Anthony Hatswell has had an article published in Value in Health, on the effects of model misspecification. The article shows where analysts should be wary of using the technique, but also that in the majority of cases it remains both accurate and unbiased.