Skip to main content

Foundational Trials

PrePARE

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Effect of Fluid Bolus Administration on Cardiovascular Collapse Among Critically Ill Patients Undergoing Tracheal Intubation: A Randomized Clinical Trial
  • Acronym: PREPARE II
  • Year: 2022
  • Journal published in: JAMA
  • Citation: Russell DW, Casey JD, Gibbs KW, et al; PREPARE II Investigators and the Pragmatic Critical Care Research Group. Effect of fluid bolus administration on cardiovascular collapse among critically ill patients undergoing tracheal intubation: a randomized clinical trial. JAMA. 2022;328(3):270-279.

Context & Rationale

  • Background
    • Peri-intubation haemodynamic instability and hypoxaemia are common in the ICU; large observational datasets report clinically important adverse peri-intubation events across diverse health systems.1
    • Induction-related vasodilation, blunting of sympathetic tone, and conversion to positive pressure ventilation can reduce venous return and cardiac output, providing physiologic rationale for pre-induction “volume loading”.
    • Randomised evidence before PREPARE II was sparse; the earlier PrePARE trial found no overall benefit of a fluid bolus, but suggested possible effect modification among patients receiving positive pressure ventilation during the procedure, motivating a focused trial in that subgroup.2
    • Despite limited trial evidence, fluid boluses were commonly incorporated into “intubation bundles”, creating a high-impact target for confirmation or de-implementation.
  • Research Question/Hypothesis
    • In critically ill adults undergoing ICU tracheal intubation with planned positive pressure ventilation between induction and laryngoscopy, does a 500 mL crystalloid bolus initiated before induction reduce cardiovascular collapse compared with no bolus?
    • Hypothesis: prophylactic pre-induction fluid administration reduces cardiovascular collapse.
  • Why This Matters
    • A prophylactic fluid bolus is simple, rapid, and low-cost; a true benefit would justify near-universal adoption, whereas no benefit supports de-implementation and fluid stewardship.
    • Clarifies whether a mechanistically plausible, widely used intervention can prevent a high-consequence complication at a narrow, high-risk procedural timepoint.
    • Informs ICU airway bundles, and highlights whether future haemodynamic trials should pivot towards vasopressor-first or physiology-guided strategies.

Design & Methods

  • Research Question: Among critically ill adults undergoing ICU tracheal intubation with planned positive pressure ventilation between induction and laryngoscopy, does a 500 mL crystalloid bolus initiated before induction reduce cardiovascular collapse compared with no bolus?
  • Study Type: Pragmatic, multicentre, parallel-group, randomised, unblinded trial conducted in 11 academic ICUs in the US; allocation stratified by ICU/site; enrolment February 2019 to May 2021; trial registration NCT03787732.
  • Population:
    • Setting: adult ICU tracheal intubations performed in participating units by operators expected to routinely intubate in that unit.
    • Inclusion: age ≥18 years; decision for tracheal intubation; planned positive pressure ventilation between induction and laryngoscopy (bag-mask ventilation and/or noninvasive ventilation during the procedure); planned use of sedative induction medication (with or without neuromuscular blockade).
    • Key exclusions: clinician judgement that a new fluid bolus was required (mandatory) or contraindicated for safe care; intubation deemed too urgent to permit trial procedures; pregnancy; prisoner/incarceration status; intubation without sedation; planned intubation without positive pressure ventilation between induction and laryngoscopy.
  • Intervention:
    • 500 mL isotonic crystalloid (type at clinician discretion) connected to existing IV/intraosseous access and started after randomisation and before induction.
    • Infused rapidly by gravity and/or pressure bag; clinicians instructed not to delay induction solely to complete infusion; infusion could continue through the procedure until the 500 mL was delivered.
  • Comparison:
    • No new fluid bolus initiated between randomisation and 2 minutes after intubation (maintenance infusions already running could continue).
    • Rescue bolus permitted if cardiovascular collapse developed, or if clinicians deemed a bolus mandatory for safe care (recorded as a protocol deviation if given outside collapse treatment in the prespecified window).
    • Other airway, induction, and haemodynamic co-interventions were not protocolised (clinician-directed).
  • Blinding: Unblinded (open-label) due to the nature of the fluid intervention; outcomes included objective components but one major component (vasopressor initiation/escalation) was clinician-driven.
  • Statistics: Initial planned sample size of 750 was calculated to detect an 8.75% absolute reduction in cardiovascular collapse (25.0% to 16.25%) with 80% power at two-sided α=0.05; a blinded sample-size re-estimation after the interim analysis increased enrolment to 1,065 to preserve power with a lower-than-anticipated event rate; primary analysis was intention-to-treat with prespecified sensitivity and subgroup analyses.3
  • Follow-Up Period: Physiologic and collapse outcomes assessed from induction to 2 minutes (vasopressor/hypotension) and to 1 hour (cardiac arrest/death); additional physiologic data to 24 hours; clinical outcomes to ICU discharge and in-hospital day 28 (or discharge, whichever occurred first).

