As presented in the previous chapter, a host of scientific issues arise when considering the inclusion of subgroups within clinical trials. Subgroup analyses can broaden the findings from a trial, but subgroups also can be too small or unrepresentative to provide useful results. Community involvement requires real partnerships, not just occasional meetings with community leaders. Involving minorities in clinical trials requires changes among many of the stakeholders in the clinical trials system, not just among researchers, government officials, or advocacy groups.
Three speakers examined many of these issues from the perspectives of academia, industry, and government. All three sectors face challenges in overcoming the barriers to clinically meaningful inclusion of minorities in clinical trials, which requires that they work together to overcome these barriers.
One way to overcome the biases evident in the underrepresentation of minorities and women in clinical trials is to make sure that every research project partners with people who are disproportionately and unjustly affected by the conditions being studied, said Carol Horowitz, associate professor of health policy and medicine at the Icahn School of Medicine at Mount Sinai. As a community partner once told Horowitz, “If you change the way you look at things, the things you are looking at change.”
Researchers and research partners can have very different perceptions of a clinical trial, Horowitz pointed out. Researchers may think that the
purpose of a project is clear and important, that the research needs to be done quickly, that the benefits are obvious and the risks minimal, and that patients should of course agree to participate. Potential research participants may have earned skepticism from past experiences, may not see the need for haste, may perceive the benefits as unclear, and may view the risks as unacceptable if the benefits are minimal. “We have to tell people honestly that we are doing the research because we do not know whether it is going to have a benefit,” said Horowitz. “That, for people, might right away be a game stopper. ‘If you do not know if it is going to work, try it out on somebody else and let me know.’ People know about historical abuses.”
Many people who are asked to participate in research know why they are sick. In a series of focus groups that Horowitz did with people with hypertension, they said that their blood pressure was high because of poverty, pollution, racism, and stress. “You are medicating a social condition,” they said, according to Horowitz. “Fix our communities, and our blood pressure will go down.”
For researchers, the research system is designed to get the patients it gets. Participants tend to be easy to contact, easy to enroll, easy to follow up, and compliant. These are usually not the patients who are underrepresented in research, Horowitz pointed out. In addition, stated and unstated design and inclusion criteria have a tendency to exclude people. For example, Horowitz cited a diabetes prevention program that required patients to keep a multiple-day food diary. In a heart failure study, a self-efficacy questionnaire asked potential patients how confident they were that they could take all their medicines as prescribed and follow a low-salt diet, and respondents with low self-efficacy were excluded.
Horowitz asked the workshop participants who had been involved with clinical trials whether they would take part in their own studies. Would they take a phone call at home? Did they have enough time to participate? Would they be available at night or on weekends? Did they have easily accessible child care? She also asked how much time researchers spend crafting their strategies and messages. Do they craft these messages within the social networks they hope to reach? Can potential participants understand the messages? Are they likely to trust a researcher? “People do not separate a hospital from a researcher,” she said. “If you do not have a good experience with the asthma clinic or the asthma practice at your hospital, you are not going to be in a study.”
She described two case studies, starting with a diabetes prevention trial in East Harlem, which is a mostly black and Latino low-income community in northeastern Manhattan. Increased physical activity and weight loss can reduce the progression to diabetes by two-thirds and eliminate the disparity between blacks, Latinos, and whites in the development of diabetes. The study looked at whether a peer-led intervention in Harlem
could prevent diabetes through lifestyle change, with not just the methods but the subject of the study being chosen by the community. Recruitment occurred, according to Horowitz, “anywhere in East Harlem where we thought people were”: churches, food pantries, senior centers, schools, health centers, and other community gathering places. Participants did not need regular health care providers in order to enroll in the trial, and half of the patients were uninsured. “We are fierce,” she added. “We have an agreement with the clinical sites that partner with us that they will accept patients. We have a list . . . of all the places that can accept people for care, places that take uninsured patients. We always try to connect people to care,” she explained. After giving their consent, participants did glucose tolerance and other blood tests, filled out a survey, participated in eight sessions, and did follow-up sessions at 3, 6, and 12 months.
