Skip to main content

Foundational Trials

PARAMEDIC

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Mechanical versus manual chest compression for out-of-hospital cardiac arrest (PARAMEDIC): a pragmatic, cluster randomised controlled trial
  • Acronym: PARAMEDIC
  • Year: 2015
  • Journal published in: The Lancet
  • Citation: Perkins GD, Lall R, Quinn T, Deakin CD, Cooke MW, Horton J, et al. Mechanical versus manual chest compression for out-of-hospital cardiac arrest (PARAMEDIC): a pragmatic, cluster randomised controlled trial. Lancet. 2015;385:947-955.

Context & Rationale

  • Background
    • High-quality manual chest compressions are difficult to sustain in real-world prehospital resuscitation, with fatigue and procedural/transport-related interruptions.
    • Mechanical CPR devices can deliver consistent compression rate/depth and may facilitate CPR during transport or procedures, but patient-centred benefit (survival with good neurological outcome) was uncertain.
    • Prior randomised evidence was mixed, and uptake of mechanical CPR in ambulance services was increasing without definitive pragmatic effectiveness data in the UK setting.
  • Research Question/Hypothesis
    • In adults with non-traumatic out-of-hospital cardiac arrest (OHCA), does the use of the LUCAS-2 mechanical chest compression device (vs manual chest compressions) improve survival to 30 days?
  • Why This Matters
    • Mechanical CPR devices are costly and require training; any benefit must outweigh risks of deployment-related interruptions.
    • Prehospital systems require clear evidence to determine whether devices should be deployed routinely, selectively (e.g., during transport), or not at all.
    • Survival and neurological outcomes provide the most clinically meaningful measures of efficacy for this system-level intervention.

Design & Methods

  • Research Question: In adults with non-traumatic OHCA attended by participating UK ambulance service vehicles, does LUCAS-2 mechanical CPR improve 30-day survival compared with manual chest compressions?
  • Study Type: Pragmatic, multicentre, cluster-randomised controlled trial (emergency vehicles as clusters; allocation determined by first trial vehicle on scene); open-label; prehospital setting across four UK ambulance services (West Midlands, North East England, Wales, South Central); recruitment April 2010–June 2013.
  • Population:
    • Adults (≥18 years) with non-traumatic OHCA attended by a participating emergency vehicle, in whom resuscitation was attempted.
    • Key exclusions included: traumatic arrest; pregnancy; age <18 years; do-not-attempt-resuscitation status; and patients for whom “recognition of life extinct”/no resuscitation was appropriate (e.g., obvious signs of death or prolonged asystolic arrest without reversible causes and without bystander CPR).
    • Cluster allocation: the first trial vehicle to arrive and attempt resuscitation determined group assignment.
  • Intervention:
    • LUCAS-2 mechanical chest compression device deployed by trained ambulance crews as soon as practicable during resuscitation, with manual compressions provided during deployment and when the device could not be used.
  • Comparison:
    • Manual chest compressions delivered by ambulance clinicians according to contemporary UK resuscitation guidelines, without routine use of a mechanical device.
  • Blinding: Treating ambulance crews were not blinded; ambulance dispatch staff and those collecting the primary outcome were masked to treatment allocation; blinding of 3‑month neurological outcome assessment was not reported.
  • Statistics: A total of 3675 patients were required (245 clusters; assumed intracluster correlation coefficient 0.01 and average cluster size 15) to detect an absolute increase in 30-day survival from 5.0% to 7.5% with 80% power at a 5% significance level; target sample size was increased to 4344 after interim sample size re-estimation. Primary analysis was intention-to-treat using fixed-effects logistic regression adjusted for prespecified covariates; complier-average causal effect (CACE) analyses were also performed.
  • Follow-Up Period: 30 days (primary); additional survival follow-up at 3 months and 12 months; neurological outcome (CPC) at 3 months.

Key Results

This trial was not stopped early. Recruitment was completed (n=4471), with a prespecified sample size increase after interim re-estimation.

