ORIGINAL INVESTIGATION Discontinuation of Nonsteroidal Anti-inflammatory Drug Therapy and Risk of Acute Myocardial Infarction Lorenz M. Fischer, MSc; Raymond G. Schlienger, PhD, MPH; Christian M. Matter, MD; Hershel Jick, MD; Christoph R. Meier, PhD, MSc Background: Systemic inflammation has been shown risk of AMI was 1.52 (95% confidence interval [CI], 1.33- to be associated with an increased risk of acute myocar- 1.74) for subjects who stopped taking NSAIDs 1 to 29 dial infarction (AMI). However, the effect of the use of days prior to the index date, compared with nonusers. nonsteroidal anti-inflammatory drugs (NSAIDs) on the The risk was highest in subjects with rheumatoid arthri- risk of AMI has not yet been well defined. We therefore tis or systemic lupus erythematosus (adjusted OR, 3.68 studied the risk of AMI during NSAID exposure and af- [95% CI, 2.36-5.74]) and for subjects who discontin- ter the cessation of NSAID therapy. ued therapy with NSAIDs after previous long-term use (adjusted OR, 2.60 [95% CI, 1.84-3.68]). Current and Methods: We conducted a large case-control analysis on past NSAID use (discontinued therapy Ն60 days prior the British General Practice Research Database. The study to the index date) were not associated with an increased included 8688 cases with a first-time AMI between 1995 risk of AMI (adjusted OR, 1.07 [95% CI, 0.96-1.19] and and 2001 and 33923 controls, matched to cases on age, 1.05 [95% CI, 0.99-1.12], respectively). sex, calendar time, and general practice attended. Conclusion: Our findings suggest that the risk of AMI Results: After adjusting for hypertension, hyperlipid- is increased during several weeks after the cessation of emia, diabetes mellitus, ischemic heart disease, rheuma- NSAID therapy. toid arthritis, systemic lupus erythematosus, acute chest infection, body mass index, smoking, and aspirin use, the Arch Intern Med. 2004;164:2472-2476 HERE IS INCREASING EVI- thors concluded that current exposure to dence that intravascular in- nonaspirin NSAIDs does not substan- flammation plays a key role tially lower the risk of AMI.5-8 in the development of ath- However, a possible limitation of these erosclerosis and acute studies is that it is difficult to distinguish Tcoronary events.1-3 Nonsteroidal anti- between the effect of the underlying in- inflammatory drugs (NSAIDs) are widely flammation—a main reason for using used for the treatment of pain and inflam- NSAIDs—and the potential NSAID effect Author Affiliations: Basel mation. They exert their effect by revers- on the AMI risk, since the 2 are highly cor- Pharmacoepidemiology Unit, ible, competitive inhibition of cyclooxy- related. Relative risks around 1.0 for cur- Division of Clinical genase (COX), an important enzyme in the rent NSAID use may also be the result of Pharmacology and Toxicology, regulation of molecular pathways of pain an NSAID effect; in other words, current University Hospital Basel, Basel, 4 Switzerland (Mr Fischer and and inflammation. In addition to COX in- NSAID exposure may lower an inflamma- Drs Schlienger and Meier); hibition, NSAIDs also decrease throm- tion-induced increased risk of AMI risk to- Cardiovascular Research, boxane A2 production, potentially lead- ward 1.0, but not below. Institute of Physiology, ing to an inhibition of platelet aggregation.4 In a recent study, we explored the effect University of Zurich, and In theory, these 2 pharmacological mecha- of current NSAID use on the risk of AMI Division of Cardiology, nisms could reduce the risk of acute myo- in 3319 cases with a first-time AMI be- Cardiovascular Center, cardial infarction (AMI) during exposure tween 1992 and 1997 and 13139 con- University Hospital Zurich, to nonaspirin NSAIDs. trols using the United Kingdom (UK)– Zurich, Switzerland In fact, several recent observational based General Practice Research Database (Dr Matter); and Boston studies explored the risk of AMI in sub- (GPRD).6 Study subjects were free of di- Collaborative Drug Surveillance 5-10 Program, Boston University jects taking nonaspirin NSAIDs. The agnosed cardiovascular or metabolic risk Medical Center, Lexington, relative risk estimates for current NSAID factors. We reported a relative risk close Mass (Drs Jick and Meier). use in these studies were consistently re- to 1.0 for current NSAID use, but ob- Financial Disclosure: None. ported to be around 1.0, and most au- served a more than 2-fold increased risk (REPRINTED) ARCH INTERN MED/ VOL 164, DEC 13/27, 2004 WWW.ARCHINTERNMED.COM 2472 ©2004 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/01/2021 of AMI for long-term users of NSAIDs who discontin- sis of the corresponding case). We also excluded controls with ued NSAID therapy before the AMI.6 a history of less than 3 years in the GPRD. The aim of the present study was to further assess the association between timing of discontinuation of NSAID EXPOSURE DEFINITION exposure and the risk of first-time AMI. For this pur- pose, we conducted another large case-control analysis Based on the number of tablets and the GP’s intake regimen of on the GPRD, including incident AMI cases from 1995 the last NSAID prescription prior to the index date, we to 2001 with or without clinical risk factors for AMI. assessed the number of days between the NSAID discontinua- tion and the index date for each case and control. A subject was defined as “current user” if the supply of the last prescrip- METHODS tion for an NSAID lasted up to the index date or beyond. Sub- jects whose therapy ended before the index date were catego- STUDY POPULATION AND DATA SOURCE rized according to the time lag between the end of therapy and the index date (1-29, 30-59, Ն60 days). Subjects were The GPRD is a large and well-validated database, which has been further classified according to the number of prescriptions for previously described in detail.11,12 Briefly, more than 3 million NSAIDs (ie, 1-19, 20-39, Ն40 prescriptions for acemetacin, residents in the UK have been registered with selected general diclofenac, diflunisal, etodolac, fenbufen, fenoprofen, flur- practitioners (GPs) who agreed to provide data for research pur- biprofen, ibuprofen, indomethacin, ketoprofen, mefenamic poses to the GPRD. The database has been the source of nu- acid, nabumetone, naproxen, piroxicam, sulindac, tenoxicam, merous epidemiological studies, and the accuracy and com- or tiaprofenic acid). pleteness of the data have been well documented and validated.13,14 Data from the GPRD have been used in several recent studies on AMI.6,15-19 The age- and sex-distribution of STATISTICAL ANALYSIS patients in the GPRD is representative of the UK population. The information electronically recorded by GPs includes pa- We conducted a matched analysis (conditional logistic regres- tient demographics and characteristics (eg, height, weight, and sion model) using the software program SAS, version 8.1 (SAS smoking status), symptoms, clinical diagnoses, referrals to con- Institute Inc, Cary, NC). Relative risk estimates (odds ratios sultants, hospitalizations, and drug prescriptions. Drug pre- [ORs]) are presented with 95% confidence intervals (CIs). scriptions are recorded in detail using a drug dictionary based For each case and control, the independent effects of vari- on the UK Prescription Pricing Authority. These codes define ous potential confounders on the AMI risk were assessed, for each prescription the active compound, the route of ad- such as body mass index (BMI, calculated as weight in kilo- Ͻ ministration, the dose of a single unit, the number of units pre- grams divided by the square of height in meters) ( 25, Ն scribed, and in most instances the intake regimen prescribed 25-29.9, 30, or unknown), smoking status (never, by the GP (eg, 3 tablets per day). Drug prescriptions are gen- exsmoker, current, or unknown), aspirin use, hypertension, erated directly from the computer and recorded in each pa- hyperlipidemia, diabetes mellitus, ischemic heart disease, tient’s computerized profile. On request, hospital discharge and other cardiac diseases (arrhythmias or congestive heart fail- referral letters are available for review to validate the diag- ure), arterial vascular diseases (claudication, stroke, transient noses recorded in the computer record. ischemic attack, or arterial thromboembolic events), kidney diseases, acute chest infection, and diseases with systemic inflammation (rheumatoid arthritis or systemic lupus ery- CASE DEFINITION AND ASCERTAINMENT thematosus [SLE]). We identified potential cases with a first-time diagnosis of AMI via computer-recorded Oxford Medical Information System RESULTS [OXMIS] codes, mapped onto International Classification of Dis- eases [ICD] codes. We searched for patients younger than 90 The analysis encompassed 8688 cases with a first-time years who had a first-time AMI between 1995 and 2001. We AMI and 33923 matched controls. Table 1 displays the excluded individuals who were registered on the database for age and sex distribution of cases and controls as well as less than 3 years before the date of the AMI (subsequently re- ferred to as index date). We reviewed the computer records of their smoking status, BMI, and presence of cardiovascu- all potential cases, whereby any information regarding NSAID lar or metabolic diseases related to an altered AMI risk. exposure was concealed. In previous studies using GPRD Patients were predominantly male (62.9%), and 50.0% data,16-18 the computer-recorded diagnosis of a first-time AMI were at 70 years or older at the date of the AMI. was validated for a random sample of approximately 450 pa- tients by reviewing hospital discharge letters. When we se- INCREASED RISK OF FIRST-TIME AMI lected cases based on a manual review of computer records and AFTER DISCONTINUATION sent for hospital discharge letters, more than 90% of cases were OF NSAID THERAPY confirmed by the presence of characteristic diagnostic crite- 16-18 ria.