Downloaded from www.ajronline.org by 121.10.203.37 on 11/04/15 from IP address 121.10.203.37. Copyright ARRS. For personal use only; all rights reserved
751
Computerized
Follow-up
Abnormalities Mammography
Detected at Screening
Debra L. Monticciolo1 Edward
A. Sickles
Obtaining of
a
follow-up
mammography
information practice,
after abnormal but
it can
be
of
interpretations tedious,
is an important
time-consuming,
and
aspect produce
suboptimal results. This study demonstrates the effect of computer-generated reminders on the collection of follow-up data for patients in whom abnormalities were detected at mammography screening. From April 1985 until December 1987, follow-up data for 1009 abnormal examinations were collected by using standard procedures, supplemented by occasional searches of pathology records and sporadic letters and telephone calls to referring physicians. Results showed that follow-up was limited to a normal physical examination in 104 women (10.3%); eight other women (0.8%) were lost to follow-up. For
over
90%
of
these
inadequately
followed
cases,
referring
physicians
were
first
contacted more than 3 months after screening. Since December 1987, follow-up data for 777 abnormal examinations were collected, supplemented by computer-generated reminders sent to referring physicians for cases apparently unresolved 3 months after screening. Only four (0.5%) of these women have had inadequate follow-up (normal physical examination alone).
We conclude
that use of a computer-generated
and more appropriate less onerous. AJR
patient
October
155:751-753,
management.
reminder
It also makes
system prompts more timely follow-up
data collection
much
1990
Timely follow-up of abnormal findings is an important aspect of any breast imaging practice, not only for ethical and self-educational purposes but from a medicolegal standpoint as well [1-4]. Obtaining pertinent follow-up data is the only way to document that appropriate management of mammographic abnormalities has occurred. Unfortunately, follow-up procedures tend to be tedious and timeconsuming [1 , 2], and standard methods (e.g., periodic phone calls by the radiologist and/or staff, repeated searches of pathology records) can produce unsatisfactory results [3]. The magnitude of the problem is likely to increase as more and more women use mammography screening. Computers are ideally suited to oversee the data management tasks of a mammography practice, including the acquisition and storage of follow-up information [5, 6]. This study was designed to determine the effects on collection of follow-up
cians Received March 1 5, 1990; sion May 1 , 1990. I Both authors: Department
accepted
after
data
whose
letters
detected
directed
at referring
at mammography
physi-
screening.
and Methods
Box
0628, University of California School of Medicine, San Francisco, CA 94143. Address reprint requests to E. A. Sickles. 0361 -803X/90/1 554-0751 © American Roentgen Ray Society
reminder
have an abnormality
revi-
Materials of Radiology,
by computer-generated
patients
From April 1 985 through February 1990, we did mammography screening examinations of 30,275 women in our Mobile Mammography Program. All women provide the name, address, and
telephone
computerized previously
number
of
a referring
database management [6]. All abnormal results
physician.
system are
reported
These
at the by
and
time
mail
and
other
data
are
of examination, by
telephone
stored
in
as described to
the
referring
a
752
MONTICCIOLO
physician
the
the nature
morning
dations
for
biopsy.
Physicians
possible
further
Downloaded from www.ajronline.org by 121.10.203.37 on 11/04/15 from IP address 121.10.203.37. Copyright ARRS. For personal use only; all rights reserved
but
From pushed
the onset
ofour
Mobile
recorded
approach, than
abnormal
examination
using
field
Mammography data
mine
existing
any follow-up
at
various
departments
institution.
intervals
physicians, In
information
procedures,
it,
until
Decem-
was
accom-
that
were
we
supplied
by
shorter than 2 months, no the disposition fields of our those for which appropriate
still
1 987, eliminating
we the
We then checked radiology, surgery,
of further
examinations
had
been
missing
was
requested
simplified computer
follow-up
searches
in our from
by telephone
our
own
referring
if necessary. data
and
the and
to deter-
done
workup
initially by mail but eventually
December
cases
receive
of communication;
data (no
for evidence
a subsequent
Any
we
Program
channels
6 months), we surveyed screening cases to identify
whether
pro-
in a separate
as
for abnormal
follow-up data had not yet been entered. computer-based records of our hospital’s pathology
as
or patients on their own initiative. To supplement
physicians
longer
and/or
as soon
is made, our computer for this
disposition
of follow-up
and
referring
this
record.
