Emergency Medical Informatics Section Newsletter - June 2007, Vol 12, #2
Vernon D. Smith, MD
The Emergency Medicine Informatics Section and its members have been quite active over the last several months! Brian Keaton, ACEP President and the section’s Board Liaison has been extremely active at the e-Health Initiative, the Markle Foundation, the American Health Information Community (AHIC), and the Centers for Disease Control, and will soon begin to serve a two-year term on the AMA Health IT Advisory Group. Drs. Barthell, Coonan, Kamens, McClay, Rothenhaus and several other section members are also active in the various standards development and certification entities, including the Emergency Care Special Interest Group (EC-SIG) and the Emergency Department Information System (EDIS) Functional Profile Workgroup of the Health Level Seven (HL7) standards development organization, the Certification Commission for Healthcare Information Technology (CCHIT), and the Health Information Technology Standards Panel (HITSP). ACEP member participation in all these groups is critical to ensure the needs of emergency physicians are considered. Please review the remainder of this article, which details selected activities within the section. To continue to meet these challenges, the section is seeking an increase in dues to support its increased activities in all of these areas. In addition, section members are sought to participate in these groups and activities. Please contact Drs. Kamens and Rothenhaus directly, or Angela Franklin c/o (firstname.lastname@example.org; (202) 728-0610) to learn how you can participate.
December Section Meeting Highlights
A meeting of the Emergency Medicine Informatics Section was held on December 12, 2006 in conjunction with the Pennsylvania EDIS Symposium in Orlando, FL. Dr. Vernon D. Smith, Secretary/Newsletter Editor and Chair Elect, chaired the meeting. In the process of approving the October 16, 2006 meeting minutes, there was quite a bit of discussion, and corrections were made to the minutes regarding Dr. Randall Case’s comments at that meeting, and confirming the approval of the proposed dues increase at the annual meeting of the section on October 16.
Unfinished business was also taken up at this meeting—the election of a Chair Elect and of a Secretary/ Newsletter Editor Elect. Dr. Rothenhaus explained that the nominating committee, consisting of the chairman, chairman-elect, immediate past chairman, and secretary, had agreed to nominate Dr. Smith as Chair Elect of the section, and Dr. Al Villarin, MD, FACEP, as Secretary/Newsletter Editor Elect. Drs. Smith and Villarin were approved by those present.
Dr. Todd Taylor, MD, FACEP, recommended that the section draft a policy statement for the ACEP Board that would address the question: "What is an adequate emergency department information system?" Significant discussion followed, and the group agreed that Dr. Taylor and Dr. Nathanson should draft proposed language.
Proposed Dues Increase Request
The topic of an increase in annual dues to support its increased activity has been an active topic within the section for some time. An increase in the annual dues was first raised and discussed at length without a vote at the 2005 annual meeting of the section as a way to support increased section activities, given that the focus on informatics had increased significantly. An increase in annual dues was proposed again for vote at the annual meeting of the section on October 16, 2006. The proposal to increase annual dues from $35 to $50 for active members was approved by the majority of members present at that time.
The increased funds will be used to support a variety of new activities by the section and its membership, including:
- Provide seed money for the section’s project fund—used to encourage and fund section members to pursue projects that endeavor to advance Emergency Medicine Informatics;
- Fund a section paper competition on the use of emergency medicine informatics, which promises to advance ACEP’s thought leadership in this area, and
- Help the section fund its continued participation in the annual Emergency Department Information Systems Symposium (http://www.isedis.com/); hosted by Pennsylvania ACEP—the course has developed over the years into the country’s premiere program for exhibition, demonstration and sales of emergency department information systems.
Please stay tuned for an update on the disposition of this issue in the near future.
Developments in Health IT
The following is a just a sampling of recent developments:
- The Integrating the Healthcare Enterprise (IHE) has created dozens of functional profiles that are for both transport, and content-based. Two emergency department work products are being prepared at IHE related to: 1) emergency department referrals, and 2) defining a standard for the electronic version of the emergency department chart. The IHE is a good place for emergency physicians and vendors to interact; physicians are able to provide input into standards and to give vendors a greater understanding of emergency department needs.
