Skip to main content

Foundational Trials

PrePARE

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Effect of a fluid bolus on cardiovascular collapse among critically ill adults undergoing tracheal intubation (PrePARE): a randomised controlled trial
  • Acronym: PrePARE (Preventing cardiovascular collaPse with Administration of fluid REsuscitation before tracheal intubation)
  • Year: 2019
  • Journal published in: The Lancet Respiratory Medicine
  • Citation: Janz DR, Casey JD, Semler MW, Russell DW, Dargin J, Vonderhaar DJ, et al. Effect of a fluid bolus on cardiovascular collapse among critically ill adults undergoing tracheal intubation (PrePARE): a randomised controlled trial. Lancet Respir Med. 2019;7(12):1039-1047.

Context & Rationale

  • Background
    • Tracheal intubation is common in critical care and is frequently complicated by acute hypotension, cardiac arrest, or death during or immediately after the procedure.
    • Pathophysiological plausibility for haemodynamic collapse includes induction-related vasodilatation/myocardial depression, loss of endogenous catecholamines, and a precipitous reduction in venous return with the onset of positive-pressure ventilation.
    • Pre-induction fluid bolus administration is embedded in many airway “bundles” and checklists, but prior evidence was indirect (bundle studies and observational cohorts) and could not isolate the effect of fluid bolus from co-interventions.
    • Equipoise was high because an unnecessary bolus could plausibly worsen oxygenation (pulmonary oedema), increase positive fluid balance, and distract from other haemodynamic mitigations (vasopressors, induction strategy).
    • Subsequent international observational evidence reinforced that peri-intubation adverse events (including cardiovascular instability) remain common in ICU practice and are associated with worse outcomes.1
  • Research Question/Hypothesis
    • Does initiation of a 500 mL isotonic crystalloid bolus immediately before induction reduce cardiovascular collapse compared with no bolus in critically ill adults undergoing tracheal intubation?
  • Why This Matters
    • The intervention is simple, cheap, and widely used; even a modest reduction in peri-intubation collapse could translate into meaningful reductions in cardiac arrest and downstream organ injury at scale.
    • Conversely, de-implementation of ineffective fluid boluses could reduce avoidable fluid exposure and refocus clinicians on physiology-directed strategies (vasopressors, ventilation strategy, drug choice).
    • The trial directly tests a single “bundle element”, improving causal inference compared with bundle/checklist studies.

Design & Methods

  • Research Question: In critically ill adults requiring tracheal intubation, does starting a 500 mL crystalloid bolus before induction reduce peri-intubation cardiovascular collapse compared with not starting a bolus?
  • Study Type: Pragmatic, multicentre, unblinded, parallel-group randomised controlled trial; nine sites across the USA (eight ICUs and one emergency department).
  • Population:
    • Setting: Six medical ICUs, one trauma ICU, one neurological ICU, and one emergency department at tertiary-care centres in the USA.
    • Inclusion: Critically ill adults (≥18 years) undergoing tracheal intubation at participating sites.
    • Key exclusions: Awake intubation planned; intubation required too immediately to permit randomisation; clinician judged a fluid bolus was required or contraindicated for optimal care; prisoners; pregnancy.
    • Screening/enrolment: 537 assessed; 511 met inclusion criteria; 337 randomised (168 fluid bolus; 169 no fluid bolus).
    • Co-enrolment: At seven sites, co-enrolment permitted in an independent RCT (PreVent) comparing prophylactic bag-mask ventilation vs no prophylactic ventilation.
  • Intervention:
    • What: Initiation of an IV bolus of 500 mL isotonic crystalloid (operator’s choice; typically normal saline or lactated Ringer’s).
    • How delivered: Bedside nurse connected the 500 mL bag to IV/IO access; infused by gravity and bag pressure; bolus continued until 500 mL infused.
    • Timing: Bolus commenced before induction; operator could start induction medications at any time after bolus initiation (bolus not required to be completed before induction).
    • Co-interventions: All other intubation aspects (drug choice, ventilation strategy, laryngoscopy, vasopressors) at clinician discretion, except for PreVent assignment where applicable.
  • Comparison:
    • What: No study-initiated crystalloid bolus between enrolment and 2 min after intubation completion.
    • Permitted care: Pre-existing infusions were continued; clinicians could administer fluids at any time for treatment of cardiovascular collapse (not a protocol violation) or if felt required for patient safety (recorded as protocol violation if within the restricted window and without collapse).
  • Blinding: Unblinded to clinicians/patients/study staff; periprocedural endpoints collected by independent observers present in-room but not participating in the procedure; primary investigators validated observer accuracy in a convenience sample (agreement 100% for primary outcome components in that sample).
  • Statistics: A total of 500 patients were planned to detect a 10% absolute reduction in cardiovascular collapse (from 25% to 15%; relative risk reduction 40%) with 80% power at a two-sided 5% significance level; primary analysis was unadjusted intention-to-treat comparison of proportions (χ² test) with prespecified secondary analyses (including interaction testing in logistic regression); one planned interim analysis with prespecified stopping rules for efficacy, safety, and futility.
  • Follow-Up Period: Primary outcome components captured between induction and 2 min post-intubation (SBP/vasopressors) and within 1 h (cardiac arrest/death); clinical outcomes through 28 days (ICU-free and ventilator-free days, in-hospital mortality); safety outcomes in the 24 h after intubation and through 3 days after enrolment (respiratory support variables, cumulative fluids and diuretics).

