Trauma Resuscitation with Group O Whole Blood Or Products (TROOP) Trial

Trauma Resuscitation with Group O Whole Blood Or Products (TROOP) Trial is sponsored by National Heart, Lung, and Blood Institute (NHLBI). A multi-center, phase 3 clinical trial comparing whole blood transfusions to blood component therapy in trauma patients with major bleeding and activation of massive transfusion protocol. The study aims to determine if whole blood improves survival outcomes.

## [](https://pmc. ncbi. nlm.

nih. gov/articles/PMC12318426/)Abstract Hemorrhage is the most common cause of potentially preventable death after injury. Balanced transfusion with red blood cells, plasma, and platelets (component therapy, CT) has been shown to reduce mortality, and is the standard of care.

Low-Titer Group O Whole Blood (LTOWB) is an attractive alternative to CT, but existing evidence comprises observational studies, and a small single center pilot randomized controlled trial, which evaluated a type of whole blood that is no longer in use.

The aim of the “Trauma Resuscitation with Low-Titer Group O Whole Blood Or Products” (TROOP) trial is to compare the effectiveness and safety of LTOWB and CT in critically injured patients predicted to require a large volume transfusion. This is a pragmatic, multicenter, Bayesian, sequential non-inferiority/superiority, randomized clinical trial, performed within 15 level I trauma centers in the United States.

We aim to randomize 1,100 injured patients to resuscitation with either CT or LTOWB. The primary outcome is 6-h mortality.

Secondary outcomes include 24-h and 30-day or hospital mortality (whichever is earlier); prespecified complications; adjudicated cause of death; time to death; length of stay (ICU and hospital); and hospital-, ventilator- and ICU-free days; the incidence of major surgical procedures; time to hemostasis in those undergoing procedures with a hemostatic component; number and type of blood products used until hemostasis is achieved (and randomized products are discontinued), as well as after hemostasis has been achieved, to 24 h post-admission; discharge destination and functional status and quality of life at hospital discharge or 30 days, as measured by Glasgow Coma Scale (GCS) and EuroQol (EQ-5D) quality of life measurement.

This large multicenter clinical trial will contribute high-level evidence on the effectiveness of Low-Titer Group O Whole Blood in the in-hospital management of trauma patients predicted to require a large volume transfusion. National Clinical Trial Identified Number: NCT05638581. Clinical trial registry: https://clinicaltrials.

gov/study/NCT05638581 First submitted 2022–11-08. **Keywords:**Trauma, Injury, Resuscitation, Low-Titer Group O Whole Blood, Component therapy ## [](https://pmc. ncbi.

nlm. nih. gov/articles/PMC12318426/)Administrative information Note: the numbers in curly brackets in this protocol refer to SPIRIT checklist item numbers.

The order of the items has been modified to group similar items (seehttp://www. equator-network. org/reporting-guidelines/spirit-2013-statement-defining-standard-protocol-items-for-clinical-trials/)Title {1}Trauma Resuscitation with Low-Titer Group O Whole Blood or ProductsTrial registration {2a and 2b}National Clinical Trial Number: NCT05638581 (Registry: Clinicaltrials.

gov)Protocol version {3}Version 1. 7; October 29th, 2024Funding {4}National Heart Lung and Blood Institute, National Institutes of HealthAuthor details {5a}Jan O.

Jansen, Center for Injury Science, Department of Surgery, University of Alabama at Birmingham, USAClaudia Pedroza, Department of Pediatrics, The University of Texas Health Science Center at Houston, Houston, Texas, USALuis Leon Novelo, Coordinating Center for Clinical Trials, The University of Texas Health Science Center at Houston, Houston, Texas, USATianyao Hao, Coordinating Center for Clinical Trials, The University of Texas Health Science Center at Houston, Houston, Texas, USAGina R.

DeWildt, Coordinating Center for Clinical Trials, The University of Texas Health Science Center at Houston, Houston, Texas, USACharles F. Coton, Coordinating Center for Clinical Trials, The University of Texas Health Science Center at Houston, Houston, Texas, USAKiran Mansoor, Coordinating Center for Clinical Trials, The University of Texas Health Science Center at Houston, Houston, Texas, USAShannon W.

Stephens, Center for Injury Science, Department of Surgery, University of Alabama at Birmingham, USAMarisa B. Marques, Center for Injury Science, Departments of Surgery and Pathology, University of Alabama at Birmingham, USAJames R. Stubbs, Mayo Clinic, Rochester, Minnesota, USAJillian R.

