Background: The pathogenesis of inflammatory bowel disease probably involves an interaction between genetic and environmental factors: cigarette smoking, appendectomy and oral contraceptives have been the factors most frequently linked to its aetiology
Aim: To analyse the association between known environmental risk factors and development of Crohn's disease in the community of Aragón, Spain.
Patients and methods: A case-control, population-based study has been carried out. All patients diagnosed with Crohn's disease in the community of Aragón from 1st February 1992 to 31st January 1995 were prospectively included. The Lennard Jones criteria were used to define the cases and selected controls among the healthy population matched with patients for age, sex and rural/urban habitat. Statistical analysis included multivariate analysis using conditional logistic regression, testing 38 different models.
Results: A total of 103 patients were diagnosed with Crohn's disease in Aragón from 1st February, 1992 to 31st January, 1995. Of these 62 patients (60.2%) with Crohn's disease were smokers, compared with 42 (40.8%) controls (p<0.001). Cigarette smoking is considered a risk factor for Crohn's disease with an odds ratio of 3.09 (95% confidence interval, 1.58-6.05). After multivariate analysis, the positive association is maintained. A dose-dependent relation could not be demonstrated. No statistical differences (p=0.50) were detected in the analysis of previous appendectomy. Use of oral contraceptive acts as a risk factor with a p=0.048; odds ratio 2, 8, 95% confidence interval: 1.009-7.774; but this association disappears in the multivariate analysis. Eight patients had a family history (3 first degree and 5 second degree relatives) versus none of the controls (p=0.002). Of the variables studied for childhood hygiene none appeared significant.
Conclusion: Smoking, family history, and oral contraceptive use, appear as risk factors for developing Crohn's disease in univariate analysis, but only smoking remains significant in all models of multivariate analysis.