Timely, Granular, and Actionable: Informatics in the Public Health 3.0 Era

Abstract

Ensuring the conditions for all people to be healthy, though always the core mission of public health, has evolved in approaches in response to the changing epidemiology and challenges.

In the Public Health 3.0 era, multisectorial efforts are essential in addressing not only infectious or noncommunicable diseases but also upstream social determinants of health. In this article, we argue that actionable, geographically granular, and timely intelligence is an essential infrastructure for the protection of our health today.

Even though local and state efforts are key, there are substantial federal roles in accelerating data access, connecting existing data systems, providing guidance, incentivizing nonproprietary analytic tools, and coordinating measures that matter most.

When cholera ravaged most of Europe nearly 2 centuries ago, a London physician, puzzled by the seemingly intractable wave of deaths, created a map. Manually plotting the location of patients who died from cholera, in 1854 Dr John Snow famously revealed the geographic patterns of disease—leading to the decisive action of addressing the water source and disrupting the disease transmission. Nearly 200 years later, a different map—of Flint, Michigan, drawn from hospital electronic health records (EHRs)—sounded the alarm about the link between elevated blood levels among children and the city’s contaminated water.¹ Although these 2 events spanned nearly 2 centuries, the core purpose of public health data and analytics remains the same: through the gathering of intelligence, directing timely actions to improve people’s health. The availability of such intelligence will depend on not only modernizing but transforming the approach to public health informatics to meet the health demands of the 21st century.

CHANGING EPIDEMIOLOGY OF MORBIDITY AND MORTALITY

Actionable, geographically granular, and timely intelligence is even more critical in our increasingly complex world as we seek to understand and improve the health and health equity of communities. Despite medical advances and escalating health care spending in the United States, the Centers for Disease Control and Prevention (CDC) recently reported that the historical steady gain in longevity has plateaued for 3 years in a row, and life expectancy declined for some segments of the population.^2,3 By collating social security death records and de-identified tax records, Chetty et al. showed that the richest 1% in the United States enjoyed a life span nearly 15 years longer than did those at the bottom 1%; the size of this longevity gap, however, depends largely on geographic location.⁴ Racial/ethnic disparities persist not only in life expectancy but also in vaccination rates, infant mortality,⁵ diabetes and obesity, exposure to pollutants,⁶ and more. Analyses of multiple large data sets further confirm the long-standing notion⁷ that health and well-being—and the unequal opportunities to achieve them—are driven not only by medical, biological, and genetic factors but, increasingly, by social determinants of health. This means that how a community structures its social, environmental, and economic conditions has significant impact on whether residents can be healthy, productive, and resilient from natural and man-made threats. Addressing these diverse priorities to support the public’s health will require cross-sectoral collaboration, integration of data streams, leveraging of advanced analytic tools, and establishment of measures that inform continuous advancements in individual and community health.

A NEW ERA FOR PUBLIC HEALTH INFRASTRUCTURE

In response to this changing epidemiology, public health is evolving its mission and scope into a new framework, Public Health 3.0.⁸ Over the past 2 decades, governmental public health agencies have increasingly relied on informatics—including data, metrics, and analytics—to meet emerging challenges. However, the capabilities of informatics—and the resources to support their creation and maintenance—are uneven at best.⁹ The already constrained public health system was further weakened during the 2007–2008 recession, with nearly 40 000 jobs lost and critical budget cuts in most jurisdictions.¹⁰

Recognizing that health requires more than traditional health care and public health approaches, local public health leaders around the country are stepping out of their more traditional boundaries and serving as chief health strategists for their communities. They are building multisectoral partnerships that are data driven, prevention oriented, and acutely focused on addressing upstream determinants of health.^11,12

Such transformation builds on earlier, highly successful evolutions of public health (see the box on page 932).⁸ In the Public Health 1.0 era (late 1800s through mid-1980s), a major expansion in diagnostic and therapeutic capabilities bolstered systematized surveillance; advances in epidemiological methods allowed tracking of the spread of diseases, mostly communicable in nature; and thoughtful record keeping and synthesis allowed evidence-based medicine to flourish. During this time, the most salient metrics for public health were mortality and life expectancy, and the organized, widely accessible vital statistics system represented a milestone for this era.

BOX 1. Evolution of Public Health Data and Informatics Needs in the Public 1.0, 2.0, and 3.0 Eras.

