CERTA

Continuous vs Routine Electroencephalogram in Adults with Altered Consciousness and No Recent Seizure. A Multi Centre Randomized Clinical Trial

Andrea O.Rossetti. JAMA Neurology 2020; 77(10):1225-1232. doi:10.1001/jamaneurol.2020.2264

Clinical Question

  • In critically ill patients with acute altered level of consciousness and no recent seizures, does continuous EEG (cEEG) compared to repeated routine EEG (rEEG) correlate with reduced mortality?

Background

  • EEG is emerging in critical care for its use in detecting convulsive and non clinical (non convulsive) seizures, status epilepticus (SE), monitoring sedation and anaesthesia, and its use in prognostication post out of hospital cardiac arrest
  • The European Society of Intensive Care Medicine and the American Clinical Neurophysiology Society guidelines suggest cEEG for critically ill patients with altered consciousness
  • This recommendation is based on low quality evidence from 2 retrospective observational trials, based on discharge diagnosis, which suggest that cEEG compared to rEEG is associated lower mortality, albeit at significantly greater cost and hospital length of stay in the cEEG group
  • High burden of seizures has been shown to be associated with worse short and long term outcome in critically ill patients
  • It is yet to be demonstrated that increased identification of seizures leads to improvement in outcome
  • This is the first randomised trial to compare cEEG to rEEG

Design

  • Multicentre randomised controlled trial
  • 1:1 randomisation through an online program stratified by site
  • Bedside clinicians were not blinded to patient allocation
    • Outcome assessors were blinded
  • Patients were recruited during local investigators’ availability (working hours, not on weekends)
  • Protocolled intervention was stopped in patients diagnosed as having seizures or SE during the study period to allow conversion to cEEG if treatment required
  • EEG results were communicated to treating teams within 2 hours of EEG start, at least 3 (working days) or 2 times per day (weekends, holidays)
  • Primary outcome was mortality at 6 months
  • Secondary outcomes were based on intention to monitor
  • Power calculation was based on previously available data from one North American Study
  • 2x 174 patients were required to achieve 80% power with 0.05 alpha error to detect a survival difference of 14% in patients undergoing cEEG compared to rEEG

Setting

  • Switzerland (4 centres)
  • Enrolment from April 2017- November 2018

Population

  • Inclusion: Patients in intensive or intermediate care units with GCS ≤11 or FOUR score ≤12 of any aetiology, referred from the treating team for EEG
  • Exclusion: Patients with recent seizures (36hours) or SE(96 hours) prior to randomisation, patients in palliative care, and those having invasive procedures within 48hours
  • 402 patients recruited
  • Well matched baseline demographics; age, gender
  • Groups differed with respect to cause of altered consciousness
    • rEEG group tended to have a lower burden of comorbidities, less prevalent hypoxic-ischaemic encephalopathy and brain trauma, and more prevalent ischaemic stroke, toxic metabolic disorders and other conditions
  • Comparing baseline characteristics of routine vs continuous groups
    • Age: 64 vs 64
    • Female: 33% vs 34%
    • Reason of admission
      • Brain injury (including cardiac arrest): 56% vs 62%
      • Medical: 33% vs 24%
      • Surgical: 9% vs 12%
    • Previous seizures (excluding seizures <36 hours or SE <96 hours pre randomisation): 10% vs 8%
    • GCS (median): 3 vs 3
    • Final neurologic diagnosis
      • Hypoxic ischaemic encephalopathy: 29% vs 32%
      • Brain trauma: 9% vs 17%
      • Intracranial haemorrhage: 22% vs 25%
      • Ischaemic stroke: 10% vs 5%

Intervention

  • Continuous EEG (cEEG) for 30-48 hours
    • Interruptions less than 2 hours were allowed e.g. for neuroimaging

Control

  • Routine EEG (rEEG)
    • Two 20-30 minute recordings over 48 hours
    • No repetition on the same day

Management common to both groups

  • EEG to commence within 4 hours of referral
  • Results communicated to treating teams regularly
    • Within 2 hours of EEG start, and 2-3 times daily
  • Protocolled intervention ceased in patients diagnosed as having seizures or SE during the intervention period
    • Subsequent treatment according to best practice (to allow conversion to cEEG if needed)
  • Management of seizures not protocolised (pragmatic trial)

