What's most interesting with that is their complete openness with describing how this came about. It would make a great case study for anyone studying QA. Sounds like they got shafted by circumstance - a problem that is rare enough to easily slip through batch testing, serious enough to warrant recall, but wouldn't show up in the individual proof load test.
If I was doing a forensic analysis of their procedures, the first thing I'd want to know was if they upped the sampling rate of batch testing when they made the production change that lead to this, and if so, by how much and for how long.
