John F. (John Fralick) Rockart.

An automated medical history system; experience of the Lahey Clinic Foundation with computer-processed medical histories online

. (page 1 of 3)
Online LibraryJohn F. (John Fralick) RockartAn automated medical history system; experience of the Lahey Clinic Foundation with computer-processed medical histories → online text (page 1 of 3)
Font size
QR-code for this ebook


m




LIBRARY

OF THE

MASSACHUSETTS INSTITUTE
OF TECHNOLOGY



WORKING PAPER
ALFRED P. SLOAN SCHOOL OF MANAGEMENT



AN AUTOMATED MEDICAL HISTORY SYSTEM:
EXPERIENCE OF THE LAHEY CLINIC FOUNDATION
WITH COMPUTER-PROCESSED MEDICAL HISTORIES*



John F . ^oqkar t , Ph . D .
Ephraim R. McLeaW, Ph.D.
Philip I. Hershberg, M.D.



March 1972



/



592-72



MASSACHUSETTS
INSTITUTE OF TECHNOLOGY
[ 50 MEMORIAL DRIVE

CAMBRIDGE, MASSACHUSETTS 02139



MASS. \KS1. TECH.



L



APR 12 1S7'
DEWEY LIBRARY



AN AUTOMATED MEDICAL HISTORY SYSTEM:
EXPERIENCE OF THE LAHEY CLINIC FOUNDATION
WITH COMPUTER-PROCESSED MEDICAL HISTORIES*

John F._Roqkart, Ph.D.

Ephraim R. HcLearl, Ph.D.

Philip I. Hershberg, M.D./

March 1972 592-72



This work was supported in part by the Lahey Clinic Foundation; by
PHS Research Grant No. 3 R18 HS00307-01 from the National Center of Health
Service Research and Development, HSMHA; by NASA Office of Occupational
Medicine, Grant NGR 22-007-203; and by the Accounting and Information Sys-
tems Research Program, UCLA.

This project was under the medical supervision of George 0. Bell,
M.D., former Chairman of the Department of Internal Medicine of the Lahey
Clinic Foundation. Assisting were members of the Lahey Clinic Systems and
Data Processing Department, particularly J. Chaney, R. Harrison, D. Katzer,
and M. Mulloy.



John F. Rockart, Ph.D., is Associate Professor of Management,
Alfred P. Sloan School of Management, Massachusetts Institute
of Technology, Cambridge, Massachusetts, and Research Associate,
Lahey Clinic Foundation, Boston, Massachusetts.



Ephraim R. McLean, Ph.D., is Assistant Professor of Information
Systems, Graduate School of Management, University of California,
Los Angeles, California.



Philip I. Hershberg, M.D. , is Assistant Professor of Medicine,
School of Public Health, Harvard University, Boston, Massachusetts,
and Research Associate, Lahey Clinic Foundation, Boston, Massachu-
setts.



634714



Lahey Clinic ATIIIS , Rockart et al,



ABSTRACT



SincG the late 1940 's, efforts have been made to assist the
physician in the time-consuming task of collecting medical history
information. More recently, computer technology has been em-
ployed in the design of automated medical history systems. Re-
searchers at a number of different locations have tried several
approaches including the use of prepunched cards, keypunched or
mark-sensed forms, and on-line terminal-based interview systems.

At the Lahey Clinic, an automated medical history system
(AMIIS) has been in operation since 1968 and as of the present time
has been administered to over 30,000 patients. The system v/as
designed to aid in patient scheduling, to assist the physician
in his history taking, and to provide a data base for research
studies .

The present Questionnaire is a 25-page booklet which is
mailed to the patient at his home in advance of his appointment.
The completed Questionnaire is .iiailed back to the Clinic where
the booklet pages are optically scanned and processed by the
computer. The resultant print-out is placed in the patient's
medical record folder where it serves as the primary history docu-
ment.

Although the Lahey system shares much in common with other
computer-based history systems, it also has a number of distinc-
tive features. These can briefly be summarized as follows:

1. The original development and operation of the AMHS
was funded from the operating revenues of the Clinic.



Lahey Clinic AIIHS , Rockart et al. 2

2. It is a routine, integral part of the Clinic procedure
and is administered to almost all nov; patients who come to the
Clinic.

