8

Key Themes and Considerations for Future National Health and Nutrition Examination Survey (NHANES)

This chapter summarizes the recap provided by Jeffrey R. Botkin, University of Utah, that took place immediately before Session 6 on December 8, 2022. Botkin also moderated the 1-hour panel discussion that took place after the session and included questions provided by the sponsors, planning committee members, other workshop participants, and the public webcast audience.

RECAP OF WORKSHOP DAYS 1 AND 2

Botkin, chair of the 2022 workshop planning committee, provided a summary of what was learned in Days 1 and 2 of the workshop. He highlighted several key elements of the National Health and Nutrition Examination Survey (NHANES) project that the planning committee felt would impact decisions about adding a genetic component.

Botkin started by noting that NHANES has provided essential information on population characteristics for citizens of the United States and has been a “tremendously valuable public health project” for many decades. He explained that because much of what is known about genetics thus far has been based primarily on individuals from European ancestry, NHANES offers a unique chance to collect genetic data from a representative sample, which is a high priority for the National Human Genome Research Institute and other entities. Turning to some of its challenges, Botkin highlighted NHANES’s relatively small sample size of about 5,000 individuals per year. “So you accrue quite a few people over time, but [NHANES’s sample size is] relatively small compared to some other projects that are ongoing

in other domains,” stated Botkin. A second challenge is that NHANES is cross sectional by design and therefore does not include any longitudinal follow-up, meaning there is no opportunity to track and recontact individuals over time.

Botkin reflected that the workshop speakers had articulated a change in attitudes over the last 15 years, such that there is currently a strong ethical obligation for investigators to return actionable variants. He underscored that returning genomic results to participants is a complex process; therefore, projects today must carefully consider whether and how to return results from the outset.

There are relatively well-defined methods for defining actionable variants, said Botkin. He noted that while there is some variation on what is considered actionable, there are currently justifiable and widely used tests of variants. One particular challenge of genomics is that the variants that are considered to be actionable change over time. “So in that way, genetic results can differ from other sorts of analyses because interpretations are somewhat fluid, in contrast to the other types of testing that NHANES does and has routinely returned to individuals as part of its study,” explained Botkin.

Botkin noted that regulatory issues must be attended to, such as Clinical Laboratory Improvement Amendments certification and validation. “FDA [Food and Drug Administration] has also been actively involved in a number of larger studies to return genomic results, genetic results, and so it’s likely that a collaboration with the FDA and approval there is going to be necessary if a genetics component is added to the study with return of results,” summarized Botkin.

Botkin explained that this workshop has highlighted research on language that helps people understand the nature of the choices that they are being offered, but understanding of the informed consent process in general remains a huge challenge. He highlighted that emerging data and understanding of the consent process can assist comprehension; for example, methods like teach-back and emerging sophistication with the eConsent process can give people access to additional resources, illustrations, and dynamic consent processes that can help with understanding.

Individual projects and studies, Botkin indicated, vary in the degree to which participants are offered the ability to decline genetic testing and are given a separate choice to decline a return of results. Some projects offer an opportunity to decline up front, while others offer a second opportunity if an actual variant is defined. Some projects exclude individuals who are not interested in genetic testing and results, while others do not.

Finally, Botkin highlighted challenges with individuals’ abilities to translate individual genetic results into clinical use. He noted that the nature of the health care system in the United States combined with the national

breadth of NHANES present unique challenges with respect to assisting participants in acting on clinically actionable results.

PANEL DISCUSSION

Botkin opened the panel discussion by reminding the audience and contributors that workshops do not provide recommendations, but individual contributors are welcome to offer their ideas. He also highlighted that this workshop is serving as one step in a much longer process of thinking through the many considerations to determine whether a genomics component can be added to NHANES and how that could be operationalized. Botkin highlighted that one key issue NHANES will need to consider is whether there needs to be a confirmatory loop in the process and what that will look like.

