3

The Intersection of Health Literacy and Precision Medicine

The workshop’s first panel session featured three presentations that provided a further grounding for the rest of the day’s discussions. Sara Van Driest, an assistant professor of pediatrics at Vanderbilt University School of Medicine, described what precision medicine is and how it has evolved over time, while William Elwood, the coordinator of the National Institutes of Health (NIH) Basic Behavioral and Social Science Opportunity Network in the NIH Office of Behavioral and Social Sciences Research (OBSSR), did the same for health literacy. Michael Wolf, a professor of medicine and the director of the Health Literacy and Learning Program at Northwestern University’s Feinberg School of Medicine, then laid out some of the issues at the intersection of these two disciplines. A discussion, moderated by Ruth Parker, a professor of medicine, pediatrics, and public health at Emory University School of Medicine, and joined by Joseph McInerney, followed the three presentations.

PRECISION MEDICINE AND ITS EVOLUTION¹

NIH defines precision medicine as an emerging approach for disease treatment and prevention that accounts for individual variability in genes, environment, and lifestyle for each person, Van Driest said. While the words genes, environment, and lifestyle may be the buzzwords found in the press

___________________

¹ This section is based on the presentation by Sara Van Driest, assistant professor of pediatrics at Vanderbilt University School of Medicine, and the statements have not been endorsed or verified by the National Academies of Sciences, Engineering, and Medicine.

releases, interviews, and grant applications, she said that she considered the key phrase in that definition to be “individual variability” because it represents the difference between precision medicine and medicine as usual.

Precision medicine is not new, and, in fact, it is already in play for some diseases, Van Driest said. The goal today, she said, is to expand the number of diseases to which precision medicine is applicable, increase the number of physicians who practice precision medicine, and grow the number of patients who can benefit from this approach. While she quoted William Osler as saying, “It is much more important to know what sort of patient has a disease than what sort of disease a patient has,” to show that the basic idea of precision medicine is more 100 years old, she credited NIH director Francis Collins for the current emphasis that precision medicine is now receiving. Collins, when interviewed in 2015 about the Precision Medicine Initiative (PMI) and getting his own genes analyzed, said the analysis revealed he had an elevated risk for developing type 2 diabetes, which motivated him to lose weight, exercise more, and eat a healthier diet (Collins, 2015).

In her overview of precision medicine, Van Driest divided the subject into precision therapeutics, diagnostics, and prognostics—a division that she acknowledged was artificial. Concerning precision therapeutics, she said that while physicians treat one patient at a time, they decide on the treatments for those patients they are treating based on data from a population, such as the patients who had enrolled in a clinical trial to test the efficacy of a drug. Precision medicine represents a change in approach that looks to find markers in an individual that predict how that individual will respond to that drug. The question that doctors ask changes, Van Driest said. “Instead of asking which treatment is best for this disease,” she said, precision medicine “asks which treatment is best for this patient.” Many factors go into answering that question, she said. For example, drug effects can vary from patient to patient, reflecting such factors as the patient’s age, sex, and ancestry; interactions with other drugs the patient takes; the environment; the exact diagnosis; and the patient’s genetics, in particular as they affect drug absorption and metabolism and the individual’s response to a particular drug.

One example of how precision medicine is being practiced today is the prevention of a second heart attack following the implantation of a stent to restore blood flow through a blocked coronary artery. After stent placement, the patients are prescribed an anti-platelet drug, such as clopidogrel, to inhibit clot formation. Every patient is treated with this type of drug because there are good population-based data showing the benefits of taking clopidogrel after receiving a stent (Sabatine et al., 2005). However, Van Driest explained, the benefit that a specific patient receives from clopidogrel depends on whether or not that patient has a specific variant

of the gene CYP2C19, which codes for a liver protein that metabolizes a number of drugs, including clopidogrel (Mega et al., 2010; Wallentin et al., 2010). Patients with certain variants of this gene are unable to metabolize clopidogrel and do not realize full clinical benefit from the drug. “Knowing this information about the genotype can help the selection of appropriate medicine because someone who has that high-risk genotype should get a different drug,” Van Driest said. Her institution now tests for this specific genetic variant and uses the results to help the clinician and patient decide on the proper course of treatment. She explained that the presence of the high-risk genotype is not a “hard stop” because other factors, such as cost of the alternate drug are also important.

