Content area
Abstract
Abstract
This report describes a hospital-wide clinical computing system that permits physicians, nurses, medical students, and other health workers to retrieve data from the clinical laboratories; to look up reports from the departments of radiology and pathology; to look up demographic data and outpatient visits; to look up prescriptions filled in the outpatient pharmacy; to perform bibliographic retrieval of the MEDLINE data base; to read, write, retract, edit, and forward electronic mail; and to request delivery of a patient's chart.
During a one-week study period, from 300 video display terminals located throughout the hospital, 818 patient care providers used a common registry of 539,000 patients to look up clinical and laboratory data 16,768 times; 477 other hospital workers used the patient registry 46,579 times.
In a separate study of 586 health care providers, 470 (80 per cent) indicated that they used computer terminals "most of the time" to look up laboratory results; in contrast, 48 (8 per cent) preferred printed reports. Of 545 hospital workers, 440 (81 per cent) indicated that the computer terminals definitely or probably made their work more accurate, and 452 (83 per cent) indicated that terminals enabled them to work faster.
The large amount of use by clinicians and their judgment that the computer has been so helpful to them suggests that a reliable, comprehensive, and easy-to-use computer system can contribute substantially to the quality of patient care. (N Engl J Med 1985; 312:756-64.)
Full text
IN the early 1960s, with the increased availability of digital computers, many of our colleagues were concerned about the potential encroachment of this new technology on the profession of medicine and on the traditional rapport between doctor and patient. Would these machines result in the dehumanizing processes that had been associated with the Industrial Revolution? Would modern times destroy the art of medicine? The debate was frequently lively, and a rejoinder that we found useful in those days, one borrowed from experimental psychologists,1 was that any doctor who could be replaced by a computer deserved to be.
Coupled with these concerns were excessive expectations. It is often true that the less direct experience people have with a machine, the more they tend to react to it with both fear and unreasonable admiration. The electronic digital computer, with its capacity to hold large amounts of data and to execute multiple, complex instructions with great speed and accuracy, is indeed an awe-inspiring device -- one that stimulates comparisons with the brain itself and fosters remarkable prophecies.
In medicine, as fear of the computer waned, expectations increased and prophecy became a substitute for accomplishment. Computer manufacturers, in turn, moved to capitalize on these great expectations. Some advertised their machines as panaceas for the medical community; they sold what they called total hospital information systems that were at best partial and contained remarkably little information. When the dust settled, in part on expensive, unused computer terminals, hospitals that had purchased these systems found that they had spent a great deal of money and received little in return -- usually partially working billing systems. From the patient's perspective, the principal difference was that the bills, although higher because of the computing costs, arrived, if at all, somewhat later than before.
Digital computers in hospitals in the United States are still used primarily for fiscal purposes, just as they were when originally deployed 20 years ago. Typically, financial data are recorded throughout the hospital on pieces of paper, aggregated in a data processing area, keypunched onto Hollerith cards or magnetic media, and fed to the computer. From these data, the computer produces bills for patients or third-party payers, payment checks for the hospital's creditors and employees, and reams of printed reports. Computers that are programmed to handle batches of financial transactions are poorly suited to interactive clinical applications, and have had little impact on the practice of medicine.
In recent years, however, investigators have demonstrated that the needs of individual hospital departments, such as the clinical laboratories,2 3 4 5 the admitting office, the medical records department,6 the pharmacy,7 and the radiology department,8 could successfully be served by computers.9 10 11 12 13 With few exceptions,13, 14 these programs were electronically isolated from one another and unable to share information, yet they pioneered the introduction of computing into the hospital environment.
In 1976, we approached the administration of Boston's Beth Israel Hospital with a plan to develop an integrated, hospital-wide clinical computing system with the following characteristics:
First, information should be captured, not on pieces of paper, but directly at computer terminals located at the point of each transaction. The computer in turn should provide immediate benefit to the person who enters this information.
Second, information captured at a terminal anywhere in the hospital should be immediately available, if needed, at any other terminal. Rather than printed reports, which become progressively out of date from the moment they are produced, terminals that provide immediate access to the most up-to-date information should be the principal means of retrieval.
Third, the response time of the computer should be rapid; for the busy physician, nurse, or medical technologist, delays that can be measured in seconds are often unacceptable.
Fourth, the computer should be reliable; in the event of a failure, the defect should be corrected within minutes, and users should never lose data.
Fifth, confidentiality should be protected; only authorized persons should have access to the data.