Outcome LUCAS-2 Manual CPR Effect p value / 95% CI Notes
Survival to 30 days (primary) 104/1652 (6%) 193/2819 (7%) Adjusted OR 0.86 95% CI 0.64 to 1.15; P not reported Intention-to-treat; fixed-effects logistic regression adjusted for prespecified covariates.
Favourable neurological outcome at 3 months (CPC 1–2) 77/1652 (5%) 168/2819 (6%) Adjusted OR 0.72 95% CI 0.52 to 0.99; P not reported Secondary outcome; CPC derived from records or face-to-face assessment (approach as reported in trial).
Return of spontaneous circulation (ROSC) 522/1652 (32%) 885/2819 (31%) Adjusted OR 0.99 95% CI 0.86 to 1.14; P not reported Secondary outcome.
Survived event 377/1652 (23%) 658/2819 (23%) Adjusted OR 0.97 95% CI 0.82 to 1.14; P not reported Secondary outcome.
Survival to hospital discharge 145/1652 (9%) 277/2819 (10%) Adjusted OR 0.87 95% CI 0.70 to 1.07; P not reported Secondary outcome.
Survival to 3 months 96/1652 (6%) 182/2819 (6%) Adjusted OR 0.83 95% CI 0.61 to 1.12; P not reported Secondary outcome.
Survival to 12 months 89/1652 (5%) 175/2819 (6%) Adjusted OR 0.83 95% CI 0.62 to 1.11; P not reported Secondary outcome.
Subgroup: survival to 30 days with initial shockable rhythm 69/376 (18%) 148/615 (24%) OR 0.71 95% CI 0.52 to 0.98; interaction P<0.05 Prespecified subgroup analysis (initial rhythm); interpret as subgroup effect estimate.
Clinical adverse events (reported) 7 events (0.4%) 0 Not reported Not reported All events occurred in LUCAS-2 group: chest bruising (3), lacerations (2), blood in mouth (2); no serious adverse events reported.
  • Primary outcome showed no survival benefit: 30-day survival was 104/1652 (6%) with LUCAS-2 vs 193/2819 (7%) with manual CPR (adjusted OR 0.86; 95% CI 0.64 to 1.15).
  • Neurologically favourable survival at 3 months was lower with LUCAS-2: 77/1652 (5%) vs 168/2819 (6%) (adjusted OR 0.72; 95% CI 0.52 to 0.99).
  • Protocol adherence was incomplete: LUCAS-2 was used in 985/1652 (60%) in the intervention arm vs 11/2819 (<1%) in the control arm; CACE analyses for 30-day survival were similar to intention-to-treat (CACE1 OR 0.92; 95% CI 0.69 to 1.21; CACE2 OR 0.87; 95% CI 0.61 to 1.23).

Internal Validity

  • Randomisation and Allocation
    • Cluster randomisation by emergency vehicle (n=418 vehicles; 147 allocated to LUCAS-2 and 271 to control), stratified by ambulance station and vehicle type.
    • Allocation was necessarily known to crews (device present or absent), but assignment was determined before the arrest and by the first trial vehicle on scene.
  • Dropout or Exclusions
    • Of 11,171 OHCA incidents, 6482 were not assessed for eligibility (recognition of life extinct or no resuscitation attempted).
    • 4689 were assessed; 218 were excluded; 4471 were included in the intention-to-treat analyses (1652 LUCAS-2; 2819 control).
    • Vital status follow-up was near-complete (1 patient with unknown survival status in the control group reported); CPC at 3 months was missing for 3 (LUCAS-2) and 4 (control).
  • Performance and Detection Bias
    • Open-label delivery could influence co-interventions and resuscitation behaviour; primary outcome (30-day survival) is objective and largely registry-based.
    • Ambulance dispatch staff and primary outcome data collectors were masked to allocation; blinding for neurological outcome assessment was not reported.
  • Protocol Adherence
    • Intervention exposure: 985/1652 (60%) received LUCAS-2; 667/1652 (40%) received manual CPR only.
    • Reported reasons for non-use in the LUCAS-2 arm included: crew not trained (78), crew error (168), device not in vehicle (26), patient unsuitable (102: 58 too large; 22 too small; 22 other reasons), device issues (14), not possible to use device (140), reason unknown (110).
    • Control contamination was low: 11/2819 (<1%) received LUCAS-2.
  • Baseline Characteristics
    • Groups were broadly comparable: age 70.4 (SD 15.5) vs 69.7 (15.6) years; male 63% vs 65%.
    • Arrest features were similar: witnessed 66% vs 68%; bystander CPR 43% vs 44%; shockable initial rhythm 23% vs 22%.
    • Prehospital timing was similar: median (IQR) call-to-arrival 6.5 (4.9–8.8) vs 6.3 (4.7–8.5) minutes; total response interval 11.4 (8.7–15.2) vs 11.2 (8.5–14.8) minutes.
  • Heterogeneity
    • Cluster design introduces between-vehicle variability; the prespecified multilevel random-effects model did not converge and analyses used fixed-effects logistic regression.
    • Inclusion of an intracluster correlation coefficient (ICC) was attempted; data suggested a very low ICC (~0.001) for survival outcomes, supporting limited clustering effect.
  • Timing
    • Allocation was operationalised at first-vehicle arrival; overall response times were similar between groups (as above).
    • Time-to-first-compression, hands-off time, and time-to-first-shock were not reported as trial outcomes, limiting direct assessment of deployment-related interruptions.
  • Dose
    • Mechanical CPR “dose” was defined by device use; detailed CPR quality metrics (compression depth, rate, fraction) were not reported.
  • Separation of the Variable of Interest
    • Exposure separation was incomplete: mechanical device use 985/1652 (60%) vs 11/2819 (<1%).
    • Resuscitation characteristics were broadly similar: adrenaline 83% vs 80%; advanced airway (supraglottic or tracheal) 88% vs 91%; defibrillation 36% vs 37%.
  • Crossover
    • Device contamination in control was rare (11/2819), but non-use in the intervention arm was substantial (667/1652 manual-only), potentially diluting any true effect.
    • CACE analyses were performed to address non-compliance and did not suggest benefit for 30-day survival.
  • Outcome Assessment
    • Primary outcome (30-day survival) was objective and ascertained with masking of outcome collectors.
    • Neurological outcome at 3 months (CPC) was clinically relevant but methods for blinding of assessors were not reported.
  • Statistical Rigor
    • Primary intention-to-treat analysis with prespecified covariate adjustment; additional CACE analyses to estimate effect in compliers.
    • Sample size re-estimation and increase in target sample size occurred mid-trial to preserve power.

