Ionized calcium concentration and outcome in critical illness*

Moritoki Egi, MD; Inbyung Kim, PhD; Alistair Nichol, PhD; Edward Stachowski, MD; Craig J. French, MB; Graeme K. Hart, MD; Colin Hegarty, BSc; Michael Bailey, PhD; Rinaldo Bellomo, MD
Objective: To assess the association of abnormalities of ionized calcium levels with mortality in a heterogeneous cohort of critically ill patients. Design: Retrospective, combined clinical and biochemical study. Setting: Four combined medical/surgical intensive care units. Patients: Cohort of 7,024 adult critically ill patients. Interventions: None. Measurements and Main Results: We studied 177,578 ionized calcium measurements, from 7024 patients, with a mean value of 1.11 mmol/L (ionized calcium measured every 4.5 hrs on average). The unadjusted lowest and highest ionized calcium reported during intensive care unit stay were signicantly different between intensive care unit survivors and nonsurvivors (p < .001). If hypocalcemia occurred at least once during the intensive care unit stay, the probability of intensive care unit mortality increased by 46%, 108%, and 150% for ionized calcium levels <1.15, 0.90, and 0.80 mmol/L, respectively. If hypercalcemia occurred at least once during the intensive care unit stay, the probability of intensive care unit mortality increased by 100%, 162%, and 190% for ionized calcium levels >1.25, 1.35, and 1.45 mmol/L, respectively. Similar trends were seen for hospital mortality. However, from multivariate logistic regression analysis, only an ionized calcium <0.8 mmol/L or an ionized calcium >1.4 mmol/L were independently associated with intensive care unit and hospital mortality. Conclusions: Within a broad range of values, ionized calcium concentration has no independent association with hospital or intensive care unit mortality. Only extreme abnormalities of ionized calcium concentrations are independent predictors of mortality. (Crit Care Med 2011; 39:314 321) KEY WORDS: calcium; ionized calcium; hypocalcemia; hypercalcemia; intensive care unit; critical illness; mortality

normal ionized calcium (iCa) concentration in blood is important to many fundamental physiologic regulatory mechanisms (15). However, abnormalities of iCa concentration are common in critically ill patients (6 9). The vast majority of these abnormalities are due to hypocalcemia (524), but hypercalcemia has also been reported (25).

*See also p. 406. From the Department of Anesthesiology and Resuscitology (ME), Okayama University Hospital, Okayama, Japan; Department of Intensive Care (IK, GKH, CH), Austin Hospital, Melbourne, Victoria, Australia; Australian and New Zealand Intensive Care Research Centre and Department of Epidemiology and Preventive Medicine (AN, MB, RB), Monash University, Victoria, Australia; Department of Intensive Care (ES), Westmead Hospital, Sydney, New South Wales, Australia; and Department of Intensive Care (CJF), Western Hospital, Melbourne, Victoria, Australia. Supported, in part, by the Austin Hospital Intensive Care Trust Fund, Melbourne, Australia and by an Enabling Grant from the National Health and Medical Research Council, Canberra, Australia (Grant No. 316001). The authors have not disclosed any potential conicts of interest. For information regarding this article, E-mail: rinaldo.bellomo@austin.org.au Copyright 2011 by the Society of Critical Care Medicine and Lippincott Williams & Wilkins DOI: 10.1097/CCM.0b013e3181ffe23e

Hypocalcemia has been associated with increased mortality in critically ill patients (5, 7, 10 17, 23, 24, 26). However, this relationship is derived from univariate analyses or relatively small single center studies (7, 10 17). Additionally, there are only a few small studies on the association between hypercalcemia and mortality in critically ill patients (1, 27). Thus, the independent relationship between abnormalities of iCa and outcome is currently poorly understood. Despite this limitation, correction of hypocalcemia with exogenous calcium has been advocated to prevent neurologic and cardiovascular complications (28, 29). However, the administration of exogenous calcium may have complications. For example, increased mortality was reported in models of sepsis (30). Thus, clinicians are left with uncertainty about the therapeutic implications of disorders of iCa homeostasis. Accordingly, we sought to assess the independent relationship among intensive care unit (ICU) admission (iCaAdm); maximal (iCaMax); minimal (iCaMin), and timeweighted average of (iCaTw) blood iCa concentration and ICU or hospital mortality in a large multicenter heterogeneous cohort of ICU patients.

MATERIALS AND METHODS

The data collection for this study was part of an established quality assurance activity. The data collection and the data analysis for this study were approved by the local institutional ethic committee, which waived the need for informed consent. The Austin Hospital Ethics Committee approved this investigation and its submission for publication. Study Population and Data Sources. The current study is a multicenter, retrospective, and observational investigation of 7024 ICU patients. Four hospitals, three in Melbourne and one in Sydney, Australia participated in this study. All patients admitted to the ICUs of these hospitals from February 2, 2000 to October 20, 2004 were included. Patients admitted during periods when arterial blood gas data were not electronically stored in a given hospital were excluded. Patients who required renal replacement therapy by using citratebased anticoagulation were excluded (31). Age, sex, use of mechanical ventilation, reason for ICU admission, and Acute Physiology and Chronic Health Evaluation II score (32) were obtained from the electronic data repositories of each ICU, using data that had been collected prospectively by trained data collectors. Coding for admission diagnosis was by means of a modied Acute Physiology and Chronic Health Evaluation III system used by the Australian and New Zealand Intensive Care Society (33). The Australian and New Zealand Intensive Care Society Centre for Outcome

