Leveraging health informatics to foster a smart systems response to health disparities and health equity challenges

Abstract

Informaticians are challenged to design health information technology (IT) solutions for complex problems, such as health disparities, but are achieving mixed results in demonstrating a direct impact on health outcomes. This presentation of collective intelligence and the corresponding terms of smart health, knowledge ecosystem, enhanced health disparities informatics capacities, knowledge exchange, big-data, and situational awareness are a means of demonstrating the complex challenges informatics professionals face in trying to model, measure, and manage an intelligent and smart systems response to health disparities. A critical piece in our understanding of collective intelligence for public and population health rests in our understanding of public and population health as a living and evolving network of individuals, organizations, and resources. This discussion represents a step in advancing the conversation of what a smart response to health disparities should represent and how informatics can drive the design of intelligent systems to assist in eliminating health disparities and achieving health equity.

1. Defining collective intelligence in public and population health management

In 2007 Carley et al. published a discussion on social computing entitled, “[Moving] from Social Information to Social Intelligence” [1]. In a complementary discussion, the same author examines what “smart agents and organizations of the future” would look like [2]. Each of these discussions explore a common theme: understanding the fields of computer science and informatics as drivers of individual and collective intelligence in fields such as public and population health [1,2]. IBM also addresses the topic of health system-derived intelligence in a report entitled, “Clinical Decision Intelligence: Medical Informatics and Bioinformatics Infrastructure for a Clinical-Decision-Intelligence System” [3]. While individual intelligence is an established construct in the literature, collective intelligence remains an evolving concept. As a result, the notion of a “smart public or population health organization” remains equally ambiguous in nature [4]. If informatics professionals should lead the charge in (1) fostering collective intelligence, (2) designing smart public and population health organizations, and (3) managing highly adaptive health systems, what measures of progress should we rely upon, and how might this change our trajectory in our approach to dealing with complex health issues like health disparities and achieving health equity?

First, we must define the term collective intelligence. Carley et al. provide a set of core competencies for informatics professionals who are attempting to model, measure, or manage collective intelligence in any public or population health setting [1]. According to Carley, some measure of collective intelligence is achieved by (1) modeling and analyzing social behavior, (2) capturing human social dynamics (e.g., within a given workflow or clinical pathway), (3) creating artificial social agents to represent both classes of individuals (e.g., providers, administrators, families, public health practitioners, policy makers, researchers) and their roles in the care process (e.g., patient navigators, care coordinators, specialty care, support service providers, public health program managers), and (4) generating and managing actionable social knowledge (i.e. how individual-tacit knowledge [an individual’s know-how or expertise] is converted into shared explicit knowledge [organizational knowledge products and artifacts]). While these descriptions still do not fully explain what collective intelligence is, they move us closer toward developing a toolkit to assess social or organizational aspects of a public or population health system. Essentially, we want to better understand how to move from a massive array of individual (person-level) patient encounters and experiences—that are typically captured in tacit forms—to collective (population-level) explicit knowledge about risk, causality, patterns of incidence and prevalence, and interventions that shape public and population health outcomes.

A critical piece in our understanding of collective intelligence for public and population health rests in our understanding of public and population health as a living and evolving network of individuals, organizations, and resources [5]. Carley explains how network analysis is critical in helping to capture, represent, and analyze social interactions, organization connections, and the flow of informational and knowledge resources among public and population health stakeholders (also referred to as agents). Thus, the characterization of social structure and relations such as organizational and social networks—typically represented via nodes and ties in network representations—are critical in an informatics professional’s arsenal when conducting research and leveraging collective intelligence to address complex healthcare delivery issues, such as ameliorating health disparities in an effort to achieve health equity. The goal of this discussion is to better understand the data, information, and knowledge demands informatics professionals might face in meeting the challenges associated with health disparities and health equity in public and population health settings.

2. Understanding health disparities and health equity terminology

To date, there is no global consensus on how to define “health disparities” or “health inequities”. While some define health disparities as differences in health outcomes (e.g., mortality, morbidity, burden of disease) between some population sub-groups, others have critiqued definitions such as these as they do not address the “systematic, plausibly avoidable health differences adversely affecting socially disadvantaged groups” [6]. This social disadvantage “refers to the unfavorable social, economic, or political conditions that some groups of people systematically experience based on their relative position in social hierarchies” [6]. In other words, disparities are not just simply differences in health behaviors and outcomes between some groups – they are differences that are systematically determined by one’s social place which is in turn determined by overarching systems of power [7,8].