Conclusion on Internal Validity: Overall, internal validity appears moderate—a large pragmatic randomised design with objective primary outcome and near-complete follow-up, but open-label delivery, substantial non-adherence (60% device use), and lack of CPR-quality/process measures limit confidence that the intervention achieved sustained separation without unintended interruptions.

External Validity

  • Population Representativeness
    • Broad inclusion of adult non-traumatic OHCA with attempted resuscitation in routine UK ambulance services; baseline characteristics typical (mean age ~70 years; ~two-thirds witnessed; bystander CPR ~44%).
    • Participating stations were described as urban/semi-urban, which may under-represent rural response/transport profiles.
    • Exclusion of traumatic arrest, pregnancy, and cases in which resuscitation was not attempted is consistent with most OHCA trials and routine practice.
  • Applicability
    • Findings are most applicable to systems considering routine mechanical CPR deployment for all eligible OHCA calls.
    • Generalisability to systems that deploy mechanical CPR selectively (e.g., for prolonged transport, catheterisation laboratory CPR, or bridging to ECPR) is more limited because the trial tested a pragmatic “device carried on vehicle” strategy with incomplete uptake and without measuring CPR process metrics.
    • Resource-limited EMS environments may have different trade-offs (e.g., fewer rescuers), where the balance between consistency and deployment interruptions could differ.

Conclusion on External Validity: Overall, external validity is good for routine UK-style prehospital OHCA care and for services contemplating widespread device implementation, but is less direct for highly protocolised “selective-use” strategies (transport/PCI/ECPR) and for markedly different EMS staffing models.

Strengths & Limitations

  • Strengths:
    • Large, pragmatic, multicentre trial embedded in routine ambulance service practice (n=4471 across four services).
    • Cluster randomisation by vehicle minimised selection bias at the point of arrest and reflected real-world implementation constraints.
    • Objective primary endpoint (30-day survival) with masking of outcome collectors and near-complete follow-up.
    • Prespecified covariate adjustment and additional CACE analyses to address substantial non-compliance.
  • Limitations:
    • Open-label intervention with potential for performance differences in co-interventions and resuscitation behaviour.
    • Incomplete intervention delivery (LUCAS-2 used in 60% of intervention-arm patients), reducing separation and potentially diluting effect estimates.
    • Key CPR process/quality metrics (compression fraction, pauses, time-to-first-shock) were not reported, limiting mechanistic interpretation.
    • Random-effects models did not converge; fixed-effects approaches were used, and clustering effects could not be modelled as originally planned (although estimated ICC was low).

Interpretation & Why It Matters

  • Routine device deployment
    • In a pragmatic UK EMS implementation strategy, routine access to LUCAS-2 did not improve survival to 30 days and was associated with lower odds of favourable neurological outcome at 3 months.
  • Implementation is the intervention
    • The trial underscores that effectiveness depends not only on device physiology but also on training, deployment feasibility, and avoiding interruptions—particularly in shockable rhythms where time-to-defibrillation is critical.
  • Targeted rather than universal use
    • For clinicians and services, PARAMEDIC supports prioritising high-quality manual CPR as standard, with mechanical CPR considered selectively when manual compressions are difficult to deliver safely or continuously (e.g., during extrication/transport), while recognising that this specific “routine deployment” strategy did not improve outcomes.