Richter, Center for Injury Science, Department of Surgery, University of Alabama at Birmingham, USAHenry E. Wang, Department of Emergency Medicine, The Ohio State University, Columbus, OHJohn B. Holcomb, Center for Injury Science, Department of Surgery, University of Alabama at Birmingham, USAStacia M.

DeSantis, Coordinating Center for Clinical Trials, The University of Texas Health Science Center at Houston, Houston, Texas, USAAnd the TROOP Trial Investigators (listed at end of paper)Name and contact information for the trial sponsor {5b}University of Alabama at Birmingham (UAB)Role of sponsor {5c}The sponsor played no part in study design; collection, management, analysis or interpretation of data. ## [](https://pmc. ncbi.

nlm. nih. gov/articles/PMC12318426/)Introduction ### Background and rationale {6a} Trauma is the leading cause of death in adults (below the age of 46) and children in the USA.

Injuries account for more years of potential life lost before age 75 than any other cause, including cancer or heart disease [1]. Bleeding is responsible for approximately one-third of trauma deaths, and around one-quarter are thought to be potentially preventable [2, 3]. Blood transfusion is an essential part of trauma resuscitation.

Early and rapid replacement of lost blood is key to reversing shock and restoring cellular perfusion and hemostasis [4]. Since the late 1970 s, blood banking practices have separated donated “whole” blood into “components”—including red blood cells, plasma, platelets, and cryoprecipitate—to allow a unit of donated whole blood to serve multiple patients.

However, severely injured patients require all of these components, and during resuscitation efforts healthcare providers now aim to concurrently administer red blood cells, plasma, cryoprecipitate, and platelets in ratios that approximate whole blood [5].

Based on the results of the Pragmatic, Randomized, Optimal Platelet and Plasma Ratios (PROPPR) trial [6] as well as extensive military experience from the wars in Iraq and Afghanistan [7], the current standard of care in trauma transfusion is the balanced administration of equal ratios of blood components. Clinical practice guidelines affirm these approaches [5]. There are pitfalls associated with balanced transfusion.

Each component contains a volume of preservatives and anticoagulants that when summed is not inconsequential [8]. The components have variable shelf-lives, platelets expire quickly, frozen plasma has to be thawed, and cryoprecipitate is often not given. It is also labor intensive to track and transfuse such large numbers of blood products.

Thus, many trauma centers struggle to provide a balanced resuscitation for patients in hemorrhagic shock [9, 10]. Even when all the blood products are on hand, keeping track of the units ordered and administered can be difficult, resulting in an unbalanced resuscitation [11].

An alternative to component blood products is low-titer group O whole blood (LTOWB)—a product that contains type O red blood cells, plasma, fibrinogen, and platelets, and only low levels of antibodies (anti-A and anti-B). The use of LTOWB is associated with several potential benefits. Most importantly, it is associated with improved survival in multiple observational studies [11–47].

LTOWB contains higher levels of hemoglobin, coagulation proteins, and functioning platelets [48], and has been shown to better mitigate against shock in animal models as well as humans [37, 45, 49]. In addition to possibly reducing mortality, LTOWB offers several substantial logistical and administrative advantages [50].

It is easier and faster for blood banks to issue whole blood, since it is contained in one bag, and the transfusion of whole blood requires fewer steps and ensures the correct ratio. A recent study has shown reduced “flow time” from making the decision to transfuse, to completion of transfusion, as well as decreased hands-on “touch-time,” fewer required steps, and fewer required checks [50].

The objectives of the TROOP trial are to compare the effectiveness and safety of LTOWB and component therapy in critically injured patients predicted to require a large volume blood product transfusion. TROOP is a pragmatic, multicenter, Bayesian, group-sequential, non-inferiority, phase III, randomized clinical trial; if non-inferiority for the primary outcome is met, superiority will be reported.

The trial’s design and development was supported by an NIH/NHLBI planning grant (1U34HL148472-01). This protocol is written in accordance with the “Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) checklist” (and item numbers are included in curly brackets). The schedule of enrollment, interventions, and assessments for the duration of the study is tabulated in Fig.

1. The trial includes injured adults with hemorrhagic shock predicted to require large volume blood transfusions, who will be randomized to receive either LTOWB or component therapy (CT; Fig. 2).

Fig. 1Schedule of Enrollment, interventions, and assessments. a This is an EFIC study.