Key Results

This trial was stopped early. Enrolment stopped on Jan 9, 2018 after the single planned interim analysis (complete data from the first 250 patients) met prespecified futility criteria.

Outcome Fluid bolus (500 mL crystalloid) No fluid bolus Effect p value / 95% CI Notes
Cardiovascular collapse (primary composite, ≤1 h) 33/168 (20%) 31/169 (18%) Absolute difference 1.3% 95% CI −7.1 to 9.7; P=0.76 Composite: new SBP <65 mm Hg (induction–2 min) OR new/increased vasopressor (induction–2 min) OR cardiac arrest/death (≤1 h)
Death within 1 h 2/168 (1%) 1/169 (1%) Absolute difference 0.6% 95% CI −1.5 to 2.7; P=0.55 Component of primary composite
Cardiac arrest within 1 h 7/168 (4%) 2/169 (1%) Absolute difference 3.0% 95% CI −0.2 to 6.3; P=0.08 Numerical imbalance; not statistically significant
New SBP <65 mm Hg (induction–2 min) 11/168 (7%) 10/169 (6%) Absolute difference 0.7% 95% CI −4.5 to 5.9; P=0.79 Component of primary composite
New or increased vasopressor (induction–2 min) 32/168 (19%) 31/169 (18%) Absolute difference 0.7% 95% CI −7.6 to 9.0; P=0.86 Component of primary composite; clinician-driven escalation
Lowest SBP (induction–2 min), mm Hg 119 (97–144) 123 (97–146) Mean difference −1.2 95% CI −8.4 to 6.1; P=0.85 Median (IQR) shown; between-group estimate reported as mean difference
Lowest arterial oxygen saturation (induction–2 min), % 94 (82–98) 95 (82–99) Mean difference −0.6 95% CI −3.4 to 2.2; P=0.67 No signal of improved oxygenation with bolus
Arterial oxygen saturation <80% (induction–2 min) 28/168 (17%) 33/169 (19%) Absolute difference −3.1% 95% CI −11.5 to 5.2; P=0.46 Prespecified secondary
Ventilator-free days (to day 28) 20 (0–25) 19 (0–25) Mean difference −0.7 days 95% CI −3.2 to 1.7; P=0.55 Exploratory clinical outcome
ICU-free days (to day 28) 16 (0–23) 14 (0–23) Mean difference −0.6 days 95% CI −3.0 to 1.8; P=0.62 Exploratory clinical outcome
In-hospital mortality 48/168 (29%) 59/169 (35%) Absolute difference −5.7% 95% CI −15.8 to 4.5; P=0.27 Not powered for mortality
Cumulative IV fluids (enrolment to 72 h), mL 2110 (1410–3118) 2030 (1320–3040) Mean difference 228 mL 95% CI −308 to 764; P=0.40 Safety endpoint; cumulative exposure converged over 72 h
Cumulative diuretic dose (0–72 h), mg furosemide equivalents 20 (0–80) 0 (0–60) Mean difference 23.4 mg 95% CI −24.5 to 71.3; P=0.34 Safety endpoint; no clear signal of clinically important fluid-related harm
  • Primary finding: Initiating a 500 mL crystalloid bolus before induction did not reduce cardiovascular collapse (20% vs 18%; absolute difference 1.3%; 95% CI −7.1 to 9.7; P=0.76).
  • Signal for limited “preload before insult” separation: In a directly observed convenience sample (n=38), median crystalloid infused before induction was 200 mL (IQR 200–325; mean 262 mL) in the bolus arm vs 0 mL (IQR 0–0; P<0.0001).
  • Heterogeneity of treatment effect with positive-pressure ventilation:
    • Non-invasive ventilation for preoxygenation: cardiovascular collapse 5/44 (11%) bolus vs 11/37 (30%) no bolus; RR 0.51; 95% CI 0.24 to 1.09; Pinteraction=0.032.
    • Bag-mask ventilation after induction but before intubation: 9/58 (16%) bolus vs 18/64 (28%) no bolus; RR 0.61; 95% CI 0.33 to 1.13; Pinteraction=0.008.
    • Among co-enrolled PreVent participants (n=201): interaction between PrePARE assignment and PreVent assignment; in those randomised to prophylactic bag-mask ventilation: 11/48 (23%) bolus vs 23/57 (40%) no bolus; RR 0.63; 95% CI 0.40 to 0.99; Pinteraction=0.021.

