Pre‐hospital transfusion of red blood cells. Part 2: A systematic review of treatment effects on outcomes

Summary The primary aim of this systematic review is to describe the effects of prehospital transfusion of red blood cells (PHTRBC) on patient outcomes. Damage control resuscitation attempts to prevent death through haemorrhage in trauma patients. In this context, transfusion of red blood cells is increasingly used by emergency medical services (EMS). However, evidence on the effects on outcomes is scarce. PubMed and Web of Science were searched through January 2019; 55 articles were included. No randomised controlled studies were identified. While several observational studies suggest an increased survival after PHTRBC, consistent evidence for beneficial effects of PHTRBC on survival was not found. PHTRBC appears to improve haemodynamic parameters, but there is no evidence that shock on arrival to hospital is averted, nor of an association with trauma induced coagulopathy or with length of stay in hospitals or intensive care units. In conclusion, PHTRBC is increasingly used by EMS, but there is no strong evidence for effects of PHTRBC on mortality. Further research with study designs that allow causal inferences is required for more conclusive evidence. The combination of PHTRBC with plasma, as well as the use of individualised transfusion criteria, may potentially show more benefits and should be thoroughly investigated in the future. The review was registered at Prospero (CRD42018084658).


| INTRODUCTION
Haemorrhage is a potentially preventable cause of death after major trauma. [1][2][3] Topical treatment is not always sufficient to control haemorrhage, since it is often non-compressible. [3][4][5] The lethal triad of acidosis, hypothermia and coagulopathy is related to haemorrhagic shock, especially when blood loss is treated with liberal crystalloid fluid replacement. 6 In damage control resuscitation (DCR), blood products are preferred over crystalloids as fluid replacement, while a degree of hypotension is accepted until haemorrhage control has been achieved. [6][7][8][9][10][11] Through transfusion of red blood cells (RBC), the infusion of large volumes of crystalloids may be avoided, as RBC provide a more effective volume expansion. Haemostasis and thrombosis are promoted 12 and oxygen carrying capacity restored. 13 In an effort to decrease mortality through haemorrhage after major trauma, prehospital transfusion of red blood cells (PHTRBC) is increasingly performed. Military medical teams have been transfusing blood products prior to arrival at a surgical unit for years. 14,15 This may partially explain survival differences between civilian casualties who require massive transfusion (60%) and military casualties (93%). 16 More recently, civilian emergency medical services (EMS) have also started to carry blood to the scene and transfuse their patients in the prehospital setting. 17,18 In part 1 of this series, we described the availability and frequency of PHTRBC around the world, and demonstrated that varying transfusion criteria are being used. 19 However, to date, little is known concerning the effects on patient outcomes. We have therefore conducted a systematic review with the aim to evaluate the effect of PHTRBC in patients treated by EMS on multiple outcomes including mortality, haemodynamic parameters, and the need for further in-hospital transfusions.

| METHODS
The review was registered at Prospero (website: https://www.crd.york. ac.uk/prospero, identification number: CRD42018084658). This systematic review was conducted in accordance with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. 20

| Information sources, search strategy and study selection
PubMed and Web of Science were searched through January 2019. The search strategy and the process of selection of studies were described previously. 19 For the purpose of this systematic review, only the manuscripts reporting outcome measures after PHTRBC (data on the hemodynamic state, coagulopathy, 24 hours RBC transfusion requirement, length of stay in hospital/intensive care unit [ICU], mortality, or occurrence of complications) were included. For a comprehensive overview of studies reporting outcomes after PHTRBC, controlled studies as well as observational studies were considered for this systematic review.

| Data extraction
A standardised data extraction sheet was developed, and after testing this on the first 20 articles, it was refined. The setting and type of transport the EMS used (civilian or military, scene or interfacility) and patient characteristics were extracted. Data regarding the effects of prehospital transfusion were collected, including haemodynamic data, coagulopathy, 24 hours RBC transfusion requirement, length of stay in hospital/ICU, mortality and occurrence of complications. Descriptions of problems that arose during PHTRBC are summarised in the text.
Bias was assessed using the Newcastle-Ottawa Scale. 21

| Data synthesis
A priori, we had planned a random-effects meta-analysis of the available evidence. However, no controlled studies were identified, and the observational studies carry a risk of residual confounding even if matching or regression-based techniques were used to control for confounding. Moreover, a high heterogeneity among the studies precluded meaningful pooling of data: Civilian and military studies are not comparable due to fundamental differences in patients, mechanisms of injury and medical facilities. Also within these two groups, differences between patient populations and patient selection, differences between healthcare systems and EMS systems (eg, logistics, staff, equipment, treatment options, response and transport times to definitive care), differences in the type of blood products used (RBC only or a combination of blood products), differences in transfusion criteria, as well as differences in outcome measures are too great to allow a meaningful combined analysis. A meta-analysis was therefore not performed.

| Selection of articles
The search in PubMed and Web of Science yielded 2172 hits after removal of duplicates. In our scoping review, 71 articles were included. 19 In total, 55 of these studies reported one or more outcome measure, and were included in this review ( Figure 1). Forty-one of these studies discussed PHTRBC in civilian medical services, 14 of which allowed comparisons of PHTRBC with a control population. Notably, four articles primarily dealt with a different study topic, but were included as they additionally provided valuable information regarding PHTRBC [22][23][24][25] (Table 1).

WHAT IS KNOWN ABOUT THE TOPIC
• Prehospital transfusion of red blood cells is increasingly used in the setting of damage control resuscitation, and transfusion criteria vary widely.
• Transfusion of red blood cells is common practice in hospitals to save exsanguinating patients. In this setting, red blood cells are often transfused along with other blood products such as plasma.

