A closer look at the real cost of obtaining and utilizing a compter-aided detection system (CAD).
is the President of Breast Health Management, Inc., Bend, OR. He
has provided consulting services to R2 Technology, Inc., and
holds a small equity position in the company.
Most physicians--and other healthcare providers who must make
decisions concerning new medical technologies or techniques--are
reluctant capitalists. While the first criteria for the purchase or
adoption of the technology/technique is clinical efficacy, to be a
sustainable addition it must also meet economic objectives. As one
radiology practice manager recently commented to the author: "Our
purchase decisions are based on how the technology advances our
ability to deliver care, not on profitability concerns. Now, tell
me about the reimbursement!"
There is nothing wrong with this approach. In fact, it is
healthy. The appropriate allocation of scarce resources is a
serious challenge in today's demanding healthcare environment. It
is important to first determine the clinical benefit that will be
conferred by a new technology or technique and then weigh that
benefit against its cost and the return on the required investment
or effort. Obviously, expensive technologies that present
substantial clinical benefit but do not provide reimbursement to
reasonably amortize the costs involved present a difficult
challenge. There are, however, technologies like the computer-aided
detection (CAD) technology for mammography that present a clear-cut
decision. The analysis that follows will illustrate the decision
process, using the ImageChecker system (R2 Technology, Los Altos,
CA) as its example, since this currently is the only FDA-approved
The clinical decision
For several years, radiologists have known that double reading
had the potential to reduce the false-negative rate in screening
Methods for double reading vary from protocols that are designed
only to increase the sensitivity of the review to those that also
employ a consensus review in an effort to control the work-up or
recall rate. Studies of the various methods have shown an increase
in the cancer detection rate of up to 15%.
Unfortunately, double reading entails a significant additional
investment in physician time and, equally important given the
current shortage of breast radiologists, practical difficulties in
Throughout the 1990s, CAD was being developed and was gaining
the interest of radiologists who saw it as a potential answer to
the staffing issues posed by double reading. In mid-1998, the U.S.
Food and Drug Administration (FDA) approved the ImageChecker. In
its labeling statement approval, the FDA found that "for every
100,000 cancers currently detected by screening mammography, the
use of the ImageChecker could result in early detection of an
additional 20,500 breast cancers."
The sensitivity of the ImageChecker had been established in the
clinical documentation presented to the FDA and had been supported
by subsequent publications.
But Freer and Ulissey
established that the ImageChecker could be integrated into clinical
practice without unduly increasing the recall rate. A complete
discussion of the Freer study is beyond the scope of this article;
however, its conclusion is important. (For more detail, see the
discussion with Freer and Ulissey on page 17 of this supplement.)
To quote directly from the study of 12,860 women, "the authors
observed the following: a) an increase in recall rate from 6.5% to
7.7%; b) no change in the positive predictive value for biopsy at
38%; c) a 19.5% increase in the number of cancers detected; and d)
an increase in the proportion of early-stage (0 and 1) malignancies
detected from 73% to 78%."
Simply stated, while use of the ImageChecker can be expected to
increase the recall rate slightly, the increase is reflected in
increased diagnosis of cancers, and those cancers will
predominately be early-stage. This finding not only proves its
clinical efficacy, but establishes it as a cost-effective
technology insofar as its impact on procedural volumes and
utilization are concerned.
The financial impact
With the clinical efficacy and utilization impact of CAD, the
technology must be examined from a financial perspective. To do
this properly, it is important to examine the additional costs that
will be incurred, and then compare these costs with the expected
additional revenue that will be generated by the technology.
The purchase of a CAD system represents a large capital
expenditure, with essentially no supply cost (there is some cost
for bar code supplies; however, these are reused and are therefore
not considered as a variable cost). The only other significant
costs are maintenance and any additional labor that may be required
to accomplish the digitization process. Table 1 includes the basic
assumptions for the model that the author utilized to analyze the
In addition to the assumptions presented in Table 1, the author
assumed that, at the volumes that would be analyzed, the model of
ImageChecker selected would require 0.5 full-time equivalent (FTE)
of a technical aide to digitize the mammographic images. Actual
practice has shown that little additional time is required for this
function; however, the model has been developed from an
intentionally financially conservative perspective.
