Background: To evaluate the safety of thiopurines in patients with inflammatory bowel disease. To identify predictive factors associated with the development of thiopurine-induced adverse events.

Methods: Long-term incidence of adverse events was estimated in patients from a prospectively maintained Spanish nationwide database using Kaplan-Meier analysis. Cox regression analysis was performed to identify potential predictive factors of adverse events.

Results: Three thousand nine hundred and thirty-one patients were included. Ninety-five percent of patients were on azathioprine. The median follow-up with thiopurines was 44 months (range, 0-420). Adverse events occurred at a median of 1 month after starting treatment. The cumulative incidence of adverse events was 26%, with an annual risk of 7% per patient-year of treatment. Most frequent adverse events were nausea (8%), hepatotoxicity (4%), myelotoxicity (4%), and pancreatitis (4%). Four patients had lymphoma. Female and Crohn's disease increased the risk of having nausea. The risk of hepatotoxicity was lower in females and higher in Crohn's disease. The risk of myelotoxicity was significantly higher in patients treated with mercaptopurine and in females. The risk of pancreatitis was higher in Crohn's disease. Overall, 17% of patients discontinued thiopurine treatment due to adverse events. Thirty-seven percent of these patients started thiopurines again and 40% of them had adverse events again.

Conclusions: As many as 1 of 4 patients on thiopurine therapy had adverse events during follow-up. A relatively high proportion of patients (17%) had to discontinue the treatment with thiopurines due to adverse events. However, more than half of patients that restarted thiopurine treatment after its discontinuation due to adverse events tolerated it. Several predictive factors for some adverse events have been identified.