Skip to main content

Foundational Trials

PROCOAG

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Efficacy and Safety of Early Administration of 4-Factor Prothrombin Complex Concentrate in Patients With Trauma at Risk of Massive Transfusion: The PROCOAG Randomized Clinical Trial
  • Acronym: PROCOAG
  • Year: 2023
  • Journal published in: JAMA
  • Citation: Bouzat P, Charbit J, Abback P-S, Huet-Garrigue D, Delhaye N, Leone M, et al; PROCOAG Study Group. Efficacy and safety of early administration of 4-factor prothrombin complex concentrate in patients with trauma at risk of massive transfusion: the PROCOAG randomized clinical trial. JAMA. 2023;329(16):1367-1375.

Context & Rationale

  • Background
    • Traumatic haemorrhage is a leading driver of early potentially preventable death; trauma-induced coagulopathy is common on arrival and is associated with higher transfusion requirements and mortality.
    • Component-based damage control resuscitation (RBC, plasma, platelets) is effective but can be logistically constrained (e.g., plasma thawing/availability) and may not rapidly correct early factor depletion.
    • 4-factor prothrombin complex concentrate (4F-PCC) is a small-volume, rapidly administered concentrate of vitamin K–dependent factors (II, VII, IX, X) that can normalise prothrombin time; trauma use is off-label and was supported mainly by observational analyses prone to confounding.
    • The central uncertainty was whether empirical early 4F-PCC adds clinically meaningful haemostatic benefit on top of contemporary transfusion protocols, and whether any benefit is offset by thromboembolic harm.
  • Research Question/Hypothesis
    • In adults with severe trauma requiring early RBC transfusion and judged at risk of massive transfusion, does early 4F-PCC (25 IU factor IX/kg; 1 mL/kg) reduce total allogeneic blood product use within 24 hours compared with placebo?
    • Is early 4F-PCC safe in this setting, particularly regarding thromboembolic complications within 28 days?
  • Why This Matters
    • If effective, 4F-PCC could provide a pragmatic, rapidly deployable haemostatic adjunct that reduces component exposure and resource burden during early trauma resuscitation.
    • A reliable prothrombotic signal would materially shift practice away from routine empirical PCC and towards phenotype-guided haemostatic therapy with explicit safety surveillance.

Design & Methods

  • Research Question: Does early in-hospital 4F-PCC (25 IU factor IX/kg) reduce 24-hour allogeneic blood product consumption, without excess thromboembolic harm, in adults with severe trauma requiring early RBC transfusion and at risk of massive transfusion?
  • Study Type: Multicentre, double-blind, randomised, placebo-controlled superiority trial (investigator-initiated), stratified by centre; conducted in 12 French designated level I trauma centres (Dec 2017 to Aug 2021) during early emergency department trauma resuscitation.
  • Population:
    • Inclusion: adults (≥18 years) with severe trauma (grade A trauma team activation), requiring RBC transfusion prehospital or within the first hour after admission, randomised within 1 hour of admission, and considered at risk of massive transfusion (Assessment of Blood Consumption score ≥2 or clinician judgement).
    • Exclusion: traumatic cardiac arrest with cardiopulmonary resuscitation ongoing at admission; pregnancy; major burns; known congenital coagulopathy/bleeding disorder; previous enrolment in PROCOAG.
  • Intervention:
    • 4F-PCC 1 mL/kg (equivalent to 25 IU factor IX/kg), administered intravenously via syringe pump at 120 mL/h as soon as possible after randomisation (median treatment start 35 minutes after admission).
    • Delivered alongside protocolised trauma haemorrhage management (including tranexamic acid and component therapy as clinically indicated).
  • Comparison:
    • Placebo (0.9% saline) 1 mL/kg, administered with identical appearance and delivery conditions.
    • Same background trauma haemorrhage management protocol as the intervention arm.
  • Blinding: Double-blind (patients, treating clinicians, investigators, outcome assessors, and statisticians); an unblinded nurse and hospital pharmacy prepared study drug to maintain masking.
  • Statistics: A total of 324 patients were required to detect a 3-unit reduction in 24-hour allogeneic blood product consumption (from mean 12 to 9 units; 25% relative reduction) with 80% power at the 5% two-sided significance level; primary analysis used a modified intention-to-treat approach (excluding patients who withdrew consent), with prespecified secondary/per-protocol analyses and no imputation for missing data.
  • Follow-Up Period: Primary end point to 24 hours; clinical and safety outcomes (including thromboembolism and mortality) to 28 days.

