Development and Validation of a Novel Laboratory-Specific Correction Equation for Total Serum Calcium and Its Association With Mortality Among Hemodialysis Patients

Abstract

Conventional albumin-corrected calcium is inaccurate in predicting ionized calcium, and hidden hypercalcemia, characterized as high ionized calcium with normal total calcium, is associated with higher mortality in hemodialysis patients. By using a national cohort of hemodialysis patients in the Unites States, a novel laboratory-specific prediction equation composed of total calcium, albumin, and phosphorus was derived from 242 patients in the South Atlantic division (adjusted R² = 0.80 versus 0.71 for the conventional equation) and then validated among 566 patients in the other divisions (adjusted R² = 0.79 versus 0.68 for the conventional equation). Compared with the conventional equation, the novel equation showed a greater correlation with intact parathyroid hormone. Its relative performance against the conventional equation was consistent across subgroups based on medications related to calcium metabolism. The novel equation also had a higher sensitivity (57% versus 34%) and an equivalent specificity (99% versus 100%) against ionized hypercalcemia at a cut-off value of 10.2 mg/dL. Sensitivity and specificity at 9.4 mg/dL was 94% and 76% (versus 87% and 82% for the conventional equation), respectively. A survival analysis in 87,779 incident hemodialysis patients showed that among patients who were categorized as having a high-normal calcium status (ie, >9.4 to 10.2 mg/dL) by the conventional equation, there appeared a trend toward higher adjusted mortality risk across higher calcium status defined according to the novel equation. Meanwhile, the mortality risk was consistent across calcium strata defined according to the conventional equation within the categories defined by the novel equation. In conclusion, in comparison to the conventional equation, a novel laboratory-specific correction equation derived for correction of total calcium performs significantly better in ascertaining hidden hypercalcemia in hemodialysis patients, and aids in identifying patients at higher risk for mortality. © 2016 American Society for Bone and Mineral Research.

Keywords: DISORDERS OF CALCIUM/PHOSPHATE METABOLISM; EPIDEMIOLOGY; STATISTICAL METHODS.

Fig. 1

Scatter plots between ionized calcium concentrations and (A) uncorrected total calcium and (B) conventional albumin-corrected total calcium among 242 patients in the derivation data set. All values are expressed as Z-scores (see Materials and Methods for detail).

Fig. 2

Bland-Altman plots of total calcium corrected by the conventional equation (EQ), EQ1, EQ2, and EQ3 in the derivation data set of 242 patients (the left panels) and the validation data set of 566 patients (the right panels). The y axis denotes the bias in each equation calculated by (total calcium-equivalent ionized calcium value) – (total calcium value corrected by a given equation). Conv. = conventional; tCa = total calcium.

Fig. 3

(A) Receiver operating characteristic curve, (B) Youden index, (C) sensitivity, and (D) specificity of total calcium corrected by the conventional equation (EQ), EQ1, EQ2, and EQ3 against ionized hypercalcemia (ionized calcium >1.32 mmol/L) in the validation data set of 566 patients. Dotted vertical lines were placed at 9.4 mg/dL and 10.2 mg/dL. Sensitivity and specificity at 10.2 mg/dL against ionized hypercalcemia was 34% and 100%, 50% and 99%, 57% and 99%, and 57% and 98% for the conventional EQ, EQ1, EQ2, and EQ3, respectively. Likewise, sensitivity and specificity at 9.4 mg/dL against ionized hypercalcemia was 87% and 82%, 90% and 78%, 94% and 76%, and 94% and 74% for the conventional EQ, EQ1, EQ2, and EQ3, respectively. Conv. = conventional.

Fig. 4

(A) Concordance and discordance of calcium status between the conventional equation (EQ) and EQ2 and their association with (B) all-cause and (C) cardiovascular mortality among 87,779 hemodialysis patients who survived the first 6 months of dialysis and who had data on total calcium, albumin, and phosphorus during months 4–6. Gray cells in A indicate groups where hazard ratios were not reported because of a limited number of patients. ^*p < 0.05 (B and C).: Conv. = conventional.