ELSI Research Methods Spotlight: Social Network Analysis

Laura M. Koehly, Ph.D., Chief & Senior Investigator at the Social and Behavioral Research Branch (SBRB) of the National Human Genome Research Institute (NHGRI), develops and applies social network methods to her study of family and community members’ behavioral responses to genetic/genomic risk information. A social network perspective is a key to the development of strategies to help individuals and families cope with disease risk information and adhere to risk-reducing behaviors. It can also help us understand factors that contribute to resilience in the caregiver role in families affected by hereditary disease. CERA staff spoke to Dr. Koehly about her work at the SBRB, gathered some insights into her process for conducting social network analysis, and explored her experiences with adapting an educational resource for use with new populations.

CERA: For those who may not know much about it, could you please start off by telling us a little bit about the NHGRI intramural SBRB? How does (or doesn’t it) interact with the extramural, NHGRI ELSI Research Program?

Dr. Koehly: The SBRB was launched in 2003 when the Human Genome Project was completed. At that point, both Francis Collins, who was the NHGRI Institute Director at that time, and Eric Green, who was the Scientific Director, recognized the value of the social and behavioral sciences in the context of genetic and genomic advancements and the need for the NHGRI intramural research program to have a dedicated branch focused on this work. Our vision and mission of the branch align with the goal of leveraging genomic science to enhance public health while promoting equity and inclusivity in research and practice. Ongoing research in the branch is further described in the table below, but in summary, SBRB faculty are conducting research across the discovery to translation continuum. Some focal areas that align with ELSI research include:

• Building tools to promote genomic literacy toward empowering individuals and communities to understand the implications of genetic information for their health and the health of their family. This can help people make informed decisions regarding genetics services and interventions, and ultimately improve health outcomes.  
• Investigating whether stigma is associated with conditions that have genetic contributions, including, for example, overweight, neurodevelopmental conditions, diabetes, and sickle cell disease. Our research can identify strategies to reduce stigma and thereby improve access to care and support for affected individuals and their families. We reduce stigma through education, advocacy, and destigmatizing narratives or amplifying individuals’ experiences and stories to increase empathy and reduce the stereotypes that underlie stigma, and working to understand the importance of changing potentially stigmatizing language.
• Characterizing the influence of social relationships on health in the context of genetics/genomics. We are researching patient-provider interactions; the flow of information within the family; patterns of encouragement by network members that influence engagement in health behaviors like screening, treatment compliance, and lifestyle factors; and support resources that foster resilience in response to caregiving stress and burden. Our understanding about these social dynamics informs interventions tailored to these social processes.

While we share interest and expertise in ELSI research, our interactions with the NHGRI extramural ELSI program are informal. We share our science at both internal and external scientific symposia and conferences and many of us have strong collaborations with extramural researchers outside of the NIH, where we sometimes work as unpaid collaborators on NIH-funded projects.

Ongoing Research at the NHGRI Social and Behavioral Research Branch

Genomic Discovery Science	Several of our faculty engage in innovative genomic research through transdisciplinary science to explore the complex interplay between genetics, behavior, and social determinants of health. This work is being done in the context of ADHD, autism, and caregiver stress and burden when caring for relatives with rare disease.
Genomics Integration into Clinical and Community Settings	One branch priority is to develop strategies for translating genomic discoveries into actionable interventions and treatments that can be effectively implemented in both clinical and community settings. Here, our goals are to bridge the gap between research findings and real-world applications and ensure that all populations can benefit from genomic advances. For example, our Immersive Simulation Program develops virtual simulations that help identify best practices for translating genomic information into the clinical setting. Virtual simulations are also being investigated as a treatment for individuals with ADHD.
Equity and Inclusivity in Genomics Research	We prioritize diversity and inclusivity in genomics research by actively engaging communities historically underrepresented in biomedical research. This includes establishing partnerships with community organizations, advocating for diverse participation in research studies, and addressing disparities in access to genetics services. One example of this work is the Democratizing Education for Sickle Cell Disease Gene Therapy project.
Connecting Research Laboratories with Communities	We facilitate meaningful collaboration between our research laboratories and community partners to ensure that our research priorities are aligned with the needs and values of the communities that we serve. This involves building trust, promoting transparency, and co-designing research initiatives to address the unique challenges and opportunities within different communities. For example, the Families SHARE toolkit was developed through partnerships with academic, medical, and community organizations, with a focus on serving historically marginalized populations.
Training the Next Generation of Scientists	The branch mentors and supports a new generation of scientists who are dedicated to advancing genomic research while upholding principles of equity, diversity, and inclusion. We do this by providing opportunities for education, training, and mentorship that emphasize the importance of ethical scientific practices, cultural sensitivity, and community engagement in genomic research and practice. We have a strong commitment to training, and eagerly bring on trainees through NIH’s wonderful training program for early career scholars. In addition, we would love to host mid-career and more established scholars who are looking for sabbatical opportunities.