Related to issues of defining health disparities, indicators used to measure and inform health disparities are also of mixed efficacy and reliability, and this may impact comparisons and interpretations of disparities between groups [9]. For example, quality of life may be measured several different ways, including using disability-adjusted life years or quality-adjusted life years [10]. Furthermore, some racial and ethnic groups are often aggregated together for ease of data collection and analysis which can lead practitioners, researchers, and policymakers to overlook health disparities that may exist within socially defined groups [11]. Data beyond just mortality should be measured, as there are several different types of social and physical determinants to health that could be driving health disparities among and between some populations, such as the built environment (e.g., safe housing, access to sidewalks), availability and access to resources (e.g., access to quality healthcare), community empowerment, education, access to employment, and systemic discrimination [10]. For example, much of the literature on disparities shows that there are educational and socioeconomic differences in health among some populations (e.g., smoking prevalence is higher among those with lower education compared to those with higher education) [9,10,12]. Additionally, some studies have shown that perceived racial and ethnic discrimination is associated with less health-seeking behavior and less preventive screenings [13,14].

As much of the existing health disparities data is collected at a larger, population-level, this data may often take significant time to actually be disseminated into the field. There remains a gap in the healthcare field of how to apply this data appropriately within the healthcare sector in communities. The health informatics field is in a potentially influential position to help contribute additional knowledge in order to help better assess and understand disparities faced by some population groups. Specifically, if designed and implemented in a “smart” fashion, public health agencies and population healthcare delivery systems can serve as knowledge coalescing centers of learning about the determinants of health disparities. This more robust stream of data, information, and knowledge can then be used to better address health disparities and provide more equitable healthcare.

Population health informatics tools, such as electronic health records, should be motivated to add a social assessment in order to capture these potential determinants and risk factors that may lead to additional health risk factors and poor health outcomes, especially for vulnerable groups. The collection and analysis of this data in tandem with larger, national datasets on health behaviors and outcomes, could help healthcare management, providers, patients, and policymakers better identify health disparities and even detect internal prejudices (e.g., implicit bias) while also potentially assessing what determinants are associated with and potentially driving these disparities. For example, after implementing a social assessment, a clinic manager may assess the data and find that there is a high appointment no-show rate among some patients. Further analysis may reveal a pattern in patient reported outcomes responses that indicate that these patients typically do not own a car or lack reliable public transportation to meet appointment demands. This type of data triangulation from sources such as scheduling logs, patient reported outcomes, utilization tracking systems, and public transit mapping systems may assist practitioners and policymakers in then designing population-level and/or high-risk health approaches to help ameliorate some of the health determinants and disparities observed [15,16].

3. Big-data challenges in health disparities and health equity

Some have defined informatics as the science of applying information-age technology to serve the specialized needs of one of several practice domains [17]. These domains can include any category of practice that comprises a public health and population health care delivery system. Hence, the domains of informatics range from public/population health informatics (which examines systems-level/population-level informatics technology practices) to biomedical informatics (which examines the information technology required to understand microcellular systems dynamics in the search for new drug discoveries, enhanced treatments, and new methods of screening and detection). Within this discussion, we will limit our conversation to four distinct domains of informatics practice which we deem critical to eliminating health disparities. These domains include, public health informatics, population health informatics, community health informatics, and consumer health informatics. Table 1 summarizes our current understanding of the landscape of issues in health disparities/health equity by each informatics domain. The table highlights what we imagine to be the critical components of a centralized knowledge-base for health disparities- and health equity- related decision making across multiple categories of stakeholders. The variety of coalescence of health-specific data sources and tools listed in our table represents a “big-data” challenge related to the integration, interoperability, and exchange needed to foster analytics, knowledge synthesis, and informed decision making. We view this big-data challenge as critical in meeting public health objectives (e.g., Healthy People targets), population health objectives (e.g., patient centered medical homes), community health objectives (e.g., needs assessments), and consumer health objectives (e.g., self-management).

Table 1.

Informatics domains, health equity challenges, and informatics strategies.

