Skip to main content

Foundational Trials

NICO

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Effect of Noninvasive Airway Management of Comatose Patients With Acute Poisoning: A Randomized Clinical Trial
  • Acronym: NICO
  • Year: 2023
  • Journal published in: JAMA
  • Citation: Freund Y, Viglino D, Cachanado M, Cassard C, Montassier E, Douay B, et al. Effect of noninvasive airway management of comatose patients with acute poisoning: a randomized clinical trial. JAMA. 2023;330(23):2267-2274.

Context & Rationale

  • Background
    • Tracheal intubation is traditionally taught for coma to “protect the airway”, often operationalised as a Glasgow Coma Scale threshold (commonly ≤8).
    • In acute poisoning, coma is frequently transient and reversible, yet intubation exposes patients to procedure-related harms, sedation/neuromuscular blockade effects, mechanical ventilation complications, and increased ICU utilisation.
    • Pre-trial evidence for intubation based on neurological score alone was predominantly observational and confounded by toxin type, trajectory, clinician preference, and system factors; randomised evidence was lacking.
  • Research Question/Hypothesis
    • In adults with suspected acute poisoning and GCS <9, but without immediate physiological instability or other urgent indications for intubation, can a protocolised restricted-intubation strategy improve outcomes compared with routine practice?
    • Hypothesis: avoiding systematic early intubation would reduce ICU admission and length of stay without increasing serious complications.
  • Why This Matters
    • Poisoned comatose patients are often young with low baseline mortality, so iatrogenic complications and resource use may dominate net benefit–harm.
    • A rigorous test of “GCS-driven intubation” has immediate implications for ED workflow, ICU bed demand, and prehospital airway decision-making.
    • Clarifying safe deferral criteria supports more physiologically grounded airway management and may reduce unnecessary invasive ventilation.

Design & Methods

  • Research Question: In adults with suspected acute poisoning and GCS <9 (without urgent indications for intubation), does a restricted-intubation strategy (noninvasive airway management with rescue intubation for pre-specified deterioration during the first 4 hours) improve a hierarchical clinical outcome compared with routine practice?
  • Study Type: Multicentre, parallel-group, pragmatic, open-label randomised clinical trial; 20 emergency departments in France (including a physician-led prehospital emergency system).
  • Population:
    • Inclusion: Adults (≥18 years) with suspected acute poisoning; comatose with GCS <9 in the prehospital setting or ED.
    • Key exclusions: Indication for immediate intubation (seizures, respiratory distress, shock); suspicion of brain injury; cardiotropic drug poisoning (beta-blockers, calcium channel blockers, ACE inhibitors); isolated reversible intoxication (opioids or benzodiazepines with naloxone/flumazenil response); inability to complete follow-up; other protocol-defined exclusions.
    • Randomisation: Sealed-envelope allocation; stratified by hospital with block balancing; block size concealed from investigators.
    • Consent: Deferred consent process; participants who subsequently opposed data use or were under legal protection were excluded from analysis.
  • Intervention:
    • Restricted intubation strategy (first 4 hours after admission or until GCS >8 for ≥30 minutes): prioritised noninvasive airway management and close monitoring, with tracheal intubation only if pre-specified emergency criteria occurred.
    • Monitoring: Systematic assessments at least every 30 minutes during the intervention window.
    • Rescue intubation criteria: seizure; respiratory distress defined as SpO2 <90% despite oxygen via nasal cannula; vomiting; shock defined as systolic BP <90 mm Hg despite 1 L crystalloid.
  • Comparison:
    • Routine practice: airway management at clinician discretion (including early intubation if preferred), without protocolised restriction during the first 4 hours.
  • Blinding: Unblinded (participants and clinicians); outcomes not blinded.
  • Statistics: A total of 240 patients were required (α 0.05; 98% power) based on simulations assuming ICU stay mean 0 vs 1 day (restricted vs routine), hospital stay mean 2 vs 4 days, and 3% mortality in both groups; primary analysis was modified intention-to-treat using a Finkelstein-Schoenfeld hierarchical composite (death, ICU length of stay, hospital length of stay), reported as a win ratio; no interim analysis planned.
  • Follow-Up Period: In-hospital outcomes through discharge, truncated at 28 days.

