Maarten Cozijnsen

37 Chapter 2 Development and validation of the MINI-index In all three cohorts endoscopic activity was captured using the SESCD and explicit demographic and laboratory data, including FC, were collected at the time of ileocolonoscopy (but not during the bowel preparation), as well as PCDAI. A total of 168 children were included in the validation cohorts (Table 1). Statistical analyses The derivation of the MINI index was based on the individual PCDAI items and the following laboratory items: hematocrit, albumin, ESR, CRP, platelets, white blood cell count, and FC. We explored various models to associate with the SESCD including when laboratory tests were entered to the models as continuous variables or grouped into categories. We used a blended mathematical–judgmental approach to determine the items; those with a p-value>0.1 in the multivariate analyses were considered for exclusion by an advisory board of 10 international experts in pediatric CD (see authors of this manuscript), thus ensuring content and face validity. Discriminative validity was assessed by the area under the receiver operating characteristic (ROC) curve (AUROC) which was also used for exploring the best cutoff to identify MH (SESCD<3), and a second cutoff to discriminate mild (SESCD 3-9) from moderate-to-severe (SESCD > 9) mucosal inflammation. Data are reported as mean ± standard deviation or median (inter quartile range [IQR]) as appropriate. Continuous data were compared using Student’s t test, or the Wilcoxon rank sum test as per the distribution normality. Spearman or Pearson correlations were used as appropriate. Categorical variables were compared using χ² or Fisher’s exact tests, as appropriate. McNemar’s test was used to compare the accuracy of MINI<8 with FC<300 μg/g to detect MH. Of the entire derivation dataset, there were 26 missing values of any individual blood test (13 CRP, 8 ESR, 4 albumin, 1 platelets) which were imputed by a regression analysis using the other blood tests corrected for age, gender and FC value. For the validation dataset, 19 missing values were imputed (2 CRP, 11 ESR, 6 albumin). The ethics committees of all centers approved the Imagekids study and the validation cohorts. Consent, and when appropriate also assent, were obtained in all cases. All authors had access to the study data and approved the final manuscript.