Key Results

This trial was not stopped early. Recruitment completed as planned (no interim analyses were reported).

Outcome 4F-PCC Placebo Effect p value / 95% CI Notes
Total allogeneic blood products within 24 h (units; RBC+FFP+platelets) 12 (6–20) 11 (7–21) Absolute difference 0.2 U 95% CI −2.99 to 3.33; P=0.72 Primary end point; median (IQR); analysed n=164 vs n=160
RBC within 24 h (units) 6 (3–9) 6 (4–9) Absolute difference −0.3 U 95% CI −1.8 to 1.3; P=0.93 Median (IQR)
Fresh frozen plasma within 24 h (units) 4 (1–7) 4 (2–7) Absolute difference 0.1 U 95% CI −1.3 to 1.5; P=0.56 Median (IQR)
Platelets within 24 h (units) 1 (0–2) 1 (0–2) Absolute difference 0.0 U 95% CI −0.3 to 0.3; P=0.83 Median (IQR)
Time to PTr <1.5 among severe coagulopathy (minutes) 0 (0–0) 0 (0–0) Absolute difference −8.5 min 95% CI −48.9 to 32.0; P=0.86 Median (IQR); competing-risk approach used in prespecified analyses
Time to haemostasis (minutes) 300 (155–635) 288 (148–534) Absolute difference 22 min 95% CI −73.3 to 73.8; P=0.96 Median (IQR); haemostasis timing is clinically determined
Mortality at 24 h 18/164 (11%) 21/160 (13%) Absolute difference −2% 95% CI −9 to 5; P=0.67 Binary outcome
Mortality at 28 days 27/164 (17%) 33/160 (21%) Absolute difference −3% 95% CI −12 to 5; P=0.48 Binary outcome
≥1 thromboembolic event by day 28 56/161 (35%) 37/157 (24%) RR 1.48 95% CI 1.04 to 2.10; P=0.03 Safety end point; passive surveillance; denominator reflects evaluable follow-up
  • Early 4F-PCC did not reduce 24-hour total allogeneic blood product use (median 12 vs 11 units; absolute difference 0.2 U; 95% CI −2.99 to 3.33; P=0.72).
  • Despite rapid administration (median treatment start 35 vs 30 minutes after admission), there was no signal of benefit for component-specific transfusion, haemostasis timing, or mortality at 24 hours/28 days.
  • A statistically significant increase in thromboembolic events was observed with 4F-PCC (35% vs 24%; RR 1.48; 95% CI 1.04 to 2.10; P=0.03).

Internal Validity

  • Randomisation and allocation:
    • Computer-generated randomisation with variable block sizes (2/4/6), stratified by centre.
    • Sequential sealed-envelope system with preparation by pharmacy (supports allocation concealment in emergency conditions).
  • Dropout or exclusions:
    • 327 randomised; 3 withdrew consent (1 in 4F-PCC; 2 in placebo); primary analysis included 324 patients (164 vs 160).
  • Performance/detection bias:
    • Double-blinding limits performance bias for transfusion decisions; the primary outcome (blood product units) is objective but can still be influenced by clinician behaviour within protocol boundaries.
    • Thromboembolic outcomes were captured by passive surveillance (greater vulnerability to variable detection and imaging thresholds across centres).
  • Protocol adherence:
    • Per-protocol population: 308/324 (95%) received study product within the first hour (159/164 vs 149/160).
    • Time from admission to treatment start: 35 (25–53) minutes with 4F-PCC vs 30 (21–47) minutes with placebo.
  • Baseline characteristics:
    • Overall injury severity and admission physiology were broadly similar (e.g., median ISS 36 vs 36; admission systolic blood pressure 89 vs 90 mmHg).
    • Prehospital systolic blood pressure was higher in the 4F-PCC arm (101 [80–120] vs 90 [70–116] mmHg), while admission lactate was similar (5.4 vs 5.6 mmol/L).
    • Coagulopathy burden on admission was high but comparable: PTr >1.2 in 65% vs 68% (among those with available PTr at baseline).
  • Heterogeneity:
    • 12 trauma centres with stratified randomisation improves balance, but transfusion intensity and imaging practices can vary materially in pragmatic haemorrhage trials.
  • Timing:
    • Randomisation within 1 hour of admission and rapid study drug initiation align with the biological window when early factor depletion might influence haemostasis.
  • Dose:
    • Fixed 25 IU factor IX/kg (1 mL/kg) provides moderate factor repletion; higher doses used in anticoagulant reversal were avoided, plausibly balancing haemostatic ambition against thrombotic risk.
  • Separation of the variable of interest:
    • Study drug exposure was clearly separated (4F-PCC vs saline placebo) under masking.
    • Relevant co-interventions were common in both arms, with measurable imbalances: tranexamic acid 76% (124/164) vs 86% (138/160); fibrinogen concentrate 3 (3–7.5) g vs 3 (3–6) g; time to first FFP 73 (48–105) vs 91 (60–133) minutes.
  • Outcome assessment:
    • Primary outcome (blood product units within 24 hours) is objective and generally complete.
    • Time to haemostasis is clinically determined and can be influenced by procedural and operational factors beyond haemostasis biology.
  • Statistical rigour:
    • Power calculation and target sample size were achieved for the prespecified effect size; primary and secondary analyses reported confidence intervals that include clinically meaningful benefit and harm for some outcomes.
    • Primary analysis was modified intention-to-treat (consent withdrawals excluded), which is common in emergency consent models but can introduce small post-randomisation exclusions.

