Search in the Abstract Database

Abstracts Search 2019

P595 Early medical therapy and risk of subsequent perianal fistula development among paediatric patients with Crohn’s disease

J. Adler*1,2, C. C. Lin3, S. Gadepalli2,4, K. Dombkowski2

1University of Michigan, Pediatric Gastroenterology, Ann Arbor, USA, 2University of Michigan, Child Health Evaluation and Research (CHEAR) Center, Ann Arbor, USA, 3University of Michigan, Health Services Research Program, Department of Neurology, Ann Arbor, USA, 4University of Michigan, Pediatric Surgery, Ann Arbor, USA


Crohn’s disease (CD) commonly causes perianal fistulas (PF). Evidence for effective PF preventive strategies is lacking. Early steroid sparing therapy (SST) use improves other CD outcomes and decreases steroid use. We sought to determine whether early SST prevents or delays PF development among children without PF at CD diagnosis.


We identified patients with CD age 5–24 years in OptumInsight Clinformatics Data Mart (2001–2014). Diagnosis required minimum 3 CD claims within 3 years of initial diagnosis. CD diagnosis date was the first occurrence of any inflammatory bowel disease claim. We required 6 months run-in period before, and 2 years continuous follow-up after diagnosis. PF was identified by previously validated claims-based case definition (perianal/genital fistula/abscess, or seton/fistulotomy). Patients were excluded if PF <90 days after CD diagnosis. SST was defined as immunomodulator and/or anti-tumour necrosis factor (TNF) medication. Early SST use was characterised within 6 months after CD diagnosis, and only if started prior to PF. PF development was compared for those with/without early SST. Propensity score (PS) matching for patient characteristics/comorbidities was used to balance baseline characteristics. Cox multi-variate regression analysis estimated hazard ratios (HR) for PF development.


We identified 2378 patients with CD (mean age 17 years, 49% female, 78% white). PF developed after diagnosis among 342 (14%). Overall, 40% initiated early SST (immunomodulator alone 28%, anti-TNF 11%); rates increased two-fold from 22% in 2001 to 45% in 2014 (trend test p < 0.001). There were differences in early SST use by sex (female 36%, male 43%; p < 0.001) and household income (greater income, higher risk SST; p = 0.002). Higher income was associated with lower risk of PF (HR = 0.55, p = 0.0495). Among patients diagnosed in later years there was a trend toward lower risk of PF with early SST (average 2001–2011: 11.5% vs. 2012–2014: 6.2%). After PS matching, 942 patients remained in each group. Early SST was not associated with risk of PF development in 3 yr after CD diagnosis (HR = 1, 95% CI = 0.75–1.34, p = 0.98). Antibiotic use was associated with 57% greater risk of developing PF (HR = 1.57, 95% CI = 1.18–2.09, p = 0.002). For each year a patient aged, risk of developing PF increased 4% (HR = 1.04, 95% CI = 1.01–1.08, p = 0.01). After PS matching, there were no longer differences by sex, income, or year of diagnosis.


We found that among paediatric CD patients without PF at diagnosis, later PF development is common. SST use soon after CD diagnosis is uncommon, but increasing. Changing trends in use of early SST may lead to decreased PF development, although this is not conclusive. Further research is needed to identify optimal treatment for PF prevention.