William
A. Murphy,
MD
Jr,
#{149} Judy
M. Destouet,
MD
#{149} Barbara
S. Monsees,
Professional Quality Assurance for Mammography Screening
P
have always been concerned about the provision of quality medical care. However, recent trends have focused even greater attention on quality of care issues and have renewed the commitment to quality assurance (QA). The result is a major thrust from within the profession and from outside interests to define appropriate QA goals and to develop quality assessment tools and programs. With regard to mammography, QA goals and assessment programs are currently best defined and developed for the technical aspects of the examination. For
example,
the American
ogy mammography gram has centered cations, technical amination, image limitation breast.
and
of radiation Importantly,
implementation
come and
A recent
HYSICIANS
mostly
gress
along
sequent
of Radiolprospecifiof the exand
the
within
to the design,
program
QA lines
improvement quality, it must
wide average nized that there is still a long before quality mammography assured for all American Although there have
has
the profession,
is voluntary. has been impressive
technical
apparent
exposure impetus,
of this
from
participation While there
College
accreditation on equipment performance characteristics,
with
con-
of nationbe recog-
way to go
culties
encountered
standards mented, jectively will be quality tion to al service Again, programs profession.
were developed and implethese were relatively easy to obquantify. The more difficult task to define QA goals and develop assessment methods for applicathe performance of the professionaspect of mammography. these commitments, tools, and must come from within the
QA
report
from
Association
Service
states:
the
Programs’ American
Council
“A
on Medical
comprehensive
pro-
gram for managing the quality of care includes not only quality assessment but quality assurance activities, designed to assist practitioners in modifying practice
behavior
found
to be deficient
by quality
assessment, to protect the public against incompetent practitioners, as well as to modify structural or resource deficiencies that may exist” (1). The Joint Commission on Accreditation of Healthcare Organizations also emphasizes the development of such standards (2). The key elements of a monitoring and evaluation program are identifying the important aspects of the
care delivered, able
enumerating
critical
lishing
indicators
thresholds
the quantifi-
of the for
care,
evaluation
indicators,
collecting
data, taking thresholds
action to improve are exceeded, and
and
Accordingly, recognize
the
care when then reeval-
important established
ally
QA program that we predict will be widely imitated and adopted.
Although
many
ultigener-
radiology
practices have performed informal medical audits in the past, the trend in the 1990s will be to develop more formal methods, giving greater attention to obtaming quantitative information and to
doing
are
mography should
the
important
interpretation, be collected?
cle by Sickles
aspects core
et al addressed
of the
Radiology
1990;
175:319-320
‘From the Mallinckrodt Institute of Radiology, 5105 Kingshighway Blvd. St Louis, MO 63110. Received February 12; accepted February 14. Address reprint requests to W.A.M. RSNA, 1990 See also the article by Sickles et al (pp 323327) in this issue.
up in a sample findings, individualization
stage,
of results
for
status, previous
weight,
family and cancer, parity,
prior
mammography.
demographic but research ests. Clinical
breast
examinations
to obtain
for
and
personal menarchal
and
are
because
negative
examinations
difficult
to document
the
plained
by
are much for the
Sickles
et al,
large numbers of examinations no abnormality was reported. equally important to discover the false-negative examinations an excellent education
more
reasons
including
exthe
for which Yet, it is and review for there-
opportunity and improved
pretations. The follow-up large number of “normals”
for conservice.
effort for the is difficult
Patients
each radi-
surgery, The
easiest
facility
We question the inclusion of most demographic data. Demographic data collected, tabulated, and evaluated for each included of breast
false-positive
cared
ologist.
screenee history
and
and time consuming.
negative
of tumor
spective interpretations regarding the presence or absence of a significant mammographic abnormality and the followup data regarding the presence or absence of breast cancer with pathologic or acceptable clinical verification. These combined data are used to construct the truth table with true- and false-positive and true- and false-negative results. The search for this information is time consuming and expensive. The numbers of
arti-
this issue
of cases with
tabulation
audit.