Alan Simon, Centers for Disease Control and Prevention’s (CDC’s) National Center for Health Statistics reminded the audience that NHANES is currently a cross-sectional survey that is not embedded in care. He provided this additional guidance and context: “There are limitations on what we can do, and with that, in thinking about how to go forward, it’s important for us to sort of break up the questions a little bit that we’re asking, I think, into the initial collection of sample when I know what I’m going to do with it immediately, what tests I might do on that right away, and I can consent you for that right away, versus the situation where I’m collecting sample and I may not know what’s going to be done on that in the future.” He highlighted that in both scenarios, one of the first decisions NHANES will need to make is whether it should be looking at actionable variants or only nonactionable variants.

Botkin explained that the planning committee was asked to focus on the primary survey (NHANES 2025) and not the biobank. Based on the limited discussions that the planning committee did have about the biobank, he saw those as “a somewhat different set of considerations.” Botkin added that the current strategy of not allowing investigators to produce actionable findings does limit the science “in ways that may be important.” He noted that deidentifying or anonymizing the samples and data is one approach to the problem, which raises some practical technical and scientific issues in terms of the value of the research going forward. Botkin said: “So it seems to me that to promote science, it would be good to just rethink some of those issues and traditions that NHANES has about the biobank and whether the new tools that are out there might lead to some better science with the samples that you have, understanding that the return of results then becomes a significant challenge there.”

Ingrid A. Holm, Harvard Medical School and Boston Children’s Hospital, asked Simon if there was any consideration of adjusting the

survey’s current structure to, for example, include a longitudinal component. Simon replied by first noting that genetics is one possible component of NHANES and it needs to look at the full set of components when thinking about structural changes. Simon said: “With that said, is there some feedback loop or information feedback that could happen on some ongoing basis with people, I don’t know…. It’s worth thinking about would it open up scientific doors that are really, really important, and then maybe we could see what is possible.”

Leslie G. Biesecker, National Human Genome Research Institute at the National Institutes of Health, observed that every study highlighted throughout this workshop had a longitudinal component. He added: “It is in the era of longitudinal dynamic engagement with participants. That is absolutely where the state of science is.” From Biesecker’s perspective, the big question is whether genomics is important for “nutritional population research” before deciding on any structural changes.

Simon responded by offering his view that there are different flavors for what interaction with participants looks like over time. He asked: “What’s ethically acceptable? Would a sort of opt-in return of results be acceptable? Is it acceptable to provide results but not give a super warm handoff to a geneticist?” He noted that there is a lot of grey area about what might be acceptable, which is part of the struggle for NHANES.

Botkin moved to the topic of community advisory boards or some other community group. “If you decide to go in the genetics direction, it would seem advisable to have that sort of interaction, moreso I think than what your tradition has been,” said Botkin. Simon replied that has been helpful feedback and is something they can see value in.

Holm next raised the question, “Why is genetics so focused on the longitudinal piece?” Biesecker responded that in genetics one thinks about three things: genes, phenotypes, and environment. He noted that genotypes change the whole game because unlike phenotypes and environmental factors, “That [a gene] is a stable attribute of an individual and the only way to make use of it and to efficiently exploit that dataset is to use it over time, because it’s stable, and you’re paying a lot for the data and the way you get value for what you pay is to use it over time.”

Dana C. Crawford, Case Western Reserve University, offered her observation that “what we are hearing here is a fundamental conflict maybe in purpose.” She recalled that NHANES is very much rooted in the public health history of monitoring trends over time and seeing to what extent public health interventions impacted measurements in the population. By comparison, genetics and return of results are very much at an individual level. “So I think it’s really that difference in the study design and the purpose of how NHANES was originally conceived and conducted over time, and maybe it’s time to revisit that if genetics is to be introduced, because I

really do think it’s an individual-level impact now,” said Crawford. Adam H. Buchanan, Geisinger, suggested that genetics can be quite helpful for looking at a population-level impact over time. Crawford agreed, noting that because some of the genetics are lower frequency, it will take more time to see those impacts, and they may not be as striking as results NHANES was able to show in the past (e.g., changes in lead levels).

