Introduction

Calcium ions play a vital role in various physiological processes relevant to the shocked trauma patient. It is an essential electrolyte responsible for cardiovascular function through enabling cardiomyocyte contraction upon its release from the sarcoplasmic reticulum [1, 2] and by maintenance of vascular tone [3]. Hypocalcaemia is shown to cause acute cardiovascular compromise through the development of dysrhythmias [4], hypotension, as well as decreased cardiac output and contractility [1, 5, 6]. Calcium is also essential in haemostasis and coagulation. The calcium ion acts as a positively charged bridge between negatively charged phospholipids and vitamin K-dependent clotting factors II, VII, IX, and X [7]. Calcium also acts as a secondary messenger in platelet function and has been significantly associated with platelet activation, aggregation, and viscoelastic clot strength [8]. Hypocalcaemia upon arrival (HUA) in trauma patients is associated with higher overall mortality [9,10,11,12,13,14,15,16,17,18] and blood product transfusion rates [10, 14, 18, 19]. As a result, calcium has been identified as part of the lethal diamond of death alongside hypothermia, coagulopathy, and acidosis [20, 21]. However, there is no robust evidence of causation over correlation.

Intravenous administration of citrate causes transient hypocalcaemia, after which ionised calcium (iCa) has been shown to return to pre-transfusion levels from a few minutes to hours [22,23,24]. Citrate is a common anticoagulant that has been used in transfusion medicine since 1914 [25]. A unit of packed red blood cells contains 3 g of citrate which can be metabolised by a healthy adult in 5 min [26]. It is often observed that patients receiving citrated blood products are hypocalcaemic, especially in cohorts of patients undergoing massive transfusion in hospital. As a result, intravenous calcium supplementation has become routine in these patients based on the knowledge that citrated blood products can induce hypocalcaemia in recipient trauma patients.

HUA has also been found to be common amongst shocked trauma patients independent of prehospital blood product administration [27]. This mechanism should be considered in shocked trauma patients.

A better understanding of the mechanisms causing hypocalcaemia in prehospital trauma patients is important as prehospital transfusion is increasingly becoming the standard of care in the bleeding trauma patient in many advanced trauma systems. An Australian study reported the use of prehospital transfusion increased over two-fold from 2010 to 2018 [28]. Prehospital transfusion is also becoming more complex as services are increasingly implementing balanced transfusion practices [29] and early activation of mass transfusion protocols [30].

This is the first systematic review combining a meta-analysis that compares arrival iCa levels in trauma patients who did and did not receive prehospital blood products. The primary aim was comparison of arrival iCa levels in trauma patients who did and did not receive prehospital blood products. The hypothesis was that hypocalcaemia is common in severe trauma patients regardless of whether they received prehospital blood products or not. Secondary aims were to compare the arrival iCa of patients receiving different blood product types and explore associations with injury severity, injury type, mortality, and in-hospital transfusion requirement.

Methods

The protocol for this review was registered in the International Prospective Register of Systematic Reviews (PROSPERO CRD42022315189). This study’s methodology is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [31].

Search strategy

A literature search of the PUBMED, MEDLINE, EMBASE, CINAHL, International Clinical Trials Registry Platform, and Cochrane Library electronic databases was performed on 31 March 2023 containing the following search terms:

("trauma patient* OR "severe trauma*" OR "major trauma" OR "major trauma patient*" OR "adult trauma patient*" OR "traumatic injur*" OR "Wounds AND Injuries"[mesh:noexp]OR "hemorrhagic shock" OR "haemorrhagic shock" OR "hypovolemic shock" OR "major bleed*" OR Shock[mesh:noexp] OR Shock, Hemorrhagic[mesh:noexp] OR Shock, Traumatic[mesh:noexp) AND (hypocalcemia OR hypocalcaemia OR calcium OR "ionised calcium" OR "ionized calcium" OR Hypocalcemia[mesh:noexp] OR Calcium[mesh:noexp]ae, bl, df).

Manual searches of reference lists were also performed and Google Scholar was searched for citations from non-indexed manuscripts. No further limitations to the search strategy were applied due to the scarcity of prehospital literature.

Inclusion and exclusion criteria

Study types that were included consisted of randomised and non-randomised controlled trials, cohort, cross-sectional, and retrospective studies. Small case series, case reports, reviews, abstracts, ecological studies, and animal studies were excluded. There was no limitation on publication period and studies in languages other than English were included if accompanied by an English translation.