Public Health 1.0	Public Health 2.0	Public Health 3.0
Characteristics of Essential Data and Informatics Infrastructure
• Counts and trends	• Exposure-outcome cohort studies and causal inferences	• Geospatial inferences and trend
• Vital statistics and registration	• Relative risks and attributable risk estimates	• Layering of data and multilevel-systems thinking
• Registry systems of tracking mortality and diseases	• Methods to control for confounding and sampling bias	• Nontraditional data sources
• Identify pathogens and mode of transmission	• Continuous outcomes and exposure	• Digital bridges that interface with other sources
• Binary exposure and binary outcomes	• Longer time frame	• Community-level indicators
• Population statistics based on sum of individuals	• Measures of disparities, quality of life, and well-being	• Capacity to leverage big data
	• Health services research
Public Health Actions Driven by Data Insight
• Developing diagnostic and therapeutic means to identify and remove pathogens	• Managing chronic disease risks through screening and behavioural change	• Coordinated multisectoral monitoring and action plan
• Coordinating actions to disrupt disease transmission such as quarantine, vaccination, and treatment	• Consistent surveillance and survey infrastructure	• Prediction modeling based on complex set of risk drivers
	• Professionalized functions and performance standards of governmental public health agencies	• Data and evidence as communication and policy tools rather than as the endpoint
		• Wellness promotion through changing the environmental, social, and economical contexts

The seminal 1988 Institute of Medicine (now National Academy of Medicine) report “The Future of Public Health” (https://www.nap.edu/catalog/1091/the-future-of-public-health), which called for strengthening the capabilities of public health authorities across the federal, state, and local levels, is a central feature of the Public Health 2.0 era. This period saw the transformation of local public heath into one with an expanded scope in disease prevention, including screening and managing risk factors for chronic diseases. Clinical services, particularly for those most at risk due to lack of insurance or social marginalization, became a core function of public health,⁸ as did disaster preparedness. The addition of these areas of responsibility caused an expansion of data collection, with increasingly sophisticated survey tools and analytics to allow for data normalization and stratification.

Although there is some variability to the informatics approach taken by Public Health 3.0 communities, there are shared features in their data and informatics needs: they are timely, granular, actionable, and multisectorial and have a strong focus on social determinants of health. For example, the New York City (NYC) Macroscope uses data from electronic health records to offer timely, reliable information about the health of the city that complements more traditional survey methodology.¹³ Other communities have sought broad-reaching strategies to address economic and environmental barriers to healthy behavior, safety, and well-being.^11,14,15

PUBLIC HEALTH 3.0 INFORMATICS INFRASTRUCTURE

To support chief health strategists in the Public Health 3.0 era,¹⁶ we propose the following principles for an upgraded data and informatics system:

1. Timely and geographically granular. Data should be collected and used in a timely manner to be relevant to public health decision-making. For regional, state, local, and tribal public health leaders and policymakers to get a dynamic picture of the people they serve, they need a dashboard that reveals trends, highlights hotspots, and uncovers disparities relevant to their jurisdiction.

2. Action-oriented. A statistic that signals a problem without apparent remedies may be a valuable academic exercise, but is often counterproductive in practice. The informatics infrastructure needs linkage to specific actions such as referral, benefit activation, and resource reallocation. A selected set of value- and evidence-based, substantive core resources should be used consistently over time to reveal accountability and local assets.¹⁷

3. Spanning multiple sectors. What creates health is more than health care; it is clean and safe environments, economic opportunity, education attainment, access to affordable, healthy food, and many others. Data that illuminate the connectivity of these systems are necessary to efficiently inform action and solidify shared accountability; they also serve as connective tissue in building new partnerships and strengthening existing ones. Rather than single-purpose, one-time data collection and use, the data infrastructure in the Public Health 3.0 world should be continuous and serve multiple efforts aimed at improving community health, whether through health care, social services, or public health actions.

4. Mindful of social determinants of health and equity. The Public Health 3.0 informatics infrastructure allows a population-level view that traces all determinants of health¹⁸—not only encounters with the health care system but also behavioral patterns, neighborhood contexts, and other social factors such as financial, food, or housing insecurity that affect health. In addition, equity lens should be embedded in determining metrics for outcomes and system performance.

