time unpacking the relative nuances of how to understand the relevant philosophical or ethical concept. So that feels a little funny to me because we’re an ELSI [project].

From the perspective of several of the ELSI researchers, the clinician researchers’ prevailing sentiment seemed to be that you have to know genetics, which is hard, whereas ELSI is optional and easy. The ELSI researcher just quoted captured this feeling succinctly in the comment, “We’re fetishizing science.”

A related issue raised in the interviews was whether we might have overcome the knowledge gap by starting the project differently. Some researchers suggested that we could have begun with a genetics boot camp, perhaps a one-day intensive workshop in genetics conducted by the clinicians. There were varying views on whether this would have been feasible. No one proposed an ELSI boot camp.

Despite these worries, we cannot, in hindsight, identify a single instance in which ELSI researchers’ lack of genetics expertise slowed or impeded any aspect of the project. To take an example from gene selection, evaluating whether an intervention is effective or acceptable does not seem to require a deep knowledge of genetics. Such evaluations depend, rather, on a more general understanding of the concept of efficacy and a holistic, empathetic conception of acceptability—things that are core elements of the ELSI researchers’ competency.

The clinician researchers’ previous collaboration.

This knowledge gap was perhaps exacerbated by the existence of NCGENES, the earlier patient-screening project in which most of the GeneScreen clinicians participated. The selection of genes for GeneScreen was, from the clinician researchers’ perspective, largely completed before GeneScreen even started. As one of them commented in an interview, the chosen genes “seemed like no-brainers when we were talking about them for NCGENES.” The clinicians acknowledged that the transition to GeneScreen required little more than cutting down the much longer NCGENES list before presenting it to the GSC group. They believed that only genes that scored high on the NCGENES clinical actionability metric should be part of the study in order to increase the net benefit of screening a healthy population. In theory, selecting genes that scored high would minimize potential harms associated with screening genes for variants whose penetrance was not well established, did not have particularly effective interventions, or whose required interventions might be seen as unacceptable by the asymptomatic GeneScreen participants. Furthermore, they believed that the healthy GeneScreen participants would be less interested in genes of marginal relevance than their counterparts in NCGENES, who were sick patients. The clinician researcher just quoted characterized the transition from NCGENES to GeneScreen as “a really smart effort to leverage the infrastructure we already had in terms of analyzing genetic test information.”

Many of the ELSI researchers, who were largely uninvolved with the NCGENES process, did not appreciate this inevitability until the selection process was over. Nor did they appreciate the fact that this inevitability effectively mooted the larger research question of whether the screening was a good idea in the first place. The key working assumptions of the clinician researchers who brought the prospective gene panel from NCGENES to GeneScreen were not transparent to the entire GSC until much later in the process. The issue was not that the selection could or should have come out differently, but that the ELSI researchers’ participation was likely futile from the start. Perhaps ironically, it was a clinician researcher who described this reality best:

[The NCGENES selection group] was primarily composed of medical professionals… We had people from a bunch of different disciplines, but there weren’t really any active participants who were complete non-medical either lay people or professionals in other disciplines. I think that may be one way that really made the NCGENES decisions perhaps too firm when they came over to GeneScreen because it was a bit of, “Well, if the medical people have already decided that these are the things that are actionable, what can the non-medical people add to that decision?”

Since the gene selection was seen as largely inevitable, spending time bringing the ELSI researchers’ genetic knowledge up to speed for a decision they could not or should not influence was perhaps seen as a waste of time.

Value-laden differences in epistemology.

A point previously made—that genetic knowledge was perceived as necessary and hard to acquire, whereas ELSI knowledge was optional and easy—was repeatedly reinforced in the meetings and interviews by comments on the two disciplines’ standards for what counts as “knowing.” Overall, the dominant perspective of the clinician researchers favored the primacy of hard science and its positivist framework versus the qualitative imprecision of ELSI methods: scientists have hard data and proof while ELSI has only argument and anecdote. Ironically, though, while clinicians sometimes disparaged ELSI evidence as “merely” anecdotal, in every meeting of the GSC, one or more clinician researchers introduced an individual story, personal or clinical, to bolster an argument.

Some ELSI researchers resisted this hard/soft dichotomy. The ELSI researcher quoted above repeated the “fetish” trope, this time challenging the apparent precision of the scoring metric:

I’m not gonna say fetishism again… [T]here is no way I think that the process that lead to the numbers was objective in any sort of bench lab kind of sense. Right. I mean we weren’t just looking at protons.

