Skip to main content

Foundational Trials

PROPPR

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Transfusion of Plasma, Platelets, and Red Blood Cells in a 1:1:1 vs a 1:1:2 Ratio and Mortality in Patients With Severe Trauma: The PROPPR Randomized Clinical Trial
  • Acronym: PROPPR
  • Year: 2015
  • Journal published in: JAMA
  • Citation: Holcomb JB, Tilley BC, Baraniuk S, et al; PROPPR Study Group. Transfusion of plasma, platelets, and red blood cells in a 1:1:1 vs a 1:1:2 ratio and mortality in patients with severe trauma: the PROPPR randomized clinical trial. JAMA. 2015;313(5):471-482.

Context & Rationale

  • Background
    • Haemorrhage is a leading early cause of death after major trauma; severely injured patients in haemorrhagic shock often require massive transfusion.
    • “Damage control resuscitation” (earlier, higher plasma and platelet ratios alongside red cells) was increasingly adopted from military and civilian observational data, but causal inference was limited by confounding and time-dependent survival bias (patients must survive long enough to receive plasma/platelets).
    • Practice varied widely (including “RBC-heavy” early resuscitation and delayed platelets), and safety concerns persisted regarding inflammatory/volume-related complications (e.g., ARDS, multiple organ failure, thromboembolism).
    • A multicentre randomised trial was needed to test whether a more balanced early component strategy improves patient-centred outcomes and remains safe.
  • Research Question/Hypothesis
    • In adults with severe trauma predicted to require massive transfusion, does transfusion using a 1:1:1 plasma:platelets:RBC strategy (during active haemorrhage resuscitation) reduce mortality compared with a 1:1:2 strategy?
    • Co-primary endpoints: all-cause mortality at 24 hours and 30 days.
  • Why This Matters
    • Massive transfusion protocols (MTPs) determine blood bank logistics, bedside workflows, and resource utilisation at scale in trauma systems.
    • Clinically meaningful effects, if present, are expected early (exsanguination prevention), but competing causes of death (notably traumatic brain injury) may dilute all-cause endpoints—making high-quality trial evidence essential for balanced interpretation.
    • Safety signals are critical because higher plasma/platelet exposure could plausibly increase pulmonary, thrombotic, or volume-overload complications.

Design & Methods

  • Research Question: Among patients with severe trauma predicted to need massive transfusion, does a 1:1:1 plasma:platelets:RBC resuscitation strategy (vs 1:1:2) reduce 24-hour and 30-day all-cause mortality?
  • Study Type: Pragmatic, phase 3, multisite, randomised clinical trial; open-label (unblinded); site-stratified randomisation; emergency department trauma resuscitation with blood bank-prepared containers delivered to bedside within 10 minutes.
  • Population:
    • Setting: 12 Level I trauma centres in North America; highest-level trauma activations.
    • Key inclusion features (operational): predicted need for massive transfusion; direct from injury scene; received at least 1 unit of a blood component within 1 hour of arrival or during prehospital transport; age ≥15 years (or weight ≥50 kg).
    • Key exclusions (examples): CPR >5 minutes pre-randomisation; devastating traumatic brain injury (GCS 3 with head AIS ≥5); emergency thoracotomy; pregnancy; prisoners; burns >20% TBSA; known objection/opt-out; expected death within 1 hour of ED admission; fourth unit of RBCs transfused before randomisation.
  • Intervention:
    • 1:1:1 group: transfusion strategy targeting 1 unit plasma + 1 platelet “dose” + 1 unit RBC during active haemorrhage resuscitation.
    • Delivery mechanism: pre-packed blood bank containers (coolers) issued sequentially; components provided in fixed ratios to facilitate immediate, protocolised delivery.
    • Operational goal: first container delivered to bedside within 10 minutes of request; container use continued during active bleeding until anatomic haemostasis or discontinuation of protocolised transfusion.
  • Comparison:
    • 1:1:2 group: transfusion strategy targeting 1 unit plasma + 1 platelet “dose” + 2 units RBC during active haemorrhage resuscitation.
    • Operational distinction: platelets were delivered later relative to RBCs compared with the 1:1:1 group, reflecting the fixed-container design.
    • Co-interventions: subsequent resuscitation (outside the protocolised container phase) and procedural haemorrhage control occurred per clinician judgement and local practice.
  • Blinding: Unblinded at bedside (visible differences in delivered products); mortality endpoints objective; cause-of-death attribution required adjudication.
  • Statistics: Initial planned sample size 580 to detect a 10% absolute reduction in 24-hour mortality (from 21% to 11%) with 90% power at a two-sided overall 5% significance level (alpha adjusted to 0.044 for interim analyses); also powered (88%) to detect a 12% absolute reduction in 30-day mortality (from 35% to 23%); trial ultimately randomised 680 patients; primary analyses were intention-to-treat with site adjustment.
  • Follow-Up Period: 24 hours and 30 days after injury/admission (mortality); in-hospital complications assessed through 30 days.

Key Results

This trial was not stopped early. Enrolment completed at 680 randomised patients.

Outcome 1:1:1 1:1:2 Effect p value / 95% CI Notes
All-cause mortality at 24 h (co-primary) 43/338 (12.7%) 58/342 (17.0%) Adjusted RR 0.75 95% CI 0.52 to 1.08; P=0.12 Risk difference −4.2% (95% CI −9.6 to 1.1)
All-cause mortality at 30 d (co-primary) 75/335 (22.4%) 89/341 (26.1%) Adjusted RR 0.85 95% CI 0.65 to 1.11; P=0.23 Risk difference −3.7% (95% CI −10.2 to 2.7); 4 patients had unknown 30-day status (3 vs 1)
Death due to exsanguination within 24 h 31/338 (9.2%) 50/342 (14.6%) Risk difference −5.4% 95% CI −10.4 to −0.5; P=0.03 Time to death due to exsanguination: 105.7 ± 73.8 vs 96.1 ± 68.7 min; P=0.50
Achieved anatomic haemostasis 291/338 (86.1%) 267/342 (78.1%) Not reported P=0.006 Anatomic haemostasis defined by surgeon (OR) or resolution of contrast blush after embolisation (IR)
Time to anatomic haemostasis 105 (64–179) min 100 (56–181) min Not reported P=0.44 Median (IQR)
Total blood products to 24 h after admission 25.5 (14.0–40.0) units 19.0 (12.0–33.0) units Not reported P<0.001 Median (IQR); includes pre-randomisation + intervention + post-intervention periods
Acute respiratory distress syndrome (to 30 d) 24/334 (7.2%) 19/338 (5.6%) Difference 1.6% 95% CI −2.2 to 5.4; P=0.42 Assessed among those surviving ≥24 h (denominators as reported)
Multiple organ failure (to 30 d) 33/338 (9.8%) 31/342 (9.1%) Difference 0.7% 95% CI −3.8 to 5.2; P=0.79 No signal for excess organ failure with higher plasma/platelet exposure
Venous thromboembolism (DVT or PE) 27/338 (8.0%) 32/342 (9.4%) Difference −1.4% 95% CI −5.8 to 2.9; P=0.53 DVT: 4.5% vs 5.3%; PE: 3.3% vs 4.4%
  • Co-primary all-cause mortality at 24 hours and 30 days was not statistically different between 1:1:1 and 1:1:2 (24 h adjusted RR 0.75; 95% CI 0.52 to 1.08; P=0.12; 30 d adjusted RR 0.85; 95% CI 0.65 to 1.11; P=0.23).
  • Signal concentrated in early haemorrhage control: exsanguination deaths within 24 hours were lower with 1:1:1 (9.2% vs 14.6%; risk difference −5.4%; 95% CI −10.4 to −0.5; P=0.03) and anatomic haemostasis was achieved more often (86.1% vs 78.1%; P=0.006).
  • Despite higher blood product exposure by 24 hours in 1:1:1 (median 25.5 vs 19.0 total units; P<0.001), major complications (ARDS, multiple organ failure, venous thromboembolism) were similar.

