Comparison of two methods for the assessment of intra-erythrocyte magnesium and its determinants 53 3 7,5 mg/L, 3,0% at 25 mg/L and 3,4% at 45 mg/L. Within run and between run coefficient of variation were 0.5% and 2.0% for the LLOQ (5 mg/L), 0.6% and 2.7% for the low quality control sample (7.5 mg/L), 0.6% and 1.1% for the medium quality control sample (25.0 mg/L) and 0.4% and 0.8% for the high quality control sample (45 mg/L). iIEM was calculated according to the formula: iIEM = (whole blood magnesium - plasma magnesium × (1 - hematocrit)) / hematocrit 14. Statistical analyses Passing-Bablok regression analysis was used to calculate slopes and intercepts for the relationship between dIEM and iIEM (expressed in mmol/L and mmol/1012 cells). BlandAltman plots were used to visualize and analyze bias between dIEM and iIEM. Data are presented as mean ± SD for normally distributed data and median [interquartile range (IQR)] for non-normally distributed data. Categorical data are presented as percentages. We used frequency distribution histograms and Q-Q plots to assess normality of our data. Data with a non-parametric distribution were log10 transformed. We tested correlations between dIEM, iIEM, plasma magnesium and 24-h urinary magnesium excretion with Pearson bivariate correlation coefficients. Linear regression analysis was performed to investigate determinants of dIEM, iIEM, plasma magnesium and 24-h urinarymagnesium excretion. Regression coe cients were given as standardized beta values, referring to the number of standard deviations a dependent variable changes per standard deviation increase of the independent variable, thereby allowing for comparison of the strength of the associations of di erent variables. Clinical parameters included BMI, waist circumference, glucose concentrations, HbA1c, cholesterol and triglyceride concentrations, blood pressure, renal function (eGFR and creatinine clearance), plasma concentrations of sodium and potassium and hematology parameters (hemoglobin, hematocrit, and MCV). Potential confounders, including age, sex, BMI, eGFR (eGFR <90 mL/min/1.73m2), plasma sodium and potassium and alcohol consumption and smoking status were taken into account. Missing data (present in data on alcohol consumption (30.9 %) and smoking status (30.4 %)) were handled with multiple imputations 24. Results are reported for imputed data, except for the baseline characteristics. We evaluated potential effect modification in the associations of determinants with iIEM, dIEM, and plasma and urine magnesium by fitting models containing both main effects and their cross-product terms. Due to the shape of the cells, it is impossible to obtain only packed cells after washing and therefore, we indexed the intracellular magnesium concentrations to erythrocyte count and dIEM and iIEM were expressed as mmol/1012
RkJQdWJsaXNoZXIy MTk4NDMw