Key Results

This trial was not stopped early. Enrolment was increased after a blinded sample-size re-estimation (per prespecified approach) rather than early termination.

Outcome Fluid bolus (n=538) No bolus (n=527) Effect p value / 95% CI Notes
Cardiovascular collapse (primary composite) 113/538 (21.0%) 96/527 (18.2%) Absolute difference 2.8% 95% CI -2.2 to 7.7; P=0.25 Composite: new/increased vasopressor or SBP <65 mm Hg (induction to 2 min) OR cardiac arrest/death (induction to 1 h)
New or increased vasopressor (composite component) 111/538 (20.6%) 93/527 (17.6%) Absolute difference 3.0% 95% CI -1.9 to 7.9 Between induction and 2 minutes; clinician-directed initiation/escalation
SBP <65 mm Hg (composite component) 21/535 (3.9%) 22/524 (4.2%) Absolute difference -0.3% 95% CI -2.8 to 2.3 Between induction and 2 minutes; denominators exclude missing SBP data
Cardiac arrest (composite component) 9/538 (1.7%) 8/527 (1.5%) Absolute difference 0.2% 95% CI -1.3 to 1.7 Between induction and 1 hour
Death (composite component) 4/538 (0.7%) 3/527 (0.6%) Absolute difference 0.2% 95% CI -0.9 to 1.2 Between induction and 1 hour
In-hospital death by day 28 218/538 (40.5%) 223/527 (42.3%) Absolute difference -1.8% 95% CI -7.9 to 4.3; P=0.55 Secondary clinical outcome
IV fluid volume from randomisation to induction (mL) 300 (150–450) 0 (0–0) Median difference 300 mL 95% CI 300 to 400 Process separation (majority did not receive the full 500 mL before induction)
IV fluid volume from randomisation to 2 min after intubation (mL) 500 (300–500) 0 (0–0) Median difference 500 mL 95% CI 499 to 500 Strong process separation by the primary outcome window
Lowest SBP (mm Hg) 116 (93–139) 113 (95–134) Median difference 3.0 95% CI -3.0 to 7.0 Between induction and 2 minutes; SBP missing in 6 patients
Change in SBP from induction to nadir (mm Hg) -7 (-26 to 0) -9 (-27 to 0) Median difference 2.0 95% CI -2.0 to 5.0 Negative values indicate a fall in SBP
Oxygen saturation <80% 79/531 (14.9%) 71/518 (13.7%) Absolute difference 1.2% 95% CI -3.3 to 5.6 Between induction and 2 minutes; denominators exclude missing saturation data
Invasive mechanical ventilation–free days (through day 28) 14 (0–25) 12 (0–25) Median difference 2.0 95% CI -10.0 to 15.0 Exploratory clinical outcome
ICU-free days (through day 28) 9 (0–22) 9 (0–22) Median difference -0.5 95% CI -9.0 to 9.5 Exploratory clinical outcome
  • Despite strong separation in early fluid exposure (median 500 mL vs 0 by 2 minutes), the 500 mL bolus did not reduce cardiovascular collapse (21.0% vs 18.2%; 95% CI compatible with modest benefit or harm).
  • The composite outcome was numerically higher in the bolus group, largely driven by vasopressor initiation/escalation rather than severe hypotension or arrest.
  • No signal of downstream clinical benefit or harm was observed in mortality or ICU/ventilator-free days (with wide uncertainty for small effects).

