Calcium administration appears not to benefit acute pancreatitis patients with hypocalcemia

Hypocalcemia occurs commonly among patients with acute pancreatitis (AP) in the intensive care unit (ICU). Calcium therapy could be used to correct hypocalcemia and maintain calcium levels, but its impact on the prognosis has not been demonstrated. Our study aimed to determine whether calcium therapy could benefit the multiple outcomes of AP patients with hypocalcemia.


| INTRODUCTION
Acute pancreatitis (AP) is a severe inflammatory disease commonly observed in the intensive care unit (ICU), with an increasing annual incidence of 33.74 cases per 10 000 person-years. 1 A total of 80% of AP cases are mild to moderate with manifestations, and one-fifth of AP cases could develop severe acute pancreatitis (SAP) with significantly high mortality (20%-40%) and severe complications, including pancreatic necrosis, shock, and organ failure. 2,35][6] Hypocalcemia is also regarded as an independent predictor of all-cause mortality among patients with chronic heart failure and chronic renal disease. 7It has been demonstrated that hypocalcemia occurred more frequently in patients during severe attacks than mild attacks of AP (86% vs. 39%, p < .001). 8owever, the exact mechanism of hypocalcemia among AP patients is unknown, and the possible mechanisms include the saponification of calcium, hypomagnesemia, and hypoparathyroidism. 6,9Clinicians recommended calcium supplementation to correct hypocalcemia, as this deviation is considered harmful.1][12] Whereas the evidence of the association between calcium therapy and the prognosis is very sparse.
Our study aimed to determine whether receiving calcium treatment could influence the prognosis of patients diagnosed with AP.This retrospective study was designed based on one large critical care database named Medical Information Mart for Intensive Care (MIMIC-IV) with a variety of high-quality clinical data from hospital monitoring systems. 13

| Setting
Our study was based on the Medical Information Mart for Intensive Care (MIMIC-IV) program made available mainly through researchers at the MIT Laboratory for Computational Physiology and collaborating research groups. 14MIMIC-IV contains the data of 383 220 patients admitted to intensive care units (ICUs) at Beth Israel Deaconess Medical Center (BIDMC) between 2008 and 2019. 13One author (Yan Tianao) passed the Collaborative Institutional Training Initiative (CITI) program and had full access to the database (Record ID 41817834).The study was an analysis of the third-party anonymized publicly available database with pre-existing institutional review board approval.Written informed consent for participation was not required for this study in accordance with national legislation and institutional requirement.Our procedures were followed in accordance with the ethical standards of the responsible committee on human experimentation (institutional or regional) and with the Helsinki Declaration of 1975.

| Study population
We enrolled 6222 patients diagnosed with "acute pancreatitis" in MIMIC-IV and only included patients with initial serum calcium levels <8.5 mg/dL at the first admission to the ICU.Patients younger than 18 years and those who spent less than 12 h in ICU were excluded.A total of 807 AP patients with hypocalcemia were included in the final cohort, of which 620 patients received calcium therapy and comprised the calcium group.The rest of the 187 patients comprised the non-calcium group without calcium administration.In the calcium group, the patients were classified into three subgroups.(1) A total of 466 patients were given parenteral calcium infusion alone.(2) A total of 33 patients were given oral medication exclusively.(3) A total of 121 patients were given the combination of parenteral and oral therapy.The sample screening process is shown in Figure 1, and the number of patients in each diagnosed title is provided in Table S1.MIMIC-IV had patients diagnosed with ICD-10 codes (International Classification of Diseases code, version 10), which was different from MIMIC-III. 14-16