Interestingly, as noted above, the one tangible success that ELSI contributors had in influencing the project—persuading the group that GeneScreen participants should not be limited to the insured—resulted from a presentation that had many of the hallmarks of scientific proof.

Related differences in professional culture.

The epistemological differences just discussed were paralleled by other differences in professional culture and values between the clinical and non-clinical (ELSI) worlds. Here, however, the effects of the differences were felt in both directions, with both clinician and ELSI researchers developing negative views of the other culture.

A specific manifestation of this cultural gap concerned whose opinion counted in the definition of acceptability in the scoring metric. From the ELSI researchers’ perspective, clinician researchers had an inexplicable reluctance to seek outside views on this issue. Even the obvious expedient of consulting clinicians from other specialties who treated patients with the relevant conditions was never tried. For example, one clinician researcher (following up on an earlier suggestion by an ELSI researcher) stated in an email, “We would like to have clinicians who take care of patients with these conditions come talk to us about what it’s like.” Others agreed with this suggestion during a subsequent meeting; one said, “It’s a different kind of perspective… it seems like it would be nice.” Prospective consultants were well known to the GeneScreen clinicians and in some cases were literally down the hall. However, those proposed collaborations never materialized, for reasons we never learned.

Difference in discussion styles were perceived negatively by both sides. One clinician researcher, for example, reported in an interview feeling uncomfortable with the sometimes adversarial nature of the ELSI researchers’ dialectic approach in meetings, sensing disrespect if not hostility in the occasionally argumentative back-and-forth. One of the ELSI researchers perceived this sensitivity among the clinicians, commenting in an interview that clinicians “aren’t as willing to get into it with each other.” Conversely, despite their willingness to argue ELSI-related points with each other, some ELSI researchers reported being uncomfortable with how the clinician researchers conversed. Specifically, they felt dissuaded from talking, and even personally offended, by the tendency of some of the clinician researchers to state positions authoritatively, if not dismissively. This may be an aspect of clinical discourse generally—that’s how doctors talk—but some ELSI researchers took it personally.

An ELSI researcher acknowledged that both sides bore some responsibility for the problem:

I think that a lot of it boils down to personalities in the sense that the bioethics group felt marginalized when it didn’t need to early on, and therefore it kind of got its nose out of joint and said “Well, we’re gonna be cynical and sneer at you, and roll our eyes from now on.” Rather than actually trying to work with the system and try to make it better… . it meant that there wasn’t a functional ELSI element in the discussion, and it wasn’t the scientists’ fault. It was our fault.

It is obvious that personalities are extremely important to the outcome of any collaborative project. What was especially striking in our clinical/non-clinical collaboration was the way in which professional norms and personal differences interacted with and reinforced each other, with an overall effect of limiting the potential of real-time ELSI influence.

The pressures of time and the grant.

The pressure of time quickly emerged as a major influence on the gene selection process and all the work that followed. Some of the ELSI researchers on the GSC had initially expected a more leisurely process with ample, if not unlimited, time to pursue issues. But the leader of the GSC introduced time pressure into the committee’s first deliberative meeting, stating, “in the next six months we need to come up with a list.” The leader (frequently backed up by the principal investigator) continued to stress time in various ways, both with respect to the need to move along within individual meetings (e.g., “we don’t want to talk the whole time about BRCA”) and the overall need to complete the entire process expeditiously. In the fourth deliberative meeting (out of six), four-and-one-half months into the GSC’s work, the leader summarized the time pressure as follows:

I think we’ve had really good discussions… We narrowed the list a lot, but in the end, and that end, the end is nigh. You know, we’re gonna have to sooner rather than later come up with a list because we gotta plan our assay. We’ve gotta, you know, budget it out and all of that kinda stuff … I would hasten to add that this is not an attempt to like railroad the, the process.

This time pressure was real, of course, given the timeline of the supporting grant and the technical factors cited by the leader. But it was also unexpected by some of the committee members, especially the ELSI members. Some of these people expressed—outside of the formal meetings—a feeling that the airing of ELSI issues had indeed been railroaded. In a later interview, an ELSI researcher summed it up this way:

It was fueled by a sense of urgency, by a sense of time pressure … We had trouble with pace, the pace of deliberations because it takes a long time to think through the implications of any particular piece, and it – we never had the luxury to do that.

The time pressure was driven largely by the schedule written into the grant. The six-month deadline for the gene list just referred to was dictated by the grant’s timeline, leading one of the project administrators to comment in a GSC meeting, “I have this tremendous tension of the time line …” And indeed there was a time line, with certain activities to be completed at specified times.

