Search in the Abstract Database

Search Abstracts 2016

* = Presenting author

P697 Highlighting of epidemic areas of Crohn’s disease in a population-based registry over 22 years: genetic or environmental cause?

M. Genin*1, 2, C. Vignal1, 3, F. Vasseur2, M. Fumery4, G. Savoye5, M. Body-Malapel1, 3, C. Preda6, H. Sarter7, L. Peyrin-Biroulet8, P. Desreumaux3, 9, C. Gower-Rousseau1, 7, 10

1University Hospital, Lille, France, 2CHRU Lille, Department of Biostatistics EA 2694, Lille, France, 3Inserm 995 LIRIC Team 5, Lille, France, 4University Hospital, Gastroenterology, Amiens, France, 5University Hospital, Gastroenterology, Rouen, France, 6UFR de Mathématiques, Lille, France, 7University, Inserm 995 LIRIC Team 5, Lille, France, 8University Hospital, Gastroenterology, Nancy, France, 9University Hospital, Gastroenterology, Lille, France, 10CHRU Lille, Department of Epidemiology, EPIMAD Registry, Lille, France


Our previous studies detected a spatial heterogeneity in standardised incidence ratios of Crohn’s disease (CD), through a population-based IBD registry over a 17-year period, pointing the existence of clusters with low and high incidence. The first aim of the present study was to determine the origin of these clusters according with CD family history (CD-FH) or not over a 22-year period (1990–2011). The second aim of this study was to compare the clinical phenotype at CD diagnosis between clusters with high and low CD incidence.


From 1990 to 2011, the Epimad population-based registry recorded 8 970 incident CD cases distributed in 273 administrative areas of Northern France. Isotonic scan statistics allowed the detection of clusters and their epicentre. Several data were collected at time of diagnosis, including gender, age, CD-FH, smoking status, phenotype (location, behaviour, and anoperineal lesion) according to Montreal classification, and diagnosis management.


High (warm colours) and low (cold colours) incidence clusters including all patients (n = 8 970) are depicted in Figure 1. Figure 2 shows the clusters detected after exclusion of patients with CD-FH (n = 1 086) prone to have a genetic burden. Seven clusters have been identified in both analyses; 4 with high incidence (2 163 CD cases; RR from 1.27 to 1.46; p < 103–) and 3 with low incidence (861 CD cases; RR from 0.69 to 0.71; p < 102–). The size of cluster with high incidence in the south-eastern part of the area was greatly reduced when excluding CD patients with CD-FH. Concerning clinical parameters; gender, median age at diagnosis, smoking status and CD phenotype were not different between clusters with high and low incidences.

Figure 1. Space clusters including all patients with CD from 1990 to 2011 (n = 8 970).

Figure 2. Space clusters including CD patients without known CD-FH from 1990 to 2011 (n = 7 884).


Four clusters of epidemic CD areas have been identified in our population-based study over a 22-year period. Only 1 cluster was reduced in size when excluding CD-FH patients. These results assumed for a high effect of environmental risk factors in the origin of these clusters. It is now our challenge to highlight them.