Use of nonsteroidal anti-inflammatory drugs and the risk of first-time acute myocardial infarction


  • Raymond G. Schlienger,

    1. Basel Pharmacoepidemiology Unit, Division of Clinical Pharmacology and Toxicology, University Hospital, Basel, Switzerland and
    Search for more papers by this author
  • Hershel Jick,

    1. Boston Collaborative Drug Surveillance Program, Boston University Medical Center, Lexington, MA, USA
    Search for more papers by this author
  • Christoph R. Meier

    1. Basel Pharmacoepidemiology Unit, Division of Clinical Pharmacology and Toxicology, University Hospital, Basel, Switzerland and
    2. Boston Collaborative Drug Surveillance Program, Boston University Medical Center, Lexington, MA, USA
    Search for more papers by this author

Christoph R. Meier, PhD, M.Sc., Basel Pharmacoepidemiology Unit, Division of Clinical Pharmacology and Toxicology, University Hospital of Basel, Petersgraben 4, CH −4031 Basel, Switzerland. Tel: + 41 61 265 88 70; Fax: + 41 61 265 88 64; E-mail:


Aims Aspirin decreases the risk of clinical manifestations of atherothrombosis. This effect is mainly due to inhibition of platelet aggregation and potentially due to anti-inflammatory properties of aspirin. To evaluate whether use of non-aspirin non-steroidal anti-inflammatory drugs (NSAIDs) may also be associated with a decreased risk of first-time acute myocardial infarction (AMI), we performed a population-based case-control analysis using the United Kingdom-based General Practice Research Database (GPRD)

Methods We identified first-time AMI-patients free of preexisting diagnosed cardiovascular or metabolic diseases. We compared use of NSAIDs prior to the index date between cases and control patients who were matched to cases on age, gender, practice and calendar time.

Results A total of 3319 cases (≤75 years) with a diagnosis of first-time AMI between 1992 and 1997 and 13139 controls (matched to cases on age, sex, general practice attended, calendar time, years of prior history in the GPRD) were included. Overall, the relative risk estimate of AMI (adjusted for smoking, body mass index, hormone replacement therapy and aspirin) in current NSAID users was 1.17 (95% CI 0.99, 1.37). Long-term current NSAID use (≥30 prescriptions) yielded an adjusted odds ratio (OR) of 1.20 (95% CI 0.94, 1.55). Stratification by age (<65 years vs≥65 years) and sex did not materially change the results.

Conclusions Our findings indicate that current NSAID exposure in patients free of diagnosed cardiovascular or metabolic conditions predisposing to cardiovascular diseases does not decrease the risk of AMI.


Aspirin has been thoroughly evaluated as an antiplatelet drug to prevent or treat atherothrombosis [1]. It has been found to prevent vascular death by about 15% and nonfatal vascular events by about 30% in a recent meta-analysis of more than 50 secondary prevention trials [2]. The antiplatelet effect of aspirin is due to irreversible inhibition of cyclooxygenase-1 (COX-1) in platelets. This results in an inhibition of platelet thromboxane A2 (TXA2) production, a potent inducer of platelet aggregation [3]. Additionally, anti-inflammatory effects by inhibition of cyclooxygenase-2 (COX-2) activity have been suggested as a potential additional mechanism of aspirin for preventing ischaemic heart disease [4, 5]. Studies have indicated that the baseline plasma concentration of C-reactive protein (CRP), a systemic marker for underlying inflammation, may predict the risk of future acute myocardial infarction (AMI) [5, 6] and that the risk reduction associated with aspirin use appears directly related to the CRP level [5]. Thus, it has been suggested that anti-inflammatory agents other than aspirin may also have a role in preventing cardiovascular disease [5].

