Introduction: Calcium is an essential co-factor in the coagulation cascade, and hypocalcemia is associated with adverse outcomes in bleeding patients, including trauma patients, women with postpartum hemorrhage, and patients with intracranial hemorrhage. In this retrospective, single-center, cohort study, we aimed to determine whether admission-ionized calcium (Ca++) is associated with higher rates of therapeutic interventions among patients presenting with acute nonvariceal upper gastrointestinal bleeding (NV-UGIB). Methods: Adult patients admitted due to NV-UGIB between January 2009 and April 2020 were identified. The primary outcome was defined as a need for clinical intervention (two or more packed cell transfusions, need for endoscopic, surgical, or angiographic intervention). Univariate and multivariable logistic regression analyses were performed to determine whether Ca++ was an independent predictor of the need for therapeutic interventions. Propensity score matching was performed to adjust the imbalances of covariates between the groups. Results: A total of 434 patients were included, of whom 148 (34.1%) had hypocalcemia (Ca++ <1.15 mmol/L). Patients with hypocalcemia were more likely to require therapeutic interventions than those without hypocalcemia (48.0% vs. 18.5%, p < 0.001). Specifically, patients with hypocalcemia were more likely to require endoscopic intervention for control of bleeding (25.0% vs. 15.7%, p = 0.03) and multiple packed cell transfusions (6.8% vs. 0.3%, p < 0.001). Additionally, they had significantly longer hospital stay (5.0 days [IQR 3.0–8.0] vs. 4.0 days [IQR 3.0–6.0], p = 0.01). After adjusting for multiple covariates, Ca++ was an independent predictor of the need for therapeutic intervention (aOR 1.62, 95% confidence interval [CI] 1.22–2.14, p < 0.001). The addition of Ca++ to the Modified Glasgow Blatchford score improved its accuracy in the prediction of therapeutic intervention from AUC of 0.68 (95% CI 0.63–0.72) to 0.72 (95% CI 0.67–0.76), p = 0.02. After incorporation of the propensity score, the results did not change significantly. Conclusion: These findings suggest that hypocalcemia is common and is associated with an adverse clinical course in patients with NV-UGIB. Measurement of Ca++ on admission may facilitate risk stratification in these patients. Trials are needed to assess whether the correction of hypocalcemia will lead to improved outcomes.

Upper gastrointestinal bleeding (UGIB) is a common medical emergency with an estimated annual incidence of 37–65 per 100,000 [1, 2]. More than 70% of UGIBs are caused by nonvariceal etiology (NV-UGIB) with gastroduodenal ulcer disease being the most common etiology [3]. Despite the development of pharmacological and endoscopic therapies for the management of NV-UGIB, it is still associated with significant morbidity and mortality [4].

Patients with NV-UGIB may present with a wide variety of clinical syndromes, from life-threatening to self-limiting bleeding. International consensus group guidelines recommend the implementation of validated prognostic scores for early risk stratification at admission [5]. However, these scores are not routinely used in clinical practice because they include many clinical and laboratory variables, which makes them cumbersome to use bedside.

Calcium plays a cardinal role in the coagulation cascade and platelet aggregation [6]. Ionized calcium (Ca++), the active form of calcium, is readily available and easily measured in the emergency department setting, as a part of blood gas analysis. Numerous studies found that hypocalcemia is associated with adverse clinical course in bleeding patients, including trauma patients, women with postpartum hemorrhage, and patients with intracranial hemorrhage [7‒10]. This study aimed to evaluate the association between hypocalcemia on admission and the need for therapeutic interventions (transfusion of two or more packed red blood cells (PCs), need for endoscopic intervention, angiography, or surgery) among patients hospitalized with NV-UGIB.

This is a retrospective cohort study, using patient data from electronic medical records. We included patients who were hospitalized with NV-UGIB at Rambam Health Care Campus (RHCC), Haifa, Israel, between January 1, 2009, and April 31, 2020, and underwent upper endoscopy. In our institution, angiography and endoscopy services are available around the clock. Chart review was conducted following published methods for retrospective studies [11]. The Strengthening the Reporting of Observational Studies in Epidemiology Statement was used to guide the reporting of this research [12].

The study population included adult patients (≥18 years old) hospitalized due to NV-UGIB, as recorded by the emergency department physician and confirmed by the researchers reviewing the charts. Only patients who underwent upper endoscopy during the index admission were included. Patients were included in the final analysis if they had Ca++ measured on admission (<1 h from admission). We excluded the following:

  • 1.

    Patients transferred from other hospitals;

  • 2.

    Patients treated with anticoagulants at home;

  • 3.

    Patients suffering from esophageal varices or chronic liver disease, which may cause coagulation abnormalities;

  • 4.

    Patients with hyperparathyroidism or hypercalcemia (Ca++ >1.31 mmol/L);

  • 5.

    Patients for whom Ca++ was measured after one or more blood product transfusions, as citrate may cause hypocalcemia.

During the study period, all the patients with suspected UGIB were treated with high-dose parenteral proton pump inhibitors before endoscopy. The duration of the treatment after endoscopy was determined by the endoscopic findings. All the other diagnostic and treatment decisions were at the discretion of the treating physician. The modified Glasgow-Blatchford score (mGBS) was calculated for each patient. This score includes heart rate, systolic blood pressure, hemoglobin, and blood urea nitrogen and was shown to predict the need for clinical interventions, with the same accuracy as the full GBS [13].