The study faced several challenges, Horowitz said. The population represented by those living in East Harlem has been underrepresented in these kinds of studies and historically has been seen as difficult to engage in research. Trying to get people to change how they move and eat is difficult. Efforts to reach out to the community sometimes were misguided. Early in the study, a poster showing an obese black man with an amputated leg and the words “Portions Have Grown: So Has Type 2 Diabetes, Which Can Lead to Amputations” was criticized for showing just one ethnic group—especially when it was revealed that the photograph had been digitally altered to remove the model’s leg.
The study sought to overcome these challenges by working in partnership with a community action board. Through this board, the community chose the topic, methods, strategies, incentives, domains for surveys, and analytic questions for the study. The community did not want a control group and opted for a delayed intervention. “We do not want to tell people they are going to get nothing, but if they can have it now or in a year, that is fine,” said Horowitz. The community called attention to such factors as residential segregation, sleep apnea, and food insufficiency, which made the study more complex but more accurately reflected the community context. Social marketing, street art campaigns, and other forms of outreach sought participants. The study provided community benefits such as employment and capacity building, and team members did not need a primary care provider, literacy, or a Social Security card in order to enroll in the research.
The study enrolled about 500 diverse patients, mostly through their community partners. The result was significant weight loss that was maintained at 1 year. “People lost weight,” said Horowitz, and “they kept it off.”
The second case study involved disparities in kidney disease. A genetic variant that is more common in African Americans than in other populations can increase the risk for kidney failure in people with high blood pressure. This variant was favored in Africa because it protects against sleeping
sickness, but in today’s world it can have harmful effects. The object of the study was to incorporate risk information into clinical care in an attempt to motivate people to improve management of their blood pressure. The study was still ongoing at the time of the workshop, but Horowitz described some of the challenges this study has faced. She was warned that studying a genetic variant that affects the health of a particular population group could contribute to racism, but that has not been her experience. Horowitz explained that “the first person I talked to was one of my colleagues, who is a pastor, who said, ‘Now maybe white doctors who see black people on dialysis won’t think it is because we didn’t try hard enough. They will recognize there is more to disease than bad behavior.’”
Recruitment was also a challenge in a patient population that is difficult to reach and schedule. An inclusive approach was again the answer, said Horowitz, with formative research to develop outreach, a stakeholder board that included patients and community leaders, and stakeholder engagement throughout. For example, the board changed aspects of the study design and worked with the IRB at Mount Sinai to gain approval of study modifications. The study also developed apps with QR-type barcodes for enrollment and verbal consent procedures to overcome low literacy levels. The study has been designed to produce actionable information, so that participants know what to do with the results of a test. In the event of a positive genetic test, for example, patients talk not only with a genetic counselor but with another patient who has had a positive test.
Horowitz concluded with several lessons drawn from these experiences. First, look through participants’ lenses, she said. Studied need to be designed with inclusiveness in mind at every step; she added that “You should have studies that people want to join.” Teams should be committed to research and include people from the target community, she said, with both proven and novel strategies being used. Evaluation of these strategies, with publication of data so other people can learn from the experiences, can lead to modifications that improve the strategies. Finally, she urged study leaders to share results with participants and communities and democratize data. Every patient who was in a study should receive a note thanking them and telling them what was learned, she said, and the data should be freely available for anyone to examine.
The number one priority for the biopharmaceutical industry is improving timely access to innovative new medicines, said Jocelyn Ulrich, senior director of scientific and regulatory affairs at Pharmaceutical Research and Manufacturers of America (PhRMA), which represents biopharmaceutical research and discovery companies in the United States. The cost of devel-
oping a new medicine currently exceeds $2 billion and takes more than a decade, she explained. Advancements in science and technology are changing the understanding of disease and the development of drugs. But the regulatory framework and acceptance of drug development tools and methodologies must evolve to keep pace with these advances and encourage the inclusion of participants in ways that are both science based and meaningful, she said.