3. Because of the desire to use the results for advanced
scheduling purposes, the Questionnaires are filled out by the
patients at their homes.

4. The success of the system is due in large measure to
the use of an interdisciplinary approach joining the skills of
the physician, management scientist, and computer specialist.

5. By design, the M\llS is an evolutionary system. The
present Questionnaire is the fourth version, emd plans are
already underway for the fifth version.

6. Accompanying these developmental activities is a strong
commitment to continuing research and evaluation. Some of the
studies v;hich have already been cpnducted have dealt with physicians'
attitudes, patients' attitudes, and content validity.



AN AUTOMATED MEDICAL HISTORY SYSTEM
Experience of the Lahey Clinic Foundation VJith
Computer-Processed Medical Histories

John F. Rock art, Ph.D.
Ephraim R. McLean, Ph.D.
Philip I. Hershberg, M.D.

Of the many competing demands upon a physician's time, the
taking of a patient's medical history is one of the most time
consuming. Although it does provide an opportunity for the phy-
sician to establish good rapport with the patient, many physi-
cians believe that it is not an altogether satisfactory means
of gathering information for a number of reasons.

For one, there is never enough time. The physician's busy
schedule does not allow him the luxury of exploring every aspect
of the patient's physical and mental condition. In every patient
interview a number of questions might be asked that would provide a
more complete history, but there is simply not enough time for
them. Fortunately, by training and by instinct, the experienced
clinician is able to move quickly to the important facts of the
patient's condition and rarely does any harm result from the
few items of dat^ that are missed. However, from the patient's
standpoint, this inability of the physician to listen to all of
his minor complaints and problems is somewhat disconcerting.

A second problem is the patient himself. Frequently he is



Lahey Clinic W4IIS , Rockart et al . 2

poorly prepared to be an effective historian. Family history is
only vaguely remembered; drugs and medications are imperfectly
recalled; and even for such simple questions as "How much coffee
do you drink?" a few moments of mental calculations are necessary.
Also, there is the nature of the questions themselves. Ques-
tions of a highly personal or intimate nature can be embarrassing
for the physician and patient alike. There is some evidence that
valuable information is not elicited because of the physician's
failure to ask certain questions or because of the patient's
failure to respond frankly. Despite the physician's best efforts,
the physician encounter can be a very stressful one for the pa-
tient, and this nervousness can be a barrier to good communica-

1
tion .

It is not the intention of this article to suggest that
there are easy solutions to these problems. However, techniques
have been developed that offer great promise in assisting in this
information-gathering process. Chief among these is the use
of medical history questionnaires.

At first such questionnaires were merely a short list of
printed questions with a space for the patient to mark his
response. The physician would scan these responses and then
proceed v/ith his own questioning. With the advent of the com-
puter, more sophisticated approaches have become possible. The
responses can be fed into the machine and be edited, summarized,
and printed out in an easy-to-read format. It is even possible
to use a computer terminal to interrogate the patient directly



Lahey Clinic MMIS , Rockart et al . 3

and then produce tlio resultant information on demand. A number
of individuals have performed research on use of computer-pro-
cessed medical histories. Although most of these efforts have
been of an experimental nature, the results have been encouraging.

At the Lahey Clinic Foundation, a program for the computer
processing of medical histories has been in operation since early
1968 and, while under continuing evaluation, is now an established
part of the Clinic routine. As of the fall of 1971, over 30,000
patient-completed Questionnaires have been processed. This "pro-
duction" aspect of the Clinic's automated medical history system - as
opposed to a one-time test program - makes it a particularly in-
teresting vehicle for study. Also, the Clinic's commitment to
research and continually improving patient care has provided a
receptive climate for an on-going appraisal of the system.

The Questionnaire itself has gone through three major re-
visions and the system has been the subject of several in-depth
studies, including three on physician acceptance and use and two
on patient attitudes. Although continuing improvements are anti-
cipated, the system is now in a sufficiently mature state that it
is possible to report on the results of the Clinic's experience.
Also, distinctive aspects of the Lahey system, aside from its
acceptance as a routine part of the Clinic procedure, merit
attention. A discussion of these features, as well as a report
on the results of the Clinic's experimental findings, v\?ill
comprise the major part of the remainder of this paper.



Lahey Clinic AIHIS , Rockart et al .