Informatics Domains	Health Disparities Elimination and Health Equity Achievement Challenges	Information Technology Tools, Data Sources, and/or Strategies
Public Health Informatics	• Establish national guidelines, standards, and objectives for health disparities and health equity (e.g., Healthy People 2010 objective) • Foster timely translation of research evidence into practice through national health information clearinghouses • Evaluate state and local health department health disparities and health equity policy and practices	• National surveillance and disease registries • State and local health department surveys • Public health program reporting • Public health research data hubs
Population Health Informatics	• Foster equitable and empowering patient-provider communication • Emphasize primary care and preventive care and reduce the use of emergency care use for non-emergencies • Increase coordination of care through the use of Patient-Centered Medical Homes (PCMHs)	• Electronic Health Records (EHR) • Health Information Exchanges (HIEs) • Clinical data warehouses • Regional research data collaborations • Hospital-based event and disease tracking through registries (e.g., medical errors, quality, safety)
Community Health Informatics	• Leverage community-based participatory research hubs and community partnerships • Establish multi-organizational coalition network partnerships (e.g., community, academic, corporate, and government) to address collective large-scale challenges	Community Health Needs Assessments (CHNA) • Community health worker summaries • Community-based knowledge exchange networks
Consumer Health Informatics	• Increase health literacy • Increase access to care • Increase emphasis on patient self-management and shared decision making	• Patient reported outcomes • Patient health surveys • Patient advocacy (e.g., lay health advisors, social workers) • Electronic patient portals • Patient Health Records (PHRs)

4. Revising the health disparities informatics strategy

In the 2009–2013 Strategic Plan, the National Institutes of Health (NIH) called for an enhanced health disparities informatics capacity [18]. The plan proposed the need for a greater or enhanced health disparities informatics capacity, one that demands we evolve health disparities research from qualitative to quantitative approaches through the use of new computational tools, increased interconnectedness, and the design of a knowledge environment that maximizes collaboration within and across institutional boundaries [18]. An increased application of health information technology “could also enable enhanced characterization of the causes and determinants of health care disparities; the design of novel and more effective clinical and behavioral health care interventions; and improvements in current interventions” [18]. Such a plan requires an operational definition that can facilitate ideas to bridge the various domains of informatics practice that make up the health disparities information ecosystem and to facilitate a smarter and more connected public and population health network of individuals, organizations, and resources working in partnership to eliminate health disparities.

Public Health Informatics Challenges –

Public health informatics represents the highest level of aggregation in our view of systems behavior, and it has typically been defined as “the systematic application of information and computer science and technology to public health system and practice, research, and learning” [17,19]. Here, the primary informatics challenge remains focused on ensuring national-, state-, and local- level public health agencies and practitioners are guided by policies and have adequate resources to respond to the national challenges. The Association of State and Territorial Health Officials (ASTHO) is a national non-profit representing public health agencies in the United States (U.S.), the U.S. Territories, and the District of Columbia, and includes over 100,000 public health professionals. In 2014, ASTHO conducted a Health Equity Minority Survey to examine each of its member states efforts in addressing minority health, health disparities, and health equity (MH-HD-HE) within their strategic plans across 10 primary measures [20]. The ASTHO MH-HD-HE capability measures and their respective averages across all participating member partners (n = 49) examined the ability to: (1) leverage and engage partners (87.8%), (2) establish policy to require health equity focus in all funding opportunities (55.1%), (3) develop health equity communication strategy (71.8%), (4) partner to enhance multi-sector capacity (83.7%), (5) develop multi-sector advocacy strategy (34.7%), (6) ensure health equity is integrated in state strategic priorities and plans (77.6%), (7) increase access to primary care (67.3%), (8) increase cultural competency/health literacy (87.8%), (9) collect and track disparities data (91.8%), and (10) increase health workforce diversity (69.4%) [20].

The ASHTO survey also examined state characteristics with respect to patterns of leadership and funding dedicated to addressing minority health, health disparities, and health equity [20]. In addition to the MH-HD-HE snapshot, states and territories typically develop their own periodic status reports on the state of disparities in their respective areas. The core public health informatics challenge in eliminating health disparities is to assume stewardship over the data, information, and knowledge systems that are required to properly synthesize streams of data from national, state-wide, regional, and local stakeholders in an effort to maintain an accurate assessment of an area’s public health status. This also includes the identification of best practices, the fostering of data sharing and information exchange protocols, the establishment of national benchmarks, and the development of integrated data and research hubs, clearinghouses, and translational communication networks that foster rapid learning and responses to national and local challenges.