| Study design
The treatment group received calcium therapy, including parenteral infusion (calcium gluconate, calcium chloride) and oral medication (calcium carbonate, calcium acetate) during the hospitalization.The time and dose of calcium treatment were also recorded in MIMIC-IV.The baseline characteristics within the first 24 h after ICU admission included gender, age, admission period, ethnicity, weight, sequential organ failure assessment (SOFA) score, simplified acute physiology score II (SAPS II) score, application of renal replacement therapy (RRT), mechanical ventilation (MV), comorbidities, laboratory tests, vital signs.The comorbidities, including congestive heart failure (CHF), chronic obstructive pulmonary disease (COPD), liver disease, renal disease, and malignant tumor, were collected on the recorded ICD-10 codes in the MIMIC-IV database.The laboratory tests were measured on the first 24 h in ICU.Considering the constantly changing of laboratory values during the acute stage, we included the most severe values in the first day from the MIMIC-IV database, including maximum lactate, maximum blood urea nitrogen (bun), maximum creatinine, minimum platelets, minimum white blood cell (WBC) count, minimum serum albumin level, minimum hemoglobin, minimum serum calcium level, and the vital signs including heart rate, respiratory rate, mean blood pressure (MBP), and temperature (°C).Furthermore, considering the correlation of serum calcium with albumin, we analyzed every patient's cCa (corrected calcium) using Payne's formula (correction of serum calcium mg/dL = total serum calcium mg/dL + 0.8 × [4.0 g/dL − albumin -g/dL]). 17

| Primary and secondary outcomes
The primary outcome was in-hospital mortality.Secondary outcomes included 28 days mortality, ICU mortality, hospital length of stay (days), and ICU length of stay (days).

| Statistical analysis
Categorical variables were presented as total numbers and percentages.Means (standard deviations) and interquartile ranges (IQR) were used to present continuous variables.The patients' characteristics were compared using χ 2 or Fisher exact tests for categorical variables and the t-test or Wilcoxon rank-sum test for continuous variables.We also used Kaplan-Meier (K-M) survival curve analysis to draw curves evaluating survival time and generate a log-rank trial comparing the differences between the two groups.In addition, we have several missing data (<20%) of the covariates and showed the percentage and amount of those missing data in Table S2.Multiple imputations were used to assume the data randomly in the MIMIC-IV and we used the Within approach, which was considered more desirable in terms of bias reduction in our research.We analyzed the influence of calcium therapy before and after multiple imputations.Propensity score matching (PSM) and propensity score-based inverse probability of treatment weighting (IPTW) methods were conducted to minimize the covariate differences and balance the baseline characteristics between the two groups. 18,19After estimating the patients' propensity scores, the Calcium group was matched to the control group as a 1:1 nearest neighbor with a caliper width of 0.05 through PSM.Moreover, IPTW adjusted the covariates and generated two virtual cohorts after weighing each patient.Standardized mean differences (SMD) can show the balance of characteristics between the two groups.Before matching, some variables' SMD was more than 0.1, which can be considered an imbalance between the two groups.After matching to test the effects of PSM and IPTW, almost all the SMD dropped to less than 0.1.
Multivariate Cox regression models were used to assess whether calcium therapy is associated with in-hospital mortality, 28 days mortality, and ICU mortality.The results are presented as hazard ratios (HRs) with 95% confidence intervals (Cls).To facilitate the clinical interpretation of our results, we adjusted for several covariates that were selected based on the impact on prognosis and the data loss rate.Multivariate linear regression models were also applied to analyze hospital outcomes of continuous variables.Subgroup analysis of the associations between calcium treatment and hospital mortality was performed on different serum calcium levels.
All the statistical analyses were performed using Rstudio (version 1.4.1106), and p < .05 was considered statistically significant, and all the data from the MIMIC database were extracted in SQL (Structured Query Language).

| Baseline characteristics
In total, 6222 patients diagnosed with acute pancreatitis in MIMIC-IV between 2008 and 2019 were screened.After exclusion, we included 807 AP adult patients, of whom 620 received calcium treatment during hospitalization (calcium group), and the remaining did not (non-calcium group).The baseline characteristics of the two groups are shown in Table 1.In the calcium group, patients had significantly higher inhospital mortality (16% vs. 4.3%, p < .001),significantly higher SOFA scores (7 [2-12] vs. 4 [1-7], p < .001),and higher lactate levels (2.7 mmol/L [0.3-5.1 mmol/L] vs. 1.9 mmol/L [0.5-2.3 mmol/L], p < 0.001) than the noncalcium group.In the calcium group, patients were more likely to receive RRT (8.2% vs. 1.1%, p < .001)and mechanical ventilation (42.7% vs. 16.6%,p < .001).As for the vital signs and laboratory tests, in the calcium group, patients have faster heart and respiratory rates, lower platelets, lower albumin, lower calcium, higher WBC, higher BUN, and higher creatinine levels.We performed PSM and IPTW methods to mitigate the estimation bias caused by imbalanced covariates between different treatment groups, and the covariates differences between groups almost disappeared.Almost all the SMD were less than 0.1, indicating that the baseline variables in the two groups have similar distributions.(Table 2).

