Electronic Health Records: Promises and Realities: A 3-Part Series Part I: The Digital Sea Change, Ready or Not

Abstract

Most American hospitals have been slow to adopt electronic health records (EHRs), despite widespread anticipation that these systems would improve both clinical performance and economic efficiency. ⁎The terms “electronic health record” and “electronic medical record” are used interchangeably, with “health information technology” or “medical informatics” denoting the discipline more generally. EHR appears preferentially here to minimize potential confusion with other acronyms in emergency medicine (EM, EMS, EMT, etc).The terms “electronic health record” and “electronic medical record” are used interchangeably, with “health information technology” or “medical informatics” denoting the discipline more generally. EHR appears preferentially here to minimize potential confusion with other acronyms in emergency medicine (EM, EMS, EMT, etc). During the 2008 campaign and then early in his presidency, Barack Obama has recommended EHRs as a tool to reduce waste and errors, a cost-saving measure, and a generator of jobs. Support for EHRs is currently one of the rare areas of credible bipartisanship, attracting the enthusiasm of Governor Bobby Jindal (R-LA) and retired House Speaker Newt Gingrich (R-GA). President Obama's $787 billion stimulus package includes $19 billion in earmarks for “meaningful use” of EHRs through the Health Information Technology for Economic and Clinical Health (HITECH) section of the American Recovery and Reinvestment Act.1Greene J. Obama's $19 billion boon to health care it: mammoth investment fasttracks electronic health records.Ann Emerg Med. 2009; 53 (Accessed June 3, 2010): A24-A27http://www.annemergmed.com/article/PIIS0196064409002443/fulltexthttps://doi.org/10.1016/j.annemergmed.2009.03.004Abstract Full Text Full Text PDF PubMed Scopus (6) Google Scholar Such incentives, combining both payments and penalties, practically guarantee that American medicine will finally move from 20th-century analog information technologies to their 21st-century digital equivalent. Steep adoption costs, measured in time and attention, as well as money, have been a particular obstacle for the small practices that remain prevalent in American medicine; the HITECH Act addresses that problem directly. But as EHR questions become “when, how, and how much” rather than “whether,” observers are offering caveats about the benefits claimed to date. The Veterans Administration's widely admired system, Veterans Health Information Systems and Technology Architecture (VistA), says David U. Himmelstein, MD, associate professor of medicine at Cambridge Health Alliance and Harvard Medical School, “is one of the few that looks [sic] to have been relatively successful.” A recent study estimated the net benefit from the VA's health informatics investments at more than $3 billion.2Byrne C.M. Mercincavage L.M. Pan E.C. et al.The value from investments in health information technology at the US Department of Veterans Affairs.Health Aff (Millwood). 2010; 29: 629-638Crossref PubMed Scopus (63) Google Scholar Last January, however, Dr. Himmelstein and colleagues offered a sobering view of the financial effects in the “most wired” hospitals among some 4,000 surveyed annually from 2003 to 2007 by Healthcare Information and Management Systems Society Analytics, drawing cost and quality data from the Medicare Cost Reports from the Centers for Medicare & Medicaid Services and the 2008 Dartmouth Health Atlas.3Himmelstein D.U. Wright A. Woolhandler S. Hospital computing and the costs and quality of care: a national study.Am J Med. 2010; 123: 40-46https://doi.org/10.1016/j.amjmed.2009.09.004Abstract Full Text Full Text PDF PubMed Scopus (153) Google Scholar Although yielding moderate benefits in quality of care, wiredness—with the crucial qualifier “as currently implemented”—did not correlate with cost savings. In Dr. Himmelstein's view, medical informatics should not be considered a financial panacea; it is an uncertain field with an urgent need to define best practices. If an information system is not built to reflect an organization's particular purpose and priorities, it is likely to obstruct them. A prominent advocate of EHR systems as a means of upholding and regularizing standards of care,4Mechanic D. Rethinking medical professionalism: the role of information technology and practice innovations.Milbank Q. 2008; 86: 327-358Crossref PubMed Scopus (51) Google Scholar sociologist David Mechanic, PhD, of the Institute for Health, Health Care Policy, and Aging Research at Rutgers University, shares many of Dr. Himmelstein's concerns. The legislative term “meaningful use,”5Brailer D.J. Interview: guiding the health information technology agenda.Health Aff (Millwood). 2009; 29: 586-594Crossref Scopus (14) Google Scholar ⁎According to David Blumenthal, MD, MPP, former National Coordinator for Health Information Technology, in an interview with his predecessor David J. Brailer, the term “meaningful use” originated with Congress and not with the Office of the National Coordinator. Dr. Mechanic says, does not distinguish between basic systems (defined as those including patient demographics, problem lists, clinical notes, prescription orders, and views of laboratory and imaging results) and a fully functional system including history and follow-up, electronic test/prescription ordering, warnings of drug interactions and contraindications, highlighting of out-of-range test levels, and reminders about guideline-based measures or problems to anticipate, given the patient's situation. The low rate of adoption by US physicians—13% for basic systems and 4% for fully functional as defined in 2007 and 2008 by one expert panel,6DesRoches C.M. Campbell E.G. Rao S.R. et al.Electronic health records in ambulatory care—a national survey of physicians.N Engl J Med. 2008; 359 (Accessed June 3, 2010): 50-60http://content.nejm.org/cgi/citmgr?gca=nejm;359/1/50Crossref PubMed Scopus (849) Google Scholar increasing only to 20.5% and 6.3%, respectively, in a 2009 report7Hsiao C.