Breast pattern classification and observer error.

Using the Wolfe breast pattern classification, a study was conducted to test the reproducibility of this classification in view of its likely significance for breast screening and rescreening programs. The results are highly encouraging, and there was good agreement. Reclassification from low risk occurred in 1% of the cases; reclassification from high risk to low risk occurred in 1.4%. There was a variance of 2.4% in the low-risk group and 0.6% in the high-risk group.