SIRS Signs: Shoddy Sensitivity in Severe Sepsis?
And I knew there was more to this story. (Chris, unlike me apparently, is rarely a pure internet troll.) A followup tweet from him sent me to this NEJM study suggesting that SIRS criteria are actually neither sensitive nor specific for sepsis (or severe sepsis). They looked at a ton of ICU patients and saw who had ≥2 SIRS critiera and ended up having an infectious etiology. The really brief conclusion is this: in this study, about 12% of patients who ended up having an infectious disease etiology for their illness didn’t have 2 or more SIRS criteria. (We usually criticize these criteria for being way too sensitive. Example: If I go to the gym and exercise, I have the systemic inflammatory response syndrome. If I go to the gym and exercise while I have a cold, I have sepsis.)
Today, the wonderful EM Literature of Note wrote his take on the study as well:
So, the traditional SIRS-criteria definition of severe sepsis, previously thought to have at least sensitivity at expense of specificity will miss 1 in 8 patients with organ failure and an underlying infection. Considering only approximately 1/3rd of patients with two or more SIRS criteria in the Emergency Department have an underlying infection, the utility of these criteria is substantially less reliable than previously thought. Sadly, I’m certain many of you are suffering under SIRS criteria-based alerts in your Electronic Health Record — and, if such alerts are introducing cognitive biases by decreased vigilance and alert fatigue, it ought to be obvious we’re simply harming ourselves and patients.
I have to say, however, that I took a slightly different take on this. Maybe I’m wrong — and I definitely see septic patients without 2 or more SIRS criteria — but it doesn’t seem to be 1 in 8. (I’d also been wanting an excuse to try out Medium after @nickgenes did, so I thought I’d give my own take.) My thoughts on the study:
- SIRS+ patients (those that had ≥2 SIRS criteria and were infected) were younger and sicker. This means a few things to me. First, I’m not at all surprised. Young people who are septic will come in screaming “I’M SEPTIC!” at you. They’re usually terribly sick when they’re septic (because they’re rarely septic, because they’re young). Second, older people are more likely to have “occult sepsis,” which I guess I’ll identify as the patient that you order a lactate on but have pretty good looking vitals (maybe 1 SIRS criteria, a fever or something) and you’re like, “Wow, really, it’s 6? Well, okay then.” Older people also are more likely to be beta blocked, and are also less likely to mount a “true” fever of 100.4 or greater. We all know this: Old people will burn you.
- There’s nothing I love more than ranting about vital signs; we’ll start with the respiratory rate. Now, these patients were in an ICU and likely intubated or had tons of monitoring on, so one would hope that there were accurate respiratory rates being documented, but even the third year medical student pre-rounding collecting vitals thinks it’s strange than every single patient’s respiratory rate overnight was “18" or “20.” 70% of patients had an abnormal respiratory rate — the same number than had an abnormal white blood cell count — so perhaps this wasn’t the case in the ICU. And as the self-proclaimed King of the Booty Temp(tm), I also have to wonder about how these temperatures were taken. We all know that core temperatures (rectal, temperature-sensing foley) are more accurate but rarely performed.
- I wonder how many patients received or had received antipyretics, IV fluids, or early antibiotics that might have changed their vital signs.
It does seem like “elevated lactate = badness” in a number of different disease processes, but I do worry that what ends up happening is “elevated lactate = sepsis.” It’s a similar reflexive “elevated troponin = MI” bit. Arguably both abnormal lab tests might make an emergency physician lean toward admission in a patient they otherwise might have discharged or not given a second thought to, but we’ve all been guilty of searching for the critical infection due to an elevated lactate when the patient’s actually having a massive GI bleed, or a tylenol overdose, or shock liver.