Outcome

  •  Primary outcome: No significant difference in mortality at 6 months between cEEG vs rEEG
    • Mortality at 6 months : rEEG 88/182 (48.4%) vs cEEG 89/182 (48.9%) RR 1.01 (95% CI 0.81 to 1.25, P= 0.92)
    • Adjusted for Charlston Comorbidity Index (CCI), cardiac arrest: RR 1.02 (95% CI 0.83 to 1.26, P=0.85)
    • No difference after exploratory stratification for hypoxic ischaemic encephalopathy, or age, baseline modified Rankin Score (mRS), CCI, traumatic brain injury, intracranial haemorrhage, severity of consciousness impairment, or time to EEG
  • Secondary outcomes: Comparing cEEG vs. rEEG group
    • Functional outcome measures
      • Cerebral performance Category (CPC) at 6 months median (range) – no significant difference
        • 2 (1 to 4) vs 2 (1 to 4)
      • ∆mRS at 6 months (difference between pre admission mRS, estimated prospectively, and outcome mRS calculated at 6 months) – significantly greater in rEEG group
        • Regression coefficient 0.65 (95% CI 0.13 to 1.16, P= 0.01)
        • median (range) ∆mRS 1 (-3 to 5) vs 1(-5 to 4)
        • ie better evolution in mRS in patients in rEEG group based on estimated pre admission mRS
  • Exploratory secondary outcomes (intention to monitor)
    • Significantly greater in cEEG group
      • Detection of ictal/interictal features without seizures
        • 69.2 vs 55.7% RR 1.24 (95% CI 1.06-1.46, P=0.009)
      • Detection of seizures
        • 15.7 vs 4.4% RR 3.59 (95% CI 1.68-7.64, P=0.001)
      • Modification rate of anti seizure drug prescription
        • 21.1 vs 11.5% RR 1.84 (95% CI 1.12-3.00, P=0.01)
    • No significant difference in:
      • Sedatives, need for EEG after intervention, infections, mechanical ventilation duration, time to death since randomisation, and survivors’ length of stay
  • Exploratory Analysis:
    • Patients without hypoxic-ischaemic encephalopathy
      • Significantly better ∆mRS at 6 months with cEEG vs rEEG
      • Regression coefficient 0.73 (95%CI 0.10-1.36, P=0.023)
      • Median (range) ∆mRS 1(-3-5) vs 1(-5-4)
      • Mortality at 6 months showed non significant trend favouring cEEG 41% vs rEEG 45.4% RR 0.9 (95%CI 0.68-1.20, P=0.483)
    • Patients with hypoxic-ischaemic encephalopathy
      • Mortality at 6 months showed non significant trend favouring rEEG:  cEEG 65% vs rEEG 55% RR 1.15 (95% CI 0.86-1.58, P=0.327)
      • Death latency in subgroup with hypoxic ischaemic encephalopathy, calculated to explore whether cEEG facilitated withdrawal of life sustaining treatment
        • cEEG median 6 days (range 0-157), vs rEEG median 8.5 days (range 0-156), p=0.07

Authors’ Conclusions

  • Despite increased detection of seizures, and higher rate of anti-seizure treatment modifications, cEEG does not seem to correlate with improved patient outcome compared with repeated rEEG
  • rEEG may represent a valid alternative to cEEG in centres with limited resources

Strengths

  • First RCT to investigate this topic
  • Near complete follow up: only 4 patients lost to follow up
  • Blinding of outcome assessors
  • Intention to monitor analysis
  • Pragmatic design
    • Recruitment and reporting in hours, consistent with resource availability at most centres
  • 15% detection of seizure in cEEG
    • Comparable to existing literature
  • Heterogenous aetiology
    • Increases generalisability of the results
  • Patient centred outcomes
    • Mortality
    • Functional outcome measures

Weaknesses

  • Heterogenous aetiology
    • Effect in subgroups may be diluted
    • Non significant 5% improvement in mortality in patients without hypoxic ischaemic encephalopathy should be investigated in future studies
  • Higher mortality rate compared to previous studies
    • May be due to high proportion (30%) with hypoxic ischaemic encephalopathy, known to have a high mortality rate
  • Bedside clinicians were not blinded to patient allocation
  • Potential selection bias as only referred patients were recruited; not all patients with altered level of consciousness were screened
  • Reporting limitations overnight which may have resulted in prolonged seizure time without intervention
    • This may dilute the benefit of cEEG
  • Not stratified for severity of altered level of consciousness
  • Interventions in response to seizure detection not protocolised
    • Including conversion to cEEG
    • But most patients received anti epileptic drugs as per the author (not published)
  • Intention to monitor analysis means crossover to cEEG not accounted for in analysis
  • Evolution of mRS is based on estimated preadmission mRS, therefore interpret with caution
  • 27 patients who were randomised later withdrew consent and their results were not included in analysis

The Bottom Line

  • In critically ill patients with altered level of consciousness cEEG compared with rEEG leads to greater detection and treatment of seizures but has no effect on mortality or functional outcome
  • Further studies may elucidate a benefit for the use of cEEG in particular patient groups e.g. those without hypoxic ischaemic encephalopathy
  • I will continue to use rEEG to in patients with altered level of consciousness and repeat rEEG if ongoing seizures are suspected. This study suggests that the significant cost (expertise, resources, time) associated with cEEG does not translate to improved outcomes

External Links

Metadata

Summary author: Tessa Garside @tessagarside
Summary date: 17th March 2021
Peer-review editor: Celia Bradford @celiabradford

Image by Gerd Altmann from Pixabay

 

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.