Early Hypercalcemia is Associated With Mortality in Critically Ill, Transfused Trauma Patients

Abstract

Introduction:

The detrimental effects of calcium derangements are well recognized in trauma patients, especially in those requiring resuscitation. Although hypocalcemia has been associated with worse outcomes, hypercalcemia incidence and impact are unknown. We hypothesized that hypercalcemia is associated with higher mortality in trauma patients requiring blood transfusion.

Methods:

Adult trauma patients who received blood products in the first 24 h at a level 1 trauma center from 2021 to 2022 and were evaluated. The institution’s trauma registry and charts were reviewed to collect demographics, calcium levels, transfusion records, and outcomes. Youden indices were used to define extreme hypercalcemia. Multivariable logistic regression was performed to determine independent associations.

Results:

Ninety-seven transfused trauma patients were included, 24 of whom received more than 4 units in 4 h. Free calcium of 7.99 had the highest Youden index for mortality and was used to define the lower limit of the extreme hypercalcemia cohort. The hypercalcemic and extreme hypercalcemia cohorts had comparable admission free calcium levels to the nonhypercalcemic cohort (4.6, 4.6, 4.7 mg/dL, P = 0.06). However, the hypercalcemic and extreme hypercalcemia cohorts had more calcium chloride administered (0, 1, 2 g, P = 0.001) and a higher number of units of blood transfused by 4 and 24 h (16 [5–93.5] versus 4 [2–8] units, P = 0.02). On univariable analysis, 30-d mortality was associated with extreme hypercalcemia, higher Injury Severity Score, and increased blood administration. On multivariable analysis, 30-d mortality remained associated with hypercalcemia (odds ratio, 17.5, confidence interval, 2.5–123.3; Wald P value 0.004) and platelets administered after 4 h (odds ratio, 2.3, confidence interval, 1.0–5.2; Wald P value 0.04).

Conclusions:

In this retrospective analysis of transfused trauma patients, extreme hypercalcemia was independently associated with 30-d mortality. Future studies should evaluate the physiologic impact of postinjury hypercalcemia.

Introduction

The transfused trauma patient presents a complex, life-threatening situation in which providers attempt to keep innumerable variables in balance.¹ Stored blood products, however, inherently create a challenge to that balance. In order to have readily available blood products, they must be stored and are done so in a citrate-based preservative solution to prevent clotting via the binding of free calcium,² subsequently making calcium physiologically inactive.³ When citrated blood products are transfused during hemorrhagic shock, the citrate continues to bind free calcium, working against the patient’s need for calcium to promote hemostasis. This chelation of calcium has led to the empiric administration of calcium during massive transfusion protocols,⁴ with the aim of increasing biologically active calcium (free calcium) that can partake in its innumerable roles, including its essential functions within the coagulation cascade.⁵

Hypocalcemia has been well documented postinjury, with literature consistently demonstrating high incidence in transfused patients,^6,7 suggesting causes to be multifactorial⁸ and finding independent associations with negative clinical outcomes, including mortality.^9–12 Where literature falls short is in addressing the balance of supplementing enough calcium during blood product transfusion to avoid hypocalcemia but not to the point of inducing hypercalcemia.^13,14 This gap in understanding the ongoing changes in serum calcium can also be attributed to most literature focusing on the first, but not the subsequent, calcium levels after injury.

Hypercalcemia, in contrast to hypocalcemia, is not well studied. While searching “hypocalcemia and transfusion” in published article titles on the National Institutes of Health PubMed, 15 manuscripts result. In comparison, while searching “hypercalcemia and transfusion” in published article titles, 0 manuscripts result. What is known is that hypercalcemia has an incidence that is thought to be low, between 3%⁶ and 22%,⁷ and is thought to be iatrogenic in nature,¹³ but the time course and associated clinical outcomes of hypercalcemia are not well described. Given this gap in knowledge, we aimed to describe the incidence of hypercalcemia in transfused adult trauma patients and its association with 30-d mortality. We hypothesized that hypercalcemia is associated with a higher 30-d mortality rate in trauma patients requiring blood transfusion.

