Can a Patient with Sepsis Be Discharged from the ED?

By Erica Remer, MD, CCDS

There has been a lively debate on LinkedIn regarding an article which came out in JAMA Open Network February 10, 2022 called Prevalence, Characteristics, and Outcomes of Emergency Department Discharge Among Patients With Sepsis, by Peltan, et. al., instigated by Dr. Ronald Hirsch and stoked by me. Dr. Cesar Limjoco posted a meme that says, “Clinical Truth: One dose of antibiotics in the ED resolves sepsis only when it isn’t sepsis,” attributed to Dr. Ryan Greiner who is a System Medical Director and Physician Advisor, and I wholeheartedly agreed. 

Ron contacted the author of the original article informing them of our discussion, and we were referred back to a commentary which had been published with the manuscript, entitled, Elucidating the Spectrum of Disease Severity Encompassed by Sepsis, by C. Rhee and M. Klompas (both MDs and MPHs). I strongly encourage you all to read both the original article and this companion piece. 

The original article stated that 16% of patients who “met sepsis criteria” were discharged from the ED, and 66% of those had urinary tract infections. The 30-day mortality for this cohort was 0.9% as opposed to 8.3% for patients admitted to the hospital. I was unable to ascertain which organ dysfunction was present in this group of patients (e.g., infection correlated with organ dysfunction). There is a graphic in the supplemental materials which demonstrates that patients who were admitted had higher increases in SOFA components than those discharged. 

The article which elicited this commentary starts off referring to it as "sepsis syndrome." Although they are probably hailing back to the days when we thought there was sepsis, severe sepsis, and septic shock on a spectrum, there is no way to code "sepsis syndrome." I refer to sepsis as the condition formerly known as "severe sepsis." All diagnoses of sepsis now should have an additional code of R65.20/1, (Severe sepsis without/with shock), which can only be captured if the documentation supports it. 

Coders are permitted to capture "severe sepsis" if the clinician specifies that the organ dysfunction is sepsis related. The macro that I recommend is: Sepsis due to [infection] with acute sepsis-related organ dysfunction as evidenced by [organ dysfunction/s]. This serves multiple purposes. First, it ensures there is a bona fide infection. Second, the patient must have organ dysfunction (not just deranged vital signs). Finally, the linkage of acute sepsis-related with organ dysfunction gives the coder permission to pick up R65.20. 

I think this also highlights the problem with this study. The way they identified patients is if body fluid cultures were obtained and antibiotics were administered, indicating an infection (prophylaxis and empiric antibiotic administration were excluded). Then, they reviewed the chart and determined if there was organ dysfunction as per SOFA compared to baseline according to labs. One of the authors told me in a personal communication that, “We did not use discharge diagnosis or clinical documentation.” I am relieved that they used organ dysfunction as opposed to SIRS, however! 

The Third International Consensus (Sepsis-3) clearly stipulates that there are no gold-standard diagnostic tests, and that sepsis is a clinical diagnosis. Organ dysfunction must be associated with the sepsis. Underlying infection (e.g., pneumonia causing hypoxemia), dehydration, other conditions like Gilbert’s disease can cause derangement of the components of the SOFA score. Infection plus organ dysfunction isn’t sepsis; infection plus sepsis-related organ dysfunction is. If the provider hasn’t made the call of sepsis and the researcher isn’t reviewing the clinical documentation, then how do they know whether the organ dysfunction was sepsis-related or not? 

We old-timers know that inherent to the diagnosis of sepsis, which is a clinical diagnosis without any gold standard diagnostic criteria, is the fact that the patient is SICK. Sepsis-2 was an attempt to facilitate clinicians to not miss the diagnosis of sepsis because it has a high mortality (just the disparate 30-day mortality rates, 0.9% vs. 8.3%, makes one question whether the discharged population really had sepsis). It was determined that too wide a net was being cast, so Sepsis-3 tried to identify the essence of what indicated a patient was SICK, hence, mandating acute organ dysfunction. When I teach, I explain that "sepsis without organ dysfunction" is called... "pneumonia" or "UTI" or "cellulitis." 

The problem with data is that the output is only as good as the person drawing the conclusion. Is the clinician ticking SOFA boxes without considering whether a dysregulated host response to infection caused the derangement? Is hypoxemia from the pneumonia itself or is it sepsis-related? Is a bump in creatinine in a UTI from the localized infection, dehydration, or from resultant sepsis? 

I think this study really demonstrated that patients with infection who are sent home from the emergency department are not as sick as patients who are admitted to the hospital. That’s good. I agree with the commentators that “the most important reason to call an infection sepsis is to emphasize the severity of the condition and trigger immediate aggressive care.” 

Perhaps we would benefit from reintroducing the criterion for a patient to appear SICK in order to consider a diagnosis of sepsis. I don't believe that we want to impose a threshold duration of organ dysfunction, as has been suggested. We may benefit from insisting there needs to be organ dysfunction not involving the infected organ/system.

There are payors who won't acknowledge the possibility that sepsis can be placed in Observation status. How common is it really (or should it be) to have a patient with genuine sepsis discharged to home from the ED? It should be about as common as it is to bill Critical Care Time in the ED for a discharged patient - it happens, but it is rare. 

I agree with Dr. Ryan Greiner, a system medical director and physician advisor, who said, "One dose of antibiotics in the ED resolves sepsis only when it isn't sepsis."