Prehospital Plasma during Air Medical Transport in Trauma Patients at Risk for Hemorrhagic Shock

Sperry et al. NEJM. Published 26th July, 2018. N Engl J Med 2018;379:315-26.
DOI: 10.1056/NEJMoa1802345

Clinical Question

  • In severely injured patients at risk for haemorrhagic shock, does prehospital plasma resuscitation, compared with standard-care resuscitation (not including plasma administration), reduce 30-day mortality?


  • Bleeding trauma patients have traditionally been managed with the prehospital infusion of crystalloids, and with component transfusion (RBCs, plasma, platelets) started once the patient arrives in hospital.
  • Recent evidence that suggests the early institution of blood components that more closely resembles whole blood may help prevent trauma coagulopathy, downstream traumatic shock, and improve outcomes (“AKA haemostatic transfusion”)
    • The PROPPR trial (2015) was a multicenter RCT of in hospital patients at level 1 trauma centers comparing massive transfusion protocols based on plasma, platelets and RBCs ratios of 1:1:1 vs 1:1:2. There was no statistically significant mortality difference on the primary 30 day mortality outcome
      • However death from exsanguination was reduced (9.2% vs 14.6% in 1:1:2 group; difference, −5.4% [95% CI, −10.4% to −0.5%]; P = .03
  • The majority of studies have focussed on in hospital patients, with few focussing on the pre hospital management
    • One exception is the COMBAT trial, published in July 2018. This was a single centre, placebo controlled, RCT of patients with haemorrhagic shock, which compared the prehospital infusion of plasma to a normal saline placebo. There was no difference in day 28 day mortality between plasma and crystalloid 15% vs 10% P=NS.
      • This study was blinded, however transport times were very rapid 16-19 mins, and it was only a single centre with smaller numbers
  • Therefore the efficacy and safety of prehospital initiation of plasma resuscitation in bleeding trauma patients in a large multicentre trial is not well defined.


  • Pragmatic, phase 3, multicentre, cluster-randomized trial
  • Team in ambulances screened and enrolled patients
  • Randomisation was of air medical services (2 to 6 bases or helicopters per service), and not individual patients.
    • These were block randomized and assigned to the plasma arm or standard care (control) arm for 1 month at a time.
    • The block scheme was varied randomly between 2, 4, and 6 month block sizes over the period of enrollment for the trial at each participating center.
  • Delayed consent approach after community consultation
  • No blinding of prehospital and hospital staff, but trial assessors were blinded
  • No commercial support for trial
  • Power calculation
    • Assuming a 1:1 randomization of 32 clusters or 16 patients each, 530 patients (504 eligible patients with complete data) would provide the trial with 88% power to detect a difference of 14 percent age points (8.0% vs. 22.0%) in 30-day mortality
    • The sample size was increased to 564 patients after 2 years and 8 months owing to a higher rate of ineligible patients in the prehospital setting than was initially estimated.
    • Statistical significance for the primary analysis (P<0.038) was adjusted for two interim analyses by the external data safety monitoring board


  • 27 pre hospital air medical transport services that transport patients to 9 participating level 1 trauma centers
  • All centers in the United States
  • Enrolment from 2014 to October 2017


  • Inclusion:
    • Blunt or penetrating trauma patients, and
    • Transported from a scene of their injury or from an outside referral ED to a participating trauma centre, and
    • At risk of haemorrhagic shock, as defined by at least one episode of:
      • Hypotension (systolic blood pressure <90mmHg) and tachycardia (HR>108/min)
      • Severe hypotension (systolic blood pressure <70mmHg)
  • Exclusion:
    • >90 years old or <18 years old
    • Unable to establish IV or IO access
    • Isolated fall from standing
    • Documented cervical cord injury
    • Known to be a prisoner
    • Pregnant
    • Traumatic cardiac arrest >5 minutes
    • Penetrating brain injury
    • Isolated drowning or hanging
    • >20% TBSA burns
    • Admitted to an outside referral hospital
    • Patient or family member objected to participation in the trial at the scene
  • A total of 501 patients were enrolled
    • 390 enrolled directly at scene, while 111 transferred from a non-trauma centre emergency department
    • 72.7% were men
    • Median age 45 years old
    • 82.4% had blunt trauma
    • High risk trauma population
      • Median injury severity score 22 (13-30)
      • Median MAP 70 Median HR 116
      • Prehospital intubation in 51.1%
      • 34.7% received prehospital red blood cells
      • 58.4% had urgent operations in initial 24hours
      • Overall mortality 29.6%
  • Assigned treatment given in 496/501 (99.0%) patients
  • A total of 42 patients were lost to follow-up (19 of plasma vs 23 crystalloid)
  • Similar baseline demographics, prehospital vital signs, and injury scores in all areas except
    • Prehospital crystalloid was lower in plasma group median 500mls (0-1250) vs 900mls (0-1500)
    • Prehospital RBC transfusion was lower in plasma group 26.1% vs 42.1%


  • Air medical bases randomized to the intervention were provided with 2 units of thawed plasma stored in coolers with a temperature between 1 to 10 degrees
  • Plasma was either group AB (“universal donor”) or group A with a low anti-B antibody titre (<1:100)
  • Plasma was initiated in prehospital setting by air transport team before other resuscitative fluids were given
  • The infusion was required to be completed once started
  • The infusion was not previously part of standard care in any of the participating sites
  • Following completion of the plasma, goal directed resuscitation as described below continued until arrival at trauma centre, including the infusion of crystalloids and or RBCs


  • Goal-directed, crystalloid-based resuscitation targeting a systolic blood pressure of ≥90mmHg
    • This was not standardised, but was guided by local  air ambulance protocols

Management common to both groups

  • In addition to above crystalloid resuscitation, 13/27 air transport teams also carried 2 units of universal donor red cells
    • RBC transfusion was indicated if after 1L of crystalloid there was ongoing
      • Hypotension with systolic blood pressure <90mmHg
      • Changes in mental status
      • Tachycardia with heart rate >120 beats per minute
      • Capillary refill >2 seconds
    • This was done in the intervention group only after the plasma infusions was complete


  • Primary outcome:
    • 30 day mortality was significantly lower in the plasma group compared to control (23% vs. 33.0%; difference−9.8%, 95% CI −18.6 to −1.0; P=0.03)
      • Number needed to treat to prevent one death at 30 days was 10
      • Multiple sensitivity analyses performed to account for missing data and significant differences remained
      • An analysis that controlled for differences in volume of crystalloid and prehospital red-cell transfusion demonstrated that administration of prehospital plasma was still associated with a reduced risk of death within 30 days after randomization (adjusted odds ratio, 0.61; 95% CI, 0.40 to 0.91; P = 0.02).
      • Kaplan– Meier survival curves showed an early separation of the two groups that began 3 hours after randomization and remained until 30 days after randomization (log-rank chi-square test, 5.70; P=0.02)
    • Subgroup analysis of primary outcome revealed lower mortality at 30 days in most prespecified subgroups, however after adjusting for multiple comparisons, no significant interactions were found.
  • Secondary trial outcomes
    • The plasma group, compared to standard care, had the following outcomes
      • 24 hour mortality 13.9% vs 22.1% P=NS
      • In-hospital mortality 22.2 % vs 32.5% P=NS
      • Median 24 hour volume of RBCs 3 (0-7) vs 4 (1-9) P=NS
      • Median 24 hour volume of crystalloid 4388 (2225-6320) vs 4500 (3000-6800) P=NS
      • Vasopressors received in first 24hours 45.2% vs 50.9% P=NS
      • Multiorgan failure 63.0% vs 57.6% P=NS
      • Median initial INR lower 1.2 (1.1-1.4) vs 1.3 (1.1-1.6) P<0.001
    • Allergic reaction or transfusion related reaction higher 2.2 vs 0.4% P=NS
      • These were classified as minor and were considered “possibly related” to the trial treatment

Authors’ Conclusions

  • In patients at risk for hemorrhagic shock, the administration of thawed plasma during prehospital air medical transport was safe and resulted in lower 30-day mortality and a lower median prothrombin-time ratio than standard-care resuscitation.