Methods

Study design

The study was reviewed and approved by the local institutional review board under Protocol #2022–0074 with a waiver of informed consent. Adult trauma patients who received blood products in the first 24 h at a level 1 trauma center from September 2021 to August 2022 were included. Patients needed only one free calcium level recorded within the first 24 h to be included in the study.

The institution’s trauma registry and patient charts were reviewed to collect demographics, free calcium levels, transfusion records, and outcomes. Free calcium was tracked from admission to initial hypercalcemia and then for time to normalized calcium. Calcium administered was tracked for the first 24 h of admission.

Definitions

Our hospital’s lab reference range for free calcium is 4.4–5.4 mg/dL. Using that reference range, “not hypercalcemic” is defined as free calcium ≤ 5.4 mg/dL and “hypercalcemic” as 5.5–7.9 mg/dL “Extreme hypercalcemia” is ≥ 7.9 mg/dL based on Youden indices (=sensitivity – (1 – specificity)], which describes the point at which sensitivity and specificity are concordantly maximized. The highest Youden index possible is 1, meaning the sensitivity is 100% and the specificity is 100%. A sensitivity and specificity both of 50% give a Youden index of 0, and a Youden index between −1 and 0 means the test performs worse than change or worse than 50% sensitivity and specificity.

Blood product transfusion requirements were defined as massive and submassive. Massive transfusion is defined as receiving five or more units of packed red blood cells and/or whole blood within 4 h, and submassive is 1–4 units of packed red blood cells and/or whole blood within 4 h.¹⁵

Total calcium administered was calculated as [3 × calcium gluconate + calcium chloride]. This was chosen to account for calcium chloride having three times the amount of elemental calcium compared to calcium gluconate.

Statistics

Demographic comparison and multivariable logistic regression analyses were performed to determine independent associations. Relevant predictors for multivariable logistic regression included age, sex, body mass index, injury mechanism, Injury Severity Score, admission calcium, hypercalcemia by laboratory reference upper limit, hypercalcemia by extreme upper limit, total calcium administered, and blood products administered. Nonparametric continuous variables were reported as median [Interquartile range] and were compared using Wilcoxon rank-sum test. All continuous variables were nonparametric based upon Brown-Forsythe test and confirmed with Welch’s test. Categorical variables were described as n (%) and were compared using Pearson chi-square tests. If data were noted to be unavailable for a case entry, that case was excluded from the specific analysis for which data were incomplete. Differences were considered statistically significant for P values < 0.05, and all tests were two-tailed.

A multivariable logistic regression model for factors associated with 30-d mortality was built with all variables with Wald P value ≤ 0.2 on univariable logistic regression analysis. The final multivariable model was reduced using backwards stepwise elimination. Covariance between evaluated predictors was assessed and adjusted for if present during multivariable modeling. All analyses were performed using JMP Statistical Discovery Pro v18 (SAS Institute, Cary, NC).

Results

Cohort summary

Ninety-seven transfused trauma patients were included. Twenty-four received massive transfusion, and the remaining 73 received submassive transfusion. Of these transfused patients, 72.2% (n = 70) were male, with a median age of 43.2 [31.6, 63.8] y. Sixty-seven percent (n = 65) had a blunt mechanism of injury, the median Injury Severity Score was 22 [14, 34], and the overall 30 d mortality rate was 13.4% (n = 13) (Table 1).

Table 1 –

Comparison of demographics, injury, calcium level, calcium administration, and mortality.

