This Article Abstract Submit a response to this article Services E-mail this article to a friend Alert me to new issues of the journal

H E A L T H  T R A C K I N G
F R O M  T H E  F I E L D  

19 January 2005  
  

E-Health: 
Steps On The Road 
To Interoperability

A large integrated delivery system takes action
on improving health care information exchange.

By Brent James

ABSTRACT:

Interoperable electronic medical records (EMRs) have the potential to produce better health outcomes while improving the efficiency of care delivery and reducing its costs. Implementation will require massive changes at all levels. In many instances, the costs of implementation could fall on one group, while savings will accrue to some other group. A successful transition strategy identifies a series of steps, where each step pays its own way, at the level of the local groups directly affected, and lays the foundation for the next step. Such a strategy implies an era in which large groups will likely play a critical role.

Almost four decades ago, Paul Beeson noted an oft-overlooked truth that lies at the heart of effective health care delivery policy. “After all,” he stated, “doctors and nurses are the only people who can possibly alter the conditions of patient care.”¹ Don Berwick defines a “chain of effect for quality” that more fully links policy to action.² Federal and state health care policy creates an environmental context for care delivery groups. Care delivery groups, in turn, create organizational contexts for clinical care delivery teams. It is the physicians, nurses, pharmacists, therapists, and technicians who make up clinical teams, however, who bridge the last, critical gap to patients. Environmental and organizational changes produce results only to the degree that they smoothly influence front-line care delivery interactions, while avoiding the friction and inefficiency of strong front-line pushback.

While philosophically driven by mission—a professional, social, and human ethic to help those with health needs—all levels of Berwick’s chain also face a financial reality. Clinicians and administrators working at the front lines, where real patients come into contact with real resources, live that balance daily. A simple rule defines front-line response to environmental and organizational change: You can’t destroy clinical productivity. A proposed change will have traction if it produces easily recognized, significant improvements to patient outcomes. Alternatively, a successful change could increase income or reduce the time and resources necessary to deliver a service. The most effective strategies improve both sides of the value equation simultaneously.

Over the past seven years, electronic medical records (EMR) systems have begun to move out of a few pioneering organizations and into broad use. Today they have passed the tipping point. Installation numbers are still small but increasing very rapidly.³ Their long-awaited expansion from research into practice arose from two parallel developments, which together address both sides of the value equation: (1) Quality improvement began to more precisely define the structure and content of effective clinical practice; and (2) medical informaticists began to more effectively blend EMR systems into clinical workflow.

Defining A Practical Structure For Clinical Information Through Quality Improvement

Around the turn of the nineteenth century, medical professionals adopted the scientific method as a means of understanding and treating disease. On that foundation, the intervening century has seen an unprecedented explosion in clinical knowledge and effective treatment.⁴ However, treatments that can heal can also harm. The Institute of Medicine (IOM) showed that preventable injuries in U.S. hospitals are a major source of patient deaths.⁵ Subsequent IOM reports documented a wide gap between theoretically possible and actual health care outcomes on a broad scale and proposed a model for national health care system reform.⁶ The performance chasm arises primarily from complexity and clinical uncertainty.⁷

Evidence-based clinical best-practice guidelines, or protocols, provide one proven approach to managing complexity. Real patients show unique differences in underlying genetics, physiology, anatomy, response to treatment, ability to participate in their own treatment, resources, preferences, and values. It is therefore impossible to build guidelines that perfectly fit every person. However, a clinical team can practice “mass customization.” It can create a shared baseline, then adapt its common approach to the needs of each patient. Typically, a clinician will change about 5–15 percent of a protocol for a particular patient. That means that the most important tool in clinical care delivery—the trained human mind—can focus its attention on that small portion of a total care process that needs modification. Such an approach not only helps clinicians manage complexity, it also greatly increases clinical efficiency.⁸

The use of clinical process management protocols to close the quality gap, while reducing the costs of care, goes hand in glove with EMR systems. EMRs provide an elegant tool to blend guidelines into clinical workflow. For example, we have never been able to implement more than four clinical protocols at one time in a single clinical setting. Beyond that point, the shuffling of paper overwhelms the care delivery team. EMRs, however, appear to have no upper limit. They have the capacity to blend data collection and decision support seamlessly into clinical workflow. In counterpoise, practice protocols structure a care delivery process and identify key data elements essential to clinical management and improvement.⁹ Quality improvement protocols define the clinical content that is essential to effective EMR use.

Using Data Automation To Improve Clinical Workflow

Implementing an automated clinical data system can be a massive undertaking. It requires changes not only to data flow, recording media, and related administrative processes, but also to the mental models that clinicians use to understand and use medical records. One way to ease the transition from paper to electronic records is to break the total transition into a series of small, more manageable, steps, where each step pays its own way and lays the foundation for the next steps.

