SDoH data infrastructure

Building robust data infrastructure and data ecosystems is a critical first step for SDoH research and practice. While SDoH factors are often categorized into two main categories – contextual SDoH (i.e., neighborhood-level or place-based factors) and individual-level social risks and needs – the integration of this information with existing healthcare data presents both challenges and opportunities [Reference Guo, Bian and Wang7,Reference Zhang, Hu and Diller8].

Contrary to the initial oversimplification, many healthcare systems have made significant strides in linking EHR data with area-level SDoH information. However, the seamless integration of this data into clinical care workflows and decision-making processes remains an area for improvement. Similarly, while an increasing number of clinics and hospitals are now screening patients for social risks and needs, the comprehensive integration and utilization of this information across the healthcare system is still evolving.

The current special issue accepted a total of 7 papers [Reference Tilmon, Nyenhuis and Solomonides9–Reference Charlson, Wells and Hollenberg15] dedicated to SDoH data, with 5 specifically focused on data infrastructure building and tool development. These contributions represent significant advancements in efforts to build SDoH data infrastructure, offering new opportunities to enhance our ability to address health disparities via SDoH research and practice. Craven et al. [Reference Craven, Highfield and Basit16] introduced their collaborative efforts on building a SDoH data infrastructure across four Texas Clinical and Translational Science Award (CTSA) institutions, promoting SDoH data standardization, harmonization, and integration into EHRs. Similarly, Abdallah et al. [Reference Abdallah, Bahroos and Angyan14] introduced their procedure for integrating social and environmental determinants (i.e., contextual SDoH) by geocoding patients’ residential addresses to geo-enrich an EHR data warehouse in a secure and compliant manner.

Three papers report on their SDoH data platform products. Tilmon et al. [Reference Tilmon, Nyenhuis and Solomonides9] presented the Sociome Data Commons (SDC), a cloud-based repository of harmonized, geospatial sociome data sets. This platform, designed to work alongside clinical data, predicts a variety of health outcome by characterizing local social, environmental, behavioral, psychological, and economic exposures. The SDC adheres to principles of governance, sustainability, inclusiveness, disaggregation, and multidisciplinary management as well as the Findable, Accessible, Interoperable, and Reusable (FAIR) principles for its data products. As a use case, they investigated social factors associated with asthma exacerbations in children residing on the South Side of Chicago, using machine learning and six SDC datasets. Holt et al. [Reference Holt, Roman, McIntosh, Kleinsorge, Holden-Wiltse and Bennett10] introduced RocHealthData.org, a web-based tool providing access to thousands of publicly available national datasets. The platform already has 1033 registered users, with 28.2% from the community. Adams et al. [Reference Adams, Gasman, Beccia and Fuentes11] described the development and evaluation of “Health Equity Explorer (H2E),” an open-source, scalable informatics tool designed to support dynamic and interactive explorations of diverse health drivers compatible with RWD structured in the Observational Medical Outcomes Partnership Common Data Model.

Two additional papers by Bensken et al. [Reference Bensken, McGrath, Gold and Cottrell17] and Telzak et al. [Reference Telzak, Levano, Haughton, Chambers and Fiori12] are focused on examining the correlation between contextual-level SDoH and individual-level SDoH. Although both studies showed suboptimal associations, the study by Telzak et al. [Reference Telzak, Levano, Haughton, Chambers and Fiori12] suggested the strength of these associations and predictive values were better in areas with high SDoH risk compared to areas with low SDoH risk.

These contributions represent significant advancements in efforts that build SDoH data infrastructure, offering new opportunities to enhance our ability to address health disparities via SDoH research and practice.

SDoH tool development

Development of tools plays a crucial role in addressing SDoH by providing innovative solutions to identify, measure, and analyze the complex SDoH factors influencing health outcomes. The papers featured in this special issue highlight a range of notable tools and methodologies designed to tackle various aspects of SDoH [Reference Bensken, McGrath, Gold and Cottrell17–Reference Li, Mowery and Ma24]. For example, German et al. [Reference German, Stirling and Gorgone21] presented an interactive visualization tool that integrates clinical, sociodemographic, and environmental data to enhance understanding of health disparities in diabetes care and outcomes. Harris et al. [Reference Harris, Fu and Wen20] introduced a Natural Language Processing (NLP) tool for identifying housing instability and supporting clinical trials related to housing issues. Huang et al. [Reference Huang, Socrates and Gilson23] developed advanced NLP models for identifying incarceration history, an important SDoH factor, from unstructured clinical notes, opening new avenues for targeted interventions and research. Resnick et al. [Reference Resnick, Montella, Brown and Elkin18] developed a terminological representation of the Assessing Circumstances & Offering Resources for Needs SDoH survey, addressing a gap in representing SDoH with standardized terminologies and enabling its use with NLP and clinical decision support systems. Bensken et al. [Reference Bensken, McGrath, Gold and Cottrell17] assessed the predictive ability of area-level SDoH measures for individual-level social risks and explored biases in social risk screening. Espinoza et al. [Reference Espinoza, Sehgal and Phuong19] proposed a Social and Environmental Determinants of Health (SEDoH) Informatics Maturity Model to help organizations assess their current level of SEDoH informatics maturity and identify areas for improvement. Ngai et al. [Reference Ngai, Maroko, Bilal, Wilder, Gordon and Richardson22] developed a method to measure concordance and discordance between individual characteristics and corresponding neighborhood-level SDoH, enhancing our understanding of their interplay and impact on health. These tools and methodologies have the potential to advance clinical and translational research by enabling researchers to better understand the complex interplay between SDoH and health outcomes, ultimately informing the development of targeted interventions to address health disparities.