Van Driest said that this type of decision support increases the rate of using alternatives to clopidogrel among people who have the high-risk genotypes (Peterson et al., 2015). While this is an interesting story, she said, the important question is whether people getting the alternative therapies are living longer, healthier lives. “We do not have that answer yet,” she said. “When you start subdividing your population by genotypes, it takes many thousands of people to see differences in results, and we have not done this long enough yet to know whether we have a real clinical benefit for patients.”

Another example involves the use of precision diagnostics to determine the appropriate therapy for malignant melanoma, an aggressive form of cancer. In the past, Van Driest said, physicians based their treatment decisions on the microscopic features of a patient’s specific melanoma. A new approach to classifying melanoma instead characterizes the genetic changes that drive the cells to be malignant. Using this approach, approximately 40 percent of malignant melanomas are found to have specific variants of a gene called BRAF, a discovery that led researchers to develop drugs that inhibit BRAF. For those melanoma patients with the clinically relevant BRAF mutations, BRAF inhibition produces dramatic positive results (Wagle et al., 2011), but as Van Driest explained, that is not the end of this story. Other cancers, such as hairy cell leukemia and certain thyroid cancers, also have these mutations. It also turns out that in most cases at least a few cancer cells evolve a way to circumvent BRAF inhibition, and the cancer returns. “Currently, these BRAF inhibitors are life extending but are not curative,” she said. “The goal is to understand more of these molecular mechanisms so we can develop a curative therapy.”

Van Driest then turned to a third area of precision medicine—precision prognostics. In the case of Francis Collins, he did not need tailored therapy for type 2 diabetes because he changed his behavior based on his genetic predisposition in order to prevent the disease from developing in the first place. Another well-publicized example of precision prognostics leading to action is the decision of actress Angelina Jolie to have her breasts and

ovaries removed as a preventive measure after she was tested and found to have a BRCA1 variant.

Unfortunately, identifying genetic variants that can be used to inform precision prognostic is challenging because the genome is full of what Van Driest called “red herrings”—variants that look as if they should cause disease but perhaps do not. For example, she and her colleagues looked at two genes associated with arrhythmia in 2,022 individuals and found 122 rare variants that Van Driest said should cause problems. However, when the data were sent to three expert laboratories that were asked to determine which of the variants would be pathogenic, they agreed on only four of them (Feero, 2016; Van Driest et al., 2016). Then, when Van Driest and her colleagues examined the electronic medical records for these individuals, fewer than 5 percent of the individuals with these supposedly harmful mutations had any signs of arrhythmia in their electrocardiograms, which is the standard diagnostic test for arrhythmia. The conclusion that one of her colleagues drew was, “At a minimum, the language for describing variations’ predictive ability should be carefully calibrated to convey, when appropriate, a probabilistic, rather than a deterministic, nature” (Feero, 2016).

Looking to the future, Van Driest said that the precision medicine community needs to start thinking like geographers, who, as she described it, are able to take different types of data, layer them, align them given the borders they know about, and draw useful inferences about the way the world works. Precision medicine is being fueled by the emergence of new types of data that are broader and deeper in their information content, and the field needs to somehow layer those data in a way that allows them to be integrated and provide comprehensive recommendations that while still probabilistic, rather than deterministic, will contribute to longer, healthier lives.

HEALTH LITERACY AND ITS EVOLUTION²

In 2000, Elwood said, health literacy was defined as the degree to which individuals have the capacity to obtain, process, and understand basic health information and services needed to make appropriate health decisions (Ratzan and Parker, 2000). By 2014 that definition had expanded somewhat to become the degree to which individuals can obtain, process, understand, and communicate about health-related information needed

___________________

² This section is based on the presentation by William Elwood, the coordinator of the NIH Basic Behavioral and Social Science Opportunity Network in the NIH OBSSR, and the statements have not been endorsed or verified by the National Academies of Sciences, Engineering, and Medicine.

to make informed health decisions (Berkman et al., 2010). The important feature to notice, Elwood said, is that health literacy is not simply about medical decisions. It involves myriad other issues such as knowing how to take a medication, how to be fit, how to monitor one’s health on a smart phone, and how to use proper ventilation when cooking to avoid exposure to particulate matter and being able to discuss with a health care provider how the environmental effects of living near a freeway can effect personal health.