Natasha Bonhomme, Expecting Health, offered her observations that members of the public are in the process of setting their expectations of what they think they may be getting from genetic information based on what they see in the media and through personal experiences. She noted that what people know and expect and how to communicate with them is changing each year and that evolution is “happening quite quickly.”

Biesecker highlighted that he and other contributors are somewhat unique in their specialized focus on the topic of return of results. “For most people, return of results is not central to the science that they’re doing, but it’s … what some of us feel is an ethical obligation, duty to rescue, when somebody has something really evil and you know about it, you should help them out,” explained Biesecker. He reiterated his view that the scientific needs and interests ought to be the driver determining whether genomics is included in NHANES; and the consideration of an obligation to return results ought to follow that.

Botkin added that NHANES is well positioned to look at a particular population risk factor that can be detected using genetic testing, looking at the prevalence, changes over time, and interventions. He explained that, from his perspective, a lot of the scientific questions are still focused on phenotype–genotype correlations. “NHANES is not going to give you that kind of data. It’s just not big enough and there’s no longitudinal followup,” stated Botkin.

Simon explained that a key question for NHANES is to what extent it is observing changes in genetics in the population as the population shifts that are responsible for changes in specific risk factors and outcomes. He offered pediatric obesity as one example of something that NHANES is tracking; it is not yet known how much the environment, versus genetics, is contributing to that phenomenon. Botkin noted that one would need to know a priori of a genetic variant that is associated with pediatric obesity. Simon noted that this line of inquiry would not be NHANES’s primary purpose.

Holm observed that NHANES is thinking more about polygenic risk scores and that type of genetic testing. Simon offered his opinion that NHANES would be best positioned to look at polygenic risk scores in addition to nongenetic risk factors, but that it has not been able to do polygenic risk scores to date. Buchanan spoke up to say that, based on this conversation, it does seem like polygenic risk scores plus nongenetic data are worthwhile places for NHANES to focus.

Buchanan next recalled that there are some acutely important things that can be reported with genetic testing. He suggested that Genomics England and the principles about near-term actionability of the information could be useful to NHANES for figuring out what genetics to capture and test, because in a cross-sectional study “it’s that near-term impact that you’re going to be most interested in, because you’re not going to have the ability to follow long-term and so forth.” He noted that genetics can be used to determine individuals who are at risk for a particular condition with a significant risk of sudden death. He added that “the newborn screening context where you want to know something sooner” may be a good fit for a particular context such as that of NHANES, where there is limited contact with people and therefore one could prioritize those things that are most important in the short term.

Audience member, geneticist, and director of the CDC’s Office of Science Muin Khoury offered more context to the discussion. He shared that his team would be submitting proposals for the retrospective analyses of the 20,000 banked specimens. He provided what he saw as a more fundamental reason why genetics should be added to NHANES: “as you monitor trends of risk factors over time or trends of prevalence of disease over time, we’re treating the population as a one-size-fits-all.” However, he next noted that the prevalence of a given risk factor or condition may vary by genotype. Khoury shared that the struggle has been with single gene disorders because only about 1 percent of the population will have a result that needs to be returned. He asked the group if there is a way to return results to 1 percent of NHANES participants within the current 60-day parameter NHANES operates within, and if so, what sort of infrastructure, pipeline for informatics, and genetic counseling is needed to do that.

Jason Vassy, Harvard Medical School and Veterans Affairs Boston Healthcare System, responded first by recalling that for sexually transmitted diseases, NHANES had a special carve-out operation for returning those results, which might be a good parallel for the return of any genetic information. Biesecker offered his opinion that any study could absolutely return a restricted set of genes to target 1 percent of the population in a time-sensitive manner. “It’s a question of if you want to do that, if you can marshal the resources,” he said. Biesecker pointed out that before making that decision, one must answer the harder question of “why do you want to do that?” Crawford responded by returning to a scenario where one goal may be to use genome-wide data to stratify according to such factors as polygenic risk scores and monitor those trends within strata over time. She pointed out that in that scenario, there is an ethical obligation to report.