CERA: Your training is in quantitative psychology. How do you think your disciplinary perspective influences how you approach behavioral research related to genomics?

Dr. Koehly: Through my training in quantitative psychology, I developed skills in statistics and social network analysis that I apply to research that aims to understand psychological phenomena. This background provides a solid foundation for investigating the role of social context on health, particularly in the context of genomics.

My early research on how genetic information flows within families accounted for qualitative and functional aspects of family relationships and highlighted the complexity of familial communication dynamics. For example, we found that kinship alone did not fully explain dissemination patterns of genetic information, which underscored the need for a more nuanced understanding of family network systems. More recently, we demonstrated the impact of introducing genomic risk information within the family system. Family health history-based risk information can activate discussions about health risks and promote engagement in preventive behaviors. This work highlighted the practical implications of understanding and leveraging family networks in health interventions. One of our current projects focuses on how families manage caregiving roles and the potential impact of the structure and function of family networks on caregivers’ mental and physical well-being. By exploring whether certain network characteristics can serve as protective factors, or exacerbate, caregiving burden, our research addresses an important gap in the literature and contributes to our understanding of the complex interplay between social relationships and health outcomes. Overall, my interdisciplinary approach can provide valuable insights into how families communicate about and cope with genetic/genomic disease risk and diagnoses. Our findings have implications for both research and practice, informing interventions aimed at improving health outcomes by leveraging these family network systems.

CERA: What information does social network analysis offer to our understanding about how people respond genetic information?

Dr. Koehly: Social network analysis offers a holistic approach to examining how a family system responds to genetic information. By uncovering the interplay between individuals (e.g., family members), their relationships, and genetic information or disease diagnoses, this approach can provide valuable insights for designing interventions, policies, and support systems aimed at promoting informed decision-making, engagement in healthy behaviors, and resilience in the face of health challenges. 

CERA: What data collection tools do you recommend for social network analysis? If more than one, what are the best scenarios in which to use each?

Dr. Koehly: Social network analysis aims to understand the structure, function, and dynamics of networks. Traditionally, there are two approaches used in the field: ego-centered network analysis and whole network analysis. These two approaches focus on different levels of analysis—and the varying viewpoints can provide different insights.

Ego-centered network analysis, also known as personal network analysis, examines the immediate or local social environment of a particular person (or ego). This approach is particularly useful for understanding how individuals are embedded within their social contexts and how these immediate connections influence the individual’s health and well-being. Often, analysis considers the composition of these networks (e.g., characteristics of the network members, including their gender, kinship, geographic proximity), the quality of the relationships between the ego and their network members (e.g., interpersonal closeness or strain), and the function of those relationships (e.g., what resources are exchanged among them, such as genetic risk information, emotional support). We have used several different approaches to collect egocentric networks. For example, as part of a detailed family health history assessment, we capture aspects of the individual’s relationship with each family member, including the quality and function of those relationships. We also ask them to enumerate other important people to capture the role of social kin (e.g., friends, coworkers, in-laws) in these social processes. An example of a tool for doing this is the Colored, Eco-Genetic Relationship Map (CEGRM), developed by June Peters and Regina Kenen. More recently, we have used the hierarchical mapping technique (also see ^1,2) within Robert Kahn and Toni Antonucci’s Social Convoy Model. This latter approach is not anchored to biological family ties like the CEGRM, but focuses more broadly on participants’ social support system, which may or may not include family.