Since 2004, OBSSR has led three special program announcements on understanding and promoting health literacy. NIH has funded and administered more than 500 investigator-initiated grants aimed in this area, Elwood said. These programs, he added, have generated more than 9,600 health literacy publications indexed in PubMed. The health literacy literature operationalizes and places this topic in clinical, personal, and health-specific settings to demonstrate that while health literacy may be rooted in an individual’s capacity to obtain, process and understand health related information, such capacity is constantly in flux given a person’s milieu, the groups to which a person belongs, the type of setting, and the wellness and disease issues that person faces at any given moment. “A person’s health literacy is dynamic,” Elwood said.

When one reviews the health literacy literature, Elwood said, several themes emerge, including

• Disease-specific contexts such as Alzheimer’s disease, diabetes and its sequelae, cancer or HIV treatment adherence;
• The setting, context, and situation in which health information is exchanged, such as in medical clinics, emergency rooms, pharmacies, and schools;
• Literacy in target population groups, such as African American health literacy or Native American health literacy; and
• Wellness issues, such as food portion control, sufficient weekly physical activity, age- and time-appropriate vaccinations, and timely cancer screenings.

The literature also emphasizes communication between individuals and through mediated means such as family discussions, shared decision-making processes, and negotiation processes to obtain informed consent or research participant consent.

Health literacy is starting to benefit from the development of research models, including the cultural and linguistic tailoring of interventions, social-network analyses, and wait-listed, randomized controlled trials. The field is also seeing the development and testing of new and existing tools to measure health literacy, to test whether health information is sufficiently

clear so that typical patients are likely to understand that information, and to assess the reliability and validity of tests across different patient populations, different types of health care providers, and different types of health care organizations.

Another theme in the literature is the hypothesis-generating research being conducted in the health literacy field. One example Elwood cited was of a qualitative study involving general practitioners, patients, and pharmacists that aimed to test different approaches for improving adherence to prescription medications among people with chronic conditions. Another study tested different types of conversations to determine which are most effective at identifying the life-sustaining treatments that Chinese-American elders envision before they become unable to speak for themselves (Fung et al., 2010).

Based on his review of the health literacy literature and the outcomes of the projects NIH has funded, Elwood concluded that health literacy is a dynamic state of being that depends on individual circumstances, experiences, mental and physical status, and even on the biophysical processes going on in the body. Communication is a substantive component of anyone’s health literacy, for communication is how people convey their needs, transmit information, and seek recognition that the meaning they have corresponds to that held by others. Health literacy research, Elwood said in closing, studies communication, thought, and other socio-behavioral processes that facilitate knowledge, abilities, and skills to obtain and maintain optimal health and well-being.

THE PRECISION MEDICINE INITIATIVE AND WHY HEALTH LITERACY MATTERS³

There are multiple ways in which precision medicine and health literacy intersect, Wolf said. One relates to how the benefits of precision medicine will be realized in the clinical setting in the future. At this intersection, the challenge will be to transmit information that results from the study of genomics between the health care provider and the patient in a way that engages the patient and enables the patient to take action. The second intersection, which Wolf, Suzanne Bakken, and Ruth Parker discussed in a commentary (Parker et al., 2016) and which was the focus of Wolf’s presentation, concerns the role health literacy will play in ensuring that the

___________________

³ This section is based on the presentation by Michael Wolf, a professor of medicine and the director of the Health Literacy and Learning Program at Northwestern University’s Feinberg School of Medicine, and the statements have not been endorsed or verified by the National Academies of Sciences, Engineering, and Medicine.

PMI can successfully recruit 1 million Americans who will be fully informed about the research in which they will engage.