Internal Validity

  • Randomisation and Allocation:
    • Computer-generated permuted blocks (2/4/6), stratified by site; allocation concealed in opaque envelopes until enrolment decision made.
    • Pragmatic enrolment at bedside; clinicians could exclude patients (urgency; fluid indicated/contraindicated), creating a selected “equipoise” cohort.
  • Drop out or exclusions (post-randomisation):
    • 0 lost to follow-up.
    • Protocol delivery: 165/168 (98%) in bolus arm received full 500 mL; 3/168 (2%) received none (urgency); in control, 2/169 (1%) received a bolus due to haemodynamic decompensation.
    • Per-protocol population: 332 patients; cardiovascular collapse 31/165 (18.8%) vs 31/167 (18.6%); P=0.95 (consistent with ITT).
  • Performance/Detection Bias:
    • Unblinded intervention; clinician behaviours (especially vasopressor initiation/escalation) could plausibly be influenced by group assignment.
    • Outcome collection by trained independent observers; in a convenience sample, agreement between independent observer and investigator for primary outcome components was 100%.
  • Protocol Adherence and Separation of the Variable of Interest:
    • High adherence to assignment (98% received bolus vs 99% did not).
    • Observed pre-induction crystalloid: median 200 mL (IQR 200–325; mean 262 mL) vs 0 mL (IQR 0–0).
    • Downstream fluid exposure converged: cumulative IV fluids to 72 h 2110 (1410–3118) mL vs 2030 (1320–3040) mL (mean difference 228 mL; 95% CI −308 to 764; P=0.40).
  • Baseline Characteristics and Illness Severity:
    • Groups broadly comparable: age 61 (47–70) vs 58 (46–68) years; APACHE II 21 (14–27) vs 20 (16–27); vasopressor receipt in 6 h pre-intubation 28/168 (17%) vs 28/169 (17%); NIV in 6 h pre-intubation 44/168 (26%) vs 37/169 (22%).
    • Indication imbalance (pragmatic enrolment): hypoxic respiratory failure 85/168 (51%) vs 69/169 (41%).
  • Heterogeneity:
    • Prespecified interaction testing identified qualitative effect modification with positive-pressure ventilation (NIV preoxygenation; bag-mask ventilation after induction; and PreVent assignment among co-enrolled patients).
    • Multiple subgroup analyses without multiplicity adjustment increase the likelihood of chance findings; however, the direction of interaction is physiologically plausible (preload dependence under positive intrathoracic pressure).
  • Timing:
    • Bolus “commenced” before induction (completion not required); the observed pre-induction volume (median 200 mL) suggests limited time for preload augmentation before the haemodynamic stressor.
  • Dose:
    • Fixed 500 mL crystalloid dose regardless of baseline preload, fluid responsiveness, or capillary leak physiology.
    • Clinicians excluded patients where fluid was clearly indicated or contraindicated, further narrowing the likely responder pool.
  • Key Delivery Aspects:
    • Procedural characteristics were similar: first-attempt success 148/164 (90%) vs 152/167 (91%); median laryngoscopy attempts 1 (IQR 1–1) in both groups; median time from induction to intubation 70 (40–102) s vs 70 (39–105) s.
    • Co-enrolment with PreVent was common: 99/168 (59%) vs 102/169 (60%).
  • Outcome Assessment:
    • Primary outcome definition combined objective haemodynamic threshold (SBP <65 mm Hg) with a clinician-mediated component (new/increased vasopressor), plus hard clinical events (cardiac arrest/death).
    • The composite increases event rate but mixes mechanistic and decision-mediated elements; interpretability depends on whether components move concordantly (they did not).
  • Statistical Rigor:
    • Preplanned interim analysis with prespecified futility boundary; trial stopped early, reducing precision and power for modest effects.
    • At interim (n=250), P=0.93 met futility boundary (P>0.60); conditional power to detect the prespecified 40% relative risk reduction if continued to n=500 was 11%; using all 337 enrolled, conditional power was 0.6%.

