Learning Health System in Crisis: Lessons From the COVID-19 Pandemic

Abstract

The coronavirus disease 2019 (COVID-19) pandemic is the gravest public health crisis that the United States has seen in more than a century. Health care delivery systems are the focal point for interfacing with COVID-19; however, many were and remain unprepared for this or similar outbreaks. In this article, we describe the learning health system (LHS) as an ideal organizing principle to inform an evidence-based response to public health emergencies like COVID-19. We further describe barriers and challenges to the realization of the LHS and propose a call to action for a substantial investment in the LHS, with a focus on public health. Specifically, we advocate for a learning health network that promotes collaboration among health systems, community-based organizations, and government agencies, especially during public health emergencies. We have approached this commentary through the unique lens of researchers embedded within a large, integrated health care delivery system, with direct experience working with clinical and operational units in response to the COVID-19 pandemic. After its initial detection in Wuhan, China in December 2019, the COVID-19 outbreak, caused by the severe acute respiratory syndrome coronavirus 2, rapidly spread across the world.1World Health OrganizationCoronavirus disease (COVID-19) weekly epidemiologic update and weekly operational update: situation report 1.https://www.who.int/emergencies/diseases/novel-coronavirus-2019/situation-reportsGoogle Scholar On March 11, 2020, the World Health Organization officially declared the COVID-19 outbreak a global pandemic,2World Health OrganizationWHO director-general’s opening remarks at the media briefing on COVID-19 – 11 March 2020.https://www.who.int/dg/speeches/detail/who-director-general-s-opening-remarks-at-the-media-briefing-on-covid-19---11-march-2020Date accessed: May 17, 2020Google Scholar the worst seen in the United States since the influenza pandemic of 1918, which claimed approximately 675,000 American lives. It is currently believed that the first COVID-19 fatality in the United States was on February 6, 2020, in California’s San Francisco Bay area, a month prior to the initially purported first death in Washington state,3Hamilton M. St. John P. Lin II, R.-G. Autopsies reveal first confirmed U.S. coronavirus-related deaths occurred in California in February. Los Angeles Times website.https://www.latimes.com/california/story/2020-04-21/autopsies-reveal-first-confirmed-u-s-coronavirus-deaths-occurred-in-bay-area-in-early-februaryGoogle Scholar suggesting that the virus was spreading earlier than experts had previously believed. In a brief time, the United States surpassed China and several other nations with the highest number of COVID-19 cases globally. Around the time of publishing this article, the United States had cumulatively reported more than 19 million COVID-19 cases and over 330,000 deaths.4Johns Hopkins University Coronavirus Resource CenterCOVID-19 dashboard by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University (JHU).https://coronavirus.jhu.edu/map.htmlDate accessed: May 21, 2020Google Scholar With only 4% of the world's population, the United States Currently Claims approximately 25% of all COVID-19 cases and 20% of COVID-19–related deaths. Across the United States, COVID-19 cases are still rising. The Institute for Health Metrics and Evaluation of the University of Washington (Seattle) estimates that cumulative deaths could climb to nearly 500,000 cases by February of 2021.5Institute for Health Metrics and EvaluationCOVID-19 projections: United States of America.https://covid19.healthdata.org/united-states-of-americaGoogle Scholar This projection is very concerning, especially as capacity and resources to treat infected individuals continue to deplete. Complicating matters, early in the pandemic, the United States lagged behind many other countries with regard to community-based testing, making it nearly impossible to implement informed risk mitigation policies to contain the spread of the virus. Health care delivery systems are the focal point for interfacing with the COVID-19 pandemic and thus are vital for mounting an evidence-based response. However, many were and remain unprepared for this or similar outbreaks. In the era of the electronic health record (EHR) and more than a decade after the Institute of Medicine (IOM) formalized the concept of the LHS, why was the United States not better prepared for its worst public health emergency in more than a century? In recent years, there has been an intensifying interest in the LHS paradigm, evidenced by the establishment of an open-access journal in 2017 dedicated to the topic, aptly named Learning Health Systems,6Wiley Online LibraryLearning Health Systems.https://onlinelibrary.wiley.com/page/journal/23796146/homepage/productinformation.htmlDate accessed: May 17, 2020Google Scholar and $40 million in grant funding awarded by the Agency for Healthcare Research and Quality and the Patient-Centered Outcomes Research Institute to conduct research on patient-centered outcomes within the context of the LHS.7Agency for Healthcare Research and QualitySupporting the next generation of learning health systems researchers.https://www.ahrq.gov/funding/training-grants/lhs-k12.htmlGoogle Scholar The IOM, now called the National