Attempts at obtaining consent and contacting legal representatives will commence as soon as the patient arrives in the emergency department, but may take time to completeFig. 2CONSORT diagram ## [](https://pmc. ncbi.

nlm. nih. gov/articles/PMC12318426/)Methods: participants, interventions, and outcomes The trial is being conducted in 15 level I trauma centers in the USA, listed in Fig.

3. Level I trauma centers are specialized, tertiary centers designated to provide definitive care for seriously injured patients.

The trial sites are high-acuity, high-volume centers that care for trauma patients with hemorrhagic shock; have existing clinical experience with LTOWB, yet maintain clinical equipoise; have blood banks that routinely have at least eight units of LTOWB on hand; have existing research infrastructure, including the ability to enroll subjects anytime, day or night; and have extensive clinical trial experience.

Sites’ stock levels of LTOWB were evaluated as part of the site set-up process, and some sites amended their ordering schedules from their respective blood centers. Fig. 3Participating level I trauma centers ### Eligibility criteria {10} The study population consists of injured patients predicted to require a large-volume blood transfusion, and who are at high risk of dying (> 15%) within hours of admission.

The inclusion and exclusion criteria are listed in Fig. 4. These pragmatic criteria are based on recently conducted trials, and strike a balance between specificity and sensitivity, to identify those patients most likely to benefit, based on their clinical characteristics.

The “Assessment of Blood Consumption” (ABC) score is a simple clinical tool for predicting the likelihood of a trauma patient requiring a large-volume transfusion [51].

The score consists of the following elements: (a) penetrating mechanism of injury; (b) systolic blood pressure in the emergency department ≤ 90 mmHg; (c) heart rate in the emergency department ≥ 120 beats per minute; and (d) positive Focused Abdominal Sonography for Trauma (FAST) scan. “Confirmed acute major bleeding,” unless the bleeding is external, typically requires computed tomography scanning.

However, in the most seriously injured patients, such confirmation is not possible, because these patients require immediate hemorrhage control (usually operatively) and could die if taken to the CT scanner before the OR. “[…] suspected acute major bleeding” refers to the global clinical Gestalt assessment conducted by a team of providers, led by experienced clinicians [52].

This type of assessment takes into consideration injury mechanism, injury pattern, vital signs, bedside diagnostic (such as radiographs and ultrasound scans), need for immediate intervention, and response to treatment [53]. Fig. 4Entry criteria We are conducting the TROOP trial under United States Food and Drug Administration (FDA) rules for Exception From Informed Consent (EFIC) (21 CFR 50.

24). The patients who meet the inclusion criteria for the TROOP trial are critically ill, at imminent risk of exsanguination and death, and incapacitated and unable to give verbal or written prospective informed consent. Given the time-critical nature of the clinical setting, it is also not possible to seek consent from legally authorized representatives.

### Additional consent provisions for collection and use of participant data and biological specimens {26b} The trial does not involve the collection of biological specimens. However, 400 patients will be enrolled in an ancillary study funded by the Department of Defense, called bioTROOP. Separate consent provisions exist for participation in bioTROOP.

## [](https://pmc. ncbi. nlm.

nih. gov/articles/PMC12318426/)Interventions ### Explanation for choice of comparators {6a} The TROOP trial compares the current standard of care, balanced component therapy, and the use of LTOWB. LTOWB is an FDA-approved product, and available in 50% of Level 1 and 2 trauma centers in the USA [54].

TROOP does not dictate either Rh positive or Rh-negative blood products, but rather reflects the standard of care as provided by each center. ### Intervention description {11a} The intervention is a resuscitation strategy based on LTOWB. LTOWB will be obtained from suppliers registered with the FDA and will have a maximum acceptable anti-A and anti-B titer of 256.

LTOWB will be delivered to the trauma bay, operating room, or interventional radiology suite in a cooler issued by the blood bank; or drawn from blood bank refrigerators in resuscitation areas. Uncrossmatched group O blood will be utilized because there is insufficient time to type the patient and crossmatch blood units, in the setting of serious hemorrhage.

Sites must have at least eight units of LTOWB in the blood bank for the study in order to enroll a patient in the trial. Treating centers will continue to provide LTOWB as long as it is available in the blood bank. In women of childbearing age [55], sites may use Rh negative (O −) or Rh positive (O +) group O whole blood, in line with local operating procedures.