SDoH associated tool implementation and knowledge generation

The landscape of SDoH research and practice is undergoing a transformative shift, driven by innovative tools and technologies that enhance our understanding and ability to address health disparities. The papers featured in this special issue showcase notable implementations that are pushing the boundaries of SDoH informatics and intervention strategies [Reference Fiori, Levano and Haughton25–Reference Parsons, Haynes and Blaes35]. From comprehensive organizational assessment frameworks [Reference Fiori, Levano and Haughton25,Reference Orakwue, Hing and Chantarat31] to cutting-edge NLP techniques [Reference Huang, Socrates and Gilson23], these advancements represent a significant leap forward in our ability to capture, analyze, and utilize SDoH data effectively.

Particularly noteworthy is the trend toward leveraging technology to overcome longstanding barriers in SDoH research and intervention. The success of remote enrollment protocols in increasing participation among underrepresented groups in clinical trials [Reference Ritchie, Turk, Holtrop, Durfee, Dickinson and Kaufmann36] is a prime example of how digital approaches can enhance inclusivity and representation. These implementations, along with efforts to standardize SDoH terminology [Reference Parsons, Haynes and Blaes35] and assess organizational readiness [Reference Espinoza, Sehgal and Phuong19], collectively point to a future where SDoH data is not just collected but also seamlessly integrated into clinical decision-making and population health strategies. As we move forward, it is imperative that healthcare systems and researchers embrace these tools and technologies, recognizing their potential to drive more equitable health outcomes and inform evidence-based policies to address the root causes of health disparities.

Summary and looking forward

Collectively, this body of work underscores the importance of leveraging cutting-edge informatics methodologies for building SDoH data infrastructure, as well as developing and implementing tools to address SDoH and reduce health disparities. Figure 1 provides a comprehensive overview of the papers related to informatics and data infrastructure that were accepted for the SDoH special issue in the Journal of Clinical and Translational Sciences. Utilizing a Learning Health Systems framework [Reference Friedman37], Figure 1 illustrates how these contributions collectively advance the integration of SDoH into clinical and translational sciences. The inner circle depicts a continuous cycle of Data, Knowledge, and Performance, which encompasses the broad topics covered in the special issue.

Figure 1. Social determinants of health-driven translational sciences: a learning health cycle from data to action.

As we reflect on the diverse range of studies presented in this special issue, several key themes emerge that will shape the future of SDoH research and practice. The importance of robust data infrastructure and standardization is evident across multiple studies, providing a solid foundation for more comprehensive analyses of health disparities. The development of sophisticated tools and methodologies, particularly in NLP, has also emerged as a significant trend. These studies not only highlight technical solutions but also raise important questions about data privacy and the ethical use of information.

The focus on implementation and real-world assessment underscores the importance of bridging the gap between theoretical models and practical applications. The studies presenting “best practices” and use cases offer a roadmap for institutions and communities looking to enhance their SDoH-related initiatives. Interestingly, while our initial call emphasized AI and machine learning methods broadly, the contributions we received primarily focused on NLP applications, suggesting that text-based data sources may be a particularly rich resource for understanding and addressing SDoH.

When examining the connections between seemingly disparate manuscripts, we see a cohesive story emerging. For instance, the work on data infrastructure by Craven et al. [Reference Craven, Highfield and Basit16] and Abdallah et al. [Reference Abdallah, Bahroos and Angyan14] provides the foundation upon which tools like the Health Equity Explorer by Adams et al. can be built [Reference Adams, Gasman, Beccia and Fuentes11]. These tools, in turn, inform implementation strategies like those described by Fiori et al. [Reference Fiori, Levano and Haughton25] and LeLaurin et al. [Reference LeLaurin, De La Cruz and Theis26]. This progression from data to tools to implementation illustrates the interconnected nature of SDoH research and the importance of a comprehensive approach.

Moreover, the studies by Bensken et al. [Reference Bensken, McGrath, Gold and Cottrell17] and Telzak et al. [Reference Telzak, Levano, Haughton, Chambers and Fiori12] highlight the complexity of these social factors and the need for nuanced approaches in both research and clinical practice. This complexity is further explored in the implementation studies, such as Miller et al.’s work in a free clinic setting [Reference Miller, Doucas and Karra27], demonstrating how SDoH research can directly impact patient care in diverse healthcare environments. Li et al. conducted a scoping review of approaches for integrating SDoH data into EHR systems, offering a comprehensive overview of current practices and future possibilities for leveraging this data in clinical settings [Reference Li, Mowery and Ma24].

The field of SDoH research is continually evolving, and as we consider its future, it’s imperative to acknowledge the critical role of community engagement, as highlighted by Cottler [Reference Cottler6]. Community engagement provides insights beyond what can be gleaned from medical records alone, ensuring that SDoH initiatives address the actual needs and priorities of the entire community, not just those represented in medical databases.

As we conclude this special issue, we are filled with a sense of optimism, tempered by the recognition of the significant work that still lies ahead. We hope this collection serves as a catalyst for ongoing innovation, collaboration, and progress in the vital field of SDoH. The integration of advanced informatics methodologies, robust data infrastructure, and community-engaged research holds the promise of creating truly transformative solutions for addressing health disparities.

We urge researchers, clinicians, informaticians, policymakers, and community leaders to embrace the challenges and opportunities that come with SDoH research. Let us commit to fostering interdisciplinary collaboration, prioritizing community engagement, and advocating for policies that address the root causes of health inequities. Together, we can build a more equitable, effective, and compassionate healthcare system that truly serves all members of society.