Referring to the appropriation of $215 million to the PMI for fiscal year 2016 as a “significant event,” Wolf said that paying attention to health literacy throughout the life of this project will be important in order to avoid generating data that cannot be used. He said that while he is enthusiastic about the effect that health literacy can have on the PMI, he is realistic enough to recognize that there are challenges the field will face as it tries to impart what it has learned with the aim of providing value to the PMI and maintaining participation rates throughout the duration of the project. The PMI, explained Wolf, is asking the 1 million participants to agree to a long-term relationship, to undergo a research medical exam and prescription assessment, to have their blood drawn, to complete health surveys, to use and share environmental exposure and lifestyle information using mobile technology, and to share information in their electronic health record. “When you put these all together, this is a lot to ask of the participants,” Wolf said.

In his view, the health literacy challenges start with building trust in the project and the researchers involved in the PMI and helping potential participants understand the PMI’s purpose and its near- and longer-term value. The next challenge will be to ensure participants are giving truly informed consent—that they know precisely what they are signing up for when they agree to participate in the PMI. How successfully these first two challenges are addressed will dictate the quality of the data that the project will get from the participants because many of those data will come from patient-reported outcomes and will thus depend on the ability to retain participants in the study. Wolf’s concern is that if trust and understanding have not been established and if participants are not fully aware of why they are participating and what is required of them, there will not only be attrition but uneven attrition across the different subpopulations that the PMI wants to enroll and follow.

The final health literacy challenge will be to continually inform participants about what the PMI is learning and to convey to them the value of being part of this national effort. “Can we convey the PMI’s purpose, importance, and value for community and for the individual participant?” Wolf asked. While some may participate for altruistic reasons—in particular, to contribute to the formation of knowledge—Wolf said that this can be a difficult concept to grasp. “I think we will find many people are looking for immediate or more contextualized value for being in the study.” He also noted that this challenge is something the research and health literacy communities have been working on for three or four decades. One piece of addressing this challenge, he said, is to help the providers and clinicians involved in this study convey that value to their patients.

As Wolf mentioned with regard to the participation challenge, the health literacy field is not starting from scratch. Three decades of health literacy research has shown repeatedly that many people, especially those with limited health literacy, misunderstand health information on prevention, disease etiology, treatment, and research consent (Ownby et al., 2015). In his opinion, Wolf said, genetics may be among the most complicated subject areas for the general public to understand—an opinion that research appears to support (Erby et al., 2008). Wolf also pointed to research from Michael Paasche-Orlow (Paasche-Orlow et al., 2003; Sugarman and Paasche-Orlow, 2006) showing that the readability and understandability of consent forms was low, and he described a 2013 follow-up on which he and Paasche-Orlow collaborated showing that the situation has not changed much in the intervening decade since those first studies were published. The advent of the Health Insurance Portability and Accountability Act, or HIPAA, has only made consent more complicated, he added. “We continue to not recognize that it is a hard and abstract challenge for people to grasp the value and need to participate, and those with low health literacy are even less likely to participate,” he said.

In fact, Wolf said, research has shown repeatedly that for many reasons, the level of health literacy cannot be separated easily from demographic and socioeconomic factors (Kressin et al., 2000) and that most psychometric tests in wide use today have not been appropriately validated for use among audiences with lower levels of literacy (Wolf et al., 2005). Research has also shown that retention in cohort studies is challenging and that participants with limited health literacy are at greatest risk for attrition (Rabbitt et al., 2004). Wolf also noted that individuals with limited health literacy have lower rates of health and mobile technology use, including the use of portals to access electronic health records (Bailey et al., 2015; Smith et al., 2015). All of these issues intersect with the goals of the PMI, Wolf said.

With regard to psychometric tests, Wolf noted that the research field has begun to recognize that how someone responds to a health questionnaire depends on that person’s health literacy. “What that means is that how someone with low health literacy understands and responds to a question may be vastly different than how a person with adequate health literacy skills responds to the same question,” Wolf said. “While this should not come as a complete surprise, it does mean that the results may be invalid.”

