Years ago, Belinda Ostrowsky, then at CDC, showed in the NEJM that VRE could be controlled through regional infection control efforts in the Siouxland region of Iowa, Nebraska, and South Dakota. David Smith, then at NIH, was motivated by this success to test whether this regional response was necessary for control of a generic hospital pathogen using mathematical models. (see his PNAS study). He found that regional coordination may be necessary. So what about MRSA?
Hajo Grundmann and others in Europe have just published a very interesting paper in PLoS Medicine with an accompanying editorial by Frank Lowy. The authors collected MSSA and MRSA samples from 450 hospitals in 26 countries during 2006–2007 and completed spa typing on all of the isolates to determine genetic relatedness. They then geographically mapped the spa types. What they found was the unlike the widely distributed MSSA, dominant MRSA spa types formed distinctive geographical clusters within regions. Check out their interactive map (here). I think this shows that regional efforts will be required to control the spread of MRSA since it appears that it's mainly spread by patients who are re-admitted to different hospitals.
I won't begin to suggest what those efforts should be, but we certainly have multiple potential interventions and a single approach won't work in all places all of the time. However, since JJ Furuno's Archives of Internal Medicine paper a few years ago found that patient self-report of a hospital admission in the previous year detected 76% of MRSA carriers on admission, we have been using this regional spread hypothesis to control MRSA at University of Maryland Medical Center. Instead of obtaining nares swabs on 100% of admissions, we've swabbed only the highest-risk patients (admissions in the past year to any hospital and/or active skin infection) and detected and isolated almost all patients who subsequently developed MRSA infections - saving lives and saving millions of dollars too. Dollars we can spend to reduce CLABSIs or SSIs or another MDRO.