Association of Non-Transfusion-Related Admission Hypocalcaemia With Haemodynamic Instability in Paediatric Major Trauma: A Retrospective Single-Centre Pilot Study

Background The ‘lethal triad’ of acidosis, hypothermia, and coagulopathy is now considered a diamond of death, with ionised hypocalcaemia (iHypoCa) contributing to cardiovascular decompensation and coagulopathy. iHypoCa may be associated with haemodynamic instability and adverse outcomes in paediatric major trauma patients. However, current data are limited. The primary aim of this pilot study was to report the association between admission iHypoCa and early hypotension on admission in a cohort of paediatric major trauma patients. Secondary aims include reporting the incidence and differential determinants of iHypoCa and the association with treatment (vasoactive agents, blood transfusion, interventional radiology (IR), or surgery) and adverse outcomes (length of stay, morbidity (Glasgow Outcome Scale), and mortality). Methods This pilot study is a retrospective analysis of paediatric major trauma patients (<16 years old) admitted to a major trauma centre (2016-2022). Patients with an admission ionised calcium level obtained before the administration of blood products were included. Multivariable logistic regression was used to assess the dichotomous endpoint of hypotension (systolic blood pressure of <80 mmHg for <1 year, <85 mmHg for one to five years, <90 mmHg for five to 12 years, <100 mmHg for >12 years) for association with hypocalcaemia and adjusted for other potential variables of interest (age, gender, Injury Severity Score, pre-hospital fluids, and acidosis). Results Admission iHypoCa was observed in 8/45 (17.8% (95% confidence interval (CI) 9.3-31.3%)) patients. Other than the adolescent age group (p < 0.05), there were no significant differences in the baseline characteristics. As a pilot study, this was not powered for statistical significance; however, point estimates of the odds of hypotension were almost three times higher for patients with iHypoCa (odds ratio (OR) 2.8 (95% CI 0.4-23.6), p = 0.33). An association between iHypoCa and the need for IR/surgery in the first 24 hours of admission was also observed (OR 10.9 (95% CI 1.4-159.4), p < 0.05). Conclusion iHypoCa was observed in approximately one in six paediatric major trauma patients at admission and may be associated with increased odds of requiring IR/surgery. Larger multicentre studies are required to clarify point estimates for treatment requirements and adverse outcomes.


Introduction
Major trauma is one of the leading causes of morbidity and mortality in children in the United Kingdom (UK) 1, 2 2, 3, 4 3, 5 1 1, 6 [1].Haemorrhage is the predominant cause of potentially survivable death in trauma, and a high proportion of these deaths occur pre-hospital [2].Consequently, the management of traumatic haemorrhage is a UK national research priority in emergency medicine [3].The 'lethal triad' of acidosis, hypothermia, and coagulopathy is now considered a 'diamond of death'; the fourth component, hypocalcaemia, is critical to trauma resuscitation [4,5].Calcium homeostasis is required for clotting, cardiac contractility, and vascular tone [4,5].Therefore, hypocalcaemia can contribute to cardiovascular decompensation and coagulopathy [4,5].
Ionised hypocalcaemia (iHypoCa) in the context of trauma may be secondary to the administration of citrated blood products, a process that is well understood [6,7].However, recent data suggest that iHypoCa may exist in patients with traumatic haemorrhage before receiving blood, and this may be further exacerbated by the citrate in the transfusion [5,8,9].Although trauma-induced iHypoCa is likely multifactorial, potential mechanisms include calcium-lactate binding, haemodilution, reduction in parathyroid hormone release, inappropriate renal calcium loss, and intracellular calcium influx in the setting of ischaemia and reperfusion [4,5].Up to 50% of adult major trauma patients have been observed to have iHypoCa, which is associated with mortality, coagulopathy, shock, and an increase in subsequent blood transfusion requirements [10][11][12][13][14]. Children may be more sensitive to iHypoCa compared to adults due to the different injury mechanisms, patterns of injury, and their maturing haemostatic system [15].However, paediatric data are limited and heterogeneous [16][17][18][19][20].
The primary aim of this pilot study is to report the association between admission iHypoCa and early hypotension on admission in a cohort of paediatric major trauma patients.Secondary aims include reporting the incidence and differential determinants of iHypoCa and the association with treatment requirements (vasoactive agents, blood product transfusion requirement, interventional radiology (IR), or surgery) and adverse outcomes (length of stay (LOS), morbidity (Glasgow Outcome Scale (GOS)), and mortality).

