Best Evidence Topics
  • Send this BET as an Email
  • Make a Comment on this BET

Predicting need for endotracheal intubation in poisoned patients

Three Part Question

[In undifferentiated poisoned adult patients] which [risk factors are suggestive of need] for [endotracheal intubation]?

Clinical Scenario

You are working in A&E and review a patient with suspected poisoning or overdose, you are unclear of the identity of the substance so can not yet confidently consult the local toxicology database but are aware of the occasional need for intubation in such patients to maintain a safe airway and oxygenation.
You wonder if there are any other clinical risk factors or predictors indicative of need for endotracheal intubation in poisoned patients which you could use to support your decision to intubate or not.

Search Strategy

Medline via Ovid interface 1946 to Week 3 August 2018
EMBASE via Ovid interface 1974 to August 23rd 2018

((poison* OR intoxicat* OR overdose) AND ((risk factor*) OR predict*) AND (intubat* OR ventilat*)).mp
Limit to human and English language

Search Outcome

A search of the EMBASE database found 665 results of which 5 were relevant to the three part question. A search of the Medline database found 207 results of which 5 were relevant to the three part question.
The 5 relevant papers identified were the same in both databases.

Relevant Paper(s)

Author, date and country Patient group Study type (level of evidence) Outcomes Key results Study Weaknesses
El-Sarnagawy et al
104 patients with acute drug overdose admitted to a tertiary toxicology unit, age >16 with GCS <15.Prospective observational studyRequirement for intubation24/104 (23%) patients required intubationDoes not document reasons for deciding to intubate. Does not specify whether treating clinicians were blinded to calculated scores and whether they used any of these scores to help decide whether to intubate or not. Does not state whether cut offs for scoring systems were retrospectively or prospectively chosen. Single centre tertiary study, risk of selection bias, unclear of referral criteria from other centres.
APACHE II scoreAPACHE II score >18 sensitivity of 66.67% and specificity of 90% for predicting need for intubation
GCSGCS ?8 sensitivity 100% and specificity of 50% for predicting need for intubation
RAPS scoreRAPS >6 sensitivity of 50% and specificity of 90% for predicting need for intubation
REMS scoreREMS >8 sensitivity of 67.66% and specificity of 100% in predicting need for intubation
Hua et al
2497 patients with acute drug overdose presenting to an urban emergency department in a tertiary hospital, age ?18Retrospective observational study (prospectively collected, secondary analysis)Intubation87/2497 (3.5%) intubatedRetrospective secondary analysis. Timing of blood gas analysis not specified nor stated whether taken prior to any intervention. Clinician judgement gold standard. Urban teaching hospital setting perhaps limits generalisation.
Indications for intubation givenPerceived worsening clinical course (50.6%), depressed mental status/coma (27.6%), failure to oxygenate (24.1%), protect airway (19.5%), failure to ventilate (14.9%), cardiac arrest (13.8%), seizing (5.7%), respiratory arrest (4.6%), other (2.3%)
Obstructive lung disease (OLD) (asthma/emphysema/chronic bronchitis) or congestive heart failure (CHF)Adjusted odds ratio 6.6 (CI 3.5 -12.3) for risk of needing intubation with OLD. No association with intubation in CHF
Arterial blood gas analysisNot analysed independently with regards to intubation but pCO2 higher and pH lower in OLD patients. pCO2 cut off of 60mmHg had PPV of 6% (CI 2.8-11.2) for predicting mortality, NPV of 99.4% (CI 98.8% - 99.8%) for predicting mortality
AgeAdjusted OR 0.97 (CI 0.96-0.98) for every year over age 18 for risk of intubation
Donald et al
26 patients with a toxicological presentation and reduced glasgow coma scale presenting to an urban emergency department, age ?14Prospective observational studyIntubation12/26 intubatedUrban emergency department limits generalisation. Gold standard clinician judgement. Observational study. Small numbers limits statistical analysis. Five non-intubated patients managed on short stay ward with airway adjuncts, staffing may not facilitate this in all centres.
Reason given for intubation (percentage of 12 intubated patients)Loss of airway protection (81.8%), failure ventilation/oxygenation (50%), predicted clinical course 66.6%.
Age Mean age 34.6 in intubated group, mean 38 in non-intubated. No further statistical analysis.
Heart rateMean 110bpm in intubated group, mean 99bpm in non-intubated. No further statistical analysis.
Systolic blood pressureMean SBP 119 in intubated group Mean SBP 112 in non-intubated.
GCS Mean 5.9 (range 3-11) in intubated group. Mean 5.5 (3-8) in non-intubated group.
Arterial blood gas analysis Tendency to worse oxygenation/ventilation in intubated group. PaO2 on high-flow O2 interquartile mean (IQM) 20.1kpa in intubated group, IQM 33.4kpa in non-intubated. PaCO2 IQM 6.5kpa in intubated group, IQM 5.64 in non-intubated group.
Duncan et al
73 patients with an acute toxicological presentation and GCS <15 admitted to an emergency department short stay ward in an urban emergency department, age ?14Prospective observational studySubsequent need for intubation after admission to short stay1/73 (GCS 14 on admission and subsequently deteriorated to GCS 6)Observational study. Clinician judgement gold standard. Possible selection bias, were only lowest perceived risk patients admitted for only observation? Some patients managed with airway adjuncts, may be unable to replicate this and observe safely in other centres. No follow-up after discharge.
GCS on admission Median GCS 11 (range 3 -14) 12 patients had GCS?8. 3 patients had GCS of 3.
Complications (cardiac/respiratory arrest or clinically significant aspiration)No episodes of aspiration/respiratory or cardiac arrest in all patients.
Chan et al
393 patients with a presentation of acute poisoning -for which an antidote was not available - presenting to an urban emergency departmentProspective observational studyNeed for intubation43/393 (10.9%) intubatedClinician judgement gold standard. No mention of what part GCS formed of clinicians’ decision to intubate. Gastric lavage used in 60% of patients, limits generalisation to modern UK practice and affects decision to opt for definitive airway protection. Gag reflex may have formed part of clinician assessment.
Relationship between GCS and need for intubation GCS ?8 sensitivity of 90 (CI 81% - 99%) for predicting intubation, specificity of 95% (CI 93% - 97%) for predicting intubation. Logistic regression analysis odds ratio 0.48 (CI 93% - 97%), twofold increase in risk of intubation for every drop in GCS.
Relationship between gag reflex and need for intubationSensitivity of 70% (CI 56% - 85%) and specificity of 100% for need for intubation