One such scheme could proceed in several steps from the automation of billing and scheduling (step 1); to the automation of laboratory and imaging work, pharmacy, and specialty consultation (step 2); to the introduction of an “electronic file cabinet” (with no consideration of encoding yet) with patient records available across multiple sites (step 3); and finally to a three-stage encoding process as the last step toward interoperability (steps 4–6).

Automated billing and lab functions. Most U.S. clinical settings have already automated patient scheduling and billing (step 1). Such automation makes good business sense: It is cheaper, faster, and more effective to schedule and bill using well-designed computer systems than to do so manually. Similarly, most major laboratories have full data automation (step 2). Modern pharmacies routinely rely on computerized systems for operative internal workflow. More and more, clinical imaging departments store images digitally, and imaging reports are prepared and stored as digital documents. In each instance, automated systems increase productivity; automation pays its own way.

Clinical note taking. EMR systems (step 3) replace handwritten or dictated clinical notes through a range of clinician-defined text modules called “hot text” or “hypertext.” Clinicians load a series of boilerplate templates (for example, a text template for a short history, an electrocardiogram report, or a neurological examination) to describe a patient encounter, then quickly modify embedded text fields to reflect accurate clinical findings (the concept of mass customization applied to clinical charting). Such tools make it possible for clinicians to document while interacting with their patients, often with patients’ direct participation. Properly implemented, such systems pay their own way through improved clinical workflow and productivity.

For example, in one large outpatient clinic group, such a system has a total hardware, software, and telecommunications cost of about $5,000 per practice per year, while transcription costs fell by $8,000–$14,000 per year.¹⁰ With less stand-alone time spent on documentation, primary care physicians can fit about one more patient visit into a workday. Physicians leave the office an hour or two earlier, because they need not remain to complete the day’s charting or to review transcription. Medical records are always immediately available, thereby avoiding the overhead associated with missing charts and the time lags associated with nonelectronic data flow. A secure Web-based interface means that on-call physicians can review complete medical records as they respond to patients’ telephone contacts, through any Web-enabled computer (for example, from their home). Clerical staffing for the group fell more than 30 percent. The group’s EMR is the foundation for a series of other productivity tools, such as applications that help clinicians review and manage laboratory results, that manage patients’ phone calls and e-mail messages, and that link a patient’s chart to the medical literature through context-sensitive help.

Initial EMR implementation (step 3) asks clinicians only to shift record keeping from paper to electronic format. It does not require that they simultaneously modify their internal mental models for clinical charting. With the transition to an electronic format complete, step 4 begins to shift the internal structure of the medical record itself. It initially asks clinicians for only two changes. It requires encoded patient medication lists, with allied allergy information. It further asks physicians to maintain a focused problem list, with major chronic conditions stored as encoded (computer actionable) fields.

Medication ordering and management. A computerized medication list represents a first step into computerized physician order entry (CPOE). By design, such electronic medication orders are complete, without missing fields, confusing abbreviations, or illegible handwriting. The EMR has the foundation it needs for full medication decision support. It can calculate ideal dosing for a particular patient, check for allergies and drug-drug interactions, and assess related laboratory values. It can transmit prescriptions to pharmacies automatically. As with each previous step, it pays its own way through clinical productivity. Physicians can more quickly pull up a “pick list” of their commonly used medications, select a drug, drop it onto a patient’s medication list, modify the dose and route, then hit “print,” than they can handwrite a prescription. Routine patient requests for refills, or for a complete set of prescriptions as they move between mail-order pharmacies, shift from a long list of handwritten scripts to a simple “highlight, hit ‘print,’ then sign.” In the large outpatient group mentioned earlier, many physicians adopted the EMR because of the efficiencies inherent in automated medication management alone.

Chronic disease management. Step 5 represents chronic disease management embedded into a clinical practice. The encoded problem lists generated in step 4 act as disease registries. Using standard encoded templates, the EMR can implement evidence-based best-practice guidelines in the form of customized decision support, such as disease-specific, guideline-driven clinical worksheets and electronic “tickler” files that apply best-practice standards and then produce lists of patients who need intervention. Step 6 anticipates a day when EMRs become the norm, with a new model for medical information storage widely in place from initial clinical training through care delivery practice.

Implications

Potential cost savings. Jan Walker and her colleagues argue that interoperable health information infrastructure could produce cost savings in excess of $77 billion per year—an amount far in excess of the estimated costs of implementation over time.¹¹ Creating that system, though, will involve a long series of one-time, indirect costs. Although the projected cost savings are compelling in the final state, successful implementation will depend upon a series of successful transition states, where the business case may not be as clear. In many instances, those who must bear the costs will not be those who will harvest the savings. Implementation will require carefully coordinated national policy and organizational planning.

