Quadratic regression of hematinic and iron indices with HAIR.

Age and gender-adjusted HAIR regression coefficients with the designated variable were calculated using the designated variable as the dependent variable (distributional graphs are provided in Figure S1) for quadratic regression, and simultaneously examining the relationships with age, gender, and HAIR. MCV, mean corpuscular volume; MCH, mean corpuscular hemoglobin; MCHC, mean corpuscular hemoglobin concentration; ln(RDW), log-transformed red cell distribution width; ln(iron), log-transformed serum iron; ln(TfSI), log-transformed transferrin saturation index; ln(ferritin), log-transformed serum ferritin. 48 datasets were available except for RDW (n = 46), iron and TfSI (n = 45), and ferritin (n = 41). Excluding outliers in the MCV, MCH and MCHC regressions did not materially alter the results (p = 0.76 for MCV, n = 44; p = 0.035 for MCH, n = 43 and p = 0.001 for MCHC, n = 36). Age, gender and iron supplement-adjusted HAIR regression coefficients with the designated variable were calculated by quadratic regression, using the designated variable as the dependent variable, and simultaneously examining the relationships with age, gender, iron supplement use, and HAIR. Iron supplements contributed significantly to the final models of MCHC and ln(RDW), reducing the respective HAIR p values in age, gender and iron supplement adjusted regression.