Sixth, the computer programs should be friendly to the user. There should be no need for user's manuals. It should be easier for the physician to obtain a laboratory result from a computer terminal than from a telephone call.
Finally, there should be a common registry for all patients. For each patient there should be one and only one set of identifying information in the computer, available at all times to authorized users and preserved, if possible, in perpetuity. Whenever an error in the common data base is detected and corrected at any terminal, that correction should be immediately available at all terminals. The common registry should be shared throughout the hospital by all programs that involve identification of a patient.
In collaboration with our colleagues throughout Beth Israel Hospital, we have developed and deployed such a hospital-wide clinical computing system. Experience gained at Beth Israel has led to our being asked to develop similar computer programs at the Brigham and Women's Hospital. In the paragraphs that follow, we describe the computer systems at Beth Israel Hospital and present data on the extent of their use and the satisfaction of their users.
Hospital and Computing Facilities
Beth Israel Hospital, a teaching hospital of the Harvard Medical School, has 452 beds and 54 bassinets. In 1984 there were 19,000 admissions and 160,000 outpatient visits. The computer programs, written in the MIIS dialect (Medical Information, Technology, Inc.) of MUMPS,15, 16 run on a network of Data General Eclipse computers that are interconnected to share 12 billion characters of disk storage. This network supports 300 video display terminals and 100 printers located throughout the hospital.
The computers are used 24 hours a day, seven days a week, except between 12 midnight and 2 a.m. twice a week, when backup copies of disk packs are made. Computer services in crucial areas, such as the emergency unit, admitting department, and medical records department, are maintained around the clock. Each computer keeps a log of all transactions on magnetic tapes; these tapes, together with the most recent copies of the disks, are available for reconstruction of data in the event of destructive malfunction. Unscheduled down time has averaged less than 12 hours per year; since 1976, when the first clinical computing program was put into use, virtually no data have been lost.
All the programs in the clinical computing system were designed, written, and implemented by our group in the Division of Computer Medicine. From 1978 to 1984 the mean annual cost for salaries, equipment, and supplies (exclusive of overhead) was $966,700 -- less than 1 per cent of the hospital's budget.
Clinical Computing System
Registration
For each of the 539,000 patients who have been cared for at Beth Israel Hospital since its medical record number system was established in 1966, the name, address, telephone number, and Social Security number and the names of parents and spouse are stored in the computers in a common registry. When a new patient arrives, whether as an inpatient or an outpatient, these data are obtained by interview and entered into a computer terminal in the patient's presence. The computer then assigns a medical record number. The computer is electronically interfaced to an automatic card maker, which produces the plastic card used to label the patient's paper record. Initial registration can occur at any of multiple stations throughout the hospital. Once in the registry, the patient's data are preserved indefinitely; thereafter, initiation of a hospital admission, outpatient visit, or any clinical procedure requires that data be located rather than reentered. The patient's identifying information can be corrected at any terminal by any authorized person who has access to the registry; the prior information, however, is preserved and can be displayed if a question of accuracy arises.
In the admitting department, terminals are used to preadmit patients, note their arrival, and track their transfer to different rooms and clinical services. Since each patient's location is known to the computer, meals, mail, medications, laboratory reports, and relatives are all directed to the patient's current location. In the outpatient departments, terminals are used to note each patient's arrival and to schedule future visits. The computer uses rules defined for each clinic to make appointments; by keeping track of all appointments, it prevents conflicts both for patients and for providers.
In the medical records department, light pens, which read bar-coded labels, track the charts as they are signed out and returned to the department. A chart can be requested from any terminal in the hospital. Each request generates a ticket, printed in the medical records department, which provides the last known location of the patient's chart; personnel from medical records no longer go to shelves to learn that a chart is somewhere else. For patients scheduled to be admitted or seen in a clinic the next day (patients whose appointments are already in the computer), requests are printed automatically the night before and appear in the order in which charts are arranged on the shelves. When a patient arrives in the emergency unit, a request for delivery of the chart is printed automatically in the medical records department.
Clinical Laboratories
The physician writes orders for laboratory examinations (and other diagnostic procedures) on the patient's chart. The orders are then transcribed onto requisitions that are affixed to the specimens. Computer terminals in the microbiology, hematology, chemistry, and blood gas laboratories communicate with the common patient registry; to register a specimen, the technologist identifies the patient in the registry and indicates which tests are to be performed. The computer then assigns an accession number, notes the date and time, and prints labels and work sheets as required.
