Yeatts

Effect of video laryngoscopy on trauma patient survival: A randomised controlled trial

Yeatts. J Trauma Acute Care Surg 2013; 75(2):212-9. doi:10.1097/TA.0b013e318293103d

Clinical Question

  • In adults requiring emergency intubation, does video laryngoscopy (VL) compared to direct laryngoscopy (DL) lead to better or worse in-hospital mortality?

Design

  • Randomised, controlled trial
  • Randomisation method, allocation concealment and treatment blinding was not described
  • Physicians allowed to exclude from randomisation without specifying reason
  • Power calculation described as 80% power with accepted Type I error rate of 0.05 to detect absolute difference in mortality of +6% or -4% from baseline of 5%, but required patient numbers not given
    • Chance of false positive 5%
    • Chance of false negative 20%

Setting

  • Single trauma centre in Baltimore, Maryland, USA
  • July 2008 to May 2010 (paused in May to November 2009)

Population

  • Inclusion: requiring intubation for any clinical reason, including airway obstruction, hypoventilation, hypoxaemia, cognitive impairment, haemorrhagic shock, and extreme pain
  • Exclusion: children (age not defined); laryngeal trauma, extensive maxillofacial injury requiring immediate surgical airway; known or strongly suspected spinal cord injury (awake fibre-optic intubation performed instead); cardiac arrest on arrival; died in trauma resuscitation unit
  • 10318 screened, 898 required intubation, 623 randomised
  • 275 intubated patients excluded:
    • 35 were children or prisoners
    • 16 VL not available
    • 14 VL chosen against DL due to teaching opportunity or clinical indication
    • 210 physician chose not to recruit but no reason given

Intervention

  • GlideScope video laryngoscope

Control

  • Macintosh direct laryngoscope
  • Rapid sequence induction was performed by unblinded physicians and described as:
    • Pre-oxygenation for at least 1 minute when possible
    • Rapid-acting sedative at physicians discretion:
      • Thiopental 4 mg/kg if haemodynamically stable
      • Adjusted thiopental dose or etomidate 0.2–0.4 mg/kg if haemodynamically unstable
    • Short-acting neuromuscular blocking agent:
      • Suxamethonium 1.5 mg/kg
    • “Maneuvers to prevent passive regurgitation” – probably cricoid pressure
    • Manual in-line stabilisation in at risk patients
  • Intubation was performed by anaesthesia nurse, resident (PGY2–6) or attending trauma anaesthetist, always under supervision of the attending doctor (otherwise known as a consultant)

Outcome

  • Primary outcome: there was no difference demonstrated between the groups in survival to hospital discharge
    • DL: 7.5%
    • VL: 9.2%
    • p-value: 0.43 by Pearson’s Chi Squared test
  • Secondary outcome: duration of intubation attempt favoured direct laryngoscopy whilst first-pass success rate was not different between the groups
    • Mean intubation duration in seconds: DL 56.5 vs VL 71.0; p-value = 0.002
    • First-pass success: DL 81% vs VL 80%

Authors’ Conclusions

  • “The use of the videolaryngosope did not improve survival to hospital discharge and was associated with longer intubation attempt duration”

Strengths

  • With the growing popularity of VL, this is an important clinical question to ask
  • This is the first randomised, controlled trial in the emergency in-hospital setting for VL vs DL
  • Statistical sensitivity analysis was attempted to confirm that no selection bias was present despite non-compliant recruitment

Weaknesses

  • No detail regarding randomisation, allocation concealment and blinding prevents any assessment of important biases
  • Massively under-powered to detect a clinically important difference.
    • The group sizes provided 80% power to detect a difference from 5% to 11%. If an intervention leads to more than double mortality, it should never reach stage 3 clinical trials!
    • If a clinically important absolute risk reduction of 2% were defined, from the observed baseline of 7.5%, then 2384 patients per group are required to provide 80% power
    • A similar number will be required for a trial of non-inferiority methodology with similar power and 2% non-inferiority limit
  • 27% of eligible patients were excluded at physicians discretion: there is a high risk of selection bias
  • Planned exclusion of those with potentially difficult airways limits generalisability: this is exactly the population group where I might reach for a VL
  • Excluding those that died in the trauma resuscitation unit may strongly bias the results: if a patient died due to failed intubation or oxygenation, then the results were discounted! Immediate death is more likely to be causally related to intubation failure compared to a death several days later.
  • The majority of intubations were performed by relatively junior residents (PGY2 ≈ FY2), which is not typical trauma management in UK

The Bottom Line

  • These data will help guide future research design and planned statistical analyses
  • The unclear risk of bias, poor compliance with recruitment and under-powered statistics means no conclusion should be drawn from these results
  • Further research is required to answer this important clinical question

Leave a Reply