If you’re lucky enough to be able to choose your hospital and have time to research your choices, you’d naturally take mortality rate into account. There’s a problem with that, and one change in how mortality rates are calculated could help solve it.
I think the authors may have missed a larger point. The variation in DNR patients may track facilities who serve more acutely ill populations. Without a more sophisticated look at the populations served, acuity of patients presenting for care, and level of patients kept versus being transferred out due to acute illness – incorporating DNR orders in the score weighting may not make them more accurate.
The interesting point for further research is whether or not DNR orders are a useful indicator or perhaps substitute measure for the other complex variables I noted above. Not having access to the study, I can’t tell whether the authors considered the point.