Study design
This study is a retrospective analysis of paediatric major trauma patients admitted to Cambridge University Hospitals NHS Foundation Trust (CUH), a major trauma centre (MTC) in the East of England between August 2016 and March 2022 (see Appendix A).Patients were included if they were <16 years old, had an Injury Severity Score (ISS) ≥ 15, Trauma Audit Research Network (TARN) positive (admission to hospital for three days or longer, intensive care, transfer for further specialist care, death), and had an ionised calcium (iCa) level taken on admission.Patients were excluded if they had a pre-hospital cardiac arrest, received a blood transfusion or exogenous calcium before the first iCa measurement, arrived at the MTC >24 hours after injury, or were treated at another hospital before transfer (secondary transfers or repatriations).The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline.

Study procedures
Paediatric major trauma patients were identified from TARN data obtained from the local site Trauma Office records with matched patient data obtained from the electronic medical record (EMR) at the local site (Epic Hyperspace Production®, Epic Systems Corporation, Verona, WI, USA).
Demographics, mechanism of injury, injury time, ISS, need for surgical management, 24-hour and 30-day mortality, functional outcome (GOS score), and hospital and paediatric intensive care unit (PICU) LOS data were obtained from the local Trauma Office records.The presence of hypotension (systolic blood pressure of <80 mmHg for <one year, <85 mmHg for one to five years, <90 mmHg for five to 12 years, <100 mmHg for >12 years; defined dichotomously based upon standard reference values) [21], and details of pre-hospital and hospital treatments (intravenous fluids, blood transfusion, exogenous calcium supplementation and whether this was before or after blood product transfusion, vasoactive agents, and IR/surgery) were obtained from the EMR.In addition, admission blood gas data and physiological observations were extracted from the EMR.Point-of-care levels of iCa were defined dichotomously as iHypoCa (iCa < 1.16 mmol/L) and normocalcaemia (iNormoCa) (iCa ≥ 1.16 mmol/L) to reflect previous literature [20].Levels of iCa were not adjusted for pH as this risks underestimating iHypoCa and the utility of adjusting this in the clinical setting is unclear [20].

Sample size
As a single-centre pilot study, a consecutive sample of 45 paediatric major trauma patients with iCa results was targeted (Appendix B).This sample size was estimated based on the extrapolation of previous patient numbers in a regional study [22].The proportioned sample size for a subsequent multicentre study is shown in Appendix C.

Outcomes
The primary outcome of the presence of hypotension (defined dichotomously based upon age-related values) was measured as the lowest recorded systolic blood pressure on emergency department admission (or as close as possible to allow a temporal link with blood gas data).
Secondary outcomes included the need for exogenous calcium, vasoactive medication, transfusion, IR/surgery within the first 24 hours of admission (defined dichotomously), and hospital/PICU LOS.Mortality within 24 hours and 30 days was also measured, as was functional outcome at 30 days (GOS score).
Basic demographics, mechanism of injury, and injury data are reported as numbers (percentages) and mean (+/-standard deviation (SD)) or median (interquartile range (IQR)).Normality was assessed using Shapiro-Wilk formal testing and quantile-normal plotting.Normally distributed parameters were compared using Student's t-test.The Mann-Whitney U test was used for non-normally distributed variables.The χ2 test was used to analyse categorical variables.
Multivariable logistic regression (supplementary appendix) was used to assess the dichotomous endpoint of hypotension for association with iHypoCa.The logistic regression adjusted for other potential variables of interest (age, gender, ISS, pre-hospital fluids, and acidosis (pH <7.35)).Wald and likelihood-ratio testing were used to evaluate different models' relative performance.Logistic regression models (Appendix D) were evaluated using post-estimation evaluation for calibration (e.g., Hosmer-Lemeshow goodness-of-fit, Cox plotting) and discrimination (e.g., cross-validated area c statistic).

Risk of bias
Previous studies have been affected by selection bias in their use of trauma team activation to include patients; this study's use of TARN criteria aims to mitigate this by capturing all relevant paediatric major trauma patients [16][17][18]20].Availability and recall bias are also relevant for documenting the handover of treatments given pre-hospital.To mitigate these biases, the original scanned pre-hospital documentation was reviewed and checked against data in the EMR.

