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

Assessing the need for intubation following tricyclic antidepressant overdose in patients with reduced level of consciousness

Three Part Question

In [adult patients who present to the ED following a psychotropic drug overdose with a reduced level of consciousness] does [endotracheal intubation versus standard treatment alone] lead to [fewer respiratory complications, reduced mortality and reduced length of hospital stay]?

Clinical Scenario

A twenty year old lady is brought to the ED by paramedics. She had been found next to empty amitriptyline tablet packets with fifty tablets missing. On arrival at the ED she is semi-conscious with Glasgow Coma Scale 8/15. Her airway is patent and there is no evidence of hypoventilation. You wonder whether endotracheal intubation is necessary or whether you can manage the overdose expectantly by placing the patient in the recovery position and administering standard treatment alone.

Search Strategy

Ovid Medline 1950 – May Week 2 2008
Ovid Embase 1980 – 2008 Week 21
(exp Overdose/ OR exp Poisoning/ OR AND (exp Intubation, Intratracheal/ OR (rapid sequence induction).mp OR OR OR (crash induction).mp OR airway AND (Glasgow Coma OR exp Coma/ OR exp Glasgow Coma Scale/ OR exp Unconsciousness/ OR (unconscious$ or semiconscious$ or obtund$ or unresponsive$).mp.) limit to human and English language

Search Outcome

62 papers were identified in Medline and 159 in Embase. Six were relevant to the three-part question.

Relevant Paper(s)

Author, date and country Patient group Study type (level of evidence) Outcomes Key results Study Weaknesses
Chan et al
393 patients who presented to the ED with a history or evidence of overdose (of a drug with no antidote) who had GCS documented at presentationRetrospective analysisGCS ≤8/15 for prediction of intubation67% of patients with GCS ≤8/15 were intubated. GCS ≤8/15 had sensitivity 90% (95% CI 81-99%) and specificity 95% (93-97%) for prediction of intubationStudy only assesses what actually happened (whether patients were intubated or not). We do not know whether it was actually necessary to intubate the patients with GCS ≤8/15. No reporting of complications in semi-conscious patients who were/were not intubated.
Relationship between GCS and intubation (logistic regression analysis)Odds ratio 0.48 (95% CI 0.4-0.59), P<0.0001 (i.e. odds of intubation increase approximately two-fold for every point decrease in GCS)
Emerman et al
United States
All 92 patients age ≥17 years who were admitted to Cleveland Metropolitan General Hospital with TCA overdose between 1975 and 1985.Retrospective analysisAssociation between GCS and complications (hypoventilation, loss of protective airway reflexes, hypotension, seizures, haemodynamically significant arrhythmias or deathSignificant association (P<0.001). GCS was significantly better than QRS interval (P<0.001)Retrospective 38 patients had a mixed drug overdose (although subgroup analysis of patients with pure TCA overdose yielded similar results) Only 92 patients included over a 10 year period.
GCS ≤8 for prediction of serious complicationsSensitivity 89%, specificity 88%. GCS≤8 was significantly more sensitive than QRS ≥100ms (P<0.05)
Sensitivity of GCS ≤8 for prediction of individual complicationsHypoventilation or loss of protective airway reflexes: 86.5%; Death, hypotension, seizures, haemodynamically significant arrhythmias: 100%
Logistic regression model for prediction of complicationsOnly GCS was a significant independent predictor of complications
Hulten et al
67 patients ≥14 years from four centres with suspected TCA overdose. Excluded if mixed overdose detected and TCA was not the major cause of symptoms. Matthew-Lawson coma grade recordedProspective diagnostic cohort studyMatthew-Lawson coma grade ≥3 for prediction of serious complications (seizures, hypotension (systolic BP <100mmHg), arrhythmias, need for intubation)Sensitivity 65%, specificity 94%Matthew-Lawson coma grade not universally accepted for assessing conscious level (GCS not recorded) Need for intubation included as an outcome. Physicians may have decided to intubate on the basis of coma grade alone, thus introducing significant bias.
Matthew-Lawson coma grade ≥2 for prediction of serious complicationsSensitivity 81%, specificity 77%
Matthew-Lawson coma grade vs. QRS duration and plasma TCA level for prediction of serious complicationsMatthew-Lawson coma grade was the strongest predictor in logistic regression model. QRS duration >100ms was more sensitive for prediction of complications (86%) but less specific (75%)
Liisanantti et al
257 patients admitted to ICU with self-poisoning of psychopharmaceutical drugs between November 1989 and October 2000 Classed as conscious (GCS 8-15) or unconscious (3-7) based on ‘approximate GCS’ 73 patients (28.4%) met criteria for aspiration pneumoniaRetrospective analysisUnconsciousness on discovery for prediction of aspiration pneumoniaOR 2.9 (95% CI 1.2 – 7.0)Retrospective ‘Approximate GCS’ used due to lack of universal use of GCS in Finland Selection bias: Only patients admitted to ICU included GCS at time of initial contact with medical services not recorded in 20.6% of cases Possible reporting bias – this centre may have noticed a particularly high rate of aspiration pneumonia in patients intubated late, prompting this analysis.
Unconsciousness in ED for prediction of aspiration pneumonia2.2 (0.9 – 5.4)
Unconscious when found and intubated on discovery for prediction of aspiration pneumonia1.8 (0.6 – 5.7)
Unconscious when found and intubated in ED for prediction of aspiration pneumonia3.4 (1.3 – 8.7)
Unconscious when found and intubated in ICU for prediction of aspiration pneumonia3.5 (1.1 – 10.7)
Mean length of hospital stayAspiration pneumonia 6.5 days (95% CI 5.3 – 7.6); No aspiration pneumonia 2.8 days (2.5 – 3.1)
Mean length of ICU stayAspiration pneumonia 1.9 days (1.3 – 2.6); No aspiration pneumonia 0.9 days (0.8 – 0.9)
Unvenir et al
356 patients who presented to the ED with antidepressant ingestion between 1993 and 2004Retrospective analysisRelationship between GCS and intubation rates34 (9.6%) patients were intubated. Low GCS was cited as the reason for intubation in 58.8% of cases. 100% of patients with GCS ≤8 were intubated compared with 5.6% of patients with GCS >8Retrospective Obvious bias in outcome reporting: Almost 60% of patients were intubated primarily because of low GCS. There was no attempt to correlate low GCS with incidence of complications
Logistic regression model for prediction of the need for intubationGCS the strongest independent predictor of need for intubation (OR 29.4, 95% CI 8.1 – 106.4). Presence of seizures was also an independent predictor of intubation. Age, gender and QRS prolongation were not independent predictors
Yanagawa et al
175 patients who were intubated following psychotropic drug overdose between January 2000 and December 2005 Patients were divided into an “early group” (extubated within 2 days) and a late group (not extubated within 2 days)Retrospective analysisMean GCS (on arrival) in early and late groupsEarly group 6.2 (SE 0.2); Late group 4.5 (SE 0.3), P=0.001Retrospective Significant selection bias: only intubated patients included No analysis of different GCS cut-offs for prediction of late extubation
Logistic regression model for prediction of “late” extubation (>2 days)GCS on arrival was an independent predictor of late extubation (OR 0.78, 95% CI 0.65 – 0.95)