In the microbiology laboratory, the computer prints protocols, individualized for the type of specimen, for technologists to follow as they prepare stains and cultures. Gross and microscopical descriptions of cultured organisms, together with their sensitivities to antimicrobial agents, are recorded at the terminals. In the blood gas laboratory, to eliminate inconsistencies among algorithms used by the different instruments, the computer corrects the measured values for the patient's temperature.17 In the hematology laboratory, white-cell differential counts, fluid-cell counts, and reticulocyte counts are performed directly on terminals, which beep when the appropriate number of cells has been counted. In the chemistry laboratory, the computer evaluates measurements of quality control specimens, which are included in each batch of patients' specimens, and advises the technologist whether to accept the results.18
Automated devices, such as the Coulter counters and Coag-a-mates in the hematology laboratory and the SMAC, COBAS, and ASTRAs in the chemistry laboratory, are connected electronically to microcomputers, which collect results and transmit them to the central computers. The microcomputers also store results for two days; in the event of failure of a central computer, results held locally can be forwarded later.
Results from automated devices, as well as those that are manually entered, are displayed on terminals in the laboratories, where they are checked by a technologist; once verified, they are released for viewing on terminals throughout the hospital. The most recent results for each patient are displayed next to the current ones; on occasion, a marked disparity between a recent and current result has enabled the technologist to detect a mislabeled specimen or the effect of an interfering substance.19, 20 A markedly abnormal result or any result that has changed substantially since the previous determination is automatically flagged; such results are rechecked by the technologist and called to the clinician at the telephone number provided by the computer.
Radiology
When a patient arrives in the radiology department, the name, address, identifying data, and billing data are, as a rule, already in the computer. The receptionist registers the examination at a terminal. The computer then prints a work card for the technologist, a label for the new film jacket, a transportation ticket that provides the patient's room number as maintained in the computer by the admitting office, and an identification card that is optically photographed at the time of the x-ray exposure to label the new films. Simultaneously, a printer located in the file room produces a request to retrieve the old films.
When the technologist indicates on a terminal that the examination has been completed, the computer adds the charge for taking the films to the patient's hospital bill. As films emerge from the dryer, they are merged with the old films and presented to a radiologist for interpretation. For 55 per cent of the examinations, the radiologist records the interpretation directly at a terminal with an interactive coding system developed by members of the radiology department21; for the remaining 45 per cent, the report is dictated and transcribed into a word processor on the same computer. Radiologists use terminals to edit and approve both coded and dictated reports, which are then released for viewing on terminals in the patient care areas. Upon approval of a report, the computer adds the charge for interpretation of the examination to the patient's bill.
Surgical Pathology
In the surgical pathology department, each specimen is registered in the computer; if the patient has had specimens analyzed previously, the program provides their identification numbers so that the slides and reports can be found. Gross and microscopical descriptions for the new specimen are transcribed into a word processor in the computer. The pathologist then reviews and edits this report at any available terminal, appends SNOMED22 diagnoses for use in future retrieval within the department, and releases the report for viewing on terminals throughout the hospital.
Printed Reports
Each evening, the computer prints clinical reports for the floors, clinics, and medical records department. For each patient, reports from the blood gas, chemistry, hematology, microbiology, urinalysis, and electrocardiography laboratories and the radiology, cytology, and surgical pathology departments are collated by the computer and printed as a single unit. Reports for the floors are sorted by room number, those for clinics by the physician's name, and those for the medical records department by the terminal digit of the patient's unit number.
Pharmacy
The computer maintains the hospital formulary. When a prescription arrives in the outpatient pharmacy, the pharmacist uses a computer terminal to identify the patient, display the medication profile, and record the new medication and its dosage. The computer then prints the label for the bottle and prices the prescription. The computer is electronically interfaced to an automatic pill counter with cells for the 50 most commonly prescribed tablets and capsules; for these medications, signals from the computer dispense the desired number of pills directly into the bottle. The computer assists with inventory by permitting surveillance of available medications, both by manufacturer and by functional class.
In the inpatient pharmacy, orders for intravenous drugs are entered into the computer, which then organizes the preparation and distribution of the medications.23
Financial Operations
Although it is possible to perform a service at Beth Israel Hospital without first registering the transaction in the computer, in most instances such registration is necessary if the patient for whom that service is performed is to be identified. Manual registration and identification procedures that were once used throughout the hospital have largely disappeared. After the service has been performed and its results have been recorded in the computer, the charge for that service is automatically posted to the patient's file within the computer. In this way, approximately 90 per cent of each patient's charges are collected as a byproduct of the clinical computing system. Each day, this information is transferred on magnetic tape to the hospital's fiscal computer (in a separate facility not within our network), which prints the bills and performs other financial operations.