Total (n = 45)
Normocalcaemia (n = 37) Hypocalcaemia (n =  Other than the adolescent age group (p <0.05), there were no statistically significant differences in the demographics and baseline clinical characteristics between the iHypoCa and iNormoCa groups.
Patients arrived at a median of 110.0 (IQR 94.0-128.0)minutes after their injury and had an iCa level measured at 11.0 (IQR 5.0-25.0)minutes after hospital arrival.
There were no differences between groups in markers of perfusion (pH, lactate) or coagulation (PT, APTT, and fibrinogen).In addition, there were no differences between groups for the adverse outcomes of mortality, morbidity (GOS ≤4), or PICU and hospital LOS (   )) received intravenous fluids ≤24 hours of admission; proportions did not differ between the iHypoCa and iNormoCa groups.Six of 45 (13.3% (95% CI 6.3-26.2%))received blood in the first 24 hours, and this was comparable between groups (Table 3).Two of 45 (4.4%) patients received exogenous calcium administration after blood product transfusion; both of these patients were in the iHypoCa group, and neither had severe iHypoCa (<1.0 mmol/L) at the time of the first blood gas measurement.

Logistic regression
None of the differential determinants included in the logistic regression model were statistically significant (age, gender, ISS, pre-hospital fluids, and acidosis) (supplementary appendix).These differential determinants were subsequently included in a multivariable logistic regression model for adverse outcomes (Table 4).

TABLE 4: Results of the logistic regression for the adverse outcomes associated with admission ionised hypocalcaemia for the 45 paediatric major trauma patients
Hypotension = systolic blood pressure of <80 mmHg for <one year, <85 mmHg for one to five years, <90 mmHg for five to 12 years, <100 mmHg for >12 years; elevated SIPA = Shock Index, Paediatric Age-Adjusted, SIPA >1.2 for zero to six years, >1 for seven to 12 years, >0.9 for >12 years; IR = interventional radiology; GOS = Glasgow Outcome Scale This multivariable logistic regression model demonstrated point estimates toward increased odds of all adverse outcomes for the iHypoCa group.However, the only statistically significant adverse outcome was the need for IR/surgery in the first 24 hours (OR 10.9 (95% CI 1.4-159.4)p ≤ 0.05).Mortality was unable to be included in the model due to the low mortality rate in this cohort, 1/45 (2.2% (95% CI 0.1-11.6%)).

Discussion
In this pilot study, admission iHypoCa was observed to be present in approximately one in six patients and was more prevalent amongst adolescents.A multivariable logistic regression model for adverse outcomes demonstrated a point estimate in the direction of iHypoCa patients having two to three times the odds of haemodynamic instability compared to iNormoCa patients; however, this was underpowered due to the small sample.In the first 24 hours, a 10-fold increased odds of IR/surgery were observed.There were also point estimates in the direction of iHypoCa being associated with all other adverse outcomes.

Incidence and differential determinants
In this study, the incidence of iHypoCa was similar to the incidence of 112/710 (15.8%) reported amongst paediatric major trauma patients in a systematic review and meta-analysis [20].However, the systematic review observed a wide range of definitions of iHypoCa (<1.00 mmol/l to <1.16 mmol/l) and variability in incidence (5.3-46.5%)[20].Conversely, the highest incidence of 66/142 (46.5%) was observed amongst the cohort described by Ciaraglia et al., for whom the definition of iHypoCa was the lowest [17,20].This may be a result of the high proportion of penetrating injuries (38/142 (26.7%)) observed amongst the patients in the study reported by Ciaraglia et al., although this was not statistically significant between the iHypoCa and iNormoCa groups [17].Studies among adult major trauma patients have also observed a wide range of definitions and incidence [5,10,13].Although a systematic review involving adult major trauma patients reported the incidence to range between 23.0% and 56.2%, two recent retrospective cohort studies comprising a larger sample size than meta-analysed data from the systematic review have also reported a variable incidence: 869/1,981 (43.9%) in an American single-centre study by Ciaraglia et al. and 3,982/30,183 (13.2%) in a European multicentre study by Helsloot et al. [5,10,13].Amongst paediatric major trauma patients, larger multicentre studies are required to assess the most appropriate cut-offs and the true incidence of admission iHypoCa.
In this single-centre study, no statistically significant difference was seen for gender, trauma mechanism, most injured region of the body, ISS, or administration of pre-hospital fluids; however, this should be interpreted with caution due to the small sample size.Age was observed to be a significant differential determinant, with a greater incidence of iHypoCa observed amongst adolescent age groups in both this single-centre study, and in the systematic review, an age group where physiology and response to trauma injury may be more like adult patients [20].Further research is required to determine whether iHypoCa in paediatric trauma should be considered different to the adult trauma population and whether there are relevant age-related cut-offs.
Studies amongst adults have shown injury severity to be significantly associated with iHypoCa [5,10,11,13].There are also a few studies, which indicate that penetrating injury and blast injury may be of significance [6,13,23].Amongst paediatric patients, a larger dataset is required to demonstrate any association with the mechanism of injury, with such data having great utility in emergency medicine with regard to trauma prealerts and subsequent preparation.Timing of iHypoCa is another important differential determinant, which may be useful in the pre-hospital and early management of trauma.This single-centre study is novel in reporting the timing of iCa measurement with regard to injury and hospital arrival times; although no significant difference was seen, this may be a result of the small sample size and the operational efficiency of the regional trauma network [22].Further work may be useful in exploring at what time iHypoCa occurs and whether this occurs at the point of injury.