Conclusion on Internal Validity: Overall, internal validity appears moderate-to-strong for the primary transfusion end point given multicentre randomisation, masking, and high protocol adherence; interpretation of thromboembolic safety is more vulnerable to outcome detection variability and competing co-interventions within pragmatic trauma care.

External Validity

  • Population representativeness:
    • Represents adult major trauma in high-resource, designated level I centres with established trauma team activation, early RBC transfusion, and access to plasma/platelets/adjuncts.
    • Exclusions (pregnancy, major burns, traumatic arrest with CPR ongoing) limit inference to these clinically important but distinct populations.
  • Applicability:
    • Findings are most applicable to systems already delivering rapid component therapy and tranexamic acid; incremental benefit of empirical PCC may differ where plasma delivery is delayed or where factor concentrates are used as first-line haemostatic therapy.
    • Generalisability is limited in resource-constrained settings where PCC availability, imaging capacity (for thromboembolism detection), and transfusion protocols differ substantially.

Conclusion on External Validity: Overall external validity is moderate; the trial is highly informative for major trauma centres with mature transfusion infrastructure, but extrapolation to different resuscitation paradigms (prehospital plasma-first, factor concentrate–first, or low-resource systems) requires caution.

Strengths & Limitations

  • Strengths:
    • Double-blind, randomised, placebo-controlled design in an emergency resuscitation context.
    • Early treatment window with rapid drug delivery, maximising biological plausibility for haemostatic effect.
    • Multicentre pragmatic execution across 12 level I trauma centres, enhancing relevance to real-world trauma resuscitation.
    • Objective primary outcome (blood product units), prespecified statistical analysis, and high protocol adherence (95% per-protocol).
  • Limitations:
    • Empirical enrolment based on transfusion requirement and massive transfusion risk rather than confirmed factor deficiency or viscoelastic phenotype; dilution of treatment effect is plausible if many patients are not PCC-responsive.
    • Primary outcome is a resource/behaviour-dependent surrogate (component use), not a direct patient-centred outcome; mortality was not powered for modest differences.
    • Co-intervention imbalances (e.g., tranexamic acid and fibrinogen use) and pragmatic transfusion decision-making may attenuate or obscure incremental effect.
    • Thromboembolic safety relied on passive surveillance, which may be influenced by centre-level imaging thresholds and competing risks.