5. Nonproprietary and interoperable. Data can improve the health of individuals and communities only if they are available to Public Health 3.0 efforts. The prevailing model in the United States—whoever collects health data (e.g., providers or payers) controls access—creates data silos and barriers to sharing. In other words, data are not interoperable, meaning organizations or programs within an organization cannot exchange data or interpret them without additional effort. Misunderstanding of Health Insurance Portability and Accountability Act (HIPAA) rules further exacerbates a culture of data blocking. Nevertheless, it is increasingly recognized that patients have a right to their own data, and they should be able to share it for public health purposes in a straightforward fashion.

CHALLENGES

For many communities, transforming their data and informatics infrastructure to a Public Health 3.0 model will prove challenging.¹⁹ Public health, especially at the local level, still lags in its adoption of information technology. Existing public health data and informatics infrastructure are often narrowly defined by disease areas (e.g., cardiovascular disease, mental health, HIV/AIDS) and, as a result, exist in silos. The federal data systems that practitioners rely on for essential statistics, although more comprehensive in health topics covered, are often based on surveys that have a time lag of several years because the data must be collected, cleaned, and de-identified. Other barriers to using these data systems for local actions include insufficient geographic granularity, required training, and restricted access policies. There is an unmet potential to leverage natural intersections among data sets collected across administrative boundaries to answer important questions in a coordinated and timely manner—for example, by creating linkages between the water sanitation and communicable disease notification systems.

Health departments continue to face resource challenges, both from local financing streams and the prospect of reduced federal public health spending. Despite promising advances such as the Big Cities Health Coalition—a forum of leaders from the nation’s largest health departments to exchange knowledge and collaborate—the absence of nonproprietary tools for data and analytics leaves actionable information out of reach for many localities. Some local public health leaders also cite the lack of skills and the pressing demand to meet statutory public health responsibilities (including, in many cases, providing safety net health care), which prevent chief health strategists from effectively coordinating community stakeholders.

One perceived barrier to achieving a Public Health 3.0 data infrastructure is the lack of interoperability between EHRs, as well as between health care providers and other entities. However, rather than a technological barrier, interoperability mostly faces business and cultural barriers. Recent data suggest that there has been substantial progress in hospital systems’ ability to send, receive, share, and integrate data.²⁰ Models such as the NYC Macroscope and the example from Flint, Michigan, however, suggest that it does not require complete interoperability or standardized EHR data for public health to reap the benefit of now-digitized health care information. But more importantly, Public Health 3.0 requires individuals and communities—not just their health care records—to be placed at the center. There are emerging innovations such as the My Health Direct information exchange in Tulsa, Oklahoma, where information from social services and social determinants of health are merged into the EHR. Moreover, innovative use of aggregated data coming directly from individuals outside of traditional health settings, such as social media, can further bolster the understanding of numerous drivers of community health and well-being.²¹

ROLES FOR THE FEDERAL GOVERNMENT

Building a Public Health 3.0 data and informatics system with the characteristics described in the “Public Health 3.0 Informatics Infrastructure” section requires stakeholders from the public and private sectors.⁸ Public Health 3.0 models will undoubtedly be championed by local and state stakeholders; however, as a major regulator, payer, provider, and developer, the federal government has a unique opportunity to set direction for measurement, collection, and usage of data in this new Public Health 3.0 informatics environment. Building on the important progress made over the past decade, we propose 5 practical, federal priorities that would support the emerging Public Health 3.0 era.

Open Existing Data Systems

Since 2008, the open data movement at the federal level has tremendously changed the default expectation of government data from closed to open. As of June 2017, there are more than 195 000 data sets hosted on Data.gov, ranging from transportation, health, and education to public safety. Its health-focused counterpart, HealthData.gov, serves nearly 180 000 visitors each year, supports community learning initiatives in which communities synthesize data across sectors to address local issues, and empowers consumers with better information on various health and health care topics.²² Such efforts provide a platform for public and private entities to conduct research, develop Web and mobile applications, and design data visualizations. We should continue such progress in opening up administrative and research data in a more timely and geographically granular fashion while safeguarding identity and privacy. One approach to achieve this goal may be to more widely employ small-area estimation methods so that large-scale surveillance data can be more readily used for local public health action.²³

