INTRODUCTION
Chairman Barton, Congressman Klink, distinguished members of the Subcommittee, thank you
for inviting us here today to talk about our plans to incorporate commercial off-the-shelf
software, known as "COTS," into our efforts to make sure Medicare pays only for claims that are
appropriate.
There are many commercially available, "off the shelf" software systems that include
computerized instructions to identify improper claims. Each computer instruction which verifies
information on a claim is called an "edit." Commercial automated systems are currently used by
many private insurers, the Department of Defense, the Veterans Administration, some state
Medicaid programs, and other public payers.
We have recently concluded a field test of one commercially available claims processing and
review system, including its edits. Results of this test suggest that national incorporation of some
of the edits in this system into Medicare claims payment systems could save Medicare up to
$465 million a year. Although actual savings may be less, I have instructed staff to take the
necessary steps to implement the edits nationally.
BACKGROUND
Medicare processes more than 800 million claims each year. We pay nearly half a million
physicians some $40 billion each year for their services. Screening out improper claims is critical
to preserving our scarce resources
Medicare already uses computer claims edits that we developed to help ensure the accuracy of
claims payment. Claims are subjected to a rigorous prepayment electronic screening process to
verify:
- beneficiary information, such as whether the patient is enrolled in Medicare and if all
co-payments and deductibles have been met;
- provider eligibility and standing with the Medicare program;
- utilization history (for example, we pay only one claim in a patient's lifetime for an
appendectomy);
- whether a beneficiary has other insurance that should pay instead of Medicare.
Since 1994, we have had a contract with AdminaStar Federal to develop edits to detect claims
with codes for services that cannot or should not be performed together or for services that
should be grouped together and paid as one item at a lower rate than if billed separately. Our
system, known as the Correct Coding Initiative, was first implemented in 1996. It includes some
93,000 computer edits and saved almost $260 million in FY 1997.
As you know, we recently concluded testing in Iowa of a commercial claims processing product
produced by the GMIS Product Group (a subsidiary of HBO & Company) known as
"ClaimCheck." We determined that the software for the claims processing aspect of this product
could not meet Medicare needs without substantial modification, and that it replicated much of
what our systems already do. In fact, the test showed that its use could significantly increase
processing time and delay payment.
However, the product contains many edits that are not similar to any in our Correct Coding
Initiative. Therefore, we focused on the edits used in the product. Medicare and our private
contractor first carefully reviewed ClaimCheck edits to make sure they conformed with
Medicare policy. This was essential, as several inconsistencies were found. GMIS made
modifications to some edits so they would conform to Medicare policy.
However, it is important to note that it was not possible to individually review all of the edits for
consistency with Medicare policy, so some ClaimsCheck edits may not have been compatible
with Medicare policies during the evaluation. Before we adopt any edits outside of an
evaluation, we must ensure that they conform to Medicare policy in order to ensure that claims
payment is consistent with Medicare policy.
SAVINGS
The testing, which concluded last fall, determined that the additional edits tested, but not the
software, result in real savings in addition to savings we are already attaining from our Correct
Coding Initiative. Projections based on results of that test suggest that implementing these
commercially available edits nationally could save as much as $465 million annually in addition
to savings generated from the Correct Coding Initiative. A General Accounting Office report
being released today provides more details of the test and its results.
The potential savings are realized from two distinct components. The first,
procedure-to-procedure edits, could save up to an estimated $205 million, based on the
approximately 500 procedure-to-procedure edits that were tested. Procedure-to-procedure edits
ensure that Medicare does not pay separately for services that should be paid for together.
For example, when a physician opens a patient's abdomen and finds nothing wrong, Medicare
pays for an exploratory laparotomy. If the physician finds an infected gall bladder, Medicare
pays for its removal. But Medicare should not pay separately for both laparotomy and gall
bladder removal at the same time. Obviously, the physician had to open the patient's stomach
to remove the gall bladder, and that's already figured into Medicare's payment for gall bladder
removal. Procedure-to-procedure edits ensure that Medicare does not pay separately for a
laparotomy and a gall bladder removal at the same time.
The second part of the commercial edits is diagnosis-to-procedure edits. Diagnosis-to-procedure
edits like those in the commercial product tested do not exist in the Correct Coding Initiative.
These edits are projected to save as much as an additional $260 million annually, based on the
approximately 900 such edits in the commercial product tested.
Diagnosis-to-procedure edits detect claims that are improper because the procedure would not
be appropriate for the given diagnosis, such as bypass surgery for a patient with a broken leg.
Some of the more sophisticated diagnosis-to-procedure edits can detect when a procedure in a
given category is inappropriate for a diagnosis in that same category, for example, bypass
surgery for simple chest pain.
PROCUREMENT AND IMPLEMENTATION
Based on the results of our test, and in consultation with Congress, I have decided that we will
proceed immediately to procure a commercially developed database of edits and implement
them nationally. These negotiations are currently underway, and I am hoping to move them
along as expeditiously as possible.
