Modelling the outcomes of different red blood cell transfusion strategies for the treatment of traumatic haemorrhage in the prehospital setting in the United Kingdom

Abstract Background and Objectives The limited supply and increasing demand of group O RhD‐negative red blood cells (RBCs) have resulted in other transfusion strategies being explored by blood services that carry potential risks but may still provide an overall benefit to patients. Our aim was to analyse the potential economic benefits of prehospital transfusion (PHT) against no PHT. Materials and Methods The impact of three PHT strategies (RhD‐negative RBC, RhD‐positive RBC and no transfusion) on quality‐adjusted‐life‐years (QALYs) of all United Kingdom trauma patients in a given year and the subset of patients considered most at risk (RhD‐negative females <50 years old), was modelled. Results For the entire cohort and the subset of patients, transfusing RhD‐negative RBCs generated the most QALYs (141,899 and 2977, respectively), followed by the RhD‐positive RBCs (141,879.8 and 2958.8 respectively), and no prehospital RBCs (119,285 and 2503 respectively). The QALY difference between RhD‐negative and RhD‐positive policies was smaller (19.2, both cohorts) than RhD‐positive and no RBCs policies in QALYs term (22,600 all cohort, 470 for a subset), indicating that harms from transfusing RhD‐positive RBCs are lower than harms associated with no RBC transfusion. A survival increase from PHT of 0.02% (entire cohort) and 0.7% (subset cohort) would still make the RhD‐positive strategy better in QALYs terms than no PHT. Conclusion While the use of RhD‐positive RBCs carries risks, the benefits measured in QALYs are higher than if no PHT are administered, even for women of childbearing potential. Group O RhD‐positive RBCs could be considered when there is a national shortage of RhD‐negative RBCs.


INTRODUCTION
Several observational studies have demonstrated survival benefits following the administration of red blood cells (RBCs) or low titre group O whole blood (LTOWB, herein referred to as RBCs) compared to saline alone in the resuscitation of traumatic haemorrhagic patients in the prehospital environment [1][2][3]. Randomized controlled trials (RCTs) that have compared prehospital blood transfusion to crystalloids have shown different outcomes due to differences in the patients' underlying demographics and clinical environments in which the studies were conducted. An RCT in which the standard of prehospital care was supplemented with two units of plasma for trauma patients transported to the hospital by helicopter showed that plasma transfusion significantly improved 30-day survival by approximately 10% [4], with the secondary analysis showing that the combination of RBC and plasma produced the highest survival rates compared to not providing any blood products while en route to the hospital [5]. A recent RCT (RePHILL trial) showed that RBC transfusion plus lyophilized plasma in the prehospital setting was not superior to saline resuscitation for improving tissue perfusion or reducing episode mortality; however, at 24 h, the adjusted average differences in mortality were 7% lower in the blood components arm (personal communication from trial investigators) [6].
In the prehospital setting, group O RhD-negative RBC is transfused to avoid harms caused by the RhD-positive blood being transfused to RhD-negative recipients because the recipient's ABO/RhD type is known. These harms include severe haemolysis and possibly death [7,8], a haemolytic transfusion reaction (HTR) following the transfusion of an RhD-positive RBC unit to a recipient with preformed anti-D [9,10], and for the RhD-negative female of childbearing potential (FCPs) who has become D-alloimmunized through RhD-positive blood, this antibody can cause haemolytic disease of the foetus and newborn (HDFN) should she later become pregnant.
However, the ability to provide RhD-negative RBCs for prehospital transfusion (PHT) is constrained by donor supply. The current demand for group O RhD-negative RBCs in England is 13% of the supply compared to the 7% frequency of group O RhD-negative in the general population. An international study of blood centre collectors found that only 10% of their RBC distributions to hospitals were group O-negative [11]. Therefore, the supply of these precious products is very limited. Several studies have modelled the overall clinical risk of foetal/neonatal outcomes following the transfusion of RhDpositive RBCs to injured RhD-negative recipients and FCPs [12][13][14][15].
However, evaluation of the risk-benefit ratio of providing group O RhD-positive RBCs to trauma patients in terms of quality-adjustedlife-years (QALYs) gained has not been evaluated before.
In this study, we compare the relative advantages of different policies for providing PHT, whereby the effect on the recipient's QALY was modelled in the following three scenarios: using RhD-negative RBCs for all PHTs, using RhD-positive RBCs for all PHTs, or not administering PHTs to injured patients.

METHODS
We modelled the impact of three scenarios on two groups of trauma patients: (i) a representative sample of all United Kingdom trauma patients in a given year, and (ii) the subset of patients who were RhDnegative females of childbearing potential <50 years old (referred as FCP), by simulating the harms experienced by patients over the remainder of their lifetimes. In the case of HDFN occurrence, we also modelled its impact on the affected babies.
A cohort was defined as a representative sample of patients who suffered a trauma in the United Kingdom in 1 year (5561 patients), as calculated previously [16]. For post-trauma life expectancy, we used the Office for National Statistics data and set this to 48 years after the accident for the entire cohort and 56 years after the accident for the subset cohort of FCPs due to their lower age at the time of injury, accounting for the average age of trauma patient and the impact of the trauma on life expectancy [17].
We modelled the reductions in health-related quality of life (HRQoL) following the trauma using a health economics approach to capture the QALYs associated with each transfusion scenario.