Internal Validity

  • Randomisation and Allocation:
    • Randomised 1:1 with stratification by ICU/site and variable block sizes; allocation concealment used sequentially numbered, opaque envelopes opened after eligibility confirmed.
    • Envelope-based concealment reduces selection bias but remains more vulnerable than fully centralised allocation if local processes are imperfect.
  • Drop out or exclusions:
    • 1067 randomised; 2 excluded after randomisation because they were ineligible (incarcerated); 1065 included in the primary analysis.
    • SBP data were missing for 6 patients for the SBP <65 component; all 6 met the composite outcome via other components.
  • Performance/Detection Bias:
    • Unblinded design; vasopressor initiation/escalation (20.6% vs 17.6%) is partly discretionary and could be influenced by clinicians’ perceptions of fluid status and trial assignment.
    • Other composite components were objective (SBP threshold, cardiac arrest, death), but these were relatively infrequent (SBP <65 ~4%; cardiac arrest ~1.6%).
  • Protocol Adherence:
    • High adherence: 535/538 (99.4%) in the fluid-bolus group received the assigned bolus; 521/527 (98.9%) in the no-bolus group received no bolus (crossovers: 3 vs 6).
    • Target physiological context achieved: positive pressure ventilation between induction and laryngoscopy occurred in 526/538 (97.8%) vs 513/527 (97.3%).
  • Baseline Characteristics:
    • Groups were broadly similar: age 61 (50–70) vs 62 (50–70) years; APACHE II 20 (14–25) vs 18 (14–25); vasopressor use within 1 hour before enrolment 107/534 (20.0%) vs 102/525 (19.4%); sepsis or septic shock 312/538 (58.0%) vs 318/527 (60.3%).
    • Baseline procedural physiology was similar: SBP at induction 128 (110–147) vs 126 (110–145) mm Hg; oxygen saturation at induction 99 (96–100) vs 99 (96–100) %.
  • Heterogeneity:
    • Pragmatic delivery across multiple ICUs increases clinical heterogeneity (induction agents, vasopressor strategy, monitoring modality), potentially diluting a true effect confined to specific physiological phenotypes.
    • Prespecified subgroup analyses did not identify a clear, reproducible subgroup with benefit (exploratory).
  • Timing:
    • Intervention started before induction by design, but the median pre-induction volume was 300 (150–450) mL, indicating that many patients did not receive the full 500 mL before induction.
    • Primary haemodynamic window (induction to 2 minutes) captures immediate instability, but may miss later hypotension occurring after early post-intubation stabilisation.
  • Dose:
    • Fixed 500 mL bolus is pragmatic but not physiology-guided; it may be insufficient in preload-dependent shock and unnecessary or harmful in volume-intolerant states (which were partly excluded by clinician judgement).
  • Separation of the Variable of Interest:
    • Randomisation-to-2-minute fluid exposure: 500 (300–500) vs 0 (0–0) mL; median difference 500 mL (95% CI 499 to 500).
    • Rescue fluid in the outcome window occurred in both arms (20 vs 31 patients received a new bolus between induction and 2 minutes), which could attenuate separation in the most unstable patients.
  • Key Delivery Aspects:
    • Airway co-interventions were similar: etomidate use 76.8% vs 78.9% (median dose 20 mg in both); neuromuscular blocker use 94.6% vs 93.5%; prophylactic vasopressor use 12.3% vs 11.8%.
    • Preoxygenation strategies and pre-procedural respiratory support were broadly similar, supporting that haemodynamic differences were not confounded by major airway practice imbalances.
  • Crossover:
    • Crossovers were uncommon (3 in bolus group; 6 in no-bolus group), limiting dilution of treatment assignment.
  • Adjunctive therapy use:
    • Induction agents, neuromuscular blockade, and use of prophylactic vasopressors were closely balanced, reducing risk that differential co-interventions explain the null result.
  • Outcome Assessment:
    • Composite included objective endpoints (SBP threshold, arrest, death) but relied heavily on clinician-directed vasopressor decisions; blood pressure measurement modality varied (arterial line vs cuff), consistent with pragmatic design.
  • Statistical Rigor:
    • Primary analysis was intention-to-treat; confidence intervals were reported for key outcomes; the trial was powered for a relatively large absolute risk reduction and may not exclude smaller but clinically relevant effects.
    • Blinded sample-size re-estimation preserved statistical power without inflating type I error when performed as prespecified.