Key Results

This trial was not stopped early. Recruitment completed with 237 randomised patients; 225 were included in the primary analysis (restricted strategy n=116; routine practice n=109).

Outcome Restricted strategy Routine practice Effect p value / 95% CI Notes
Primary outcome: hierarchical composite (death, ICU length of stay, hospital length of stay) Not applicable (pairwise ranking) Not applicable (pairwise ranking) Win ratio 1.85 95% CI 1.33 to 2.58 12,644 pairwise comparisons; wins 8,166; losses 4,404; ties 74; 5 comparisons excluded due to missing data
In-hospital death 0/116 (0%) 0/109 (0%) Not estimable Not reported No deaths in either group
ICU admission 46/116 (39.7%) 72/109 (66.1%) OR 0.23 95% CI 0.12 to 0.44 Absolute difference −29.2% (95% CI −41.0 to −17.4)
ICU length of stay (hours) Median 0 (IQR 0 to 18.5) Median 24.0 (IQR 0 to 57.0) RR 0.39 95% CI 0.24 to 0.66 Duration from ICU admission to ICU discharge
Hospital length of stay (hours) Median 21.5 (IQR 10.5 to 44.5) Median 37.0 (IQR 16.0 to 79.0) RR 0.74 95% CI 0.53 to 1.03 Censored at day 28
Any tracheal intubation (process separation) 19/116 (16.4%) 63/109 (57.8%) Not reported Not reported Prehospital 3.4% vs 25.7%; ED 4.3% vs 22.9%; ICU 8.6% vs 9.2%
Mechanical ventilation 21/116 (18.1%) 65/109 (59.6%) OR 0.12 95% CI 0.06 to 0.24 Absolute difference −42.5% (95% CI −54.1 to −30.9)
Pneumonia 8/116 (6.9%) 16/109 (14.7%) OR 0.43 95% CI 0.18 to 1.05 Absolute difference −7.8% (95% CI −15.9 to 0.3)
Adverse event from intubation attempt 7/113 (6.0%) 16/107 (14.7%) OR 0.37 95% CI 0.15 to 0.95 Denominators reflect those with an intubation attempt captured
First-pass failure (intubation attempt) 1/113 (0.9%) 14/107 (13.1%) OR 0.06 95% CI 0.01 to 0.46 Absolute difference −12.2% (95% CI −18.8 to −5.6)
  • Large protocol separation was achieved: intubation 16.4% (restricted) vs 57.8% (routine), and mechanical ventilation 18.1% vs 59.6%.
  • The primary hierarchical composite favoured the restricted strategy (win ratio 1.85; 95% CI 1.33 to 2.58), driven mainly by reduced ICU admission and ICU length of stay; hospital length of stay effect was uncertain (RR 0.74; 95% CI 0.53 to 1.03).
  • No deaths occurred; pneumonia was numerically lower but imprecise (OR 0.43; 95% CI 0.18 to 1.05), while intubation-attempt adverse events and first-pass failure were lower in the restricted group.

