Skip to main content

Foundational Trials

FAST

Image: Shutterstock / luchschenF

Publication

  • Title: Screening Frequency and Spontaneous Breathing Trial Techniques
  • Acronym: FAST-NAWC (Frequency of Screening and Spontaneous Breathing Trial Technique Trial—North American Weaning Collaboration)
  • Year: 2024
  • Journal published in: JAMA
  • Citation: Burns KEA, Wong J, Rizvi L, Lafreniere-Roula M, Thorpe KE, Devlin JW, et al. Screening Frequency and Spontaneous Breathing Trial Techniques: A Randomized Clinical Trial. JAMA. 2024;332(21):1808-1821.

Context & Rationale

  • Background
    • Liberation from invasive mechanical ventilation can occupy a large proportion of ventilator time, and delays plausibly increase ventilator-associated complications and resource use.
    • Routine practice includes readiness screening followed by spontaneous breathing trials (SBTs), yet the “dose” (how often screening is mandated) and the optimal SBT technique remain contested.
    • Two widely used SBT approaches differ in physiological loading: low-level pressure-supported SBTs (often with PEEP) vs unsupported T-piece trials; clinical trials before FAST yielded mixed signals, and practice variation persisted.
    • Guideline panels have permitted several SBT approaches and prioritised protocolised liberation, but left uncertainty regarding whether increased screening frequency accelerates successful extubation. 7
  • Research Question/Hypothesis
    • Does protocolised more frequent screening (vs once-daily screening) and/or pressure-supported SBT with PEEP (vs T-piece SBT) shorten time to successful extubation in invasively ventilated adults?
    • The factorial design assumed no mechanistic reason for interaction between screening frequency and SBT technique (i.e., independence of main effects). 12
  • Why This Matters
    • Even modest reductions in time to sustained unsupported breathing could reduce exposure to sedation, delirium risk, ventilator-associated infections, ICU length of stay, and downstream costs.
    • “Liberation bundles” are widely implemented; FAST tested two modifiable components that are routinely protocolised and are feasible targets for quality improvement.
    • Factorial trials can efficiently test multiple strategies, but only if interactions are rare or interpretable—making FAST methodologically important for trialists as well as clinicians.

Design & Methods

  • Research Question: In invasively ventilated adults, do more frequent readiness screening (vs once daily) and/or pressure-supported SBTs with PEEP (vs T-piece SBTs) reduce time to successful extubation (sustained unsupported breathing for ≥48 hours)?
  • Study Type: Multicentre, pragmatic, open-label, 2×2 factorial randomised clinical trial conducted in 23 North American ICUs; central randomisation stratified by centre using variable block sizes.
  • Population:
    • Adults (≥18 years in US; ≥16 years in Canada) receiving invasive mechanical ventilation for >24 hours; able to trigger breaths; enrolled early to allow protocolised screening and SBT application.
    • Key physiological entry thresholds included FiO2 ≤0.70 and PEEP ≤12 cm H2O.
    • Key exclusions included mechanical ventilation ≥2 weeks, tracheostomy, profound neurological impairment (GCS ≤6), or prior SBT at equivalent settings (to preserve “first-test” integrity).
    • Enrolment was operationally constrained to weekday daytime (reflecting the protocolised screening workflow).
  • Intervention:
    • Screening frequency factor: “More frequent” protocolised screening at least twice daily (06:00–08:00 and 13:00–15:00), with additional screening permitted at clinician discretion; missed screens could be completed within 6 hours.
    • SBT technique factor: Pressure-supported SBTs that included PEEP, with pressure support >0 to ≤8 cm H2O and PEEP >0 to ≤5 cm H2O; SBT duration 30–120 minutes.
  • Comparison:
    • Screening frequency factor: Once-daily screening (06:00–08:00).
    • SBT technique factor: T-piece SBTs (unsupported breathing via T-piece); during COVID-19–positive cases, T-piece SBTs could be performed while connected to a ventilator circuit without inspiratory support to minimise aerosolisation (protocol update). 2
  • Blinding: Unblinded clinicians and bedside staff; extubation decisions were clinician-driven within protocolised screening/SBT structure.
  • Statistics: Power calculation: 760 patients were required to detect a 1-day median reduction in time to successful extubation (from 5.0 to 4.0 days) with 80% power at a 5% two-sided significance level (assumed no interaction); primary analysis used a modified intention-to-treat approach; time-to-event analyses used Cox modelling with a prespecified interaction test (pairwise contrasts used if interaction detected). 12
  • Follow-Up Period: In-hospital outcomes through day 60 (including 90-day mortality reporting); selected patient-centred outcomes assessed at 6 months (EuroQol-5D and Functional Independence Measure) in those alive and providing consent.