Inclusion criteria for study participants were as follows: trauma patients with a reported mean or median Injury Severity Score (ISS) > / = 15 who had an iCa measured on arrival to hospital. Exclusion criteria for study participants included patients who were not being treated for severe trauma and those who did not have an iCa measured on arrival to hospital. Studies were grouped into those consisting of patients who received prehospital product, those who received no prehospital product, and comparative studies wherein some patients received prehospital product and some did not.

Two reviewers (TR and AB) independently performed title and abstract screening and a third reviewer (BB) was consulted to resolve any discrepancies. This process was performed with the aid of Rayyan [32], a systematic review software program that utilises artificial intelligence to assist in filtering and sorting references.

Data item and data extraction

Two authors (TR and AB) performed data collection from the full-text articles with the aid of a standardised form. The following data were extracted:

  • General study information—publication year, location, study type, the total number of participants

  • Participant characteristics—ISS score and injury mechanism (blunt or penetrating)

  • Clinical information—prehospital time, on-scene and ED vital signs, shock index, GCS, mortality, prehospital and in-hospital transfusion requirement, and type

  • Laboratory information—lactate, INR, base deficit, pH, and arrival iCa.

Arrival iCa was defined as the initial iCa tested upon arrival to hospital via arterial or venous blood gas. Severe trauma was defined as a reported mean or median ISS > / = 15. Where required, study authors were contacted for clarification or further data provision.

Quality assessment

Risk of bias assessment was manually performed by 2 independent authors (TR and AB). The ROB-2 [33] and Newcastle–Ottawa Scale (NOS) tools [34] were used for assessing randomised and non-randomised trials respectively. The ROB-2 tool grades studies as being either “high risk” of bias, “low risk” of bias, or having “some concerns.” The NOS grades studies from 0 to 9 stars based on a studies selection, comparability, and outcome. Studies that scored > / = 7 were deemed to be “good” quality, 2–6 “fair” quality, and < 2 “poor” quality.

Statistical analysis

The primary outcome of interest was iCa level on arrival to hospital. Secondary outcome measures were too heterogenous and incomplete to allow robust statistical analysis with the exception of deriving a mean ISS amongst the cohort of studies included. Analysis for the primary outcome was performed for all patients, as well as the subset of patients who did not receive prehospital blood transfusion. Comparative analysis of prehospital blood transfusion and no transfusion was also performed if studies reported both groups. iCa reported as median or a range were converted to mean and standard deviation (SD) to facilitate meta-analysis using the methods of Luo et al. [35] and Wan et al. [36].

iCa concentration was pooled using the Hartung-Knapp-Sidik-Jonkman random-effects meta-analysis of proportions for both overall aggregation and between-group comparisons [37]. A Bayesian meta-analysis was further performed to explore the robustness of the results of the between-group comparison, noting the likely scarcity of studies. The Bayesian meta-analysis was performed using vague priors (zero for mu and half-normal with scale 0.5 for tau). Results are presented as mean difference (MD) to provide clinically interpretable values. 95% confidence intervals (CIs) and 95% credible intervals (CrI) are also presented. P values less than 0.05 were considered to be significant.

Quantitative heterogeneity was assessed using I2. If I2 > 50%, suggesting significant heterogeneity, then study-level characteristics (study size, median year of recruitment, continent of practice) were explored with meta-regression to identify contributions to heterogeneity.

All statistical analysis was performed with R (version 4.1.0, R Foundation for Statistical Computing, Vienna, Austria), and packages meta (version 6.2–1) and bayesmeta (version 3.2).

Results

Description of systematic review search

Out of 2435 unique citations 34 full-text articles were reviewed by two authors (TR and BB) resulting in 14 being included in the systematic review [8,9,10,11, 14,15,16, 19, 38,39,40,41,42,43], three of which were included in the primary meta-analysis [15, 39, 40]. The remaining 20 studies were excluded due to the timing of iCa measurement, absence of iCa measurement, incorrect population, or absence of a full text.

Seven authors were contacted [8, 11, 17, 39, 44,45,46] to request further information regarding inclusion criteria such as iCa measurement and injury severity. Four authors replied, resulting in the inclusion of 3 studies [8, 11, 39] and 1 study being excluded [46]. One study of hypocalcaemia in a military setting [47] initially appeared to meet criteria, however, was excluded due to unclear injury severity due to a lack of ISS reporting. The search and selection process has been outlined by a PRISMA flow diagram (Fig. 1).