Access and Confidentiality
Each authorized user gains access to the clinical computing system by means of a unique, computer-assigned, confidential code, or key. The computer maintains a dictionary of keys, the names of their owners, and the specific computer programs for which they are authorized. Each key permits access only to programs appropriate to its owner, and then only from terminals appropriate for those programs. Thus, a key assigned to an admitting officer permits access only to programs used by the admitting department, and then only from terminals in that department. Providers' keys, which are used to retrieve clinical information, laboratory results, and radiologic interpretations, can function from any terminal in the hospital.
If someone tries to use an illegal key or a legal key in an inappropriate location, after a small but random number of tries the terminal springs a "keytrap": it beeps the Morse code for "SOS" 18 times and renders the terminal unusable until our laboratory is called. It is a principle in our hospital that patients' data are to be respected, that keys are to be kept confidential.
Some keys permit only looking at data, some permit data to be entered and edited, and some permit new keys to be issued. After a user has completed a transaction at the computer, such as scheduling an appointment, assigning a room, or correcting a patient's identifying information in the registry, the key must be entered as a personal signature. The program then records the date, the time, the name of the person who performed the transaction, the program used, and the location of the terminal. Whenever any provider looks up any patient's clinical data, that information is recorded and is, upon request, made available to the patient or to the patient's physician.
Patient Care
Clinical data are retrieved with the PATIENT CARE option. To use this option, the provider must identify the patient, which can be done with a name, a medical record number, a fiscal number, a room number, or a nursing station. When a patient's name is typed in the form SMITH, JOHN, the program searches the registry of 539,000 patients and responds with the names and identifying data of all patients with that name, followed by the names and identifying data of all patients whose names have a similar spelling or a similar sound. If the list is too long for the screen, additional identifying information, such as sex and age, is requested. To restrict a search to hospitalized patients, the user can type the last name alone.
Once the patient has been identified, the computer displays a list of options (Fig. 1), which are available to all authorized providers for use in the care of their patients. If the provider chooses the option MOST RECENT RESULTS the computer displays all results that are available for inpatients for the previous two days or all that are available for ambulatory patients since the most recent admission or clinic visit.
Data from the blood gas, chemistry, hematology, and microbiology laboratories are available as soon as the tests have been performed. Results of tests performed on the day of the search, on any earlier day, or during any range of dates can be requested. Results are displayed in reverse chronological order, with the most recent ones first (Fig. 2). If the user selects the DEMOGRAPHICS & VISIT HISTORY option the computer displays the dates of service, the initial problem, the attending physician, and the final diagnosis for outpatient visits, emergency unit visits, and hospital admissions. Choosing the PHARMACY option results in a display of prescriptions that have been filled by the hospital pharmacy, and choosing RADIOLOGY or SURGICAL PATHOLOGY results in a display of examinations performed and their diagnostic interpretations.
The history of outpatient visits and admissions is available from the terminals for 15 months; laboratory results, radiology and surgical pathology reports, and outpatient pharmacy prescriptions are available for a minimum of six months. After erasure from the active disk files, this information is maintained on magnetic tape for use in retrospective analysis if the need arises.
Utility Programs
A set of utility programs, which can be requested from any terminal in the hospital, is available to all authorized users of the clinical computing system. A KEY option permits supervisors to issue keys to employees and allows users to change their own keys in case of a suspected breach in security. An option labeled HOW TO USE THE COMPUTER TERMINAL offers instructions for the beginner and reminders for the more experienced user. A TELEPHONE option provides the telephone number, beeper number, and room number of each member of the staff and of each department. This information is continually updated, and periodically the computer prints a listing, which is then reproduced as the hospital's telephone book. PaperChase,24 a computer-based bibliographic retrieval program, permits users to search the medical literature at any time of the day or night. For authorized providers, consultation programs offer advice on the management of electrolyte and acid-base disorders25 and on the use of anti-biotics and new medications in the hospital's formulary. In addition, a RECORD REQUEST option permits them to request delivery of a patient's chart.