WHAT IS NEW
• Literature does not show consistent evidence for beneficial effects of PHTRBC on mortality, shock on arrival to hospital, trauma induced coagulopathy, length of stay in hospitals or intensive care units.

WHAT ARE THE FUTURE KEY QUESTIONS FOR FUTURE WORK ON THE TOPIC
• Can individualised transfusion criteria provide additional benefit?
• Can the combination of PHTRBC with other blood products, such as plasma, reduce mortality in severely bleeding patients in the prehospital setting?
We included 14 articles discussing PHTRBC in military medical services, of which four studies had a control population. Four articles reported prehospital transfusion as an additional topic, while primarily discussing another aspect of their study 70,71,74,76 (Table 1).
No randomised trials were identified; all studies were observational.
The potential overlap of patients in different manuscripts was substantial (Tables 2 and 3). The bias assessment of the studies is shown in Table 4.

| Civilian
Unadjusted data from one study suggested that less PHTRBC patients died compared with non-receivers, 37 while another unadjusted study found no difference. 32 In propensity score-matched trauma patients, Brown et al 28,29 found a significant advantage of PHTRBC on either 24 hours or

| Military
In military services, almost all studies included patients who possibly received other prehospital blood products besides RBC.
Two retrospective studies compared trauma PHT recipients to non-receivers, and found significantly lower mortality in PHT patients (either overall, 24 hours or 30-day mortality) (Table 5). 65,72 One of these studies subsequently focused on those patients who survived the first 24 hours; the beneficial effect on 30-day mortality was no longer present. 65 This concurs with a large retrospective study (with a partially overlapping study population), where the odds for "killed in action" (KIA) mortality (death before arrival at treatment facility) was 83% lower for casualties who needed and received prehospital transfusion, compared with patients who needed but did not receive a prehospital transfusion. 63 3.4 | Outcome-Shock after transfusion

| Civilian
Six observational studies compared vital parameters or POCT results before and after transfusion. 32 41 Lawton LD (2012) 58 Case report  30 Lyon R (2017) 49 London Rehn M (2018) 35 Rehn M (2019) 36 Weaver AE ( deficit and lactate levels on arrival to hospital, and used these to calculate the odds of shock. They found that in PHTRBC patients, these odds were significantly lower than in matched control patients (Table 5).

Israel
Barkana Y (1999) 68 Chen J (2017) 69 (Table 5).  32 and there was one possible adverse reaction in a trauma patient who developed shortness of breath, which was interpreted as secondary to volume overload. 42 A case report has described two "near miss" incidents where haemolysis of donor cells occurred during transport, when the units had not been packed correctly. 59 Patients transfused before arrival to hospital were more likely to be hypothermic 23 and have lower calcium levels, 49 but there was no significant difference in the occurrence of acute respiratory distress syndrome (ARDS)

| Outcome-Length of stay in ICU/in hospital
in PHTRBC, non-PHTRBC and PHTRBC+PHT-plasma patients. 22,29 As in civilian services, seven military studies have reported no adverse reactions to PHTRBC. 62,64,66,67,69,74,75 One possible transfusion reaction is described, in which a patient developed a fine rash on their trunk after one unit of RBC. 68 Seven incidents were reported where the blood products were found to have an out-of-standard temperature. 66

| DISCUSSION
This systematic review summarises the literature regarding the effects of PHTRBC on several outcome measures.
Overall, evidence of an effect of PHTRBC on outcomes is of limited quality. Notably, no controlled studies were identified, and all studies were observational. Therefore, all reported treatment effects must be interpreted with care. Confounding is likely-and residual confounding cannot be excluded in those studies that attempted to control for confounding-such that causal inferences on the effect of PHTRBC on outcomes are essentially not possible. 77 Nonetheless, in the absence of controlled trials, these studies represent the best available evidence and may at least provide some insight about possible associations between PHTRBC and outcomes.
The high heterogeneity of the studies was a second factor, which impedes the interpretation of the reported data. As we had expected, patients transported by civilian and military services differed considerably with respect to injury type, injury severity and mortality rates.
However, also within these groups, heterogeneity in injury type, injury severity, type of transport, transfusion criteria and type of intervention prevented meaningful meta-analysis. Differences between study outcomes might at least in part be explained by these factors.

| Mortality
Results on overall mortality are conflicting, and we found no consistent

| Adverse events
There have been few instances of transfusion reactions being reported after PHTRBC. Fortunately, transfusion reactions in the general population are rare, with urticaria occurring in 1% to 3% of patients, febrile non-haemolytic transfusion reaction and cardiac overload in <1%, and all other transfusion reactions in <0.1%. 83

| Strength and limitations
We performed a thorough search with broad inclusion criteria. therefore unlikely that we have overlooked data that would significantly alter our conclusions. Additionally, despite a different patient population, mechanism of injury and medical facilities, we did not exclude military services but decided to report on them separately, thereby including an important source of data on PHTRBC.
No controlled trials were identified, most observational studies were retrospective, and high heterogeneity also precluded meaningful data pooling. While this at first glance seems a limitation of this systematic review, it is rather a limitation of previous research and in fact an important outcome of this study; this finding highlights the lack of high-quality outcome studies, and suggests that randomised trials are needed for more conclusive evidence on the causal relationship between PHTRBC and patient outcomes.

| CONCLUSION
This systematic review revealed that despite increasing use of PHTRBC by civilian EMS, high-quality evidence for beneficial effects is still lacking. In the absence of high-quality data, it seems reasonable to assume that massively bleeding patients may benefit from PHTRBC. This assumption is supported by several observational studies that do suggest possible beneficial effects on mortality. This may especially be true when PHTRBC is combined with plasma administration. PHTRBC also appears to improve haemodynamic parameters, but there is no evidence that shock on arrival to hospital is averted,