Previous Wor k in the F ield

The first questionnaire to come into general use was the

2
Cornell Medical Index (C.M.I.) . Devised by Brodman in the late

1940 's, it consists of a form containing 195 questions which is
given to the patient immediately before the office visit. Fur-
ther processing is not required, although some efforts have been

3
made to introduce a computer-processed version. The excimining

physician quickly scans the patient's responses cind then proceeds

with his own questioning. Since its introduction, the C.M.I.

has undergone almost no changes in either its composition or in

question v/ording. It continues to enjov widespread popularity;

4
it is estimated that more than 300,000 are administered annually.

At special multiphasic health check-up laboratories within

the Kaiser-Permanente Medical Centers in San Francisco and

Oakland, California, Collen has used a medical history question-

5,6
naire for more than 20 years. In its present form, it consists

of two parts: a deck of 204 prepunched cards designed for re-
view of systems information with a single yes-or-no question
printed on each card, and a pencil-and-paper questionnaire for
past history. For the first part, the patient indicates his
response by dropping each card into the "yes" or "no" section of
a divided letter box. The positive responses are then sorted
and are available immediately for the physician's review. The
responses on the questionnaire form are keypunched and are added
to the patient's medical record later. Although not yet an
integral part of the Kaiser Plan throughout California, this



Lahoy Clinic AJIHS , Rockart ot al. 5

multiphasic screening program, including the computer-processed

medical history information, is recognized as one of the leaders

in the field.

In a pioneering effort at the University of Wisconsin,

Slack and his co-workers developed the first on-line computer-

7
based medical history system. Using a LINC laboratory computer,

questions were presented to patients by means of a cathode ray
tube screen. Patients keyed in their responses through a type-
writer-like keyboard; and, at the end of the session, the results
were summarized and printed out for the examining physician. The
use of an on-line computer provided the ability to have extensive
branching. The response to one question would determine, to a
limited extent, the next question that was to be asked. In Slack's
system, there wore more than 50 3 questions, but the maximum number
that could be asl:ed of any one patient was 320. Although Slack
has recently joined the staff of Beth Israel Hospital in Boston,
this experimental v/or]; is still continuing at Wisconsin.

At the Mayo Clinic, /layne ot al . have explored a number of

1
approaches to automated history taking. The first system was,

in many regards, the most sophisticated from a technical stand-
point. Both photographic and cathode ray tube screens were used;
the former to display color pictures of various parts of the body
and the latter to present questions. The patient responded by
means of a computer light-pen. By touching the CRT screen with
the pen, the patient's answers were recorded. Throughout this



Lahey Clinic AMHS , Rockart ot al . 6

session at the computer terminal (slightly over an hour per
patient) , a full-time attendant was available to provide assis-
tance. Although quite advanced technically, this system was also
extremely expensive and only about four patients per day could
be processed; it has since been discontinued.

A more recent development at Mayo has been the use of a three

8,9
level pencil-and-paper questionnaire. Recognizing both the

benefits of multiple-level question branching (as was possible
in the former on-line version) and the greater economies of com-
puter batch processing, the system v/as designed to use the com-
puter to tailor-make a more detailed second and, if necessary,
third level questionnaire based upon the patient's responses to
the preceding level. The first questionnaire has been tested on
more than 3,000 patients and has been well received.

In evaluating various techniques of questionnaire administra-
tion, the Mayo group tried three approaches for presenting the
questions: a deck of prepunched cards (as is used by Collen) ,
a questionnaire that v/as manually keypunched, and another ques-
tionnaire that was designed to be optically scanned. They found
that the latter method was the most satisfactory from an overall
standpoint .

