Abstract & Commentary

Synopsis: Discontinuation of nonsteroidal anti-inflammatory drugs may increase the risk of first acute myocardial infarction in the near term. The risk is greatest for long-time users of these drugs and patients with systemic inflammatory diseases.

Source: Fischer LM, et al. Arch Intern Med. 2004;164:2472-2476.

In 2002, Schlienger, Jick, and Meier, co-authors of this study, made an unexpected discovery: Long-time users of nonsteroidal anti-inflammatory drugs (NSAIDs) had a 2.7 times increased risk of first-time acute myocardial infarction (AMI) within 30 days of discontinuation of the drug.1 That was a case-controlled study that used the United Kingdom General Practice Research Database (GPRD) and examined data from 1992 to 1997 with 3319 cases. The current study returns to the GPRD, examining data from 1995 to 2001. Inclusion criteria were age younger than 90 years and first-time diagnosis of AMI. For each case, 4 matched controls were randomly selected. Cases and controls were excluded if they were entered into the database for less than 3 years before the index date of the case’s AMI. Drug exposure was classified as nonuser, current user, user discontinued 1-29 days before AMI, discontinued 30-59 days, and discontinued 60 or more days. Users were grouped according to number of NSAID prescriptions: 1-19, 20-39, and 40 or more. There were 8,688 cases and 33,923 controls. Patients were mostly male (62.9%) and elderly (50% were 70 years or older).

As expected, patients who were currently smoking and those who formerly smoked had statistically significant increased odds ratios ([ORs] of 2.07 and 1.31, respectively) for first AMI compared to those who never smoked. Other statistically significant factors (and their ORs) were: body mass ratio > 30 (1.21), hypertension (1.26), hyperlipidemia (4.21), diabetes mellitus (1.84), ischemic heart disease (2.72), arrhythmias/congestive heart failure (1.46), arterial thrombosis (1.25), kidney diseases (1.23), systemic lupus erythematosus (2.80), and rheumatoid arthritis (1.47). An acute chest infection increased the risk of a first AMI, with the greatest risk among individuals who had an infection 1 to 4 days before the AMI (OR, 3.49). The risk attenuated with increasing time since the infection, but was still significantly increased even 10 to 14 days out (OR, 1.47).

Current NSAID users had a non-significant trend toward increased risk of first AMI. Subjects who had discontinued use within 1-29 days had an increased risk across all 3 prescription groups (OR, 1.52). The ORs increased with increasing prescription use and reached 2.60 for 40 or more prescriptions. For patients who had discontinued NSAID use within the last 30 to 59 days, the OR was 1.44. For patients who had discontinued use for 60 days or more, only those who had 40 or more prescriptions had a statistically significant increased risk (1.36).

When looking at diagnosis and drug discontinuation, patients with ischemic heart disease (IHD) who had stopped taking NSAIDs within 1 to 29 days had an OR of 2.85 compared to 1.46 for those without IHD who stopped within the same time period. Current aspirin use was protective of first AMI with an OR of 0.83. Patients who suffered from rheumatoid arthritis (RA) or systemic lupus erythematosus (SLE) who were not using NSAIDs had an OR of 1.66 compared to patients who were nonusers and didn’t have RA or SLE. Patients with RA or SLE who stopped taking NSAIDs had an OR of 3.68 for the 1 to 29 days following discontinuation. Those that continued use had an OR of 1.26.

Comment by Allan J. Wilke, MD

The GPRD is rapidly becoming my favorite clinical database. Imagine a database of more than 3 million people enrolled in primary care practices, complete with demographics, vitals, symptoms, diagnoses, hospitalizations, and detailed prescription records! Of course, it isn’t perfect. Prescription of a drug is not the same thing as taking a drug, and the database would not account for OTC NSAID use. Additionally, this type of study (case-controlled) cannot prove causal relationships; perhaps the reason that the patients quit using NSAIDs had more to bear on their eventual AMI than did the NSAIDs themselves.

Earlier studies2 have not shown NSAIDs to be protective of AMI, and, of course, use of rofecoxib (Vioxx®) has been associated with an increased risk of AMI3 and has been withdrawn from the market. (Fischer and colleagues note that during the time studied in this report COX-2 inhibitors were not in wide use, and there were not enough patients taking them to analyze their risk of AMI.)

Assume, though, that discontinuation of NSAIDs did cause the AMIs. Is there a plausible explanation for this? As reviewed in a recent Internal Medicine Alert,4 Pai and colleagues have shown that markers for inflammation, especially C-reactive protein, are elevated in coronary artery disease.5 Risk of AMI and stroke are both elevated after systemic respiratory infections, especially during the first 3 days of infection.6 Patients in this study with RA or SLE, who were not taking NSAIDs, were at higher risk of AMI, and that risk increased with discontinuation. Could the NSAIDs be keeping the lid on inflammation? On the other hand, current use of aspirin, but not NSAIDs, was protective of first AMI. In light of the preceding, aspirin’s effect probably is not secondary to its anti-inflammatory properties, and platelet inhibition seems less likely, because NSAIDs also inhibit platelets, albeit not irreversibly like aspirin.

So where does this leave us clinically? We should put this into some perspective. The ORs associated with NSAID use were not as large as those associated with hyperlipidemia, IHD, and acute chest infection, but they were similar to the risk of smoking. It seems prudent that we avoid sudden discontinuation of NSAIDs, especially in patients at risk for heart disease for other reasons, in patients who have used NSAIDs for long periods of time, and in patients with systemic inflammatory diseases, and we counsel our patients about the increased risk during the 2-month window after discontinuation.

Forewarned is forearmed.

Dr. Wilke, Associate Professor of Family Medicine, Medical College of Ohio, Toledo, OH, is Associate Editor of Internal Medicine Alert.


1. Schlienger RG, et al. Br J Clin Pharmacol. 2002;54: 327-332.

2. Solomon DH, et al. Arch Intern Med. 2002;162: 1099-1104.

3. Solomon DH, et al. Circulation. 2004;109:2068-2073.

4. Kuritzky L. Internal Medicine Alert. 2005;27:7.

5. Pai JK, et al. N Engl J Med. 2004;351:2599-2610.

6. Smeeth L, et al. N Engl J Med. 2004;351:2611-2618.