Because the costs are, for all intents and purposes, fixed, the
additional cost increment added by the ImageChecker varies directly
with volume (within the capacity of a given unit or number of
units). Table 2 illustrates the cost per mammogram reviewed at
varying annual volumes within the capacity of the equipment
modeled. It should be noted that CAD review is an all-or-nothing
proposition. Selective use of CAD on only a portion of screening
mammograms would create significant standard-of-care issues.
Notice the variation on Table 2 between the columns for each
daily volume. This model has assumed starting volume equivalents as
shown, with a growth rate of 5% per year. The increased cost that
occurs in year 2 is due to the maintenance/update charges that are
assumed to commence in year 2.
The revenue picture is not as clear as the cost side. Congress
legislated reimbursement for CAD (effective from April 1, 2001 to
December 31, 2001) equivalent to $15.00 ($4.80 professional/$10.20
technical) in excess of the statutory screening mammography fee.
The Center for Medicare and Medicaid Services (CMS--formerly the
Healthcare Finance Administration) recently announced an increase
in CAD reimbursement for 2002 to an expected rate of approximately
(The 2001 Medicare conversion rate is used in this computation, and
the conversion rate is expected to increase for 2002. The
reimbursement code is Gxxx4.) The American Medical Association also
recently announced the issuance of a separate CPT code for CAD,
76085, which will be used as an add-on code with the screening
mammography code (76092).
Comparing the 2002 Medicare reimbursement rate with the costs
listed in Table 2, it is clear that the breakeven point for CAD is
slightly more than 15 mammograms per day. Unfortunately, this also
assumes that all payors will reimburse at the Medicare rate and,
sadly, that is not the case. As mentioned previously, payors tend
to drag out the process of establishing reimbursement for new
technologies. While R2 Technology has an organized reimbursement
support effort and has been very successful in obtaining
reimbursement, obtaining approval happens one company at a time, on
a regional and even local level. It is expected that most, if not
all, payors will reimburse for CAD within 1 year.
The expectation of rapid reimbursement acceptance is based upon
the fact that breast disease has perhaps the highest emotional,
political, and media profile of any U.S. healthcare issue. Women
make most of the healthcare decisions in the United States,
including the selection of a healthcare plan for their families.
The issue of CAD reimbursement cannot be hidden. Approximately 60
million women in the United States are eligible for an annual
screening mammogram. CAD analysis is a clinically proven,
adjunctive study for screening mammography that applies to all
screening mammography patients and therefore has a very large
It is the nature of commercial reimbursement systems to be very
tentative in their acceptance of new technologies. These companies
are, after all, businesses, and the engine of business is profit,
even if that business is medicine.
They are also pragmatic and, while they can delay reimbursement for
CAD temporarily, there is a very real risk that prolonged denial
may have a negative effect on subscriber enrollment. That is a risk
that will prove to be too great.
For the sake of analysis, however, assume that commercial payors
do not pay for CAD and that the only source of additional CAD
revenue is from patients who agree to self-pay a $25 charge for the
service. (Standard-of-care/risk-management considerations dictate
that all screening mammograms receive the benefit of CAD.) Further
assume that only 40% of non-Medicare patients self-pay. Given these
assumptions, Table 3 illustrates the net revenue produced, after
deducting the incremental cost of CAD, at each volume in excess of
a breakeven volume. It is assumed that 30% of the mammography
screening population is age 65 or older and enrolled in
Despite the relatively dismal assumptions included in the model
for Table 3, CAD appears to be cost-effective even at decidedly low
volumes. In fact, the lowest volume on Table 3 is just slightly
higher than full utilization of one mammography unit operating on a
20-minute screening schedule. Reports from early adopters of the
ImageChecker system who have instituted self-payment programs
indicate high patient acceptance rates. Increasing coverage by the
popular media is likely to increase this acceptance
and strengthen the pressure on payors to provide reimbursement.