Controversies & Subsequent Evidence

  • Pragmatic cluster design and non-compliance
    • The published protocol specified a pragmatic, vehicle-cluster design to mirror real deployment constraints and prespecified additional analyses to address non-adherence1.
    • Non-compliance (40% manual-only in the intervention arm) was a defining feature of the “strategy” tested and motivated CACE analyses; these did not show benefit for 30-day survival.
  • Mechanistic concerns: interruptions and shockable rhythms
    • Contemporary commentary highlighted that mechanical CPR may improve consistency but could introduce pauses during deployment and might delay critical early actions (especially defibrillation), potentially explaining absence of benefit in pragmatic settings2.
    • A prespecified subgroup analysis suggested lower 30-day survival with LUCAS-2 among patients with initial shockable rhythm (OR 0.71; 95% CI 0.52 to 0.98; interaction P<0.05), amplifying concerns about time-critical defibrillation within a device-deployment workflow.
  • Correspondence: transport and in-hospital care
    • Correspondence queried whether differential transport with ongoing CPR and subsequent in-hospital management could influence interpretation of trial outcomes3.
    • The authors’ reply emphasised post-randomisation selection issues and reported survival among those transported with CPR ongoing: 11/640 (1.7%) in the LUCAS-2 group vs 11/1081 (1.0%) in the control group4.
  • Health economic implications
    • A within-trial and model-based economic evaluation using PARAMEDIC trial data concluded that manual CPR dominated LUCAS-2 (poorer outcomes and higher costs with LUCAS-2) in most analyses, indicating poor value for money for routine deployment in OHCA5.
    • The NIHR Health Technology Assessment programme monograph reporting PARAMEDIC also integrated clinical and economic findings, supporting the interpretation that routine mechanical CPR deployment does not improve outcomes and is unlikely to be cost-effective at system level6.
  • Evidence syntheses and evolving guidance
    • Systematic reviews/meta-analyses incorporating PARAMEDIC alongside other large pragmatic RCTs have generally found no improvement in survival outcomes with mechanical CPR devices in OHCA, with heterogeneity driven by device type and deployment strategy78.
    • A Cochrane update concluded that mechanical CPR is not superior to manual CPR overall, but may be a reasonable alternative when high-quality manual compressions are not possible or are dangerous for rescuers, emphasising close monitoring to minimise no-flow time and avoid defibrillation delays9.
    • Contemporary guideline updates (e.g., 2025 AHA and 2025 ERC ALS guidance) incorporate large device trials and continue to frame mechanical CPR as a selective tool that must be deployed without compromising time-critical actions (especially in shockable rhythms)1011.

Summary

  • PARAMEDIC was a large pragmatic, vehicle-cluster randomised trial (n=4471) across four UK ambulance services testing routine access to the LUCAS-2 device vs manual CPR for adult non-traumatic OHCA.
  • There was no improvement in 30-day survival: 6% (LUCAS-2) vs 7% (manual); adjusted OR 0.86 (95% CI 0.64 to 1.15).
  • Favourable neurological outcome at 3 months (CPC 1–2) was lower in the LUCAS-2 group: 5% vs 6%; adjusted OR 0.72 (95% CI 0.52 to 0.99).
  • Intervention delivery was incomplete: only 60% of intervention-arm patients received LUCAS-2; CACE analyses did not suggest benefit among compliers.
  • Reported serious adverse events were absent; minor clinical adverse events occurred only in the LUCAS-2 arm (7 events, 0.4%).

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • the Lancet paper reported 95% confidence intervals but did not routinely report P values.

Overall Takeaway

PARAMEDIC demonstrated that, in a pragmatic UK prehospital implementation strategy, routine access to mechanical CPR (LUCAS-2) did not improve 30-day survival and did not improve longer-term survival, with a signal towards worse favourable neurological outcome at 3 months. The trial re-framed mechanical CPR as an implementation-sensitive, selective adjunct rather than a universal replacement for high-quality manual CPR, particularly where time-critical defibrillation must be protected.

Overall Summary

  • Large pragmatic UK cluster-RCT: LUCAS-2 did not improve 30-day survival vs manual CPR.
  • Intervention delivery was incomplete (60% received LUCAS-2), and compliers-only analyses did not show benefit.
  • System-level message: prioritise manual CPR quality; use mechanical CPR selectively where manual compressions are unsafe or impractical.

Bibliography