The primary outcome of this study was a composite endpoint of the need for therapeutic intervention (transfusion of two or more PCs, endoscopic intervention, urgent angiography, or surgery). Secondary outcomes included the components of the primary outcome, length of hospital stay, need for intensive care unit admission, and mortality. We selected variables as potential risk factors for severe NV-UGIB based on literature review and clinical plausibility [14‒16].

In our hospital, Ca++ is included in the results of either venous or arterial blood gas analyses (there is a good correlation between venous and arterial values) [17]. During the study period, all the blood gas analyses were performed using GEM 3500 blood gas analyzer (Instrumentation Laboratories, MA, USA). The analyzer reports Ca++ values normalized to a pH of 7.4. As alkalosis decreases protein binding, it causes a decrease of 0.05 mmol/L in the Ca++ level for each 0.1 increase in pH [18]. Therefore, we report Ca++ corrected to the actual pH of the patients. The normal range of Ca++ in RHCC is 1.15–1.3 mmol/L.

Statistical Analysis

Patients’ characteristics were summarized with descriptive statistics. Multivariate forward stepwise logistic regression was performed to assess the relationship between patient characteristics, risk factors, and outcomes. Variables were selected as candidates for the multivariate analysis based on the level of significance of the bivariate association (p < 0.1). We used all the available data from our databases within the study time frame. Missing data were handled using list-wise deletion. Model discrimination was measured using the receiver operating characteristic-derived area under the curve (AUC). AUCs were compared using the DeLong et al. method. A p value <0.05 was considered statistically significant. Propensity score matching was performed to adjust the imbalances of covariates (specifically the prevalence of congestive heart failure [CHF], chronic kidney disease [CKD], and serum creatinine levels) between the groups. Data analysis was conducted with Statistical Package for the Social Sciences, version 25.0 (IBM SPSS Statistics for Windows, version 25.0, Armonk, NY, USA: IBM Corp.) and MedCalc for Windows, version 15.0 (MedCalc Software, Ostend, Belgium).

After applying inclusion and exclusion criteria, 434 patients were included in the final analysis. Hypocalcemia (Ca++ <1.15 mmol/L) was recorded in 148 (34.1%) patients (Fig. 1). Most of these patients had mild to moderate hypocalcemia, and only 23/148 (15.5%) had Ca++ <1 mmol/L.

Fig. 1.

Study flow chart. ED, emergency department; NV-UGIB, nonvariceal upper gastrointestinal bleeding.

Fig. 1.

Study flow chart. ED, emergency department; NV-UGIB, nonvariceal upper gastrointestinal bleeding.

Close modal

The clinical and laboratory characteristics of the entire cohort are presented in Table 1. The median age of the patients in our cohort was 67.9 years (IQR 51.3–79.3), and 69.6% (302) were male. A high rate of comorbidities was apparent. The median mGBS was 7.0 (IQR 4.0–10.0). Patients in the normal and low Ca++ groups were comparable regarding age, gender, and prevalence of comorbidities, except for CHF and CKD, which were more prevalent in the low Ca++ group (21.6% vs. 9.1%, p < 0.001 and 15.5% vs. 9.4%, p = 0.06). Creatinine values were significantly higher in the low Ca++ group (1.24 [IQR 0.9–1.91] vs. 0.97 [IQR 0.8–1.33], p < 0.001). In addition, there were small but statistically significant differences in pH, lactate, and PTT values between the groups (Table 1).

Table 1.

Clinical and laboratory characteristics of the entire study cohort and patients with normal and low Ca++