Haemodynamic instability
Only two studies involving paediatric major trauma patients have explored the association between iHypoCa and haemodynamic instability, both looking at the Shock Index, Paediatric Age-Adjusted (SIPA) [17,18].Ciaraglia et al. observed 28/66 (42.4%) of iHypoCa patients to be haemodynamically unstable, and similar to this single-centre study, this was associated with a threefold increased odds of haemodynamic instability (OR 3.6 (95%CI 1.7-7.7))[17].By contrast, Epstein et al. observed 6/24 (25.0%) of iHypoCa patients to be haemodynamically unstable and that this was not associated with increased odds of haemodynamic instability (OR 0.9 (95%CI 0.3-2.3))[18].Among adult major trauma patients, both hypotension and a significantly worse shock index amongst iHypoCa (iCa < 1.0 mmol/l) have been observed across studies [5,10,12,13].All these studies had a lower iHypoCa cut-off than in this single-centre study.The aetiology and degree of tachycardia in paediatric patients can also be variable [24].Therefore, larger multicentre studies would benefit from exploring whether the observed point estimate towards haemodynamic instability is statistically significant with a larger sample size and at what cut-off for iHypoCa haemodynamic instability is seen.

Laboratory abnormalities
In this single-centre study, no significant differences were seen between the iHypoCa and iNormoCa groups for the markers of physiological derangement, such as pH and lactate or markers of clotting.This is similar to results from the paediatric systematic review and meta-analysis [20].By contrast, adult major trauma patients with iHypoCa have been observed to be at greater risk for acidosis, hyperlactatemia, and raised base deficit, potentially reflecting impaired perfusion and ischaemic injury [10,13].This is physiologically significant as it further potentiates iHypoCa due to pH-dependent calcium-lactate binding and stimulation of parathyroid hormone secretion, and thus also affects the other components of the 'diamond of death' [4].
The difference in results between paediatric and adult studies may reflect the paucity of paediatric data compared to adult data or may suggest an underlying physiological difference based on age group, which requires further exploration.
Studies of adult major trauma patients have shown markers of clotting to be heterogeneously reported across studies, and iHypoCa has been observed to be associated with coagulopathy, with studies reporting significantly increased INR, PT, and APTT [5,9,10,12,13].Mechanistically, calcium is an essential co-factor in the clotting cascade and is important for platelet activation and aggregation [25].Minimal research has included iHypoCa major trauma patients with thromboelastography (TEG) or rotational thromboelastometry (ROTEM) measurement, which may be due to this not being widely used in clinical practice [26].Future studies would benefit from incorporating viscoelastic measurements to appreciate the relationship between iHypoCa and coagulopathy better.

Treatments received
Although not powered for statistical significance, this single-centre pilot study suggested a point estimate towards increased odds of blood transfusion requirements and receipt of vasoactive medications.By contrast, other studies in both paediatric and adult trauma populations have demonstrated that iHypoCa patients have a statistically significant increased transfusion requirement [5,9,13,18,20] and requirement for vasoactive medications [13].This may be important when planning treatment and resuscitation strategies.Current guidelines recommend maintaining normal iCa levels in the bleeding trauma patient [27].In addition, studies have demonstrated that iHypoCa pre-transfusion can be significantly worsened following the administration of even a single unit of citrated blood products [6].Therefore, recognition and early treatment of iHypoCa is prudent.
This single-centre study is novel in reporting the need for IR/surgery.The odds of requiring IR/surgery in the first 24 hours of admission in the iHypoCa group were nearly 11-fold, and this was statistically significant.This is again significant in considering which patients may require more aggressive treatment and also has significance when considering the wider physical and psychological morbidity that occurs in children undergoing such treatments [28].Since this was significant in the multivariate analysis but not univariate analysis, there is a suggestion that other differential determinants, such as injury severity, may be relevant, and this would benefit from further exploration in a larger multicentre study.

