Accurate assessment of calcium status in the critically ill requires measurement of ionized calcium rather than total calcium

Disturbance of blood calcium concentration is a common feature of many critical illnesses; prevalence studies suggest for example that 50-85 % of patients admitted to intensive/critical care have some degree of hypocalcemia (reduced blood calcium concentration). So monitoring and correction of blood calcium, where necessary, is part of the routine care delivered to all patients admitted to intensive/critical care units. 

The calcium present in blood plasma/serum comprises three fractions: around 40 % is bound to protein (principally albumin); around 10 % is complexed with a range of anions, including bicarbonate, phosphate and citrate; and the remaining 50 % is free (unbound, non-complexed) ionized calcium. This third fraction is the physiologically active, and therefore the clinically relevant fraction.

It is widely accepted that calcium status is best assessed by measurement of plasma/serum ionized calcium (Ca²⁺) concentration, which in health is maintained within the approximate reference interval of 1.15-1.30 mmol/L (4.6-5.2 mg/dL). Unfortunately, the ion-specific electrode assay used to determine ionized calcium is currently only available on blood gas analyzers and other similar low-throughput point-of-care platforms.

In the central laboratory, where high-throughput biochemical profiling analyzers are used to process the vast majority of clinical samples, calcium status is assessed by measurement of serum/plasma total calcium (tCa) concentration, i.e. the sum of the concentration of all three fractions (protein-bound calcium plus complexed calcium plus free ionized calcium). In health serum/plasma tCa concentration is maintained within the approximate reference interval: 2.20-2.60 mmol/L (8.8-10.4 mg/dL). 

The validity of using tCa measurement as a proxy for the clinically important ionized calcium fraction depends on the constancy of the % of total calcium that is in the ionized state, i.e. on the constancy of the equilibrium between ionized and bound/complexed calcium. A number of factors, most notably serum albumin concentration and pH, can affect this equilibrium; if serum albumin or pH are abnormal, then tCa is, potentially, a less than reliable predictor of ionized calcium. It is routine practice to employ one of several formulas to correct tCa concentration for abnormality in serum albumin concentration but there is evidence to suggest that these formulas are invalid during critical illness.

Additionally, disturbance of acid-base (abnormal pH) is common in the critically ill. The notion, suggested by these considerations, that tCa may be a less than reliable indicator of ionized calcium during critical illness has been given credence by evidence from a number of studies in recent years. Results of this recently published study contribute to the accumulating evidence that the calcium status of critically ill patients should be assessed using ionized rather than total calcium measurement. 

The stated aim of this retrospective study was to determine whether the concentration of total serum calcium (tCa), either corrected for albumin or not, can predict critically high or low ionized serum calcium values in critical illness. Critically low and high ionized calcium concentrations were defined respectively as: ≤0.79 mmol/L (3.16 mg/dL) and ≥1.58 mmol/L (6.32 mg/dL).

Study investigators employed the Medical Information Mart for Intensive Care (MIMIC) III database. MIMIC-III is a freely accessible database of the clinical details of all patients admitted to critical care units at a large US tertiary care hospital, Beth Deaconess Medical Center in Boston Massachusetts, between 2001 and 2012. This huge database holds the clinical details (including all laboratory/point-of-care test results) of 46,520 critically ill patients (including 38,645 adults and 7875 neonates). 

Investigators interrogated the MMIC-III database to identify all instances in which patients had total calcium and albumin measured within 1 hour (before or after) ionized calcium measurement. A total of 12,118 such instances involving 4699 critically ill patients were retrieved. Serum albumin concentration was used to generate an ”albumin corrected” total calcium result. So, each of the 12,118 sets retrieved for statistical analysis comprised three matched elements: an ionized calcium result, an uncorrected total calcium result and an albumin-corrected total calcium result. 

Spearman correlation analysis revealed positive correlation between ionized calcium and total (uncorrected) calcium although the correlation coefficient (r), just 0.51 (P<0.01) indicates that overall, uncorrected total calcium concentration does not well reflect ionized calcium concentration. 

Of the 12,118 ionized calcium results, 103 (0.85 %) were ≤0.79mmol/L (critically low) and 92 (0.76 %) were ≥1.58 mmol/L (critically high). Receiver operating characteristic (ROC) curve analysis was used to estimate the predictive accuracy of total calcium (both uncorrected and corrected) to identify these critically low and critically high ionized calcium samples. This revealed that total calcium is a less than reliable predictor of critically low ionized calcium and that albumin correction of total calcium has no improving effect; areas under the curve (AUC) for corrected and uncorrected total calcium to predict critically low ionized calcium were respectively, 0.69 (95 % CI: 0.61-0.76) and 0.70 (95 % CI: 0.63-0.78).

However, ROC analysis revealed that total calcium (whether corrected or not) is a reliable predictor of critically high ionized calcium; AUCs for corrected and uncorrected total calcium to predict critically high ionized calcium were respectively, 0.98 (95 % CI: 0.97-1.00); and 0.97 (95 % CI: 0.95-1.00). 

Further statistical analysis is presented in this paper which supports the authors’ overall conclusion that predictive accuracy of total calcium, either corrected or uncorrected for critically low ionized calcium is not reliable. However, the predictive accuracy of total calcium for critically high ionized calcium is reliable. Albumin correction of serum total calcium was found not to affect (improve) the ability of total serum calcium to predict either critically low or critically high ionized calcium in this critically ill study population.