Shorr Score for MRSA Pneumonia

Why is it so difficult to nail down a practical and useful definition for HCAP; i.e., one that can reliably be used to make clinical decisions?

I think the issue with the concept of healthcare-associated pneumonia is that everybody is concerned about resistance in pneumonia that presents to the hospital, but everyone faces a different pre-test probability. For example, in the United Kingdom, where there is no MRSA, your pre-test probability for MRSA in pneumonia patients who come to the hospital is essentially zero. So of course you’re not going to agree on what the score should be, because you don’t think it’s a problem in the first place. But in the United States, where MRSA is much more prevalent, and you’re taking patients from nursing homes or patients who’ve been recently hospitalized, you’ve got to sort out how to do that.

It’s the classic dilemma of “where you sit determines what where you stand”, and so a lot of the divergence about the concept of HCAP is based on what you perceive as the probable pathogens in those patients. And if you look at the most recent guidelines on nosocomial pneumonia from the Infectious Disease Society of America and the American Thoracic Society, they’ve really abandoned the concept of HCAP, because it can lead to too much broad and indiscriminate use of antibiotics, but they say you should adopt the concept of looking at patients presenting to the hospital with risk for resistant pathogens. Which, if you think about it, is how the concept of HCAP came to be in the first place, and strikes me as sort of a false dichotomy. No one’s saying that these pathogens don’t exist; we’re just trying to figure out ways to more precisely define who these patients are so we don’t overtreat unnecessarily, but more importantly, so we don’t miss patients who have these pathogens and give them antibiotics that are not going to be active, and expose them to a higher risk of death.

What pearls or pitfalls do you have for using the Shorr Score? Specifically, do you know of any cases when it’s been applied or interpreted inappropriately?

I don’t know of many cases where it’s been applied inappropriately, but the pitfall is to understand is that we validated the score in a population where there was an intermediate pre-test probability of MRSA as a pathogen. A percentage of the patients who showed up to our ICU were that sick and had MRSA pneumonia—it wasn’t 100% or 80% or even 50%, but it wasn’t 2% either. The point is, it was a moderate pre-test probability population, which is really where you need these tools to help you discriminate. At the same time, they’re validated in a population that’s very sick, where if you miss giving the antibiotic, there’s a price to pay. We’re not talking about this for a low-risk skin infection, where you have other choices, and if it takes you a day to figure it out, there’s no mortality associated with it. So it’s important to remember that we’re talking about a specific disease state in a specific risk population, with a specific pre-test probability.

What does the Shorr Score add over clinician gestalt in deciding to start antibiotics?

I think it adds a lot, because at this point, clinician gestalt is to just treat everyone for everything. And that leads to antibiotic resistance, side effects, cost—in most U.S. hospitals right now, if you’re sick they will reflexively give you piperacillin-tazobactam and vancomycin. My joke is, that’s the jello on the tray for the patient. We can’t do that anymore. It’s how we got into this mess in the first place.

Are there any adjustments or updates you would make to the score based on new data or practice changes?

The only update is that the score is now more refined in terms of looking at a broader array of pathogens that might be resistant instead of just MRSA. There are still people who come to the hospital with pure community-acquired pneumonia, and all they need is cefTRIAXone and azithromycin. They don’t need piperacillin-tazobactam or imipenem and vancomycin. So I think that’s been the biggest refinement—validating the risk score more broadly as a tool for looking at resistant infections generally, and doing it in a completely different data set.

The score uses ICU admission as a dichotomous variable to represent severity of illness. How did you choose this, and was there any consideration given to validated scores like the PSI Score?

We picked ICU admission—it’s certainly true that what gets you into my ICU might not be what gets you into your ICU, but we picked it because it’s easy for a clinician to apply immediately. To use a score where you have to calculate another score is not very useful. But you pretty much know if a patient’s going to the ICU or not within 30 minutes of evaluating them. We did look at other tools for severity, but they didn’t provide any better resolution. Those risk scores are co-linear: higher PSI Score, higher likelihood of going to the ICU. Because it didn’t add anything to the score, and actually made the score more cumbersome, we went with the simplest approach that didn’t sacrifice any accuracy.

How do you use the Shorr Score in your own clinical practice?

It’s very limited, because I only see the patients who come to the ICU. If you’re in the ICU for severe pneumonia, our pre-test probability for MRSA is already 20-30%, so we’re going to treat you empirically if you come to the ICU. It’s really a tool for the ER prescriber or for patients on the floor, so I can’t really comment on how it’s used outside the ICU.  

What other research are you working on at the moment that you’re particularly excited about?

We’re looking at some outcomes data to help people understand what the burden is of missing a severe resistant infection and not giving it the right antibiotic up front, and helping people understand that perhaps the most expensive infection out there is not the one that we treat with the most expensive antibiotic, but the one that’s treated with an antibiotic that doesn’t kill the bug. If you have MRSA pneumonia and you give them cefTRIAXone, great, you used an inexpensive antibiotic. But if the patient survives, they’re now going to take a long time to get better, and that’s going to make their length of stay higher and their chance of being readmitted higher. That may be a logical mentality from the budget standpoint, but it’s bad for the system.

Andrew Shorr, MD, MPH, is an associate director of pulmonary and critical care medicine and the chief of the Pulmonary Clinic at MedStar Washington Hospital Center in Washington, DC. Dr. Shorr's research interests include resistant pathogens and healthcare-associated bacteremia, and he has published more than 140 original studies.