For example, many local physician groups rely upon income generated through imaging and laboratory testing as a key component of their financial balance sheet. Those “redundant tests” may be a key part of their income stream and financial viability. Even if such groups are philosophically committed to ideal, nonwasteful care, they must somehow make a bottom line as they move through the transition. How will they book sufficient operational savings to cover not just the new data systems but also lost revenues from reduced testing rates? To fully realize potential savings, care delivery organizations will also need to rebuild other parts of their organizations. For example, clerical services must change. Clinical savings from better care—which Walker and colleagues correctly identify as potentially the largest pocket of savings—will require that the organization learn and implement effective methods for process management. Such change entails its own transitional costs. As mechanisms are found to ease these transitions, actual cost savings may turn out to be far less than general, system-level estimates.

Technical challenges. Similarly, Walker and colleagues correctly note that Level 3 implementation will require that someone build and maintain a series of eight to twenty software interfaces per major external provider. During the transition phase, that estimate may be conservative. For example, a single outpatient oncology EMR application required eight interfaces to fully link. The oncology subsystem is but one of almost sixty such subsystems in use in our integrated delivery system. Those interfaces must be in place before a local care delivery group can begin to make the transition.

Physicians often balk at EMRs (step 3 systems) that do not integrate clinical pathology, pharmacy, and imaging information (step 2 information). They perceive that the overhead of shifting among different systems and manually moving information damages clinical productivity. The transition will thus necessarily move through a “big system” phase. Large organizations—integrated delivery systems, large physician groups, or community-level consolidators—are better positioned than smaller organizations to align resources and to deal with the complex organizational structures necessary to build and maintain complicated interface systems. When Walker’s Level 4 interoperability is in place, it is easy to imagine a solo physician buying an EMR package much as he or she might purchase Microsoft Office today—but that era is not yet here.

Need for national standards. Finally, this discussion relies upon the eventual creation and implementation of interoperability standards. At present, the cost of building interfaces can equal or surpass that of purchasing a major clinical data subsystem. Interoperability standards would reduce or eliminate those costs. Although outgoing Health and Human Services Secretary Tommy Thompson identified national standards for key subcomponents (for example, LOINC for lab data; NCPDP for pharmacy data; and HL7 CDA for text-based imaging reports), the country still lacks incentives that will lead software vendors to implement such changes. Those incentives must include certification standards. Today it is still possible to buy software that claims to be standard-compliant but that cannot interoperate with software from other vendors that claim compliance with the same national standard.

A growing national consensus holds that EMRs can eliminate waste, increase operational efficiency, and improve clinical outcomes. Some have called for direct subsidies to help care provider groups make the transition. However, investment in critical infrastructure—interoperability standards, certification standards, market and regulatory incentives, and viable transition plans that focus on front-line health professionals—may prove far more effective in achieving real change.

NOTES

1. P.B. Beeson, “Special Book Review: Sickness and Society,” by Raymond S. Duff and August B. Hollingswood, Yale Journal of Biology and Medicine 41, no. 5 (1968): 240.
2. D.M. Berwick, “A User’s Manual for the IOM’s ‘Quality Chasm’ Report,” Health Affairs 21, no. 3 (2002): 80–90.
3. David C. Classen, First Consulting Group, personal communication, December 2004 (in reference to an internal survey conducted through the Scottsdale Institute, 2004, regarding rates of adoption of EMRs and CPOE).
4. B.C. James, “Quality Improvement Opportunities in Health Care: Making It Easy to Do It Right,” Journal of Managed Care Pharmacy 8, no. 5 (2002): 394–399.
5. L.T. Kohn, J.M. Corrigan, and M.S. Donaldson, eds., To Err Is Human: Building a Safer Health System (Washington: National Academies Press, 1999).
6. Institute of Medicine, Crossing the Quality Chasm: A New Health System for the Twenty-first Century (Washington: National Academies Press, 2001).
7. D.M. Eddy, Clinical Decision Making: From Theory to Practice (Boston: Jones and Bartlett Publishers, 1996).
8. James, “Quality Improvement Opportunities.”
9. B.C. James, “Information System Concepts for Quality Measurement,” Medical Care 41, no. 1 Supplement (2003): I-71–I-79.
10. Based on the author’s personal experience in implementation of the Clinical Workstation outpatient EMR within Intermountain Health Care’s 100-plus clinic settings, 2002–2004. For more information, e-mail the author, bjames{at}ihc.com.
11. J. Walker et al., “The Value of Health Care Information Exchange and Interoperability,” Health Affairs, 19 January 2005,
content.healthaffairs.org/cgi/content/abstract/hlthaff.w5.10.

Brent James (bjames{at}ihc.com) is vice president for medical research and executive director of the Institute for Health Care Delivery Research, Intermountain Health Care, in Salt Lake City, Utah.

DOI: 10.1377/hlthaff.w5.26
©2005 Project HOPE–The People-to-People Health Foundation, Inc.