Is SIRS really that bad?

One of the major headlines that accompanied sepsis 3 last week was that SIRS misses 1 out of 8 patients with sepsis in the ICU. This fact was touted as a major reason to abandon SIRS in favour of SOFA. I think it is worthwhile taking a closer look at that data. The reference is:

Kaukonen KM, Bailey M, Pilcher D, Cooper DJ, Bellomo R. Systemic inflammatory response syndrome criteria in defining severe sepsis. The New England journal of medicine. 372(17):1629-38. 2015. PMID: 25776936 [free full text]

This is a large retrospective study that looked at a database that covers more than 90% of the ICUs in Australia and New Zealand. Of all ICU admissions, they identified a cohort of 109,663 that were coded as having an infection and also met their definition of organ failure – which is the now familiar score of 2 or more on the SOFA score. Of these patients, 88% were SIRS positive, and 12% were SIRS negative. Or, as it has been widely reported, ‘SIRS missed 1 in 8 ICU patients with infection and organ failure’.

Were any of these really misses?

These patients were all identified in a database because they were coded as “sepsis”. So the clinicians looking after these patients knew they all had sepsis. They were in the ICU being treated for sepsis. Clinicians used clinical judgement and were able to make that diagnosis even though they didn’t fulfill all the SIRS criteria. So although the authors of this paper, retrospectively and rigidly applying SIRS criteria, would classify these patients as SIRS misses, the clinicians at the bedside actually diagnosed them all.

Furthermore, you might ask what is the gold standard that they were being compared to? There is no gold standard test for sepsis. So what these authors compared SIRS to was SOFA. However, as discussed in my previous post on sepsis 3.0, there is little reason to think SOFA is any better than SIRS. (Also see the excellent post by Josh Farkas on PulmCrit.) You could just as easily have decided SIRS was the gold standard, and then compared how many patients SOFA missed.

What were their outcomes?

If SIRS negative patients were less obvious, and therefore diagnosed later, that might lead to worse outcomes. So what were the outcomes in the two groups? If you were SIRS positive, your mortality was 24.5%, whereas if you were SIRS negative, your mortality was 16.1% (p<0.001). So SIRS negative patients do better. SIRS negative patients also had a shorter ICU stay (85 vs 57 hours, p<0.001), shorter hospital stay (13 vs 11 days, p<0.001), and higher rate of discharge to their own home (54% vs 62%, p<0.001).

What is SIRS supposed to do for us in infectious patients? It is supposed to help us predict which patients are more likely to die. It did exactly that in this cohort.

Finally, the mortality of both SIRS positive and negative patients decreased over time, which is further evidence that the current management of sepsis, based on the traditional SIRS definitions, is working.

So what about this 1 in 8?

You could play this off as SIRS missing 1 in 8 (against a faulty gold standard). But you could also point out that this would translate into a sensitivity of 88%, which doesn’t sounds so bad. But what you should really focus on is that all of these patients were in the ICU. They were all diagnosed and treated. The only thing that these authors have done is identify a sepsis subgroup – SIRS negative patients – that has better outcomes.

This paper in no way supports a change to SOFA. If anything, the trend demonstrating a decreasing mortality over time supports sticking with our current – SIRS based – sepsis management.

References

Kaukonen KM, Bailey M, Pilcher D, Cooper DJ, Bellomo R. Systemic inflammatory response syndrome criteria in defining severe sepsis. The New England journal of medicine. 372(17):1629-38. 2015. PMID: 25776936 [free full text]

Singer M, Deutschman CS, Seymour CW. The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA. 315(8):801-10. 2016. PMID: 26903338[free full text]