The association between ionized and total magnesium in critically ill patients: a prospective cohort study.

Abstract

Most clinical laboratories measure total magnesium (tMg) which includes both the ionized (iMg) and bound forms. Only the ionized component is biologically active, and it may be influenced by acid-base status, plasma protein concentrations and electrolyte disturbances, potentially altering how well tMg reflects iMg. This multicenter observational study evaluated the relationship between tMg and iMg using paired samples collected within a 2-hour window. Association was assessed with linear regression and Pearson's correlation coefficient, and agreement with Bland-Altman analysis. Ionized magnesium constituted 77.9% of tMg (95% CI 77.5%-78.4%). There was a strong correlation between iMg and tMg (r = 0.916, p < 0.0001). Bland-Altman analysis showed that tMg was, on average, 0.20 mmol/L higher than iMg (95% LoA 0.03-0.37 mmol/L). Proportional bias was observed (slope = 0.22, p < 0.001), with increasing discrepancies at higher concentrations. Stepwise multiple linear regression analysis including acid-base parameters, albumin and electrolytes, was used to identify variables independently associated with ionized and total magnesium concentrations. In this multivariable model, pH and ionized calcium were independently associated with both ionized and total magnesium. In conclusion, ionized and total magnesium were strongly correlated. However, proportional bias indicated that total magnesium increasingly deviates from the biologically active ionized fraction at higher concentrations. Therefore, derived equations to estimate one measure from the other are not reliable for clinical use.