Total (n = 434)Ca++ ≥1.15 mmol/L (n = 286)Ca++ <1.15 mmol/L (n = 148)p value
Age, years (IQR) 67.9 (51.3–79.3) 66.9 (50.4–78.1) 67.4 (53.8–80.1) 0.25 
Male gender, n (%) 302 (69.6) 200 (69.9) 102 (68.9) 0.83 
Comorbidities 
 Ischemic heart disease, n (%) 123 (28.3) 80 (28.0) 43 (29.1) 0.81 
 Diabetes mellitus, n (%) 132 (30.4) 84 (29.4) 48 (32.4) 0.51 
 CKD, n (%) 50 (11.5) 27 (9.4) 23 (15.5) 0.06 
 Chronic obstructive lung disease, n (%) 35 (8.1) 21 (7.3) 14 (9.5) 0.44 
 Hypertension, n (%) 233 (53.7) 147 (51.4) 86 (58.1) 0.18 
 CHF, n (%) 58 (13.4) 26 (9.1) 32 (21.6) <0.001 
 Cerebrovascular disease, n (%) 84 (19.4) 48 (16.8) 36 (24.3) 0.06 
 Antiplatelet treatment, n (%) 112 (25.8) 68 (23.8) 44 (29.7) 0.18 
Vital signs at admission 
 Arterial oxygen saturation, % (IQR) 98 (96–99) 98 (96–99) 98 (95–99) 0.19 
 Heart rate, bpm (IQR) 90 (80–105) 90 (80–105) 90 (80–104) 0.94 
 Systolic blood pressure, mm Hg (IQR) 129 (113–142) 128 (114–140) 130 (109–148) 0.45 
Laboratory at admission 
 Hemoglobin, g/dL (IQR) 10.6 (8–12.7) 10.8 (8–12.7) 10.25 (7.7–12.85) 0.51 
 Platelets count, μL (IQR) 250 (192–313) 249 (187–311) 254 (197–327) 0.42 
 Creatinine, mg/dL (IQR) 1.03 (0.81–1.5) 0.97 (0.8–1.33) 1.24 (0.9–1.91) <0.001 
 Blood urea nitrogen, mg/dL (IQR) 30.0 (20–45) 28.0 (19–42.5) 31.8 (21.9–50.9) 0.05 
 Lactate, mmol/L (IQR) 2.0 (1.3–3.2) 1.9 (1.3–2.9) 2.3 (1.5–4.1) 0.02 
 pH (IQR) 7.39 (7.34–7.42) 7.38 (7.34–7.41) 7.4 (7.36–7.44) <0.001 
 Prothrombin time, s (IQR) 11.6 (10.9–12.6) 11.6 (10.9–12.53) 11.5 (10.8–12.6) 0.48 
 Partial thromboplastin time, s (IQR) 25.8 (23.5–29.7) 25.2 (23.2–28.3) 27.85 (23.85–32.45) <0.001 
 Ca++, mmol/L (IQR) 1.18 (1.12–1.21) 1.2 (1.18–1.23) 1.1 (1.04–1.13) <0.001 
mGBS (IQR)a 7.0 (4.0–10.0) 7.0 (4.0–10.0) 7.0 (3.0–10.0) 0.15 
Endoscopy within 24 h from admission, n (%) 227 (52.3) 147 (51.4) 80 (54.1) 0.67 
Endoscopic findings and therapy 
 Normal endoscopy, n (%) 50 (11.5) 37 (12.9) 13 (8.8) 0.26 
 Esophagitis, gastritis, or duodenitis, n (%) 113 (26) 77 (26.9) 36 (24.3) 0.64 
 Peptic ulcer disease, n (%) 177 (40.8) 117 (40.9) 60 (40.5) 0.98 
  Forrest Ia, Ib, and IIa, n (%) 42 (23.7) 26 (22.2) 16 (26.7) 0.64 
  Forrest IIb, IIc, and III, n (%) 135 (76.3) 91 (77.8) 44 (73.3) 0.64 
 Mallory-Weiss syndrome, n (%) 20 (4.6) 12 (4.2) 8 (5.4) 0.74 
 Angioectasia (including Dieulafoy), n (%) 22 (5.1) 14 (4.9) 8 (5.4) 0.99 
 Polyps and neoplasms, n (%) 28 (6.5) 18 (6.3) 10 (6.8) 0.98 
 Cameron or marginal ulcer, n (%) 16 (3.7) 9 (3.1) 7 (4.7) 0.57 
 Other findings, n (%) 8 (1.8) 2 (0.7) 6 (4.1) 0.04 
Endoscopic therapy 
 No intervention, n (%) 352 (81.1) 241 (84.3) 111 (75.0) 0.03 
 Endoscopic intervention, n (%) 82 (18.9) 45 (15.7) 37 (25.0) 0.03 
 Epinephrine injection, n (%) 8 (1.8) 2 (0.7) 6 (4.1) 0.04 
 Clip application, n (%) 11 (2.5) 4 (1.4) 7 (4.7) 0.08 
 Argon plasma coagulation, n (%) 12 (2.8) 10 (3.5) 2 (1.4) 0.33 
 Electrocoagulation multipolar (BICAP), n (%) 2 (0.5) 1 (0.3) 1 (0.7) 0.79 
 Combined therapy (epinephrine injection and clip application), n (%) 48 (11.1) 28 (9.8) 20 (13.5) 0.31 
Clinical outcome 
 Need for therapeutic intervention, n (%) 102 (23.5) 53 (18.5) 49 (48.0) <0.001 
 Packed cells transfusion, n (%) 203 (46.8) 125 (43.7) 78 (52.7) 0.08 
 Two or more packed cells transfusion, n (%) 11 (2.5) 1 (0.3) 10 (6.8) <0.001 
 Need for emergent surgery or angiography, n (%) 21 (4.8) 10 (3.5) 11 (7.4) 0.11 
 Need for “second-look” gastroscopy, n (%) 50 (11.5) 31 (10.8) 19 (12.8) 0.65 
 Need for additional gastroscopy within 30 days, n (%)* 16 (4) 8 (2.9) 8 (6.1) 0.21 
 Intensive care unit admission, n (%) 62 (14.3) 42 (14.7) 20 (13.5) 0.85 
 Length of hospital stay, days (IQR) 4.0 (3.0–7.0) 4.0 (3.0–6.0) 5.0 (3.0–8.0) 0.01 
 All-cause in-hospital mortality, n (%) 76 (17.5) 43 (15) 33 (22.3) 0.08 
