Skip to main content
. 2013 Oct 16;8(10):e76516. doi: 10.1371/journal.pone.0076516

Table 4. Multiple regression of log transformed (ln) hepcidin.

Variable Regression coefficient (95% CI) Standard error p value
A) Ferritin restricted
Ferritin 0.04 (0.023, 0.058) 0.0083 <0.0001
Ferritin2 −0.00015 (−0.00028, −0.000066) 0.000039 0.001
B) Complete model
Ferritin 0.036 (0.028, 0.044) 0.0037 <0.001
Ferritin2 −0.00011 (−0.00015, −0.000076) 0.000017 <0.001
Hepcidin:ferritin 1.88 (1.43, 2.34) 0.22 <0.001

A) Model using ferritin alone. 95% CI, 95% confidence intervals. Overall model parameters: sum of squares 17.2; 20 degrees of freedom; mean square 0.86; variance ratio (F) 11.0; adjusted r2 = 60.0; overall p value for model = 0.0001. B) The final model for ln(hepcidin), generated by stepwise multiple regression in data from 21 HHT patients. Overall model parameters: sum of squares 17.2, 20 degrees of freedom, mean square 0.86, variance ratio (F) 81.5, adjusted r2 = 92.4, overall p value for model <0.0001. No other captured variable contributed to the model.