Patient engagement at appropriate times throughout the drug development process has a number of benefits, Ulrich observed. During the design phase, the inclusion of patient perspectives allows for designs that encourage participation and reduce burdens on patients and caregivers, which creates the potential to improve recruitment and retention. Patient-focused drug development tools, if accepted by patients and regulators, can improve efficiency and provide data that are meaningful to patients, including novel clinical endpoints, clinical outcome assessments (including patient-reported outcomes), and benefit–risk assessments.
Scientific factors should be the primary drivers when considering the need for specific racial, ethnic, or geographic groups to be represented in a trial, said Ulrich. This decision needs to be informed by nonclinical, preclinical, and early clinical data and by biological and medical knowledge about the disease prevalence and mechanism of action of the compound, she added. In addition, when significant differences in drug metabolism are expected, specific subpopulations need to be included in early-phase development to better understand pharmacokinetics, safety, and dose–response data.
The overarching principle for considering the need for specific racial, ethnic, or geographic groups to be represented in a trial is that sample sizes for trials should be driven by the primary objectives and hypotheses of a study, Ulrich noted. From a scientific perspective, it should not be necessary for studies to be powered to be significant in any specific subgroup unless there is a biologically plausible reason to do so, she said, and sponsors must balance the need to make treatment options available in a timely manner with the need to enroll a defined subpopulation.
Another overarching principle is that prespecification and documentation of the plan for subgroup assessment in the protocol are essential for interpreting the results, Ulrich explained. Remaining questions about subgroup differences then need to be monitored and answered, she said. In some cases, these differences may be most effectively studied through the use of real-world evidence after approval, as when there are cases of low incidence rates or where it may not be practical to enroll a particular population. The context can drive the balance between pre- and postapproval information collection. For example, better understanding of patient tolerance for risk may create more willingness to accept uncertainty in return for earlier access to the medicine in a broader population.
Innovative tools and approaches are being developed and used in the drug development space, Ulrich pointed out. For example, adaptive designs may use accumulating data to modify aspects of a study as it continues without undermining the validity and integrity of the trial. Such designs could be designed to drop a study arm if prespecified safety or efficacy measures are not met, so that more patients can be enrolled in an arm that would be beneficial to them. “From an ethical perspective, that is a great advancement in drug development, rather than being given the choice between placebo or standard of care versus the active treatment,” said Ulrich.
Inclusion and exclusion criteria generally are developed based on the intended patient population for the investigational treatment and the questions that the study seeks to answer. These criteria need to allow for the identification of a well-defined population and minimize patient risks without being unnecessarily restrictive, Ulrich said. Arbitrary limits, such as 65 years as an upper limit for age, may generally be inappropriate, and the exclusion of patients with common concomitant illnesses should be done only if this is expected to affect treatment effect or safety, she added.
As pointed out in the previous chapter, standard categories of race and ethnicity are widely acknowledged to be inadequate for understanding meaningful genetic differences. Self-identified race and ethnicity generally correlate with population groups but not necessarily with an individual’s distinct genetic background, which can lead to confusion and debate about the relationship between the risk of disease, treatment options, and self-identified race. Adding country of origin in addition to race may be one way to disaggregate the data and get more meaningful data for patients. Also, drug development is a global enterprise, Ulrich reminded the workshop participants. Multiregional clinical trials with non-U.S. data constitute more than 70 percent of applications to FDA.
Drug development tools, such as biomarkers and surrogate endpoints, are indicators of biological processes that may be used to identify patients who are more likely to benefit from investigational treatments, Ulrich explained. Use of such tools may increase the efficiency of the drug development process and provide meaningful results back to patients.