Variable	Not hypercalcemic n = 65 (median [IQR], or %)	Hypercalcemic n = 27 (median [IQR], or %)	Extreme hypercalcemic n = 5 (median [IQR], or %)	P value
Age	50.6 [45.8–55.4]	41.7 [34.3–49.2]	33.6 [16.0–51.3]	0.043*
Sex, female	24.6%	33.3%	40.0%	0.57
Mechanism, blunt	69.2%	70.4%	20.0%	0.07
Body mass index	29.1 [27.3–30.8]	28.6 [26.0–31.2]	26.6 [20.6–32.6]	0.72
Injury Severity Score	24.4 [20.9–27.9]	28.1 [22.7–33.6]	26.4 [13.8–39.0]	0.51
Admission free calcium (mg/dL)	4.6 [4.4–4.8]	4.6 [4.3–5.1]	4.7 [3.4–6.7]	0.86
Calcium chloride given (g)	0 [0–1]	1 [0–3.2]	2 [1.5–3.5]	<0.0001*
Calcium gluconate given (g)	2 [0–5]	1 [0–4]	2 [0–2]	0.21
Total calcium given	1.33 [0.33–2]	2 [1–2.33]	2.67 [1.83–3.83]	0.016*
30-d mortality	10.8%	11.1%	60.0%	0.0072*

Cohorts include not hypercalcemic (below upper limit of lab reference range), hypercalcemic (above lab reference range upper limit but below extreme hypercalcemia cutoff), extreme hypercalcemia (above extreme cutoff).

IQR = interquartile range.

^*Denotes P < 0.05.

Defining and describing hypercalcemia

Thirty-two patients (33%) were hypercalcemic as defined by exceeding the upper limit of the lab reference range for free calcium, with a median initial hypercalcemic value of 5.9 [5.6–6.4] mg/dL, which occurred at a median of 1.7 [1.0–5.0] h after admission. Patients spent 1.9 [0.7–8.2] h with an elevated free calcium level before normalizing.

Free calcium of 7.99 mg/dL had the highest Youden index of 0.42 for 30-d mortality (Fig.), which then defined the lower limit of extreme hypercalcemia. Five percent of transfused trauma patients (n = 5) had at least one free calcium level that exceeded the extreme hypercalcemic threshold. Extreme hypercalcemic patients had a median initial hypercalcemic calcium of 8.4 [8.0–9.9] mg/dL, which occurred 1.4 [0.5–1.9] h after presentation and normalized on average 1.9 [0.5–13.9] h thereafter.

Fig. –

Youden index of free calcium and 30-d mortality. Youden index was highest at 0.4231 for a free calcium of 7.99 mg/dL. Youden index sensitivity – (1 – specificity) 0.500 – 0.0769 0.4231. Area under the curve was 0.63.

Not hypercalcemic versus hypercalcemic versus extreme hypercalcemia

Patients were then compared as three cohorts: not hypercalcemic (n = 65), hypercalcemic (n = 27), and extreme hypercalcemia (n = 5). Patients had comparable sex distribution and body mass index. Age differed between the three cohorts, trending younger in hypercalcemic and extreme hypercalcemic patients (50.6, 41.7, 33.6 y, P = 0.04). Patients groups had similar injury mechanisms and Injury Severity Scores. Interestingly, median admission free calcium values were comparable between the three cohorts (4.6, 4.6, 4.7 mg/dL, P = 0.86). Calcium gluconate administered was comparable across cohorts, while calcium chloride (0, 1, 2 g, P < 0.0001) and total calcium (P = 0.016) differed. Finally, the 30-d mortality rate increased by group from not hypercalcemic to hypercalcemic to extreme hypercalcemia (10.8, 11.1, 60.0%, P = 0.007) (Table 1).

As for transfusion comparisons between the three cohorts at the 4 h time point, patients with hypercalcemia and extreme hypercalcemia were more likely to be massively transfused (13.9, 44.4, 60.0%, P = 0.0014), have a higher amount of total blood products transfused (2, 7, 16 units, P < 0.001), and have received a higher amount of packed red blood cells, fresh frozen plasma, and platelets transfused compared to not hypercalcemic patients (P < 0.05) (Table 2). At the 24-h time point, hypercalcemic and extreme hypercalcemia patients continued to have more total blood products transfused, including packed red blood cells and fresh frozen plasma compared to nonhypercalemic patients (P ≤ 0.0001) (Table 2).