Total (n = 434)Ca++ ≥1.15 mmol/L (n = 286)Ca++ <1.15 mmol/L (n = 148)p value
Age, years (IQR) 67.9 (51.3–79.3) 66.9 (50.4–78.1) 67.4 (53.8–80.1) 0.25 
Male gender, n (%) 302 (69.6) 200 (69.9) 102 (68.9) 0.83 
Comorbidities 
 Ischemic heart disease, n (%) 123 (28.3) 80 (28.0) 43 (29.1) 0.81 
 Diabetes mellitus, n (%) 132 (30.4) 84 (29.4) 48 (32.4) 0.51 
 CKD, n (%) 50 (11.5) 27 (9.4) 23 (15.5) 0.06 
 Chronic obstructive lung disease, n (%) 35 (8.1) 21 (7.3) 14 (9.5) 0.44 
 Hypertension, n (%) 233 (53.7) 147 (51.4) 86 (58.1) 0.18 
 CHF, n (%) 58 (13.4) 26 (9.1) 32 (21.6) <0.001 
 Cerebrovascular disease, n (%) 84 (19.4) 48 (16.8) 36 (24.3) 0.06 
 Antiplatelet treatment, n (%) 112 (25.8) 68 (23.8) 44 (29.7) 0.18 
Vital signs at admission 
 Arterial oxygen saturation, % (IQR) 98 (96–99) 98 (96–99) 98 (95–99) 0.19 
 Heart rate, bpm (IQR) 90 (80–105) 90 (80–105) 90 (80–104) 0.94 
 Systolic blood pressure, mm Hg (IQR) 129 (113–142) 128 (114–140) 130 (109–148) 0.45 
Laboratory at admission 
 Hemoglobin, g/dL (IQR) 10.6 (8–12.7) 10.8 (8–12.7) 10.25 (7.7–12.85) 0.51 
 Platelets count, μL (IQR) 250 (192–313) 249 (187–311) 254 (197–327) 0.42 
 Creatinine, mg/dL (IQR) 1.03 (0.81–1.5) 0.97 (0.8–1.33) 1.24 (0.9–1.91) <0.001 
 Blood urea nitrogen, mg/dL (IQR) 30.0 (20–45) 28.0 (19–42.5) 31.8 (21.9–50.9) 0.05 
 Lactate, mmol/L (IQR) 2.0 (1.3–3.2) 1.9 (1.3–2.9) 2.3 (1.5–4.1) 0.02 
 pH (IQR) 7.39 (7.34–7.42) 7.38 (7.34–7.41) 7.4 (7.36–7.44) <0.001 
 Prothrombin time, s (IQR) 11.6 (10.9–12.6) 11.6 (10.9–12.53) 11.5 (10.8–12.6) 0.48 
 Partial thromboplastin time, s (IQR) 25.8 (23.5–29.7) 25.2 (23.2–28.3) 27.85 (23.85–32.45) <0.001 
 Ca++, mmol/L (IQR) 1.18 (1.12–1.21) 1.2 (1.18–1.23) 1.1 (1.04–1.13) <0.001 
mGBS (IQR)a 7.0 (4.0–10.0) 7.0 (4.0–10.0) 7.0 (3.0–10.0) 0.15 
Endoscopy within 24 h from admission, n (%) 227 (52.3) 147 (51.4) 80 (54.1) 0.67 
Endoscopic findings and therapy 
 Normal endoscopy, n (%) 50 (11.5) 37 (12.9) 13 (8.8) 0.26 
 Esophagitis, gastritis, or duodenitis, n (%) 113 (26) 77 (26.9) 36 (24.3) 0.64 
 Peptic ulcer disease, n (%) 177 (40.8) 117 (40.9) 60 (40.5) 0.98 
  Forrest Ia, Ib, and IIa, n (%) 42 (23.7) 26 (22.2) 16 (26.7) 0.64 
  Forrest IIb, IIc, and III, n (%) 135 (76.3) 91 (77.8) 44 (73.3) 0.64 
 Mallory-Weiss syndrome, n (%) 20 (4.6) 12 (4.2) 8 (5.4) 0.74 
 Angioectasia (including Dieulafoy), n (%) 22 (5.1) 14 (4.9) 8 (5.4) 0.99 
 Polyps and neoplasms, n (%) 28 (6.5) 18 (6.3) 10 (6.8) 0.98 
 Cameron or marginal ulcer, n (%) 16 (3.7) 9 (3.1) 7 (4.7) 0.57 
 Other findings, n (%) 8 (1.8) 2 (0.7) 6 (4.1) 0.04 
Endoscopic therapy 
 No intervention, n (%) 352 (81.1) 241 (84.3) 111 (75.0) 0.03 
 Endoscopic intervention, n (%) 82 (18.9) 45 (15.7) 37 (25.0) 0.03 
 Epinephrine injection, n (%) 8 (1.8) 2 (0.7) 6 (4.1) 0.04 
 Clip application, n (%) 11 (2.5) 4 (1.4) 7 (4.7) 0.08 
 Argon plasma coagulation, n (%) 12 (2.8) 10 (3.5) 2 (1.4) 0.33 
 Electrocoagulation multipolar (BICAP), n (%) 2 (0.5) 1 (0.3) 1 (0.7) 0.79 
 Combined therapy (epinephrine injection and clip application), n (%) 48 (11.1) 28 (9.8) 20 (13.5) 0.31 
Clinical outcome 
 Need for therapeutic intervention, n (%) 102 (23.5) 53 (18.5) 49 (48.0) <0.001 
 Packed cells transfusion, n (%) 203 (46.8) 125 (43.7) 78 (52.7) 0.08 
 Two or more packed cells transfusion, n (%) 11 (2.5) 1 (0.3) 10 (6.8) <0.001 
 Need for emergent surgery or angiography, n (%) 21 (4.8) 10 (3.5) 11 (7.4) 0.11 
 Need for “second-look” gastroscopy, n (%) 50 (11.5) 31 (10.8) 19 (12.8) 0.65 
 Need for additional gastroscopy within 30 days, n (%)* 16 (4) 8 (2.9) 8 (6.1) 0.21 
 Intensive care unit admission, n (%) 62 (14.3) 42 (14.7) 20 (13.5) 0.85 
 Length of hospital stay, days (IQR) 4.0 (3.0–7.0) 4.0 (3.0–6.0) 5.0 (3.0–8.0) 0.01 
 All-cause in-hospital mortality, n (%) 76 (17.5) 43 (15) 33 (22.3) 0.08 