| Primary outcome
A total of 107 patients (13.3%) died in the hospital.Patients in the calcium group had higher in-hospital mortality (16.0% vs. 4.3%, p < .001)and longer median length of stay in the hospital (19 (0-38) days vs. 10 (0-20) days, p < .001)than patients in the non-calcium group.The Kaplan-Meier survival curve showed no significant survival benefits before and after matching (log-rank test: p = .064PSM: p = .34IPTW: p = .30)between the two groups.(Figure 2) We also performed multivariate Cox regression to investigate whether calcium therapy could affect the prognosis of those AP patients.The results showed that calcium administration has no association with inhospital mortality of AP patients with hypocalcemia (HR: 1.03, 95% Cl: 0.47-2.27,p = .942,population: 807 patients) (Figure 3).Moreover, we performed IPTW and PSM to reduce bias caused by imbalanced covariates between the two groups, and the results showed no beneficial effect (PSM: HR: 1.25, 95% Cl: 0.33-4.69,p = .746,population: 302 patients; IPTW: HR: 1.52, 95% Cl: 0.68-3.42,p = .312,population: 1575 patients).Multivariate analysis in 388 patients without missing data was also performed to increase the robustness of our study, and we achieved similar results (HR: 1.03, 95% Cl: 0.29-3.61,p = .962)(Tables S3 and S4).

| Additional analysis
To investigate the influence of exclusive parenteral calcium infusion on the prognosis, we analyzed 466 patients who received exclusive infusion (calcium gluconate or calcium chloride) without oral medication.The multivariate Cox regression showed that parenteral infusion had no beneficial effects on hospital mortality (HR: 1.12, 95% Cl: 0.54-2.30,p = .767)in the Table S5.However, patients receiving both parenteral and oral therapy had different results.The multivariate Cox regression before and after matching showed a significant beneficial effect of combination therapy in-hospital mortality (HR: 0.04 95% Cl: (0.004-0.40) p < .001) in the Table S6.
Patients with various degrees of hypocalcemia have different disease severity and prognosis.To illustrate the impact of calcium therapy on AP patients with different initial serum calcium levels, we classified patients into three subgroups by various degrees of hypocalcemia: mild group (8.0 ≤ calcium ≤ 8.5), moderate group (7.1 < calcium < 8.0), severe group (calcium ≤ 7.1).As initial serum calcium levels decreased among groups, hospital mortality increased significantly (mild: moderate: severe 8.4%: 11.5%: 19.6%).We found that calcium therapy could not benefit hospital mortality of patients with different degrees of hypocalcemia through multivariate Cox regression before and after matching in Table S7.K-M survival curves were shown in Figure S1, and no significant survival benefits occurred in different groups.To evaluate the corrected calcium (cCa) with prognosis, we classified patients into two groups with different degrees of cCa.(466 patients with cCa <8.5 mg/dL; 339 patients with cCa ≥8.5 mg/dL) K-M analysis showed no significant survival differences between groups with different degrees of cCa, and multivariate Cox analysis showed that calcium therapy could not benefit those patients with cCa <8.5 mg/ dL (Figure S2; Table S8).From the results above, patients with different degrees of hypocalcemia could not benefit from calcium therapy.
F I G U R E 2 Kaplan-Meire survival curve analysis between calcium therapy group and non-calcium therapy group before and after matching from the MIMIC-IV database (a) Survival curve analysis between the calcium therapy group and the non-calcium therapy group before matching from the MIMIC-IV database.(b) Survival curve analysis between the calcium therapy group and the non-calcium therapy group after PSM matching from the MIMIC-IV database.(c) Survival curve analysis between the calcium therapy group and the non-calcium therapy group after IPTW matching from the MIMIC-IV database.IPTW, inverse probability of treatment weighing; PSM, propensity score matching.
We divided through the timing of calcium therapy into three subgroups, including <24 h group, 24-48 h group, and 48-120 h group.Interestingly, calcium therapy in all different timing (<24 h, 24-48 h, 48-120 h) could not benefit those patients in the Table S9.Additionally, organ failure and other severe complications are likely to occur during the early stage of AP. 3 The sequential Organ Failure Assessment (SOFA) score is often used to assess the severity of critical illness in the ICU.To determine whether patients with various disease severity could benefit from calcium therapy, we divided patients into the severe subgroup (SOFA ≥8) and the mild group (SOFA <8) and performed subgroup analysis.A total of 287 patients comprised the severe group, and their mortality was much higher than the mild group (mortality: high SOFA vs. low SOFA: 28.9% vs. 4.6%).After multivariate Cox regression, both the severe and mild groups could not benefit from the calcium therapy (Table S9).In addition, to avoid the interference of calcium fluctuation caused by RRT, we included and analyzed 754 patients without RRT use, and the result showed that calcium administration could not be beneficial to patients without RRT in Table S9.