Table 2 –

Comparison of transfusion requirements at 4 and 24 h.

Variable	Not hypercalcemic n = 65 (median [IQR], or %)	Hypercalcemic n = 27 (median [IQR], or %)	Extreme hypercalcemia n = 5 (median [IQR], or %)	P value
4 h time point
Massive transfusion (yes)	13.9%	44.4%	60.0%	0.001*
Total blood products (units)	2 [1–4]	7 [2–11.5]	16 [1.5–93.5]	0.0004*
Whole blood (units)	0 [0–1]	0 [0–1]	0 [0–3.5]	0.81
Packed red blood cells (units)	1 [0–3]	3 [2–6]	10 [1–40.5]	0.0002*
Fresh frozen plasma (units)	1 [0–2]	2 [0–5]	5 [0.5–45]	0.004*
Platelets (units)	0 [0–0]	0 [0–0.25]	0 [0–5]	0.04*
24 h time point
Total blood product (units)	3 [2–6]	9 [5–17]	16 [5–93.5]	<0.0001*
Whole blood (units)	0 [0–1]	0 [0–1]	0 [0–3.5]	0.81
Packed red blood cells (units)	2 [1–3]	5 [3–8]	10 [3–40.5]	<0.0001*
Fresh frozen plasma (units)	1 [0–2]	3 [1–6]	5 [2–45]	0.0001*
Platelets (units)	0 [0–0]	0 [0–1]	0 [0–5]	0.06

Cohorts include not hypercalcemic (below upper limit of lab reference range), hypercalcemic (above lab reference range upper limit but below extreme hypercalcemia cutoff), and extreme hypercalcemia (above extreme cutoff). Massive transfusion is defined as > 4 units of packed red blood cells plus whole blood in 4 h.

IQR = interquartile range.

^*Denotes P < 0.05.

Multivariable analysis

To control for confounding variables, multivariable analysis was performed. On univariable analysis, 30-d mortality was associated with Injury Severity Score ≥ 22, higher admission calcium level, extreme hypercalcemia, increased blood product administration at 4 h, increased platelets administered at 4 h, and total calcium administered (P < 0.20) (Table 3). These variables were then put into multivariable analysis, in which 30-d mortality remained independently associated with extreme hypercalcemia and number of platelet units administered at 4 h (P < 0.05) (Table 3).

Table 3 –

Univariable and multivariable analysis for 30-d mortality.

Variable	Univariable			Multivariable
	OR	95% CI	P value	OR	95% CI	P value
Injury Severity Score ≥ 22	2.72	0.78–1.06	0.10	-	-	>0.05
Admission free calcium	1.66	0.77–3.59	0.19	-	-	>0.05
Extreme hypercalcemia	12.30	1.83–82.7	0.01	17.48	2.48–123.27	0.004*
Total blood products administered at 4 h	1.05	1.00–1.11	0.06	-	-	>0.05
Total platelets administered at 4 h	2.00	0.92–4.32	0.08	2.31	1.03–5.21	0.04*
Total calcium administered	1.43	0.09–2.27	0.13	-	-	>0.05

Variables with Wald P value < 0.02 were included in multivariable analysis.

OR = odds ratio; CI = confidence interval.

^*Denotes P < 0.05.

Discussion

Summary

This study found hypercalcemia to be present in 33% of transfused trauma patients. Patients with hypercalcemia most often presented normocalcemic and become hypercalcemic after admission. Hypercalcemic patients were also found to have received more calcium chloride but not calcium gluconate, received more blood products at 4 and 24 h, and had higher 30-d mortality rates compared to not hypercalcemic patients. On multivariable analysis, postinjury extreme hypercalcemia and platelets transfused at 4 h were independently associated with 30-d mortality.