obtained this
into
by
patients
contains
the
primarily
aspiration,
their
an initially empty data field to store of the case. We then enter the pertinent
that
collection
describes
workup.
a record
completing
1 987,
report
imaging,
interpretation
disposition
information
thereby
The
as well as our recommen-
to contact
suggested
creates
database
follow-up
ber
the
an abnormal
linked
examination.
additional
are urged
automatically
the ultimate
the
of the abnormality,
evaluation:
to complete
Whenever gram
after
and location
AND
mail
records
screening
examination.
still
apparently We
use
unresolved the
cians who have not yet responded For purposes
of this study,
of women who were completely of
women
whose
examination. up. As 40-70% [7, 8], physical follow-up
subsequent
list
4 months
to telephone
or more referring
to a computer-generated
we carefully
documented
lost to follow-up evaluation
Both of these outcomes
was
after physi-
letter. the number
and also the number limited
represent
to
inadequate
physical
follow-
of screening-detected breast cancers are nonpalpable examination alone represents deficient and improper
for abnormalities
identified
October
1990
For over 90% of these 1 1 2 cases, referring physicians were first contacted for follow-up disposition data more than 3 months after initial mammography screening. During the second study period, there were 777 abnormal interpretations among 1 7,548 mammography screening examinations. The computer-based follow-up program has been beneficial in several respects. Our secretarial staff no longer has to conduct searches of hospital records from the departments of radiology, surgery, and pathology, and the letters requesting follow-up data that were sent manually to referring physicians are now generated automatically by computer. In addition, many physicians’ offices that previously required such letters began sending us follow-up information on their own initiative. We now send letters for only 20% of our abnormal cases (40% prior to computer-based operation), and we have to telephone for still missing data only two to three times per month (fewer than 1 0% of abnormal cases now, 20% prior to computerization). Among all 777 abnormal cases, none has been lost to follow-up and in only four (0.5%) was subsequent evaluation limited to a normal physical examination. The comparison of follow-up dispositions before and after institution of our computer-based program is shown in Table 1.
for
that had been
abnormal
AJR:155,
collection requests
carried out by our secretarial staff. Instead, once a month our computer program began to automatically generate the letters requesting follow-up information from physicians, but only for those records whose disposition fields remained empty 3 months after mammography screening. Only one such letter is mailed per case, but each month the computer program also prints a list of all information
SICKLES
by mammography.
Results During the initial study period, there were 1009 abnormal interpretations among 1 2,727 mammography screening examinations. In order to collect follow-up disposition information for the abnormal cases, we had to conduct multidepartmental computer searches in our hospital for 75% of cases, send letters to referring physicians for 40% of cases, and subsequently telephone doctors’ offices for data still missing in 20% of cases. These procedures were time-consuming (at least 8 hr per month) and thus were done only sporadically. Often delays of up to 6 months occurred before efforts were begun to acquire follow-up disposition information. By that time, some patients had moved, occasionally without leaving a forwarding address or telephone number. As a result, followup is unknown for eight cases (0.8%). Furthermore, 104 women (1 0.3%) had inadequate follow-up, consisting of nothing other than a normal physical examination of the breast.