An ELECTRONIC MAIL option permits a person to send a message either to an individual or a group of people. A laboratory supervisor can send a request to an individual technologist or broadcast a description of a new procedure to all technologists; a chief resident can broadcast a message to the house staff that an instructive, unscheduled conference is about to occur; and the director of the blood bank can broadcast a message to everyone who signs on to the computer that there is an urgent need for a particular type of blood. Mail can be sent immediately, posted for future delivery, or retracted if not yet read and the writer has had second thoughts. Users can also ask whether a message has been read, view messages previously sent, and read their old mail. At the time of sign on, each user is informed about unread mail and offered the opportunity to read it then or later. Mail that is received can be held in an arbitrary number of queues for future use, or it can be edited and forwarded to someone else.
In some cases the computer itself generates messages. When a patient is admitted to the emergency unit or to the hospital, the patient is automatically placed on a list that belongs to the attending physician. A physician who sees a patient in consultation can add that patient to his or her list. Thereafter, when the patient returns to the hospital, a message to that effect is automatically sent to all physicians on whose list that patient appears. Physicians have told us that they very much appreciate this feature; it informs them whenever one of their patients has returned to the hospital, even if under the care of a colleague, and it eliminates the need to call the emergency unit repeatedly to find out whether an expected patient has arrived.
Evaluation
The randomly assigned key that identifies each user provides an electronic signature for each transaction with the computer. We have used these signatures to determine the frequency with which programs are used by members of the hospital staff. In addition, we have surveyed users for their opinion of the programs by means of a computer-administered questionnaire, developed in Converse (a program for use in computer-based interviews)26 and incorporated into the signon procedure. In evaluating the use of the programs, we have differentiated between obligatory users, such as departmental and laboratory personnel whose jobs require interaction with the computer (the inpatient admitting officer, the medical record administrator, and the laboratory technologist), and voluntary users, who could if they so desired rely on printed reports or telephone calls for their clinical information (the physician, the nurse, and the medical student).
Obligatory Users
During the week of April 6 to 12, 1984, 477 departmental and laboratory personnel gained access to the common registry of 539,000 patients 46,579 times, an average of 6654 times per day (Table 1). Since identification of a patient in the common registry is typically followed by additional interactions with the computer, access to the census is a conservative measure of the use of the computer in these departments and laboratories.
In a study conducted between November 1982 and January 1983, departmental and laboratory personnel were asked for their opinion concerning use of the computer -- whether it was helpful or harmful to them in their work. Each user who signed on was offered an opportunity to answer questions immediately, to defer the questions until a later date, or to avoid them altogether. Those who deferred were again asked to participate after a delay of at least 24 hours. Those who refused were asked to reconsider after a delay of at least one week. Those who refused twice were not asked again. Of the 677 persons asked, 545 (80 per cent) agreed, sooner or later, to participate. As shown in Table 2, the majority of them indicated that use of the computer terminals enabled them to work faster and made their work easier, more accurate, and more interesting.
Voluntary Users
During the week of April 6 through 12, 1984, 818 staff physicians, house officers, clinical fellows, nurses, medical students, and health assistants used the PATIENT CARE option 16,768 times (Table 3). They looked up data on hospitalized patients 12,688 times and data on outpatients 4080 times. During that week, 834 patients were in the hospital; providers used computer terminals to look up data concerning 684 of them. The daily census averaged 455 patients; data on each hospitalized patient were looked up an average of four times per day. On the medical service, there were six inquiries per patient per day, whereas on the obstetrical service, there was one.
Among the 28 house officers and students assigned to work on the inpatient medical service during the week of the study, each intern looked up data an average of 20 times per day (range, 6 to 29), each resident 6 times per day (range, 2 to 10), and each medical student 7 times per day (range, 2 to 11). Among the 14 house officers assigned to work on the inpatient surgical service, each intern looked up data an average of 11 times per day (range, 3 to 23), and each resident 8 times per day (range, 3 to 13).
Inquiries during the week about hospitalized patients were asymmetrically distributed among the laboratories and departments. Chemistry results were looked up 2037 times, hematology results 1710 times, microbiology results 433 times, and blood gas results 270 times. Radiology reports were looked up 672 times, surgical pathology reports 128 times, demographic and visit-history 434 times, and outpatient pharmacy profiles 32 times. In addition, MOST RECENT RESULTS (an option that provides, in a single selection, recent results from the chemistry, hematology, microbiology, and blood gas laboratories, as well as reports from radiology and surgical pathology) was used 6972 times. Since we counted each use of this option as a single inquiry, the number of inquiries reported probably understates the number of telephone calls that would have been required to obtain the same data.
Of the 16,768 inquiries, 11,850 were made between 8 a.m. and 6 p.m., the remaining 4918 during other hours. The mean number of inquiries per weekday, 2672, exceeded that per weekend day, 1704.