At the Massachusetts General Hospital, Grossman et al . have

10,11
explored the use of on-line history taking. Their work was done

with the use of several Teletype terminals connected on-line to



Lahey Clinic AI'IIS , Rockart et al. 7

a central computer. In the two years of operation at the Medical
Diaqnostic Center of the Massachusetts General Hospital, nearly
600 patients were processed, most of them on an outpatient basis.
As with other on-line systems, the use of question branching causes
the number of questions asked of a given patient to vary, in this
case from as few as 9 to as many as 107. Patient acceptance was
favorable .
Rationale _for^ t.he_ Lahey Automated Medical Histor y System

The Lahey Clinic Automated Medical History System (AMHS)
differs from its computer-based predecessors in that the T^H
Questionnaire was designed to be mailed to patients in advance
of their Clinic visit. This wa:3 done so that the results of
the Questionnaire could be used by the Clinic in the scheduling
of patients to the appropriate specialists, and so that the AMH
print-out could be available at the time of the patient's appoint-
ment to aid the physician in his history taking. These two rea-
sons formed the primary justification for the development of tlie
AMHS. A third reason was to establish a data base for research
purposes .
Pat_ient Scheduling

The Lahey Clinic is comprised of approximately 100 physicians,
each practicing in 1 of 25 medical specialties. In its desire
to keep patient v;aiting time to a minimum and to eliminate as
much as is practicable the need for multiple Clinic visits, the
Lahey Clinic makes every effort to schedule both the primary and
secondary appointments in advance of the patient's arrival. This
approach is in contrast to other large clinics where no special-



Lahey Clinic AMIIS , Rockart et al . i

ist appointments are made until after the patient arrives at

the clinic and has been seen l^y the primary physician.

In order for the Lahey appointment office to make these

appointments in advance, the appointment secretaries must try

to obtain some information from the patients as to the nature of

their complaints either from the contents of their letters or

from short telephone conversations. Despite the difficulty of

this task and the fact that the secretaries have no special

medical training, their years of experience, coupled with v/ritten

guidelines , have enabled this advanced scheduling system to be

reasonably successful. However, cancellations and "work-ins"

are inevitable and the Clinic has long searched for ways whereby

the procedure could be improved. The AJIH Questionnaire provided

the potential for such an improvement. It v;as felt that if the

results of the computer-processed Questionnaires could be made

available to the appointment secretaries in a convenient form,

they would have much better information upon which to base their

decision as to the most appropriate physician for the patient to

see. As of this time, the use of the A^IHS for patient scheduling

12
is still on a trial basis, but the results are encouraging.

Aid jto_ the Physi cian

Although the preceding benefit, that of scheduling assis-
tance, was the first reason advanced in support of the AMIIS ' s
development, the second benefit, that of providing aid to the
physician in his own history taking, has been given increasing
attention by the Clinic. The development of the system was
particularly timely from the physicians ' point of view because



Lahey Clinic 7V!1IIS , Rockart ct al .

at about the same time it was being introduced, the number of
residents available to assist in work-ups was being substantially
reduced. Thus many physicians were more willing to give the
system a try than might otherwise have been the case.

As might be expected, the physicians have used the Question-
naire in different ways and have thus derived differing benefits
from its presence in the medical record. Some have realized a
distinct time saving, while others have used the same amount of
time as formerly but have been able to make a more complete eval-
uation of the patient. A few, of course, feel that the AMHS
has not been helpful to them at all. In a recent survey of
the physicians' attitudes toward the AMIIS , the following benefits
were each noted by a large number of physicians with regard to
the system:

1. Fewer questions need be asked of the patient, especially
on family and social history.

2. Less writing is necessary; also, those work-ups per-
formed by other Clinic physicians are easier to read.

3. The automated medical history provides a good starting
point for more detailed questioning; it gives the phy-
sician a "head start. "

4. The history is "more complete."

5. The patient, having been forced to think about his
problems beforehand, becomes a better historian.

6. The history is helpful in bringing to light problems
that lie in areas other than the physician's own spe-
cialty.



Lahey Clinic AI^HS , Rockart et al . 10

7. Finally, even those physicians who wish to take their
own histories entirely have suggested that it does pro-
vide "a check on my own history."
Re searc h

The development of the AMIIS has provided a rich opportunity
to explore the fundamental process of medical history questioning
and its role in the determination of a diagnosis. Several pro-
jects have been undertaken which attempt to assess question

validity and the role of individual questions in contributing to

13,14
a final diagnosis.

