United States citizens have followed the Ebola outbreak in West Africa for nearly 6 months, with concern mounting about the disease arriving on US shores, apparently for good reason. The readiness of the health care system for Ebola was challenged by the very first case. When Thomas E. Duncan went to the Texas Health Presbyterian Hospital in Dallas in September, health care workers reportedly obtained and recorded his travel history, but the patient was nonetheless discharged home without being diagnosed as having Ebola.1 The press in part focused on whether the electronic health record (EHR) contributed to the missed diagnosis, but the right question to be asking is how a modern computer system should perform in this circumstance. The EHR appears to have performed exactly as expected. However, as a generally one-size-fits-all technology, it was designed for typical patients. For many patients, travel history is not especially relevant. However, for Duncan, it was the single most important aspect of the case: he had recently traveled from Liberia.
Public Health and the IT Status Quo
What should the public expect from the $48 billion appropriation to promote health information technology (IT) adoption made under the HITECH provisions of the American Recovery and Reinvestment Act? One goal is greater adoption of EHRs. That is happening. The expenditure, multiplied many-fold by investment of dollars and time across health care, has prompted more than half of physicians to use EHRs—up from 5% in 2008. However, EHRs purchased under the meaningful use program are not yet nimble enough to rearrange the display of health information based on either patient or public health context and they do not communicate with public health. Five years after the enactment of meaningful use, public health officials still reach clinicians and hospitals through traditional dispatches and media alerts. Not emphasizing the most salient data under an extenuating circumstance may ironically distract a busy clinician from an urgent task at hand.2 Under the meaningful use program, financial inducements were offered to physicians and health systems that use certified technology, but federal committees defined criteria for certification, not purchasers or physician users. Compounding the problem is that public health, largely absent from the table in defining requirements, remains mostly locked out of the point of care, barely able to exploit the newly deployed health information technology (HIT) infrastructure.
“Little Data” Only at the Point of Care
The United States has purchased an infrastructure that shows a clinician one patient at a time, emphasizing the information previously entered in the EHR. The vast population data accumulating in EHRs are not yet brought to bear on individual clinical decision making, nor are the “big data” available from across the health ecosystem.3 Public health data resources—in this case, for example, geomapped Ebola incidence that could be used in a clinical decision support system to update the prior probability that a patient is at risk—are separated from the care process. Patients deserve the improved diagnosis that could be gained from contextual awareness. Physicians deserve public health data delivered just in time, targeted to the patient at hand, that will protect them and their patients.
Building a Facade on Existing Infrastructure
There is a way forward, leveraging the massive investment in EHRs by the government, clinicians, and health systems. A ready technical approach could create an innovation layer over existing HIT structure—one resembling familiar consumer IT. What if, in the midst of a crisis in which workflows, policies, procedures, and operations must be altered, the Centers for Disease Control and Prevention (CDC) could distribute an app to emergency departments as easily as a software developer submits an app to the Apple App Store? Such a hypothetical app could reshape emergency department triage workflow to emphasize travel history (perhaps on every page) and to immediately recommend rapid assessment and isolation exposure if there is a combination of fever and recent travel to Ebola-affected regions. To some extent this is happening at airports, not with an app but by human interaction, when passengers from selected countries are screened for illness. The app could prominently flag the patient for all subsequent clinicians. The app also could manage all phases of the epidemic by calculating an estimated risk for each patient based on EHR data on symptoms and laboratory findings, combined with the epidemiologic context based on regional incidence rates of Ebola; probability of disease could be calculated in the context of prior probability. The app could be updated, just as smartphone apps constantly are, as the CDC adapts to an evolving or different epidemic.
This is not science fiction. In 2009, a fundamental shift was proposed, recasting health IT as a platform that can run “substitutable” apps.4 In other words, EHRs should work like a smartphone in that an app could be readily added to or deleted from any EHR. The Office of the National Coordinator for Health Information Technology funded the SMART Platforms Project to design the innovation layer for health IT—one that connects apps in a standard and consistent fashion to the underlying EHR data.5,6 SMART (Substitutable Medical Apps and Reusable Technology) is being used in patient care. For instance, an app for patients with hypertension runs at Boston Children’s Hospital inside the Cerner EHR, a technology not initially designed to run apps. Following suit, multiple systems have been rapidly modified to run the SMART interface, including Intermountain Healthcare’s longitudinal medical record, the core Cerner product, and a version of the Veterans Health Administration’s Vista EHR.7 The Centers for Medicare & Medicaid Services, major health care delivery systems, the pharmaceutical industry, and big data companies have joined to support and implement SMART widely. The standards-based, free SMART programming interface opens a doorway to the point of care for innovators in care redesign, public health, and research.
Getting to the Point of Care at Scale
The hypothetical CDC app described illustrates the public good in having a common interface to health system data and workflows: an app written once can run nearly anywhere. (This contrasts with the status quo having to custom modify each EHR installation.) The app could run directly inside of existing proprietary EHRs or on a “side car”—a data warehouse containing real-time data extracted from EHRs. The CDC or another innovator could release a single app and affect the point of care nationwide.