Conclusion on Internal Validity: Moderate. Randomisation and outcome ascertainment were robust, and crossover was minimal, but the unblinded design, early stopping, and limited pre-induction volume separation (median 200 mL delivered before induction) constrain the ability to exclude smaller effects and complicate mechanistic inference for “true” preload optimisation.

External Validity

  • Population Representativeness:
    • Enrolled a pragmatic ICU/ED cohort in tertiary-care US centres with typical indications (hypoxaemic respiratory failure, altered mental status) and moderate illness severity (APACHE II ~20–21).
    • However, selection was strongly influenced by clinician judgement: of 511 eligible, 71 excluded due to urgency, and 88 excluded because clinicians felt fluid should be given (33) or withheld (55), limiting representation of patients with clear hypovolaemia/shock or overt volume overload.
  • Applicability:
    • Most applicable to the “grey zone” peri-intubation population where clinicians perceive equipoise about fluid bolus administration.
    • Less applicable to settings where intubation urgency precludes trial enrolment (e.g., peri-arrest, severe hypoxaemia with imminent collapse), and to patients in whom fluid is clearly indicated (e.g., profound hypovolaemic shock) or contraindicated (e.g., acute cardiogenic pulmonary oedema).
    • Transferability across health systems is reasonable for similar ICU staffing and rapid access to vasopressors; effects may differ where haemodynamic rescue is delayed or fluid type/availability differs.

Conclusion on External Validity: Generalisability is moderate for elective/controlled ICU and ED intubations in resourced settings where equipoise about fluids exists, but is limited for the sickest and most fluid-sensitive patients who were systematically excluded.

Strengths & Limitations

  • Strengths:
    • Pragmatic randomised design isolating one common airway “bundle” element.
    • Multicentre enrolment across ICU subtypes and an ED; high protocol adherence and complete follow-up.
    • Clinically meaningful composite endpoint anchored to an objective haemodynamic threshold and hard events.
    • Independent in-room observers and validation of outcome ascertainment accuracy.
    • Prespecified subgroup framework including biologically plausible interaction with positive-pressure ventilation.
  • Limitations:
    • Stopped early for futility; reduced precision, limited power for smaller effect sizes, and reduced ability to robustly evaluate subgroup interactions.
    • Unblinded; vasopressor components of the composite are clinician-mediated and potentially susceptible to performance bias.
    • Intervention was initiation of a bolus, not completion before induction; observed median pre-induction delivery was only 200 mL.
    • Selection of an “equipoise” cohort (excluding urgency and clear fluid-indication/contraindication) may have removed patients most likely to benefit or be harmed.
    • Substantial co-enrolment with PreVent complicates interpretation where biological interaction between positive-pressure ventilation and preload is plausible.

Interpretation & Why It Matters

  • Routine prophylactic fluids
    • For most critically ill adults in whom clinicians have equipoise about fluid administration, a routine 500 mL crystalloid bolus commenced before induction did not reduce cardiovascular collapse.
    • Given common fluid stewardship goals, a default “bolus for everyone” approach before ICU intubation is not supported.
  • Mechanism-based practice
    • The trial underscores that peri-intubation collapse is multifactorial; a small-volume bolus started immediately before induction may be insufficient to modify physiology meaningfully.
    • Haemodynamic optimisation likely requires individualisation: assessment of preload responsiveness, vasopressor readiness, induction agent selection, and ventilation strategy.
  • Positive-pressure ventilation interaction
    • Effect modification with NIV/bag-mask ventilation is biologically coherent (increased intrathoracic pressure may render patients more preload-dependent), but the subgroup findings are not definitive given early stopping and multiple comparisons.

