(Wells et al., 1998). They found overall high agreement among the pathologists for assignment to diagnostic category (kappa coefficient = 0.71) and very high agreement for differentiation of benign versus malignant breast disease (kappa coefficient = 0.95.) Other studies have found that the adequacy of specimens obtained from fine needle aspiration and stereotactic core biopsy varies widely, as do their reported sensitivity and specificity (Acheson et al., 1997; Hayes et al., 1996; Stolier, 1997). Yet no published studies have explored the issues that affect the quality of these procedures as performed in the diverse clinical facilities across the United States.

The pathology report is the critical link between pathologist and clinician. Deficits in the pathology report may represent problems with communication or deficiencies in the pathologic evaluation itself. Several studies suggest that variation exists in the quality of pathology reports for breast cancer specimens that warrants further evaluation. In the mid to late 1980s, three reports, although conducted in different populations, suggested serious underreporting of pathology information. Less than one-quarter of biopsies, for example, had documentation of lymph node dissection (Table 4.3).

More recent data appear to show gains in some aspects of pathology reporting; however, room for improvement remains. In 1995, the College of American Pathology in its Q-Probes Study reviewed 20 breast biopsy specimens and the corresponding pathology reports from 434 voluntarily participating surgical pathology laboratories in the United States, Canada, and Australia (Nakleh et al., 1997):

• In 92 percent of malignant cases the margin status was reported (necessary to determine if all of the malignancy was removed by the procedure), and 77 percent of reports contained the lesion size.
• Approximately 75 percent documented whether estrogen and progesterone receptor status had been evaluated.

A single-institution study of needle localization breast biopsy reported even lower rates of documentation for this important information in the pathology report (Howe et al., 1995). Only 33 percent of reports in this study commented on the margins of the lesion, and estrogen receptor status was determined for only 68 percent of the cases. Such deficits in pathology reports have been the target of quality improvement projects. Hammond and Flinner (1997) reduced the number of incomplete breast cancer pathology reports in a large urban pathology practice in Salt Lake