Statistically significant (p < 0.05) values are in bold and italicized.

IQR, interquartile range.

*405 patients survived to additional gastroscopy or 30 days.

amGBS incorporates hemoglobin, blood urea nitrogen, systolic blood pressure, and pulse rate.

Primary Outcome

A total of 102 patients (23.5%) underwent therapeutic intervention during their hospitalization (transfusion of ≥2 PCs, endoscopic intervention, emergent angiography, or surgery). Patients with low Ca++ required more therapeutic interventions (48.0% vs. 18.5%, p < 0.001). The results of the bivariate and multivariate analyses for the correlation between patients’ variables and primary outcome are summarized in Table 2. After adjustment for multiple covariates, hypocalcemia was independently associated with the composite endpoint of the need for therapeutic intervention with an aOR of 1.62 (95% confidence interval [CI] 1.22–2.14, p < 0.001) for each 0.1 mmol/L decrease in Ca++. The only other variables that were independently associated with the primary outcome were mGBS (aOR 1.20 [95% CI 1.12–1.29], p < 0.001) and platelet count (aOR 1.01 [95% CI 1.00–1.01], p < 0.001). The AUCs of mGBS and Ca++ for prediction of therapeutic intervention were 0.68 (95% CI 0.63–0.72, p < 0.001) and 0.61 (95% CI 0.56–0.65, p = 0.001), respectively. The addition of Ca++ to mGBS slightly, but significantly, improved its accuracy, leading to AUC of 0.72 (95% CI 0.67–0.76) (p = 0.02).

Table 2.

Results of the bivariate and multivariate analyses of the correlation between need for clinical intervention (packed cells transfusion, need for endoscopic, surgical, or angiographic intervention) and demographic factors, comorbidities, vital signs, and laboratory test results at admission

Odds ratio (95% CI)p value
Age, years 1.01 (1.00–1.03) 0.03 
Male gender 1.13 (0.69–1.85) 0.62 
Comorbidities 
 Ischemic heart disease 1.54 (0.96–2.47) 0.08 
 Diabetes mellitus 1.06 (0.66–1.71) 0.81 
 CKD 1.31 (0.68–2.54) 0.43 
 Chronic obstructive lung disease 1.55 (0.73–1.31) 0.26 
 Hypertension 0.72 (0.46–1.13) 0.16 
 CHF 2.07 (1.15–3.73) 0.02 
 Cerebrovascular disease 1.20 (0.69–2.07) 0.52 
 Antiplatelet treatment 1.63 (1.00–2.64) 0.05 
Vital signs at admission 
 Arterial oxygen saturation, (%) 1.01 (0.94–1.08) 0.88 
 Heart rate, bpm 1.00 (0.99–1.02) 0.69 
 Systolic blood pressure, mm Hg 0.98 (0.97–0.99) <0.001 
Laboratory at admission 
 Hemoglobin, g/dL 0.81 (0.75–0.88) <0.001 
 Platelets count, μL 1.004 (1.002–1.001) <0.001 
 Creatinine, mg/dL 1.09 (0.95–1.25) 0.25 
 Blood urea nitrogen, mg/dL 1.01 (1.01–1.02) <0.001 
 Lactate, mmol/L 1.09 (1.01–1.17) 0.02 
 pH 0.41 (0.07–26.61) 0.82 
 Prothrombin time, s 1.04 (0.98–1.10) 0.18 
 Partial thromboplastin time, s 0.99 (0.96–1.02) 0.65 
 Ca++, mmol/L* 1.4 (1.1–1.78) <0.001 
mGBS (IQR)a 1.61 (1.24–2.10) <0.001 
Multivariate analysis 
mGBS (IQR)a 1.20 (1.12–1.29) <0.001 
Platelets count, μL 1.01 (1.00–1.01) <0.001 
Ca++, mmol/L* 1.62 (1.22–2.14) <0.001 
Area under the receiver operating characteristic curve 0.73 (0.69–0.78) <0.001 
Odds ratio (95% CI)p value
Age, years 1.01 (1.00–1.03) 0.03 
Male gender 1.13 (0.69–1.85) 0.62 
Comorbidities 
 Ischemic heart disease 1.54 (0.96–2.47) 0.08 
 Diabetes mellitus 1.06 (0.66–1.71) 0.81 
 CKD 1.31 (0.68–2.54) 0.43 
 Chronic obstructive lung disease 1.55 (0.73–1.31) 0.26 
 Hypertension 0.72 (0.46–1.13) 0.16 
 CHF 2.07 (1.15–3.73) 0.02 
 Cerebrovascular disease 1.20 (0.69–2.07) 0.52 
 Antiplatelet treatment 1.63 (1.00–2.64) 0.05 
Vital signs at admission 
 Arterial oxygen saturation, (%) 1.01 (0.94–1.08) 0.88 
 Heart rate, bpm 1.00 (0.99–1.02) 0.69 
 Systolic blood pressure, mm Hg 0.98 (0.97–0.99) <0.001 
Laboratory at admission 
 Hemoglobin, g/dL 0.81 (0.75–0.88) <0.001 
 Platelets count, μL 1.004 (1.002–1.001) <0.001 
 Creatinine, mg/dL 1.09 (0.95–1.25) 0.25 
 Blood urea nitrogen, mg/dL 1.01 (1.01–1.02) <0.001 
 Lactate, mmol/L 1.09 (1.01–1.17) 0.02 
 pH 0.41 (0.07–26.61) 0.82 
 Prothrombin time, s 1.04 (0.98–1.10) 0.18 
 Partial thromboplastin time, s 0.99 (0.96–1.02) 0.65 
 Ca++, mmol/L* 1.4 (1.1–1.78) <0.001 
mGBS (IQR)a 1.61 (1.24–2.10) <0.001 
Multivariate analysis 
mGBS (IQR)a 1.20 (1.12–1.29) <0.001 
Platelets count, μL 1.01 (1.00–1.01) <0.001 
Ca++, mmol/L* 1.62 (1.22–2.14) <0.001 
Area under the receiver operating characteristic curve 0.73 (0.69–0.78) <0.001 