314

Crit Care Med 2011 Vol. 39, No. 2

and Resources Evaluation independently collected information on clinical outcomes. The Australian and New Zealand Intensive Care Society Centre for Outcome and Resources Evaluation Adult Patient database is a highquality database whose features have been previously described (34). iCa Concentration Measurements. All iCa concentrations in the ICU were measured by arterial blood gas analyzer (Rapilab, Bayer, Sydney, Australia). The iCa data for this study were corrected to a pH of 7.4. The iCa data reported in this study were stored and retrieved electronically. Blood samples were collected in standard preprepared, heparinized blood gas syringes. The analyzer measured whole-blood samples at 37C. Trained nursing staff performed all blood analysis. Laboratories in the participating hospitals comply with standards of the National Association of Testing Authorities (35) and the Royal College of Pathologists of Australia (36). The characteristic of the systems are the following: mean SD: 0.009; mean coefcient of variation (SD/mean): 0.008; average bias: 0.041. iCaMax and iCaMin calcium concentrations were obtained, each indicating the highest or lowest value recorded while in ICU. In addition, to avoid the potential surveillance bias due to the increased arterial blood gas monitoring in the nonsurvivor group, an iCaTw was calculated. All measured iCa values and their corresponding sampling time were taken into account for its calculations. The iCaTw was calculated assuming a linear trend between individual measurements and giving a time value to such measurements. For example, an iCa of 1.1 mmol/L at 10 AM followed by an iCa of 1.3 at 2 PM, would yield a value of 1.2 mmol/L weighted by 4 hrs. The sum of such weighted values would then be divided by total hours of observation to deliver the iCaTw for a given individual during ICU admission. The iCaAdm was dened as the rst value measured within 24 hrs of ICU admission. As the normal range iCa for each hospital was between 1.15 and 1.25 mmol/L, mild, moderate, and severe hypocalcemia were dened as 1.15, 0.9, and 0.8 mmol/L, respectively. We also dened mild, moderate, and severe hypercalcemia as 1.25, 1.35, and 1.45 mmol/L, respectively. We calculated additional iCa indices for the assessment of uctuations in iCa. The absolute change in iCa (iCa) was calculated as the absolute difference between iCaMax and iCaMin. The percentage relative change in iCa (%iCa) was calculated according to the following formula: [(iCaMax iCaMin)/iCaMin] 100. Approach to Hypocalcemia and Hypercalcemia in Each Hospital. There was no protocol for intravenous calcium therapy to correct hypocalcemia in the four study hospitals. However, its use was reported to be typically only for severe hypocalcemia associated with bleeding. Intravenous calcium infusion was also used per protocol in patients treated with

citrate continuous renal replacement therapy (31). These patients were excluded from this study and do not appear in our analysis. Serum Phosphate, Albumin, and Creatinine Measurements. The phosphate and albumin levels were measured every morning by using a phosphomolybdate complex colorimetric technique and a bromcresol purple dye binding colorimetric technique. Statistical Analysis. The primary outcomes for this analysis were ICU and hospital mortality. Univariate analysis for comparison between groups was performed by using the chisquared test for proportions, Students t test for normally distributed outcomes, and Wilcoxons rank sum test for nonparametric data. Results are reported as means (SD) or medians (interquartile range) as appropriate. As the relationship between iCa and mortality was not linear in nature, iCa was divided into bands of 0.1 mmol/L in width and treated as a categorical variable. We dened the iCa reference range for such analyses as being between 1.1 and 1.2 mmol/L in this analysis.

Multivariate analysis was employed with all available predictors of mortality included in the models (gender, age, Acute Physiology and Chronic Health Evaluation II score, mechanical ventilation, surgical admission, and diagnosis type). To determine whether iCa effects were consistent across subgroups, all interactions between measures of iCa and other variables in the model were examined. Results from the multivariate models have been reported by using odds ratios (ORs) with 95% condence intervals (CIs). All analysis was performed by using SAS version 9.2 (SAS Institute, Cary, NC). A twosided p value of .05 was considered to be statistically signicant.

RESULTS
We studied 7,024 consecutive patients admitted to four ICUs. Their mean age was 61.5 yrs, and their mean Acute Physiology and Chronic Health Evaluation II

Table 1. Comparison of intensive care unit survivors and nonsurvivors Nonsurvivors (n 852) 478 (56.1%) 65 (15.7) 27 (8) 743 (87.2%) 227 (26.6%) Survivors (n 6172) 3727 (60.4%) 61.1 (18.1) 15.9 (6.8) 3465 (56.1%) 2805 (45.4%)

Characteristic Male sex Age Acute Physiology and Chronic Health Evaluation II score Number (%) on mechanical ventilation Surgical patients Reason for intensive care unit admission Cardiac and vascular Thoracic and respiratory Trauma Neurologic Gastrointestinal tract diseases Other Intensive care unit stay (days) iCaMin concentration (mmol/L) iCaMax concentration (mmol/L) iCaTw concentration (mmol/L) iCa concentration at admission (mmol/L) Incidence of mild hypocalcemia (iCa 1.15 mmol/L) Incidence of moderate hypocalcemia (iCa 0.9 mmol/L) Incidence of severe hypocalcemia (iCa 0.8 mmol/L) Incidence of mild hypercalcemia (iCa 1.25 mmol/L) Incidence of moderate hypercalcemia (iCa 1.35 mmol/L) Incidence of severe hypercalcemia (iCa 1.45 mmol/L) iCa (mmol/L) %iCa (%)

p .017 .001 .001 .001 .001 .001 .001 .18 .001 .001 .001 .001 .82 .001 .001 .013 .57 .001 .001 .001 .001 .001 .001 .001 .001