- On April 17, section Board Liaison Dr. Brian Keaton testified before the National Committee on Vital and Health Statistics (NCVHS) on the topic of "Consumer Controls for Sensitive Health Records" in the Nationwide Health Information Network (NHIN) environment. In his testimony, Dr. Keaton emphasized the need for balance between patient’s right to control sensitive information, and emergency physicians’ ability to deliver the best possible medical care. NCVHS is expected to make a recommendation to Secretary Leavitt on this topic in the next several months.
- The Certification Commission for Healthcare Information Technology (CCHIT) has announced it will expand and realign its program to certify technology used in electronic health records (EHRs) systems. In addition to the existing work groups that are addressing certification criteria for office- and hospital-based EHRs, CCHIT will add a work group addressing emergency departments (EDs). On February 7, the CCHIT prioritized certification of ED EHRs for its expanded certification work, finding that EHR adoption in the ED had great potential impact because of the ED setting's high annual healthcare costs ($100B), high readiness, and the potential for a "quick start" due to a nearly complete HL7 profile, which identifies functions that must be contained in an EHR system for EDs. An ED Workgroup has been formed to help carry out this work.
- The Office of the National Coordinator for Health Information Technology (ONC) is planning its first trials of the Nationwide Health Information Network (NHIN), which will connect state and regional health groups to share electronic health records (EHRs) nationwide to create a "network of networks." On June 5, ONC released Request for Proposal (RFP) Number 07EASRT070057 entitled "National Health Information Network Trial Implementation," which is due by 2:00 pm EDT, Monday, July 9, 2007. According to the RFP cover letter, awardees will provide support to ONC for the National Health Information Network (NHIN) Trial Implementation. The trial implementations must demonstrate core services implementation of summary patient record exchange as per the AHIC Emergency Response EHR use case (for which ACEP and section members have provided significant input) and must demonstrate two of the other AHIC use cases: EHR – Lab Results; Consumer Empowerment – Registration and Medication History; Biosurveillance; Patient Access to Clinical Information; Quality, or Medication Management. Contractors may be asked to implement additional use cases by modification to the contract ONC will award ten contracts. The initiative is in response to President Bush' goal of creating EHRs for most Americans by 2014. In the first of many steps toward this goal, ONC awarded contracts to Accenture, Computer Science Corporation, IBM and Northrop Grumman in November 2005 to develop prototypes through partnerships with regional health care organizations. The trials will test how the system would operate on a larger scale. Additional details regarding the RFP may be found at: http://www.fbo.gov/spg/HHS/PSC/DAM/07EASRT070057/SynopsisP.html.
Section members are urged to become involved in the work in all these areas, particularly the EC-SIG and EDIS functional profile work, as well as involvement in significant, longer term work, which is being led by Drs. Kamens, Rothenhaus and several other section members on revising the Data Elements for Emergency Department Systems (DEEDS). Please contact Drs. Kamens and Rothenhaus directly, or Angela Franklin at ACEP (email@example.com; (202) 728-0610) to find out how to participate.
Michael Gillam, MD, FACEP, would like to alert section members that the National Institute for Medical Informatics (NIMI) is providing several grant opportunities. NIMI would provide $1,000 to grantees for interesting projects, and $3,000 for more complex projects. Those interested must submit a single page description to Dr. Gillam for consideration.
In addition, Microsoft will sponsor one-year fellowships, with compensation in-line with a resident’s salary and located in Washington DC. Opportunities for a single month, or "as many months as desired" electives are also available. Participants would work in emergency medicine informatics in Washington, DC. Those interested must submit a letter of intent and curriculum vitae. Please contact Dr. Gillam directly, or Angela Franklin at ACEP (firstname.lastname@example.org; (202) 728-0610) for additional information.
