The Authors Respond:
Drs. Werner, Romsing, and Nielsen raise some concerns about the validity of our recent findings. 1 We would like to clarify that one of the conclusions of our paper 2 was that acetaminophen was associated with an increased risk of upper gastrointestinal bleeding and perforation (UGIC) at daily doses starting at 2 mg and above. We saw no increased risk at doses lower than 2 mg.
Specifically, the correspondents suggest three possible sources of bias: (1) confounding by use of selective serotonin reuptake inhibitors (SSRIs), (2) confounding by indication, and (3) misclassification of exposure. We respond below to each of their points:
1. Neither adjusting for SSRI use nor exclusion of SSRI users modifies the association between high-dose acetaminophen and UGIC. Therefore, confounding by SSRI use was not an important source of bias. One correction: we had previously reported 3 a relative risk of 15.6 (95% confidence interval from 7 to 37) among patients taking concomitantly a nonsteroidal antiinflammatory drug (NSAID) and a selective serotonin reuptake inhibitor (SSRI). Perhaps these authors confused our point estimate with the upper bound of its confidence interval.
2. We, too, were concerned about the preferential prescribing (or avoidance) of NSAIDs or acetaminophen to patients with gastrointestinal symptoms. For that reason we manually reviewed patients’ profiles to confirm the indication for use. Of note, pain related to osteoarticular conditions was the indication in 80% of acetaminophen users. Moreover, no important differences were found between indications for NSAIDs or for acetaminophen. Secondly, we adjusted our estimates for the effects of a patient's upper gastrointestinal medical history, including symptoms of gastritis and dyspepsia. These adjustments did not explain the association between acetaminophen and UGI complications. Therefore, confounding by indication does not appear to represent a main source of bias in our study.
3. One would hope that prescribing habits in clinical practice would more strictly follow accepted guidelines such as those recommended in the British National Formulary, but concordance between real life and guidelines is seldom perfect. We would expect some degree of noncompliance, as well as over-the-counter use. In general, we would expect over-the-counter use to be more common among patients who did not receive prescriptions and were consequently classified as nonusers. Therefore, this type of misclassification would be expected to bias the results towards the null, and the relative risk reported for high-dose acetaminophen could underestimate the true magnitude of the association.
Scientific evidence, particularly in the area of drug safety, does not come only from well performed prospective controlled randomized trials but also from well performed observational studies like many of the ones published in this journal. Novel findings should always be subjected to criticism and replication. Whether or not the potential increased risk observed for high-dose acetaminophen and UGIC withstands this scrutiny, as did the epidemiologic evidence for NSAIDs and UGIC, only time will tell.
Luis Alberto García Rodríguez
1. Werner MU, Rømsing J, Nielsen PR. Acetaminophen and upper gastrointestinal complications [letter]. Epidemiology 2002; 13: 605–606.
2. García Rodríguez LA, Hernández-Díaz S. Relative risk of upper gastrointestinal complications among users of acetaminophen and nonsteroidal anti-inflammatory drugs. Epidemiology 2001; 12: 570–576.
3. de Abajo FJ, García Rodríguez LA, Montero D. Association between selective serotonin reuptake inhibitors and upper gastrointestinal bleeding: population based case-control study. BMJ 1999; 319: 1106–1109.