Achieving greater diversity in clinical trial participation faces significant practical barriers that add to the scientific complexities mentioned earlier, Ulrich said. According to one survey that she cited, 57 percent of patients would prefer to receive information about trials from their primary physician, but only 20 percent do. The study design or process may be too complex, with lengthy informed consent and too many visits and procedures. Staff may not be adequately trained to deliver linguistically and literacy-appropriate information, and patients may fear or mistrust researchers. Participation may involve lost time or wages and create child
care or transportation problems, and patients may lack awareness or have misperceptions about the clinical trial process. The lack of proximity to research sites and trial processes that may only be conducted at an individual site may also cause difficulties.
The biopharmaceutical industry has been engaged in improving awareness and knowledge about clinical trials. For example, PhRMA partnered with the National Minority Quality Forum in 2014 to launch “I’m In,” which is an advocacy, social media, and educational campaign.1 PhRMA is a member of the Coalition for Clinical Trials Awareness, and it sponsored the first Center for Information and Study on Clinical Research Participation AWARE for All event in Washington, DC, in 2015. In addition, PhRMA also has created a website to provide educational resources and information about clinical trials.2
Some of the scientific challenges mentioned could be addressed by innovative designs and methodologies in combination with advanced drug development tools. These collectively have the potential to accelerate drug development while obtaining evidence-based information about subgroups in an efficient manner. However, to realize this potential, a multistakeholder approach is needed to address the practical barriers to participation that affect people from historically underrepresented populations, Ulrich concluded.
The Patient-Centered Outcomes Research Institute (PCORI) is very interested in strategies for ensuring broad participation in clinical trials, said David Hickam, program director of the clinical effectiveness research program for PCORI. The institute funds comparative clinical effectiveness research that engages patients and other stakeholders throughout the research process. Hickam described PCORI’s research interest as studies of the real-world effectiveness of clinical interventions that have made their way into clinical practice. PCORI strives to produce research results that are applicable to broad population groups, even if treatment effects are heterogeneous, and it disseminates information about the results of this research.
Comparative effectiveness research compares alternative approaches to the management of a clinical condition. The starting point for such research, said Hickam, is looking at the choices that patients and clinicians make in deciding among management options. Those choices provide
insights into what information is needed by decision makers and where there are important gaps in the available clinical evidence. The research defines important patient subgroups up front, recognizing disparities in their health and health care and the origins of those disparities. It then seeks to define the outcomes that are important to patients, including benefits and harms.
Comparative effectiveness research also is characterized by patient and stakeholder engagement. Studies are designed to integrate with routine clinical or office operations and minimize disruption to participants’ daily routines. It seeks to refine recruitment strategies and deal proactively with recruitment issues.
PCORI believes that comparative effectiveness research should be carried out by interdisciplinary teams, Hickam said. Once a study starts, the team works together, so that when problems arise, multiple perspectives can be brought to bear on those problems, including the perspectives of patients. The stakeholder and patient partners also can participate in the monitoring of data and safety issues, and they can help to think about how to make the studies more efficient by capitalizing on existing resources, such as electronic health records, claims databases, or data networks.
PCORI has initiated more than 200 studies based on this model, which has revealed certain barriers that can arise in the course of such research, said Hickam. First, researchers can be inexperienced in building and managing interdisciplinary teams. Also, partners sometimes are kept at arm’s length, with overly formal meetings and a reluctance by investigators to acknowledge problems. Meetings may be monthly or even quarterly instead of the regular interactions that are important for engagement. When problems do come up, researchers may be reluctant to share bad news with partners. As a result, “The people who could actually help you troubleshoot some of your problems are not in the loop with the problems they are encountering,” said Hickam, adding that “Researchers have a tendency to want to reassure the stakeholder partners that everything is going well.” This is human nature, he said, but it emphasizes the need to think about how multidisciplinary teams are put together.
Another problem arises when partners do not have sufficient access to clinical sites or enough time or resources to participate sufficiently. Partners may be asked to volunteer their time to the project in an otherwise busy workday, and even the best-intentioned plans sometimes can have trouble getting carried out. These barriers can be overcome, but the community needs to consider these problems so it can move forward together, said Hickam.