Addressing these challenges, Wolf said, requires exercising some common sense and taking time to think about these challenges before diving right into the research process. “If we are going to make such a large investment in funds to have this incredible cohort, and to move this unprecedented project forward, we just want to make sure it gets down right,” he said. Doing it right, as he put it, requires first operationalizing the mean-

ing of a “patient-powered” study, a phrase that appears often on the PMI website but whose meaning Wolf said he has no clear understanding of. Next, it is important to test the clarity and effectiveness of any recruitment messages before unveiling them to the public. “If we cannot articulate and explain the value of being involved in this study, then we can expect to see the problems I just mentioned,” he said.

It will be important, he added, to recognize the need for a sizable investment in “shoe leather strategies” to ensure that communities in medically vulnerable and underserved areas in the country, where low health literacy is likely to predominate and where participation rates are historically low, are not underrepresented in the PMI Cohort. The problem, however, is there are not many success stories in the literature regarding recruitment from these vulnerable and underrepresented communities.

To be successful, Wolf said, the PMI will have to validate any instruments in the health survey to make sure that all concepts can be well understood across all levels of health literacy. It will also have to consider multifaceted approaches to retaining people in the study over the long term and to properly train and perhaps even retrain participants in the use of mobile technology (Chan et al., 2014). As a final comment, Wolf said he wondered if assessing health literacy should be a component of the PMI. “If we know this is going to be a factor that could lead to disparity, then we need to measure it to make sure that there are no differences,” he said.

DISCUSSION

Ruth Parker started the discussion by summarizing the important messages she had heard from the workshop’s first four speakers. Joseph McInerney spoke about how genetics content relates to an ability to understand life, and Van Driest stressed that precision medicine requires probabilistic rather than deterministic thinking. Given what a big challenge numeracy is in the United States, Parker said she was concerned that many Americans will find probabilistic thinking, which carries the connotation of statistics and risk and probability, to be a significant challenge. Elwood, she continued, spoke of the public good that comes from health literacy research and of the commitment of funders to continue enabling the research that will improve the communication of health information to taxpayers. Concluding, Parker said that Wolf had challenged the health literacy community to get involved in the PMI and had noted the problems that would likely arise should health literacy not be an important consideration in the PMI. She said that she agreed with Wolf that measuring health literacy should be included in the PMI but added that she was worried that the “train has already left the station.”

She then asked the panelists for their ideas on how best to convey

to the public the benefits it stands to gain from participating in the PMI. McInerney said he would emphasize the dual benefits to the community and individual. “There are contributions that one can make by participating in this research that may benefit the individual involved,” he said, “but we hope the prospect of benefiting the larger community is much greater.” He also said that he wants potential participants to understand the potential of the PMI to produce new knowledge that will integrate information about biological variation with what is known about human physiology to generate new insights about health and disease, with an emphasis on health that equals the emphasis on disease. In addition, he said it is important to explain uncertainty to potential participants.

Van Driest said she would like to see the PMI community engage with meteorologists to help with developing explanations for uncertainty, prediction, and probability and would like for the research community to think about how to explain the idea of a confidence interval in a way that potential participants can grasp. Elwood said he would like to see an effort aimed at encouraging potential participants to ask questions and speak up when they do not understand a concept presented to them, while at the same time empowering researchers and practitioners to encourage the public to ask more questions. Wolf provided a similar response. “I want to find ways to empower the public to ask questions, to get engaged, to not accept confusion,” he said. Toward that end, he said he believes there is a need to find avenues to insert the public into the research process to a greater extent than has been tried before with other large-scale projects so that the public does not just have a voice but actually helps in some regards to determine the PMI’s direction by providing real-time feedback that is assessed over time, not just in a cross-sectional manner.

Linda Harris from the U.S. Department of Health and Human Services asked when the PMI was going to start. Van Driest replied that the initial 1-year effort to develop pilot studies, communication tools, and recruitment strategies began in late February 2016.

Cindy Brach from the Agency for Healthcare Research and Quality said that she thought the term “precision” presents a challenge because it implies an illusion of certainty, which runs counter to Van Driest’s emphasis on probabilistic rather than deterministic information. Brach then asked the panel to comment on the importance of genetic information relative to other types of information that could be connected to health, such as environmental impacts. Van Driest said the PMI is focusing on genetics because it is the area of science currently providing the most new data to drive research. “We are starting to routinely collect genetic or genomic information as part of clinical trials and to be able to flesh out the effects of different genotypes within study populations,” she said. The research goal of the PMI, she added, is to use genetic information as a model for how to

collect, analyze, and apply data about other factors, such as environmental impacts, so as to make medicine more precise. Van Driest also questioned whether 1 million people would be a big enough population to sort out all of the contributions to health and disease from all of the factors that could be involved.

