End-Tidal CO2 in the Diagnosis of Pulmonary Embolism


The content you’re about to read or listen to is at least two years old, which means evidence and guidelines may have changed since it was originally published. This content item won’t be edited but there will be a newer version published if warranted. Check the new publications and curriculum map for updates

Author: Anu Mitra / Codes: A2 / Published: 05/11/2014

Riaz I, Jacob B., Pulmonary embolism in Bradford, UK: role of end-tidal CO2 as a screening tool., Clin Med. 2014 Apr;14(2)

  • Objective: To evaluate whether EtCO2 measurement can be used at the bedside to exclude pulmonary embolism (PE).
  • Design: This was a prospective diagnostic study set in a British acute medical admissions ward. Patients suspected of having a PE ie., those with CT pulmonary angiogram (CTPA) were enrolled.
  • Tests: The reference test was CTPA. The index test was an oral hand-held capnopgraph. Patients also had Wells scoring done and D-dimers taken in those patients for whom it was indicated. The researcher conducting the capnography was blinded to the CTPA findings.
  • Results: 100 patients were enrolled, of whom 38 turned out to have a PE. At cutoff of 450mcg/L D-dimer had sensitivity of 96%, specificity 53%. At a cutoff of 4.3kPa, EtCO2 had a sensitivity of 100%, specificity 68%
  • Strengths: A clear, unambiguous design and protocol using a recognised reference standard. There was blinding of the researcher.
  • Limitations: Validity patients werent blinded. EtCO2 measured up to 24 hours after admission: ?attrition effect after anticoagulation started. The patients had already been through a selection process ie., Wells/D-dimer, to be referred for CTPA. No sample size calculation.
    Generalisability: This is a very different population from the many undifferentiated chest pain/dyspnoea/haemoptysis/collapse cases we see in the ED. This is a higher prevalence population. Positive predictive value would be even lower in the lower prevalence population, although negative predictive value would remain high. Also question marks remain about the reliability of the test itself. Is it accurate if nasal prongs arent used?
  • Summary: This is a well-designed paper. Results are interesting and have reasonable validity for an acute medical admissions population. We think this paper demonstrated proof of concept.
  • Will it change our practice? As Emergency Physicians we dont think this has quite enough external validity what I call the so what factor for us to adopt it wholesale. It may have a place as a quick rule-out test at triage prior to deciding upon doing Wells/D-dimer, and may be more effective than the PERC score as it is not dependent upon age. We would recommend a similar study performed on our undifferentiated ED population.

Podcast Notes

Anu Mitra Catherine Gant
Talking about Riaz and Jacob
Background: Pulmonary Embolism Keeps Emergency Physicians Up At Night
  • PE is not uncommon (60-70/100,000), compare with sepsis (66), AMI in men (154)
  • PE is often fatal. It is a medical emergency.
  • PE can be a diagnostic quandary
  • There are conflicting pressures:
    • Importance of rule-out: we dont want to be sending people with pathology
    • Limited resources: we dont have the resources to admit and test everyone
  • We all know the problem with D-dimers
  • PERC has helped although with our ageing population it may apply less and less with time
  • PE is an example of V-Q mismatch
  • Low V-Q – shunt: ACPE, asthma
  • High V-Q alveolar deadspace: PE, COPD
  • Poor perfusion
  • Poor gas exchange
  • Low EtCO2
So this sounds like a very important study then! Shall we go through it?
Objective: To evaluate whether EtCO2 can be used to exclude PE at the bedside

  • Prospective, diagnostic study


  • Adults in AMU or ward in an English NHS hospital
  • These were patients referred to the medics suspected of having PE
    • High Wells
    • Low Wells and D-dimer
  • ie., all those having a CTPA (which is how they got their patients)
Reference test

  • CTPA

Index test

  • EtCO2 measured by oral capnography
    • Nostrils not clipped
    • 10s normal resps
    • 3 readings
  • Wells Score
  • D-dimer
    • In those in whom it was done

  • Researcher took EtCO2 readings
  • Blinded to CTPA result
  • Within 24 of onset of symptoms
  • Wells also calculated if not already done so
  • D-dimer results recorded

  • 100 patients (age 18-93)
    • 38 had PE, 62 had no PE
  • Wells Score:
    • PE: 2.93
    • Non-PE: 2.7
    • AUC: 0.52
    • Is this relevant? is Wells a diagnostic test
  • D-dimer:
    • 64/100 had done
    • Cutoff >275 mcg/L
      • Sens 100%, spec 20%
    • Cutoff >450 mcg/L
      • Sens 96%, spec 53%
    • Mean for PE pts: 1855; non-PE: 912
    • AUC: 0.82
  • EtCO2:
    • Ran the test performance using different cut-off figures
    • Cutoff 3 kPa
      • Sens 16%, spec 89%
    • Cutoff 4.3kPa
      • Sens 100%, spec 68%
    • Cutoff 6 kPa
      • Sens 100%, spec 8%
    • For cutoff 4.3 kPa
      • PPV 66%, NPV 100%


  • EtCO2 measurement is a quick, safe, non-invasive way of excluding PE.

  • Clear, unambiguous objective
  • Simple design
  • Relevant pathology (PE)
  • Tests are those we use and understand
  • Good reference standard
    • Unrelated to the index standard
  • Measurement clear
  • Validated EtCO2 capnograph
  • Appropriate statistical analysis
Limitations Internal Validity

  • Sample: patients already selected to have CTPA
    • Already had risk stratification (Wells, D-dimer)
  • Patients not blinded to the CTPA result
  • Delay in measuring EtCO2
    • Effects of LMWH/heparin after 24h
  • Sample size calculation not done
External Validity

  • Population: AMU/ inpatient
  • High prevalence population
  • Not undifferentiated ED
  • Sensitivity and specificity do NOT depend on prevalence of the disease
  • PPV and NPV DO depend on prevalence:
    • PPV prevalence
    • NPV 1/ prevalence
  • What does this mean in the ED suspected-PE population?
    • PPV even lower
    • NPV still high
  • Population: AMU/ inpatient
  • High prevalence population
  • Not undifferentiated ED
  • EtCO2 equipment?
    • Oral
    • ?nasal leak
    • Cant use cutoff of 4.3 kPa as different method of sampling to CXH ED
  • Delay in measuring EtCO2
    • No delay with ED patients

  • Interesting concept
  • Proof of concept (in this population)?
  • Very difficult to justify translating to the ED population
  • Cant use 4.3 kPa cut-off either
  • Suggestions for further study?
Will It Change Our Practice?


Leave a Reply