Back to Top
Call for Essays for Longevity and Tenure Awards
Angela Franklin, Section Liaison
The ACEP Section of Careers in Emergency Medicine is soliciting nominations for an award for emergency physicians in the following two categories:
- A Longevity Award for the physician with the longest active career in emergency medicine.
- A Tenure Award for the physician with the longest active career in the same emergency department.
Recognition is also given to those physicians who are still actively practicing emergency medicine after 20, 25, 30, and 35 years. Further details are available online at www.acep.org (click "Member Section" then "Sections"). Members do not have to belong to the Careers Section. Nominations are due by July 9 to Tracy Napper c/o email@example.com, fax: 972-580-2816, or by mail to:
Section of Careers in Emergency Medicine
PO Box 619911
Dallas, TX 75261-9911
Back to Top
Welcome New Informatics Section Members
Maj. Christofer A Strode, MD - APO, AP
Brian Laux, MD PhD FACEP - Bath, NY
Ian W Cummings, MD, PhD - Port St Lucie, FL
Jaime Lent, DO - Everett, PA
John C Brown, MD, FACEP - Frisco, TX
John E Braden, MD - Miami, FL
Jonathan A Finkelstein - Staten Island, NY
Juan Juarez, Jr., MD - League City, TX
Jason C Morgan, MD - Drexel Hill, PA
Michael M-T Nguyen - Sherwood Park, Canada
Michael P Tull, DO - Troy, MI
Teresa Sullivan Dolan, MD - State College, PA
Walter W Jacquemin, MD, FACEP - Alpena, MI
Back to Top
Optimal Laboratory Data Display
Jonathan A. Handler, MD, FACEP
Craig F. Feied, MD, FACEP
Michael Gillam, MD, FACEP
Mark Smith, MD, FACEP
Computers have long played a major role in the reporting of laboratory data. Computerized medical records have been in existence since 1959. In fact, it was the need to efficiently report laboratory results that often led to the first computer terminal installations in the hospital clinical arena. While there have been some advances in the display of medical laboratory data over these decades, many if not most hospitals are still using legacy reporting systems with unfriendly user interfaces and suboptimal data display formats. Even the more advanced data display systems, such as those using a graphical user interface are often difficult to use and do not optimally display the needed data.
Despite the ubiquity of laboratory data display systems, the heavy use hospitals place on these systems and the critical function they serve, relatively little study has been done and relatively little has been written on the subject of optimal data display. In turn, the advances in this area have been small despite huge advances in the technology available to display information.
Defining certain principles of data display will help to identify an optimal data display system. Our focus will be on the display of laboratory data to emergency department personnel. This physician population is often under the greatest time pressure to make decisions with the aid of laboratory data, and makes these decisions without the benefit of a prior history with the patient. We propose that a system that works well in such a data-hostile environment will work well in almost any other area of the hospital.
The primary principle of data display is that all data is not of equal importance. We refer to this inequality as being "non-democratic".
Laboratory tests directly relevant to the patient’s likely disease process are generally more important than collateral tests ordered to rule out unlikely diagnoses or those that come packaged with the relevant test as part of a "panel". For instance, if a physician is concerned about hyperkalemia in a patient, the chloride level that comes packaged with the potassium level in a "chemistry panel" may be less important than the potassium level itself.
Laboratory results that are unexpected are generally more important than those that are expected. For instance, an amylase that a physician thinks will probably be normal but in actuality is ten times the normal value is more important than an elevated serum glucose in a patient already known to have an elevated "fingerstick" glucose.
Laboratory results that are abnormal and diagnostic of serious disease are generally more important than those that are normal or diagnose minor diseases. As an example, an elevated troponin level is usually more important than an elevated uric acid level.
Laboratory results that deviate greatly from the norm are usually more important than those that minimally deviate. Thus, a serum sodium level of 130 is low, but it is not as important as a serum sodium level of 105.
