Social determinants of health (SDOH) and social risk have long been key considerations in real-world studies focused on medication adherence and health outcomes. Typically, however, research has evaluated either individual variables – for example age, race and ethnicity – or other publicly available community assessment tools.
This is starting to change as the traditional approach fails to capture nuances across communities and populations. A recent seminal retrospective cohort study investigating medication adherence in patients with atrial fibrillation (AF) utilized domain level social risk data to capture the contours and concentrations of social risk across the United States. The study, conducted in collaboration with Pfizer, leveraged Socially Determined’s community level social risk scores across five core domains: economic climate, food landscape, housing environment, transportation network and health literacy. It combined these with third party claims data and other individual level variables from that third party data asset, such as sex, age, geographic region, insurance type, and comorbidities. Related to insurance type, nearly 95% of patients included in the study were commercially insured, demonstrating that insurance is not the determinant to adherence.
What separates this study from previous health outcomes studies is that it delves deeper into the link between individual patient and population data, identifying gaps and allowing life science organizations and healthcare provider organizations to adopt a more informed, actionable approach to their patient care strategies.
Understanding adherence through data
A primary objective of the study was to understand the connection between social determinants of health and medication adherence for patients with AF. The study found that “elevated levels of health literacy risk at the community level were associated with lower odds of being persistent on a prescribed oral anticoagulant for atrial fibrillation at 90 and 180 days and of having a proportion of days covered of ≥0.8 after controlling for individual risk factors”. In other words, the analysis revealed that low health literacy is associated with poor adherence.
While healthcare experts have long suspected a link between health literacy and adherence, having the data that demonstrates the correlation is important and specifies data-driven rationale for focusing efforts around health literacy.
Approaching social risk at a domain-specific level provides insight into the unique factors influencing that risk and outcomes, which helps to inform investment, intervention design, measurement and the ability to scale programs that are most likely to have an impact.
To really understand issues on the ground, the aforementioned risk scores are calculated at a
specific geospatial unit represented as a hexagon, where each hexagon is 200 meters or 400 meters in diameter, depending on the population density of that area. For the purposes of the study, the data was aggregated to a three-digit ZIP code level to align with the granularity of the claims data. For context, a three-digit ZIP code level represents a relatively large geographic area like a large portion of a major city. It is noteworthy that the correlation between health literacy and adherence is evident even at the ZIP3 level. More granular research focused on narrower geographic units or at the de-identified individual level would inevitably provide even more powerful insights into the variation of social risk to guide action.
These insights into cohorts of patients could revolutionize how life sciences companies and
healthcare organizations identify and develop health care provider engagement initiatives, patient engagement and education programs, community-based collaboration initiatives and more.
Taking social determinants of health data to the next level
Interest in examining the impact of SDOH and social risk factors is growing. The AF study illustrates the opportunity to take analysis beyond those traditional socio-demographic data elements, often categorized as SDOH data, to understand the true community level attributes and possible socio-ecological factors that might be barriers to adherence.
It is often stated that 80% of health outcomes are influenced by factors other than clinical care. But the life sciences and broader healthcare industries tend to rely on clinical data from claims history as an example to tell the story of a person, which provides no insight into the myriad of life factors that inevitably impact health outcomes.
Domain-specific social risk scores and associated contextual data both for more defined geographic areas and individuals provide full visibility into the factors that may be contributing to outcomes. With deeper insights, healthcare decision-makers can take a more holistic, equitable, actionable approach to improving adherence, rather than relying solely on broad demographic data that doesn’t provide the full picture of the issues faced.
Social risk data continues to evolve, paving the way for a more comprehensive, and nuanced,
understanding of how to address social risk and social determinants of health consistently and efficiently to ultimately improve outcomes and advance equity for communities and individuals.