In total we identified five retrospective analyses of patients who had been admitted following psychotropic drug overdoses and one prospective diagnostic cohort study that investigated the association between Matthew-Lawson coma grade and serious complications following tricyclic antidepressant overdose. Although the studies have significant weaknesses, a strong correlation has consistently been shown between level of consciousness and the development of serious complications including death, hypoventilation and aspiration pneumonia following drug overdose. Of interest, both Hulten et al and Emerman et al showed that TCA drug levels are of little use for predicting complications especially when coma grade and QRS width were taken into account. Further it would seem that level of consciousness is a stronger independent predictor of complications than QRS width. The evidence strongly suggests that patients with GCS ≤8 should undergo intubation at an early stage in the ED. Results from the retrospective study by Liisanantti et al suggest that intubation at the earliest possible opportunity may reduce complication rates. Further, in the study by Emerman et al GCS ≤8 was only 86.5% sensitive for prediction of hypoventilation or loss of protective airway reflexes. Thus intubation may still be necessary for some patients with GCS >8 from a pragmatic patient safety viewpoint.

Editor Comment

Abbreviations: TCA, tricyclic antidepressant; ED, Emergency Department; GCS, Glasgow Coma Scale; OR, odds ratio; CI, confidence intervals; ICU, Intensive Care Unit; SE, standard error

Clinical Bottom Line

Patients who present to the ED following psychotropic drug overdose with GCS ≤8 should undergo intubation at the earliest opportunity. Some patients with GCS >8 may also need intubation.

Level of Evidence

Level 2 - Studies considered were neither 1 or 3.


  1. Chan B; Gaudry P; Grattan-Smith TM; McNeill R The use of Glasgow Coma Scale in poisoning The Journal of Emergency Medicine 1993; 11: 579-82
  2. Emerman CL; Connors AF; Burma GM Level of consciousness as a predictor of complications following tricyclic overdose Annals of Emergency Medicine 1987; 16: 326-30
  3. Hulten BA; Adams R; Askenasi R; Dallos V; Dawling S; Volans G; Heath A Predicting severity of tricyclic antidepressant overdose Journal of Toxicology: Clinical Toxicology 1992; 30(2): 161-70
  4. Liisanantti J; Kaukoranta P; Martikainen M; Ala-Kokko T Aspiration pneumonia following self-poisoning Resuscitation 2003; 56: 49-53
  5. Unverir P; Atilla R; Karcioglu O; Topacoglu H; Demiral Y; Tuncok Y A retrospective anlalysis of antidepressant poisonings in the Emergency Department: 11-year experience Human & Experimental Toxicology 2006; 25: 605-12
  6. Yanagawa Y; Sakamoto T; Okada Y Recovery from a psychotropic drug overdose tends to depend on the time from ingestion to arrival, the Glasgow Coma Scale, and a sign of circulatory insufficiency on arrival American Journal of Emergency Medicine 2007; 25: 757-761