Laboratory results also differ in importance based on temporal relationships. For instance, given 100 serial glucose checks taken over 3 weeks, the most important one is probably the most recent one, especially if it was done today. The next most important one is likely the second most recent glucose check. All-time high and all-time low levels may be more important than randomly selected levels from the distant past.
Other factors play into importance as well. Spurious results, improperly drawn results, and results suffering from other errors are not as important as results without these errors.
Current systems suffer seriously from a failure to take into account the non-democratic nature of laboratory data.
While distant data may not be as important as the most recent data, the aggregation of such data can be helpful to identify important trends. For instance, a slowly dropping hematocrit can help more quickly identify a patient with a slow, occult gastrointestinal bleed. The trend of the white blood cell count can be helpful in following the progress of patients with infectious diseases such as pneumonia. Most often, trends are clearly identified when the data is presented in graphical format.
Current systems rarely display data in graphical format.
While most physicians are familiar with the normal range of values for common tests, this range can vary from hospital to hospital, and even from day to day. In addition, for less common tests, the normal range may not be known. It is important to provide the clinician with some orientation for a given laboratory value with regard to its normal range.
Most current systems suffer from an inappropriate segregation of the data. While a handwritten chart can efficiently and effectively summarize all the data and present it at a glance, many (if not most) computer reporting systems segregate the data according to the machine on which it is run. For instance, a troponin level and a creatinine phosphokinase level may be reported separately because they are run on different machines, even though the clinician considers them related and important data to be taken together.
In fact, it is almost always desirable to display all the data at once so that comparisons, contrasts, relationships and patterns can be easily discerned. This is extremely rare on most systems, which often display data segregated by both time and data type, making comparisons and pattern recognition difficult or impossible. Data aggregation allows the data to become relational and thus more useful to the clinician.
A number of models have been proposed to improve data display.
One interesting model from Germany proposed the display of data in a circular format in this model, only data relevant to a certain disease process would be displayed in the data circle. Data that should have been obtained for evaluation of this disease process would deform the circle inward, and abnormal data would deform the circle outward. This model was interesting in that it was extremely unique, but suffered from numerous limitations. The model failed to provide immediate access to the magnitude or direction (high or low) of any result. Further, it imposed relationships between different types of data on physicians, rather than allowing physicians to relate different parts of the data on their own. Finally, temporal relationships could not easily be made, as each set of laboratory data represented a unique snapshot in time.
Another interesting model came from a well-known authority in data display, Edward Tufte. This model allows the display of every single piece of laboratory data for a patient on a single page, or just a few pages. Each data point is plotted inside a small square, with high and low ranges clearly shown to provide orientation. With this model, the data is relational, trends can be visualized, and the data is aggregated both by type and temporally.
While excellent conceptually, the model suffers from a few weaknesses, the most notable being that the data is treated almost democratically. The current value of a test holds a special position in that it sits above the graph of old results. However, the majority of space on the page is devoted to old (and therefore less relevant) results, and one cannot easily view and compare current data (scanning the current data is difficult). Neither all time high and all time low values, nor the normal range for the test are numerically demarcated. It is not easy to see at a glance whether the current value for a test is normal or abnormal. The graphical data is presented on a time line separated into two parts. The larger part on the right side is devoted to more recent events, and the smaller part on the left to more distant events. This change in the graph makes it difficult to read, and still overemphasizes past events.
We have seen many other models of data display suggested in the literature, as well as those present on existing laboratory reporting systems. Unfortunately, so far all of the systems we have encountered have suffered similar shortcomings.
Given the prevalence of inadequate data display systems, we were pleased to stumble upon a system of data display that seemed to closely meet our defining principles for optimal laboratory data display. The system is based on another Edward Tufte method for data display, but one intended for financial data. We modified the system slightly to meet the needs of laboratory data display. The system consists of a tiny graph of all prior data, which we have modified to include bars displaying the normal range for the test. Adjacent to the graph is the name of the test, its normal range, the current value, the immediately prior value, and the lifetime high and low values for the data. If the current or immediately prior values are abnormal, they are preceded by an "H" or an "L" to demarcate whether the value is high or low.