There are challenges to implementing an apps model. The workflows must be carefully developed to avoid unintended consequences—such as the missed travel history in the Dallas case—and to promote efficient care delivery. Apps given access to health system data must be vetted for efficacy, accuracy, utility, safety, privacy, and security. If the app runs on a server at an external entity, HIPAA may require business associate agreements (BAAs) to be in place between the two institutions.
Addressing Ebola and Health IT Now
Not technical barriers but a pervasive socioadministrative-regulatory inertia slows progress in health IT. Simple actions taken now could advance health IT as the current Ebola epidemic unfolds but also deliver wider value. For example, diagnosis of streptococcal pharyngitis was substantially improved by integrating data about the local incidence of streptococcal disease and calculating disease risk based on prior probability of disease.8 Hundreds of thousands of antibiotic doses per year could potentially be avoided using these epidemiologically adjusted diagnostic models. Electronic health records are not yet capable of delivering those incidence data into a decision support system at the point of care, but the apps model readily allows data “mash-ups” and novel forms of decision support. To facilitate response to enterovirus D-68—a pathogen with a changing case definition now possibly including flaccid paralysis in rare cases9—a common apps interface to EHRs could enable rapid nationwide uptake of a triage and management app, one that could be updated as the epidemic and clinical picture evolves. Such technological feasibility would also be helpful when the next epidemic arrives. Potential next steps should include
Standardize on a programming interface between data and apps. The SMART platform specification, created under a $15 million federal investment, is a good place to start.
Create the necessary apps functionality. Clinicians, informatics experts, and representatives from the CDC, the World Health Organization, the US Agency for International Development, and nongovernmental organizations could collaborate to design workflows and data displays to improve diagnosis and management apps that work for physicians providing care.
Liberate data for contextualized diagnosis. Using the open.fda.gov initiative as a model, public health data resources could be identified and made available in computable formats so external data sources can be combined with EHR data to provide clinical and public health intelligence to treating physicians.
Ready the point of care. Institutions with real-time data warehouses could adopt the SMART application programming interfaces and begin running apps. The largest EHR vendors, several of which have invested in SMART and SMART-inspired programming interfaces, could lead the way in responding to Ebola by upgrading as many installations as possible to support public health apps, as a first-use case.
With Ebola moving across the globe, this aggressively paced response may be achievable in a short time frame. Even if it takes more time, the steps outlined could rapidly transform current-stage HIT into a platform that may turn the point of care into a place for innovation, efficiency, and improved outcomes.
Corresponding Author: Kenneth D. Mandl, MD, MPH, Harvard Medical School, Boston Children’s Hospital, 300 Longwood Ave, Boston, MA 02115 (kenneth_mandl@harvard.edu).
Published Online: October 20, 2014. doi:10.1001/jama.2014.15064.
Conflict of Interest Disclosures: The author has completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Funding/Support: Research on diagnosis enhanced by epidemiologic context was funded by Center of Excellence in Public Health Informatics award P01HK000088 from the CDC and by grants 1G08LM009778 and R01 LM007677 from the National Library of Medicine, National Institutes of Health. The SMART Platforms Project was funded under the Strategic Health IT Advanced Research Projects (SHARP, a congressionally appropriated program) with award 90TR000101 from the Office of the National Coordinator of Health Information Technology. Member organizations of the SMART Advisory Committee (http://smartplatforms.org/advisory-committee/), including the Hospital Corporation of America, Lilly, Surescripts, the Advisory Board Company, MyHealthBook, Polyglot System Inc, and the BMJ Group, provide philanthropic support to the Boston Children’s Hospital, which funds SMART development.
Role of the Funder/Sponsor: The funders had no role in the preparation, review, or approval of the manuscript or decision to submit the manuscript for publication.
2.Middleton
B, Bloomrosen
M, Dente
MA,
et al; American Medical Informatics Association. Enhancing patient safety and quality of care by improving the usability of electronic health record systems: recommendations from AMIA.
J Am Med Inform Assoc. 2013;20(e1):e2-e8.
PubMedGoogle ScholarCrossref 3.Weber
GM, Mandl
KD, Kohane
IS. Finding the missing link for big biomedical data.
JAMA. 2014;311(24):2479-2480.
PubMedGoogle Scholar 5.Mandl
KD, Mandel
JC, Murphy
SN,
et al. The SMART Platform: early experience enabling substitutable applications for electronic health records.
J Am Med Inform Assoc. 2012;19(4):597-603.
PubMedGoogle ScholarCrossref 8.Fine
AM, Nizet
V, Mandl
KD. Improved diagnostic accuracy of group A streptococcal pharyngitis with use of real-time biosurveillance.
Ann Intern Med. 2011;155(6):345-352.
PubMedGoogle ScholarCrossref 9.Pastula
DM, Aliabadi
N, Haynes
AK,
et al. Acute neurologic illness of unknown etiology in children—Colorado, August-September 2014.
MMWR Morb Mortal Wkly Rep. 2014;63(40):901-902.
PubMedGoogle Scholar