Fig. 1
figure 1

PRISMA diagram of search and inclusion processes

Full size image

Description of the included studies

Of the 14 studies included in the systematic review, 2 of these studies consisted of a transfusion group only [38, 42], 7 consisted of a non-transfusion group only [8,9,10,11, 14, 16, 43], and 5 were comparative studies with a transfusion and non-transfusion group [15, 19, 39,40,41], 3 of which presented sufficient data to permit statistical analysis [15, 39, 40]. General study characteristics and risk of bias assessment are presented in Table 1. Patient characteristics such as injury severity, mechanism of injury, as well as laboratory parameters are presented in Table 2.

Table 1 Summary of studies reporting iCa in severe trauma patients who did and did not receive prehospital blood products
Full size table
Table 2 Summary of patient characteristics and laboratory parameters in studies reporting iCa in severe trauma patients who did and did not receive prehospital blood products
Full size table

Meta-analysis results

Nine studies reported iCa on arrival to ED (Fig. 2), with a mean of 1.08 mmol/L (95% CI 1.02–1.13; I2 = 99%; 2087 patients). Subgroup analysis of patients who did not receive prehospital blood transfusion had a mean iCa of 1.07 mmol/L (95% CI 1.01–1.14; I2 = 99%, 1661 patients).

Fig. 2
figure 2

Forest plot of iCa concentration (mmol/L) on arrival to ED, grouped by whether prehospital blood was administered or not. Multiple exclusive subgroups may be present in the same study

Full size image

Patients who received prehospital blood transfusion in 3 comparative studies (Fig. 3) had a slightly lower iCa on arrival compared to those who did not receive transfusion (mean difference − 0.03 mmoL/L, 95% CI − 0.04 to − 0.03, I2 = 0%, p = 0.001, 561 patients). This is consistent with the result of the Bayesian meta-analysis using vague priors, with posterior probability of the MD of − 0.03 (95% CI, − 0.23 to 0.18). The posterior probability that prehospital transfusion reduced iCa on arrival to ED was 76.8%.

Fig. 3
figure 3

Forest plot of iCa (mmol/L) concentration on arrival to ED, comparing those who received prehospital blood transfusion versus those who did not. SD, standard difference

Full size image

Eleven studies reported ISS [9,10,11, 15, 16, 19, 38,39,40,41,42] with an aggregated mean of 27 (95% CI 26.5–27.5; I2 = 98%; 2698 patients). There was significant heterogeneity, variation, and overall, not enough granularity to compare ISS for transfused vs non-transfused groups. We were not able to regress ISS against iCa outcomes either without patient-level raw data.

Discussion

To our knowledge, this is the first systematic review and meta-analysis that compared the arrival iCa of trauma patients who received prehospital blood products with those who did not. Overall, the results highlight that hypocalcaemia is a common metabolic disturbance in the trauma patient, outside of blood transfusion. Quantitative analysis of the available comparative studies revealed a statistically significant difference (mean difference 0.03 mmol/L) (Fig. 3) between the arrival iCa of patients who did and did not receive prehospital blood products. Patients in both the blood transfusion and non-transfusion groups were hypocalcaemic upon arrival at hospital by nearly all published definitions; suggesting that whilst blood products do play a role in the pathogenesis of hypocalcaemia in this population, it is not the sole cause. This is in keeping with a recent systematic review that looked at shocked trauma patients who did not receive prehospital blood products [27]. These findings suggest that we have possibly overemphasised the concept of citrate-induced hypocalcaemia alone, especially in the context of the prehospital setting where transfusion volume is limited and hence citrate load is also limited prior to hospital arrival. The literature base has likely overlooked the concept of shock-induced hypocalcaemia, irrespective of transfusion.

Whilst the exact mechanism of hypocalcaemia in trauma is yet to be elucidated, various sources suggest potential mechanisms other than citrate chelation, such as dilution [43, 48, 49], colloid binding [50], lactate binding [21], acidosis [9, 51], ischaemic reperfusion [52,53,54,55], and impaired parathyroid hormone secretion or action [56,57,58,59]. In fact, proinflammatory cytokines such as IL-6 have been found to suppress parathyroid hormone at clinically relevant concentrations [16, 56]. Notably, IL-6 levels have been found to be significantly increased in relation to the severity of trauma [60,61,62,63,64,65,66]. This is of particular interest as inflammatory cytokines in the form of DAMPs (damage-associated molecular proteins) have been suggested as a potential target for immunomodulatory therapy to reduce mortality and morbidity in acute trauma and haemorrhagic shock [67, 68].

Administration of calcium in severe trauma is viewed by some as a double-edged sword. Phosphatidylserine (PtdSER), a negatively charged phospholipid plays an important role in coagulation. It is transported to the outer leaflet of the plasma membrane in platelets where it acts as a base for positively charged calcium ions to form a bridge between PtdSER and the negatively charged dicarboxyl glutamic acid residues in the vitamin K-dependent coagulation factors, thus enhancing the activation of prothrombin to thrombin [69,70,71]. However, calcium ions also play a role in cellular apoptosis following injury. In response to calcium-dependent stimuli, PtdSER is known to have an important role in the regulation of apoptosis. Calcium triggers the exposure of PtdSER on the outer cell membrane, which then acts as an “eat me” signal and causes apoptosis of cells [72].

In a recently published observational paper [73], the authors concluded that trauma-induced disturbances in ionized calcium levels correlate parabolically with coagulopathy, transfusion, and mortality. They concluded that “iCa2 + levels change dynamically and are more a reflection of severity of injury and accompanying physiological derangements, rather than an individual parameter that needs to be corrected as such” [73]. It is possible that hypocalcaemia could be a normal physiologic stress response to support coagulation or to limit apoptosis following injury and flooding the system with too much calcium could make things worse.

Another important finding from this review is a lack of consistency in the existing literature concerning the definition of hypocalcaemia in trauma. Various sources define hypocalcaemia from < 1 mmol/L [14,15,16] to < 1.2 mmol/L [47, 74] and anything in between (Table 1). Many multijurisdictional hospital MTPs use iCa targets of > 1.1 mmol/L. Consensus is needed to aid future investigations. Based on the findings of this SRMA, we recommend a standardised definition of hypocalcaemia to aid further collaborative research. Given most MTPs target iCa > 1.1 mmol/L, it would seem logical to use that as the definition of hypocalcaemia in trauma.

We also recommend that trauma centres routinely collect and record arrival iCa levels in their trauma registries to allow future collaborative research to progress. In many trauma systems/registries, this is not a routinely collected variable. Lastly, as it is clear that hypocalcaemia is an important factor in the trauma patient, we suggest that randomised controlled trials comparing the administration of calcium, trends of calcium levels, and its effect on mortality, morbidity, and transfusion demands in severe trauma are needed. Options could include (1) a prehospital RCT investigating single-dose IV calcium versus placebo in shocked trauma patients and (2) an RCT comparing two different calcium doses, in particular where up front/empiric transfusion is performed.

Limitations

There are several limitations which make it difficult to make more robust conclusive statements, and which highlight the need for further high-quality prospective studies. As expected, there was a paucity of high-quality prehospital transfusion-related literature sources. Many were composed of small sample sizes and had a moderate-to-high risk of bias. This systematic review highlighted 5 comparative studies; however, only 3 had the required data for meta-analysis (Fig. 3). Whilst our analysis did highlight a statistically significant difference in arrival iCa between the transfusion and non-transfusion groups, the results were heavily weighted (90.3%) towards a single study [15]; a post hoc analysis of 2 prehospital plasma randomised controlled trials [75, 76] which if removed, yielded a non-significant result. Their study also utilised back calculation as a method to calculate standard deviation, an accepted yet imprecise methodology. That said, the findings of prehospital transfusion and shock-induced hypocalcaemia are undoubtedly present.

There was also significant heterogeneity amongst the studies in the reporting and definition of iCa on arrival to hospital. Blood product volume, and type, and hence citrate load, varied within and between studies due to the different policies and practices of prehospital services as well as the fact that most studies were retrospective in nature. This is common in trauma research. Interestingly, one study adjusted for citrate load by comparing the arrival iCa to the amount of citrate transfused and found no association [39]. There was also a lack of information on the prehospital phase in the studies, notably timings which intuitively should be reported from point of injury with respect to iCa levels over time.

Conclusion

Prehospital blood transfusion and injury severity are associated with hypocalcaemia on arrival to hospital, the hypocalcaemia effect more pronounced in the former group. There is a need for an internationally accepted definition of hypocalcaemia in trauma to allow further research and benchmarking. There is also a need for routine collection of iCa levels on hospital arrival in trauma registries and correlation with other variables from point of injury such as prehospital vital signs, injury severity, prehospital times, time to haemostatic resuscitation, and haemorrhage control. This systematic review and meta-analysis provide reasonable argument and equipoise for future RCTs in calcium administration in severe trauma patients.