259 (30.4%) 161 (18.9%) 13 (1.5%) 130 (15.3%) 102 (12%) 187 (21.9%) 3 (27) 0.99 (0.14) 1.23 (0.16) 1.11 (0.11) 1.11 (0.14) 779 (91.4%) 156 (18.3%) 57 (6.7%) 296 (34.7%) 106 (12.4%) 37 (4.3%) 0.23 (0.18) 26.7 (26.9)

1320 (21.4%) 1289 (20.9%) 433 (7%) 626 (10.1%) 1434 (23.2%) 1070 (17.3%) 3 (25) 1.03 (0.11) 1.19 (0.11) 1.11 (0.08) 1.11 (0.11) 5431 (88%) 601 (9.7%) 172 (2.8%) 1299 (21%) 318 (5.2%) 95 (1.5%) 0.16 (0.13) 17.3 (19.3)

iCa, ionized calcium concentration; iCaMax, iCa maximum; iCaMin, iCa minimum; iCaTw, iCa time-weighted average; iCa, absolute change in iCa; %iCa, percentage relative change in iCa. Reported as mean (SD), median (25% quartile and 75% quartile), or n (%). The iCa was calculated as the absolute difference between iCaMax and iCaMin. The %iCa was calculated according to the following formula: (iCaMax iCaMin)/iCaMin 100.

Crit Care Med 2011 Vol. 39, No. 2

315

Table 2. Comparison of hospital survivors and nonsurvivors Nonsurvivors (n 1518) 869 (57.2%) 66.2 (15.2) 24.3 (8) 1191 (78.5%) 438 (28.9%) 389 (25.6%) 329 (21.7%) 37 (2.4%) 208 (13.7%) 240 (15.8%) 315 (20.8%) 3 (27) 1 (0.13) 1.22 (0.14) 1.11 (0.1) 1.11 (0.13) 1366 (90%) 248 (16.3%) 81 (5.3%) 487 (32.1%) 153 (10.1%) 51 (3.4%) 0.22 (0.16) 24.0 (24.6) Survivors (n 5506) 3336 (60.6%) 60.3 (18.3) 15.3 (6.6) 3017 (54.8%) 2594 (47.1%) 1190 (21.6%) 1121 (20.4%) 409 (7.4%) 548 (10%) 1296 (23.5%) 942 (17.1%) 3 (25) 1.03 (0.11) 1.19 (0.11) 1.11 (0.08) 1.11 (0.11) 4844 (88%) 509 (9.2%) 148 (2.7%) 1108 (20.1%) 271 (4.9%) 81 (1.5%) 0.16 (0.13) 16.9 (19.0)

Characteristic Male sex Age Acute Physiology and Chronic Health Evaluation II score Number (%) on mechanical ventilation Surgical patients Reason for intensive care unit admission Cardiac and vascular Thoracic and respiratory Trauma Neurologic Gastrointestinal tract diseases Other Intensive care unit stay (days) iCaMin concentration (mmol/L) iCaMax concentration (mmol/L) iCaTw concentration (mmol/L) iCa concentration at admission (mmol/L) Incidence of mild hypocalcemia (iCa 1.15 mmol/L) Incidence of moderate hypocalcemia (iCa 0.9 mmol/L) Incidence of severe hypocalcemia (iCa 0.8 mmol/L) Incidence of mild hypercalcemia (iCa 1.25 mmol/L) Incidence of moderate hypercalcemia (iCa 1.35 mmol/L) Incidence of severe hypercalcemia (iCa 1.45 mmol/L) iCa (mmol/L) %iCa (%)

p .019 .001 .001 .001 .001 .001 .001 .26 .001 .001 .001 .001 .001 .001 .001 .51 .98 .001 .001 .001 .001 .001 .001 .001 .001

iCa, ionized calcium concentration; iCaMax, iCa maximum; iCaMin, iCa minimum; iCaTw, iCa time-weighted average; iCa, absolute change in iCa; %iCa, percentage relative change in iCa. Reported as mean (SD), median (25% quartile and 75% quartile), or n (%). The iCa was calculated as the absolute difference between iCaMax and iCaMin. The %iCa was calculated according to the following formula: (iCaMax iCaMin)/iCaMin 100.

Figure 1. Intensive care unit (ICU) and hospital mortality according to the lowest ionized calcium (iCa) concentration during ICU stay. The bars indicate the ICU (black) and hospital (white) mortality of each 0.1-mmol/L concentration band as determined by the lowest ionized calcium concentration. The error bars indicate the SE for mortality. The number of patients in each band is indicated at the bottom. *p .05 compared with the reference range (1.10 1.20 mmol/L).

