11] Leeds Gastroenterology Institute, St James's University Hospital, Leeds, UK  Leeds Institute of Biomedical and Clinical Sciences, University of Leeds, Leeds, UK.
2Gastroenterology Division, Health Sciences Center, Farncombe Family Digestive Health Research Institute, McMaster University, Hamilton, Ontario, Canada.
3Gastroenterology Department, St Joseph's Healthcare, Hamilton, Ontario, Canada.
There are few data concerning the utility of symptoms and signs at first presentation in predicting a diagnosis of ulcerative colitis (UC) or Crohn's disease (CD). We conducted a study to examine this issue in secondary care.
We collected complete symptom, colonoscopy, and histology data prospectively from 1,981 consecutive adult patients with lower gastrointestinal symptoms at two hospitals in Hamilton, Ontario. Assessors were blinded to symptom status. The reference standard used to define the presence of UC or CD was according to accepted histological criteria. Patients without UC or CD served as controls. Sensitivity, specificity, and positive and negative likelihood ratios (LRs) were calculated for individual items from the clinical history, as well as combinations of these.
In identifying 302 patients with inflammatory bowel diseases (IBD), positive LRs for individual items ranged from 1.18 (incomplete emptying) to 2.30 (passage of stools more than four times per day at least most of the time) and negative LRs from 0.70 (bloody stools) to 0.96 (incomplete emptying). Combinations of items had a high specificity, but at the expense of sensitivity. Items that were independent predictors of IBD after logistic regression analysis were family history of IBD, younger age, passage of stools more than four times per day ≥75% of the time, urgency most of the time, and anemia.
Individual items from the clinical history are not helpful in predicting a diagnosis of UC or CD. However, this may be because some items lacked sufficient detail. Combinations of symptoms and computer models had a high specificity, but overall were only modestly useful diagnostically. Future studies should evaluate biological markers in combination with symptoms to improve accuracy.