In a study conducted between May 1982 and January 1983, 94 staff physicians, 152 house officers, 19 medical students, and 487 nurses, each of whom had used the PATIENT CARE option at least 10 times, were asked to give their opinion of the computer and the degree to which the computer terminals and printed reports were helpful to them in looking up laboratory information. Once again, at the time of signing on, each user was offered the opportunity to answer the questions immediately, to defer them until a later date, or to avoid them altogether. Of those who were asked, 76 staff physicians (81 per cent), 120 house officers (79 per cent), 17 medical students (90 per cent), and 373 nurses (77 per cent) agreed to participate and completed the questionnaire. Of these, 470 (80 per cent) indicated that they used the computer terminals "most of the time" to look up laboratory results (Table 4), and 524 (89 per cent) found the computer terminals "very helpful" in doing so (Table 5). Staff physicians, house officers, medical students, and nurses gave similar estimates for the helpfulness of the computer. On the other hand, when asked whether they would use computer terminals or printed reports more often, on the assumption that the printed reports would always be readily available, staff physicians were almost equally divided, whereas 78 house officers (65 per cent), 13 medical students (76 per cent), and 265 nurses (71 per cent) indicated that they would continue to use the terminals more often.
At the conclusion of the questions, users were asked to comment, if they wished, on their reaction to the computer; 133 providers, 23 per cent of those who agreed to participate in the survey, took the time to type comments. "The computer is one of the greatest assets of Beth Israel Hospital," wrote one enthusiastic house officer. "I just find it more convenient (and fun!) to look up the lab results I need using the computer terminal," typed a nurse from one of the medical units. "When I was an intern here, we didn't have the computer system like this. If we had, I estimate that I would have saved myself an hour a day at least," commented a medical resident. "A computer [terminal] should be installed in the dialysis unit," typed a staff nephrologist (we had one installed as quickly as possible).
Discussion
A distinguishing feature of the clinical computing system at Beth Israel Hospital is the magnitude of its use. During a one-week study period, from terminals located throughout the hospital, 477 departmental and laboratory personnel gained access to the common registry of patients 46,579 times. During the same week, 818 patient care providers used this same registry to look up clinical and laboratory data an additional 16,768 times. That same week, physicians, nurses, medical students, and hospital personnel used the computer to perform 1423 bibliographic searches of the biomedical literature and to deliver 9185 pieces of electronic mail.
The magnitude of use stems in part from the dependability of the computer system. As users learned that the programs were reliable and that the computers were almost never down, they came to rely increasingly on terminals for help in their work, requesting new features and suggesting additional programs. When the laboratory programs were first implemented (before terminals were deployed on nursing units), interns and residents quickly discovered that results were available on terminals in the laboratories as soon as the tests had been performed. They descended from their floors and confiscated terminals, sometimes to the consternation of the technologists. In response to the demands of the house staff, terminals were deployed on the floors and in the clinics. Throughout the hospital, the number of programs and terminals increased, and new users were attracted; eventually, virtually every employee who could be helped by the computer had a key and used it regularly.
The common registry of patients, which is available to all programs and shared by the entire hospital, provides the foundation for the clinical computing system. All data related to patients are deposited into this registry. Medical record numbers and other identifying information are maintained indefinitely; the remaining information is maintained for a minimum of six months. From any terminal in the hospital, the physician can look up data from the clinical laboratories (microbiology, blood gas, chemistry, and hematology); obtain reports from the departments of radiology and pathology; obtain demographic data and visit histories; look up prescriptions filled in the outpatient pharmacy; perform bibliographic retrieval of the MEDLINE data base; read, write, edit, retract, and hold electronic mail; request the delivery of a patient's chart; and find out whether a patient of interest has arrived in the emergency unit or has been admitted to the hospital. To perform many functions, it is necessary to identify the patient only once. If the information is not found, there is virtually no need to check another file in the same or another computer, to consult a card file, to look at microfilm, or to call another department.
Information about a patient that is entered at any terminal is immediately available (if appropriate) at every other. A patient who registers in the emergency unit need never provide the same identifying information again; the information is available when needed in other areas, such as the admitting department, the radiology department, and the clinical laboratories. At any terminal in their laboratories, authorized technologists can look up the results of any patient's tests, whether performed in their own or in another laboratory. Thus, in the microbiology laboratory, a supervisor can retrieve information about antibiotics from the pharmacy data, white-cell counts from the hematology data, and spinal fluid values from the chemistry data. In addition, all computer systems are integrated: they all "talk to one another." The admitting department is automatically notified when a patient to be admitted has had a chest film within the past six months; duplication of routine chest films is thereby avoided.