Statistically significant (p < 0.05) values are in bold and italicized.

CI, confidence interval.

*Odds ratios for Ca++ are presented as OR (95% CI) for each 0.1 mmol/L decrease in Ca++.

amGBS incorporates hemoglobin, blood urea nitrogen, systolic blood pressure, and pulse rate.

To reduce a possible bias caused by covariation between Ca++ levels, CHF, and CKD, propensity score matching was performed based on these variables (prevalence of CHF and CKD and serum creatinine level). After the incorporation of the propensity score in a 1:1 ratio (109 patients in each group), the results did not change significantly (online suppl. Table S1, S2; for all online suppl. material, see https://doi.org/10.1159/000534522).

Secondary Outcomes

Endoscopy within 24 h from admission was performed in 227 out of 434 patients (52.3%). Endoscopic findings, including high-risk stigmata (Forrest Ia – IIa ulcers), were comparable across the groups. However, patients with low calcium on admission required more endoscopic interventions for control of bleeding (25.0% vs. 15.7%, p = 0.03), specifically more epinephrine injections (4.1% vs. 0.7%, p = 0.03) and a trend in need for hemostatic clip application (4.7% vs. 1.4%, p = 0.08). In addition, patients in the low-calcium group had a higher rate of multiple PC transfusions (6.8% vs. 0.3%, p < 0.001) and longer hospital stay (5.0 days [IQR 3.0–8.0] vs. 4.0 days [IQR 3.0–6.0], p = 0.01). Patients with low Ca++ required more emergent surgery or angiography (7.4% vs. 3.5%, p = 0.11), “second-look” endoscopy (6.1% vs. 2.9%, p = 0.21), and had higher all-cause in-hospital mortality (22.3% vs. 15.0%, p = 0.08). However, these differences were not statistically significant. The intensive care unit admission rate was not different between the groups (14.7% vs. 13.5%, p = 0.85). The differences in clinical outcomes between the groups are summarized in Figure 2.

Fig. 2.

Compression of clinical outcomes rates between patients with admission Ca++ ≥1.15 mmol/L and Ca++ <1.15 mmol/L. *, statistically significant. PCs, packed red blood cells.

Fig. 2.

Compression of clinical outcomes rates between patients with admission Ca++ ≥1.15 mmol/L and Ca++ <1.15 mmol/L. *, statistically significant. PCs, packed red blood cells.

Close modal

A high rate of hypocalcemia and its association with worse clinical outcomes were first described in severe trauma patients [7]. A concentration-dependent association between Ca++ level and in vitro clot strength, as well as the development of coagulopathy, was also demonstrated among trauma patients [19, 20]. Similar findings were also found in other bleeding patients, such as women suffering from postpartum hemorrhage and patients with intracranial hemorrhage [8‒10].

Our study shows that among high-risk patients hospitalized due to acute NV-UGIB, low Ca++ levels measured within 1 h of admission and before any intervention were significantly and independently associated with the need for therapeutic intervention. Ca++ is a simple, cheap, and readily available laboratory variable that can potentially improve the triage and facilitate clinical decisions among patients admitted with NV-GIB. Although Ca++ can be used by itself as a risk stratification tool, its combination with other physiological parameters or scores (such as mGBS) seems more promising.