-J. Beatty P.C. Hing E.S. et al.Electronic medical record/electronic health record use by office-based physicians: United States, 2008 and preliminary 2009. NCHS Health E-Stat. National Center for Health Statistics, December 2009.http://www.cdc.gov/nchs/data/hestat/emr_ehr/emr_ehr.htmGoogle Scholar—is on the verge of changing, but later adopters may not outperform early adopters in choosing (and purposefully using) more capable software. EHR efforts overseas, although ahead of ours in some respects, have also occasionally been hesitant. In The Netherlands, adoption of some form of EHR is estimated at 98%,8Commonwealth Fund2006 International Health Policy Survey of primary care physicians in seven countries. Commonwealth Fund 2006 International Symposium on Health Care Policy, Washington, DC, November 1-3, 2006.http://www.commonwealthfund.org/usr_doc/2006_IHP_Charts_10-30-06.pdfGoogle Scholar but a third of physicians have objected to participation in the proposed Elektronisch Patiëntendossier central registry on privacy/security grounds, even with a patient opt-out feature.9van Baardewijk L.J. Electronic health record in The Netherlands: afraid of the unknown.Amsterdam Law Forum. 2009; 1 (Accessed June 3, 2010): 41-46http://ojs.ubvu.vu.nl/alf/article/download/93/158Google Scholar The British National Health Service has moved toward “a relatively centralized emergency system” allowing interoperability between practices, Dr. Himmelstein notes. Last April, however, the NHS Connecting for Health program suspended its rollout until “greater public and professional awareness” could spread.10Timmins N. NHS suspends electronic record project.Financial Times. April 16, 2010; (Accessed June 3, 2010)http://www.ft.com/cms/s/0/9c0377b2-498e-11df-9060-00144feab49a.htmlGoogle Scholar Dr. Himmelstein names the Kaiser Permanente HealthConnect system, Boston's Brigham and Women's Hospital, Salt Lake City's LDS Hospital (part of Intermountain Healthcare), and the Regenstrief Institute in Indianapolis as additional national leaders in EHRs. However, he cautions, even these strong reputations still rest more on anecdotal reports than rigorous study. As a wider range of hospitals wires up, he says, “the incentives make it almost mandatory for most places to use vendor-sold systems, and I think that's very problematic. In order to get the financial incentives, you have to install a certified system … and those are exactly the systems that, at this point, there's no evidence have been useful.” Choosing between locally developed, clinically tailored systems and off-the-shelf products—“substantially billing tools on which you've got clinical things hung”—Dr. Himmelstein fears that the latter will offer many organizations a path of least resistance. “It's akin to saying, ‘You’ve got to have x-ray machines,' and what we're going to do is install x-ray machines all over the country without radiology departments.” When any discipline converts to digital technologies, a continuum exists between tools that replicate paper-based or analog processes, expanding on some capabilities but not qualitatively transforming them, and tools that create entire new spaces for, and hence modes of, professional practice. Not every promise is borne out (conjectures in the 1980s about the “paperless office” being one familiar example), and not every activity derives unalloyed benefit from the conversion. Telephony, suggests Tom Love, PhD (a cognitive scientist, former research director for several major corporations who worked on the development of the first fully distributed digital telephone switching system for ITT, and cofounder and chief executive officer of the informatics firm ShouldersCorp, currently consulting on efforts to update VistA), has gained immensely from the analog-to-digital change. The transition from typewriters to computers, for those who remember it, was likewise intuitively transformative. Music, on the other hand, is arguably one field in which the consequences of going digital have been mixed at best and, to some ears, disastrous. In Dr. Love's words, there are situations in which “your iPod and my Steinway don't match.” What types of EHR systems might help hospitals achieve and maintain a Steinway level of sophistication, particularly in the emergency department (ED), even while generating iPod-style efficiencies? Consider 2 conjectural scenarios involving American hospitals a decade or 2 from now. Hospital A converted its cumbersome paper records system to EHRs through careful consultation between physicians and information technologists, with an orderly stepwise rollout process incorporating users' feedback into system design. This institution's first-adopter departments, including the ED and selected local primary care partner groups, invested considerable time troubleshooting an extensive database organized to benefit both present-day clinical care and long-range research. The resulting system, an institutionally