Internal Validity

  • Randomisation and allocation:
    • Hospital-stratified, block-balanced randomisation with concealed block size.
    • Allocation by sealed envelopes (pragmatic concealment but more vulnerable than centralised electronic allocation).
  • Dropout or exclusions (post-randomisation):
    • 12/237 randomised excluded from analysis due to deferred consent issues or legal protection measures: 9/121 (7.4%) in the restricted group vs 3/116 (2.6%) in routine practice.
    • This constitutes a modified intention-to-treat population rather than a strict ITT analysis.
  • Performance/detection bias:
    • Unblinded clinicians and patients.
    • Key outcome components (ICU admission and ICU/hospital discharge timing) are susceptible to clinician and system decision-making.
  • Protocol adherence and delivery:
    • Restricted strategy applied for 4 hours after admission or until GCS >8 for ≥30 minutes; after this period, management reverted to routine practice.
    • Among 19 intubations in the restricted group, 16 were triggered by emergency criteria; timing of emergency-triggered intubations: 4 within 30 minutes, 8 between 30 minutes and 2 hours, and 4 between 2 and 4 hours.
    • 3/19 intubations in the restricted group were performed for “routine practice” rather than protocol-defined deterioration.
  • Baseline characteristics:
    • Groups were broadly comparable: median age 33 vs 34 years; median GCS 6 vs 6; suspected mixed poisoning 59.5% vs 67.9%.
    • Low baseline risk population: no in-hospital deaths, limiting the ability to evaluate mortality or rare catastrophic harms.
  • Heterogeneity:
    • Multicentre ED-based enrolment (20 sites) improves robustness, but practice patterns (including prehospital intubation) likely varied.
    • Trial design stratified by hospital, but detailed screening logs were not reported in the main manuscript.
  • Timing:
    • The intervention was time-limited (first 4 hours), matching the period during which airway decisions are most discretionary in this phenotype.
    • In the restricted group, most rescue intubations occurred within 2 hours of admission, suggesting monitoring identified early deterioration when it occurred.
  • Separation of the variable of interest:
    • Tracheal intubation: 16.4% (19/116) vs 57.8% (63/109).
    • Mechanical ventilation: 18.1% (21/116) vs 59.6% (65/109).
    • ICU admission: 39.7% (46/116) vs 66.1% (72/109).
  • Outcome assessment and statistical rigour:
    • Primary outcome used a pre-specified Finkelstein-Schoenfeld hierarchical approach (death, ICU length of stay, hospital length of stay), reported as a win ratio with 95% CI.
    • Key safety outcomes (pneumonia, intubation attempt adverse events) were reported with effect estimates, but event counts were small.

Conclusion on Internal Validity: Moderate. Randomisation and large between-group separation support causal inference for reduced intubation and ICU utilisation, but post-randomisation exclusions, unblinded care, and decision-sensitive outcome components increase susceptibility to performance and measurement bias—particularly for ICU admission and length-of-stay endpoints.

External Validity

  • Population representativeness:
    • Median age in the early 30s with no deaths suggests a comparatively low-risk poisoned coma cohort.
    • Commonly suspected agents included alcohol and benzodiazepines; cardiotropic drug poisonings and isolated reversible intoxications were excluded.
  • Applicability:
    • Conducted in France with a physician-led prehospital emergency system; intubation patterns and monitoring capacity may differ substantially in paramedic-led EMS models.
    • Safe replication requires an environment capable of frequent reassessment (≥30-minute intervals) and rapid rescue intubation by skilled clinicians.
    • Generalisability is uncertain to older, comorbid patients; toxin profiles with delayed deterioration; and resource-limited settings where close monitoring outside ICU is difficult.

Conclusion on External Validity: Moderate. Findings are most generalisable to young-to-middle-aged poisoned coma patients without early physiological instability in systems able to provide close ED monitoring and rapid airway rescue; extrapolation to higher-risk toxins, different EMS structures, or constrained monitoring environments is uncertain.

Strengths & Limitations

  • Strengths:
    • Pragmatic, multicentre randomised design addressing a high-variation, dogma-driven clinical decision.
    • Clear operationalisation of a restricted-intubation strategy with pre-specified rescue criteria and structured monitoring.
    • Large separation in exposure (intubation and mechanical ventilation) enabling meaningful inference about strategy effects on utilisation outcomes.
    • Appropriate hierarchical composite approach for competing outcomes in low-mortality populations.
  • Limitations:
    • Open-label design with outcome components strongly influenced by clinician/system decisions (ICU admission and length-of-stay).
    • Modified intention-to-treat due to deferred-consent exclusions, with differential exclusions between groups.
    • Low event rates and no deaths, limiting precision for rare but critical harms (aspiration, cardiac arrest, catastrophic airway events).
    • Predominance of alcohol/benzodiazepine intoxication and a physician-led prehospital system may constrain generalisability.
    • Intervention applied for only the first 4 hours; later trajectory management was not protocolised.

