FORT LAUDERDALE, FlaWith the first results from its
Oncology Outcomes Database, the National Comprehensive Cancer Network
(NCCN) has taken a step to move beyond the guideline
development process into a process of connection, William
McGivney, PhD, CEO of the network, said at its Fourth Annual
Conference. That critical step is the implementation of
guidelines through performance measurement.
Dr. McGivney introduced Jane Weeks, MD, of Dana-Farber Cancer
Institute, who is the principal investigator for the NCCNs
outcomes and data project.
The NCCN began the project with a breast cancer pilot study,
involving five member centers, to get proof of concept
and to develop and test techniques for data collection, aggregation,
and analysis, Dr. Weeks said.
Patients eligible for the pilot are those with newly diagnosed breast
cancer receiving some or all of their primary oncologic care in an
NCCN institution. Were collecting sociodemographic data,
clinical data, and outcomes data, Dr. Weeks said.
Everybody in our database gets followed in perpetuity.
Data collection began on July 1, 1997, and the database now includes
1,542 cases. Most patients presented with early-stage disease.
There were only six patients in the whole database we were
unable to stage, Dr. Weeks said. This compares very
favorably with typical tumor registry data and, I think, reflects the
very aggressive training of our data collection staff at the sites,
and the coordination to ensure its accuracy.
The data show an extremely large spread among institutions in
the proportion of patients who are participating in clinical
trials, she said. For stage II patients, for example, it ranges
from 9% to 53%. This is important data to feed back to the
institutions to give them some sense of how their trial participation
rates compare with those of other similar institutions, she commented.
The first patterns of care analysis of the dataset involved the use
of breast-conserving surgery by stage. We found that among
women with stage I disease, 63% received breast-conserving surgery as
opposed to mastectomy. The range is quite wide, indicating clear
differences in patterns of care within the NCCN institutions,
The guidelines allow either type of surgery, reflecting the view that
survival outcomes are comparable and that it is appropriate to tailor
therapy to patients preferences. Do these differences
across institutions reflect patient preferences or physician
preferences and local patterns of care? she asked.
Were interested in feeding this data back and making sure
that all institutions are really relying on patients
preferences to choose between these equally efficacious
The guidelines recommend use of axillary lymph node surgery for full
staging in all patients. In stage 1, the rate is 86%, and in stage 2,
it is 96%. So this is routinely being done in our institutions,
although not absolutely in all patients, she said.
A Provocative Finding
The data also showed a significant association between age and use of
lymph node surgery. Older women are less likely to get axillary
procedures, she said, and this is provocative. It
means one of two things, she said: That physicians are using
legitimate discretion in not performing these procedures in older
women, in which case that ought to be reflected in our
guidelines, or that physicians may be undertreating some
Thats a key difference, Dr. Weeks said, and I
think its a nice example of how very quickly were in a
position to provide feedback to the guideline writers and work with
them to make sure that the guidelines really reflect optimal care.
Among women treated with breast-conserving surgery, 89% received
radiation therapy, again a measure of guideline adherence, with a
range across institutions from 80% to 100%.
Within the mastectomy group overall, 12% of patients received
postmastectomy radiation, ranging from 6% to 18%. The critical
question here would be whether the high-risk patients identified by
the guidelines as needing radiation therapy after mastectomy were, in
fact, the 12% that got it, Dr. Weeks said, adding that such an
analysis is planned for the future.
Data on adjuvant chemotherapy show different choices (doxorubicin or
no doxorubicin) being made based on stage and whether the
chemotherapy was administered within the NCCN or outside. Many
of us have long suspected that there might be more use of
Adriamycin-containing regimens in the centers than in the
communities, which are still a bit more reliant on CMF; it will be
interesting to see how that evolves over time, she said.
Dr. Weeks next presented some data looking at actual adherence to
guidelines for adjuvant therapy by selectively looking at the groups
of patients for whom there are specific recommendations that do not
involve patient or physician preferences. Among stage I patients with
small tumors, the guidelines specifically recommend no adjuvant
therapy (or enrollment in a clinical trial). In this subset, 58 of 71
patients (82%) had care that was adherent with the guidelines.
A Good Number or a Bad Number?
What do we make of that? Dr. Weeks asked. Is that a
good number or a bad number? The simple answer is that we dont
know because nobody else is doing this. It is really critical to
start collecting this sort of data so we can establish an appropriate
level of adherence.
The appropriate level, she continued, almost certainly, is not
100%. There needs to be some allowance for patient preferences and
special circumstances. She added that the NCCN does not require
physicians to record reasons for deviations from the guidelines.
In general, for a first-time look at adherence with no prior
feedback, Dr. Weeks called the 82% compliance figure remarkably
high. In other subsets, adherence to adjuvant therapy
guidelines was even higher, ranging up to 93%.
To sum up, she said that it is reassuring to know that we can
get these data, look at it, analyze it, and present it back to our
institutions. The analyses to date show that most care is
adherent to the guidelines, she said, but there is some room
for improvement, which suggests that there is a value in doing this.
Within the leeway provided by the guidelines, institutions are making
some different choices. The database will allow us to begin to look
at how those choices translate into differences in outcomes.
Dr. Weeks noted that the NCCN member institutions have been very
interested in receiving feedback on their patterns of care.
Theyre not frightened by these data, not threatened by
them. They really want to see what theyre doing, she said.
The NCCN plans to continue to expand the pilot program to include all
NCCN institutions and has begun work on the next outcomes database,
for non-Hodgkins lymphoma. Finally, as the database grows, the
network hopes to establish new partnerships with entities likely to
have an interest in the data, such as regulatory and accrediting
bodies, the pharmaceutical and biotech industries, and insurers.