Discussion: Can Real-Time ELSI Succeed?

Our cataloguing of “successes” and “challenges” might imply that success has a self-evident meaning in this context, or at least “you know it when you see it.” That implication is a gross oversimplification, of course. Any effort to evaluate success is fraught with definitional peril, as well as unavoidably subjective.

With those caveats, we would nonetheless characterize GeneScreen’s real-time ELSI participation a success in terms of process. That is, we succeeded simply by being there, by coming and going as we wanted, and by voicing our concerns when and where we saw fit. In the words of one commentary on embedded ELSI projects, “ELSI researchers gain by being embedded in the trenches with the scientists themselves, rather than in the remote echo chamber of the ivory tower” (Meirom et al. 2016, 33). In GeneScreen, ELSI was indeed in the trenches, and—unlike in some of the horror stories in the ELSI literature—no one was fired, silenced, or shunned. It is a separate question whether and how the research project benefited from the ELSI presence. As ELSI researchers observed in their interviews, there is a difference between having a seat at the table and having a chance to exert tangible influence.

Outside of our real-time focus, ELSI efforts did succeed in other important ways. The ELSI researchers were leading participants in a systematic review of the harms-benefits literature concerning one candidate gene (Prince et al. 2017) and, most notably, had control over the design and analysis of participant interviews and surveys. ELSI participants thus exerted a primary influence on how harms and benefits would be defined and how the success or failure of the screening would be measured. (An initial report (Butterfield et al. 2018) on returning negative results reveals that GeneScreen disproportionately enrolled participants with an elevated risk for the conditions being screened and that GeneScreen’s communication about the nature of screening and meaning of negative results was inadequate.)

ELSI researchers also directed the CAB process, shaping the agenda and participating in each meeting. Interestingly, CAB members in their meetings echoed many of the concerns raised by ELSI researchers within the GSC and ECC regarding the propriety of genes on the panel, the acceptability of related interventions, the understanding of screening by various populations, and the time pressures imposed by the grant. As events unfolded, the real-time influence, or lack thereof, of ELSI researchers and CAB members was remarkably similar. On reflection, these similarities may warrant examination in future studies that aim to include both perspectives.

If, however, we assess real-time ELSI success in terms of substance, our assessment is far more negative. That is, if we ask, “was the GeneScreen project changed in any substantive way by real-time ELSI participation?,” the answer is no, with the single exception of the insurance issue. It seems in hindsight that the same decisions would have been made regardless of the ELSI researchers’ presence. In fact—playing directly into a widespread scientific stereotype of ELSI—the principal measurable effect that our presence had was probably to make things take longer.

But there is a counterpoint: perhaps ELSI researchers did not effect changes because changes did not need to be made. In the case of gene selection, maybe the NCGENES clinician researchers had gotten it right in the first place. In hindsight, the authors, as ELSI participants, could not argue against the inclusion or exclusion of any particular gene that was discussed by the GSC. As far as we knew, the choices were correct. But “as far as we knew” did not go very far. The problem was an oversight in process and communication: the premises employed by NCGENES in selecting genes were not apparent to the new actors (ELSI researchers) present in the GSC. Thus, while exploration of these premises did make GeneScreen more deliberative in the early stages, and particularly animated with respect to the concept of “acceptability,” the effect of that deliberation was only to confirm what the clinician researchers already knew.

In this reading of events, real-time ELSI succeeded, even if the larger project was ultimately undercut by technical failures that no one could have foreseen. And those specific failures might have been headed off had there been even more ELSI-style deliberation. An initial stance of devil’s advocate skepticism about—rather than enthusiasm for—new technology might have led to earlier scrutiny of the purveyor of the new technology that never was.

These considerations raise the further question of whether the ELSI researchers should reasonably have expected anything else. That is, GeneScreen was a gene screening project, and most of the key genetic issues had been answered reliably in NCGENES, so should the ELSI researchers not have expected to be relegated to a subsidiary role? There are two answers to this question. First, in our retrospective analysis of events, we have been repeatedly struck by the virtual absence of expectations on both sides of the clinical divide. Specifically, ELSI researchers had little sense of their appropriate role, and—regrettably in hindsight—did not take the time at the outset to define it, beyond the decision that we would be present everywhere. Clinicians, for their part, often seemed puzzled as to why we were there at all, given the apparently successful precedent of NCGENES. Second, to reiterate a point made earlier, given the framing of GeneScreen’s research questions, the ELSI researchers could reasonably have expected to debate the underlying question of whether the screening should be done, rather than merely fine-tuning a prepackaged project.

