NCCN Physicians Get an ‘A’ for Guidelines Concordance

FORT LAUDERDALE, Fla—The 17 institutions of the National Comprehensive Cancer Network (NCCN) are adhering closely to their own practice guidelines for breast cancer, according to a report given by the NCCN Outcomes Committee.

“For the first time out of the box, I give them an ‘A,’” outcomes committee chair Jane C. Weeks, MD, said at a press briefing held in conjunction with the NCCN’s Fifth Annual Conference.

Dr. Weeks is associate professor of medicine, Harvard Medical School, and director of the Center for Outcomes & Policy Research, Dana-Farber Cancer Institute.

This was the first study of guidelines concordance since the NCCN began its first outcomes database 3 years ago of women with newly diagnosed breast cancer. The committee’s report studied the records of the first 3,137 women enrolled in the database and determined concordance with the treatment they received, compared with the treatment recommended in the guidelines for the year they were treated.

One example of highly concordant practice was in the use of adjuvant chemotherapy to treat breast cancer patients with stage IIA and IIB disease who were node positive and hormone-receptor negative. In this group, 130 women (98%), were receiving appropriate chemotherapy. The range of concordance among the NCCN institutions was 93% to 100%.

Analysis of areas where concordance with the guidelines was low revealed some examples of possible weaknesses in the guidelines themselves.

For instance, the guidelines state that women with ductal carcinoma in situ (DCIS) with widespread disease should undergo a mastectomy and no axillary lymph node dissection.

The outcomes report found that 89% of the women in the sample of 85 had appropriately received a mastectomy. However, a large majority of women were also receiving axillary lymph node dissection. Indeed, only 35% of women had not undergone axillary lymph node dissection.

“The reason is that the women who are getting mastectomy are also getting a level 1 node dissection,” Dr. Weeks said. “Surgeons feel that level 1 node dissection should be considered appropriate care. This is not specifically addressed in the guidelines, but may have been what the guidelines panel intended.”

She said that feedback from her committee to the breast cancer panel should result in clearer, more appropriate guidelines. “We want to see evolution toward very high rates of concordance,” she said.

In another instance, the guidelines lagged slightly behind newly published research on tamoxifen (Nolvadex) as chemoprevention. For example, the outcomes committee found low concordance (68%) with the recommendation for no adjuvant therapy in women with stage I, IIA and IIB breast cancer, negative nodes, and tumors less than 1 cm in size.

Most of these women were, in fact, not receiving chemotherapy, but instead of no adjuvant therapy, they were receiving tamoxifen, presumably for chemoprevention.

Physicians who adopt changes in practice early will always be ahead of printed guidelines, Dr. Weeks told ONI. “When the guidelines are implemented, they will catch up to the early adapters and encourage late adapters to follow,” she said, adding that physicians at the NCCN centers are, on average, early adapters.

When Physicians Disagree

The outcomes committee also found at least one example where the lack of concordance seemed to suggest strong physician disagreement with specific guidelines. The guidelines recommend that patients with stage IIIA breast cancer with one to three positive nodes undergo a mastectomy and receive chemotherapy, radiation to the chest wall and supraclavicular areas, and tamoxifen for 5 years (if estrogen-receptor status is positive or unknown). Concordance with this recommendation was only 59% in this sample of 70 women. The problem centered around the recommendation to irradiate the chest wall: Only 37% of the women were receiving radiation. “This is an educational opportunity within our institutions,” Dr. Weeks said.

Patient refusal of treatment seems to account for relatively few instances of nonconcordance, according to a pilot study conducted at three NCCN centers. The study analyzed 286 examples of nonconcordance that it found at those centers, Dr. Weeks reported. Patient refusal accounted for only 9% of the examples. “We were surprised that patient refusal accounted for such a small percentage,” she said.

Special patient characteristics—such as age or presence of other disease—that precluded recommended treatment accounted for 41% of the cases of nonconcordance. “That is appropriate decision making—to deviate from the guidelines when they are not appropriate for the patient,” she said.

Disagreement with the guidelines accounted for 35%. No reason (8%) and follow-up data missing (7%) accounted for the rest. These last two are “targets for improvement,” Dr. Weeks said.