score was 17. Of these patients, 43.2% were surgical cases. ICU and hospital mortality were 12.1% and 21.6%, respectively. There were 177,578 iCa measurements with a mean value of 1.11 mmol/L, a minimum of 0.29 mmol/L, and a maximum of 3.32 mmol/L (iCa measured every 4.5 hrs on average). We detected at least one episode of mild hypocalcemia (1.15 mmol/L) in 88.4% of patients, moderate hypocalcemia (0.9 mmol/L) in 10.8%, and severe hypocalcemia (0.8 mmol/L) in 3.3%. Similarly, we detected mild hypercalcemia (1.25 mmol/L) in 22.7% of patients, moderate hypercalcemia (1.35 mmol/L) in 6.7%, and severe hypercalcemia (1.45 mmol/L) in 2.0%. When considering patients who had both hypo- and hypercalcemia, 16.9% of patients were found to have both mild hypocalcemia and mild hypercalcemia, while only 1.0% of patients experienced moderate hypocalcemia and moderate hypercalcemia, and only 0.24% were found to have had both severe hypocalcemia and severe hypercalcemia. Clinical characteristics and the distribution of patients in predened iCa bands among ICU survivors and nonsurvivors are shown in Table 1. The iCaMin and iCaMax concentrations were signicantly different between ICU survivors and nonsurvivors (p .001). The incidence of hypocalcemia and hypercalcemia was signicantly higher in nonsurvivors than survivors. The iCaTw concentrations were also signicantly different; however, the strength of the difference was small. There were no statistically signicant differences in iCaAdm in nonsurvivor compared with ICU survivors. Table 2 describes the demographics and characteristics of the hospital survivors and nonsurvivors, showing similar results to those reported for ICU survival. Hypocalcemia and Hypercalcemia in the ICU. If hypocalcemia occurred at least once during ICU stay, the probability of ICU mortality was increased by 46% (OR 1.46 [95% CI 1.131.87]; p .003) for iCa levels 1.15; 108% (OR 2.08 [1.71 2.52]; p .0001) for iCa levels 0.90, and 150% (OR 2.50 [1.84 3.41]; p .0001) for iCa levels 0.80 when compared with patients without such abnormalities. Similar trends were seen for hospital mortality. Severe hypocalcemia was signicantly more likely in male, older, and high severity patients with cardiovascular or respiratory failure. We identied 42 patients who experienced extreme hypocalcemia (iCa 0.7 mmol/L) and conCrit Care Med 2011 Vol. 39, No. 2

316

Figure 2. Intensive care unit (ICU) and hospital mortality according to highest ionized calcium (iCa) concentration during ICU stay. The bars indicate the ICU (black) and hospital (white) mortality of each 0.1-mmol/L concentration band as determined by the highest ionized calcium concentration. The error bars indicated the SE for mortality. The number of patients in each band is indicated at the bottom. *p .05 compared with the reference range (1.10 1.20 mmol/L).

Figure 3. Intensive care unit (ICU) and hospital mortality according to time-weighted average of ionized calcium (iCa) concentration during ICU stay. The bars indicate the ICU (black) and hospital (white) mortality of each 0.1-mmol/L concentration band as determined by the time-weighted average of calcium concentration. The error bars indicate the SE for mortality. The number of patients in each band is indicated at the bottom. *p .05 compared with the reference range (1.10 1.20 mmol/L).

rmed that only nine of these patients received intravenous calcium. If hypercalcemia occurred at least once during ICU stay, the probability of ICU mortality increased by 100% for iCa levels 1.25 (OR 2.00 [1.712.33]; p .0001); 162% for iCa levels 1.35 (OR 2.62 [2.07 3.30]; p .0001); and 190% greater for iCa levels 1.45 mmol/L (OR 2.90 [1.97 4.28]; p .0001) compared with the cohort without such abnormalities. Similar trends were seen for hospital mortality. Severe hypercalcemia was signicantly more likely in
Crit Care Med 2011 Vol. 39, No. 2

male, older, and higher severity patients. It was also signicantly more common in postoperative patients. Figures 1 4 show the relationship between each iCa index (iCaMin, iCaMax, iCaTw, and iCaAdm) and mortality according to each 0.1-mmol/L band. They all demonstrate that ICU and hospital mortality increased at both extremes but that there was only a small change in mortality for less extreme abnormalities. Time Course of the iCa in ICU. Among the 229 patients who experienced severe

hypocalcemia (0.8), 61 (26.6%) also experienced hypercalcemia (1.25) while in ICU. The mortality of the 61 patients who experienced both conditions was 21.3% compared with 26.2% in those who only experienced severe hypocalcemia (p .45). Among the 132 patients who experienced severe hypercalcemia (1.45), 86 (65.1%) also experienced hypocalcemia (1.15). Their mortality was 26.7%, compared with 30.4% in the 46 patients who only experienced severe isolated hypercalcemia (p .65). Fluctuations of iCa in ICU. ICU nonsurvivors had a signicantly higher iCa and %iCa than ICU survivors (iCa, 0.23 vs. 0.16 mmol/L, p .001; %iCa, 26.7% vs. 17.3%, p .001, respectively) (Table 1). There was a similar difference between hospital nonsurvivors and survivors (iCa, 0.22 vs. 0.16 mmol/L, p .001; %iCa, 24.0% vs. 16.9%, p .001, respectively) (Table 2). Relationship of iCa Levels with pH, Serum Phosphate, Albumin-Ionized Magnesium Levels, and Kidney Function. There were 3,219 patients (45.8%) for whom we were able to obtain daily phosphate, albumin, and ionized magnesium levels. The highest serum creatinine concentration during rst 24 hrs after ICU admission was also obtained among 4,127 patients (58.8%). There was a signicant, but weak correlation between iCa and pH, phosphate, albumin, and ionized magnesium levels (pH: r .10, p .0001; phosphate: r .08, p .0001; albumin: r .14, p .0001; ionized magnesium: r .62, p .0001). There was a signicant, but weak, correlation between iCaAdm and the highest creatinine level during the rst 24 hrs of ICU admission (r .079, p .0001). Multivariate Analysis. After adjustment for other predictors of mortality, there was a signicant relationship between iCaMin and ICU and hospital mortality (Table 3). This trend was driven by the mortality of patients with very low iCaMin values (0.8 mmol/L for ICU mortality and 0.9 mmol/L for hospital mortality). There was a signicant relationship between hospital and ICU mortality and iCaMax (Table 4). This result was again primarily dependent on patients with very high iCa levels ( 1.4 mmol/L). The iCaTw concentration was independently related to mortality (ICU mortality p .001 and hospital mortality p .0003) (Table 5). This result was mostly dependent on patients with iCaTw values 1.0 mmol/L. Finally, iCaAdm was not signicantly related to mortality. As there were no signicant
317

interactions between iCa indices and other predictors of mortality, there was no evidence to suggest that their relationship differed across different subgroups.