Academy of Medicine, conceptualized the LHS in 2007 as a novel approach to health care delivery, evidence generation, and a conduit toward value-based care.8Olsen L. Aisner D. McGinnis J.M. The Learning Healthcare System: Workshop Summary. National Academies Press, Washington, DC2007Google Scholar The LHS is an environment in which “science, informatics, incentives, and culture are aligned for continuous improvement and innovation, with best practices seamlessly embedded in the delivery process and new knowledge captured as an integral by-product of the delivery experience.”8Olsen L. Aisner D. McGinnis J.M. The Learning Healthcare System: Workshop Summary. National Academies Press, Washington, DC2007Google Scholar The LHS paradigm, as envisaged by the IOM, has 3 core elements: (1) foundational elements of data infrastructure, including the ability to collect and (re)use data, primarily from the EHR; (2) care improvement targets that assist learning and health through clinical decision-making activities; and (3) a supportive policy environment in which financial incentives reward high-value care and promote performance transparency. The LHS paradigm is an ideal organizing principle to inform a unified and data-driven response to national public health emergencies like COVID-19, given that the cornerstone of an LHS is its data used to inform and improve care. The LHS can facilitate a continuous learning cycle of generating data (collected via the EHR, patient registries, or other sources), interpreting data using robust analytics, planning and coordinating data-driven solutions, and implementing an informed and systematic response. In 2011, the IOM released a report on digital infrastructure for the LHS to facilitate continuous learning.9Institute of MedicineDigital Infrastructure for the Learning Health System.in: Grossmann C. Powers B. McGinnis J.M. National Academies Press, Washington, DC2011Google Scholar The report brings into the fold the aforementioned data sources, as well as other digital information including health portals, electronic monitoring devices, and biobanks. Each of these additional sources “adds important capacity for clinical care, clinical and health services research, public health surveillance and intervention, patient education and self-management, and safety and cost monitoring.”9Institute of MedicineDigital Infrastructure for the Learning Health System.in: Grossmann C. Powers B. McGinnis J.M. National Academies Press, Washington, DC2011Google Scholar Presently, the EHR is the most mature of all the data sources proposed by the IOM and holds the most promise for data collection and analytics in the context of an LHS. Thus, we primarily focus on EHR data in this commentary. In the following sections, we discuss the role of each of the core elements of the LHS paradigm in responding to the COVID-19 pandemic and other public health crises in general. Health systems that successfully operationalized the LHS paradigm would be well prepared to serve as sentinels for disease surveillance, enabling early detection in shifts away from the norm, and well positioned to anticipate an impending rise in the epidemiological curve of an outbreak within their communities. To achieve this goal, the necessary data infrastructure, namely the EHR, must be in place. Over the past decade, there has been a steady adoption of EHRs. Between 2008 and 2017, office-based health care professionals using an EHR increased from 42% to 86%.10Office of the National Coordinator for Health Information TechnologyHealth IT Dashboard: office-based physician electronic health record adoption.https://dashboard.healthit.gov/quickstats/pages/physician-ehr-adoption-trends.phpGoogle Scholar This shift is primarily attributed to the American Reinvestment and Recovery Act of 2009, and specifically the Health Information Technology for Economic and Clinical Health (HITECH) Act,11Index for excerpts from the American Recovery and Reinvestment Act of 2009 (ARRA). Division A—Appropriations Provisions. Title XIII. Health Information Technology (HITECH Act). 111th Congress of the United States of America.https://www.healthit.gov/sites/default/files/hitech_act_excerpt_from_arra_with_index.pdfDate accessed: May 19, 2020Google Scholar which has incentivized EHR adoption through financial payments for meeting specific “meaningful use” metrics and penalties for not meeting these metrics.12Centers for Disease Control and PreventionPublic health and promoting interoperability programs.https://www.cdc.gov/ehrmeaningfuluse/introduction.htmlDate accessed: May 19, 2020Google Scholar With widespread adoption of the EHR, health systems—especially large integrated systems—are able to generate vast amounts of data that can be repurposed for disease surveillance. Moreover, data from other existing sources can be used to track key resources, including staffing, beds, ventilators, medications, testing kits, and other vital supplies. With data infrastructure and prespecified metrics in place, health systems are poised to conduct surveillance, which should begin prior to the start of a pandemic. Continuous monitoring of metrics can be used to detect emerging signals, applying sophisticated methods such as artificial intelligence and machine learning. For example, real-time analysis of increases in expected cases of pneumonia, intensive care unit bed occupation, utilization of ventilators, or even cases of suspected influenza (compared with previous years) could reveal important signals that a novel and potentially communicable disease is circulating and further investigation is needed. Using the EHR, a recent study has reported the ability to detect signals of COVID-19, evidenced by excess visits with the word “cough” documented as the primary reason for a health care visit, as well as hospitalizations for acute respiratory failure.13Elmore J.G. Wang P.