I like that you ran the numbers, and think you have a good question.

One remark in the original article said she had access to files and samples for many cases that she was not primary on. Perhaps that's the issue -- the ability to taint results across the whole departmental workload.