Journal Logo



Are Phenotypes a Magic Bullet for Sepsis?

Walker, Graham MD

Emergency Medicine News: May 2020 - Volume 42 - Issue 5 - p 15
doi: 10.1097/01.EEM.0000666316.57389.fe
    sepsis, phenotypes

    There's a constant battle in medicine between practice styles—standardization and customization. Standardization is order sets and doing the same thing for every patient every time because it (hopefully) delivers better care, either by being faster, less error-prone, or less costly.

    I don't think anyone questions standardization of care—timeouts before surgery, ECGs for chest pain, order sets for STEMI care. But critics quickly point to the opposite approach, customization of care. What if that STEMI is actually a GI bleed causing critical ischemia? Shouldn't we as physicians take a history, practice medicine, and customize the care to the individual patient? Can't robots do our job if we're only practicing standardized care?

    Nowhere in emergency medicine can this tension be felt more acutely than in sepsis care. I have strongly criticized Sepsis-3 and the proposed one-hour bundle in the past because we can't agree on a definition of sepsis except in retrospect. I feel like a robot when delivering sepsis care, mostly due to regulatory requirements (thanks, standardization) mandating early antibiotics, blood cultures, and aggressive fluids for anyone who might be septic (again, whatever that actually means).

    Patients with true septic shock who are critically ill are certainly deserving of this aggressive care, but a lactate of 2.1 mmol/L in the already fluid-overloaded patient is the bane of our combined existence as emergency physicians. The 30 mL/kg fluid order almost always makes me roll my eyes and let out an audible, not-under-my-breath sigh because I feel my arm is so twisted by the vast quality, metric, and regulatory industrial complex that all hospitals have had to develop over the past 20 years.

    But this will not be another rant about sepsis quality metrics. It will be about hope. Hope for personalized sepsis care. And I found that hope on Twitter.

    A New Hope

    I recently clicked on a curious article suggesting there might be different phenotypes of sepsis. This was not from some clickbait crackpot BuzzFeed article; it was from one of the leading sepsis researchers in the world, who conducted a retrospective analysis of data on 63,858 patients using machine-learning techniques.

    Chris Seymour, MD, MSc, an associate professor of critical care and emergency medicine at the University of Pittsburgh School of Medicine, published this article after similarly struggling to figure out how to deliver better sepsis care. (JAMA. 2019;321[20]:2003; Sepsis carries a high risk of death and hospitalization worldwide, but we still grapple with finding a magic bullet for sepsis. Dr. Seymour wanted to know why sepsis trials fail.

    The hypothesis behind Dr. Seymour's work is interesting and thought-provoking, and I think it will resonate with emergency physicians. He analyzed a ton of septic patient variables and tried to determine if patients could be grouped or clustered based on the abnormalities or problems they had. His analysis found four separate phenotypes that tended to have different clustered characteristics. He called them alpha, beta, gamma, and delta.

    The exact characteristics of each phenotype are not particularly important for this article; it's more just the idea that they might exist. Patients with the alpha phenotype, for example, tended to have more neurological, pulmonary, and hematologic changes, while patients with the beta phenotype tended to have more renal and inflammatory changes.

    Outcomes were also markedly different among the groups. Let's just say you do not want the delta phenotype, which had the highest rate of ICU admission and mortality (85% and 32%, respectively), compared with the alpha phenotype's mere 25 percent ICU admission rate and two percent mortality rate. I'll take the 16-fold lower death rate, please.

    I'm not sure if I see these exact subtypes play out in my day-to-day practice of emergency medicine, but I certainly see different patient presentations that we all call sepsis:

    • There are the patients who look much worse than their numbers; they eyeball poorly, you're waiting for their blood pressure to tank, and you're surprised their lactate is only 2.7 mmol/L.
    • Then there are the patients whose numbers look much worse than they do, like the occult sepsis little old lady who is smiling and telling you about her dog and her travels to Italy and has normal vitals besides a mildly elevated temperature and a lactate of 6 mmol/L.
    • There are also the slam-dunk-obviously-septic-shock patients who you're preparing to intubate upon arrival while hoping they don't code on you.
    • Or the patients with a lactate of 3 mmol/L that just won't clear or is slowly trending up by their sixth hour boarding in your ED.

    I was particularly excited to learn about these phenotypes. Could it be true? Could we start treating those with the alpha phenotype differently from those with the beta phenotype? I had a fascinating conversation with Dr. Seymour, which you'll find here next month.

    Share this article on Twitter and Facebook.

    Access the links in EMN by reading this on our website,

    Comments? Write to us at

    Dr. Walkeris an emergency physician at Kaiser San Francisco. He is the developer and co-creator of MDCalc (, a medical calculator for clinical scores, equations, and risk stratifications, which also has an app (, and The NNT (, a number-needed-to-treat tool to communicate benefit and harm. Follow him on Twitter @grahamwalker, and read his past columns at

    Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.