Novel methods of predicting ionized calcium status from routine data in critical care: External validation in MIMIC-III

Abstract

BackgroundLow ionized calcium (ICa) is prevalent and prognostic in critical care, but poorly detected by either total calcium (TCa) or albumin-corrected TCa (cTCa). We recently derived models of ICa (Pred-ICa) and low ICa (ProbHYPO) in critical care that adjust TCa for binding to albumin and small anions—represented by the anion gap’s components. On internal validation, they outperformed cTCa in diagnosing low ICa. Two other new anion gap-based models of ICa, derived in renal patients, have not been validated. This study tested the external validity of these 4 new models in the Medical Information Mart for Intensive Care III (MIMIC-III) database. MethodsWe identified 4105 patients in MIMIC-III with ICa measured on an arterial blood gas panel within 20 min of chemistry panel measurements of TCa, albumin, sodium, chloride, and total carbon dioxide. The 4 models and cTCa were assessed by their diagnostic discrimination for low ICa (<1.10 mmol/l) and high ICa (>1.32 mmol/l), and by the agreement between predicted and observed values. ResultsPred-ICa and ProbHYPO had the best discrimination and agreement. ConclusionsPred-ICa and ProbHYPO were externally validated in MIMIC-III. They can help clinicians efficiently decide when to order direct ICa testing in critical care.