| DISCUSSION
Our study is the first retrospective research concentrating on the influence of calcium supplementation on the multiple outcomes of AP patients with hypocalcemia.Our research found that calcium therapy could not benefit AP patients' multiple outcomes, including in-hospital mortality, 28-day mortality, and ICU mortality.Interestingly, calcium administration was correlated with prolonged length of stay in the hospital and ICU.In the subgroup analysis, exclusive parenteral calcium infusion was not associated with improved outcomes in AP patients.However, the combination of parenteral infusion and oral calcium medication could benefit those patients.There is very sparse evidence on the exact association between calcium administration and acute pancreatitis patients with hypocalcemia.
The strength of our study includes performing matching approaches (PSM and IPTW) to minimize the covariates differences.Although at admission to the ICU, patients in calcium group showed significantly serum calcium levels, higher severity scores, higher interventions proportion, and other worse vital signs.After PSM and IPTW, the baseline differences between groups were narrowed and nearly disappeared.Several subgroup analyses were conducted in our research to further investigate the correlation between calcium therapy and outcomes.Although parenteral calcium is always regarded as the primary choice to correct calcium derangement, we found no correlation between exclusive parenteral calcium infusion and in-hospital outcomes in our study.Furthermore, we found that the combination of parenteral infusion and oral therapy was associated with improved hospital mortality, and we speculated the following reason.Many patients with acute pancreatitis are prohibited from the diet in the acute stage, and most severe or moderate patients can only receive parenteral calcium supplements.We consider that the positive results might be related to the fact that individuals who could take medication orally might have better gastrointestinal conditions and severity of disease than those who only receive parenteral calcium.Additionally, we also classified patients into subgroups depending on different serum calcium levels and corrected calcium levels.Interestingly, whether those patients presented with mild, moderate, or severe hypocalcemia, all three groups could not benefit from the calcium administration.Considering the correlation of serum calcium with albumin, we analyzed every patient's cCa (corrected calcium) using Payne's formula, and we similarly found no association between corrected calcium levels and prognosis.Considering the potential impact caused by the differences in the timing of early therapy, we extracted the start time of calcium therapy.Interestingly, our results showed no association between the prognosis and the different timing of therapy, including <24, 24-48, and 48-120 h groups.According to our results, calcium treatment could not benefit AP patients with hypocalcemia.
To weigh the pros and cons of calcium supplementation, we list previous investigations on human or animal models below, and the exact impacts of calcium therapy have remained contentious.Severe hypocalcemia is regarded as an independent predictor of critically ill patients' mortality, and those patients in whom serum calcium fail to be normalized within the first 4 days might be associated with increased mortality (38% vs. 19%). 20,21Several studies have stated that calcium therapy appears not to provide any benefit and even has a detrimental effect on patients with critical illness or animal models.The study by Chung et al. found that during the preanhepatic phase, excessive parenteral calcium administration could contribute to a serum calcium surge, which is responsible for pancreatic acinar cell damage and the BAP (biochemical acute pancreatitis) in patients after liver transplantation. 22The study by Collage et al. suggests that calcium administration does not provide any benefit and even imposes harm on septic patients. 23ur study suggests that calcium therapy does not benefit AP patients and is even associated with prolonged LOS in the hospital and ICU.
