Customize your JAMA Network experience by selecting one or more topics from the list below.
The scope and types of data available in social networks represent growing potential resources for deciphering and addressing the social determinants of health. Between 2005 and 2018, the use of social media among individuals in the United States increased from 5% to 69%.1 In 2016, the Pew Research Center found that Facebook was the most popular platform, used by 68% of the US population (76% of whom visited the site daily, engaging for nearly an hour a day), followed by Instagram, Pinterest, LinkedIn, and Twitter.1 Social media user profiles represent archives of people’s activities and connections, curated in real time. Furthermore, the relationships between and among users are dynamic social network structures that intertwine people, places, and their interests.2 This information has the potential to improve understanding of the social determinants of health in a deeper way than prior data sources have ever allowed, and, in turn, identify novel targets that could meaningfully influence health outcomes.
Remarkable advances in medical science, clinical care, and therapeutics over the last 60 years have established the current understanding of the “nature” side of disease. The “nurture” components of disease also have been explored, revealing strong associations between social support and health outcomes. However, prior investigations into social determinants have often been limited by self-reported information based on reductionist instruments with standardized responses. Moreover, social determinants are complex, and entail networks and behaviors that are best revealed by what actually occurs in life, rather than the perception of these complex relationships. Individual and network data available within social platforms therefore have the potential to elucidate the understanding of social determinants of health and could offer measurable, actionable insights into how disease can be prevented. To date, only limited direct links between medical and complex social network data have been made.
The roots of interest in social determinants of health can be traced back in part to the World Health Organization (WHO), which, in 1946, notably defined health as “complete physical, mental, and social well-being.”3 Since then, a number of national and international efforts have increased awareness of the field, such as the WHO’s Commission on Social Determinants of Health and initiatives by the MacArthur and Robert Wood Johnson Foundations.4-6 From a research perspective, numerous studies have shown associations between social determinants of health and life expectancy.7 While these studies reflect the overall importance of social determinants of health in relation to health outcomes, an important limitation of current research on social determinants of health is that many of the identified factors, such as income and education, cannot easily be changed.8 This is in part because of a lack of granularity in understanding the person and his/her community, compounded by the potential flaws introduced by survey self-reported social/behavioral variables as opposed to observed factors. Access to information that captures the habits, behaviors, and networks of individuals has been limited in the existing body of work as these parameters relate to health outcomes. As such, it is not surprising that actionable variables continue to be elusive.
Better approaches are needed for accessing information about observed habits, behaviors, and networks to foundationally understand their relationship with health and health outcomes. Despite the exponential increase in the role of social media in the daily life of individuals around the world over the last decade, most studies have not directly evaluated social variables from social network sources in relation to clinical outcomes.
Evaluating social network data in combination with increasingly available digital health care data (such as from large, national clinical registry programs or electronic health records) could lead to novel, more nuanced understanding of social and behavioral variables that account for the interplay of the individual and the network in relation to health outcomes. These may transform the traditionally held social determinants of health, including education, income, housing, and community, to encompass a more granular tech-influenced definition, ranging from simple factors, such as numbers of online friends, to complex social biomarkers, such as timing, frequency, content, and patterns of posts and degree of integration with online communities. With data related to millions of users, network effects may amplify the total range of patterns and associations.
The promise of this kind of research endeavor not only involves its scale, but also involves the richness of possible connections that can be explored with machine learning and other evolving “big data” methodologies. Given that users interact daily with these platforms, the potential for intervention may exceed the immutable nature of the social factors that have been explored to date. For example, could suicidal ideation be identified at the individual and network level with enough advance warning and accuracy to stage a peer-driven intervention? What if risk-factor prediction models used to identify patients at high risk for opioid addiction could be refined sufficiently for physicians to reliably use these models as a means to revise care management plans? Could individuals at the highest risk for cardiovascular mortality be identified, and then families and close networks be tasked with responsibilities for each other using the networks to facilitate increased social integration and, thereby, reduce mortality? These data could also provide novel markers of early disease, as suggested by assessing keystroke patterns as a precursor for Parkinson disease.9
However, along with significant potential for benefit, this kind of research also harbors potential risk and clear challenges. As was done for the Human Genome Project, the first step will be to establish the legal and ethical framework for this endeavor. Social network data raise unique challenges to deidentification beyond the typical demographic identifiers. For example, when an individual posts a simple phrase, the exact composition of words used can become a form of identification in its own right; developing the techniques to deidentify this kind of data will require thoughtful approaches. In addition, a combination of physical and software-driven isolation needs to tightly control access to the data. Concurrently, investment in research towards the creation of “synthetic data sets” (ie, data sets that maintain associations but have the original data removed) may serve to advance security and privacy for the next iteration of this research.
The United States and many other countries are in national dialogues about privacy and whether data collected for personal reasons can be used in aggregate to advance knowledge in meaningful ways. The prospect of social media data propelling research into social determinants of health should serve to stimulate the development of methods that respect individuals, seek consent whenever possible, and always protect security and privacy in the quest to learn more about how best to promote health.
Once acceptable privacy and security standards have been established, the next step will be to forge the public and private partnerships necessary to link social media data with clinical and health outcomes. A number of additional steps will be required between this point and data analysis, and even once accomplished, the findings will be retrospective and associative. Nonetheless, embarking on the earliest stage of harnessing social network data is necessary to determine if those data hold the promise to advance understanding of social determinants of health in a truly novel way. When combined with deep, powerful, and interactive social tools, perhaps it will be possible to rewrite the source code for nurture.
Corresponding Author: Freddy Abnousi, MD, MBA, MSc, Facebook Inc, 1 Hacker Way, Menlo Park, CA 94025 (email@example.com).
Published Online: December 20, 2018. doi:10.1001/jama.2018.19763
Conflict of Interest Disclosures: Dr Abnousi is the Head of Healthcare - Research at Facebook Inc. Dr Krumholz reports contract work for the Centers for Medicare & Medicaid Services, grants through Yale from the US Food and Drug Administration, Medtronic, and Johnson & Johnson, serving as the chair of the Cardiac Scientific Advisory Board at United Health, the Life Sciences Board at IBM Watson Health, the advisory board at Element Science, the Physical Advisory Board on Aetna, and is the founder of Hugo. Dr Rumsfeld reports no conflicts of interest to disclose.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Abnousi F, Rumsfeld JS, Krumholz HM. Social Determinants of Health in the Digital Age: Determining the Source Code for Nurture. JAMA. Published online December 20, 2018. doi:10.1001/jama.2018.19763
Create a personal account or sign in to: