*2.8. Statistical Analyses*

Statistical analyses were performed using Statgraphics Centurion XVI software and Graph-Pad Prism 7.00. *p*-values less than 0.05 were considered statistically significant. Normality was verified with the Kolmogorov–Smirnov test. Normally distributed and continuous variables were expressed as means ± standard deviation (SD). Non-normally distributed variables were expressed as median and interquartile range. Comparison of continuous variables between the two groups were performed using the Student's t-test (normally distributed) or Mann–Whitney test (non-normally distributed). Di fferences between proportions were assessed using the chi-square or Fisher exact test.

The area-under-curve (AUC) of the receiver operating characteristic (ROC) curves for each oxidative stress biomarker were used to characterize their utility for discriminating CD patients from healthy subjects.

A stepwise multiple linear regression analysis with backward selection was performed with oxidative stress markers as the dependent variables and the following factors as independent variables: age at surgery, gender, diagnostic delay, smoke habit, CDAI, disease location, disease behavior, disease duration, extra-intestinal disease, perianal disease, first clinical presentation, recurrence, number of surgeries, allergies, and family history of IBD.

An oxidative score was calculated and consisted of four components. For oxidative markers (TBARS, CO, AGEs, and AOPP) values below the median value were assigned 0 point and those above the median value, 1 point. For FRAP values, the point assignment was the reverse (0 below the median and −1 above the median value). According to the sum of the four components, patients were allocated to three oxidative score categories: low, medium, and high.