4.1. Population health informatics challenge

Population health informatics—a close relative of public health informatics—”comprises organized activities for assessing and improving the health and well-being of a defined population. Population health is practiced by both private and public organizations. The target [priority] ‘population’ can be a specific geographic community or region, or it may represent some other ‘denominator,’ such as enrollees of a health plan, persons residing in a provider’s catchment area, or an aggregation of individuals with special needs” [21]. Here, we are using population health informatics to refer to a specific patient population of any given healthcare delivery setting. Thus, for this discussion, the population health informatics professional is focused on the task of supporting health disparities elimination and promoting health equity within a given patient population. However, because the determinants of health disparities and health inequalities typically fall outside the realm of traditional healthcare delivery (e.g., genetic pre-disposition, SES, environmental justice, etc.), we want to emphasize a sub-category of health disparities: performance-based health disparities, which we deem well within the strategic parameters of healthcare delivery. Operationally, we define “performance-based” as the form of health disparities that are directly or indirectly associated with decisional events that shape the delivery of care from early prevention and health promotion communication efforts to end-of-life measures. Along the continuum of care, decisions/actions are made/taken by individuals, referred to as actors or agents. Such interactions among agents may lead to actions or choices that serve as trigger events of an adverse health outcome (e.g., racially divisive comments/behaviors from healthcare professionals, denial of care, non-adherence to medication protocols, failure to screen eligible populations, loss to follow-up care, compromised patient-provider communication, culturally insensitive treatment, biased treatment protocol assignment, and medical errors). These initiating conditions may help promote individual (point-in-time) deviations in guideline-concordant care that we argue if (1) repeated and (2) left unattended, may foster the proliferation of population-level health disparities.

In the arena of population health informatics, the informatics professional is challenged to synthesize data from a variety of healthcare delivery systems that includes EHR data on clinical encounters, event and disease registries collecting data on adverse reported events, disease status, quality and safety surveys, and utilization trends. The objective here is to align traditional quality, cost, safety, and efficiency measures with global health disparities measures to identify a core set of measures for a given patient population that can then be monitored and tracked overtime. The precise challenge is in devising a means of capturing data through proxy measures that are not easily obtained or volunteered (e.g., unstructured provider notes, patient experiences, verbal and non-verbal communications, and opinioned based assessments). In other words, the objective here is to reduce or to eliminate performance-based disparities. As such, we have to track these occurrences at or near the point of occurrence throughout the continuum of care in order to prevent individual instances from growing into population-level adverse health events. In the absence of ‘near’ real-time reporting of patient encounters, this is nearly impossible. The added challenge is that much of the data that is required to shape our understanding of health disparities is at times highly subjective and not easily captured in any data system. This compounds the informatics challenges. As discussed previously, health disparities may stem from a variety of factors, such as socioeconomic status, racial and ethnic biases, and/or racism, which may lead to differential access and delivery of care, differential treatment, and/or diminished trust and communication between patients and providers. Given these factors, the informatics professional is charged with the task of collating data, information, and knowledge resources on specific clinical encounters and population-level trends that can detect, respond to, and even predict the presence of episodic or systematic differential care within these health systems. To our knowledge, there is no national event registry system in place to monitor and track events such as these. In the absence of such tracking systems, population health managers are relegated to retrospective data analysis and patient surveys that search for historical patterns of disparate or inequitable care.

4.2. Community health informatics challenge

Community health informatics encompasses the domain of informatics that seeks to meet the data, information, and knowledge needs of a pre-defined community of stakeholders. This can be a specific geographic region (e.g., a neighborhood, township, or tribal community) or a population sub-group (e.g., racial/ethnic, religious, gender, orientation). This generic definition does not differ widely from the definition of population health informatics used above. However, in this section, the primary informatics data source that informs community health disparities and associated programs, policies, and interventions is the Community Health Needs Assessment (CHNA). The informatics challenge at the community level can be easily reduced to a single task of ensuring the representativeness, completeness, accuracy, and timeliness of community needs assessments. This includes examining the reporting protocols, data sources, data gathering, data analysis, and CHNA report generation. With respect to health disparities, generating and maintaining a knowledge-base that informs relevant and impactful programs, policies, and intervention strategies is a focal challenge.

It is this last aspect of a timely, relevant, and impactful knowledge-base that proves most difficult in community health settings. One issue revolves around trying to collect and collate data from a variety of community stakeholders and partnerships that while committed to the general ideas of community health, may not always be equally willing or able to share institutional data of their specific patient activities. As such, community networks are plagued by uneven patterns of data sharing and information exchange. Additionally, many community health collaborative/coalition settings struggle with limited resources, unclear boundaries for accountability, market share competition, and inconsistent streams of federal and state funding for community programs, policies, and interventions. An additional challenge is in organizing and coordinating a plethora of individual or institutional academic research activities that are typically uncoordinated and at times, highly intrusive or disruptive. This has been tempered to some degree by Community-Based Participatory Research (CBPR)-focused research collaborations that are aimed at resolving some of these longstanding issues.