Adverse outcomes
Amongst paediatric patients, the functional outcome (GOS score) may be a more meaningful outcome than mortality due to the relatively low overall mortality in this cohort [29].This is shown when comparing mortality outcomes between paediatric and adult studies, with paediatric studies not demonstrating a mortality difference, whilst adult studies showed a difference across studies [5,10,12,13,14].However, amongst adult major trauma patients, Helsloot et al. demonstrated that iCa levels have a parabolic relationship with adverse outcomes, with both iHypoCa and hypercalcaemia being associated with poor outcomes [13].Moreover, a systematic review of empirical calcium administration in cardiac arrest suggested that this may be associated with harm [30].As such, the effect of calcium administration is unclear and empirical administration is not recommended.Further studies exploring whether exogenous administration of calcium for paediatric patients with admission iHypoCa improves outcomes are recommended.

Limitations
The small sample size and single-centre retrospective design limited this study as it was underpowered to detect differences in the primary outcome; however, point estimates were useful in hypothesis generation for future multicentre studies.As a reflection of the small sample size, there was also a small number of penetrating injuries, which limits generalisability, and mortality was so low that this was not able to be formally analysed.These may have been important outcomes to include in a multivariate analysis.However, as a single-centre pilot study, this methodology is appropriate, and these limitations can be overcome by future multicentre expansion of the study and accordant adjustment of models of analysis.Given the large number of explanatory variables relative to the number of occurrences of the objective variable, the accuracy of the logistic regression model may have been affected.The multicentre expansion of this study would benefit from evaluating the AIC to optimise the model and evaluate its accuracy.
Bias associated with the retrospective design was mitigated by a manual review of the scanned and electronic medical records and quality checking of data extraction by a second author (supplementary appendix).However, although there was a detailed review of the records, extraction of the most injured regions may have missed neck injuries and/or parathyroid injuries, which could directly influence iCa levels.
In addition, details on the type of surgery/IR required were not explored.Despite such limitations, this study is novel in considering a number of potential confounders related to pre-hospital treatments and timing of measurement that have not been considered across previous paediatric studies and may be relevant

Appendix D Logistic Regression Modelling
Logistic regression modelling for the differential determinants of admission ionised hypocalcaemia is shown in Table 5.

TABLE 5: Logistic regression for the differential determinants of admission ionised hypocalcaemia for 45 paediatric major trauma patients
The logistic regression modelling for the adverse outcome of hypotension associated with admission ionised hypocalcaemia is shown in Table 6.

6: Logistic regression for the adverse outcome of hypotension associated with admission ionised hypocalcaemia for 45 paediatric major trauma patients
The logistic regression modelling for the adverse outcome of elevated shock index associated with admission ionised hypocalcaemia is shown in Table 7.The logistic regression modelling for the adverse outcome of the need for interventional radiology or an operation associated with the admission ionised hypocalcaemia is shown in Table 10.

TABLE 3 : Pre-hospital treatments and treatments within the first 24 hours for 45 paediatric major trauma patients, dichotomised by the ionised calcium level on arrival to the emergency department
Hypocalcaemia = ionised calcium <1.16 mmol/L; normocalcaemia = ionised calcium ≥ 1.16 mmol/L; IQR = interquartile range; SD = standard deviation; IR = interventional radiology

TABLE 7 : Logistic regression for the adverse outcome of elevated shock index (paediatric age- adjusted) associated with admission ionised hypocalcaemia for 45 paediatric major trauma patients
The logistic regression modelling for the adverse outcome of the need for blood transfusion associated with admission ionised hypocalcaemia is shown in Table8.

TABLE 8 : Logistic regression for the adverse outcome of the need for blood transfusion in the first 24 hours associated with admission ionised hypocalcaemia for 45 paediatric major trauma patients
The logistic regression modelling for the adverse outcome of the need for vasoactive medications associated with admission ionised hypocalcaemia is shown in Table9.

TABLE 10 : Logistic regression for the adverse outcome of the need for interventional radiology/operation in the first 24 hours associated with admission ionised hypocalcaemia for 45 paediatric major trauma patients
The logistic regression modelling for the adverse outcome of Glasgow Outcomes Scale ≤ 4 associated with admission ionised hypocalcaemia is shown inTable 11.