-C. Kerr K.F. et al.Excess patient visits for cough and pulmonary disease at a large US health system in the months prior to the COVID-19 pandemic: time-series analysis.J Med Internet Res. 2020; 22: e21562Crossref PubMed Scopus (9) Google Scholar These data must get into the hands of decision makers early and often so that planning is proactive, swift, and evidence based. Timeliness during disease outbreaks is vital, as we have learned from the current pandemic. Had initial shelter-in-place policies been put into effect even a week earlier, tens of thousands of lives could have been saved.14Pei S. Kandula S. Shaman J. Differential effects of intervention timing on COVID-19 spread in the United States.Sci Adv. 2020; 6: eabd6370Crossref PubMed Scopus (134) Google Scholar During a public health crisis, communication and transparency across health care systems, public health agencies, and other institutions are needed to facilitate strategic and coordinated responses. Health care systems must not operate in silos during such times. Rather, information should be shared. We offer several pertinent examples. First, during the early stages of an epidemic, if one health care system detects an emerging signal, quickly alerting other health systems and public health departments, especially those in the same geographic region, may help to halt the spread of infection and contain an outbreak early. Second, as the epidemiological curve rises and resources become increasingly limited within one health care system (eg, intensive care unit beds or ventilators reaching maximum capacity), a contingency plan to channel patients and/or resources to other facilities could accommodate the need. Third, health care systems could notify government agencies in real time about shortages of testing kits, medication, and personal protective equipment to alert those responsible for local, state, or national supply chains. With the core elements of the LHS in place, ideally health care systems should be well-equipped for challenges like the COVID-19 pandemic—early and rapidly identifying cases (signal detection), modeling and assessing the potential impact, and proactively planning a systematic and iterative response—informed largely by warnings from other modern-day, global disease outbreaks (severe acute respiratory syndrome [2003], influenza A virus subtype H1N1 [“swine” flu; 2009], Middle East respiratory syndrome [2014]).15World Health OrganizationEmergencies preparedness, response: disease outbreaks by year.https://www.who.int/csr/don/archive/year/en/Date accessed: May 17, 2020Google Scholar This scenario, however, was not the case in the context of the current pandemic for health care delivery systems in the United States or the health care system as a whole. This failure raises questions about why prior warnings were not heeded and whether the LHS, after 13 years, is still an abstract concept rather than a reality. Notwithstanding efforts to encourage the materialization of the LHS in the United States, there is little evidence of its successful implementation.16Budrionis A. Bellika J.G. The learning healthcare system: where are we now? a systematic review.J Biomed Inform. 2016; 64: 87-92Crossref PubMed Scopus (124) Google Scholar Historically, one barrier has been lack of information technology, or the so-called foundational elements of data infrastructure of the LHS paradigm, namely the EHR. Despite the advancements achieved through the HITECH Act, EHR adoption in the United States is not complete, with lower use among health care professionals in rural or impoverished communities.17Mack D. Zhang S. Douglas M. Sow C. Strothers H. Rust G. Disparities in primary care EHR adoption rates.J Health Care Poor Underserved. 2016; 27: 327-338Crossref PubMed Scopus (19) Google Scholar,18Gold M. McLaughlin C. Assessing HITECH implementation and lessons: 5 years later.Milbank Q. 2016; 94: 654-687Crossref PubMed Scopus (85) Google Scholar Even with an EHR in use for clinical care, many health care professionals and systems do not have the resources and infrastructure to repurpose data for applied analytics. Thus, for some institutions it is nearly impossible to use these data to detect early signals of disease outbreaks or to quickly assess facility resourcing needs. Again, professionals in rural and impoverished communities are less likely to have the requisite resources to conduct analytics and thus are at a disadvantage during disease outbreaks like COVID-19, exacerbating already prevalent inequities. This issue underscores the importance of data sharing. Another barrier to the implementation of the LHS has been the structure of the US health care system itself. Although the LHS is considered a conduit to value-based care, a supportive policy environment that embodies value-based care and transparency is required to promote the development and successful implementation of an LHS. The US health care system, or more accurately