How well predicted ionized calcium detects calcium problems in dogs

Abstract

BACKGROUND: Abnormal total calcium (tCa) values do not necessarily imply dysregulated ionized calcium. OBJECTIVES: We aimed to evaluate the diagnostic performance of predicted ionized calcium (piCa) regarding true calcium status in dogs with abnormal tCa. METHODS: This was a cross-sectional multicenter study. piCa and its prediction interval (PI) were calculated in 114 dogs, from three different hospitals, with either increased (62) or decreased tCa (52), All dogs also had ionized calcium and a biochemical profile available. The sensitivity, specificity, predictive values, and diagnostic discordance of piCa to confirm ionized hypercalcemia (iHyperCa) and ionized hypocalcemia (iHypoCa) were calculated using logistic regression analysis. RESULTS: iHyperCa was found in 28% and 66% of hyperphosphatemic and non-hyperphosphatemic dogs with tCa above the reference interval upper limit, respectively. The piCa correctly classified dogs with iHyperCa in 72.2% of those with hyperphosphatemia and 93.2% of those without hyperphosphatemia. Comparatively, elevating the tCa threshold to 12 mg/dL properly classified dogs 50% and 75% of the time in hyperphosphatemic and non-hyperphosphatemic dogs, respectively. iHypoCa was found in only 31/52 (60%) dogs with decreased tCa. The piCa correctly classified 55.2 to 100% of dogs with iHypoCa depending on the hospital. The PI demonstrated high sensitivity for iHyperCa (100%) and high specificity for both iHyperCa (100%) and iHypoCa (100%). CONCLUSIONS: Evaluating tCa alone does not reliably determine ionized calcium status. Even with hyperphosphatemia, piCa and its PI represent a reliable alternative to interpret abnormal tCa values when ionized calcium measurements are not available. However, if the tCa reference interval is notably different from 7.6 to 11.4 mg/dL, piCa values might be under/overestimated.