  • Multi centre, randomized controlled trial
  • Important relevant question as majority of trauma related deaths due to haemorrhage occur early, so prehospital interventions are important to research
  • Primary analysis was adjusted to account for prehospital crystalloid and RBC transfusion differences, therefore reducing any confounding these may have had on the results.
  • Population identified was relevant to the study question, with a high proportion having haemorrhage that required transfusion and surgery.
  • Intervention was relatively simple to deliver and would be reproducible in systems already with capacity to deliver blood products
  • Early separation of the KM effect makes its treatment effect more plausible


  • Imbalances of prehospital crystalloid and RBC transfusion complicate interpretation
    • Plasma group had less crystalloid and RBC transfusions compared to standard care
    • Probably as a result of the volume that was also given in plasma (300ml plasma x 2 = 600mls)
    • However when a multivariate regression model was used to adjust for these factors, it did not effect the overall result – adjusted odds ratio 0.61; 95% CI, 0.40 to 0.91; P = 0.02
  • No blinding of prehospital and hospital clinicians increased the risk they changed their behaviours and biased the results
  • Questionable biological plausibility
    • A 10% drop in mortality is a very large reduction for a single intervention of 600mls of plasma
  • Unclear mechanism of action
    • The large mortality benefit appears to have occurred very early (see KM curves) and a 8.2% mortality difference by 24hours
    • The difference in INR (1.2 vs 1.3), although statistically significant, is unlikely to account for all of this mortality difference.
    • Other potential mechanisms (?less crystalloid, less RBCs, anti-inflammatory effects) not clarified
    • The total volume given prehospital was not reported, but was probably similar in both groups, and blood pressure and heart rates very similar
Allocation Crystalloid (mls) Plasma (mls) RBC (% transfused) Prehospital total (mls)
Plasma 500 600 26% ?Similar
Crystalloid 900 0 42% ?Similar
  • Resuscitation protocols were not standardized, but instead were dictated by local protocols, which may have resulted in  variation in delivered treatments
  • Generalisability to other air medical transport systems is not clear
    • Currently many air transport services do have the capacity to carry blood products (14/27 did not the trial)
    • Results may not be generalisable to other system with longer or shorter transport times (i.e. > or < 41 minutes)
    • AB plasma is a rare resource (only accounts for 3-4% of the blood donor pool), and is therefore may be difficult to access in different systems
    • Many centres do not use group A with low antibody titres, making availability of the plasma potentially more difficult
    • Majority of patients had blunt trauma, so results not generalisable to penetrating trauma
  • No mention of the use of prehospital tranexamic acid
  • Economic analysis not performed
    • Although plasma is relatively inexpensive in most systems (for example, AU$280/unit), it requires a lot of resources and logistics to make it work safely in prehospital setting.

The Bottom Line

  • The early (prehospital) initiation of plasma results in a significant fall in mortality in patients at high risk of bleeding without a significant increase in harm.
  • Further blinded studies are needed to confirm this benefit, and to test its applicability in different trauma systems.

External Links


Summary author: Aidan Burrell
Summary date: 01 August 2018
Peer-review editor: Duncan Chambler

One comment

  • Gianluca Castellani

    I am quite impressed by the difference in result between Pamper and Combat. Severity of patients is similar, mortality is consistently different.
    Do we have data about transport time in PAMPer?
    Thanks for your fantastic website!
    Gianluca Castellani, Milan.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.