Patients in the LTOWB arm may be given plasma, platelets, and cryoprecipitate (or other factor concentrates), as deemed necessary, but not red blood cells. Patients allocated to “control” treatment will receive CT (which is the current standard of care) [5, 56].

Blood components will also be delivered to the trauma bay, operating room, or interventional radiology suite in a cooler issued by the blood bank, or drawn from blood bank refrigerators in resuscitation areas, utilizing their standard procedure.

Following the PROPPR trial, Level I trauma centers have adopted massive transfusion protocols, aiming to deliver high ratios of units of plasma and platelets for every unit of red blood cells given upon arrival. Many sites now also keep LTOWB or components in the Emergency Department.

Based on local practice, the precise ratios transfused will vary, and will be recorded, but the exact ratio will not be prescribed, in keeping with the pragmatic philosophy of the trial. Patients in the CT arm may also receive any component or factor concentrates, as deemed necessary, but not whole blood.

The administration of randomized blood products will cease (by notifying the blood bank that no further randomized products are required) when anatomical hemostasis has been achieved (confirmed by the surgeon), and resuscitation has been completed (defined as the resolution of transfusion need; agreed by surgeon and anesthesiologist) [57].

### Criteria for discontinuing or modifying allocated interventions {11b} Patients who have been randomized to the LTOWB arm may receive CT when stocks of LTOWB have been exhausted. Patients who have been randomized to the CT arm may be given LTOWB if there are no platelet units available, and the patient is deemed to require platelets.

### Strategies to improve adherence to interventions {11c} Sites have been extensively trained on how to implement the intervention, during an in-person site initiation visit with clinicians, as well as blood bankers. Adherence is monitored using the CRF, and non-adherence results in additional educational measures being put in place. Randomized blood products should be available at the bedside within 10 min of request.

### Relevant concomitant care permitted or prohibited during the trial {11d} Inclusion in the TROOP trials does not require alteration to usual care pathways, other than transfusion strategy. All relevant medical care is permitted, and all other concurrent and subsequent care will follow standard practices and local protocols. At no time will study procedures slow down the delivery of blood products to bleeding patients.

### Provisions for post-trial care {30} There is no provision for post-trial care beyond the standard care provided to severely injured patients, except for Rh negative women of childbearing age who received Rh-positive blood. These women will be followed up after approximately 3 months and evaluated for alloimmunization. The primary outcome is mortality at 6 h.

For the purpose of determining the end of the 6-h time period, the clock will start when the patient is enrolled/randomized, by opening the sealed cooler, or the ED refrigerator (following pre-randomization). Because of the short-term nature of the outcome, we expect no loss to follow-up for the primary endpoint. Participants will be followed up to discharge, or 30 days, whichever is earlier.

Secondary endpoints include 24-h and 30-day/hospital mortality (whichever is earlier); complications; cause of death; time to death, length of stay (ICU and hospital); hospital-, ventilator-, and ICU-free days; incidence of major surgical procedures; time to hemostasis in those undergoing procedures with a hemostatic component; number and type of blood products used until hemostasis is achieved (and randomized products are discontinued) and after hemostasis is achieved (up to 24 h post-admission); discharge destination and functional status and quality of life at hospital discharge or 30 days, as measured by Glasgow Coma Scale (GCS) and Euroqol (EQ-5D) quality of life measurement.

Pre-specified complications include acute kidney injury; ventilator-associated pneumonia; multi-organ failure; transfusion-related hyperkalemia; transfusion-related hypocalcemia; transfusion- associated circulatory overload; transfusion-associated lung injury; acute respiratory distress syndrome; symptomatic and asymptomatic deep vein thrombosis; symptomatic and asymptomatic pulmonary embolism; bleeding after hemostasis requiring intervention; stroke; myocardial infarction; abdominal compartment syndrome; transfusion-related allergic reactions; febrile non-hemolytic transfusion reaction; systemic inflammatory response syndrome; sepsis; and alloimmunization in women of childbearing age.

### Participant timeline {13} The participant timeline is shown in Fig. 1. If the trial continues to full enrollment, non-inferiority will be tested first.

Then if LTOWB is declared non-inferior, probability of superiority is calculated. The probability threshold for declaring success for non-inferiority is explained below. As we will report but not declare superiority if non-inferiority is met, no superiority threshold is specified.

Screening, recruitment, and baseline data collection take place in the trauma bays of the emergency departments. Once all sites have been activated, the trial is expected to enroll 7. 7 patients per week (0.

