Risk Stratifying the Potentially Septic Patient in the ED

A 75-year old male with no reported past medical history is found in his apartment by his grandson with confusion and generalized weakness…

The Setup: A 75-year old male with no reported past medical history is found in his apartment by his grandson with confusion and generalized weakness. He arrives in your emergency department via EMS with hypotension, tachycardia, and a fever. An infection is undoubtedly a component so you immediately initiate measures to stabilize and further evaluate, while contemplating his ultimate disposition. Since your ICU bed acquisition typically requires hours to days, you think about requesting one immediately following initial evaluation realizing that the Intensivists will resist until further data are available.
Based upon findings available at the initial bedside evaluation, can potentially septic patients be risk-stratified for short-term prognosis?

Your Choices:
(1) Await definitive diagnostic testing before calling the ICU.
(2) Calculate ICU-derived critical care scores such as the APACHE, MPM0-III, or SAPS.
(3) Calculate an EM-derived and validated sepsis prognostic score.
(4) Forget any formulas or prognostic scores, waiting, or hurting your Intensivists’ delicate sensibilities – just pick up the phone and admit them immediately.

The Evidence:

The Bottom Line: Because 40% of ICU patients with sepsis are admitted from the ED, a prognostic screening instrument derived and validated on this unique population could focus resource allocation towards those most likely to benefit. Although multiple instruments have been developed for ICU patients, few have been validated in the ED setting. In the current study, admitted ED patients with SIRS criteria at one of four geographically disparate university hospitals were prospectively recruited. Research assistants, not clinicians, collected the MEDS variables which were not disclosed to or utilized by the treating physician. Among the 385 enrolled patients, 9% died and 11% required mechanical ventilation. The primary outcome was 28-day mortality as ascertained by telephone follow-up or hospital record review.


The MEDS score was derived in 2003 and is a nine-component instrument (Table 1) which has been validated in multiple settings demonstrating increasing mortality as the score increases (Table 2). Additionally, each MEDS variable has demonstrated at least moderate reliability meaning that if two physicians use the MEDS to risk-stratify the same patient they will likely obtain the same score. Finally, when compared head-to-head, MEDS compares favorably to other scoring systems and to serum lactate levels.

The Caveats: While the MEDS score is a Level II Clinical Decision Rule validated for wide-spread use across varied ED settings, no impact analysis has been performed. In other words, in actual clinical practice nobody knows if physicians or patients will accept this rule, appropriately apply and interpret it, or if MEDS will alter patient-important outcomes such as inappropriate, risky therapy or mortality. In fact, recent research has suggested that MEDS may not reliably risk-stratify the sickest septic patients. However, the more severely ill septic patients, defined as fluid-refractory hypotension (MAP < 65 mm Hg) or lactate ≥ 4 mmol/L, are probably not the mystery population we struggle to identify early. Others have retrospectively identified APACHE II, SAPS, and MPM as superior tools. None of these instruments were prospectively derived or validated upon ED patients, though. Future research should combine the MEDS score with clinical gestalt, in addition to an ever-expanding array of sepsis biomarkers (lactate, C-reactive protein) to further assess and perhaps enhance this prognostic tool.

Click on the image to view the high-res PDF



The Outcome: Your patient has persistent hypotension despite your early goal directed therapy saline bolus of 20 mL/kg and you subsequently note hypoxia accompanying a pulmonary infiltrate. Unfortunately, your hospital performs automated CBC so bandemia is never reported, though you suspect he might manifest bands given his WBC 27,000. Nonetheless, his age, persistent hypotension, hypoxia, and confusion still net him a MEDS score of 11. Your critical care colleagues are called immediately after his chest x-ray and blood counts return. Provided his diagnosis and MEDS score, they admit him without argument and he is en route to the ICU within 90-minutes of ED arrival.

Christopher Carpenter, MD, MSc, EPM’s Chief Clinical Editor, is on faculty with BEEM (Best Evidence in Emergency Medicine), and is an assistant professor at Washington University in St. Louis




  1. Mike Klevens on

    As much as I like CDRs and fancy outcome calculators, I’m not a big fan of this one. This guy needed the unit from the beginning. The MD knew it and so did the fair reader. While proper ICU utilization is important, there are sometimes that you send the ICU a borderline case that may not be “worthy”. You send it because you had a feeling. A “blink” moment if you will. And…9 times out of 10…you will be right.

  2. Adan R Atriham on

    I agree with Mike. This MEDS score, and for that matters any other score or decision instrument/rule, if it has more than 4 or 5 variables becomes really unpractical. If the patient looks sick, is tachycardic and hypotensive, then the patient is sick and needs to go to the unit. I personally have problems keeping all this scores and rules fresh in my brain. I would have to read them every other day.

  3. Chris Carpenter on

    Mike & Adam: Good points. I too find Clinical Decision Rules difficult (impossible) to memorize for bedside application. I routinely use PDA applications (like Pepid) to recall the specifics and complement my computational accuracy. For those who do not use handheld devices, the Washington University in St. Louis Journal Club website archives the CDR specifics for each of these decision aids we discuss (5-years and counting). These can be accessed from the web anytime (I use them every shift for resident and patient teaching tools, as well as memory aids for my rapidly aging brain!).

    CDR recall issues aside, though, I strongly believe that the MEDS score has a place in the EP’s diagnostic armamentarium since a) many septic patients presenting to ED’s worldwide are not receiving guideline directed care because they are not being recognized (Am J Emerg Med 2006; 24(5): 553-559)so MEDS may help raise awareness of these patients, particularly during lengthy ED length-of-stay patients; b) the sheer volume of septic patients we will encounter will continue to increase with the aging baby-boomers’ demographic tsunami while the number of ICU beds remains static (Ann Emerg Med 2006; 48(3): 326-331) so we will need to carefully allocate those few beds we do have available using best-evidence risk-stratification to do so. Additionally, we will need to ensure that future interventional trials are comparing apples to apples in treating septic patients so the MEDS score (or other sepsis-related prognostic tools) will help us to compare one patient population to another from trial-to-trial.

    Great commentary. Keep it coming. Thanks!

    Chris Carpenter, MD, MSc
    EP Monthly Chief Clinical Editor
    Washington University in St. Louis School of Medicine
    Director, Evidence Based Medicine
    Pepid, Senior Editor

Leave A Reply