Computer systems at Beth Israel Hospital are easy to use. New users are given preliminary instructions and are asked to type a question mark when they need further help. No courses have been given to teach people how to use the computer terminals, and no users' manuals have been written. Insofar as possible, the programs have been developed to suit the specific needs of the individual user while providing all users with the advantage of shared information in an integrated system. Once formats on terminals and printed forms are developed, they can be edited without compromising the internal organization of the programs. Accordingly, our programmers work directly with prospective users; they create prototypical formats on the computer, which can then be tailored into useful products. The one constant response to the programs has been the continual request for change; even the most farsighted users change their minds and think of additional useful features as they become increasingly familiar with the potential of the computer. In our experience, detailed plans prepared in advance are likely to be obsolete by the time the programs are implemented.
Years ago, Watson, Thorndike, and Skinner taught that behavior is shaped by its consequences.1, 27, 28 If the consequences are reinforcing, the behavior is strengthened; if not, the behavior tends to disappear. Behavior that pays off becomes part of our repertoire. The answer to the question, "Will physicians interact with computers?" is yes, if the interaction is helpful. When computer manufacturers ask, "How can we get physicians to use computers?" they are more likely to mean, "How can we get physicians to buy computers?" A better question would be, "How can we make the computers helpful?" If the computer is not helpful, if nothing worthwhile happens when the keys are pressed, the terminals and the computer will gather dust. If, on the other hand, the computer is programmed to help the physician, nurse, medical student, and laboratory technologist, our data indicate that it will be heavily used.
Supported in part by grants (HS 04050) from the National Center for Health Services Research and (LM 03393) from the National Library of Medicine, U.S. Public Health Service.
We are indebted to Dr. Mitchell T. Rabkin, president of Beth Israel Hospital, and Mr. David Dolins, executive vice-president of Beth Israel Hospital, whose continual support and encouragement have been essential to our work; to Ms. Karen DeMartino, Ms. Mary McNulty, and Ms. Mary Jean Moore for their editorial assistance; to Dr. Douglas Porter for his help with the evaluation studies; to Drs. Anthony L. Komaroff and Charles Safran for their suggestions for improvement of the manuscript; and to our many colleagues in the Division of Computer Medicine for their assistance with this project.
1. Skinner BF. The behavior of organisms:an experimental analysis. New York:Appleton-Century-Crofts, 1938.
2. Hicks GP, Gieschen MM, Slack WV, Larson FC. Routine use of a small digital computer in the clinical laboratory. JAMA 1966;196:973 -8.
3. Collen MF. The multitest laboratory in health care of the future. Hospitals 1967;41 (9 ):119 -25.
4. Lindberg DA. Collection, evaluation and transmission of hospital laboratory data. Method Inf Med 1967;6:97 -107.
5. Kunz LJ, Poitras JW, Kissling J, et al. The role of the computer in microbiology. In:Prier JE, Bartola J, Friedman H, eds. Modern methods in medical microbiology:systems and trends. Baltimore:University Park Press, 1974:181-93.
6. Weed LL. Medical records that guide and teach. N Engl J Med 1968;278:593 -600, 652-7.
7. Gouveia WA, Diamantis C, Bamett GO. Computer applications in the hospital medication system. Am J Hosp Pharm 1969;26:140 -50.
8. Lodwick GS, Turner AH, Lusted LB, Templeton AW. Computer-aided analysis of radiographic images. J Chronic Dis 1966;19:485 -96.
9. Warner HR. The role of computers in medical research. JAMA 1966;196:944 -9.
10. Barnett GO. Computers in patient care. N Engl J Med 1968;279:1321 -7.
11. Lindberg DAB. The computer and medical care. Springfield, Ill.:Charles C Thomas, 1968.
12. Garfield SR, Collen MF, Feldman R, Soghikian K, Richart RH, Duncan JH. Evaluation of an ambulatory medical-care delivery system. N Engl J Med 1976;294:426 -31.
13. Warner HR, Olmsted CM, Rutherford BD. HELP -- a program for medical decision-making. Comput Biomed Res 1972;5:65 -74.
14. Simborg DW, Chadwick M, Whiting-OKeefe QE, Tolchin SG, Kahn SA, Bergan ES. Local area networks and the hospital. Comput Biomed Res 1983;16:247 -59.