5 patients per site per week). Inclusion in the TROOP trial is determined by physiological characteristics. Patients are randomized individually, within 90 min of arrival in the Emergency Department.

The choice of enrollment window is based on a secondary analysis of the PROPPR trial, which showed that patients enrolled after 90 min had different characteristics and outcomes [60]. Patients are considered enrolled and randomized when the cooler or blood bank refrigerator is opened, in the trauma bay/emergency department. The 6-h and the 24-h time windows are based upon the time of cooler/refrigerator opening.

## [](https://pmc. ncbi. nlm.

nih. gov/articles/PMC12318426/)Assignment of interventions: allocation ### Sequence generation {16a} The randomization mechanism has been designed to have minimal impact on the speed with which life-saving blood products can be provided by the blood banks. Randomization is stratified by site with 1:1 (equal) randomization to LTOWB or CT, using a randomized permuted block design within site.

The randomization assignments are computer-generated by the DCC. They are then physically provided to and stored in each blood bank, separate from the trauma bay. A random assignment scratch-off sheet is always available to study teams.

When notified by clinical staff of the arrival of a potential TROOP trial patient, the blood bank technologist will consult the TROOP Trial Randomization binder, which contains sequentially numbered randomization sheets. Allocation is determined by “scratching off” a latex covering (similar to a bank PIN number or lottery number) which reveals which arm the patient has been pre-randomized to.

This, in turn, will determine whether the cooler that will be provided to the bedside will be filled with LTOWB, or components. The cooler is sealed with adhesive tape. In recent years, trauma centers have increased the accessibility of blood products, by locating refrigerators in resuscitation areas.

In addition to the above cooler-based mechanism, TROOP allows for LTOWB and blood products from such refrigerators to be used once the patient has been allocated to a study arm. Patients are not enrolled/randomized until the seal is broken, and the cooler (or ED refrigerator) is opened.

This process promotes the early activation of the MTP (so that blood is available if needed), while also ensuring that patients who—contrary to the initial impression—do not require transfusion are not enrolled in the trial. If a patient is not enrolled/randomized, and the cooler is returned to the blood bank unopened, the randomization number will be retired, and a fresh number will be used for the next trial patient.

Randomization to the two arms will be evaluated regularly for balance by the DCC. Every 9 months, fresh randomization lists will be provided to sites. This process maintains an equal likelihood of receiving either treatment over the course of the trial.

### Concealment mechanism {16b} The study protocol seeks to maintain treatment blinding to the maximum extent possible until the moment of enrollment/randomization. We will achieve this by having all centers use similar opaque blood transport coolers. Allocation is concealed until the first cooler is opened.

Coolers are closed and sealed with adhesive tape. Since platelets are typically stored at room temperature, they are often transported in a separate external pouch outside of the cooled compartment. To maintain blinding of the LTOWB group, decoy saline fluid bags are placed in the platelet compartment of the coolers containing whole blood.

Research staff present in the trauma bay will be made aware of the patient’s allocation by the blood bank, but will not disclose this information to the treating clinicians, unless blood has to be obtained from a refrigerator (prior to the TROOP cooler arriving from the blood bank).

Once the cooler/refrigerator (following allocation) has been opened, clinical teams implement the intervention by transfusing the blood products (LTOWB or CT) in the cooler. Additional coolers are sent by the blood bank, as required. ## [](https://pmc.

ncbi. nlm. nih.

gov/articles/PMC12318426/)Assignment of interventions: blinding ### Who will be blinded? {17a} Clinicians are blinded until the cooler/refrigerator (following allocation) has been opened. After the cooler/refrigerator has been opened, TROOP is an open-label study, and neither patients nor investigators are blinded.

### Procedure for unblinding if needed {17b} There is no need for an unblinding procedure. ## [](https://pmc. ncbi.

nlm. nih. gov/articles/PMC12318426/)Data collection and management ### Plans for assessment and collection of outcomes {18a} All data are collected by dedicated research staff, and participating sites are required to maintain around-the-clock research staff coverage.

### Plans to promote participant retention and complete follow-up {18b} The trial intervention pertains only to inpatient management, and the primary outcome is assessed early. Retention is therefore not anticipated to be problematic.

Data are entered directly into eCRFs and housed at UTHealth servers located within the most secure area of the UTHealth network, with on-site and off-site backup and a geographically remote fail-over facility supporting robust disaster recovery processes. Details of transfused units are collected using barcode scanners reading International Society of Blood Transfusion (ISBT) 128-formatted barcode labels affixed to each unit.

