F

Large Genetic Cohort Studies: A Background¹

BACKGROUND

For decades, large-scale prospective studies have been carried out with the goal of accurately assessing the relationships between biomedical factors, environmental exposures, and health outcomes for both their study participants² and source populations (Willett and Colditz, 1998). For example, extensive personal data collected for the National Health and Nutrition Examination Survey (NHANES) led to the discovery of the association between high cholesterol levels and heart disease,³ while the Framingham Heart Study made clear the heart health risks of tobacco smoking.⁴ Recently, technological advancements have opened the door for the incorporation of genetic data into these types of studies, allowing for the discovery of specific pathogenic and protective genotypes through genome-wide association studies (GWASs).

___________________

¹This background paper was prepared by Roundtable interns during the summer of 2015, Andy Castro (Northwestern University) and Lauren Nahouraii (Duke University), and shared with the participants in advance of the workshop.

²The use of the terms “cohort” and “participant” are currently being reconsidered by the genetics research community, but since there is no current consensus on an alternative, these were the terms chosen for use in this article.

³National Health and Nutrition Examination Survey, http://www.cdc.gov/nchs/nhanes.htm (accessed June 13, 2016).

⁴Framingham Heart Study, https://www.framinghamheartstudy.org (accessed June 13, 2016).

Using genetic data in studies like NHANES and the similar Wisconsin Longitudinal Study⁵ has been advantageous because of the large volume and various types of phenotypic information on record for comparison. Similar methods allowed the Framingham Heart Study to find added success in identifying pathogenic variants and confirming the functions of candidate genes (Framington Heart Study, 2016). Seeing the potential of this model for drug development, pharmaceutical companies have started collaborating with organizations that have access to large databases of genomic information. For example, 23andMe’s partnership with Genentech seeks to find a genetic cause of Parkinson’s disease by using data from 600,000 consenting customers, and Amgen’s acquisition of DeCode Genetics allows access to data for over half of Iceland’s adult population (Chen, 2015). Similarly, Geisinger Health System has partnered with Regeneron to sequence and research 100,000 participants as part of its MyCode Community Health Initiative.

Other countries have begun longitudinal research using genomic and other data as well, using centralized systems like the UK Biobank (Manolio et al., 2012), Qatar Biobank,⁶ and Danish Civil Registration System (Schmidt et al., 2014). These offer examples of how other countries’ health care infrastructures may allow researchers easier and less expensive access to linked biomedical information because the information can all be found in one place. Alternatively, less centralized consortium models can compile data from similar studies, such as the National Cancer Institute’s Cohort Consortium, which unites more than 40 cohorts consisting of more than 4 million people,⁷ and the Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE), which performs meta-analyses and GWASs for 10 separate cohorts.⁸ The Precision Medicine Initiative (PMI) of the National Institutes of Health (NIH) may adopt the consortium method in part by incorporating ongoing studies with direct volunteer recruitment, pursuing a goal of enrolling over 1 million Americans (Collins and Varmus, 2015). The scale and scope of new endeavors in large cohort research may provide an opportunity to learn from previous and ongoing studies.

___________________

⁵Wisconsin Longitudinal Study, http://www.ssc.wisc.edu/wlsresearch/gwas (accessed June 13, 2016).

⁶Qatar Biobank for Medical Research, http://www.qatarbiobank.org.qa (accessed June 13, 2016).

⁷National Cancer Institute Cohort Consortium, http://epi.grants.cancer.gov/Consortia/cohort.html (accessed June 13, 2016).

⁸Cohorts for Heart and Aging Research in Genomic Epidemiology, http://www.chargeconsortium.com, (accessed June 13, 2016).

Public Support and Willingness to Participate

The success of large genetic cohort studies relies on the enrollment and retention of participants, and achieving this may be aided by understanding and responding to public perception about participation. Research by Kaufman et al. (2008) found that among a representative sample of 4,659 Americans, 84 percent supported conducting genetic research, and 60 percent said they would participate. This was consistent across most racial/ethnic demographics, with only American Indian/Alaskan Native respondents showing relatively less support (65 percent). Similar surveys have shown that participant support may be directly correlated with both education and income level. These surveys also show significantly higher approval among urban respondents (80 percent) than rural respondents (73 percent) and a higher likelihood of participation among Spanish speakers (61 percent) than English speakers (56 percent) (Bollinger et al., 2014; Kaufman et al., 2008; PMI Working Group, 2015).

Kaufman et al. (2008) assessed what factors would incentivize participation in a large genetic cohort study and found that the return of individual results had the most positive impact on recruitment. Ninety percent of participants desired the return of all of their results, even if they offered no clinical utility (Kaufman et al., 2008). More recent surveys support these findings, with 82 percent of respondents in one study indicating “it would be interesting to receive the results of the study” (PMI Working Group, 2015) and 90 percent indicating that “learning information about [their] health” was either somewhat or very important in influencing their decision to participate (Bollinger et al., 2014).