What is the reimbursement environment likely to be in a year?
Using the term "year" is rather generic. Some areas of the country
are adopting CAD rapidly, and may mature earlier than areas in
which the adoption, and consequent pressure on payors, is slower.
Some areas of the nation also have more aggressive reimbursement
environments, and there is wide variation in payment for medical
services across the United States. If, however, a few assumptions
can be made with respect to reimbursement levels, a picture of the
future can be drawn.
Table 4 illustrates the financial results of applying the
following assumptions in the author's model.
Patient insurance coverage:
Commercial coverage: 45%
Note that the charge for CAD interpretation must be set
significantly higher than the Medicare reimbursement rate because a
significant percentage of payors will set payment rates as a
percentage discounted from the "charge." The charge is almost never
paid at the full rate. Commercial insurance is assumed to receive a
discount of 40% of charges. Self-pay patients are given a discount
of 30% off of charges. The capitated category includes patients for
whom a new mammography rate cannot be negotiated or who, for
whatever reason, will not pay any additional amount for CAD. While
this percentage will undoubtedly decrease over time, it is included
at this high level in the interest of maintaining a conservative
bias to the financial projections.
Obviously, most practices and centers will fall in the lower end
of this spectrum. It is especially important that a practice with
20 patients per day is well above the breakeven point for CAD,
because this enables this important technology to be "driven down"
to the nation's smaller communities that sometimes are not afforded
the benefits of technological advance. This fact also makes it
incumbent upon all payors to reimburse at levels that are high
enough to make CAD economically viable. Although Table 4
illustrates that there is a substantial economic incentive for the
implementation of CAD at the higher volume levels, the majority of
mammograms in this nation are delivered to patients by lower volume
practices. With a technology of this clinical significance, it is
important to protect access for all patients, not just those in
Why CAD works economically
From a financial perspective, mammography has always been a
marginal procedure. Not only has the reimbursement been low, but
federal and state regulations and requirements specific to
mammography have driven the cost of delivery to a level at which
most practices consider screening mammography to be a financial
loss. The recently proposed physician fee schedule governing
Medicare fees creates a substantial increase in reimbursement for
screening mammography in 2002, but the increases will most likely
cover only the cost of delivering and interpreting the
In view of the foregoing, it may seem questionable how CAD, a
low-reimbursing procedure, can produce such a substantial economic
effect. The answer is really fairly simple. A large part of the
cost of a mammogram is administrative. The problem with screening
mammography is that, almost by definition, the patient receives
only the one procedure during a given visit and that all ancillary
cost (scheduling, reception, registration, billing, filing, etc.)
must be borne by the reimbursement for the one service.
Computer-aided detection, other than the costs enumerated in Table
1 and its supporting text, adds nothing to the cost of the
patient's visit. Because there is no per-patient cost added, net
revenue per patient rapidly increases with volume, and moves
straight to the bottom line.
Computer-aided detection passes both the clinical and the
financial tests. It is not only a solution for the few high-volume
mammography centers in the United States, it can be accommodated in
almost every practice that can keep a mammography unit busy. It is
truly one of those rare healthcare solutions that benefits everyone
it touches. Patients benefit from earlier detection of breast
cancers, and therefore decreased mortality from breast cancer.
Payors benefit in that the decreased staging of the CAD-detected
cancers allows lower morbidity and decreased treatment costs. The
community benefits from healthier, more productive women whose
cancers have been detected earlier and who can be treated
successfully. Finally, providers benefit in that this clinically
valuable technology can be provided to patients in a manner that
does not detract from the practice or center financially.
Opportunities like this are rare--they should be recognized and
embraced wherever and whenever possible. *