Conclusion on Internal Validity: Overall, internal validity appears moderate to strong, supported by randomisation, minimal attrition, and strong exposure separation; the main limitation is the unblinded, clinician-influenced vasopressor component of the primary composite and the incomplete delivery of the full bolus before induction in many patients.

External Validity

  • Population Representativeness:
    • Represents ICU intubations in academic centres with high uptake of modern supportive practices (high neuromuscular blocker use, frequent noninvasive ventilation/HFNC use around intubation).
    • Clinician equipoise was required: patients in whom clinicians deemed a bolus mandatory or contraindicated were excluded, limiting applicability to extreme hypovolaemia or marked volume intolerance.
  • Applicability:
    • Intervention is highly feasible in most ICUs (500 mL crystalloid via existing access) and requires minimal additional resources.
    • Findings apply specifically to intubations with planned positive pressure ventilation between induction and laryngoscopy; generalisability to ED, theatre, prehospital settings, or intubations without intra-procedural positive pressure ventilation is uncertain.

Conclusion on External Validity: External validity is moderate for adult academic ICU intubations with planned positive pressure ventilation and clinician equipoise about fluids, but is limited for patients at physiological extremes and for non-ICU or non-PPV intubation contexts.

Strengths & Limitations

  • Strengths:
    • Large pragmatic multicentre RCT in a high-risk procedural window with minimal loss to follow-up.
    • Clear biological rationale and targeted enrolment of patients with planned positive pressure ventilation (97% received PPV between induction and laryngoscopy).
    • Strong separation in early fluid exposure (median 500 mL vs 0 by 2 minutes) and minimal crossover.
  • Limitations:
    • Open-label design with a primary composite substantially influenced by clinician vasopressor decisions (susceptible to performance/detection bias).
    • Many patients received less than the full bolus before induction (median 300 mL pre-induction), potentially under-testing “pre-induction loading” as practised in some bundles.
    • Fixed-volume intervention without physiology-guided selection; exclusion of patients with mandated bolus or contraindication may reduce detectable benefit in the most fluid-responsive phenotypes.
    • Powered for a relatively large absolute risk reduction; confidence intervals do not exclude smaller effects.

Interpretation & Why It Matters

  • Clinical practice
    • A routine 500 mL crystalloid bolus initiated before induction should not be expected to prevent peri-intubation cardiovascular collapse in ICU patients with planned positive pressure ventilation.
    • Given the composite was largely driven by vasopressor initiation, the results support an approach prioritising readiness for vasopressors and careful induction strategy over universal prophylactic fluids.
  • Mechanistic insight
    • Even with strong overall fluid separation by 2 minutes, physiological outcomes (lowest SBP; change in SBP) did not show a clinically meaningful improvement, suggesting that immediate peri-induction instability is not reliably prevented by a fixed bolus alone.
    • Induction hypotension likely reflects multiple interacting factors (vasoplegia, myocardial dysfunction, ventilatory mechanics, sedation depth), not all modifiable through preload augmentation.
  • Trial design implications
    • Future trials may need physiology-guided selection (e.g., preload dependence, ventricular function) and/or alternative haemodynamic strategies (e.g., prophylactic vasopressors) rather than universal fluid boluses.

