In view of what's been happening in Baghdad over the last few years, and the fact that various estimates for the number of violent deaths differ by a factor of ten, how reliable are her statistics?

It's always a good idea to question the validity of any statistics. For example, it's possible that the statistics are wrong. It's also possible they're right but it's due to detection bias. By that, I mean if medical services are open to more people than before, then more people might be diagnosed with cancer than before while the actual rate of occurrence is the same. In that case, it doesn't mean more people are getting sick than before, it just means that more people are getting proper diagnosis. Before, they may have just died on unknown causes.

Depleted uranium is used as an anti-armor munition. It isn't used all that commonly. I don't recall all that many armored actions in Baghdad, so widespread use of DU there seems unlikely.