Internal Validity

  • Randomisation and allocation
    • Site-stratified randomisation; allocation executed through blood bank preparation of protocolised containers.
    • Patients were randomised at the point of container seal break (operationally linking allocation to initiation of protocolised transfusion).
  • Dropout, exclusions, and missing data
    • Randomised: 680 (338 in 1:1:1; 342 in 1:1:2); randomised blood products transfused to 669 patients.
    • Primary endpoint completeness: 24-hour mortality complete; 30-day status unknown for 4 patients (3 vs 1), with prespecified sensitivity analyses in the supplement showing no material change in inference across missing-data scenarios.
    • Functional outcome (GOS-E) was recorded in a small subset (30 vs 28), limiting interpretability of this exploratory measure.
  • Performance and detection bias
    • Unblinded bedside delivery was unavoidable given visible differences in component sequencing.
    • Primary outcomes (mortality) were objective; cause-of-death attribution required adjudication, reducing (but not eliminating) subjectivity in mechanism-specific endpoints.
  • Protocol adherence and separation of the variable of interest
    • Mean percentage of intervention blood products transfused out of order: 4% (1:1:1) vs 7% (1:1:2); P=0.01.
    • During the intervention phase, separation was achieved for ratios: median plasma:RBC ratio 1.0 (1:1:1) vs 0.5 (1:1:2); median platelets:RBC ratio 1.5 vs 0.4.
    • By 24 hours, total blood products were higher in 1:1:1: median 25.5 (IQR 14.0 to 40.0) vs 19.0 (12.0 to 33.0) units; P<0.001, reflecting greater plasma/platelet delivery.
    • Supplemental process measures: platelets were administered to 98.5% (1:1:1) vs 59.9% (1:1:2); plasma median 7 (IQR 4–11) vs 5 (3–9) units; platelets median 12 (8–17) vs 6 (0–12) units; RBC median 9 (6–17) vs 9 (6–16) units.
  • Baseline characteristics and prognostic enrichment
    • Severe injury burden and shock markers were similar: median Injury Severity Score 26.5 vs 26; shock physiology common (e.g., SBP ≤90 mm Hg in 38.5% vs 39.8%; base excess −8.0 ± 7.6 vs −8.5 ± 7.7).
    • Only ~45–47% ultimately met the trial definition of massive transfusion (≥10 RBC units in 24 hours): 45.3% vs 46.8%.
  • Heterogeneity and site effects
    • Multicentre design across 12 trauma centres; primary analyses were adjusted for site and formal tests did not identify heterogeneity for 24-hour or 30-day mortality.
  • Timing and dose
    • Randomisation occurred rapidly after ED arrival: median 27.5 (IQR 17.0 to 47.0) vs 25.5 (16.0 to 41.0) minutes.
    • Intervention delivered via pre-packed containers designed for immediate bedside transfusion; time to anatomic haemostasis was similar between groups (105 vs 100 minutes; P=0.44).
  • Adjunctive therapies and contamination
    • Use of tranexamic acid and non-TXA procoagulants was similar (TXA 18.9% vs 19.9%; other procoagulants 5.6% vs 4.4%).
    • Post-intervention care permitted clinician-directed transfusion and haemostatic management, which can reduce between-group separation later in the resuscitation course.
  • Outcome assessment and statistical rigour
    • Co-primary outcomes prespecified; intention-to-treat analysis; site-adjusted comparisons; interim monitoring with alpha adjustment.
    • Mechanism-specific outcomes (exsanguination) and complications were systematically assessed; no excess ARDS, multiple organ failure, infection, or thromboembolism was detected.

Conclusion on Internal Validity: Overall, internal validity appears strong, supported by pragmatic site-stratified randomisation, high completeness of mortality follow-up, objective co-primary endpoints, and demonstrable early separation in transfusion ratios; limitations are mainly inherent to the open-label delivery model and later-phase resuscitation contamination.

External Validity

  • Population representativeness
    • Typical of high-acuity trauma resuscitation: median age 34 years; predominantly male (~75%); blunt trauma ~51–55% and penetrating ~45–49%.
    • Enrolled from highest-level trauma activations; physiologic shock common (SBP ≤90 mm Hg ~39%; HR ≥120 ~44%).
    • Key exclusions (devastating isolated TBI, burns, pregnancy, prolonged CPR, emergency thoracotomy) limit applicability to those subgroups.
  • Applicability across systems
    • Most applicable to mature trauma systems with 24/7 blood bank capacity for rapid issue of thawed plasma and platelets in pre-packed containers.
    • Generalisability may be constrained in resource-limited settings (platelet availability, plasma thaw times, cold-chain logistics) and where prehospital transfusion/whole blood strategies dominate.

Conclusion on External Validity: External validity is moderate-to-high for Level I/major trauma centres operating MTPs with rapid component availability, but is more limited for settings without immediate access to plasma and platelets or for excluded phenotypes (e.g., devastating isolated TBI, major burns, pregnancy).

Strengths & Limitations

  • Strengths:
    • Large, pragmatic, multicentre randomised trial in time-critical trauma haemorrhage resuscitation (a domain historically dominated by observational inference).
    • Operationally realistic intervention (blood bank-prepared containers; early bedside delivery), increasing relevance to MTP implementation.
    • Co-primary outcomes prespecified; high completeness of mortality follow-up; complications systematically collected.
    • Demonstrated early separation in delivered plasma and platelet ratios during the active transfusion phase.
  • Limitations:
    • Unblinded intervention with potential for co-intervention differences, particularly after the protocolised transfusion phase.
    • All-cause mortality endpoints may dilute haemorrhage-specific effects because traumatic brain injury and other non-haemorrhagic causes contribute substantially to deaths.
    • Only ~45–47% met massive transfusion criteria, implying imperfect prognostic enrichment and possible attenuation of detectable treatment effect.
    • Exploratory functional outcome data (GOS-E) were available in a small subset, limiting patient-centred recovery inference.