Interpretation & Why It Matters

  • Clinical implication
    • Routine empirical early 4F-PCC for trauma patients “at risk of massive transfusion” is not supported: transfusion requirements were unchanged, while thromboembolic events increased (RR 1.48; 95% CI 1.04 to 2.10).
    • Where PCC is considered, the risk–benefit calculus likely favours phenotype-guided administration (e.g., demonstrable factor deficiency/viscoelastic signature) rather than blanket use.
  • Mechanistic interpretation
    • 4F-PCC primarily targets vitamin K–dependent factor repletion and may normalise prothrombin time without addressing fibrinogen depletion, platelet dysfunction, endothelial injury, and hyperfibrinolysis that drive traumatic bleeding.
    • In a setting already delivering component therapy, PCC may add limited incremental haemostatic effect while increasing thrombin generation capacity and thrombotic risk.
  • Trial design lesson
    • Trauma haemostasis interventions may require tighter biological enrichment (rapid assays/viscoelastic triggers) and stronger separation from co-interventions to detect plausible benefits.
    • Safety end points (thrombosis) should ideally incorporate prespecified, standardised ascertainment to mitigate detection variability across centres.

Controversies & Subsequent Evidence

  • Correspondence highlighted tension between pragmatic enrolment and biological targeting:
    • Published correspondence in JAMA questioned whether empirical PCC in a heterogeneous “risk of massive transfusion” population risks treating patients without PCC-responsive haemostatic deficits, potentially diluting benefit and exposing low-risk patients to thrombosis.12
    • The authors’ reply emphasised the trial’s pragmatic intent and cautioned against routine empirical PCC given the thromboembolic signal.3
  • Interpretive debate: “PT correction” vs “haemostatic efficacy”:
    • Commentary in Intensive Care Medicine underscored that PCC may improve conventional coagulation metrics but does not substitute for fibrinogen/platelets and may have limited incremental value when modern component-based protocols are already in place.4
    • Further critique argued that the observed thromboembolic signal materially constrains empirical PCC use and strengthens the case for more selective indications and monitoring.5
  • Mechanistic follow-up:
    • An ancillary PROCOAG analysis using thrombin generation methodology reported procoagulant shifts after 4F-PCC administration, providing a plausible mechanistic link to the higher thromboembolic event rate observed in the parent trial.6
  • Evidence synthesis remains constrained by study design heterogeneity:
    • A 2023 systematic review and meta-analysis of PCC in trauma-induced coagulopathy emphasised that the evidence base is dominated by observational studies with heterogeneous dosing/indications and inconsistent thromboembolic ascertainment, limiting causal inference.7
    • A published critique of pooled analyses highlighted methodological fragility and the risk of over-interpreting low-certainty pooled estimates in this domain.8
  • Guideline alignment and practice impact:
    • The European trauma bleeding guideline advises against empirical haemostatic factor therapy when monitoring is available and emphasises targeted therapy guided by clinical and laboratory/viscoelastic assessment.9
  • Subsequent trial evidence:
    • In the FiiRST-2 RCT (2025), early fibrinogen concentrate plus PCC was not superior to frozen plasma for 24-hour allogeneic blood product use, reinforcing uncertainty about routine early factor-concentrate strategies and maintaining focus on patient selection and safety.1011

Summary

  • In 324 analysed patients with severe trauma requiring early RBC transfusion and judged at risk of massive transfusion, early 4F-PCC did not reduce 24-hour allogeneic blood product use (median 12 vs 11 units; P=0.72).
  • There was no significant signal of benefit for component-specific transfusion, time to haemostasis, or mortality at 24 hours or 28 days.
  • 4F-PCC was associated with more thromboembolic events by day 28 (35% vs 24%; RR 1.48; 95% CI 1.04 to 2.10; P=0.03).
  • The trial is most informative for modern, high-resource trauma systems already delivering rapid component therapy and tranexamic acid.
  • PROCOAG materially shifts the balance of evidence away from routine empirical PCC in trauma and towards phenotype-guided haemostatic strategies with explicit safety monitoring.

Overall Takeaway

PROCOAG provides high-level randomised evidence that empirical early 4F-PCC (25 IU/kg) added to contemporary trauma resuscitation does not reduce 24-hour blood component use. The observed increase in thromboembolic events shifts the risk–benefit balance against routine PCC in unselected “at-risk” trauma patients and supports a move towards targeted, phenotype-guided haemostatic strategies with explicit safety surveillance.

Overall Summary

  • Empirical early 4F-PCC did not reduce 24-hour transfusion burden but increased thromboembolic events.
  • Modern trauma haemostasis likely requires biology-enriched selection (rapid assays/viscoelastic phenotype), not blanket factor-concentrate use.

Bibliography