Iatrogenic nature of hypercalcemia

This study noted that patients who were hypercalcemic during admission presented to the emergency department with a free calcium level that was within normal limits. The median time to initial hypercalcemic level was approximately 6 h after presentation for all hypercalcemic patients, but earlier, at 1.4 h for the extreme hypercalcemia alone cohort. As for time to normalize serum calcium levels, patients with hypercalcemia, including the extreme cohort, took approximately 2 h to no longer be hypercalcemic. This is in contrast to the pattern seen with hypocalcemic patients, who tend to be hypocalcemic before in-hospital interventions, including transfusion.¹⁶ Hypocalcemia has been shown to have several independent predictors including penetrating mechanism, increased Injury Severity Score, and packed red blood cell and fresh frozen plasma administration.⁸ In summary, hypocalcemia is early and multifaceted―both iatrogenic in nature because of citrate administration,¹³ and also driven by injury mechanism. Our study suggests that hypercalcemia is induced after admission and is speculatively more iatrogenic in nature than hypocalcemia.

Ratio of calcium to citrated blood product

The iatrogenic nature of calcium derangements is at least in part due to the empiric dosing during blood product transfusion. Our study found calcium chloride and total calcium administered to be higher in hypercalcemic cohorts. Optimal dosing of calcium supplementation during transfusion is unknown, although prior studies have attempted to answer this question. The first obstacle is to define how much citrate is in various blood products, with platelets and fresh frozen plasma have the highest concentration, followed by packed red blood cells,^17,18 albeit unreliable and widely variable.^17,19 The second obstacle is to define a ratio of citrate administered to calcium supplemented to avoid both hypo- and hypercalcemia. In one study, via retrospective review of transfused trauma patients who required an operation, found no optimal cutoff was demonstrated to predict severe hypocalcemia; however, the authors did find a ratio of 0.903 mmol of administered calcium per citrated blood product could differentiate hypercalcemic patients from the rest.¹³ To account for the difference in citrate levels across blood products, a study attempted to utilize a calcium administered–to–citrate molar ratio but found no relationship between mortality and calcium dose corrected for citrate load.¹⁴ Our hospital’s calcium supplementation is provider- and situation-dependent; most often supplemented based on point-of-care testing in the acute setting, including in the trauma resuscitation bay and the operating room. Altogether, if hypercalcemia is a largely iatrogenic phenomenon, citrate concentrations in blood products and calcium administration during transfusion must be actively taken into consideration during resuscitation to better understand and avoid calcium derangements.

Limitations

The authors acknowledge the following limitations. First, this study utilized a modest sample size. This sample, however, was meant to serve as a pilot to evaluate the incidence and negative associations of hypercalcemia, which it was able to identify using an extreme hypercalcemic cohort. Utilizing an extreme cohort is not uncommon in hypocalcemia studies,⁷ but expanding the sample size in the future will potentially negate the need for an extreme cohort and subsequently be more widely clinically applicable. In addition, our finding of extreme hypercalcemia being independently associated with 30-d mortality had an odds ratio of 17, with a wide-ranging confidence interval of 2–123, which could potentially be attributed to sample size and blood sample timing after calcium administration. Second, this study did not compare hypercalcemia against hypocalcemia, but rather highlighted associations of hypercalcemia alone. Future studies will incorporate hypocalcemia to compare odds ratios as well as incorporate hypo- and hypercalcemia into synchronous multivariable analyses. Third, to better understand the potentially iatrogenic nature of hypercalcemia, nontransfused patients will be incorporated as an additional control cohort and the time course of normo-, hypo-, and hypercalcemia will be further delineated in future investigations. Finally, hypercalcemia is intertwined with resuscitation, and although this study included transfusion status, additional markers of shock were not included. Future studies could include markers of hemorrhagic shock, such as lactic acid.

Conclusions

Calcium derangement studies have previously focused primarily on hypocalcemia and its subsequent correction. This study highlights the importance of hypercalcemia―both in noteworthy incidence in transfused patients postinjury and through its independent association with 30-d mortality. Future studies need to expand upon these findings in a larger cohort and evaluate the physiologic implications of postinjury hypercalcemia.