Skip to main content

Foundational Trials

DOSE VF

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Defibrillation Strategies for Refractory Ventricular Fibrillation
  • Acronym: DOSE VF
  • Year: 2022
  • Journal published in: The New England Journal of Medicine
  • Citation: Cheskes S, Verbeek PR, Drennan IR, et al. Defibrillation Strategies for Refractory Ventricular Fibrillation. N Engl J Med. 2022;387(21):1947-1956.

Context & Rationale

  • Background
    • Refractory VF/pulseless VT after multiple defibrillation attempts is associated with very low survival, despite otherwise “favourable” shockable physiology.
    • Standard practice (repeated shocks in a fixed pad vector) can fail because the delivered current vector may not adequately capture critical myocardial mass, and escalating energy alone may not overcome impedance/current pathway limitations.
    • Two pragmatic alternatives gained traction before DOSE VF:
      • Vector change (VC): changing pad position (typically to anterior–posterior) to alter the current pathway through the myocardium.
      • Double sequential external defibrillation (DSED): two shocks delivered in very rapid sequence from two defibrillators using two pad vectors, aiming to increase delivered energy/current density and/or exploit different vectors.
    • Pre-trial evidence was dominated by case series/observational designs and low-certainty syntheses, leaving equipoise around patient-centred outcomes and implementation safety1.
    • An internal pilot randomised phase demonstrated feasibility of delivering VC and DSED in paramedic services, but was not powered for definitive outcome estimates2.
  • Research Question/Hypothesis
    • In adult out-of-hospital cardiac arrest with VF/pulseless VT refractory to three standard shocks, does switching the defibrillation strategy (VC or DSED) improve survival to hospital discharge compared with continued standard defibrillation?
  • Why This Matters
    • Refractory VF is a time-critical phenotype where a small absolute survival gain translates into substantial population benefit.
    • Defibrillation strategy is a modifiable early intervention that could reduce reliance on more resource-intensive salvage pathways (e.g., ECPR) and standardise escalation after “shock failure”.
    • The trial tests a high-impact systems intervention (equipment + choreography + training), informing protocolised escalation in EMS systems.

Design & Methods

  • Research Question: Among adults with OHCA and refractory VF/pulseless VT after three standard shocks, does switching to DSED or VC (from shock 4 onwards) improve survival to hospital discharge versus continued standard defibrillation?
  • Study Type:
    • Pragmatic, cluster-randomised, crossover, three-arm trial in the prehospital setting (paramedic services as clusters).
    • Investigator-initiated, multicentre (six paramedic services), single-province (Ontario, Canada).
    • Open-label at point of care (blinding not feasible for pad placement and dual-defibrillator choreography).
  • Population:
    • Setting: Adult out-of-hospital cardiac arrest treated by participating paramedic services.
    • Core inclusion trigger: VF/pulseless VT persisting after three consecutive standard defibrillation attempts with 2-min CPR cycles between shocks.
    • Key exclusions (protocol-level): Not reported in the main manuscript in a single consolidated list; operational exclusions included inability to deliver the assigned strategy (e.g., equipment availability) or not meeting refractory shock criteria prior to randomisation.
  • Intervention:
    • DSED arm: Two defibrillators; initial anterior–lateral pads retained; additional anterior–posterior pads applied; two sequential shocks delivered <1 second apart (anterior–lateral then anterior–posterior) at maximum energy (ZOLL: 200 J; LIFEPAK: 360 J).
    • VC arm: Switch pad position to anterior–posterior and deliver subsequent shocks at maximum energy (ZOLL: 200 J; LIFEPAK: 360 J).
  • Comparison:
    • Standard defibrillation: Continued anterior–lateral pad position using a single defibrillator; shocks delivered according to device escalating energy schedules initially and maximum energy thereafter (ZOLL: 120/150/200 J for shocks 1–3; LIFEPAK: 200/300/360 J for shocks 1–3).
  • Blinding:
    • Open-label for clinicians/paramedics (pad vectors and dual-defibrillator delivery are not concealable).
    • Patient-centred endpoints (survival to discharge) are objective; neurological outcome relies on discharge assessment (susceptible to some measurement variability).
  • Statistics:
    • Power calculation (protocol): 930 patients (310 per arm) required to detect an 8% absolute increase in survival to hospital discharge (from 12% to 20%) with a two-sided alpha of 0.05 and >80% power; no multiplicity adjustment for pairwise comparisons3.
    • Analysis: Primary analysis reported as adjusted relative risks (intention-to-treat at the patient level) accounting for the cluster crossover structure (as reported in the main manuscript).
  • Follow-Up Period:
    • Through hospital discharge (primary endpoint) and neurological status at discharge (modified Rankin scale).