Data collected from transfusion units for each patient are aggregated and validated on-site, then securely uploaded to a document management system within the electronic data collection system provided by UTHealth. These data are validated a second time and imported by an automated process into an eCRF supporting post-randomization transfusion data collection. Manual eCRF entry of transfusion unit data is available as a backup process.

Data collected during the course of the research will be kept strictly confidential and only accessed by members of the trial team. Participants are allocated an individual study number upon randomization. Participants details are stored on a password-protected database and only accessible to the study team.

Participant’s data is fully anonymised for analysis and reporting. ### Plans for collection, laboratory evaluation, and storage of biological specimens for genetic or molecular analysis in this trial/future use {33} See above “Additional consent provisions for collection and use of participant data and biological specimens {26b}”. There will be no biological specimens collected.

However, 400 patients will be enrolled in an ancillary study funded by the Department of Defense, called bioTROOP [61]. ## [](https://pmc. ncbi.

nlm. nih. gov/articles/PMC12318426/)Statistical methods ### Statistical methods for primary and secondary outcomes {20a} The primary hypothesis of the trial is that the use of LTOWB for in-hospital trauma resuscitation is as effective as CT (no worse than a 2% absolute increase in 6-h mortality).

If non-inferiority is determined, we will present the probability that the use of LTOWB is more effective than CT (in terms of 6-h mortality).

The primary outcome of 6-h mortality (and all binary outcomes including 24-h mortality, in-hospital/30-day mortality, and in-hospital/30-day complications) will be analyzed with the same logistic regression model including intervention group as a covariate, and site as a random intercept:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}${\text{logit(}p}_{ij})={\beta }_{0}+{\beta }_{1} {\text{intervention}}_{i}+{\gamma }_{j}$\end{document} \left(\text{logit}( p\right)_{\mathit{ij}} \left.

\right) = \beta_{0} + \beta_{1} \text{intervention}_{i} + \gamma_{j} Where the index $\mathit{ij}$ indexes the patient $i$ in site $j$, $\gamma_{j}$ is the effect of site $j$, which follows a $\text{Normal}( 0 , \sigma^{2} \left. \right)$ distribution. $\beta_{0}$ is an intercept, and $\beta_{1}$ is the log odds ratio for LTOWB vs CT.

LTOWB is coded = 1 and CT is coded = 0. Posterior distributions of risk differences (RDs) will be derived from Markov chain Monte Carlo posterior draws of the model parameters, as is standard under Bayesian analysis.

For all binary outcomes, $\beta_{0}$ and $\beta_{1}$ are converted to a risk difference (difference in proportions) by calculating \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}${\text{RD}}=\frac{\text{exp}\left\{{\beta }_{0}+{\beta }_{1}+\gamma \right\}}{1+\text{exp}\left\{{\beta }_{0}+{\beta }_{1}+\gamma \right\}}-\frac{\text{exp}\left\{{\beta }_{0}+\gamma \right\}}{1+\text{exp}\left\{{\beta }_{0}+\gamma \right\}}$\end{document} \text{RD} = \frac{\text{exp} \left{\beta_{0},+,\beta_{1},+,\gamma\right}}{1 + \text{exp} \left{\beta_{0},+,\beta_{1},+,\gamma\right}} - \frac{\text{exp} \left{\beta_{0},+,\gamma\right}}{1 + \text{exp} \left{\beta_{0},+,\gamma\right}} where $\gamma$ is shorthand for the averaged random site effects (which is 0 by the model specification).

The posterior median estimate of RD will be presented with 95% credible interval (CrI). For the primary outcome, non-inferiority will be declared if the probability that the RD is less than 2% is > 97. 5%.

That is, there is a 97. 5% probability that 6-h mortality rate in LTOWB is no higher than 2% the rate in the component therapy group: Prob(RD $< 0. 02$) > 97.

5%. If non-inferiority is declared, the Bayesian posterior probability of superiority will be reported. This is denoted as a “sequential evaluation.

” Statistical analyses will be conducted in R. Bayesian models will be fit using Markov Chain Monte Carlo (MCMC) methods implemented in Stan via R packages “rstanarm” and “brms. ” Each model will be fit using three MCMC chains.