The public’s clear interest in receiving individual data and health information could be seen as challenging by some researchers, as the process of returning results is costly and time consuming and can be bound by institutional review board restrictions. There may also be a notion among researchers that they should “protect” participants from possible misconceptions regarding the therapeutic value of research data (Bollinger et al., 2014; Kaufman et al., 2008). However, Bollinger et al. (2014) showed that if participants are informed of the fact that the return of results may slow the progress of a study, participants become more willing to compromise and accept only medically actionable information. This could indicate a “middle ground” that is palatable to both participants and researchers for future large-cohort studies. MyCode plans to return results on genetic variants that have been identified as actionable

by the American College of Medical Genetics and Genomics, and patients will then be referred to precision health clinics for appropriate genetic counseling.

Additionally, both the Kaufman et al. (2008) research and a survey presented by the PMI Working Group (2015) identified monetary compensation as the second most important incentive for participation, with 80 percent of respondents ranking it as either somewhat or very important. Kaufman et al. (2008) found no difference in the effect of a $200 incentive on participation rates between those earning less than $25,000 per year and those earning more than $75,000 per year. A review of population-based cohort studies also showed that increasing monetary incentives had the largest positive effect on the retention rates of participants (Booker et al., 2011).

Use of Modern Technology in Recruitment and Data Collection

Using various technological innovations may help in enrolling and following up with study participants. The Kaiser Research Program on Genes, Environment, and Health and the Vanderbilt University BioVU program use models of recruitment for cohort studies that are built into their existing electronic health record and patient registries, allowing for a simpler linkage of information and follow up. However, differences in medical terminology can cause confusion when medical histories are being assessed, and many institutions do not currently possess the technological infrastructure to support interoperability (Manolio et al., 2012). Other methods of recruitment using e-mail and text messaging services have been shown to be cost-effective and popular with younger participants, but they produce lower response rates than mailed requests. The use of social media is an inexpensive, though time-consuming, method to inform and update participants on recent developments, and it allows the participants to promote the study they are involved in and encourage their friends and family to participate as well (Toledano et al., 2015). A survey of participants in the Qatar Biobank showed that 72 percent decided to contribute based on recommendations from friends and family, demonstrating the influential power of “word of mouth” on recruitment (Qatar Biobank, 2015).

The European Prospective Investigation into Cancer and Nutrition (EPIC) study has experienced success in recruitment and data collection by using touch-screens instead of interviews or paper forms (Manolio et al., 2012). 23andMe has employed a user-friendly interface for its web-

site that prompts the user with questions about his or her health history.⁹ Similarly, the collection of large amounts of various types of biomedical data from wearables (e.g., Fitbit, Apple Watch) and smartphones offers new ways to engage the public. While the specifics of how to effectively implement these technologies are still being studied, 60 percent of respondents to the 2015 NIH survey said that, given the opportunity, they would use a mobile device to submit health data at least once a day (PMI Working Group, 2015).

Consent, Data Sharing, and Privacy

It is generally difficult during the initial consent process of a large-cohort study to foresee what sorts of future studies might also use the data collected for that study. Thus, researchers who conduct such studies generally prefer to use broad consent models as a way of reducing the financial and time costs of re-consenting participants (NHGRI, 2005). While 64 percent of respondents to the 2015 PMI survey mentioned above said they would give broad consent, 73 percent said that they would prefer a dynamic model that would allow personal control over exactly who could use their data and how. “Layered” or “tiered” consent models ensure that participants are informed of their options, but the process can be time-consuming for clinicians and researchers.

To alleviate this problem, innovative models like Genetic Alliance’s Platform for Engaging Everyone Responsibly (PEER)¹⁰ are being implemented. PEER is an online, user-friendly interface that educates research participants about the practices and goals of individual studies so that they can selectively decide who can access their data and for what purposes. Users also determine their own privacy settings by either keeping their data anonymous or allowing it be linked back to them for use in clinical care. Models similar to PEER could be useful in streamlining the consent process while empowering participants to take control of their health data.

___________________

⁹23andMe Research Portal, https://www.23andme.com/23andMeResearchPortal (accessed June 13, 2016).

¹⁰Platform for Engaging Everyone Responsibly, http://www.geneticalliance.org/programs/biotrust/peer (accessed June 13, 2016).

Engagement of Health Care Providers

The main obstacles that physicians cite as preventing them from participating in research and engaging their patients are: research that does not fit their practice, high work burden/pressure, and unfamiliarity with the study (lack of understanding of research objectives) (Arends et al., 2014). Within the current health care reimbursement restructuring, physicians also lack the time to devote to studies (Robitaille et al., 2014).

Physicians also perceive barriers to integrating genetics services, which may in turn hamper the recruitment of the physicians’ patients to participate in scientific research. Some of the barriers that have been identified are a lack of genetic knowledge and skills; ethical, social, and legal implications (ESLIs); health care system inadequacies; and lack of scientific evidence (Mikat-Stevens et al., 2015). In addition, there is a lack of awareness among primary care physicians about the Genetic Information Nondiscrimination Act, which was passed in 2008 to prohibit the use of genetic information in health insurance and employment. If physicians remain skeptical about the potential of personal genetic information to negatively affect patients’ insurability or further coverage, they are unlikely to recruit their patients for research (Mikat-Stevens et al., 2015).

