Morbidity and mortality of the self-poisoned patient is often related to respiratory complications resulting from the ingested drugs.  As a result, these patients may require intensive care and monitoring.  The ability to predict the clinical course of a patient would be helpful to potentially prevent subsequent complications.

Currently, there does not exist a  clinical measurement that has been shown to be helpful in outcome prediction in the self-poisoned patient. Previous studies have looked at various scales of consciousness (e.g. GCS) with poor predictive value. Noninvasive end tidal CO2 (ETCO2) has been helpful in monitoring and detecting early ventilatory depression in iatrogenically sedated patients when patients have ETCO2 greater than 50mmHg or more than 10mm Hg increase from baseline.  This is the first prospective study looking at the effectiveness of hypercapnia predicting early respiratory depression in self-poisoned patients.

Clinical Question

What is the predictive value of capnography (ETCO2 > 50mmHg) in detecting early complications in the self-poisoned patient?


Self-poisoned patients ages > 18 years old admitted to the ED. Patients must have ingested at least an opioid, sedative/hypnotic, antipsychotic or antidepressant drugs.


Noninvasive capnography via an oral-nasal Capnostream device


Primary: Sensitivity and specificity of >/= 50 mmHg ETCO2 to detect “early complications” in the self-poisoned patient (1. Hypoxia requiring > 3L NC to maintain O2 > 94% 2. Bradypnea (< 10breaths/min) 3. Intubation requiring ICU admission 4. Ventilator failure requiring administration of antidote) Secondary: Ability of ETCO2 to predict early complications at various time points after ED admission: 1min, mean and max during first 10 and 30 min as well as change of > 10mmHg from baseline. Comparison between capnography and “consciousness scales” including GCS, Alert, Verbal, Pain, Unresponsive scale and the Ramsay score in their effectiveness to predict early complications in the self-poisoned patient. Correlation between ETCO2 and arterial CO2.


Blinded, single center, prospective, convenience sample


Patients not requiring monitoring and admitted directly to psychiatry, patients who were completely unresponsive or were in respiratory arrest or had significant respiratory depression requiring immediate intervention on arrival. Also excluded were patients who were pregnant.

Primary Results

Critical Results

  • Primary Outcome (ETCO2 >/= 50mmHg at any time to predict early complication)
    • Sensitivity: 46% (95% CI 29 to 63%)
    • Specificity: 80% (95% CI 73% to 86%)
    • Positive predictive value: 33% (95% CI 20% to 48%)
    • Negative predictive value: 88% (95% CI 81% to 92%)
  • Secondary Measures
    • Absolute change of > 10mmHg from baseline sensitivity 20% (95% CI 9% to 37%) and specificity 68% (95% CI 60% to 75%)
    • Maximum ETCO2 within 10 min of inclusion had best predictive value for early complications
    • Consciousness scales had better predictive value than ETCO2 >/= 50mmHg
    • The median PaCO2 was not significantly different between patients experiencing early complications vs. patient not experiencing complications
    • The median ETCO2 was not significantly different between patients experiencing early complications vs. patients not experiencing complications


  • First prospective blinded study in the self-poisoned patient
  • Patient centered study


  • Single center study
  • Significance of low and decreasing ETCO2 levels were not analyzed even though almost all patients experienced this phenomenon and the authors attributed hypopnea to be the most common type of drug induced hypoventilation in their population.
  • All patients received 1L NC at baseline per the Ethics Committee Guidelines
  • No consideration given to commonly ingested poisons e.g. salicylates in the self-poisoned patient that also significantly alter respiratory status

Author's Conclusions

“Capnometry in isolation does not provide adequate prediction of early complications in self-poisoned patients referred to the ED. A dynamic minute-by-minute assessment of ETCO2 could be more predictive.”

Our Conclusions

This is a small single center study that looked at hypercapnea’s ability to predict clinical outcome in the self-poisoned patient. Elevation in ETCO2 alone is inadequate to predict ventilatory compromise in the self-poisoned patient. There are 2 main types of central hypoventilation: bradypnea and hypopnea. They have opposite effects on ETCO2. It is important to differentiate the type of hypoventilation that the patient is manifesting and monitor accordingly in combination with patient’s overall mental and clinical status.

Potential Impact To Current Practice

Capnometry has been the standard of care for monitor of patients with the potential for respiratory decompensation. Though most practitioners, whether consciously or unconsciously, also take into consideration the patient’s overall mental status. This study has too many limitations to condemn capnometry as a whole especially when we do not know the significance of low ETCO2 in clinical outcome. From the article it does seem that hypercapnea may not be very predictive of complications of the self-poisoned patient, but will unlikely change the application of capnometry.

Bottom Line

ETCO2 > 50mmHg alone poorly predicts early complications in the self-poisoned patient. If capnometry is used, ETCO2 changes in value as well as its waveform in combination of other available clinical values should be considered.