Clinical observation of the temporal association between crack cocaine and duodenal ulcer perforation

HYPOTHESIS: To determine if a cause-effect relationship exists between crack cocaine use and duodenal ulcer perforation (DUP). PATIENTS AND METHODS: A retrospective study was conducted of all patients undergoing emergency surgical management for peptic ulcer disease over a 6-year period at a large inner-city municipal teaching hospital. The hospital records of 78 consecutive patients presenting with complications of peptic ulcer disease between April 1990 and April 1996 were reviewed. Group A (n = 24) consisted of patients with confirmation of crack cocaine usage within 8 hours of clinical presentation; group B (n = 54) consisted of patients with no antecedent history of crack cocaine use. Demographic data, timing of drug use, clinical presentation, laboratory and radiographic findings, toxicology screening, operative findings, and postoperative course were compared between the two groups. RESULTS: Both groups revealed a similar gender distribution, tobacco use, prior peptic ulcer symptoms, and laboratory findings. Group A patients were younger (t test, P = 0.01) and more likely to present with perforation, whereas patients in group B presented with a combination of symptoms (chi square, P = 0.03). Duodenal ulcer perforation was present in 75% of patients in group A compared with 46% of patients in group B (chi square, P = 0.04). Group B patients had a significantly longer hospital stay compared with those in group A (t test, P = 0.01). Both crack cocaine and alcohol are independent predictors of duodenal ulcer perforation. CONCLUSIONS: Patients with recent use of crack cocaine and/or alcohol are more likely to present with duodenal perforations. Although a temporal association between crack cocaine use and duodenal ulcer perforation was demonstrated, this study does not confirm a cause-effect relationship. A prospective cohort study is needed to clarify the pathogenesis of this potential cause-effect relationship.