The long history of genetic counseling also argues for a focus on genetics, said McInerney, who added that there is a significant literature from the genetic counseling community on how to communicate uncertainty and probability. “We should look to that literature, and we have some genetic counselors in the room who can help us understand that,” he said. Referring to Brach’s comment about emphasizing genetics versus environmental factors, McInerney said he hopes that PMI does not reinforce the age-old notion of genes versus environment. “If there is an underlying message from the Precision Medicine Initiative, perhaps it can be that it is always genes and environment,” he said.

Elwood commented on the importance of looking beyond specific racial and ethnic target populations because of the genetic mixing that has occurred in the United States. As an example, he noted that while Tay-Sachs syndrome occurs predominantly in descendants of Ashkenazi Jews, the Jewish diaspora has led to genetic intermixing in the United States between Jews and non-Jews. “There could be a fifth-generation descendant of Ashkenazi Jewish people who does not identify as Jewish at all, and yet it is through precision medicine that that person might get picked up as vulnerable to Tay-Sachs,” Elwood said. “This is an example of gene–environment interaction in terms of the social environment and identity and part of what makes us uniquely American.” Precision medicine, he added, can provide insightful data because ethnic and racial identity is such a fluid construct in the United States.

With the payoff from the PMI coming in the future, Terry Davis from the Louisiana State University Health Sciences Center asked what messages the public should be getting today about this initiative. McInerney replied that the message should be that there may be some benefit to the individual, but that the greater hope is that the PMI will provide a great deal of information that will benefit everyone. In his opinion, he said, that is the central message inherent in all biomedical research, whether it is a specific clinical trial for a new colon cancer therapy or the PMI. Van Driest said she thought that an important message would be one explaining that the promise of precision medicine is really about potential, not a guarantee. As an example of what she is worried might happen, she referred to comments she hears about the Human Genome Project not producing great advances for the public—which is not true, she said, but which is the public’s perception. She also said she thought it important to spread the message to clinicians and patients that genomic information is already being used to

understand individual response to some commonly prescribed medications. “Communicating those kinds of successes will help motivate recruitment and retention,” she said. Another important message to convey, she added, is that the PMI will help provide a better understanding about how individual variation is important to human health. “It is important to send the message that if I only study people that look, act, and behave like me, I am not going to learn anything important,” Van Driest said.

Wilma Alvarado-Little from Alvarado-Little Consulting asked if there was any consideration being given to the importance of knowing one’s family history, which can be a challenge for first-generation immigrants and refugees in particular. Van Driest said that there is some good work ongoing on how to take, document, and share family histories. One aspect of this is to empower patients when they are uncertain about family medical history to call a relative right then who might have that information. As far as how that was being woven into the PMI, Van Driest said she was not sure, and neither was McInerney. McInerney did note, though, that the American Society of Human Genetics is working to educate health care professionals about better approaches to taking a family history and is also working with the insurance industry to promote the idea that primary care providers should be reimbursed for the time needed to take a good family history. “We know that most primary care providers are not going to take a three-generation family history the way a trained geneticist would,” he said. His organization is also working to educate primary care physicians about certain red flags that come up in family histories, such as the early onset of disease and the presence of multi-focal disease, both of which can indicate a significant genetic contribution. He also said that there are projects under way to help individuals collect their family histories before they go to the doctor and that electronic health records need to incorporate family histories in a more consistent manner.

In the last comment in the discussion session, Ernestine Willis from the Medical College of Wisconsin expressed her concern that the PMI might not truly be community-based, patient-powered, participatory research. She also said she hoped that the failures of the National Children’s Study would emphasize to those leading the PMI how critical it is for the research community to be fully committed to returning data to the community and sharing the knowledge gained through the public’s participation in the project.