This system has numerous advantages over any other proposed system we have seen. It is extremely efficient, allowing thousands of pieces of data to be aggregated onto a single page. The arrangement of the data means that despite its density, the physician can easily process it. The bold type and different coloring for the test name and current value allows the clinician to see at first glance the most important information on the page. The columnar arrangement of the data allows rapid scanning of the data, and makes the data relational by displaying all the data together on a single page in a way that allows rapid comparisons.
Abnormal values are immediately identified by the presence of an additional letter ("H" or "L"), as well as by falling outside the normal range bars on the graph. The current value, the immediately prior value, and the all time high and low values for a test are shown both in the graph as well as in numerical format. All of these things make the data appropriately non-democratic, assigning greater value and space on the page to more important pieces of information.
Orientation is provided in three forms – by giving the normal range for the data in numerical format, in graphical format, and by highlighting abnormal values with additional comments ("H" or "L").
Trend analysis is provided in two ways. First, the immediate trend is easily discerned by comparing the adjacently located current and immediate past values. Long-term trends can be clearly visualized on the small graph of the data.
Unlike both the German model and Tufte’s initial model for medical data display, this system is instantly understood by its viewers, who have seen similar presentations of data before, but never combined and organized in exactly this way.
We have proposed a new system for showing laboratory results that meets all the principles of optimal data display. We are unaware of any other system that displays data as efficiently or effectively as the one we propose. Further study and test implementation of such a system is warranted.
Back to Top
My Problem with the EMR
Curtice Wong, MD
Manhattan Beach, CA
As much as I might wish for a return to the Good Old Days when a patient’s own physician was the repository of all knowledge regarding the patient in front of them, it is a wish to return to a time that was headed for the history books before I graduated medical school. The Electronic Medical Record (EMR), as it exists today, is a reflection of our profession’s origins rather than the way we practice today. The EMR owes its form to Medicine’s humble beginnings where the practice of medicine existed between one doctor and one patient. It is still a document that records the interaction of the patient with a member, or institution, providing care. It is an island of data. Though a more unified whole than in the past, the contained data is largely location dependent. One doctor’s office. One hospital. One integrated delivery network. Delivery of healthcare in the current environment is a team-based, multi-site, collaborative effort that still functions with 19th century documentation tools even though our EMR vendors would have us believe otherwise.
When a patient appears at an Emergency Department (ED) for unscheduled care, the dataset available to providers at that location is dependent on
- Whether the patient is able to give details of past medical/surgical problems, medications being taken, and allergies
- Data stored on-site
- Whatever "hardcopy" or electronic data that might accompany our patient.
The connection to lab, historical, medication, or radiographic data created outside the immediate retrieval range of the provider’s EMR is laborious. Current schema for the exchange of or access to off-site data involves various user authentication schemes and data exchange protocols. A truly accurate/useful EMR would allow an authenticated user to open a patient’s medical record and view pertinent data on the patient independent of where the user is located. Current data exchange and transfer protocols rely on the so-called client-server model of computing technology, a method "so last century" in the language of the street.
From a clinical standpoint, so-called Web 2.0 applications that have begun to appear online show the promise of the software-as-a-service model of computing in which applications have moved from the local user’s personal computer/hard drive/local storage/server to a computer networked to a larger virtual computer containing applications and storage that exists online allowing new forms of collaboration and interaction between care providers and patients. For example, Allscripts™ is offering free e-prescribing tools (http://www.nationalerx.com/) that promise to link all the major Pharmaceutical Benefits Managers (PBMs), pharmacy data clearinghouses and providers to a common database through RxHub (www.rxhub.net) that might contain all of a patient’s medication history and the provider who wrote the prescription. A connection to this data source effectively solves a large part of the Medication Reconciliation issue. Also in the application is a type of mini problem-list that links a diagnosis to the prescribed medication. Another application getting some buzz is Practice Fusion, http://www.practicefusion.com/ that offers practitioners the prospect of an entirely free EMR online. EMR vendors have been slow to recognize this evolution as have application developers.