Botkin steered the conversation to the issue of a study’s ethical obligations to report results to people who have important findings. He said that sending a letter to people telling them to talk to their doctor about this may

not be adequate in this day and age. “I understand the complexity of dealing with lots of health systems across the country, lots of people who are uninsured, et cetera, don’t have clinicians, but are there ways to be creative and think about partnering with local health departments, the local professional communities, in ways that would allow those results to be more assuredly in the hands of people who are going to help the participants out?” asked Botkin.

Holm responded first by sharing that she has a grant to use eMERGE to look at the impact on health care providers of receiving unexpected data from research. “It’s anxiety provoking for health care providers, because they don’t really understand it, then they need … to manage it, and they need to know someone to go to to help them out [sic],” she explained. Holm continued by saying that there seems to be more empirical evidence that underscores this same point. She offered that the All of Us Research Program’s efforts to deliver information to health care providers effectively may not be adequate.

Botkin asked for the panelists to offer a feasible way for genetics professionals to engage with NHANES and outline what the role could be for those specialists. Biesecker responded first that the genetic expertise would be needed up front to carefully design the research questions, as well as potential pipeline issues with interpretation. He explained that “you need people with that expertise, which you can sort of beg and borrow to help design it, but once you want to implement it, then you need to have people on the staff and in the program who know how to do this from the very first step to the very last step, and that includes a range of genetic and genomics expertise: genetic counselors, some medical genetics consultations, certainly laboratory and technical genetics and genomics people, statistical geneticists.” Holm agreed with Biesecker and returned to the model being used by Genomics England, in which people who specialize in the given condition return the information to families. She reiterated a common theme in the workshop that there are not enough genetic counselors available in the United States to meet the demand.

Buchanan shared that Geisinger has been involved in triaging the use of genetics experts. “But one of the other striking things to me about the conversations was in addition to genetics expertise, there are other types of expertise that really have been useful for a lot of the programs that we heard about, communications experts and health communications experts, for example,” recalled Buchanan. He found the information about user testing of a study’s genetic components to be very instructive and thought that could be relevant for the NHANES context. Botkin agreed with Buchanan, noting that some of the consent and related forms shared during this workshop “are much more contemporary in the way of helping people understand what you really need to know to make a choice.” Holm agreed that

the All of Us Research Program is a good model for explaining information about genetics in lay terms.

Botkin next asked the panelists to reflect on the extent to which tiered consent and choice about receiving genetic findings makes sense in the NHANES context and to what extent such choices could impact the representativeness of the data. Holm noted that several presentations and discussions pointed out that people generally like getting results back. Biesecker agreed with Holm and remarked that it is a question of biases for or against receiving findings that may be baked into the consent process and whether or not those biases are compatible with the scientific question being asked. He noted that it is more resource intensive to offer a choice.

Simon clarified that NHANES does allow people to say no to any part of the exam or interview that they do not want to do. He reinforced that the issue of representativeness and the impact of offering choices is a major ongoing concern for NHANES. He added that NHANES does honor the very few requests made by people who want to participate in the study and have their specimens collected and analyzed, but do not want to see their results, noting that this scenario is covered in the current form.

Before closing the panel discussion, Botkin shared a comment from a webcast participant that it may be worth looking at the history of the NHANES epidemiological follow-up study, which followed people from NHANES from 1971 to 1980, to inform the question of whether a longitudinal NHANES is worthwhile.

Simon asked if panelists could comment on other barriers they have encountered, such as securing appropriate staff. Buchanan responded that he feels “it is important to make the resources equitably available and closing the gaps for care in a way that is consistent across the board.” Biesecker noted that genomics tends to be a very collaborative discipline and warned against trying to rely on a single expert to handle all the facets.