Connect Existing Data Systems

A Public Health 3.0 data infrastructure must not only allow but encourage queries on gaps and possible actions across health, human services, and other health-related social domains. This requires systematically and strategically creating data linkages or crosswalks between physical health, mental health, and community health indicators within existing data systems. One such example is the linked National Health and Nutrition Examination Survey (NHANES) and Housing and Urban Development (HUD) data, providing research opportunities to access the interaction between housing and health. The Commission for Evidence-Based Policymaking was established by the bipartisan Evidence-Based Policymaking Commission Act of 2016 (PL 114–140) to increase the availability and use of government data while protecting privacy. Such an initiative exemplifies a thrust to unleash the usability of existing data and research to strengthen federal enterprise’s opportunity to advance the public’s health.²⁴ Finally, the CDC should continue its work with the private sector to make subcounty-level data—including health, health care, human services, environmental exposure, and other social determinants of health—available, linkable, accessible, and usable.

Provide Guidance and Technical Support

Public and private funding mechanisms can be deployed to seed innovations in data linkages and dissemination. Guidance on HIPAA and technical support on legal language for data use agreements can be utilized at the federal level, in collaboration with local and state informatics officers. The Department of Health and Human Services (HHS) Office for Civil Rights should continue to develop and disseminate clear guidance on private and secure data use, as well as to support civil rights compliance in addressing unequitable social conditions and discriminatory practices. At a time when nontraditional sources of data for health—including administrative records, retail, education, and digital media—are gaining importance, there should be some guidance on the oversight and stewardship structure to protect privacy and confidentiality. Public health organizations across the federal, state, and local levels can help identify best practices of repurposing health and nonhealth data, disseminate stewardship principles, serve as convener to bring stakeholders to the table, and actively communicate with the individuals who generate the data and whose lives are affected by them.²⁵ Specifically, the HHS should leverage opportunities such as Healthy People 2030 and the National Committee on Vital and Health Statistics to elevate metrics related to social determinants as leading health indicators, to define community-level indicators that describe these determinants, and to explore models that incentivize the integration of public health and clinical information.

Incentivize Open Source Analytic Tools

Today, the biggest gap for data-driven public health action is not necessarily the lack of data but the analytic infrastructure that helps make sense of data. This is particularly challenging for smaller jurisdictions with limited resources. Ready-to-use data derivatives that do not require extensive data assembly, curation, and analysis by highly trained staff are critical. This could include predictive models for risk stratification or analytic tools to quantify return on investment by addressing downstream, midstream, or upstream drivers of health and illness. One example is the ReThink model,²⁶ which allows local public health leaders to predict the health and cost impacts of investing in various, sometimes combined or layered interventions and which enhances prevention, health care, and public health infrastructure. Federal agencies should work with public health leadership, philanthropic organizations, and the private sector to support the development of nonproprietary tools based on geographic information systems and other analytic methods for frontline public health practitioners.

Measure What Matters Most

Specific to public health practices, coordination means consistently focusing on factors most important to better health and health care by building national capacity to identify, standardize, implement, and revise core measures that allow for tracking not only of individual health but of community health. The federal government, including the HHS and the Departments of Veterans Affairs and Defense, should embrace the Vital Signs: Core Metrics for Health and Health Care Progress from the National Academy of Medicine to steward the convergence and alignment of measures.^27–29 Federal entities responsible for developing measures and goals, including the NCVHS and Healthy People, should coordinate their work, focusing particular attention on developing community-level indicators, relevant at the local level.³⁰ The HHS should continue to identify gaps in the federal data infrastructure relating to race/ethnicity, language, gender identity, or sexual orientation in existing surveys. When feasible, governmental and nongovernmental stakeholders at all levels—national, state, local, and tribal—should collect standardized, reliable data to consistently monitor the state of the public’s health and health disparities.

CONCLUSIONS

We envision the Public Health 3.0 era as being powered by locally relevant and actionable intelligence that is grounded by timely, granular data, parsimonious yet essential metrics, and smart analytics that prepare communities for both present and future challenges. This is not an academic concept, but a necessary evolution called for by practitioners in the field.²⁰ Such an intelligence infrastructure will require federal leadership and action, resources, and coordination to align incentives and put common standards in place. Gaps remain in available data on some dimensions of health and well-being; overall, however, the most important work ahead is to make sense of existing data systems across multiple sectors. We believe that this is the dawn of a new era of public health informatics. By consciously moving away from process-centered data systems toward people- and community-centered ones, the vision of ensuring the conditions under which everyone can be healthy is within reach.