Controversies & Subsequent Evidence

  • Composite endpoint sensitivity to clinician behaviour: The primary outcome was dominated by “new or increased vasopressors” (≈18–21%), raising concern that open-label clinician thresholds could meaningfully influence event classification and dilute an underlying physiological effect on true hypotension.4
  • Intervention timing versus intended mechanism: The median pre-induction volume was 300 mL, so many patients did not receive the full 500 mL before induction; critique focused on whether PREPARE II tested “pre-induction loading” or “peri-induction fluid exposure”.5
  • Population selection and physiology mismatch: Excluding patients where clinicians judged a bolus mandatory/contraindicated improves safety and equipoise but may preferentially remove patients most likely to benefit (clear hypovolaemia/preload dependence) or be harmed (overt volume intolerance), potentially attenuating effect size.6
  • Induction medication strategy as an interacting determinant: Critique highlighted fixed dosing patterns (e.g., etomidate median 20 mg in both arms) and the broader issue that sedation depth and vasoplegia may overwhelm any modest haemodynamic effect of a 500 mL bolus, particularly in septic physiology.6
  • Investigator response: The authors emphasised high protocol adherence, strong early fluid separation by 2 minutes, and that the sample-size re-estimation was performed blinded to preserve power; they also clarified that the enrolled cohort remained clinically high risk (≈60% sepsis/septic shock; ≈20% on vasopressors prior to enrolment).7
  • Guidelines after PREPARE II: Subsequent Society of Critical Care Medicine guidance for rapid sequence intubation in critically ill adults emphasised individualised haemodynamic optimisation and does not endorse routine prophylactic fluid boluses for all patients, consistent with the PREPARE II findings.8
  • Subsequent syntheses: Systematic reviews evaluating pre-intubation fluid infusion strategies have not demonstrated a consistent reduction in peri-intubation haemodynamic collapse with routine fixed-volume fluids, reinforcing the likelihood of strong treatment-effect heterogeneity and the need for targeted strategies.9

Summary

  • In 1,065 ICU patients undergoing tracheal intubation with planned positive pressure ventilation between induction and laryngoscopy, a 500 mL crystalloid bolus did not reduce cardiovascular collapse (21.0% vs 18.2%; absolute difference 2.8%; 95% CI -2.2 to 7.7; P=0.25).
  • Severe hypotension (SBP <65 mm Hg) and cardiac arrest were uncommon (~4% and ~1.6%); the composite was largely driven by vasopressor initiation/escalation.
  • There was no evidence of improved clinical outcomes, including in-hospital mortality by day 28 (40.5% vs 42.3%; absolute difference -1.8%; 95% CI -7.9 to 4.3; P=0.55).
  • Protocol adherence and exposure separation were strong by 2 minutes (median 500 mL vs 0), but the median volume delivered before induction was 300 mL, raising mechanistic and implementation questions.
  • Findings support individualised haemodynamic preparation (including vasopressor strategy) rather than universal prophylactic fluids for ICU intubation in patients where clinicians have equipoise about bolus administration.

Overall Takeaway

PREPARE II provides high-quality evidence that a routine 500 mL crystalloid bolus initiated before induction does not prevent cardiovascular collapse during ICU tracheal intubation when positive pressure ventilation is planned between induction and laryngoscopy. The trial challenges a common airway-bundle practice and shifts attention towards individualised haemodynamic optimisation and alternative strategies (including vasopressors) rather than universal fluid loading.

Overall Summary

  • In ICU intubations with planned positive pressure ventilation, a 500 mL crystalloid bolus did not reduce cardiovascular collapse and did not improve clinical outcomes.
  • Event classification was largely driven by clinician vasopressor escalation, underscoring the importance of haemodynamic strategy and potential susceptibility to performance bias in unblinded airway trials.
  • The findings support de-implementation of routine prophylactic boluses and prioritisation of tailored haemodynamic preparation (including vasopressor readiness) before induction.

Bibliography