The critical data that must be collected to achieve a valuable audit are the pro-
data
screening mammography. The components of that medical audit included patient demographic data, careful follow-up in all cases with positive findings, followIndex terms: Breast neoplasms, diagnosis, 00.32 #{149} Breast radiography, quality assurance, 00.11 #{149} Breast radiography, utilization, 00.11 Cancer screening #{149} Editorials #{149} Radiology and radiologists, observer performance
be mandatory for inclusion basic mammography
The medical audit described by Sickles et al likely underestimates the magnitude of false-negative mammographic inter-
of mam-
and what The
medical
in rests tinuing
so regularly.
What
facts may not in a responsible
mammography even these
original interpretation triggers the specific follow-up and because they are a small subset of the total. The true- and false-
to by
et al,
sional mately
when and where previous was performed. However,
generally
reported in this issue of Radiology (3). The medical audit of their screening mammography program includes many of the elements of a profesSickles
programs should seek patient information concerning personal and family history of breast cancer, the performance and results of prior breast surgery, and
true-
the effectiveness
it is highly foundation
the
estabof the
evaluating
uating the care to assess of the actions.
pro-
services are women. been many diffi-
as technologic
Medical
MD
for at the
underwent
same
mammography,
follow-up window
effort of time
may not be where
they
making
the
even harder. The narrow during which follow-up
is completed is inadequate to ensure the discovery of all false-negative results. This problem is compounded by extrapo-
lation of the false-negative fraction from a subset that may not be representative of the population screened. However, double reading all the “normal” mammograms might increase sensitivity but would be a massive effort of, as yet, unproved
more
benefit
costly
reading
that
would
screening
a statistically
result
program. valid
subset
in a
Double could
low-risk
data are superfluous to all programs with special interscreening mammography
Abbreviation:
QA
=
quality
assurance.
319
provide an estimate of the false-negative fraction attributable to a radiologist’s performance. Local, statewide, and national breast cancer registries with data that were cornplete, reliable, and timely might decrease the magnitude and duplication of effort encountered when overlapping screening programs registries
all seek follow-up might facilitate
data. a more
Such accurate
estimate of false-negative results. With proper safeguards, registries could enhance mammography audits, and their further
development
aged. Another reported
should
be
important
aspect
of the
Sickles
et al is the
by
encour-
audit
stratifica-
tion of data for each participating radiologist. For assessment of practice performance to be valuable and assurance to be effective, the data must be radiologist specific. Grouped data mask individual variations
requiring
practice
modification.
Identification of individual variance beyond threshold limits would trigger appropriate educational efforts. What are the appropriate thresholds for group and individual performance when interpreting mammograms? Current practice is to develop the audit indicators and thresholds locally. It is notable that in the report
by
Sickles
et al,
the
results
of their
medical audit are kept in local context. This approach may be adequate for the near future. However, it is likely that radiologists will eventually want to know
how
their
performance
compares
with
a
more general standard. Organizations outside of radiology may also request this information. The specialty of radiology will need, therefore, to address a national definition of assessment indicators and suggest thresholds at some future date, probably sooner rather than later. When the specialty considers national standards, issues are likely to include (a) ranges for the number of breast cancers expected in prevalence screens and in incidence screens, (b) the size and stage of each cancer, (c) the lymph node status, (d) the false-negative rate, (e) the truepositive biopsy rate, and (f) the rate of palpable cancers. However, any national standards must allow for local variations
due to patient age distribution
mix that would depend on and self-selection biases,
as well as differences in how “indeterminate” interpretations are defined and how they are assigned to positive and negative categories. What action should be taken when thresholds are reached or exceeded? It must be remembered that a medical audit or a professional QA program as generally conceived or specifically described for screening mammography is primarily a self-educational effort. There is no inherent intent
for
any
aspect
of this
program
to
elicit a punitive action. Performance of the medical audit is its own justification, for a review of personal results is the single best source of self-education. Moreover, radiologists must be aware that false-negative results will never completely disappear because some factors
320
Radiology
#{149}
are beyond us has
own
our
the
personal
or more
control.
opportunity
our
through
a screening
a more makes
of
one
programs.
