How the National Health and Nutrition Examination Survey (NHANES) might go about returning genomic results to individuals was the subject of the next workshop session, with presentations by Kathryn Porter, National Center for Health Statistics; Kelly Edwards, University of Washington; and Barbara Biesecker, National Human Genome Research Institute. Wylie Burke, University of Washington, chaired this session.
Kathryn Porter began the session by presenting the NHANES view on the session topic. The NHANES program has considerable experience in reporting medically relevant results such as high cholesterol and high blood lead. When the program makes referrals to care providers, the providers typically know how to respond to study participants. However, there are some findings that health care providers may not know what to do with, Porter said. She offered the example of a study participant with a high urine arsenic level who receives a reporting letter saying that the level is high and that they should follow up with their provider. The provider then contacts NHANES and says “I am unfamiliar with this, what should I do?” This situation is difficult for the participant, for the doctor, and for NHANES, and necessitates a good deal of back and forth to properly address.
According to Porter, it seems that there would be similar issues with
regard to reporting gene variants. NHANES might proceed with developing a reporting letter to participants that they could share with their health care providers, but the providers are likely to be on tight visitation schedules. The sudden appearance of a Centers for Disease Control and Prevention (CDC) letter detailing an actionable variant and the need to do something is likely to create problems, Porter suggested. NHANES may need input from primary care providers on this concern. Further, given the importance of participant preferences that was discussed during the workshop, she said it is important to note that the NHANES program has never done any studies or surveys of its participants’ preferences.
Porter explained that the survey program has considered several plans for returning results, one being a retrospective plan dealing with people who were previously consented. NHANES would need to develop procedures for returning results given that the prior consents stated that no results would be reported. Then, she added, there are the prospective considerations; there currently is a hiatus on obtaining permission to store specimens, and new consent language would be required to move forward.
With regard to the retrospective plan, she said NHANES received documented consent between 3 and 23 years ago for a sample of 26,000 adults. In the earlier years, the consent was only to store blood for future testing. There was no mention of genetic studies, and no return of results was stated on the consent form. NHANES has not resolved the ethical and logistical issues of reporting clinically actionable or medically actionable results years after DNA samples were collected, including how this would feel to a participant. She said the question remains about whether the results would be valid, although she noted the consensus seems to be that they would be if genetic tests were done in CLIA (Clinical Laboratory Improvement Amendments)-certified laboratories.
There is no active tracking of past survey participants, Porter explained. Records include their last mailing address, street address, and phone number. Some NHANES data have been linked to mortality files to ascertain who is still living. The survey program could get a listing of those participants still alive with their last best-known address, she noted, and one possibility is to contact them again by mail. This would not be a re-consent but rather a notification, she clarified. There might be a multistage process, the first step being to let participants know about procedural changes. Next, if there were results or developments that participants should know about, they would receive a letter. There could be subsequent phone contact to discuss specific results and provide counseling.
Before undertaking such an approach, Porter said the NHANES program would first have to determine a universe of clinically actionable
gene variants and reassess them annually. The assessment would require recruiting and convening a medical advisory panel of genetic clinicians, research scientists, bioethicists, and possibly genetic epidemiologists. If the DNA bank is reopened to new research proposals, Porter said, the survey program would need to obtain full-time genetic counseling expertise to support operations, enhance its computing infrastructure, and address a variety of language-related issues for participants who do not speak English or Spanish. Everything, she stressed, would be contingent on having a sound reporting protocol. NHANES would not want the potential research proposals to drive the reporting plan, and therefore would require that the reporting plan be developed prior to considering research proposals.
Porter described two plans that have been considered going forward in time. The first involves restarting the collection and storage of DNA specimens, with appropriate modifications of the consent process and perhaps a menu of participant choices. The second would not involve banking specimens, but would consider whole-genome sequencing as part of the live NHANES data collection. This is wishful thinking at this point in time, she observed, but may make the most sense for recruiting participants who would be told they are going to have a physical exam, a bone scan, an oral health exam, have blood drawn, and also some genetic tests, and that they will receive all the findings. Porter said this plan would be very expensive, bureaucratically difficult, and likely require major funding partners to help with lab logistics and staffing requirements.
Kelly Edwards pursued the differences in the return of results between retrospective and prospective collections. She framed her remarks in terms of what is guiding behavior today by invoking the idea that people are governed by laws, technologies, markets, and norms, and asking which of these is moving faster and slower. At least in the area of genomics, she said, it seems clear that technologies and markets, and in some case norms, have moved past current legal structures and regulatory processes. Edwards noted that researchers and practitioners are running to catch up and running to build a common infrastructure and standards of excellence, treating the regulations as a floor for behavior.
In Edwards’ view, it is important to focus on participant expectations when talking about retrospective and prospective data collections, and on managing the expectations that people have about participation. When thinking about how participant expectations arise, she noted many people who work with ethical considerations and with institutional review boards (IRBs) think that expectations come from consent forms. A
good deal of effort goes into creating and finessing those forms, Edwards observed, and the creators think they are conveying what participants need to know and will understand. However, as known from a wide range of social science, people’s expectations about what is going to happen in the course of a project come from many different sources: their own experience with past projects, assumptions about what might happen, what they hope will happen, and sometimes from misunderstandings or miscommunication.
Edwards pointed out that the workshop has highlighted the issue of how research activities are different from clinical care or public health activities and has considered the trade-offs regarding the obligation to return results. She suggested that the debate over obligations for and obligations against a duty to return results may have reached a saturation point, and what is most important are the expectations and values guiding a particular data collection that will help to make a decision going forward. She highlighted three mechanisms for getting input from community members on research initiatives.
