Adjustment of serum total calcium for albumin concentration: Effects of non-linearity and of regression differences between laboratories

It has been reported that the relationship between serum total calcium and albumin concentrations in hospital patients deviates from linearity at low albumin concentrations. We searched a large laboratory computer data base for adult patients with discretionary requests for serum calcium analysis but with no other data suggestive of disturbances of calcium homeostasis, and collected a minimum of 100 calcium values at each of a wide range of albumin concentrations. We confirmed the earlier observation, but found that the use of a single regression to derive an adjustment of total calcium for albumin gave only small differences of no clinical significance. To investigate whether equations to adjust total calcium for albumin can be transferred between laboratories, three laboratory computers were searched for calcium requests from patients likely to have a low prevalence of calcium disturbances. The regressions of total calcium on albumin differed significantly between laboratories, but within each laboratory gave adjusted calcium values identical with those in health. Although the errors resulting from applying an equation from one laboratory to another were small and unlikely to be of major clinical significance, we recommend that where possible laboratories should derive adjustment equations from their own data.