Efficacy of Screening Breast Examination

Abstract & commentary

Synopsis: Barton and colleagues review the literature regarding the parameter involved with clinical breast examination (CBE) and suggest a particular approach. The role of clinical examination in contrast to mammography is emphasized.

Source: Barton MB, et al. Does this patient have breast cancer? The screening clinical breast examination: Should it be done? How? JAMA 1999;282:1270-1280.

The scheduling of a screening or diagnostic mammographic examination is often predicted on physician signs of symptoms of breast cancer, the most notable being a dominant mass or lump. Screening for breast cancer involves both physical examination and mammography, with studies showing complementary roles in increasing overall sensitivity for breast cancer detection. Many radiologists conduct physician examination of the breast as a complement to the imaging study or as a primary effort; others rely on the results of the primary care physician.

Barton and colleagues review the literature regarding the efficacy of breast cancer detection by clinical examination, trying to reconcile studies that are both inconsistent and nonstandardized with respect to the methods of physical exam. Reviewing published studies, contacting investigators for clarification of issues, and evaluating successful techniques using silicone breast models, Barton et al estimate the sensitivity of clinical breast exam (CBE) as 54% and specificity at 94%. The analysis is complicated by studies that have included both mammography and physician’s examination so that evidence of the efficacy of physical examination is consequently indirect. Dismissing mammography and histologic examination as gold standards for the accuracy of CBE—mammography because of Barton et al’s contention that it is unreliable, and histology because of the necessary absence of tissue for normal examination—Barton et al seek sufficient follow-up for a defined period in assessing their benchmarks. The likelihood ratio of a positive CBE result is reported at 10.6 and the likelihood ratio of a negative test as 0.47. Noting that only one study has ever described an employed technique, Barton et al examine different factors relating to the women, their tumors, and the examiner in reaching conclusions regarding current practice. For example, in five studies, mean examination duration was always longer for examiners with higher sensitivity. The highest sensitivity in human studies (69%) was achieved in the National Breast Screening Study in Canada, in which examiners took between five and 10 minutes to complete the exam of both breasts. Sensitivity increased with patient age. Barton et al detail a preferred approach, concluding that indirect evidence supports the effectiveness of CBE in screening for breast cancer. They describe and illustrate a vertical-strip search pattern with proper positioning and movement of the fingers, indicating that proper CBE requires at least three minutes per breast. Barton et al present data to indicate that both professional and lay examiners can improve their sensitivity on silicone breast models having mastered the described techniques.

Comment by R. James Brenner, MD, JD

Like the weather, everyone talks about CBE, but little is known regarding its effect. Many patients receive only a cursory exam or one by a clinician unfamiliar with proper CBE, complicating rather than simplifying the imaging strategy. The first prospective screen trial for breast cancer that showed a positive effect for screening—the Health Insurance Plan Study of New York in 1963—found a large percentage of cancers by CBE, rather than mammography. The survey of the field of CBE by these authors is useful in recognizing the limitations of assessing outcome, and tainted by previously established biases and selective reference regarding the efficacy of screening mammography. The NBSS study cited frequently is controversial; mammography was judged unsatisfactory by an independent panel and an inexplicable number of advanced breast cancers were placed in the "screening mammography" group. This approach may help elevate the importance of CBE in Barton et al’s view but lacks balance and validity. Such bias is found in surveying other trials. For example, where British screening mammography is only carried out every three years, the increased likelihood of interval cancer detection by clinical examination may be anticipated (most interval cancers are detected in the third year, which may indicate an insufficiently long interval for mammography rather than an increased efficacy of CBE). Barton et al’s conclusion that properly performed CBE can detect at least 50% of asymptomatic cancers is bold but not beautiful as a benchmark; perhaps it is better stated as a goal. The odds ratios predicted are likely inflated, as the complementary role between CBE and mammographic interpretation—which Barton et al mention—as well as the low incidence of a given studied population, may account for their findings. As has been previously studied, subthreshold mammographic features may rise to a threshold diagnosis when CBE is positive and ultrasound may be diagnostic of a palpable finding when the mammographic features are normal. The described technique, suggesting an average of three minutes per breast, adds to the available manuals for CBE for the interested radiologist who performs physician examination.