The fact that GeneScreen had so many challenges implementing real-time ELSI is troubling: if it was difficult to do in our circumstances, how likely is real-time ELSI to succeed elsewhere? GeneScreen was, after all, intended to be an ELSI project that involved genomic research, not the other way around. One conclusion to draw is that without intending to, the project shifted from real-time embedded ELSI to the more common model where ELSI follows a genomic experiment.

For others involved in real-time ELSI, we hope our experiences can prove useful. Studying the dynamics of GeneScreen revealed a tendency for team members to devalue the knowledge and skills brought to the table by ELSI researchers. Among both clinician and ELSI researchers, genomic knowledge seemed to hold more value. One way of addressing this is, as several members of GeneScreen suggested after the fact, to begin projects with a “genetics boot camp.” But we suggest also including an “ELSI boot camp” where different team members could explain how they approach ethical, legal, social, or policy issues associated with the project at hand. Even among ELSI researchers, skills and expertise tend to be exceptionally diverse, so all team members, not just clinician researchers, are likely to benefit from a boot camp designed to introduce, for example, substantive knowledge about ELSI issues and overviews of the legal and bioethics methods common to the field. Acquainting clinician researchers with these fundamentals could create a framework for the subsequent critical examination of ELSI issues. Although we did not encounter specific instances in which clinicians’ lack of ELSI knowledge led to an undesirable outcome, educating them about ELSI approaches might have forestalled or mitigated some of the epistemological and cultural conflicts discussed earlier. It also could have validated legal and bioethical methods of argument as acceptable companions to the discourse favored by clinicians.

Another lesson from the GeneScreen experience flows in the opposite direction: if ELSI researchers want to be persuasive to scientists, they may need to adopt some of the conventions of medical and scientific discourse. The one concrete ELSI success pointed to by many GeneScreen team members was the presentation addressing whether the uninsured ought to be eligible to join GeneScreen. The presentation proved persuasive for several reasons, including that the presenter had practical experience she could relate in the way that clinicians might relate medical case experiences—here, assisting clients with BRCA1&2 pathogenic variants to respond to health insurance companies’ denial of coverage for clinical follow-up. The presentation also used hard facts and relied on specialized knowledge, particularly of insurance laws. Finally, it argued a single point–that the uninsured should be included because having insurance provides no guarantee that follow up care will be covered—that specifically refuted the prior assumption of the research team.

Dealing with the time pressures inevitably associated with grant funding is a very difficult challenge. It points to a larger square peg-round hole issue: the problem of a scientific agency, with its positivist culture, funding—and thus imposing requirements on—an enterprise (real-time ELSI) that by its nature should be open-ended, loosely structured, and discursive. In this culture the assumption is that “success” is not measured by time spent thinking but rather by subjects enrolled—even in a study aimed to consider whether (as well as how) an adult screening panel might be employed in patient care. There is no obvious solution to this disjunction as even the nimblest ELSI researcher needs time to address dilemmas. One might suggest moving ELSI funding to an agency that funds research in the humanities and qualitative social sciences, where loose timelines are understood. But this would disconnect real-time ELSI from the scientific research enterprise, relegating it to outside observer status and thereby undercutting its defining characteristic. Again, there is no obvious solution beyond a probably futile plea for funding agencies to be flexible in imposing deadlines and benchmark requirements. Instead, funders and research colleagues must recognize that ELSI issues could require the same flexibility, in deadlines and benchmarks, as, say, a malfunctioning test or failing recruitment plan.

Conclusion

Our experience has led us to believe that real-time ELSI need not be contentious. On the contrary, it can be deliberative, collaborative, and even collegial. Scientific research projects with embedded ELSI researchers can—indeed, are likely to—contemplate ethical, legal, and social implications at greater length and in greater depth than they otherwise would have. Discrete elements of a project may be altered and improved. But is real-time ELSI likely to change scientific research in any fundamental way? Probably not, we conclude. To the bottom line: do the benefits of real-time ELSI exceed its costs? Our answer here is an unequivocal yes. The costs are both tangible (principally, the actual funding and the potential delays in carrying out significant research) and intangible (including mutual aggravation and ensuing mistrust between the two sides). But they are outweighed by the possibility of each side enhancing the other’s understanding of its goals, methods, and sensibilities, with long-term gains for both scientific research and ELSI.