DISCUSSION
We conducted a four-center retrospective, observational study of 7,000 patients and 170,000 iCa measurements to investigate the association between iCa concentrations and mortality in critically ill patients. We found that mild hypocalcemia

was more common than mild hypercalcemia and that it occurred in close to 90% of patients studied. However, severe hypocalcemia and hypercalcemia were uncommon, occurring in only 2% to 3% of patients. On univariate analysis, hypo- and hypercalcemia were associated with higher mortality; however, after adjusting for other risk factors, the relationship between iCa concentrations and ICU and hospital death was limited in extent and strength and mostly dependent on the effect of extremely high

Figure 4. Intensive care unit (ICU) and hospital mortality according to ionized calcium (iCa) concentration at admission in ICU. The bars indicate the ICU (black) and hospital (white) mortality of each 0.1-mmol/L concentration band as determined by the ionized calcium concentration at admission. The error bars indicated the SE for mortality. The number of patients in each band is indicated at the bottom. *p .05 compared with the reference range (1.10 1.20 mmol/L).

(1.4 mmol/L) or extremely low (0.9 mmol/L) values. Before this study, most investigations of hypocalcemia and its association with outcome in ICU patients had been small and single center in nature. In one of the two largest studies, Choi and Hwang (24) studied 255 trauma patients in a single center. They dened mild and severe hypocalcemia as iCaAdm level between 0.89 and 1.14 mmol/L and 0.89 mmol/L, respectively. By these denitions, mild and severe admission hypocalcemia occurred in 76.5% and 20.8% of patients. Nonsurvivors had signicantly lower iCaAdm compared with survivors. On multivariate analysis, an iCa Adm level 0.88 mmol/L was an independent predictor of mortality. Ha stbacka and Pettila (5) investigated 941 patients in a single center retrospective observational study. They dened mild and severe hypocalcemia as an iCa level between 0.90 and 1.15 mmol/L and 0.90 mmol/L, respectively, and found that mild and severe hypocalcemia during ICU stay occurred in 85.0% and 10.2% of patients, values that are remarkably similar to ours. Nonsurvivors had signicantly lower iCaAdm and iCaMin concentrations. In our study, we conrmed, in a large heterogeneous cohort, that lower iCa levels are associated with unfavorable patient outcomes on univariate analysis. We then performed multivariable analysis to assess the independent association between hypocalcemia and mortality. We found an independent but weak association between iCaMin and

Table 3. Multivariate analysis for intensive care unit and hospital mortality using iCaMin Hospital Mortality Odds Ratio (95% Condence Interval) 0.97 (0.841.12) 1.02 (1.011.02) 1.15 (1.141.16) 1.81 (1.512.16) 0.64 (0.520.78) 0.93 (0.731.17) 1.05 (0.841.31) 1.09 (0.691.71) 1.55 (1.172.05) 1.08 (0.821.44) 1.55 (1.062.27)a 1.51 (1.122.02)a 1.12 (0.91.4) 1.03 (0.851.26) 1.08 (0.721.62) Intensive Care Unit Mortality Odds Ratio (95% Condence Interval) 0.88 (0.731.06) 1.01 (11.01) 1.17 (1.151.18) 2.85 (2.163.77) 0.84 (0.651.1) 0.99 (0.751.32) 0.88 (0.661.17) 0.66 (0.331.31) 1.28 (0.911.82) 0.68 (0.460.99) 1.82 (1.162.48)a 1.41 (0.971.85) 1.35 (1.011.80)a 1.06 (0.811.39) 1.24 (0.732.09)

Variable Gender (male vs. female) Age (yrs) Acute Physiology and Chronic Health Evaluation II score Mechanical ventilation Surgical patients Reason for admission Cardiac and vascular Thoracic and respiratory Trauma Neurologic Gastrointestinal tract diseases iCaMin reference (1.11.2) 0.8 0.80.9 0.91.0 1.01.1 1.2

p .66 .0001 .0001 .0001 .0001 .52 .65 .72 .0024 .58 .03

p .18 .02 .0001 .0001 .20 .97 .39 .24 .16 .05 .04

iCaMin, minimum calcium concentration during intensive care unit stay. a Indicates that the range in which iCaMin was independently associated with increased mortality.

318

Crit Care Med 2011 Vol. 39, No. 2

Table 4. Multivariate analysis for intensive care unit and hospital mortality, using iCaMax Hospital Mortality Odds Ratio (95% Condence Interval) 0.97 (0.841.12) 1.02 (1.011.02) 1.15 (1.141.16) 1.86 (1.552.23) 0.63 (0.520.78) 0.91 (0.721.15) 1.03 (0.831.29) 1.07 (0.681.69) 1.51 (1.142) 1.09 (0.821.44) 2.09 (1.213.59)a 1.4 (1.081.8)a 1.22 (1.021.45)a 1.2 (0.941.52) 1.76 (1.282.41)a Intensive Care Unit Mortality Odds Ratio (95% Condence Interval) 0.88 (0.731.06) 1.01 (11.01) 1.17 (1.151.19) 2.99 (2.273.95) 0.84 (0.641.09) 0.98 (0.741.29) 0.86 (0.651.15) 0.65 (0.331.28) 1.24 (0.881.75) 0.69 (0.471.01) 1.92 (0.933.95) 1.41 (11.99)a 1.09 (0.871.37) 1.12 (0.831.51) 1.82 (1.272.62)a

Variable Gender (male vs. female) Age (yrs) Acute Physiology and Chronic Health Evaluation II score Mechanical ventilation Surgical patients Reason for admission Cardiac and vascular Thoracic and respiratory Trauma Neurologic Gastrointestinal tract diseases iCaMax reference (1.11.2) 1.0 1.01.1 1.21.3 1.31.4 1.4

p .68 .0001 .0001 .0001 .0001 .43 .78 .76 .004 .56 .0007

p .17 .02 .0001 .0001 .19 .87 .32 .21 .22 .06 .01

iCaMax, maximum calcium concentration during intensive care unit stay. a Indicates that the range of iCaMax was independently associated with increased mortality. Table 5. Multivariate analysis for intensive care unit and hospital mortality with iCaTw Hospital Mortality Odds Ratio (95% Condence Interval) 0.96 (0.831.11) 1.02 (1.011.02) 1.15 (1.141.16) 1.87 (1.562.24) 0.64 (0.520.78) 0.92 (0.731.16) 1.06 (0.851.33) 1.09 (0.71.72) 1.56 (1.182.06) 1.08 (0.821.44) 3.52 (1.737.15)a 1.68 (1.252.26)a 1.11 (0.941.31) 1.10 (0.871.39) 1.38 (0.842.25) Intensive Care Unit Mortality Odds Ratio (95% Condence Interval) 0.87 (0.721.05) 1.01 (11.01) 1.17 (1.151.19) 2.98 (2.263.93) 0.85 (0.651.1) 0.98 (0.741.3) 0.89 (0.671.19) 0.67 (0.341.33) 1.29 (0.911.82) 0.68 (0.460.99) 4.22 (1.889.49)a 1.7 (1.172.46)a 1.29 (1.041.6)a 1.17 (0.871.58) 1.36 (0.762.43)

Variable Gender (male vs. female) Age (yrs) Acute Physiology and Chronic Health Evaluation II score Mechanical ventilation Surgical patients Reason for admission Cardiac and vascular Thoracic and respiratory Trauma Neurologic Gastrointestinal tract diseases iCaTw reference (1.11.2) 0.9 0.91.0 1.01.1 1.21.3 1.3

p .58 .0001 .0001 .0001 .0001 .48 .58 .70 .002 .58 .0003

p .15 .03 .0001 .0001 .21 .90 .43 .25 .15 .05 .001

iCaTw, time-weighted average of calcium concentration during intensive care unit stay. a Indicates that the range of iCaTw was independently associated with increased mortality.

iCaMax and a stronger and more signicant relationship between iCaTw and ICU and hospital mortality. However, most of this association was driven by patients with extreme hypocalcemia. There are even fewer published studies that assess the incidence and association of hypercalcemia in critically ill patients. Forster et al (27) reported hypercalcemia (iCa 1.33 mmol/L) in 15 of 100 (15%) surgical ICU patients with no difference in mortality reported between patients with or without hypercalcemia. Zaloga et al (1) reported hypercalcemia (total serum Ca 10.5 mg/dl or iCa 1.27 mmol/L) in 8%
Crit Care Med 2011 Vol. 39, No. 2

of 156 surgical/medical intensive care patients with no difference in mortality between patients with or without hypercalcemia. No studies have assessed the independent relationship between ionized hypercalcemia and outcome. Our observations demonstrate that, using a variety of indices, ionized hypercalcemia has an association with ICU and hospital mortality, which is similar in strength to that of ionized hypocalcemia. Hypocalcemia of critical illness has been treated with calcium chloride (29). However, intravenous calcium administration might cause skin necrosis (37), cardiac ar-

rhythmias (38, 39), and cellular injury (40, 41). Furthermore, calcium administration has been reported to increase mortality in septic animal models (30, 42). Despite the high incidence of hypocalcaemia, there is no randomized controlled trial to assess the risks and benets of calcium supplementation in hypocalcemic critically ill patients; it is therefore unclear whether hypocalcemia (mild/moderate or even severe in the absence of clinical signs) should be treated. By demonstrating an almost symmetrical U-shaped relationship between iCa levels and ICU and hospital mortality with weak independent association out319

side of extreme low and high range, our ndings suggest that disorders in ionized calcemia are most likely to represent levels of physiologic derangements, which reect underlying illness severity (marker of illness severity). Such weak association differs in both nature, internal coherence, and strength from that seen, for example, with indices of glycemia (4750) and does not logically justify specic interventions to treat iCa (except, perhaps, at extremes). Until randomized controlled study become available, no correction of ionized calcium, at least between 0.9 and 1.4 mmol/L, appears physiologically or clinically rational. In many centers, iCa values are now typically and automatically included in arterial blood gas measurement reports, even though they are not the driving force to sampling. Our data may be useful to clinicians in appreciating the mortality rates associated with certain level of iCa. Such information can be incorporated in the decision making process as additional markers of prognosis. The causes of abnormalities in iCa are not clearly understood. Some have described an association with altered parathyroid hormone levels (9, 43). In this study, we were not able to observe parathyroid hormone level. It is also possible that changes in phosphate levels would affect iCa levels. In about half of our cohort, daily phosphate levels were available to assess any such link and only a weak relationship was found. A possible explanation for the relationship between iCa concentration and outcome might relate to a possible effect of hypercalcemia on kidney dysfunction in critically ill patients (44) . However, iCa levels at admission had only a very weak association with serum creatinine in the rst 24 hrs. These ndings suggest that abnormalities of iCa might occur because of multiple factors and might only be markers of acute illness rather than contributors to morbidity (45, 46). There are limitations in this study. Our study is retrospective in design and thus potentially subject to systematic error and bias. However, the clinical and electronic data were collected prospectively, are numerical in nature, and were measured independently; thus they were not amenable to selection bias or unintended manipulation. We studied only four centers, and our ndings may not be generalizable. However, we note that all other studies of iCa so far have been single center in design and that our study is
320

one order of magnitude greater than previous investigations. Furthermore, there is no information of citrate infusion accompanied with blood transfusion, which might affect our results. Data on such confounding factors should be collected in further prospective studies. To our knowledge, this study is the largest and rst multicenter examination of the epidemiology and outcome associations of hypo- and hypercalcemia. Our observations are in broad agreement with the literature for hypocalcemia and expand our understanding of hypercalcemia. Although the clinical benets and safety of correcting high or low iCa levels can only be fully assessed in a randomized controlled trial, until such time, our ndings also have practical implications. In summary, in a large, heterogeneous, multicenter cohort of critically ill patients, we found a limited, independent, U-shaped association between ionized hypo- and hypercalcemia and mortality. We also found that this association was mostly dependent on patients with extremely low or high iCa levels. Our results suggest that abnormalities of iCa concentrations are likely a marker of illness severity rather than an independent contributor to mortality, especially for levels between 0.9 and 1.4 mmol/L. For these patients no correction of iCa appears physiologically or clinically rational. Further studies now appear desirable to conrm or refute our observations.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

REFERENCES
1. Zaloga GP, Chernow B, Cook D, et al: Assessment of calcium homeostasis in the critically ill surgical patient. The diagnostic pitfalls of the McLean-Hastings nomogram. Ann Surg 1985; 202:587594 2. Auffant RA, Downs JB, Amick R: Ionized calcium concentration and cardiovascular function after cardiopulmonary bypass. Arch Surg 1981; 116:10721076 3. Drop LJ: Ionized calcium, the heart, and hemodynamic function. Anesth Analg 1985; 64: 432 451 4. Zaloga GP, Chernow B: Hypocalcemia in critical illness. JAMA 1986; 256:1924 1929 5. Ha stbacka J, Pettila V: Prevalence and predictive value of ionized hypocalcemia among critically ill patients. Acta Anaesthesiol Scand 2003; 47:1264 1269 6. Aderka D, Schwartz D, Dan M, et al: Bacteremic hypocalcemia. A comparison between the calcium levels of bacteremic and nonbacteremic patients with infection. Arch Intern Med 1987; 147:232236 7. Zivin JR, Gooley T, Zager RA, et al: Hypocal18.

19.

20.

21.

22.

23.

24.

cemia: A pervasive metabolic abnormality in the critically ill. Am J Kidney Dis 2001; 37: 689 698 Mu ller B, Becker KL, Kra nzlin M, et al: Disordered calcium homeostasis of sepsis: Association with calcitonin precursors. Eur J Clin Invest 2000; 30:823 831 Lind L, Carlstedt F, Rastad J, et al: Hypocalcemia and parathyroid hormone secretion in critically ill patients. Crit Care Med 2000; 28:9399 Desai TK, Carlson RW, Geheb MA: Prevalence and clinical implications of hypocalcemia in acutely ill patients in a medical intensive care setting. Am J Med 1988; 84:209 214 Chernow B, Zaloga G, McFadden E, et al: Hypocalcemia in critically ill patients. Crit Care Med 1982; 10:848 851 Carlstedt F, Lind L, Rastad J, et al: Parathyroid hormone and ionized calcium levels are related to the severity of illness and survival in critically ill patients. Eur J Clin Invest 1998; 28:898 903 Zaloga GP, Chernow B: The multifactorial basis for hypocalcemia during sepsis. Studies of the parathyroid hormone-vitamin D axis. Ann Intern Med 1987; 107:36 41 Carlstedt F, Lind L, Wide L, et al: Serum levels of parathyroid hormone are related to the mortality and severity of illness in patients in the emergency department. Eur J Clin Invest 1997; 27:977981 Burchard KW, Gann DS, Colliton J, et al: Ionized calcium, parathormone, and mortality in critically ill surgical patients. Ann Surg 1990; 212:543549; discussion 549 550 Broner CW, Stidham GL, Westenkirchner DF, et al: Hypermagnesemia and hypocalcemia as predictors of high mortality in critically ill pediatric patients. Crit Care Med 1990; 18:921928 Ward RT, Colton DM, Meade PC, et al: Serum levels of calcium and albumin in survivors versus nonsurvivors after critical injury. J Crit Care 2004; 19:54 64 Ammori BJ, Barclay GR, Larvin M, et al: Hypocalcemia in patients with acute pancreatitis: A putative role for systemic endotoxin exposure. Pancreas 2003; 26:213217 Weir GC, Lesser PB, Drop LJ, et al: The hypocalcemia of acute pancreatitis. Ann Intern Med 1975; 83:185189 Pitchumoni CS, Agarwal N, Jain NK: Systemic complications of acute pancreatitis. Am J Gastroenterol 1988; 83:597 606 Klein GL, Herndon DN: Magnesium decit in major burns: Role in hypoparathyroidism and end-organ parathyroid hormone resistance. Magnes Res 1998; 11:103109 Szyfelbein SK, Drop LJ, Martyn JA: Persistent ionized hypocalcemia in patients during resuscitation and recovery phases of body burns. Crit Care Med 1981; 9:454 458 Henderson SA, Graham HK, Mollan RA: Serum and other calcium fractions in patients after severe musculoskeletal trauma. Clin Orthop Relat Res 1992; 306 311 Choi YC, Hwang SY: The value of initial

Crit Care Med 2011 Vol. 39, No. 2

25.

26.

27.

28.

29.

30.

31.

32.

33.

ionized calcium as a predictor of mortality and triage tool in adult trauma patients. J Korean Med Sci 2008; 23:700 705 Lind L, Ljunghall S: Critical care hypercalcemiaa hyperparathyroid state. Exp Clin Endocrinol 1992; 100:148 151 Oberleithner H, Greger R, Lang F: The effect of respiratory and metabolic acid-base changes on ionized calcium concentration: In vivo and in vitro experiments in man and rat. Eur J Clin Invest 1982; 12:451 455 Forster J, Querusio L, Burchard KW, et al: Hypercalcemia in critically ill surgical patients. Ann Surg 1985; 202:512518 Kraft MD, Btaiche IF, Sacks GS, et al: Treatment of electrolyte disorders in adult patients in the intensive care unit. Am J Health Syst Pharm 2005; 62:16631682 Jankowski S, Vincent JL: Calcium administration for cardiovascular support in critically ill patients: When is it indicated? J Intensive Care Med 1995; 10:91100 Zaloga GP, Sager A, Black KW, et al: Low dose calcium administration increases mortality during septic peritonitis in rats. Circ Shock 1992; 37:226 229 Egi M, Naka T, Bellomo R, et al: A comparison of two citrate anticoagulation regimens for continuous veno-venous hemoltration. Int J Artif Organs 2005; 28:12111218 Knaus WA, Draper EA, Wagner DP, et al: APACHE II: A severity of disease classication system. Crit Care Med 1985; 13:818 829 Australian and New Zealand Intensive Care

34.

35.

36.

37.

38. 39.

40.

41.

42.

43.

Society: Available at: http://www.anzics.com.au. Accessed November 15, 2010 Stow PJ, Hart GK, Higlett T, et al: Development and implementation of a high-quality clinical database: The Australian and New Zealand Intensive Care Society Adult Patient Database. J Crit Care 2006; 21:133141 National Association Testing Authorities: Available at: http://www.nata.com.au/. Accessed November 15, 2010 The Royal College of Pathologists of Australia: Available at: http://www.rcpamanual.edu.au/. Accessed November 15, 2010 Weiss Y, Ackerman C, Shmilovitz L: Localized necrosis of scalp in neonates due to calcium gluconate infusions: A cautionary note. Pediatrics 1975; 56:1084 1086 Levy MN: Role of calcium in arrhythmogenesis. Circulation 1989; 80:IV23IV30 Carlon GC, Howland WS, Goldiner PL, et al: Adverse effects of calcium administration. Report of two cases. Arch Surg 1978; 113: 882 885 Cheung JY, Bonventre JV, Malis CD, et al: Calcium and ischemic injury. N Engl J Med 1986; 314:1670 1676 Schanne FA, Kane AB, Young EE, et al: Calcium dependence of toxic cell death: A nal common pathway. Science 1979; 206:700 702 Malcolm DS, Zaloga GP, Holaday JW: Calcium administration increases the mortality of endotoxic shock in rats. Crit Care Med 1989; 17:900 903 Baines PB, Thomson AP, Fraser WD, et al:

44.

45.

46.

47.

48.

49.

50.

Hypocalcaemia in severe meningococcal infections. Arch Dis Child 2000; 83:510 513 Escalante CP, Weiser MA, Finkel K: Hyperphosphatemia associated with phosphoruscontaining laxatives in a patient with chronic renal insufciency. South Med J 1997; 90: 240 242 Song SK, Karl IE, Ackerman JJ, et al: Increased intracellular Ca2: A critical link in the pathophysiology of sepsis? Proc Natl Acad Sci U S A 1993; 90:39333937 Rose S, Sayeed MM: Superoxide radical scavenging prevents cellular calcium dysregulation during intraabdominal sepsis. Shock 1997; 7:263268 Egi M, Bellomo R, Stachowski E, et al: Intensive insulin therapy in postoperative intensive care unit patients: A decision analysis. Am J Respir Crit Care Med 2006; 173:407 413 Egi M, Bellomo R, Stachowski E, et al: Blood glucose concentration and outcome of critical illness: The impact of diabetes. Crit Care Med 2008; 36:2249 2255 Egi M, Morimatsu H, Toda Y, et al: Hyperglycemia and the outcome of pediatric cardiac surgery patients requiring peritoneal dialysis. Int J Artif Organs 2008; 31: 309 316 Egi M, Bellomo R, Stachowski E, et al: Variability of blood glucose concentration and short-term mortality in critically ill patients. Anesthesiology 2006; 105:244 252

Crit Care Med 2011 Vol. 39, No. 2

321