The difficulty in making a diagnosis of sepsis

Question 27 from the first paper of 2007 stumped the candidates by asking them why it might be difficult to make the clinical or laboratory diagnosis of sepsis in the critically ill patient. One might surmise that many of them have never experienced any difficulty in making that diagnosis. LITFL has an excellent page on this topic which is enough for most people. If one were in need of a literature reference, the one best article would probably be this 2006 Belgian paper by Vandijck Decruyenaere and Blot.

In brief, the aforementioned difficulties can be summarised thus:

  • Old diagnostic SIRS/sepsis criteria were non-specific for infection
  • New "Sepsis 3" criteria also have methodological flaws, eg. they rely on SOFA scoring
  • Infectious and non-infectious causes of shock may coexist
  • Source of infection is hard to find (no history in the comatose patient, clinical findings obscured by critical illness)
  • There are no specific clinical signs in sepsis
  • Critically ill malnourished elderly or immunocomromised patients do not manifest typical signs of sepsis 
  • The laboratory markers of sepsis are non-specific (eg. CRP, WCC, even procalcitonin)
  • Serology, PCRs and cultures take many hours to process
  • Antibiotics obscure culture results
  • Imaging may be non-specific (eg. infected vs. sterile collection)

In other words, our definitions are crap, our tests suck, and our patients are too complex to be pidgeonholed into narrow diagnostic categories.

More detail and specific criticims of the disgnostic criteria can be found in these specific chapters:

The purpose of this chapter is to help candidates answer questions along the line of "discuss the difficulty in making the diagnosis of sepsis", which requires three major sections:

  • Difficulty related to applying the old SIRS/sepsi criteria
  • Difficulty related to applying the new Sepsis 3 criteria
  • Difficulty unrelated to diagnostic criteria

Application of defunct sepsis/SIRS criteria

Why would these criteria be difficult to apply to the diagnosis of sepsis in the ICU?

To go through them systematically:

  • These are all nonspecific features; in scientific-sounding words, the specificity of the SIRS diagnostic criteria is poor for SIRS, and especially poor for infective SIRS.
  • The temperature, heart rate resp rate WCC and so forth can all be increased because of any number of physiological disturbances, many of which are neither infectious nor inflammatory.
    • Not all infected patients are febrile. 10% are hypothermic and 35% are normothermic.
    • Tachyardia is completely empty of any SIRS-predictive information, as it can represent hypovolemia, anxiety, pain, or inotrope effect. Plus you can mask it with β-blockers.
    • Tachypnoea is closely related to metabolic acidosis, but metabolic acidosis does not reliably predict SIRS or sepsis.
  • Some patient groups will not be able to manifest many of these (or any of these):
    • Malnourished patients
    • Immunocompromised
    • Therapeutically cooled
    • Elderly patients
    • Hypothyroid patients
  • Infectious and non-infectious causes of SIRS may co-exist

The savvy candidate will have ignored much of this, as it is a discussion of outdated parameters, a definition which has now been rendered obsolete.

Difficulties unrelated to diagnostic criteria

The elusive nature of the infectious source

  • The patient is sedated and cannot participate in history-taking.
  • There are no generic features of 'infection" to look for.
  • Specific features of infection may be obscured by other clinical findings (eg. is this SIRS due to an infected pancreatic pseudocyst, or merely a non-infected one?)
  • Criteria for the diagnosis of an infectious source are frequently not available (eg. central line associated bacteraemia, ventilator-associated pneumonia, etc).

The laboratory markers of sepsis are non-specific

  • All biochemical markers of sepsis can become elevated due to non-infectious causes.
    • Not only that, but they can be falsely decreased in the presence of severe infection; for instance, CRP may be low if the patient has cirrhosis; or they may have no WCC rise because they are neutropenic.
  • Serology tests and PCR studies may be inconvenient, because:
    • They may not be organism-specific
    • They may require convalescent samples
  • The gold standard for infection is a positive culture, but:
    • Frequently the infected tissue or body fluid cannot be easily obtained for culture.
    • Even if you get a sample, you may not grow the organism in a culture.
    • You may have already given the correct antibiotics, which have sterilised the sample before it was collected.
    • Whichever organism you grow may not be the responsible organism anyway.
      • A positive sample culture may represent colonisation rather than infection
    • In any case, cultures take several days to grow, whereas the events in severe septic shock unfold over the course of hours.


Oh's Manual is listed as one of the references; however the sepsis chapter never answers this actual question, unhelpfully remarking that "the initial presentation of severe sepsis and septic shock is often non-specific and ‘cryptic’..."

Levy, Mitchell M., et al. "2001 sccm/esicm/accp/ats/sis international sepsis definitions conference." Intensive care medicine 29.4 (2003): 530-538.

Vandijck, D. M., J. M. Decruyenaere, and S. I. Blot. "The value of sepsis definitions in daily ICU-practice." Acta Clinica Belgica 61.5 (2006): 220-226.