The informatics professional may be challenged to find innovative ways to collect and synthesize data, information, and knowledge resources across multiple organizations and settings (both health and non-health specific). However, this is required to ensure that health equity programs, policies, and interventions are properly informed. This issue was on full display during the recent Flint water crisis [22]. This crisis is easily termed a social justice issue, as a largely minority, low-income, and limited-resource community was repeatedly exposed to unhealthy levels of lead in their water. While the details of this case study are still under investigation, it is clear that there was a disconnect and systemic power struggle between several different agents, including local, state, and national public health officials and organizations, local and state elected officials, clinical providers, and the vulnerable community at large [22]. A much more highly tuned community-based health status alert system that searches for aberrant patterns in community health status reporting, streamlines clinical data and cross-references data with community locations to search for clustered events, and allows for electronic checks and balances when encountering poor or failed inspections might have helped minimize or avert this public health disaster.

4.3. Consumer health informatics

Consumer health informatics examines individual (person-level) health decision making needs [23]. Additionally, this field examines the socio-technical factors associated with informed and empowered decision making and communication in an effort to eliminate health disparities [23]. This can include, but is not limited to, fostering greater health literacy, managing language barriers, resolving cultural differences by promoting diversity and inclusion, fostering an atmosphere of mutual trust and shared decision making, and seeking ways to improve patient and provider communication [23]. A core challenge health professionals seek to address in patients and health consumers is how to provide consumers with adequate data, information, and knowledge resources to improve their ability to manage their healthcare. Data systems such as electronic patient portals, patient health records, and web-based knowledge navigation are designed to guide patients, their families, and general health consumers in making better health decisions. The core informatics challenge here is in helping health consumers curate large volumes of data and information (e.g., finding what they need when they need it; defining measures of reliability of the information; helping consumers to design their own formal or informal decision models and risk assessments to make healthcare decisions).

Currently, many patient advocates and patient reported outcomes assessments point to cases where the data that a patient needs is not readily available, is poorly organized, or is too overwhelming to curate in a timely manner [24]. These scenarios result in knowledge gaps where patients make critical health decisions under less informed and uncertain conditions [24]. Studies also demonstrate that these knowledge gaps are more prevalent in patient populations that are more at risk for health disparities in the course of care [24]. These populations are typically from minority, low-income, and/or low-resourced communities that often present with more aggressive forms of some diseases [25]. Here, the core informatics challenge is to design and test informatics tools customized to health consumers of all literacy levels and backgrounds that can then be further tailored to a variety of cultural, language, and ethnic backgrounds in order to foster better health decision making [25]. Such tools should be adept at triangulating data, information, and knowledge resources from a variety of sources, and tailoring it to match the needs of the priority health consumer population.

5. Conclusion

This presentation of collective intelligence and the corresponding terms of smart health, knowledge ecosystem, enhanced health disparities informatics capacities, knowledge exchange, big-data, and situational awareness are a means of demonstrating the complex challenges informatics professional face in trying to model, measure, and manage an intelligent and smart systems-level response to health disparities. We outlined public and population health disparities challenges across four distinct domains of informatics. We also introduced a concept of performance-based health disparities that we operationally define in our models as those that are generated or triggered by breaches or defects in the continuum of care that may negatively influence the quality, cost, safety, efficiency, or effectiveness of population health and health promotion.

Informaticians are challenged to design health IT solutions for complex problems but are achieving mixed results in demonstrating a direct impact on health outcomes; additionally, issues of health disparities and of health equity still remain [26]. This mixed evidence on the relation of health IT to disparities may be in part due to the design of linear solutions for an ever-changing, complex, and adaptive health system. The NIH challenged informaticians to create a more robust enhanced health disparities informatics approach [18]. Such a computational and systems perspective requires informatics professionals to derive new metrics that account for more than the mere adoption and use of technology. Our previous studies, for example, evaluated health performance with respect to network measures of information diffusion, agent cohesion, patterns or collaboration, knowledge resource availability and use, knowledge sharing tendencies, and knowledge absorption rates [27,28]. We identify these as potential ‘smart’ network measures for use in public health practice and population healthcare delivery evaluation. This perspective involves asking a more fundamental question: how much smarter are we as a result of the adoption and use of a new strategy or technology? And, correspondingly, in an effort to eliminate health disparities, how can informatics and varying levels of smartness and population health performances be used? This discussion represents a step in advancing the conversation of what a smart response to health disparities should represent and how informatics can drive the design of intelligent systems.