professional
izing place
each
performance
of these
Establishing phy
However, to improve
QA
mammogra-
program
or formal-
already in is because mammography is the single best radiologic cancer detection test we have available. A quality technical study coupled with a quality interpretation is capable of regularly detecting very small cancers and favorably affecting patient morbidity and mortality. Radiologists are responsible for optimizing their professional performance and maximizing the benefit to the public. The prospect of instituting a professional QA program for screening mammography raises a number of additional considerations and questions. Among these are the ongoing nature of the process, the physical
informal program good sense. This
integration examination,
and QA of breast the potential
uses
and abuses of surveillance results, and the costs of such programs. Programs as described in this editorial are intended to be ongoing. The need for feedback never ends, but the design of the program need not always be the same. Any program should be reviewed periodically to determine if it meets current goals, and if not, changes should be
adopted. In our opinion, the issue of whether or not to include a physical examination in a mammography screening program is controversial cal factors
and (4).
dependent For example,
upon many loif all screen-
ees are referred by physicians following an office physical examination, it is costly and unnecessary to repeat the physical examination at mammography. Those who combine mammography and physical examination in their programs must realize they are responsible for the physical examination quality to the same extent that they are responsible for radiologic technical and interpretive quality. We question the ability to perform an adequate physical examination with the patient in a seated or standing position only. If a technologist or a nurse performs the examination, the radiologist must be
aware
of his/her
quality
of the
Radiologists should
responsibility examiner’s
who
review
the
on this
this
to be
structured
for practitioners such
systems
to “ensure
conducting who
are
acting
to reim-
The costs of mammographic professional QA programs could become great. To some degree, the details of the individual program will determine the costs. However, for any program, an employee will have to be assigned to the task. The major costs relate to locating and validating the medical care events occurring after mammography, and these take place outside the radiologist’s office. Therefore, telephone, stationery, postage, and computer costs must be added to the personnel
costs. The individual practice will have to decide how to pay these costs. Moreover, QA costs must be kept to a minimum and in line
with
the
concept
of low-cost,
ac-
cessible, mass-screening mammography. Eventually, we all will have to decide if the benefits are worth the added costs. For now, we believe it is appropriate to initiate
emphasize
programs
as discussed
the voluntary,
nonpunitive
nature
we think
that in many
above.
We
educational,
of the
process,
instances
and
such
pro-
grams may actually validate current quality performance. We believe that as more practices initiate screening mammography programs, their design should indude technical and professional QA aspects. These QA efforts need not be cum-
bersome radiologists screening
and should
not dissuade
from providing mammography.
low-cost U
References 1.
2.
3.
4.
and the
associated medicolegal aspects (5). Once gathered and analyzed, will QA data be used outside the department? While it is not the intent of such programs for data to be employed for external purposes, creative individuals may be tempted to do so. As indicated by the Council on Medical Service, it is important for a comprehensive audit such as
to tie results
bursement. In addition, competitive practices might attempt to use their data to gain a market advantage. These problems should be avoidable.
practice
of agency
laws
the data and desire
for the
performance.
rely
Even though the reasons to formalize ongoing review processes are altruistic and common sense, there are potential dangers. For example local, state, or federa! agencies could become interested in
5.
Council on Medical Service, American Medical Association. Guidelines for quality assurance. JAMA 1988; 259:2572-2573. Athanasoulis CA, Thnall JH. Standards of radiology practice: an approach to development. Radiology 1989; 173:613-614. Sickles EA, Ominsky SH, Sollitto RA, Galyin HB, Monticciolo DL. Medical audit of a rapid-throughput mammography screening practice: methodology and results of 27,114 examinations. Radiology 1990; 175:323-327. Monsees B, Destouet JM, Evens RG. The self-referred mammography patient: a new responsibility for radiologists. Radiology 1988; 166:69-70. James AE Jr. Greeson T, Price RR, et al. Le-
gal and ethical
issues
in a technologic
pline: the agency relationship. diol 1986; 21:815-817.
Invest
disciRa-
immunity
or applying in good
faith” (1). In fact, many state legislatures have enacted regulations meant to protect the confidentiality of quality assessment activities.
May
1990