Misclassification of sex: assessing an automated sex edit and the example of male breast cancer

Many first names are highly predictive of sex.These can be used to identify miscodes in public health data systems. A list of such names was developed using publicly available Social Security files. Because miscodes tend to be random, they are much more likely to accrue in rare categories. For example, about 100 times more women than men develop breast cancer, thus it is about 100 times more likely that a breast cancer case will be incorrectly miscoded as male. Results from the Florida Cancer Data System show that this was indeed the case, and rates of male breast cancer had been artificially inflated by about 15% as a result.