Hickam also noted that PCORI requires reports on the demographics of the people participating in the studies it supports. In some cases, the racial and ethnic composition of participants turned out to be different
than planned in the original applications, which may result from investigators concentrating their recruitment activities in narrow settings, such as university clinics. “One of the things that we have tried to do with projects is to push for more broad-based recruitment strategies,” he noted.
Hickam also talked about the movement for pragmatic clinical trials. Developed as an alternative to conventional, tightly controlled clinical trials, pragmatic clinical trials have been seen as a way to evaluate the effectiveness of clinical interventions in real-world settings (Thorpe et al., 2009). They typically are designed to minimize exclusion criteria that act as a barrier to participation by some patients. Thus, all patients with a condition are invited to enroll, with few if any exclusion criteria. These studies are conducted in real-world settings, such as during regular clinical visits, rather than in specially designed clinical research units.
Pragmatic clinical trials look at the effectiveness (whether a treatment works under typical circumstances) rather than the efficacy (whether a treatment works under optimal conditions) of interventions, typically investigating treatments that are in common use, even if those interventions are complex. Pragmatic clinical trials provide a way to make head-to-head comparisons, whether between drugs, surgical procedures, or other interventions. Hickam added that PCORI has been insisting that comparisons to “usual care” be approached with caution. If a usual care condition is used, it should be carefully defined and measured. He added, “Generally, we try to steer people to more active comparators rather than usual care comparators.” Placebos can be used in pragmatic clinical trials, though the pharmacy has to be brought in as a partner.
Pragmatic clinical trials have limitations, Hickam noted. They often do not provide sufficient outreach to patients who are uninsured or have other problems with access to care. Methods used to measure outcomes also often have lower precision with regard to such factors as time courses, resulting in a loss of power.
PCORI also is funding initiatives that focus on learning about best practices for trial recruitment. Its comparative effectiveness research methods program is looking at novel approaches to engagement and to informed consent. Its disparities research program is examining patient navigation and literacy and numeracy barriers.
Hickam briefly touched on statistical approaches for analyzing the heterogeneity of treatment effects. Stratified analyses separate important subgroups but increase the risk of spurious findings. Multivariate analysis with interaction terms reduces the risk of spurious findings but lacks power, requiring substantially larger sample sizes. Post-hoc analyses of subgroups should be considered exploratory, he observed.
“Patient and stakeholder engagement in clinical trials is an important emerging strategy,” said Hickam. “It can prevent and correct problems with
recruitment. [But] it requires the commitment of investigators and skills in how to manage interdisciplinary teams and approaches.”
One topic that arose during this discussion session and several other times during the workshop involved the diversity of the researchers and research leaders conducting clinical trials. Ulrich said that broadening the representation of underrepresented minorities in research has been “a constant topic of conversation” within industry. For example, PhRMA has been working with minority physician groups such as the National Hispanic Medical Association and the National Minority Quality Forum to support policies that can diversify the research community. This is particularly important, she said, to increase the referral rate from physicians and participation rates from underrepresented populations.
A related issue is how to support investigators as they are engaging communities, which often takes longer than the usual schedule of research grants. Horowitz pointed out that pilot grants can help generate stakeholder engagement while pilot data are being gathered, whereas other kinds of grants are designed for research teams that are already in place. Hickam added that PCORI has started a program called Pipeline to Proposals, which is a tiered system that offers a limited amount of funding for front-end work to pull together community partnerships.
Ulrich also pointed out, in response to a comment, that recruitment can be a challenge in industry, which has budgetary and time constraints. But the use of more patient-focused drug development tools can increase inclusion and enable industry to go beyond working with the same sites over and over. It will take time for such changes to permeate the entire system of sites, investigators, and procedures, but the pendulum is starting to swing toward more inclusive studies.