Background: An association between tobacco smoking and outcomes in IBD has been established. We aimed to investigate the impact of smoking status at IBD diagnosis on clinical outcomes including corticosteroid (CS) and thiopurine (TP) use as well as the need for surgery using a nationally representative research database.
Methods: Using the UK Clinical Practice Research Datalink, which contains data for approximately 8% of the UK population, we identified incident cases of patients with IBD between 1998 and 2014. Prescription data for IBD medications including CS and TP were obtained. Surrogate markers for disease severity included early CS use (within 3 months of diagnosis) in ulcerative colitis (UC) and the modified Beaugerie Index in Crohn's disease (CD). Smoking status at diagnosis was defined using Read codes for tobacco consumption. Patients were classified as either smokers or non-smokers at IBD diagnosis. The surgical endpoints were colectomy in UC and first intestinal resection (IR) in CD. Medical endpoints included CS use and TP use. Kaplan Meier survival analysis was used to study differences in the rates of surgery and medication use between smoking subgroups. A multivariate Cox regression model was used to calculate the risk of surgery given smoking status at IBD diagnosis.
Results: 12302 patients with IBD were identified (UC 7497, CD 4805). There were fewer smokers at diagnosis with UC compared to CD (16.5% vs 30.2%, p<0.0001). The proportion of male smokers at diagnosis was higher in UC than CD (56.4% vs 41.2%, p<0.0001). There were no differences in disease severity indices between smokers and non-smokers in either UC or CD. The 1, 3 and 5 year cumulative probability for IR in CD was 5.3%, 9.0% and 10.1% in smokers compared to 4.1%, 6.0% and 7.7% in non-smokers (log rank test for trend, p=0.008). The 1, 3 and 5 year cumulative probability of CS use in CD was 27.1%, 34.6% and 39.4% in smokers and 24.2%, 30.7% and 35.2% in non-smokers (p=0.01). Furthermore, the 1, 3 and 5 year cumulative probability of TP use in CD was 24.3%, 34.1% and 35.3% in smokers compared to 21.7%, 28.5% and 31.9% in non-smokers (p=0.001). In patients with UC, there was no difference in the rates of colectomy, CS or TP use between smokers and non-smokers. In the multivariate Cox regression analysis for patients with CD, smokers at IBD diagnosis had a 27% increased risk of IR compared to non-smokers (HR 1.27, 95% CI 1.02–1.56, p=0.03). No association was found in a similar analysis for colectomy in UC patients.
Conclusions: In Crohn's disease, smoking at diagnosis is associated with increased steroid use, thiopurine use and intestinal resection. Conversely, smoking at diagnosis in ulcerative colitis does not appear to impact on these outcomes.