The Impact of Mammography in a Public University Affiliated Hospital in an Urban Community

To the Editor:Previous published studies have shown considerabledifferences in mammography outcomes when auditingscreening versus diagnostic examinations (1–3); someof these differences have been shown to be statisticallysignificant (2,4). In this study, we present a compre-hensive medical audit of our mammography service atHarbor-UCLA Medical Center (Torrance, California),which is a public university affiliated hospital in anurban community to assess the efficacy of our practicein detecting clinically occult and early-stage breastcancer.This study was reviewed by our local InstitutionalReview Board. The consent requirement was waived.11,822 mammograms (screening and diagnostic) wereperformed from January 1, 2002 to December 30,2004 at Harbor-UCLA medical center of which 4,775(40%) were screening mammograms, and 7,047(60%) were diagnostic mammograms.Table 1 summarizes our outcome data for 4,775screening and 7,047 diagnostic mammography exam-inations. Out of a total of 252 cases of malignancydiagnosed in this study period, 72 (29%) were diag-nosed by screening mammography. The average age atdiagnosis was 52.8 (SD = 11.7) years with age rangebetween 20 and 90 years old. There were significantlymore abnormal findings (BIRAD categories 0, 4, 5) inthe screening group. However, we observed a signifi-cantly higher percentage of nodal metastasis and aver-age tumor size among the breast cancers in thediagnostic group, as expected.Table 2 summarizes the TNM staging of all the252 cancers segregated by initial diagnostic method;i.e., screening versus diagnostic mammography. Thistable fairly demonstrates that the screening groupmalignancy is diagnosed at an earlier stage. There issignificantly higher percentage of breast cancer casesclassified as stage ‘‘0 or 1’’ in the screening group. Onthe other hand, there are significantly higher percent-age of stage 2, as well as stage ‘‘3 or 4’’ in the diag-nostic group.Three cases were reported as ‘‘unable to be staged’’,two of which were histologically diagnosed as infil-trating ductal carcinoma and one was malignant epi-thelial hemangio-endothelioma.The most prevalent cancer histopathology observedwas invasive ductal carcinoma; 107 cases (59%) inthe diagnostic group and 30 (42%) in the screeninggroup. As expected, ductal carcinoma in situ wasabout four times more common in the screening popu-lation as compared with the diagnostic population(p-value < 0.001).We demonstrated a cancer detection rate of 29%by screening mammogram at our center, which closelycompares with the observed rate of 27% in a studydone by Sohlich et al. in fixed-site facilities at UCSFinvolving 40,691 screening and 11,114 diagnosticexaminations (4). However our results show signifi-cantly more advanced cancer at time of detection inboth screening and diagnostic groups as compared totheir outcome. In the screening group, Sohlich et al.reported 7% nodal metastasis and 89% stage 0 or 1cancer. We observed 17% nodal metastasis and 63%stage 0 or 1 cancer. In the diagnostic group theyreported 14% nodal metastasis and 75% stage 0 or 1cancer. Our study demonstrated 43% nodal metastasisand 18% stage 0 or 1 cancer.When comparing the above mentioned data(between our outcome and outcome reported by Soh-lich et al.), there is a statistically significant lower per-centage of stage 0 or 1 cancer in screening anddiagnostic groups in our patient population as com-pared to the data reported by Sohlich et al. (Fischer’sexact test p-values < 0.001). In other words, in a sig-nificant proportion of cancer diagnoses in our patientpopulation were made at a later (AKA moreadvanced) stage (stage 2 or above) as compared to theUCSF patient population.Furthermore, Sohlich et al. postulates expectedcombined outcome for various screening:diagnostic