During hemorrhage, hypocalcemia may be caused by the blood loss itself, leading to calcium loss, intracellular flux secondary to ischemia and reperfusion, as well as by impaired calcium homeostasis and increased sympathetic activity [7]. The cornerstone of treatment for hemorrhagic shock, blood transfusion, may further exacerbate hypocalcemia as all the blood products contain citrate, a calcium chelator. Citrate metabolism may be significantly impaired in shocked patients who may have reduced hepatic function secondary to hypoperfusion and hypothermia. No studies have determined whether the low Ca++ is a cause of these outcomes or just a surrogate marker and whether calcium administration has any impact on patients’ clinical course. However, Ca++ monitoring and hypocalcemia correction are recommended by numerous trauma guidelines and review articles [21, 22].

To the best of our knowledge, only one study evaluated the association between Ca++ and clinical outcomes of patients with UGIB. This study was published by our group and was based on the same cohort of patients [23]. Our previous study included 1,345 patients with NV-UGIB for whom Ca++ was measured 24 h before or after endoscopy and did not include endoscopic findings or interventions. We found hypocalcemia to be independently associated with the increased risk for multiple blood transfusions and the need for urgent surgical or angiographic interventions. However, the study did not include endoscopic interventions, the mainstay of treatment of patients with UGIB. These limitations were addressed in the current study, in which we manually reviewed all the medical files, included only patients admitted due to NV-UGIB and those with Ca++ measured on admission, and included endoscopic findings and interventions.

Our current study was not designed to evaluate whether the association between Ca++ is causative and whether hypocalcemia aggravates coagulopathy, which may worsen the bleeding or preclude hemostasis. Furthermore, we cannot conclude from the available data whether correction of hypocalcemia in patients with NV-UGIB will have any effect on their clinical course. However, considering the minimal risk and costs of calcium monitoring and hypocalcemia correction, it should be considered, as suggested by recent trauma guidelines [22].

Our study has some additional limitations. First, Ca++ was not measured systematically in all patients admitted with NV-UGIB but at the discretion of the treating physician, as part of blood gas analyses. Most probably, this test was performed only in clinically unstable patients or those with more significant hemorrhage. Therefore, our findings may not apply to all the patients with UGIB but only to the high-risk population. Second, all the treatment decisions were based on clinical judgment, making standardization challenging. We used the cutoff of two PC transfusions as an outcome to ameliorate the effect of this cofounder. Third, we did not include extremely unstable patients who were transferred directly to the operating theater without endoscopic evaluation. However, in the real world, the risk stratification of such patients is irrelevant since they require immediate intervention. Fourth, despite our efforts to minimize potential confounding factors related to Ca++ levels, it is important to acknowledge that CHF and diuretic medications, particularly furosemide, as well as renal failure, are potential contributors to hypocalcemia. Both of these conditions are associated with adverse clinical outcomes.

Among patients hospitalized due to NV-UGIB, hypocalcemia, when measured on admission, is significantly and independently associated with increased risk of adverse course and need for therapeutic interventions. The addition of Ca++ to mGBS may increase its accuracy in risk stratification of such patients and improve their triage. Further prospective trials are needed to validate this relationship and identify possible therapeutic measures.

The study was approved by the Institutional Review Board at RHCC (approval number RMB-0094-20). The need for informed consent was waived by the RHCC Institutional Review Board.

The authors have no conflicts of interest to declare.

No funding sources to be reported.

A.Ko., F.M., A.K., and D.E. contributed to conceptualization and methodology, writing – original draft preparation, and writing – review and editing; A.Ko., F.M., E.M., and D.E. contributed to data collection and to data analysis and visualization. A.Ko. and F.M. equally contributed as first authors. A.K. and D.E. equally contributed as last authors.

Additional Information

Alexander Korytny and Fares Mazzawi equally contributed as first authors.Amir Klein and Danny Epstein equally contributed as last authors.

The data presented in this study are available on request from the corresponding author. The data are not publicly available for privacy reasons. Further inquiries can be directed to the corresponding author.