Because of the anticipated use of the Questionnaire to
assist in patient scheduling, a continuing Study is being made
of the value of each question or series of questions in determin-
ing v;hich specialist is most appropriate for the patient to
see. The Questionnaire responses of several thousand patients
have been correlated V'/ith the final diagnoses that were subse-
quently made by Clinic physicians in order to discover which
questions are most valuable in indicating a particular specialty.
For example, preliminary results indicate that patients who re-
spond positively to the question "Do you find it necessary to
prop yourself up (with extra pillov/s or in a chair) in order
to sleep?" are more likely to have allergic disease (for example,
stuffy noses) rather than cardiologic disease despite the classic
medical thesis that orthopnea implies heart disease. It is
hoped that objective statistical analysis can replace the present
subjective criteria that are being used in question selection.



Lahey Clinic AIHIS , Rockart ct al . 11

In another study, focusing primarily on cardiologic condi-
tions, matched (by age and sex) sets of patients have been
established, one group having one of several cardiac conditions
and the other group being free from such problems. Using sta-
tistical techniques, the responses of each group to certain
questions are being analyzed to determine those questions - in
association with particular laboratory tests — which are most
useful in screening for cardiac disease.

A recent study has looked at the sensitivity and specificity

15
of 20 questions drawn from the M-IU Questionnaire. While some

questions proved to have a high correlation with the diagnoses

they were designed to suggest (e.g., joint and bone pains) , other

questions were found to be very poor predictors of diagnoses.

Such well-established questions as recent onset of orthopnea,

prominent eyes, and incidence of chest pain which increases with

sv/allowing had very low correspondence (less than 7 percent)

with the supposedly related conditions.

A final project, still in the planning stage, is concerned

with v/hether it is possible to use some of the data from the

Questionnaire to determine whether certain laboratory tests,

roentgenograms, cardiologic studies, eind other tests should be

performed. At present, no tests are ordered until after the

patient has had his first appointment v;ith the physician. It

is possible, however, that a few basic tests can be performed

prior to this first visit. In this way the physician would

have more complete information available to guide him in further



Lahey Clinic AMIS, Rockart et al. 12

evaluation of the patient's condition.

Description of the A IlHS

Since the first Questionnaire was administered in early
1968, three major revisions have been made- version V, the
fourth revision, is now under active development. The first
tv;o versions, drawn largely from the questions on the Massachu-
setts General Hospital on-line questionnaire, were experimental
in nature and were administered on a limited basis. Version III
v/as the first full-scale test and v;as administered to about
12,000 patients. It contained 392 "yes" and "no" questions
and a free-form answer sheet for chief complaint, drugs, and
so forth. The code numbers of the positive responses were
keypunched and these numbers were matched by the computer with
a response-symptom file. The resultant print-out was a list
of these symptoms. Little editing was possible and only a few
of the questions involved the use of any qualifiers, that is,
additional questions which helped determine the duration, severity,
or exact location of a complaint. In addition, the fact that
each Questionnaire had to be keypunched manually severely limited
the number of patients who could be processed on a daily basis.

The present version of the Al'^U Questionnaire (version IV)
was first distributed to patients in the spring of 1970 and
incorporates the experience of the three preceding versions.
In its present form, the MIHS operates as follows:

When a nev; patient contacts the Clinic for an appointment,
one of the secretaries finds the first available time in the



Lahey Clinic AJTHS , Rockart et al . 13

appropriate physicians' schedules and makes the necessary appoint-
ments. Preregistration material is then mailed to the patient;
and, if the appointment is more than ten days away, an AMII
Questionnaire is also sent in the same envelope. If there arc
loss than ten days, there is usually insufficient time to assure
the return of the Questionnaire in time for it to be processed
and inserted into the patient's medical record. Ilore than 85
percent of the patients v/ho receive the Questionnaire complete
it and return it to the Clinic.

The Questionnaire itself consists of a 25-page booklet v/ith
IGO questions covering family and social history, former illnesses,
and a review of systems. However, many of the questions have
several parts, and a better measure of its true length is the
number of possible responses, of which there are G19. Two sample
pages are shown in Figure 1. These pages illustrate how question
branching is achieved; if the cinswer to question 48 is negative,
the next several questions can be skipped. Most patients finish
the entire Questionnaire in less than an hour's time. In addition
to these multiple-response questions, the first page of the book-
let also contains a space for the patient to describe, in his own
v/ords , his chief complaint and any past hospitalizations, allergic


1 3

Online LibraryJohn F. (John Fralick) RockartAn automated medical history system; experience of the Lahey Clinic Foundation with computer-processed medical histories → online text (page 1 of 3)