Key Results

This trial was not stopped early. Recruitment exceeded the prespecified target (797 analysed vs target 760).

Outcome Strategy A Strategy B Effect p value / 95% CI Notes
Time to successful extubation (primary); main effect (screening frequency) More frequent screening (n=395); time not reported (pooled) Once-daily screening (n=402); time not reported (pooled) HR 0.88 95% CI 0.76 to 1.03; P=0.12 Group medians (days): once-daily+PS 2.0; once-daily+T 3.1; more frequent+PS 3.9; more frequent+T 2.9 (interaction P=0.009).
Time to successful extubation (primary); main effect (SBT technique) Pressure-supported SBT+PEEP (n=393); time not reported (pooled) T-piece SBT (n=404); time not reported (pooled) HR 1.06 95% CI 0.91 to 1.23; P=0.45 Primary endpoint required pairwise contrasts due to significant interaction (see below).
Time to successful extubation (primary); pairwise contrast (interaction) More frequent screening + PS SBT: median 3.9 d (95% CI 2.9 to 4.7); n=195 Once-daily screening + PS SBT: median 2.0 d (95% CI 1.7 to 2.7); n=198 HR 0.70 95% CI 0.50 to 0.96; P=0.02 Interaction P=0.009; HR <1 indicates longer time (slower extubation) in Strategy A.
Time to successful extubation (primary); pairwise contrast (within once-daily screening) Once-daily screening + PS SBT: median 2.0 d (95% CI 1.7 to 2.7); n=198 Once-daily screening + T-piece: median 3.1 d (95% CI 2.7 to 4.8); n=204 HR 1.30 95% CI 0.98 to 1.70; P=0.08 Interaction-driven contrast; HR >1 favours Strategy A (faster extubation).
Time to first successful SBT; pairwise contrast (within once-daily screening) Once-daily screening + PS SBT: median 0.8 d (95% CI 0.7 to 1.0); n=198 Once-daily screening + T-piece: median 1.1 d (95% CI 0.9 to 1.9); n=204 HR 1.4 95% CI 1.1 to 1.8; P=0.007 P value among groups=0.01; interaction P=0.04.
Duration of ventilation in ICU (hours); pairwise contrast (within PS SBT arms) More frequent screening + PS SBT: median 82.2 h (IQR 23.4 to 164.1); n=195 Once-daily screening + PS SBT: median 46.7 h (IQR 18.7 to 95.7); n=198 IRR 1.5 95% CI 1.5 to 1.6; P<0.001 P value among groups=0.02; interaction P=0.006.
ICU length of stay (days); pairwise contrast (within PS SBT arms) More frequent screening + PS SBT: median 6.2 d (IQR 3.4 to 12.4); n=195 Once-daily screening + PS SBT: median 5.0 d (IQR 2.7 to 8.2); n=198 IRR 1.2 95% CI 1.1 to 1.3; P<0.001 P value among groups<0.001; interaction P<0.001.
ICU mortality More frequent screening + PS SBT: 34/195 (17.4%) Once-daily screening + PS SBT: 31/198 (15.7%) OR 1.1 95% CI 0.6 to 2.3; P=0.96 P value among groups=0.76; interaction P=0.50.
Reintubation in the ICU Once-daily screening + PS SBT: 16/198 (8.1%) Once-daily screening + T-piece: 23/204 (11.3%) OR 0.7 95% CI 0.3 to 1.7; P=0.70 P value among groups=0.50; interaction P=0.50.
Ventilator-associated pneumonia in the ICU Once-daily screening + PS SBT: 24/198 (14.0%) Once-daily screening + T-piece: 32/204 (18.1%) OR 0.7 95% CI 0.3 to 1.6; P=0.72 P value among groups=0.18; interaction P=0.49.
  • Main effects were not statistically significant for screening frequency (HR 0.88; 95% CI 0.76 to 1.03; P=0.12) or SBT technique (HR 1.06; 95% CI 0.91 to 1.23; P=0.45), but a statistically significant interaction was present for the primary endpoint (interaction P=0.009).
  • Within pressure-supported SBT arms, protocolised more frequent screening prolonged time to successful extubation (median 3.9 vs 2.0 days; HR 0.70; 95% CI 0.50 to 0.96; P=0.02) and increased ventilation duration and ICU stay.
  • Harms were uncommon: adverse events related to screening/RSBI occurred in 2/198 (1.0%), 2/204 (1.0%), 4/195 (2.0%), and 3/200 (1.5%) across the four groups; adverse events related to SBT conduct occurred in 0/198 (0%), 5/204 (2.5%), 3/195 (1.5%), and 1/200 (0.5%) across the four groups.