The poisoned patient presenting to the emergency department presents a unique challenge to clinical decision making. A degree of diagnostic and prognostic uncertainty is frequently inherent to the clinical situation; a poisoned patient with reduced conscious level may leave the clinician unclear of the substance(s) or reliant on the accuracy of a collateral history. The purpose of this search was therefore to look for clinical predictors common to poisoned patients which are suggestive of need for intubation. GCS seems the most obvious scoring system to turn to and perhaps inspired by the convention in trauma to intubate <8, Chan et al found a cut-off of GCS <8 had sensitivity of 90% and specificity of 95% in predicting intubation. In Chan et al’s study clinician judgement was the gold standard with no reported indications for intubation, if local practice used a GCS cut-off as part of the decision to intubate, this could demonstrate a self-fulfilling prophecy. El-Sanargawy et al found a sensitivity of 100% and specificity of 50% for GCS <8 to predict need for intubation but had the same limitations as Chan et al. Chan et al also looked at the gag reflex and found good specificity and poor sensitivity for predicting need for intubation; again it is not clear whether clinicians used gag reflex to inform their assessment and provoking a gag reflex in a patient with reduced consciousness may come with its own risks of vomiting. Of note, of the 18/55 patients with GCS <8 not intubated in Chan et al’s study, none had any respiratory complications and indeed, the experiences of Duncan et al showed that it is possible to safely observe selected poisoned patients with GCS<8 with simple airway adjuncts and no subsequent aspiration if a suitably staffed clinical observation area is available. Selection of such patients in Duncan et al’s study was by experienced clinicians but factors affecting a patient being suitable for solely observation were not clear, nonetheless this does serve as some proof of concept. Hua et al found presence of pre-morbid obstructive lung disease had an odds ratio of 6.6 for predicting need for intubation but no link with the presence of congestive heart failure. Hua et al also found an odds ratio of 0.97 for predicting need for intubation for every year gained over the age of 18, hypothesised to be due to more dangerous substance use in younger age groups. Physiological parameters show some promise and El-Sanargawy et al found an abbreviated version of APACHE II, the REMS score, had sensitivity of 67.66% and specificity of 100% for predicting need for intubation with a REMS >8. Donald et al found a tendency towards pO2 and higher pCO2 on arterial blood gas analysis in patients requiring intubation (albeit without statistical analysis) and although El-Sanargawy et al looked only at link with mortality, they found a pCO2< 60mmHg (7.9Kpa) had good negative predictive value and a pCO2>60mmHg had poor positive predictive value for subsequent mortality, perhaps suggestive of a way of identifying higher risk patients. GCS = glasgow coma scale, APACHE II = acute physiology and chronic health evaluation, REMS = rapid emergency medicine score, RAPS = rapid acute physiology score,

Editor Comment


Clinical Bottom Line

Decisions regarding intubation in poisoned patients remain primarily based on clinical judgement as to perceived failure to ventilate, oxygenate and protect the airway. GCS can be used to form part of the assessment but with adequately supervised observation areas, suitable patients selected by experienced clinicians with GCS <8 can be closely observed with simple airway adjuncts if required. Obstructive lung disease should heighten consideration of intubation and arterial blood gases tending towards hypercapnea and hypoxia may be supportive of need for intubation. Use of physiological parameter scoring systems such as the REMS score show promise but require prospective evaluation.

Level of Evidence

Level 3 - Small numbers of small studies or great heterogeneity or very different population.


  1. El-Sarnagawy et al Comparison of different scores as predictors of mechanical ventilation in drug overdose patients Human and Experimental Toxicology 2017; Vol. 36(6) 539–546
  2. Hua et al Endotracheal Intubation after Acute Drug Overdoses: Incidence, Complications, and Risk Factors J Emerg Med Jan 2017; 52(1): 59–65
  3. Donald et al Predictors of the need for rapid sequence intubation in the poisoned patient with reduced Glasgow coma score Emerg Med J 2009; 26:510–512
  4. Duncan et al Decreased glasgow coma scale does not mandate endotracheal intubation in the emergency department J Emerg Med 2009; 37:451–455
  5. Chan et al The use of glasgow coma scale in poisoning J Emerg Med 1993; 11:579-582