Each chain will have at least 5000 burn-in samples and at least 50,000 additional iterations. We will use a neutral prior distribution for the treatment effect centered at 0 with a 95% prior credible interval (CrI) of 0. 25–4.

0 in the OR scale corresponding to Normal(0, 0. 70 2) in the log OR scale. This prior is based on the largest effect sizes identified for major outcomes in randomized trials [62].

For the intercept term, we will use a $\text{Normal}( 0 , 10^{2} \left. \right)$ prior and a weakly informative Normal(0, 1) prior (excludes OR > 7 or < 0. 14) for any other covariates included in the model in secondary analyses.

A half-Normal(0, 0. 5) prior will be used for the standard deviation $( \sigma \left. \right)$ of the random site effect, which is a weakly informative prior that has been shown to perform well in the literature [63].

The primary outcome will also be analyzed under an enthusiastic and skeptical prior, to serve as sensitivity analysis and to represent a range of a priori views. The enthusiastic prior is based on expert opinion, derived from a formal elicitation exercise [58], and is centered at OR of 0. 77, 95% CrI 0.

17–3. 5; Normal(− 0. 27, 0.

78 2) in the log OR scale. The skeptical prior is based on the non-inferiority margin and is centered at OR of 1. 20, 95% CrI 0.

3–4. 7; $\text{Normal}( 0. 18 , 0.

70^{2} \left. \right)$ in the log OR scale. For the secondary mortality outcomes, the analysis will proceed as described for the primary outcome.

The overall rate of complications in this population is known to be high (89%) and we hypothesize that LTOWB will not increase this rate by more than 3% for which we will evaluate non-inferiority. ### Interim analyses {21b} Two interim analyses are planned to monitor safety of the patients in the study.

The first interim analysis will be conducted after 370 (one-third of total enrollment) and the second after 740 (two-thirds of total enrollment) patients are randomized equally to LTOWB or CT. At each interim analysis, we will calculate the posterior probability of treatment effect: $P_{harm}$(at current):current posterior probability of higher 6-h mortality in the LTOWB group compared to component therapy group.

$P_{ben}$(at current N): current posterior probability of lower 6-h mortality in the LTOWB group compared to component therapy group. At each interim, the following actions may result: Stop the trial for harm if $P_{harm}$> 90%. Stop the trial for benefit if$P_{Ben}$> 99.

5%. Continue accrual to the next interim or final analysis. Thus, the trial will stop for harm if there is convincing evidence of higher 6-h mortality with LTOWB.

The trial can stop for benefit if there is overwhelming evidence of benefit of LTOWB. Follow-up of currently enrolled patients will continue to the final planned time point. Final analyses will include outcomes from all enrolled patients.

No interim futility analyses will be conducted since they are not appropriate for evaluating therapies already in use as is the case for LTOWB and CT [64]. ### Methods for additional analysis (e.g., subgroup analyses) {20b} We will conduct pre-planned subgroup analyses to evaluate heterogeneity of treatment effect on 6-h mortality.

Subgroup analyses will include main effects and interaction term between intervention group and the following baseline variables: sex (male vs female), leukoreduction vs no leukoreduction, age of blood products, prehospital blood products (LTOWB vs CT vs none), injury mechanism (blunt vs penetrating), impact of transport time, presence of severe traumatic brain injury (head AIS ≥ 3) vs no/mild traumatic brain injury, mode of transport (air vs ground), prehospital advanced airway management vs no pre-hospital advanced airway management, impact of prehospital fluid administration, comorbidity measured using the Charlson Comorbidity Index, pre-injury oral anticoagulant use, pre-injury antiplatelet use, shock severity measured using base deficit, pre-existing need for dialysis, injury severity, and age of patient.

For the binary moderators, an indicator for yes/no will be coded as 1/0, and the interaction with treatment will be included in the model. Variables with multiple levels (e.g., quantiles) will be coded as 3 indicator terms, and the resulting 3 interaction terms with treatment will be included in the model.

The conservative Bayesian approach of Dixon and Simon will be applied as it allows for a priori specification of how likely (or unlikely) it is for subgroup differences to be present, and to shrink the subgroup estimates to the overall mean treatment effect [65]. For subgroup analyses, the prior distributions for main effects will be the same as for the primary endpoint analysis. A Normal (mean = 0, var = 0.

7 2) will be used for interaction terms. Point estimates of posterior median treatment effect and 95% credible intervals for each subgroup will be reported along with probability of non-inferiority using a forest plot. ### Methods