Lots of detail, and then big things like how not to conduct a statistical analysis, and then things that aren't backed up but you're under no obligation to back things up in private correspondence, are you?
As a rapid example, baby C. Evans claims that there was no time for the hospital to check the child's bile aspirates before he died, but the hospital's own records note that there was a missed opportunity to do this and it should have been done in the time available. He claims that the child's resuscitation was normal, but again, the hospital records note failure to get a chest rise, problems opening the airway, and failure to intubate. That kind of thing. But in many cases it's just rehashing what he said at the trial, reasonably enough, and while we know the counterarguments, I wouldn't expect to see them here.
I guess that's the nature of published correspondence.