Robitaille et al. (2014) designed a systematic process for recruiting physician–patient dyads in practice-based research networks (PBRNs) and found that there are two main components of successfully recruiting physicians—a personal connection and participant buy-in. Additionally, Long et al. (2014) showed that physicians who work in clinical units where colleague physicians recruit participants are more likely to recruit participants themselves. Peer coaching by “physician champions” could be a valuable avenue of not only recruiting clinicians but also keeping them engaged and recruiting patients (Long et al., 2014).

Addressing Potential Disparities and Building Trust

Discrepancies in health status and care among racial/ethnic groups are well documented (Lavizzo-Mourey et al., 2005; Yancy et al., 2005), but efforts to understand and address these problems have been hindered by low levels of minority participation in health studies (Levkoff and Sanchez, 2003; Moreno-John et al., 2004). A review by Yancey et al. (2006) analyzed the factors that influence minority recruitment and retention in research studies. The findings revealed a lack of trust in inves-

tigators and the government, resulting from suspicions that participants would be mistreated for the benefit of scientists’ careers and not their community’s health. To address this issue, the involvement of researchers in the community has been shown to build trust and increase study participation. Community involvement may be achieved through the appointment of local outreach workers to communicate the potential benefits of research to the public and by the employment of minority investigators to serve as “cultural insiders.” The identification and use of specific hubs of activity, such as places of worship, can also help build trust in research efforts and serve as places of recruitment (Yancey et al., 2006).

Another approach to increasing the involvement of underrepresented populations in research is the creation of community advisory boards that consistently meet to discuss a study’s progress and any potential concerns by the public. This provides an external perspective and allows for participants to actively involve themselves in the study’s ethical implementation (Manolio et al., 2012). According to the 2015 NIH survey, 71 percent of respondents felt that participants and researchers should be “equal partners” in the study, with a majority agreeing that participants should help decide what research is appropriate and what to do with study results (PMI Working Group, 2015).

Participant retention can be improved by taking the time to have a consistent follow up to personally engage participants and by creating materials that are culturally tailored (Yancey et al., 2006). Data collection centers that are easily accessible, particularly in rural areas, have demonstrated success in retaining participant involvement (Iredale et al., 2005). While the use of regularly timed incentives could be considered in communities that express a desire for them (Yancey et al., 2006), the belief that the participant is doing something beneficial for his or her future generations may be all the incentive that is needed.

For consortium models, it may be important for researchers to acknowledge the importance of local community engagement, since many ongoing cohort studies are invested in their own customized methods that have proven locally successful. If they are to be assimilated into a larger consortium, researchers may need to explore a balance between a centralized data collection system and the localized methods of specific sub-cohorts (Manolio et al., 2012).

REFERENCES

Moreno-John, G., A. Gachie, C. M. Fleming, A. Napoles-Springer, E. Mutran, S. M. Manson, and E. J. Perez-Stable. 2004. Ethnic minority older adults participating in clinical research: Developing trust. Journal of Aging and Health 16(5 Suppl):93s–123s.

NHGRI (National Human Genome Research Institute). 2005. Design considerations for a potential United States population-based cohort to determine the relationships among genes, environment, and health: Recommendations of an expert panel. https://www.genome.gov/Pages/About/OD/ReportsPublications/PotentialUSCohort.pdf (accessed March 8, 2016).

PMI (Precision Medicine Initiative) Working Group of the Advisory Committee to the NIH Director (ACD). 2015. Survey presented at the Participant Engagement and Health Equity Workshop, July 1–2, 2015, Bethesda, MD.

Qatar Biobank. 2015. “Word of mouth” sees 72% of community contribute to Qatar Biobank. http://www.qatarbiobank.org.qa/media-center/news-detail?item=25 (accessed March 8, 2016).

Robitaille, H., F. Legare, and G. Tre. 2014. A systematic process for recruiting physician–patient dyads in practice-based research networks (PBRNs). Journal of the American Board of Family Medicine 27(6):740–749.

Schmidt, M., L. Pedersen, and H. T. Sorensen. 2014. The Danish Civil Registration System as a tool in epidemiology. European Journal of Epidemiology 29(8):541–549.

Toledano, M. B., R. B. Smith, J. P. Brook, M. Douglass, and P. Elliott. 2015. How to establish and follow up a large prospective cohort study in the 21st century—Lessons from UK cosmos. Public Library of Science One 10(7):e0131521.

Willett, W. C., and G. A. Colditz. 1998. Approaches for conducting large cohort studies. Epidemiologic Reviews 20(1):91–99.

Yancey, A. K., A. N. Ortega, and S. K. Kumanyika. 2006. Effective recruitment and retention of minority research participants. Annual Review of Public Health 27:1–28.

Yancy, C. W., E. J. Benjamin, R. P. Fabunmi, and R. O. Bonow. 2005. Discovering the full spectrum of cardiovascular disease: Minority health summit 2003: Executive summary. Circulation 111(10):1339–1349.

This page intentionally left blank.