First, community advisory boards are the classic mode by which large research initiatives obtain community input on policies. She referred to guidance on the CDC Website that addresses the various challenges with community advisory boards: How can projects develop a representative collection of participants who can provide guidance? How can projects help their community advisory board be informed enough about the issues without co-opting them into the research enterprise? How can boards maintain enough perspective to provide fresh participant perspectives? How can projects work with community advisory boards so that boards might be able to speak more broadly than just for themselves?
A second, different kind of engagement that Edwards discussed is the process of intensive community deliberation, events that involve representative groups and engage them deeply over a period of time. This process has been used successfully in California to help generate biobank policy, with community participants giving feedback directly to biobank owners and researchers and influencing policy on the fly.
Edwards noted that a third approach is beginning to emerge. Instead of asking a representative board for input, individuals are asked to directly develop and manage their own preferences. The Genetic Alliance’s Platform for Engaging Everyone Responsibly (PEER) is an example, using private access technology. She said such personalized tools allow individuals to set their own preferences about data sharing, data access, re-contact, and communication about returning results.
As an example of building a governance process, Edwards described an Institute of Medicine ([IOM], 2012a) study that was tasked with developing suggestions for how to deal with an historic collection of 90 million
specimens dating back to the Civil War that was housed at the Armed Forces Institute of Pathology. She explained that it is an existing collection, the samples had been collected for clinical pathology purposes, and the still-living participants do not know that their samples are in the collection. The issues before the IOM committee were whether this collection can be used for future purposes, including research, and if so under what circumstances. Re-consent was impractical given the size of the specimen collection, and the question was what else could be done.
She explained the proposed solution involved a governance process that focused on ways of doing as much de-identified research as possible, and on the establishment of a third-party review system for all data access requests, including a scientific advisory board and data access committee. The IOM committee did not look in depth at the return-of-results issue, but did propose the idea of an entity similar to the Data and Safety Monitoring Board, an independent group of experts that advises the National Institute of Dental and Craniofacial Research and its study investigators. The IOM committee envisioned an entity that would have a mixture of stakeholders, including primary care doctors, study participants, and researchers. This group would look at requests involving return of results on a case-by-case basis and make determinations about whether a given result rose to the level of actionability and how to go about disclosing the result. The entity would be independent of any investments of the data repository or investments of the research itself.
Because re-contact and re-consent were so impractical, she said the IOM committee emphasized Internet communication and other tools to get the word out about this resource and to make as much publically available as possible in terms of aggregate findings and ongoing research activities. It was also deemed important to make it as straightforward as possible for people to opt out of this activity if an individual or family member wanted to remove a sample. The committee recognized that all of these governance plans needed to be under constant review and responsive and dynamic in light of changing norms, laws, and technologies.
Barbara Biesecker addressed several practical aspects of returning genomic results from NHANES. The main issues she discussed involved how to consent people, how to help people make informed choices if they want to learn information, the ways in which results are returned, and how NHANES might help participants act on the results.
As Edwards had noted with regard to consent, people enter into studies with their own mental models, prior experiences, and beliefs, and therefore the consent itself probably has little to do with their decisions.
Biesecker commented that people can appreciate uncertainties and the tentative nature of information, and they do embrace the idea of multiple causes. An obvious question is whether or not NHANES participants want genomic information, and she said the answer is “we do not know because we have not asked.” Research done with different cohorts at the National Institutes of Health (NIH) suggests people do indeed want genetic information, Biesecker pointed out. For example, ClinSeq participants can discriminate different types of information. They are most interested in actionable variants and carrier results, but are almost as interested in information that they cannot do much about (Facio et al., 2013). Parents of children enrolled in NIH studies are also eager to learn about secondary variants (Sapp et al., 2014).
Biesecker asserted that this is an era of advocacy rights for patients and citizens, an era of partnerships and transparency in the delivery of health care. She said people increasingly desire better access to their medical records, and, in general, place increasing value on staying healthy and taking personal responsibility for health. Data suggest that if information about individual health becomes knowable, people want to know it. If they are not sure they want to know it, they at least want to be offered the choice.
Questions have been raised about people’s overall ability to learn about or understand genetic information. Hence, she said, one might ask if NHANES participants will internalize the genetic information that may be provided. Early research from other studies suggests more likely than not, she said, as the personalized nature of the results makes it likely that people will have further conversations about interpretation. It does not take a great deal of time for people to sort out what they need to learn, Biesecker suggested. They do not need to learn about DNA composition and where chromosomes are situated, but will seek out practical outcomes, the estimated chances of getting an illness, what could potentially be done to prevent or mitigate the risk, and what specialist they need to talk to in order to come up with a plan. Participants may learn about certain findings as effectively from a Web-based platform as from an in-person consultation, she commented. An ongoing large randomized control trial to return carrier results from ClinSeq is now under way to understand differential outcomes when participants are randomized to a Web-based platform versus a genetic counselor who provides education.
Biesecker noted that it is still unclear if NHANES participants will likely use the information they receive. Respondents are likely to tell their providers, who can help ensure follow-up on their recommendations. However, as she pointed out, it is difficult to induce people to change their health behaviors even with evidence-based interventions. As of 2014, she said, many people likely need the interpretative help of a genetic coun-
selor or someone trained by a genetic counselor or medical geneticist in order to understand genomic information. Counseling usually is done in person, and if a project like NHANES were to incorporate genetic counseling, the cost implications would be considerable. That said, Biesecker mentioned that better follow-up can be integrated into the general health care system, and more innovative use of social media and other Internet resources can assist people to take control over their information and not leave matters to a specialist or sub-specialist.