Key Results

This trial was stopped early. The trial was terminated during the COVID-19 pandemic after 405 patients were randomised (planned sample size 930).

Outcome Intervention (DSED or VC) Standard defibrillation Effect p value / 95% CI Notes
Survival to hospital discharge (primary) DSED: 38/125 (30.4%) 18/136 (13.3%) Adjusted RR 2.21 95% CI 1.33 to 3.67 Global three-group comparison: P=0.009; DSED vs VC: adjusted RR 1.29 (95% CI 0.81 to 2.05)
Survival to hospital discharge (primary) VC: 31/144 (21.7%) 18/136 (13.3%) Adjusted RR 1.71 95% CI 1.01 to 2.88 Fragility index (VC vs standard) for survival: 1
VF termination after first eligible shock DSED: 105/125 (84.0%) 92/136 (67.6%) Adjusted RR 1.25 95% CI 1.09 to 1.44 Physiological plausibility: higher termination rates align with survival signal
VF termination after first eligible shock VC: 115/144 (79.9%) 92/136 (67.6%) Adjusted RR 1.18 95% CI 1.03 to 1.36 Termination benefit observed despite smaller survival effect estimate
Return of spontaneous circulation (ROSC) DSED: 58/125 (46.4%) 36/136 (26.5%) Adjusted RR 1.72 95% CI 1.22 to 2.42 ROSC signal consistent with defibrillation efficacy
Return of spontaneous circulation (ROSC) VC: 51/144 (35.4%) 36/136 (26.5%) Adjusted RR 1.39 95% CI 0.97 to 1.99 Confidence interval crosses 1.0
Good neurological outcome at discharge (mRS ≤2) DSED: 34/124 (27.4%) 15/134 (11.2%) Adjusted RR 2.21 95% CI 1.26 to 3.88 Denominators differ due to missing neurological outcome data (not reported in detail)
Good neurological outcome at discharge (mRS ≤2) VC: 23/142 (16.2%) 15/134 (11.2%) Adjusted RR 1.48 95% CI 0.81 to 2.71 Neurological outcome estimate imprecise
  • DSED was associated with higher survival to discharge (30.4% vs 13.3%) and higher rates of VF termination (84.0% vs 67.6%) and ROSC (46.4% vs 26.5%).
  • VC improved VF termination (79.9% vs 67.6%) and showed a survival signal (21.7% vs 13.3%), but the survival estimate was fragile (fragility index 1) and confidence intervals were close to 1.0.
  • CPR process measures and resuscitation co-interventions were broadly similar across groups, supporting (but not proving) separation of the defibrillation strategy as the primary active difference.

