P488 Development and validation of a risk prediction tool for inflammatory bowel disease (IBD) diagnosis in patients presenting in primary care with abdominal symptoms, a UK based study

Abstract

Abstract Background There are no prediction models for a diagnosis of inflammatory bowel disease (IBD) in primary care. An IBD risk prediction tool has the potential to reduce the length of time patients have undiagnosed IBD symptoms and improve IBD clinical outcomes. Therefore, our aim was to develop and validate a risk prediction model for the diagnosis of IBD, ulcerative colitis (UC) and Crohn’s disease (CD) in men and women separately. Methods A population-based retrospective open cohort study using Clinical Practice Research Datalink (CPRD) Aurum database was undertaken between 1st January 2010 and 31st December 2019, including patients aged 18 years or older. Patients were followed from first presentation with lower gastrointestinal (GI) symptoms potentially related to IBD to the earliest of IBD diagnosis, loss to follow-up, death or study end. 2,054,530 patients were in the model derivation cohort and 673,320 patients in the validation cohort. Cox proportional hazards models were used to derive separate risk equations in men and women for IBD, UC and CD. Candidate predictors included demographic factors (age, sex, smoking, body mass index, Charlson comorbidity score, loperamide use), family history of IBD, co-existing conditions (anxiety, depression, irritable bowel syndrome (IBS), haemorrhoids), extraintestinal manifestations (EIM) (mouth ulcers, ophthalmic, primary sclerosing cholangitis, dermatological), investigations (hemoglobin, mean corpuscular volume, platelets, albumin, vitamin B12, ferritin, C-reactive protein, erythrocyte sedimentation rate, calprotectin level). Measures of calibration and discrimination (C-statistic and D-statistic, higher values indicate better discrimination) were determined in men and women separately in the validation cohort at 1,2,3 and 5 years after symptom presentation. Results In the derivation cohort, 15,105 (0.7%) patients had an IBD diagnosis (10.014 UC (66.3%) and 5,088 CD (33.7%)). The IBD prediction model included 41 and 40 predictors, and the UC model 37 and 36 predictors, in women and men respectively; the CD model included 37 predictors in both men and women. C-statistics and D-statistics in men were as follows: IBD model 0.77 and 1.74; UC model: 0.81 and 1.95; and CD model: 0.77 and 1.95, respectively. Similar values were observed in women.The measures of discrimination showed that the prediction models reliably differentiated patients with and without IBD, UC and CD in both sexes. Model calibration was good, tending to overestimate at higher risk thresholds in validation cohort. Conclusion A risk model of patient demographics, symptoms and investigations performs well for IBD, UC and CD and may help in prioritising suspected IBD referrals in symptomatic subjects in primary care.