the multiple systems that comprise the overall health care system, is largely a fee-for-services delivery model that has been slow to embrace value-based payment structures. During a public health crisis, health care systems must view each other as partners rather than competitors. Yet a competitive marketplace that rewards volume creates disincentives for necessary communication and coordination among health care systems. Even when organizations are willing to share data, privacy rules imposed by the Health Insurance Portability and Accountability Act (HIPAA) and other regulations, while necessary to protect patient and data have to some information J. L. barriers to the of health information or Q. PubMed Scopus Google Scholar Although we a of the US health care system time has been in the adoption of value-based care in recent J. of have to value-based care, payment Scholar Between and the number of implementing value-based care increased from 3 to J. of have to value-based care, payment The of value-based 2019 Healthcare Scholar This can be attributed to the for and Act of which the for and to health care professionals for of care as to for and Act of accessed: May 19, 2020Google Scholar have been over the past decade in the development of data that promote data with the Health Insurance Portability and Accountability and analytics across health care systems, and in the The the National Patient-Centered Clinical Research and the Health Care Systems Research are 3 such and Patient-Centered Clinical Research accessed: May 17, 2020Google Scholar These are that should inform can to rapidly generate data needed to detect signals and a response to disease outbreaks. The health care from its 2007 when the IOM first conceptualized the LHS With widespread adoption of information and toward value-based care, than the time and where the LHS has the potential to knowledge into action an additional barrier to To this we propose that the of Health and or a process through which health care systems are on their ability to and share data in to be as an LHS. This would a for which health systems can and and needed to achieve the In the context of COVID-19, it has become that public health surveillance is early analysis of the US response to this pandemic from The Times to technology, a public health system, and A. et for this Times Scholar The Times and was the data the the systems, the for Disease Control and and and public health information that could help We should be able to on health care systems, not only to care for patients during public health emergencies to in an national response. With this situation in we the LHS paradigm as a learning health network The on the of the LHS, which to organizations, by and coordination among The concept of such a network is not et A. from learning healthcare systems to learning health care Res. PubMed Scopus Google Scholar proposed a from an LHS to a learning health care which a core the of and The learning health care on the LHS by into the fold or individuals to promote a more health care delivery A. from learning healthcare systems to learning health care Res. PubMed Scopus Google Scholar This is a in the of a network that government agencies, especially those responsible for public health. In et et a network to the development of Learning Health 27: PubMed Scopus Google Scholar previously that have been to health for specifically for and of an is novel in that it a of public health and disease surveillance as a and public health To this the can the aforementioned continuous learning cycle of generating and interpreting data, planning and coordinating data-driven solutions, and implementing systematic with each as one of a integrated The should of health care and to potential across organizations development of that can be across organizations and existing resources, including the EHR, registries, and data and The development of an is and to Even when we are able to barriers of data infrastructure, a supportive policy and knowledge into the of and With regard to we propose that the may be by and/or as of a to a more robust public health infrastructure in the United For we the as a network that is by from The would for in the network and for data and To facilitate data analytics across multiple systems, a data model should be with one by the to serve as a data coordinating of the should comprise organizations with data infrastructure that for data as well as those that are of for example, may health care delivery systems and public health agencies at the local, state, and may community-based organizations, such as patient to that the patient is and agencies, which are prepared to of and in especially during of with regard to government agencies should be required to in the may These incentives should ideally be at the health care delivery systems with the necessary infrastructure to data to be incentivized through as an LHS, as previously and/or with for and for and organizations, the ability to data, could be incentivized to with to data organizations may be to as of the data, to the to which to from the information by the In this we have an from the of the LHS, to a more coordinated and informed national response to public health emergencies like COVID-19. This pandemic has been and continue to be an important for the US health care We are at a for we to the next of COVID-19 and to we this call to

Topics & Concepts