We speculate some potential mechanisms in the unfavorable impact of calcium therapy on AP patients with hypocalcemia.Initially, the role of Ca 2+ is pivotal in the pathogenesis and pathophysiology of pancreatitis, and the sustained elevation of intracellular calcium is responsible for pancreatic acinar cell injury and death. 24,25espite patients with hypocalcemia having relatively low serum calcium levels, studies have demonstrated that their intracellular Ca 2+ is still elevated.Moreover, recent insight has suggested that hypocalcemia of SAP might play a protective role for the acinar cells by depleting the extracellular supply of calcium. 9If receiving extra calcium interventions, AP patients would have higher cytosolic Ca 2+ , resulting in more intracellular trypsinogen activation and further injury or necrosis of the pancreas.In addition, calcium therapy appears not to increase iCa (ionized calcium) concentrations significantly, and calcium levels could not be normalized by calcium supplementation alone. 21,26Compared with the spontaneous elevation in iCa that occurred in similar patients without calcium treatment, the study suggested that supplementation's influence was relatively small. 26Considering that most severe AP patients are complicated with hypoperfusion, pancreatic microcirculation disturbances, and tissue hypoxia, calcium supplementation could not improve hemodynamics impairment and even increase the risk of organ dysfunction.
There are several limitations in our study.First, due to limited previous research on calcium therapy in pancreatitis, speculation can only be based on the pathophysiological role of calcium in pancreatitis and studies of calcium supplementation in acute diseases such as sepsis.Second, several details after calcium administration are insufficient, such as the correction of the calcium and the progress of the disease.Propensity matching cannot equal the baselines of the patients in a precise sense, resulting in insufficient analysis.Additionally, due to the limitations in the MIMIC-IV database, we could not extract the data out of the hospital due to the inadequate follow-up time, resulting in the neglect of the advantages impact of calcium administration on AP patients. 13,14inally, because our research was based on a larger sample size observational database, randomized control trials are required to investigate calcium therapy on pancreatitis in the future. 5| AP patients hypocalcemia do not benefit from calcium administration.Calcium therapy has no association with multiple mortality and is significantly associated with prolonged LOS in the hospital and ICU.

F I G U R E 3
Effect of calcium therapy on primary outcome in acute pancreatitis patients from the MIMIC-IV database before and after matching through multivariate Cox regressions.HR, hazard ratio; IPTW; inverse probability of treatment weighing; PSM, propensity score matching.F I G U R E 4Effect of calcium therapy on secondary outcome in acute pancreatitis patients from the MIMIC-IV database before and after matching through multivariate Cox regressions.(a) Effect of calcium therapy on 28-days mortality and ICU mortality in acute pancreatitis patients from the MIMIC-IV database before and after matching through multivariate Cox regressions.(b) Effect of calcium therapy on hospital and ICU length of stays in acute pancreatitis patients from the MIMIC-IV database before and after matching through multivariate linear regressions.HR, hazard ratio; ICU, intensive care unit; IPTW, inverse probability of treatment weighing; PSM, propensity score matching.
Study sample screening process of 807 acute pancreatitis patients with hypocalcemia from MIMIC-IV database.ICU, intensive care unit.

1
Baseline characteristics of the included patients from the MIMIC-IV database.
Baseline characteristics and SMD between treatment groups of acute pancreatitis patients from the MIMIC-IV database after PSM and IPTW matching.
T A B L E 2