Internal Validity

  • Randomisation and Allocation
    • Central randomisation stratified by centre with variable block sizes; allocation concealment was operationally robust at the point of assignment (before bedside team awareness).
    • Factorial design increased efficiency but increased dependence on the “no interaction” assumption (later violated for the primary outcome).
  • Drop out or exclusions
    • 837 patients were randomised; 797 were included in the primary analysis (modified intention-to-treat).
    • Correspondence clarified that 28 patients were excluded after randomisation and 12 were withdrawn due to lack of permission to use data, generating debate about the risk of bias from post-randomisation data loss. 45
  • Performance/Detection Bias
    • Unblinded design; clinicians controlled extubation timing (and willingness to act on screening results), a plausible pathway for interaction/behavioural effects.
    • Primary outcome was relatively objective (sustained unsupported breathing ≥48 hours), but still depends on clinician decisions (extubation/discontinuation timing; reintubation threshold).
  • Protocol Adherence
    • Separation by screening “dose” was substantial: mean screenings per patient-day were 3.9 (SD 2.7) and 4.8 (SD 4.0) in once-daily arms vs 7.9 (SD 4.9) and 7.6 (SD 4.5) in more frequent arms.
    • Major screening violations were low-to-moderate (41/803 [5.1%], 47/1016 [4.6%], 112/1694 [6.6%], 85/1633 [5.2%]); major SBT violations ranged from 3.8% to 10.2% across groups.
  • Baseline Characteristics
    • Groups were well balanced: mean age ~61–64 years; ~59% male; median SOFA 5–6; predominant indication acute respiratory failure (~55–62%).
  • Heterogeneity
    • 23 ICUs with pragmatic delivery; clinicians could vary SBT duration (30–120 minutes) and select pressure support/PEEP within ranges, potentially diluting mechanistic contrasts.
  • Timing
    • Protocolised screening windows (especially the weekday daytime operationalisation) risk systematic delays in patients improving overnight/weekend, which may affect generalisability and could interact with staffing models.
  • Dose
    • Pressure-supported SBTs allowed PS up to 8 cm H2O and PEEP up to 5 cm H2O (both >0); this may approximate clinical practice but introduces within-arm heterogeneity.
  • Separation of the Variable of Interest
    • Mean SBTs per patient-day were higher in more frequent screening arms: 3.3 (SD 2.5) and 2.6 (SD 1.7) vs 1.9 (SD 1.2) and 2.1 (SD 1.7) in once-daily arms.
  • Outcome Assessment
    • Time-to-event endpoints were analysed with survival methods; binary outcomes used effect estimates with confidence intervals and interaction testing.
  • Statistical Rigor
    • Power target was met/exceeded (797 analysed), and interaction-driven pairwise contrasts were performed once a statistically significant interaction was identified.
    • The credibility of the unexpected interaction is central to interpretation (see Controversies), and was examined using a structured interaction-credibility framework. 6

Conclusion on Internal Validity: Overall, internal validity appears moderate-to-strong: randomisation and separation of screening “dose” were clear, but lack of blinding, clinician-driven extubation decisions, and post-randomisation exclusions/withdrawals introduce plausible bias pathways—especially relevant given the unexpected interaction.

External Validity

  • Population Representativeness
    • Broad adult ICU case-mix (medical/surgical) with moderate severity; applicable to typical invasively ventilated adults beyond the first 24 hours.
    • Key exclusions (e.g., profound neurological impairment, tracheostomy, very prolonged ventilation) limit inference for neurocritical care and chronic/prolonged weaning populations.
  • Applicability
    • North American staffing models (respiratory therapist availability) may influence feasibility and effect of “more frequent” screening; interaction could behave differently in other systems.
    • Allowing broad PS/PEEP ranges and variable SBT duration increases pragmatic relevance, but makes translation to rigidly standardised SBT protocols less direct.
    • COVID-19 protocol adjustments were incorporated, but COVID-positive enrolment was small, limiting certainty for that subgroup.

Conclusion on External Validity: Generalisability is moderate for adult medical/surgical ICUs with comparable staffing and weaning infrastructure, but is limited for neurocritical care, prolonged weaning cohorts, and settings where screening frequency is constrained by workforce.

Strengths & Limitations

  • Strengths:
    • Large, multicentre, pragmatic factorial design addressing two common, protocolisable weaning strategies in routine ICU care.
    • Clinically meaningful primary outcome (sustained unsupported breathing) rather than surrogate SBT success alone.
    • Prospectively published protocol and statistical analysis plan (including a subsequent COVID-era update). 12
  • Limitations:
    • Open-label implementation with clinician-driven extubation decisions (risk of behaviour-mediated effects).
    • Modified intention-to-treat with post-randomisation exclusions/withdrawals (debated in correspondence). 45
    • Within-arm variability in PS/PEEP settings and SBT duration may reduce biological contrast between techniques.
    • 6-month outcomes were available only for a minority (consented survivors), increasing risk of attrition bias for long-term endpoints.

