Incidence of symptomatic liver function abnormalities in a cohort of NSAID users

Background — A small number of published reports have provided estimates of incidence rates of symptomatic liver dysfunction in NSAID users, and most information has pertained to cases admitted to the hospital. Methods — We studied members of a health maintenance organization who were age 20 to 64 years in 1989–1991. Beginning with health insurance claims data for over 68,000 users of NSAIDs, and obtaining abstracts of medical records of potential cases, we ascertained the frequency of liver function abnormalities, including only cases with documented symptoms treated either as outpatients or hospital inpatients. Results — Overall, symptomatic liver dysfunction was rare, with 25 cases confirmed in the cohort of 68,028 people (0.37 cases per 1000 users of NSAIDs). For people who received a prescription NSAID in the previous six months, the crude incidence rate was 24 cases in 46,636 person-years, or 0.51 per 1000 person-years. For people who did not have a prescription NSAID in the previous six months, the crude incidence rate was 1 case in 13,303 person-years, or 0.075 per 1000 person-years. Of the 25 confirmed cases, 19 (76%) had alanine aminotransferase and/or aspartate aminotransferase at least 1.5 times the upper limit of normal values; the remaining six had abnormal results but of lesser magnitude. In five cases (20%) physician's notes document that a diagnosis of drug-induced hepatic dysfunction was considered. Only 3 (12%) of the 25 cases were admitted to the hospital. Conclusion — Although symptomatic liver dysfunction was relatively more frequent during the six months following NSAID use than at later times, the overall risk was low. Given the limitations of the available data, the apparent association with NSAID use may be confounded by other causes of liver dysfunction that could not be fully evaluated, such as other hepatotoxic medications, viral infections, or other medical conditions.