15. Greenes RA, Pappalardo AN, Marble CW, Barnett GO. Design and implementation of a clinical data management system. Comput Biomed Res 1969;2:469 -85.
16. Bowie J, Barnett GO. MUMPS -- an economical and efficient time-sharing system for information management. Comput Programs Biomed 1976;6:11 -22.
17. Andritsch RF, Muravchick S, Gold MI. Temperature correction of arterial blood-gas parameters:a comparative review of methodology. Anesthesiology 1981;55:311 -6.
18. Westgard JO, Barry PL, Hunt MR, Groth T. A multi-rule Shewhart chart for quality control in clinical chemistry. Clin Chem 1981;27:493 -501.
19. Strauchen JA, Alston W, Anderson J, Gustafson Z, Fajardo LF. Inaccuracy in automated measurement of hematocrit and corpuscular indices in the presence of severe hyperglycemia. Blood 1981;57:1065 -7.
20. Saah AJ, Koch TR, Drusano GL. Cefoxitin falsely elevates creatinine levels. JAMA 1982;247:205 -6.
21. Leeming BW, Simon M, Jackson JD, Horowitz GL, Bleich HL. Advances in radiologic reporting with computerized language information processing (CLIP). Radiology 1979;133:349 -53.
22. Cote RA, ed. SNOMED:systematized nomenclature of medicine. 2nd ed. Vol. 1. Numeric index. Skokie, Ill.:College of American Pathologists, 1979.
23. Pope RA, Mattson CJ, Janousek J, Slack WV. A computer-based IV admixture system. Methods Inf Med 1982;2165-9.
24. Horowitz GL, Bleich HL. PaperChase:a computer program to search the medical literature. N Engl J Med 1981;305:924 -30.
25. Bleich HL. The computer as a consultant. N Engl J Med 1971;284:141 -7.
26. Bloom SM, White RJ, Beckley RF, Slack WV. Converse:a means to write, edit, administer, and summarize computer-based dialogue. Comput Biomed Res 1978;11:167 -75.
27. Watson JB. Behaviorism. New York:WW Norton, 1924.
28. Thorndike EL. The fundamentals of learning. New York:Teachers College, Columbia University, 1932.
Figure 1:
Printout of Options Available on the Video Display Terminal for Physicians, Nurses, and Medical Students to Use in the Care of their Patients.
The CHEMISTRY option has been chosen.
[Image Omitted: See PDF]
Figure 2:
Printout of Results from the Chemistry Laboratory, Displayed in Reverse Chronological Order (Most Recent Values First), for a 33-Year-Old Woman with Diabetic Ketoacidosis and Severe Hyperglycemia.
As soon as specimens are registered in the laboratory, the pending results are listed as PND. All abnormal values are indicated by asterisks (*). The pound sign (#) indicates that the creatinine value, though no longer abnormal, has changed by an unusually large amount (delta check). The footnote "C" designates critical values, those whose urgency warrants the physician's attention when obtained.
Data can be displayed for the present day, for any earlier date, or for any range of dates. Since results are available throughout the hospital as soon as they are available in the laboratories, telephone calls to the laboratories have been dramatically reduced.
[Image Omitted: See PDF]
Table 1:
Use of Patient Census by Departmental and Laboratory Personnel from April 6 through April 12, 1984.
[Image Omitted: See PDF]
Table 2:
Effect of Computer Terminals on Work as Evaluated by Departmental and Laboratory Personnel.
[Image Omitted: See PDF]
Table 3:
Use of Patient Care Options by Providers from April 6 through April 12, 1984.
[Image Omitted: See PDF]
Table 4:
Frequency of Use of Computer Terminals and Printed Reports as Estimated by Staff Physicians, House Staff, Medical Students, and Nurses.
[Image Omitted: See PDF]
Table 5:
Helpfulness in Looking up Laboratory Information as Evaluated by Staff Physicians, House Staff, Medical Students, and Nurses.
[Image Omitted: See PDF]
From the Charles A. Dana Research Institute and the Harvard-Thorndike Laboratory, Beth Israel Hospital, the Division of Computer Medicine, Department of Medicine, Harvard Medical School at the Beth Israel and Brigham and Women's Hospitals, and the Department of Pathology, Harvard Medical School at the Beth Israel Hospital. Address reprint requests to Dr. Slack at the Beth Israel Hospital, 330 Brookline Ave., Boston, MA 02215.
Copyright Massachusetts Medical Society Mar 21, 1985