Internal Validity

  • Randomisation and Allocation:
    • Cluster randomisation with crossover reduces within-service contamination and training drift, but introduces risk of period effects and requires adequate cluster count.
    • Only six paramedic services (clusters) participated, limiting protection against cluster-level confounding despite crossover.
  • Drop out or exclusions:
    • 450 patients met initial eligibility assessment; 45 were excluded before randomisation (not meeting criteria for refractory VF/pulseless VT).
    • 405 patients were randomised; survival outcome was missing for 2 patients; modified Rankin score was missing for 3 patients.
  • Performance/Detection Bias:
    • Open-label delivery could influence co-interventions; however, objective outcomes and similar CPR metrics reduce (but do not eliminate) this concern.
    • Neurological outcome (mRS at discharge) may be subject to inter-rater/record-based variability; detailed adjudication procedures were not reported in the main manuscript.
  • Protocol Adherence:
    • Assigned strategy was delivered in 87.7% of patients overall.
    • Adherence by arm: standard 115/136 (84.6%); VC 125/144 (86.8%); DSED 115/125 (92.0%).
    • Non-adherence mechanisms were operational (e.g., switching to standard or VC when the second defibrillator was not immediately available for DSED).
  • Baseline Characteristics:
    • Age (mean ± SD): standard 63.2 ± 12.9; VC 63.7 ± 12.5; DSED 63.0 ± 12.6 years.
    • Male sex: standard 84.6%; VC 86.1%; DSED 87.2%.
    • Bystander-witnessed arrest: standard 60.3%; VC 76.4%; DSED 66.4% (imbalance plausibly prognostic).
    • Bystander CPR: standard 54.4%; VC 62.5%; DSED 56.8%.
  • Heterogeneity:
    • Geographic and system-level practice heterogeneity across services is inevitable; crossover mitigates but does not erase cluster-level structural differences.
    • Only three crossover periods and early termination increase susceptibility to time-varying confounding (e.g., pandemic-related changes in OHCA epidemiology and systems performance).
  • Timing:
    • Time from 911 call to first defibrillation shock (mean ± SD): standard 9.0 ± 3.5 min; VC 9.1 ± 3.2 min; DSED 9.0 ± 3.3 min.
    • Time from paramedic arrival to first shock (mean ± SD): standard 1.5 ± 1.1 min; VC 1.3 ± 0.8 min; DSED 1.5 ± 1.2 min.
    • Intervention was protocolised to begin after the third standard shock; trial data support broadly similar early timelines across arms.
  • Dose:
    • DSED delivered two maximum-energy shocks in rapid sequence (ZOLL: 200 J + 200 J; LIFEPAK: 360 J + 360 J) using two pad vectors.
    • Whether DSED’s “dose” is optimal (energy, ordering, inter-shock delay, pad vectors) remains uncertain and was not individually titrated.
  • Separation of the Variable of Interest:
    • Number of standard shocks delivered (mean ± SD): standard 7.4 ± 3.0 vs VC 4.2 ± 2.1 vs DSED 3.9 ± 1.4.
    • Defibrillation success signals separated early: VF termination after first eligible shock was 67.6% (standard) vs 79.9% (VC) vs 84.0% (DSED).
  • Key Delivery Aspects:
    • CPR process measures were similar across groups: chest compression fraction 89.7 ± 5.9% (standard) vs 90.2 ± 5.3% (VC) vs 90.8 ± 4.7% (DSED).
    • Co-interventions were similar: amiodarone use 86.0% vs 84.7% vs 88.8%; mean epinephrine doses 2.4 ± 1.4 vs 2.2 ± 1.3 vs 2.3 ± 1.3.
  • Outcome Assessment:
    • Primary outcome (survival to discharge) is objective and robust to ascertainment bias.
    • Neurological outcome at discharge (mRS ≤2) is clinically relevant but depends on accurate hospital documentation and scoring.
  • Statistical Rigor:
    • Trial did not reach the planned sample size (405/930), increasing risk of overestimation and imprecision.
    • No multiplicity adjustment was applied for pairwise comparisons (protocol-level decision).
    • Fragility indices were low for the VC survival comparison (fragility index 1), underscoring result instability.

Conclusion on Internal Validity: Overall, internal validity appears moderate: randomisation and crossover strengthen causal inference, and protocol separation/co-interventions were credible, but early stopping, small cluster count, open-label delivery, and prognostic imbalances (notably bystander-witnessed status) limit certainty about the magnitude and reproducibility of effects.

External Validity

  • Population Representativeness:
    • Represents adult OHCA with shockable rhythms progressing to “refractory” VF/pulseless VT after three shocks within organised EMS systems.
    • Observed bystander witness and CPR rates may be higher than some regions, potentially affecting absolute benefits.
  • Applicability:
    • VC is relatively low-cost and implementable in most settings with minimal additional equipment.
    • DSED requires two defibrillators, extra pads, choreography, and training; feasibility may vary (single-tier vs dual-tier EMS, rural/remote response models).
    • Results apply primarily to out-of-hospital settings; extrapolation to in-hospital arrest requires caution.

Conclusion on External Validity: Generalisability is reasonable for EMS systems that can deliver protocolised escalation after three failed shocks, especially for VC; broader adoption of DSED depends on equipment availability, training, and local governance around dual-defibrillator deployment.