Interpretation & Why It Matters

  • Clinical implication
    In selected poisoned comatose adults (GCS <9) without early instability, a restricted-intubation strategy can substantially reduce intubation and mechanical ventilation without a detected increase in pneumonia or mortality, though safety precision is limited by low event rates.
  • Systems implication
    The strategy is not “no intubation”; it is a commitment to frequent monitoring with protocolised rescue triggers—implementation requires staff bandwidth, monitoring capability, and rapid access to expert airway management.
  • Conceptual implication
    The trial challenges a neurological-score threshold as a standalone mandate and reframes airway decisions around physiological trajectory, aspiration risk factors, and the safety profile of the local intubation environment.

Controversies & Subsequent Evidence

  • Endpoint choice and interpretability:
    • The hierarchical composite incorporated ICU admission/ICU length of stay and hospital length of stay, outcomes that can be directly altered by the decision to intubate and admit to ICU, raising concerns about circularity and “resource-use dominance” in a low-mortality cohort.123
    • The trial observed zero deaths, which removed the “hardest” hierarchy component and increased reliance on length-of-stay metrics for separation of the primary outcome.1
  • Population, setting, and generalisability:
    • The cohort was dominated by alcohol and benzodiazepine intoxication, and the intervention required close observation; correspondence highlighted uncertainty about applying this approach to toxins associated with prolonged coma or delayed deterioration, and to systems without robust ED monitoring capacity.14
    • The authors’ reply emphasised that the strategy presupposes a setting where clinicians can monitor closely and intubate rapidly when rescue criteria occur.6
  • Attribution of intubation-related harms:
    • Letters emphasised that routine-practice intubations occurred more frequently in the prehospital setting, and differences in environment and operator experience could contribute to observed differences in first-pass failure and intubation-attempt adverse events, complicating interpretation that harms were intrinsic to the “routine” strategy rather than the conditions under which intubations occurred.23
    • The reply acknowledged the difference in intubation location between groups as a plausible contributor to procedural outcomes and reinforced the intent of “restricted intubation” as avoidance of non-indicated early intubation, not avoidance of rescue airway management.6
  • Safety signal uncertainty:
    • Correspondence questioned whether pneumonia and aspiration events were sufficiently captured and whether a time-limited intervention window could miss later complications related to aspiration risk or toxin trajectory.5
  • Subsequent evidence synthesis:
    • Post-trial reviews highlight that NICO remains a key randomised datapoint in an evidence base otherwise dominated by observational designs, with wide practice variation and low event rates; they support moving beyond GCS thresholds alone while underscoring the need for further prospective evaluation across toxin profiles and system contexts.78

Summary

  • In adults with suspected acute poisoning and GCS <9 (without early physiological compromise), a restricted-intubation strategy over the first 4 hours improved a hierarchical composite outcome (win ratio 1.85; 95% CI 1.33 to 2.58).
  • Protocol separation was substantial: intubation 16.4% vs 57.8% and mechanical ventilation 18.1% vs 59.6% (restricted vs routine practice).
  • Benefits were largely expressed through reduced ICU admission (39.7% vs 66.1%) and shorter ICU stay (median 0 vs 24 hours); hospital length-of-stay reduction was uncertain.
  • No deaths occurred; pneumonia was numerically lower but imprecise (OR 0.43; 95% CI 0.18 to 1.05).
  • The main interpretive tension is whether reduced ICU utilisation reflects improved patient-centred outcomes versus decision-linked composite structure; replication in other systems and higher-risk poisonings is needed.

Overall Takeaway

NICO provides randomised evidence that, in carefully selected poisoned comatose adults without early physiological compromise, a restricted-intubation strategy over the first 4 hours can markedly reduce intubation and ICU utilisation and improve a hierarchical outcome. The trial’s landmark status lies in challenging a score-threshold dogma, but its open-label design, decision-sensitive composite endpoint, and low event rates mean that safety and generalisability require cautious interpretation and further validation.

Overall Summary

  • Restricted intubation reduced intubation (16.4% vs 57.8%) and mechanical ventilation (18.1% vs 59.6%).
  • Primary composite favoured the restricted strategy (win ratio 1.85), driven by ICU admission and ICU length of stay differences.
  • No deaths occurred; pneumonia and other harms were infrequent and imprecisely estimated.

Bibliography