Discussion Clearly there was more timely and appropriate follow-up of screening-detected abnormalities during the second study period. It is possible that this change was caused by factors other than computerization of follow-up data collection, since the two parts of our study were conducted sequentially rather than concurrently. However, because the change occurred abruptly, just at the time when computer-based operation began, we conclude that the computer program itself accounted for most if not all of the observed improvement in patient management. Several plausible mechanisms can help to explain how computerized procedures prompt more timely and more appropriate follow-up. By automatically generating reminder letters every month, no longer than 3 months after mammography screening, the computer program does not allow any abnormal case to be ignored for an extended period of time. Furthermore, the monthly regularity of these reminders not only shows our referring physicians that we are tracking all
TABLE 1: Disposition of Abnormal Interpretations of Follow-up Data Collection Procedures
as a Function
No. (%) Disposition
Standard Procedures
Unknown
8
Physical examination Additional
and/or
Biopsy Total
breast
alone
imaging
Computer-Generated Reminders
(0.8)
0
1 04
(1 0.3)
596
(59.1)
4 (0.5) 509 (65.5)
301
(29.8)
264
aspiration 1 009 (1 00.0)
(34.0)
777 (100.0)
Downloaded from www.ajronline.org by 121.10.203.37 on 11/04/15 from IP address 121.10.203.37. Copyright ARRS. For personal use only; all rights reserved
AJR:155,
October1990
COMPUTERIZED
MAMMOGRAPHY
of their patients with screening-detected abnormalities, but it also suggests that we are concerned that proper follow-up occurs as soon as possible after initial examination. Our assumption of some responsibility for follow-up also seems to have prompted referring physicians to cooperate more readily with requests for follow-up data. Since our introduction of computer-based reminders, many more physicians’ offices routinely send us copies of the reports from any additional procedures induced by abnormal screening interpretations. Our experience suggests that the interval between mammography screening and mailing of reminder letters should not exceed 3 months. Longer intervals appear to result in too many cases of inadequate follow-up, as we observed during the first phase of our study. Furthermore, when reminder letters are sent within 3 months of screening, one can expect virtually no abnormal cases to be completely lost to followup. An additional advantage of a computer-based reminder system is that it requires very little time to operate. Our secretarial staff now devotes less than an hour each month to collection of follow-up disposition information. Most of this time is spent placing reminder letters in mailing envelopes. We also need to make many fewer telephone calls for incomplete data because of improved physician compliance with the entire follow-up system. In summary, a mammography screening practice should not limit its activities to the detection of abnormalitiesensuring complete and timely follow-up of these lesions is also important. Although referring physicians have primary responsibility for patient management, radiologists are intimately involved as well. Whenever we find unsuspected ab-
SCREENING
FOLLOW-UP
753
normalities we become, de facto, the initiators of further workup. Any effort that we expend toward facilitating this evaluation not only helps the patient and referring physician but also enhances our own mammography screening practice. Furthermore, it can be highly instructive to analyze followup disposition information; such an exercise in self-assessment is probably the most effective method of refining mammographic skills [8]. There is no question that collecting follow-up data can be difficult. However, our experience mdicates that the use of a computer-generated reminder system
makes the process much less onerous and also prompts more timely and more appropriate patient management.
REFERENCES Evens AG. The self-referred mammography for radiologists. Radiology 1988:166:69-70 Sickles EA. Mammography screening and the self-referred woman. Cornmentary. Radio!ogy 1988:166:271-273 Robertson CL, Kopans DB. Communication problems after mammographic screening. Radio!ogy 1989:172:443-444 Edelstean G. Communication problems after mammography screening. Letter to the editor. Radiology 1989:173:879 Sickles EA, Weber WN, Galvin HB, Ominsky SH, Sollitto RA. Mamrnography screening: how to operate successfully at low cost. Radio!ogy 1986;160:95-97 Sickles EA. The use of computers in mammographic screening. Radio! C/in North Am 1987:25: 101 5-1030 Seidman H, Geib 5K, Silverberg E, LaVerda N. Lubera JA. Survival experience in the Breast Cancer Detection Demonstration Project. CA 1987;37:258-290 Sickles EA, Ominsky SH, Sollitto RA, Galvin HB, Monticciolo DL. Medical audit of a rapid-throughput mammography screening practice: methodology and results of 27,1 14 examinations. Radio!ogy 1990;175:323-327
1 . Monsees B, Destouet JM, patient: a new responsibility
2. 3.
4. 5.
6. 7.
8.