Modelling impact on patients suffering trauma
For each three resuscitation scenarios (prehospital RhD-positive RBC transfusion, or RhD-negative RBC transfusion, or no PHT) and the two patient groups (overall cohort and FCPs), we 'allocated' patients to six health states in each year following transfusion ( Figure 1):   [1,2] demonstrating survival benefits following PHT of injured patients. The probability of suffering transfusion harms was based on a previously published model [15]. Key inputs used for the model are provided in Table A1 in Supporting information. The RePHILL trial performed in England showed that prehospital RBC transfusion plus LyoPlas was not superior to saline resuscitation in reducing the composite outcome of episode mortality and lactate clearance in trauma bleeding patients [6]. However, the study's design and findings have been criticized, including the selection of the mortality time point in the primary outcome, the length of time elapsed from injury until the administration of the study intervention, the highly injured nature of the patients and their correspondingly high death rate [21,22]. There was, however, a 7% absolute risk reduction (25% relative risk reduction) among the PHT recipients at 3 h compared to those who were resuscitated with saline. Thus, the sensitivity analysis was designed to include this risk reduction.
Where a patient was predicted to have a future child with severe HDFN morbidity due to RhD-positive RBCs during the resuscitation, we modelled a QALY impact to both the patient and child, where the impact to the patient is based on the impact to HRQoL observed for mothers of children with Cerebral Palsy, as there were no other published data on the impact of HRQoL for mothers of children with HDFN [23][24][25]. For patients that experience foetal death due to HDFN caused by anti-D that was formed following receipt of RhD-positive RBCs during trauma resuscitation, we only modelled a QALY impact on the child based on studies that showed that there was no HRQoL impact to these mothers [26]. In our sensitivity scenarios we considered the impact on the mothers of foetal death.

Modelling impact on babies born to patients who suffered trauma
When RhD-positive RBCs were transfused, we calculated the potential impact on babies born with HDFN or on those who died because of HDFN. Whereas the specific harms to patients were expressed as     dicted to occur in this group, as they were not exposed to the RhDantigen through PHT. Using the valuation approach outlined in the Green Book [27], a publication from the United Kingdom Treasury that provides guidance on how to appraise policies, programmes and projects, the resulting loss of QALYs from not providing RBCs for prehospital resuscitation would be valued at just over £1billion compared to providing RBCs of any RhD type over the lifetime of these patients (Table 1).
Of the 5561 patients in England who received a PHT, 100 would have been FCPs who were at risk of a future pregnancy affected by HDFN [15]. The provision of RhD-positive RBCs would be expected to lead to very few patients in this cohort experiencing any of the

Sensitivity analyses
Sensitivity analyses, given a degree of uncertainty relating to some of the key model inputs, were performed by adjusting some of these parameters (Table 3)

DISCUSSION
Previous models of transfusing RhD-positive RBCs to injured trauma patients of unknown RhD-type have presented the risks of adverse events as probabilities of their occurrence [12][13][14]. This study is unique because it quantified the relative benefit of PHT in terms of £400. This is substantially lower than the value of the QALY benefit.
PHTs provide 2.93 discounted QALYs per patient, which at the cost of £400, is the cost per QALY of £137 and is substantially below the NICE guide of £20,000-30,000 per QALY.
Comparing the discounted monetized QALYs per patient between recipients of RhD-negative RBCs and RhD-positive RBCs, the use of RhD-negative blood provides about £125 more value per patient than RhD-positive blood. However, the equivalent figure for RhD-negative FCPs is £6369, indicating that a policy of providing RhD-negative blood for FCPs provides benefits and should be pursued where possible. Hospitals in England currently pay the same amount for RhD-negative and RhD-positive group O blood. However, RhD-negative blood is in higher demand, and blood services spend additional resources to recruit and retain more RhD-negative donors. The totality of these costs is not well described in the literature, but it seems unlikely to exceed the value of providing RhD-negative blood to FCPs. Furthermore, it is less clear what the totality of these costs will be for the whole patient group and whether recruitment costs will outweigh patient benefits, and further research is required to understand these.
This study has several limitations. First, since the model was built using data from the literature, the extent to which these findings can be generalizable depends on the generalizability of the original data.
Secondly, it would have been good to have provided a projection on the effect of blood transfusion on QALYs based on the patient's injury severity; however, all studies/trials have pooled the transfusion effect for all types and injury severity, and we were not able to quantify this in QALY terms. Thirdly, this study was designed to describe the effect of PHT on the 24-h mortality rate, as this time-point is the most likely to be modifiable by PHT, as beyond the 24 h, the cause of death in trauma patients is typically factors other than exsanguination. Indeed, the American National Heart, Lung, and Blood Institute and Department of Defence have recently supported the development of shortterm outcomes (3-6 h mortality) for adult trauma patients based on recent RCTs [29]. In the United Kingdom, the RCT of whole blood versus standard of care in the prehospital setting, will use the composite of mortality or massive transfusion at 24 h as its primary outcome.
Therefore, the use of 24 h mortality in this model took all these development into consideration.
In conclusion, this study quantified the change in QALYs based on two PHT strategies compared to not providing PHTs to injured patients. While a small number of patients were predicted to expe-