Interpretation & Why It Matters

  • Clinical practice
    • Protocolised “more frequent” screening should not be assumed to accelerate liberation, particularly in systems using pressure-supported SBTs with PEEP; in FAST, this combination prolonged time to successful extubation (HR 0.70) and increased ventilation hours and ICU stay.
    • Once-daily screening remains a defensible default when combined with structured SBTs, pending replication and implementation research.
  • Mechanistic inference
    • The observed interaction suggests workflow/decision coupling: increasing screening frequency may change clinician thresholds, SBT repetition, or extubation timing rather than simply revealing readiness earlier.
    • Pressure-supported SBTs (with PEEP) may “pass” earlier or be perceived as less demanding, which could paradoxically delay definitive extubation decisions when screening is frequent and SBTs are repeated.
  • Trial methodology
    • FAST is a cautionary exemplar for factorial trials in complex care pathways: unanticipated interaction can undermine interpretability of main effects and requires disciplined credibility assessment and replication planning.

Controversies & Subsequent Evidence

  • Unexpected interaction (primary endpoint)
    • The interaction (P=0.009) was not anticipated mechanistically and creates interpretive tension: the “dose” of screening cannot be evaluated independently of SBT technique.
    • Editorial commentary emphasised that the interaction might reflect complex clinician behaviour and workflow effects, advocating caution before prescriptive implementation changes. 3
    • Structured credibility assessment tools for effect modification (ICEMAN) highlight that unexpected interactions warrant scepticism unless supported by replication, consistency, and plausible mechanism. 6
  • Post-randomisation exclusions and modified intention-to-treat
    • Published correspondence challenged the post-randomisation exclusions/withdrawals and the potential for bias introduced by excluding randomised participants from analysis. 4
    • The authors’ reply clarified that 28 patients were excluded after randomisation and 12 were withdrawn due to lack of permission to use data, but acknowledged the modified intention-to-treat analytic set. 5
  • How FAST fits the SBT-technique evidence base
    • Post-FAST evidence syntheses continued to show that SBT technique effects are modest and context dependent; a large systematic review/meta-analysis suggested pressure-supported (vs T-piece) SBTs may increase extubation success without higher reintubation, while highlighting heterogeneity and sensitivity to influential trials. 10
    • A network meta-analysis of alternative SBT techniques ranked several approaches but underscored uncertainty due to trial quality, clinical heterogeneity, and indirectness—supporting the interpretation that implementation context matters. 11
  • Guideline evolution after FAST
    • Recent practice guidance continues to endorse protocolised liberation and permits multiple SBT techniques, emphasising careful selection of SBT method, monitoring, and the need to avoid unnecessary prolongation of ventilation. 89

Summary

  • FAST was a large, pragmatic 2×2 factorial ICU trial testing two protocolisable weaning strategies: screening frequency (once daily vs more frequent) and SBT technique (pressure-supported with PEEP vs T-piece).
  • Main effects were not statistically significant for time to successful extubation, but an unexpected interaction was identified (P=0.009).
  • More frequent screening combined with pressure-supported SBTs prolonged time to successful extubation (median 3.9 vs 2.0 days; HR 0.70; P=0.02) and increased ventilation duration and ICU stay.
  • Mortality and major complications (reintubation, ventilator-associated pneumonia) were similar across strategies; adverse events attributable to screening/SBT were uncommon.
  • The interaction and modified intention-to-treat analytic set have driven methodological debate, highlighting the need for replication and careful implementation research.

Notes

  • Directionality (as reported by trialists): for time outcomes, HR >1 favours the first-listed strategy (faster event); for IRR, values <1 indicate fewer hours/days (shorter duration/LOS); for OR, values <1 favour the first-listed strategy (lower odds of the outcome).

Overall Takeaway

FAST is landmark because it challenged a widely held quality-improvement intuition—that more frequent protocolised readiness screening necessarily accelerates ventilator liberation—by demonstrating an unexpected, clinically meaningful interaction with SBT technique. Its findings caution against implementing “more screening” as a standalone performance metric, and instead foreground the importance of the combined weaning strategy, clinician behaviour, and workflow in determining successful extubation timelines.

Overall Summary

  • Main effects were neutral, but a significant interaction altered interpretation of both interventions.
  • More frequent screening combined with pressure-supported SBTs prolonged extubation time and increased ventilation duration/ICU stay.
  • Harms were uncommon and mortality did not differ meaningfully across strategies; replication and implementation research remain priorities.

Bibliography