OBJECTIVE. We examined differences between blinded and retrospective reviews of screening mammograms obtained before a mammogram that resulted in the diagnosis of an impalpable breast carcinoma. MATERIALS AND METHODS. We reviewed 152 previous mammograms in 73 patients in whom impalpable breast carcinomas were subsequently detected on later mammograms. The earlier studies were interpreted in two ways: (1) blindly (without knowledge that carcinoma was subsequently detected) and (2) retrospectively (with the mammogram showing the carcinoma for comparison). The two interpretations were then compared with regard to the presence of carcinoma, recommendations for biopsy, parenchymal density, histologic characteristics of the tumor, lymph node status, and film quality. RESULTS. When we did a blinded review of the mammograms obtained before the diagnostic mammograms, the previous study was interpreted as showing evidence of carcinoma in 30 patients (41%). For the remaining 43 patients (59%), the findings on the most-recent previous mammogram were interpreted as normal or benign by the blinded reviewers; however, the retrospective reviewers thought evidence of cancer was visible in 25 of these patients (34%). Differences between blinded and retrospective interpretations were statistically significant. In patients in whom evidence of tumor was thought to be present on retrospective review but not on blinded review, the majority of mammographic abnormalities were asymmetric densities on the most-recent previous examination. This was true whether or not the retrospective reviewers thought that the mammographic finding warranted earlier biopsy. The histologic characteristics and lymph node status among patients in whom mammograms were interpreted retrospectively as showing evidence of tumor were no different from those among patients with no evidence of tumor. CONCLUSION. Our results show that impalpable breast carcinomas are frequently evident in retrospect on previous mammograms. However, because many are manifested only as an asymmetric density, these may not necessarily be true radiologic errors. Failure to detect a retrospectively visible abnormality on a screening mammogram is not necessarily negligent, and retrospective reviews do not reflect the everyday practice of screening mammography.
