Printer Friendly

QMR, the Quick Medical Reference 2.2.2.

QMR is a knowledge base containing more than 634 diseases oriented toward inpatient medicine, lacking common outpatient diagnoses, such as dysfunctional uterine bleeding and ankle sprain. Its purpose is to assist with differential diagnosis in cases that are diagnostically challenging. QMR is a "frame-based" expert system, in that it represents each disease as a matrix of numerical ratings for a set of clinical findings. To add a disease to QMR's database, researchers review the literature to compile clinical findings relevant to the disease and then assign a numerical rating to each finding for sensitivity, positive predictive value, and clinical importance.

QMR's installation is spartan but functional. The user must enter a serial number; otherwise, there is not copy protection. Like most DOS programs, QMR has a windowed text interface and a menu bar with drop-down menus. It requires minimal typing skills and familiarity with certain important keystroke combinations. Mouse support is optional. Windows and Macintosh versions of QMR have recently become available but are not included in this review. QMR has a useful, complete online tutorial. It is possible to use the tutorial as an example and jump back to the program, although navigation of these links can be confusing.

I tested QMR on an IBM PS/2-50, a clone DX 486-66, and a Compaq Pentium 60 MHz PC. Screen changes generally took less than a second on each computer. In contrast, QMR's major competitor, Iliad for Windows (Applied Informatics, Salt Lake City, Utah), took many times longer to perform similar functions.

QMR can perform four main functions: (1) generate a differential diagnosis from a set of clinical findings; (2) create a simulation consisting of a set of clinical findings and a "correct" diagnosis; (3) act as an electronic textbook of medicine, profiling diseases by findings, suggested work up, related diagnoses, and relevant journal articles; and (4) link automatically to MEDLINE using Grateful Med software (this requires purchase and installation of Grateful Med).

Learning QMR requires several hours of work on the tutorial, and enough time and experience to enter clinical findings accurately and completely.(1)(2) There are 4653 clinical findings, including patient history, physical examination, and laboratory tests. Many of the findings are quite specific and sometimes expressed in ungrammatical, awkward phrases; the lack of general, familiar expressions could be daunting to the casual user. "Term Completer" accepts a text fragment and presents a list of matching findings from which the user chooses. For example, typing "murmur" produces a list of 50 findings, such as "Heart Murmur Systolic Left Sternal Border Increased after Premature Contraction." It is worth scanning the Findings Index to get a feel for the working and range of findings available in QMR. Once QMR has generated a differential diagnosis, users can further ask QMR to suggest additional findings that may differentiate among proposed diseases to further define the differential diagnosis.

The process of entering, analyzing, and reanalyzing a patient case in QMR requires significant amounts of time, an important limitation in outpatient settings. QMR is helpful in analyzing difficult cases, if the clinician feels it is worth 15 minutes. Many busy clinicians will find that phoning a human consultant is more user-friendly and less labor intensive than entering data into QMR. However, QMR can provide a rich learning experience, especially for students.

With the assistance of specially trained clinician QMR operators, QMR has functioned successfully as an electronic consultant for difficult internal medicine inpatient cases in research settings.(3) A recent study evaluated the clinical performance of four computerized diagnostic systems (QMR, DXplain, Iliad and Meditel) against 10 expert clinicians. The developers of each program entered 105 patient cases. QMR selected the correct diagnosis as its number one choice about 20% of the time, and listed the correct diagnosis among its top 30 differential diagnosis list about 50% of the time.(4) QMR's performance was surprisingly similar to that of the other three programs, except that it tended to get the number one choice right more often, but had a less complete top-30 list. Each program suggested an average of two diagnoses that the expert had not considered but agreed was a valid addition to the differential diagnosis.

QMR and Iliad have almost identical features, except that Iliad's clinical simulation module provides extensive, real-time feedback on the user's performance, while QMR does not. QMR provides the automated link to the MEDLINE database, which Iliad lacks.

QMR is an inpatient internal medicine differential diagnosis generation program that suggests diagnoses even expert clinicians may not have considered. It cannot be relied on to select a single correct diagnosis. It requires an investment of time to become proficient and to enter each patient's data. These requirements restrict its clinical use to difficult cases. Its simulation mode, disease profiles, and linkage to MEDLINE provide unique educational opportunities. I would recommend it for teaching programs and practices that commonly deal with internal medicine diagnostic dilemmas.


(1.)Bankowitz RA, Blumenfeld BH, Janosky J. A prospective analysis of inter-rater agreement between a physician and a physician's assistant in selecting QMR vocabulary terms. In: Stead WW. Proceedings: the Fifteenth Annual Symposium on Computer Applications in Medical Care. New York: McGraw-Hill, 1991:609--13.

(2.)Bankowitz RA, Blumenfeld BH, Giuse Bettinsoli N, et al. User variability in abstracting and entering printed case histories with Quick Medical Reference (QMR). In: Stead WW, ed. Proceedings: the Eleventh Annual Symposium on Computer Applications in Medical Care. New York: IEEE Computer Society Press, 1987:68--73.

(3.)Bankowitz RA, McKneil MA, Challnor SM, Parker RC, Kapoor WN, Miller RA. A computer assisted medical diagnostic consultation service. Ann Intern Med 1989; 110; 824--32.

(4.)Berner ES, et al. Performance of four computer-based diagnostic systems. N Engl J Med 1994; 330:1792--6.
COPYRIGHT 1994 Quadrant Healthcom, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1994 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Ontai, Sidney
Publication:Journal of Family Practice
Article Type:Evaluation
Date:Dec 1, 1994
Previous Article:Office Practice of Medicine, 3d ed.
Next Article:Pearls.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters