Albumin adjustment of total calcium not justified

Almost all (99 %) of the approximate 1 kg of calcium in the adult human body is contained in bone, but a small fraction (approximately 350 mg) circulates in blood plasma at a concentration of around 2.5 mmol/L. This circulating calcium comprises three fractions: around half (50 %) is free ionized calcium; the rest is bound to proteins, principally albumin (40 %) and complexed to a range of anions (10 %). Only the free ionized fraction is physiologically active and arguably of clinical significance. 

There are two broad methods of measuring calcium in blood. The first measures total calcium (bound plus ionized calcium) in blood plasma or serum, approximate reference range 2.20-2.60 mmol/L, and the second measures ionized calcium only, usually in anticoagulated whole-blood samples, approximate reference range 1.10-1.30 mmol/L. 

It is widely acknowledged that ionized calcium is the preferred method because this is the physiologically important fraction. However, this assay is not available on large automated machines that process blood plasma/serum samples in the central laboratory and total calcium is the method used here. Ionized calcium measurement is currently largely restricted to blood gas analyzers and other low-throughput point-of-care analyzer platforms. 

The validity of using total calcium measurement as a proxy for the clinically important ionized calcium fraction has been thought to depend on the extent to which plasma albumin concentration (the principal calcium-binding protein) deviates from normal. This notion underpins a conventional wisdom that interpretation of total calcium results should take account of the patient’s albumin concentration.

Over the years a number of formulae for calculation of ”albumin-corrected” total calcium have been proposed and it is normal practice in many laboratories to employ one or other of these formulae and report ”albumin-corrected” total calcium results rather than simply the ”uncorrected” measured total calcium result. This widespread practice is now challenged by the results of a recently published study conducted at Uppsala hospital in Sweden. 

Investigators retrieved calcium results data for all instances at their hospital during an 8-year period (2005 to 2013) when ionized calcium, total calcium and albumin were measured on the same sample. In total, this generated 16,897 data sets for statistical analysis. They employed six different formulae to calculate six different corrected total calcium values for each of the 16,897 retrieved total calcium results. 

Using the ionized calcium results as reference, they compared agreement between ionized calcium and uncorrected total calcium results with agreement between ionized calcium and each of the six corrected total calcium results. The statistical method of comparison was intraclass correlation coefficients (ICC), which generates values from 0 (no agreement) to 1 (perfect agreement). 

Analysis of this massive data base revealed “substantial agreement” between ionized calcium and uncorrected total calcium (ICC = 0.85). Equal or worse level of agreement was evident when ionized calcium result was compared with corrected total calcium; depending on the correction formula used, ICC values ranged from 0.45 to 0.81. 

A second method of analyzing the data was also employed. Investigators categorized all ionized calcium results to one of three groups: hypocalcemic (ionized calcium <1.10 mmol); normocalcemic (ionized calcium in the range of 1.10-1.30); and finally, hypercalcemic (ionized calcium >1.30). Using uncorrected total calcium results, calcium status (hypocalcemia, normocalcemia or hypercalcemia) was correctly identified in 82 % of cases. By contrast, corrected total calcium results predicted correct calcium status in lower percentage of cases. 

In summary, the authors of this study find no evidence that albumin correction of calcium results adds useful clinical information; they suggest that the practice should be abandoned.