1.
van Leerdam
ME
.
Epidemiology of acute upper gastrointestinal bleeding
.
Best Pract Res Clin Gastroenterol
.
2008 Apr
22
2
209
24
.
2.
Abougergi
MS
,
Travis
AC
,
Saltzman
JR
.
The in-hospital mortality rate for upper GI hemorrhage has decreased over 2 decades in the United States: a nationwide analysis
.
Gastrointest Endosc
.
2015 Apr
81
4
882
8.e1
.
3.
Kamboj
AK
,
Hoversten
P
,
Leggett
CL
.
Upper gastrointestinal bleeding: etiologies and management
.
Mayo Clin Proc
.
2019 Apr
94
4
697
703
.
4.
Stanley
AJ
,
Laine
L
.
Management of acute upper gastrointestinal bleeding
.
BMJ
.
2019
364
l536
.
5.
Gralnek
IM
,
Dumonceau
JM
,
Kuipers
EJ
,
Lanas
A
,
Sanders
DS
,
Kurien
M
.
Diagnosis and management of nonvariceal upper gastrointestinal hemorrhage: European Society of Gastrointestinal Endoscopy (ESGE) Guideline
.
Endoscopy
.
2015 Oct
47
10
a1
46
.
6.
Al-Koussa
H
,
AlZaim
I
,
El-Sabban
ME
.
Pathophysiology of coagulation and emerging roles for extracellular vesicles in coagulation cascades and disorders
.
J Clin Med
.
2022 Aug
11
16
4932
.
7.
Vasudeva
M
,
Mathew
JK
,
Groombridge
C
,
Tee
JW
,
Johnny
CS
,
Maini
A
.
Hypocalcemia in trauma patients: a systematic review
.
J Trauma Acute Care Surg
.
2021
;
90
(
2
):
396
402
.
8.
Epstein
D
,
Solomon
N
,
Korytny
A
,
Marcusohn
E
,
Freund
Y
,
Avrahami
R
.
Association between ionised calcium and severity of postpartum haemorrhage: a retrospective cohort study
.
Br J Anaesth
.
2021
;
126
(
5
):
1022
8
.
9.
Epstein
D
,
Freund
Y
,
Marcusohn
E
,
Diab
T
,
Klein
E
,
Raz
A
.
Association between ionized calcium level and neurological outcome in endovascularly treated patients with spontaneous subarachnoid hemorrhage: a retrospective cohort study
.
Neurocrit Care
.
2021 Apr
35
3
723
37
.
10.
Morotti
A
,
Charidimou
A
,
Phuah
CL
,
Jessel
MJ
,
Schwab
K
,
Ayres
AM
.
Association between serum calcium level and extent of bleeding in patients with intracerebral hemorrhage
.
JAMA Neurol
.
2016 Nov
73
11
1285
90
.
11.
Kaji
AH
,
Schriger
D
,
Green
S
.
Looking through the retrospectoscope: reducing bias in emergency medicine chart review studies
.
Ann Emerg Med
.
2014
;
64
(
3
):
292
8
.
12.
von Elm
E
,
Altman
DG
,
Egger
M
,
Pocock
SJ
,
Gøtzsche
PC
,
Vandenbroucke
JP
.
The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies
.
Lancet
.
2007
;
370
(
9596
):
1453
7
.
13.
Blatchford
O
,
Murray
WR
,
Blatchford
M
.
A risk score to predict need for treatment for upper-gastrointestinal haemorrhage
.
Lancet
.
2000 Oct
356
9238
1318
21
.
14.
Cheng
DW
,
Lu
YW
,
Teller
T
,
Sekhon
HK
,
Wu
BU
.
A modified Glasgow Blatchford Score improves risk stratification in upper gastrointestinal bleed: a prospective comparison of scoring systems
.
Aliment Pharmacol Ther
.
2012 Oct
36
8
782
9
.
15.
Kim
MS
,
Choi
J
,
Shin
WC
.
AIMS65 scoring system is comparable to Glasgow-Blatchford score or Rockall score for prediction of clinical outcomes for non-variceal upper gastrointestinal bleeding
.
BMC Gastroenterol
.
2019
;
19
(
1
):
136
.
16.
Tham
J
,
Stanley
A
.
Clinical utility of pre-endoscopy risk scores in upper gastrointestinal bleeding
.
Expert Rev Gastroenterol Hepatol
.
2019 Dec
13
12
1161
7
.
17.
Bilkovski
RN
,
Cannon
CM
,
Adhikari
S
,
Nasr
I
.
Arterial and venous ionized calcium measurements: is there a difference
.
Ann Emerg Med
.
2004
44
4
S56
.
18.
Bushinsky
DA
,
Monk
RD
.
Electrolyte quintet: calcium
.
Lancet
.
1998 Jul
352
9124
306
11
.
19.
Vasudeva
M
,
Mathew
JK
,
Fitzgerald
MC
,
Cheung
Z
,
Mitra
B
.
Hypocalcaemia and traumatic coagulopathy: an observational analysis
.
Vox Sang
.
2020
;
115
(
2
):
189
95
.
20.
Ho
KM
,
Yip
CB
.
Concentration-dependent effect of hypocalcaemia on in vitro clot strength in patients at risk of bleeding: a retrospective cohort study
.
Transfus Med
.
2016
;
26
(
1
):
57
62
.
21.
Ditzel
RM
,
Anderson
JL
,
Eisenhart
WJ
,
Rankin
CJ
,
DeFeo
DR
,
Oak
S
.
A review of transfusion- and trauma-induced hypocalcemia
.
J Trauma Acute Care Surg
.
2020 Mar
88
3
434
9
.
22.
Spahn
DR
,
Bouillon
B
,
Cerny
V
,
Duranteau
J
,
Filipescu
D
,
Hunt
BJ
.
The European guideline on management of major bleeding and coagulopathy following trauma: fifth edition
.
Crit Care
.
2019
;
23
(
1
):
98
.
23.
Korytny
A
,
Klein
A
,
Marcusohn
E
,
Freund
Y
,
Neuberger
A
,
Raz
A
.
Hypocalcemia is associated with adverse clinical course in patients with upper gastrointestinal bleeding
.
Intern Emerg Med
.
2021 Mar
16
7
1813
22
.