We will continue to expand and refine our Correct Coding Initiative system. Our contract with
AdminaStar Federal for this work does not expire until July 1999. AdminiStar's ongoing edit
development work will eventually be integrated into our efforts to explore other claims
processing edit solutions.
As Administrator of HCFA, I am committed to making use of off-the-shelf products whenever
feasible. We are proceeding, and I have instructed staff to explore the use of commercially
available software products in other functions of the agency. There are, however, some
remaining issues.
Year 2000 Compliance
First, I must stress that implementation of commercial edits cannot interfere with our number
one priority, which is making sure that all our computer and information systems, and all our
contractors' computer and information systems are Year 2000 compliant.
We will implement the selected procedure-to-procedure codes nationally as quickly after
procurement as our Year 2000 obligations allow. We will implement diagnosis-to-procedure
edits once they have been thoroughly reviewed to ensure that they are consistent with Medicare
policies. We will get back to this committee with a timetable for implementation once
procurement is complete.
Disclosure
Another remaining issue is the question of whether to disclose these edits for public review. We
make our Correct Coding Initiative edits and the underlying coverage policy on which they are
based available for review by anyone. The American Medical Association in fact helped
validate the Correct Coding Initiative edits.
However, we may not be able to release commercial edits themselves for public review because
of ownership issues and the need to stay ahead of unscrupulous providers. There may be
advantages to nondisclosure, in that it could help deter physicians from gaming the Medicare
anti-fraud and abuse system. Other payers, including the Department of Defense, Veterans
Administration, and several state Medicaid programs, use commercial edits without releasing
the actual edits to the public.
But, while some edits do relate to fraud and abuse, the majority do not. For example, the
commercial software tested denies payment for shaving of knee cartilage when a physician
performs certain other arthroscopic procedures on a knee. It considers the shaving to be
"incidental" because the leg was already prepped and draped and the arthroscope in position in
the knee joint for the other surgery, and the shaving requires little additional physician time,
skill or risk. The physician billed correctly, but the second procedure does not warrant a separate
payment.
Not releasing edits for public review, particularly where physician gaming is not an issue, may
be a matter of concern to physicians who have been allowed to review our Correct Coding
Initiative edits. The AMA House of Delegates has passed a resolution criticizing use of what it
calls "black box" edits that are not released for public review.
At the very least, we will make the underlying coverage policy available to anyone, and provide
the coverage policy rationale when any claims denial is based on a commercial edit. Enough
information will be made available for the physician to understand why claims have been
denied.
This issue is currently a matter for negotiation as we procure commercial edits. We are
interested in this Committee's views on public disclosure, and look forward to working with
you on this sensitive issue.
Uncertainty About Savings
Another concern is uncertainty about savings projections based on our test in Iowa. Not all edits
tested were examined to ensure consistency with Medicare policy, and we must conduct a
thorough review of all edits before implementation to ensure consistency (in this process will
consider modifying policy if a given edit helps meet program goals). Our Correct Coding
Initiative now includes some edits with the same function as some of the commercial edits
tested, so savings based on those edits are already being achieved. Physician diagnostic coding
may improve as they see that coding errors lead to claims denials, and with fewer errors there
would be correspondingly fewer savings. And appeals could increase because of claims denied
by the commercial edits. These factors could lead to reduced savings from initial
implementation of commercial edits.
However, we are not stopping our evaluation of commercial edits with just one database of
edits. Adding more edits could increase savings further. To ensure that we reap the full benefit
of commercially available edits, we will conduct a full and open procurement next year so that
we can see a wide range of available products.
Fraud, Waste, and Abuse
Commercially developed edits do help identify claims that should not be paid as submitted, and
so will be an important addition to our efforts to make sure Medicare makes only appropriate
payments. However, it is important to understand the limitations of these edits in addressing the
problems of fraud, waste, and abuse in Medicare. Edits can only identify inappropriate
claims based on information submitted in the claim. The latest Chief Financial
Officer's (CFO) audit found that, based on information submitted in the claim,
just 2 percent of Medicare payments are inappropriate. Still, this 2 percent totals
some $3 to $4 billion.
However, by looking behind the claims at the medical records to support them, the CFO audit
found that 11 percent of all Medicare payments -- some $20 billion -- were inappropriate. The
vast majority of these inappropriate payments were found only when human reviewers
requested and reviewed the provider's actual medical documentation and found that it did not
support the claim as submitted to Medicare. Even with the help of more sophisticated
commercial edits, we would not have found most of these errors without human
intervention.
CONCLUSION
Commercially developed edits should be an important addition to our efforts to
protect Medicare program integrity and ensure that we make only appropriate
payments. The caveats that I have discussed merit close monitoring. But
savings from these edits, even if less than projected based on the results of our
test, should still be substantial and well worth the relatively small cost of
procurement and implementation. We will maintain close consultation with this
committee as we proceed. I thank you for your continued interest and support as
we proceed as expeditiously as possible. And I would be happy to answer your
questions.