In contrast, whole network analysis, also known as global network analysis or network-level analysis, considers the structure of relationships or interactions amongst members of a bounded group. Using this approach, the entire network is considered the unit of analysis, with a focus on examining the overall structure, properties, and dynamics of this network rather than focusing on individuals within the group or subgroups. Whole network analysis typically involves studying network-level structural features—such as density, centralization, clustering, or modularity—which provide insights into larger patterns and structures that represent connections across communities, the flow of information, or influence processes within the network. The method used to measure whole networks depends on the scope of study and identifying the boundaries of the social system under study. Data sources may include online platforms, such as social media groups, surveys and interviews, or direct observation of interactions within groups. For example, we recently completed some family health history community education programs and are mapping the group interaction process to obtain measures of engagement for our evaluation. To summarize, ego-centered network analysis zooms in on the immediate social environment of an individual, whereas whole network analysis takes on a broader view. Both approaches can be valuable to gain a comprehensive understanding of a social system.

In our research, we focus on family systems, using this social network lens. One challenge working with family networks is that they are unbounded, limiting our ability to use a whole network approach. To address this challenge, we use a multi-informant approach where we ask multiple family members to map the composition, quality, and function of their personal networks, and then we link these ego-centered networks across participants within the same family. This allows us to consider both the immediate social environment within which each family member is embedded, in addition to broader structural features arising when these egocentric networks are linked. In a recent paper, Jasmine Manalel and I describe this approach as interconnected convoys.

CERA: Do you have any advice related to either the sampling plan, recruitment, data collection, or analysis for ELSI researchers who may be considering adding social network analysis to their methodological toolkit?

Dr. Koehly: Integrating a social network perspective into scientific inquiries requires a shift in focus from individual attributes to the relationships and connections between individuals. The methodologies used for the sampling plan, recruitment, data collection, and analysis will largely depend on the scope of the research. However, here are a few thoughts based on some of our projects:

• Sampling plan: The sampling plan is primarily shaped by the specific research questions being asked. A critical consideration is whether the focus is on the immediate social environment surrounding an individual or on the broader social network features. For instance, in our Families SHARE (Sharing Health Assessment and Risk Evaluation) evaluation, our aim was to explore whether participants engaged in discussions about our workbook designed to help people assess their risk for several conditions using their family history and collected family health history information within their immediate family network. To address this, we used a standard single-informant, ego-centered network design, sampling individuals within our focal communities. Conversely, in Project Risk Assessment for Mexican Americans (Project RAMA) our interest was in understanding the intergenerational transfer of family health history information. Thus, we sampled ‘family triads’, which primarily consisted of parents and their adult children. This approach enabled us to examine potential variations in family responses to four different family health history feedback conditions and assess whether patterns of intergenerational information transfer resulting from these conditions correlated with updates to family health history information over the course of the intervention. In other studies where our focus is on understanding diffusion of information within the broader family system or capturing diverse perspectives on experiences for individuals caring for someone living with a genetic disease, we have used a snowball sampling design, which involves asking participants to refer other family members to participate in the study. It is worth noting that, on average, our team finds snowball sampling successful in recruiting multiple informants within a family approximately 50% of the time. We have the highest success when working with established research cohorts where relationships have been built through long-term partnerships between researchers and cohort members
• Recruitment: We recruit across diverse settings by fostering partnerships with extramural collaborators, clinician-scientists, patient advocacy groups, and community partners. One of our key priorities is to actively involve individuals from communities that have been historically underrepresented in biomedical research. To ensure success in this endeavor, we prioritize the assembly of a project team that is representative and inclusive.
• Data collection: Collecting network data can be time-intensive. Typically, the assessment involves two components. First, participants are tasked with enumerating members of their network and providing relevant information about each individual member. Often, we build this enumeration process around a family health history assessment. Second, participants are asked to characterize the quality and function of their relationships with each network member. To reduce participant burden, researchers should carefully consider the number of relational variables being asked and focus primarily on aspects that are related to their study objectives. Although surveys have been effective for collecting network data, we find that the interview format can offer valuable qualitative insights into questions about the quality and function of relationships and provide participants with the opportunity to introduce new network members in response to the questions, thereby reducing the risk of overlooking key individuals. As a result, we have developed an approach where participants enumerate their network members through survey and elaborate on the quality and function of those relationships during interviews.
• Analysis: Once collected, the data need to be prepared for analysis. In social network analysis, the focus is on relationships or connections between people. Given this, social network data are often stored in two different files: an attribute file and an edge file. The attribute file contains information about each network member, such as their demographic characteristics. An edge file contains information about the relationships between the participant and each of their network members. This separation enables researchers to clearly distinguish between individual characteristics and relational data, making the dataset more manageable and facilitating analysis. These files are linked through unique IDs for each participant and each unique network member. If network data has been collected from multiple family members, then network members across informants need to be linked by using the same unique ID. This linkage enables structural analysis of the broader network system instead of analysis that considers each informant’s network in isolation. 