specific tweak of a major commercial product based on open-source system architecture, is known as Finely Entrained Automation and Teleclinical Services, or FEATS. A patient arriving in this hospital's ED, even for the first time, in a communicative state or otherwise, does not arrive as a stranger. From either a wallet card or cross-checkable demographic and biometric information, the emergency physicians can get access to the patient's history, including detailed results from recent ED visits, other clinical examinations, ECGs, laboratory tests, and imaging procedures performed elsewhere, both those that may need repeating, such as blood tests, and those that do not, such as recent chest radiographs; medications past and present, with known responses, sensitivities, contraindications, and insurance-related formulary information and the patient's most convenient pharmacies; and even genomic data relevant to potential personalized treatments. Patients do not always remember what conditions they have had, physicians they have consulted, and evaluations they have undergone, but the system readily jogs their memory. The simple matter of ascertaining a patient's identity, notes John T. Finnell, MD, MSc, associate professor of emergency medicine at Indiana University and research scientist at the Regenstrief, is not always so simple, particularly because a national health identification card, amid legislative contention over immigration and other related issues, may always be a “political hot potato.” Regenstrief researchers have developed probabilistic identification algorithms that make universal identification cards a moot point, Dr. Finnell says, incorporating combinations of demographic and biometric information to connect each patient to an internal global identification. Fingerprints alone do not suffice, particularly in the ED, because of changes over time and the possibility of hand injuries. The radiofrequency identification (RFID) VeriChip that Beth Israel Deaconess Medical Center/Harvard Medical School emergency physician and chief information officer John Halamka, MD, MS, voluntarily had implanted11Halamka J. What are the benefits and risks of fitting patients with RFID devices? John Halamka's viewpoint: RFID devices enable patients to be stewards of their own health data. Medscape Today.http://www.medscape.com/viewarticle/567776_4Google Scholar in a 2004 pilot project, Dr. Finnell reports, offers certain advantages in personal data control but seems unlikely to catch on in a nation with large numbers of Orwell readers. Each new procedure or medication order in a clinical encounter at Hospital A automatically populates the FEATS record, generating instant alerts about drug interactions or allergies. Incorporation of the Health Level Seven International interoperability standards lets these physicians share data securely across institutional and international borders. Instant onscreen translation functions eliminate delays when patients need an interpreter. Wristbands using a hybrid of infrared and RFIDs allow effortless patient tracking across shift transfers, when responsibilities for a patient change and communication gaps can correlate with errors. (Using both technologies is a purposeful form of redundancy: RFIDs reportedly work poorly in a curtained environment.) Analysis of these data points allows ED managers to recognize queuing patterns and adjust staffing levels to minimize crowding and delays. The combination of templates, narrative fields, direct data input from instruments, voice-recognition technology, and customizable macros minimizes workflow friction while building rich and integrated clinical records. Illegible notes by clinicians who handwrite some 40 notes a day, with any resemblance to John Hancock's signature fading by about the 20th, are a thing of the past. Within a standard onscreen dashboard that presents information consistently and minimizes searching time, access to different levels of data is tailored to clinical roles: paramedics and nurses can see the information they need to do their work effectively and quickly, whereas physicians in decisionmaking positions can read expanded or filtered versions of a standard Mayo-style patient record that support nuanced clinical thinking without attempting to replace it. When patients consent, video recordings of some outpatient encounters provide detailed visual archives, used to review complicated cases and extend the scope of teaching rounds. Consent processes themselves are enhanced by built-in prompts and the capacity to process patients' voiceprints and signatures. As the legal owner of the record, each patient has the opportunity to examine a read-only copy, easing privacy concerns, along with promoting patients' responsibility for their health. Outcome measurements in this hospital have steadily increased; duplication, inefficiencies, and errors have decreased. Observers inside and outside the institution attribute these gains variously to the continuity, coordination, and standardization of care afforded by FEATS, including the capacity to incorporate changing clinical guidelines promulgated by the American College of Emergency Physicians, The Joint Commission, and other professional groups, even the latest findings reported in journals or as fast-breaking epidemiologic information. Administrators initially balked at the startup costs, but this system, like a well-planned and fully commissioned green building, paid off its own premium within a few years and has continued to