Controversies & Subsequent Evidence

  • Early stopping, precision, and the “null” interpretation:
    • The trial stopped at the first interim analysis for futility; this appropriately avoided further enrolment under low conditional power but reduced precision around clinically smaller benefits or harms.
    • The confidence interval for the primary outcome (−7.1% to 9.7%) leaves uncertainty for modest benefit or harm; the cardiac arrest component numerically favoured no bolus (4% vs 1%; P=0.08).
  • Timing and biological potency of the tested exposure:
    • The tested strategy was bolus initiation before induction, not necessarily completion; in observed cases the median pre-induction delivery was 200 mL, raising concern that the intervention did not reliably increase preload before the haemodynamic stressor.
    • Editorial critique highlighted that clinician exclusions (urgency; fluid indicated/contraindicated) likely removed patients most sensitive to fluid loading, increasing the probability of a null overall estimate.2
  • Co-enrolment with PreVent and interpretability of interaction signals:
    • 201/337 patients were co-enrolled in PreVent; the identified interaction between positive-pressure ventilation and fluid bolus creates plausible biological interdependence and complicates causal interpretation of subgroup effects.
    • Published commentary noted that, where interaction is plausible, co-enrolment without factorial design can undermine interpretability of each intervention’s effect estimate.2
  • Follow-up RCT evidence:
    • PREPARE II (JAMA, 2022) evaluated a similar question in a larger pragmatic trial and likewise did not demonstrate a reduction in cardiovascular collapse with pre-induction fluid bolus administration.3
  • Systematic reviews and guideline incorporation:
    • Recent systematic review/meta-analysis synthesising interventions aimed at peri-intubation hypoxaemia/hypotension concluded that evidence for routine pre-intubation fluid boluses is not robust and should be interpreted in the context of patient physiology and co-interventions.4
    • Society of Critical Care Medicine clinical practice guidelines for rapid sequence intubation incorporated this trial era’s evidence base and emphasise physiology-directed haemodynamic preparation rather than routine fluid bolus for all patients.5

Summary

  • In this pragmatic US ICU/ED RCT, starting a 500 mL crystalloid bolus before induction did not reduce peri-intubation cardiovascular collapse (20% vs 18%).
  • The trial stopped early for futility at the single planned interim analysis, limiting precision for modest effects and subgroup credibility.
  • Protocol adherence was high, but pre-induction separation was modest: median 200 mL infused before induction in observed cases vs 0 mL.
  • Subgroup analyses suggested interaction with positive-pressure ventilation (NIV preoxygenation or bag-mask ventilation), but these findings are hypothesis-generating.
  • Subsequent larger trial evidence and contemporary guidelines support abandoning routine prophylactic fluid bolus for all peri-intubation patients in favour of physiology-driven haemodynamic preparation.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • The DSMB’s futility decision was supported by very low conditional power (11% if continued to n=500 at interim; 0.6% using all n=337), reflecting both the observed effect size near zero and reduced ability to detect the prespecified 40% relative risk reduction.

Overall Takeaway

PrePARE demonstrated that, in an “equipoise” cohort of critically ill adults undergoing tracheal intubation, routinely commencing a 500 mL crystalloid bolus before induction did not reduce peri-intubation cardiovascular collapse and did not produce clear downstream benefit. The trial’s pragmatic design, objective haemodynamic endpoint, and subsequent corroboration by later evidence have helped shift practice away from default fluid loading and towards physiology-based haemodynamic preparation.

Overall Summary

  • Routine 500 mL pre-induction fluid bolus: no reduction in cardiovascular collapse (20% vs 18%).
  • High protocol adherence but modest pre-induction separation (median 200 mL delivered before induction in observed cases).
  • Hypothesis-generating interaction with positive-pressure ventilation; not definitive due to early stopping and multiple comparisons.

Bibliography