Stephen J.LurieMD, PhD, Senior EditorIndividualAuthor
In Reply: Drs Finlayson and Birkmeyer suggest
that some information, however imperfect, is better than no information because
"more often than not, however, patients choosing 5-star hospitals will select
a hospital with better performance." This perception is incorrect. We found
that 1-star hospitals (the lowest rated) and 5-star hospitals (the highest
rated) were statistically indistinguishable in 92% of pairwise comparisons,
and 1-star hospitals had better risk-standardized mortality rates than 5-star
hospitals in only 5% of comparisons. In fact, in only 3% of comparisons did
the 5-star hospitals have significantly better outcomes than 1-star hospitals.
Furthermore, Finlayson and Birkmeyer's concerns about direct hospital comparisons
due to small sample sizes can be addressed by the use of multilevel models
that "borrow" power across hospitals.
Krumholz HM, Rathore SS, Chen J. Should Consumers Trust Hospital Quality Report Cards?—Reply. JAMA. 2002;287(24):3206-3208. doi:10.1001/jama.287.24.3206