Strengths & Limitations

  • Strengths:
    • Pragmatic cluster-crossover design aligned with EMS implementation realities.
    • Clinically meaningful primary outcome (survival to discharge) with concordant mechanistic secondary outcomes (termination, ROSC).
    • Demonstrated feasibility and high protocol adherence for a complex intervention (DSED).
    • Comparable CPR quality and drug use across arms supports separation of the defibrillation strategy.
  • Limitations:
    • Early termination (405/930 planned) increases imprecision and risk of overestimating benefit size.
    • Small number of clusters (six services) increases susceptibility to residual confounding and period effects.
    • Open-label delivery; neurological outcome at discharge may be less robust than longer-term, blinded functional assessment.
    • Post-ROSC/in-hospital care was not standardised and could contribute to outcome differences, particularly for neurological endpoints.
    • Harms/adverse events were not comprehensively reported beyond feasibility/safety observations in the main manuscript.

Interpretation & Why It Matters

  • Clinical implication
    • After three unsuccessful shocks for VF/pulseless VT in OHCA, protocolised escalation to an alternative defibrillation strategy (particularly DSED) can plausibly improve survival to discharge in capable EMS systems.
  • Mechanistic coherence
    • Benefits were accompanied by higher VF termination and ROSC, suggesting a primary effect on defibrillation efficacy rather than downstream care alone.
  • Implementation lens
    • VC offers a simpler systems change; DSED requires two defibrillators and choreography but showed the largest effect estimates.

Controversies & Subsequent Evidence

  • Early stopping and effect-size inflation risk:
    • Stopped at 405/930 planned, with wide confidence intervals and potential for overestimation of treatment effect sizes, particularly for survival and neurological outcomes.
    • The accompanying editorial emphasised that the effect sizes are compelling but require careful systems implementation and further confirmation in other EMS contexts4.
  • Cluster-crossover complexity and small cluster count:
    • Only six clusters raises vulnerability to time-varying confounding (including pandemic-era system changes) despite crossover.
    • Protocol-level decisions (e.g., no multiplicity adjustment) and the pragmatic nature of EMS delivery influence interpretability for methodologists3.
  • VC versus DSED interpretation:
    • VC demonstrated improved VF termination but a less secure survival estimate (fragility index 1 for survival vs standard), making it challenging to separate “true benefit” from chance in an underpowered, early-stopped setting.
    • DSED showed the most consistent pattern across termination, ROSC, survival, and neurological outcomes, but is operationally more demanding.
  • Evidence synthesis trajectory:
    • Pre-DOSE VF syntheses found very low certainty evidence for DSED and emphasised the need for randomised trials15.
    • More recent network meta-analytic work has incorporated DOSE VF and has focused on comparative effectiveness across standard defibrillation, VC, and DSED, while still being constrained by the limited number of randomised datasets6.

Summary

  • DOSE VF tested two escalation strategies (VC and DSED) after three failed standard shocks for refractory VF/pulseless VT in OHCA, using a pragmatic cluster-randomised crossover design.
  • The trial stopped early (405/930 planned) during COVID-19 but showed higher survival to discharge with DSED (30.4%) and VC (21.7%) compared with standard defibrillation (13.3%).
  • DSED produced concordant improvements in VF termination (84.0% vs 67.6%) and ROSC (46.4% vs 26.5%), supporting biological plausibility for survival benefit.
  • VC improved VF termination (79.9% vs 67.6%) but the survival estimate was fragile (fragility index 1) and imprecise.
  • Implementation is the key translational challenge: VC is simple to deploy; DSED demands two defibrillators, pad logistics, choreography, and training.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • VC is often the lowest-barrier escalation (pads + protocol); DSED requires two defibrillators, extra pads, choreography, and governance around safety and training.

Overall Takeaway

DOSE VF is a landmark prehospital resuscitation trial because it randomised a systems-level defibrillation escalation strategy at the point where standard care typically becomes repetitive and ineffective. Despite early termination, the consistent pattern of improved defibrillation success, ROSC, and survival—strongest for DSED—provides the clearest randomised signal to date that “how we shock” after three failures may meaningfully alter outcomes, while also highlighting the methodological and implementation constraints that demand careful local adoption and further confirmation.

Overall Summary

  • After three failed shocks for VF/pulseless VT OHCA, switching strategy (especially DSED) improved survival to discharge versus continuing standard defibrillation.
  • VC is simpler to implement but its survival estimate was fragile in this early-stopped trial.
  • Adoption hinges on system capability: training, pad logistics, and (for DSED) access to two defibrillators.

Bibliography