Interpretation & Why It Matters

  • Clinical practice
    • PROPPR supports early, balanced component resuscitation as a haemorrhage-control strategy: more patients achieved anatomic haemostasis and fewer died of early exsanguination with 1:1:1, without an observed penalty in ARDS, multiple organ failure, or thromboembolism.
    • Because co-primary all-cause mortality endpoints were neutral, the primary “practice impact” is best framed as improved early haemorrhage control rather than definitive all-cause survival benefit.
  • Mechanistic signal
    • The physiologic target is early correction of trauma-induced coagulopathy and provision of haemostatic substrate contemporaneously with red cells—achieved via higher plasma and platelet ratios during the active transfusion phase (plasma:RBC 1.0 vs 0.5; platelets:RBC 1.5 vs 0.4).
  • Trialists and methodologists
    • PROPPR illustrates both the necessity and the difficulty of maintaining separation in emergency resuscitation trials: later resuscitation contamination can dilute mortality effects even when early protocol delivery differs substantially.
    • It also highlights endpoint selection challenges in trauma: early haemorrhage-specific endpoints may be more sensitive to transfusion strategy than 30-day all-cause mortality.

Controversies & Other Evidence

  • Primary endpoint neutrality vs haemorrhage-specific benefits
    • Published correspondence emphasised that the co-primary all-cause mortality outcomes did not differ significantly between groups and cautioned against over-interpreting secondary endpoints in isolation, while the authors’ reply defended the clinical relevance of haemorrhage-specific outcomes and protocol fidelity.12
  • Endpoint timing and analytic framing
    • A later Bayesian re-analysis of PROPPR data reframed inference around probability of benefit rather than dichotomous significance testing, reporting a high posterior probability of mortality benefit for 1:1:1 despite neutral frequentist primary results; an accompanying commentary discussed the pragmatic appeal and limitations of Bayesian approaches in trauma research.34
  • Guideline synthesis and implementation
    • Contemporary European guidance on traumatic major bleeding incorporates the PROPPR-era evidence base and endorses early balanced blood component resuscitation within damage control strategies, while recognising system constraints and the need for rapid haemorrhage control alongside transfusion.5
  • Persisting practical debates
    • Optimal “ratio-based” component therapy is now often considered alongside alternative paradigms (e.g., whole blood, viscoelastic-guided resuscitation, prehospital plasma), making PROPPR foundational but not definitive for all modern damage control resuscitation models.

Summary

  • In 680 severely injured trauma patients predicted to need massive transfusion, 1:1:1 vs 1:1:2 plasma:platelets:RBC did not significantly reduce co-primary 24-hour (12.7% vs 17.0%) or 30-day all-cause mortality (22.4% vs 26.1%).
  • 1:1:1 produced a clinically coherent early haemorrhage-control signal: fewer deaths from exsanguination within 24 hours (9.2% vs 14.6%) and more frequent achievement of anatomic haemostasis (86.1% vs 78.1%).
  • Protocol separation was demonstrable during the active transfusion phase (median plasma:RBC 1.0 vs 0.5; platelets:RBC 1.5 vs 0.4), but later resuscitation could attenuate differences.
  • Total blood product exposure by 24 hours was higher with 1:1:1 (median 25.5 vs 19.0 units), yet major complications (ARDS, multiple organ failure, thromboembolism) were similar.
  • PROPPR remains the landmark multicentre RCT underpinning ratio-based damage control resuscitation and informing modern MTP design and guideline recommendations.

Overall Takeaway

PROPPR established that a protocolised early 1:1:1 plasma:platelets:RBC strategy is feasible and safe in real-world trauma resuscitation, and that it improves early haemorrhage control (haemostasis and exsanguination deaths) without clear evidence of excess organ dysfunction. While co-primary all-cause mortality endpoints were neutral, the trial’s rigorous methodology and pragmatic delivery model made it the key randomised foundation for contemporary massive transfusion protocol practice.

Overall Summary

  • PROPPR supports early balanced component resuscitation (1:1:1) as a haemorrhage-control strategy with neutral all-cause mortality but reduced early exsanguination and no evident safety penalty.

Bibliography