CERA: We listened to the Genetics Unzipped podcast and read the NIH press release about your Families SHARE workbook, which you developed to help people assess their risk for heart disease, diabetes, breast cancer and colorectal cancer using their family history. The tool was initially developed with an educated, middle class, white sample, but you refined it following a formal assessment of its value and usability for low-income African Americans in the Washington, D.C. and Baltimore area. For other researchers who might be considering a similar undertaking, could you please give us a sense of some of the necessary updates the study pointed out to you?

Dr. Koehly: There have been several updates to our Families SHARE toolkit based on our evaluations, including modifications to the workbook and new components.

• We expanded the workbook to include prostate cancer, particularly given its impact on individuals who identify as Black or African American. This community is disproportionately affected by the disease, and our efforts are aligned with serving their needs.
• We translated the workbook from English into Spanish, Haitian-Creole, and Hausa to expand utilization into other communities. 
• In response to our evaluation within low-resourced Black communities in D.C. and Baltimore, we developed a community education program. Our participants emphasized the need for such a program, indicating it would offer an opportunity to address questions and foster supportive relationships among community members using the workbook. Importantly, we learned that often the experts in the room are our participants, who can share strategies for engaging in conversations about family health history and disease risk and encouraging health promoting behaviors within their families. To support this initiative, we developed a moderator’s guide and, in partnership with Georgetown University’s Office of Minority Health and Health Disparities Research, piloted the program in 2021.
• Participants in the community education program highlighted the importance of ongoing education to reinforce the learning objectives of the Families SHARE toolkit. In response, we created a companion video that walks users through the workbook. This video includes interactive knowledge checks with immediate feedback for users. Moreover, the video is available in both English and Spanish. We are currently piloting the video with our partners at Georgetown University.
• We developed a curriculum guide to facilitate the integration of Families SHARE into the Community Engaged Clinicians program at Florida International University as an educational module. This guide is designed to assist clinical teams in effectively utilizing the Families SHARE workbook with families from diverse cultural backgrounds.
• Soon we will be taking Families SHARE to Nigeria. In collaboration with our partner at Bayero University, we adapted the workbook to ensure cultural appropriateness for Nigerian families. Additionally, we developed three workshops aimed at introducing Families SHARE to key community partners in this initiative: community health workers, schoolteachers, and community women’s groups. Our objective is to engage families through their healthcare providers, involve children through science curriculum integration, and empower mothers and grandmothers to curate their family’s health history and foster healthy home environments.

CERA: Do you have any advice for researchers who are thinking about adapting an educational resource or data collection tool for use with a new population?

Dr. Koehly: We entered each evaluation and adaptation of the workbook through collaboration with our academic and community partners and a focus on incorporating feedback from the communities involved. This approach is essential for creating effective and culturally sensitive resources, and by actively engaging with members of the communities we serve, we can ensure that components of the tools we are adapting address real needs and reflect the diverse perspectives and voices within those communities. Doing so not only helps in developing a more accessible resource, but also cultivates trust and ownership among our partners that we hope will lead to continued utilization of the toolkit.

CERA: Do you have any advice about identifying a journal or publishing formal evaluations or validations of educational or data collection tools?

Dr. Koehly: This is always a challenge! We try to locate publication outlets that reach audiences who can integrate our findings into their own work. Our approach typically involves examining the publication outlets of the papers we cite or papers that informed our work. We then assess the scope of these journals to determine if they align with the aims of our paper. If this initial exploration doesn’t lead us toward a suitable outlet, we often brainstorm within our team and with colleagues to explore alternative possibilities.

If you would like to be featured in the ELSI Research Methods Spotlight, please email the CERA Team at [email protected].

 

---

¹Kahn, R. L., & Antonucci, T. C. (1980) Convoys over the life course: Attachment, roles, and social support. In P. B. Baltes & O. G. Grim (Eds.), Life Span Development and Behavior (Vol. 3, pp. 253-286). Academic Press.
²Antonucci, T. C. (1986). Measuring social support networks: Hierarchical mapping technique. Generations: Journal of the American Society on Aging, 10(4), 10-12.