Feb. 7th CIO Insight (Ziff-Davis)…
Robert Carter, executive vice president and CIO of Memphis-based FedEx, the $32 billion shipping company, describes the difference between the early days of the Internet and the current Web 2.0 as "a shift from destination to connection. The initial onslaught of the Internet was all about destination. You'd go to FedEx.com or Amazon.com. But Web 2.0 is about connectedness. The Internet is becoming the Global Interconnect." That emphasis on connection is not only translating to more blogging and more wikis, but is also transforming the way software applications are created in the form of Web services. Says Carter: "Web services, the modular components that allow systems to interact and connect with each other, are a primary enabler of this move toward more connection."
I forgot how I first encountered the Coghead website; but, it offers a glimpse of tools just appearing on the web that give end users the kind of programming tools that allow them to build online database applications that allow collaboration in ways that client-server apps cannot. There are drag and drop visual application builders that give end-users the ability to develop and build applications without requiring them to learn C++, Java, or Perl. I do not know the degree of customization allowable through direct programming or scripting. As these tools mature, members of the Informatics Section might want to do some exploratory research in the area of putting together the infrastructure that might allow the dream of instantly available, location-independent data access. Until now, application development and deployment were at the mercy of software engineers who often did not know the requirements of their end user.
Some non-medical Web 2.0 resources that illustrate how this novel ‘platform’ might actually facilitate collaborative care in the current environment:
Some links to where venture capital is going in this article:
Web 2.0 applications for the EM community? Azyxxi solved the problem of data retrieval on an institutional level where disparate data "silos" were combined into a unified whole. The EM community needs the same kind of middleware glue for the Internet or an application that supplants the present generation of EMRs. The Informatics Section may want to consider what infrastructure requirements might be needed to influence the migration of users and important data to a more broadly accessible collaboration-friendly computing platform. Next, define the initial ‘content’ that the majority of users with a vested interest would want to make it usable. Discharge summaries? Lab data? Radiology images, or just the radiologist’s report? Billing application? Problem/Diagnosis lists? Medication lists? EKG's?... CCR/CCD documents? The Informatics Section has an incredible talent base that could work on the security, user groups, data feeds and table content for a RHIO-oriented computing utility.
Some problems looking for a solution:
- Create the equivalent of the Web’s URL; by accessing a medical record, a physician’s computer/computing system uses the patient’s ‘resource locator’ to search for both local and remote data.
- The equivalent of the Internet’s "Domain Name Service" that would use the "patient URL" to query remote computers.
- There is no incentive for the sharing or uploading of the data needed to make this system work. Nor is there a regulatory mandate.
- Whether there is an actual application or a server that would cache the desired data, there is no person or entity to maintain the infrastructure of such a data warehouse were it to be established. Funding needed to "keep the lights on."
- Creating demand. What is the "Killer App" that will drive demand? Do we need a study that shows that having this data available improves care, decreases overall costs to the system, or decreases ED evaluation times? For some commentary from "Digital Healthcare and Productivity" from the Cambridge Healthtech Institute, see their newsletter of March 20, 2007. This may provide a starting point in the development of such an application.
Back to Top
This publication is designed to promote communication among emergency physicians of a basic informational nature only. While ACEP provides the support necessary for these newsletters to be produced, the content is provided by volunteers and is in no way an official ACEP communication. ACEP makes no representations as to the content of this newsletter and does not necessarily endorse the specific content or positions contained therein. ACEP does not purport to provide medical, legal, business, or any other professional guidance in this publication. If expert assistance is needed, the services of a competent professional should be sought. ACEP expressly disclaims all liability in respect to the content, positions, or actions taken or not taken based on any or all the contents of this newsletter.