generate savings through high performance. Aggregate clinical data, stripped of individual patient identifiers and searchable through a logical, accessible query language, have also made Hospital A a more fruitful field for both academic research and real-time biosurveillance. At Hospital B, in contrast, a ballyhooed step into the future has backfired. The off-the-shelf EHR system, chosen largely on cost grounds and developed with little clinician involvement, consumes inordinate amounts of time in data entry (that of physicians and nurses, as well as scribes); widespread perceptions that the system slows down ED activity have been confirmed with stopwatches. It has made some personnel dependent on it and aroused resistance among others, sharpening generational and philosophical schisms within the staff. A cumbersome interface presents users with too much irrelevant data while making essential information hard to find. Porous security systems, hacked in a well-publicized incident shortly after system adoption, have not only ignited brushfires of litigation over unauthorized data disclosure—expanded in scale by the HITECH Act's extension of Health Insurance Portability and Accountability Act provisions to business associates, as well as “covered entities”—but also made some privacy-conscious patients reluctant to offer complete medical histories to their physicians, fearful not of insurance rescission (now illegal) but that their sensitive pharmacologic, psychiatric, and infectious disease information might “go viral” in the nonmedical sense, through the Internet. Others have gamed the patient identification protocols so as to receive care (sometimes including controlled painkillers) under someone else's name,12Moore J. Why pay for health insurance when you can steal it?.National Public Radio. March 3, 2010; (Accessed June 3, 2010)http://www.npr.org/templates/story/story.php?storyId=123977187Google Scholar not only free-riding on insurance but also unwittingly contributing misleading data to the clinical record. Users have occasionally found a single patient record listing more than one blood type. Hospital B's system offers cookie-cutter decision support that is redundant in the presence of horse hoof beats yet inappropriate for the occasional zebra. It frequently crashes, bringing departments close to a state of panic. Its interoperability features are largely a moot point because its security problems have led to guarded, haphazard populating of content fields; the need to negotiate a patchwork of state privacy laws has complicated the transfer of different components of patients' EHRs; continuity-of-care gains, though not negligible, have not lived up to expectations. And because its designers were not considering the potential for personalized care in the era of data-intensive genomic research, it is not open-ended enough to assist clinicians in customizing treatments with the precision found in more expandable systems. In this respect, it was approaching obsolescence before it was out of the box. This system's chief virtue is in suggesting and recording billing codes, and it has raised institutional gross income accordingly, but the costs of its installation, maintenance, and dedicated staffing—wags on the clinical floors call it DEFEATS, the Defensively Exhaustive Full-Employment Act for Techies and Scribes—make it a net money loser for the hospital. Most important, in Hospital B and all those that will come to resemble it after the nation achieves nominally universal “meaningful use” in 2014, EHR technology will have complicated medical practice without yielding measurable improvement in clinical outcomes. ShouldersCorp's Dr. Love, who has seen some industries adapt to information technology well and others badly, frames the EHR challenge not as a competition between human and artificial intelligence—no machine threatens to render human judgment obsolete, particularly in the complex biomedical realm—but as the task of making the inevitable system design decisions align with one's professional mission. “I think we've learned in the software industry in the last 40 years that sitting down and building an expert system to duplicate the functionality of human beings is a really hard, intractable problem,” he says. “I don't want the computer sitting between me and the patient when I'm interviewing them or doing an examination. On the other hand, I don't mind having the computer sitting there on the side, providing me with reminders and assistance and, if it can do it fast enough, allowing me to enter information in real time as I'm gathering it. I haven't run into situations in which doctors are in any way concerned about ‘Is this robot taking my job?’” If a cinematic cautionary image is illustrative here, it may not be Frankenstein or HAL from 2001, an artificial creation attacking its creator, but Terry Gilliam's Brazil, where the future dystopian state's machinery is relentlessly overcomplicated and absurdly inept. “You've got to build [a system] with maximal flexibility in mind,” Dr. Love continues, “sit down and have long serious conversations and brainstorming sessions: ‘How might this system be asked to change in the next 20 years?’ You'll still miss some big ones, but you'll at least have accommodated some of the more likely changes that come along … . Over the next 30 to 50 years in the medical world, we'll see more radical change than we've seen in the communications world in the last 30 to 40 years. It seems that there's a perfect storm a-brewing here.”