Non-aspirin nonsteroidal anti-inflammatory drugs (NSAIDs) are widely used in a variety of disorders associated with inflammation and acute or chronic pain. The principal pharmacological mechanism by which NSAIDs exert their therapeutic effect is by reversible, competi-tive  COX-inhibition.  With  the  exception  of  newer, more selective COX-inhibitors [7], most currently used NSAIDs inhibit COX-1 and COX-2 nonselectively [8]. By reversible inhibition of COX-1, NSAIDs too may decrease the production of   TXA2 in platelets and inhibit platelet aggregation [1, 9]. In addition, due to their anti-inflammatory effects NSAIDs might also reduce the risk of AMI through reduction of chronic systemic inflammation. However, information is scarce whether nonaspirin NSAIDs may have beneficial cardioprotective effects of clinical relevance. In a recent nested case-control analysis exposure  to  nonaspirin  NSAIDs  did  not  alter  the  risk of first-time myocardial infarction in postmenopausal women compared with nonusers of NSAIDs [10]. The study included women with major risk factors for myocardial infarction such as diabetes mellitus, hypertension, and family history of coronary heart disease [10]. Another recent epidemiological study in patients 50–84 years of age including patients with preexisting cardiovascular disease found no evidence that exposure to nonaspirin NSAIDs reduces the risk of serious coronary heart disease events [11].

The aim of the present study was to evaluate further whether current and/or long-term exposure to nonaspirin NSAIDs may modify the risk of first-time AMI in both men and women 75 years of age or younger who are free of diagnosed cardiovascular or metabolic conditions predisposing to cardiovascular disease.


Study population and data source

Data were derived from the United Kingdom-based General Practice Research Database (GPRD), which has been previously described in detail elsewhere [12–14]. Since 1987, more than 3 million residents in England and Wales have been registered with selected general practitioners (GPs) who have agreed to provide data for research purposes to the GPRD. The age- and sex-distribution of the patients enrolled is representative of the entire UK-population. The information recorded includes patient demographics and characteristics (e.g. height, weight, smoking status), symptoms, clinical diagnoses, consultant referrals, hospitalizations, and drug prescriptions.  A  coded  drug  dictionary  based  on  the UK Prescription Pricing Authority dictionary is used for recording prescriptions including information on the route of administration, dose, and number of tablets for each prescription. The GPs generate prescriptions directly from the computer and they are recorded in each patient's computerized profile. On request, hospital discharge and referral letters are available for review to validate the diagnoses recorded in the computer record.

The GPRD is administered by the Medicines Control Agency. The database currently encompasses some 30 million person-years of follow-up; it has been the source for numerous epidemiological studies in recent years, and the accuracy and completeness of these data have been well documented and validated [15, 16]. GPRD data have  been  used  in  several  recent  studies  investigating risk factors for AMI [17–20] or effects of NSAIDs [21, 22].

Case definition and ascertainment

Potential cases were selected on the basis of a first-time diagnosis of AMI (by computer-recorded Oxford Medical Information System [OXMIS] codes, mapped onto International Classification of Diseases [ICD] codes) between January 1, 1992, and October 31, 1997. We restricted the study to patients who were ≤75 years of age at the time of the diagnosis of AMI (index date), and who were free of metabolic or cardiovascular diseases predisposing to AMI. Therefore, all patients with a diagnosis of AMI, angina pectoris, unexplained chest pain, cardiac arrhythmias, congestive heart failure, stroke, intermittent claudication, venous thromboembolism, chronic renal disease, hypertension, hyperlipidaemia, diabetes mellitus, or connective-tissue disorders recorded >60 days before the AMI were excluded. All cases had to be registered on the database for at least 3 years before the index date. Any information with regard to exposure to NSAIDs was concealed when the records were reviewed to identify potential cases.

In previous studies using GPRD data [17–20], the computer-recorded diagnosis of a first AMI was validated for a random sample of approximately 450 patients by reviewing hospital discharge letters. The validity of the AMI-diagnosis was confirmed for a high percentage (>90%) of cases identified on computer and after review of computer records by at least two of the following documented diagnostic criteria: Characteristic chest pain, characteristic changes in the electrocardiogram, characteristic serial rises in the concentrations of cardiac enzymes, an arteriogram documenting a recent coronary occlusion, or fibrinolytic therapy. Therefore, we decided to include all the potential cases that we identified through a manual review of computerized patient records.


Four controls were matched to each case on age (same year of birth), sex, the practice attended, and calendar time by using the same index date (i.e. the date of the AMI-diagnosis of the corresponding case) for matched controls as for cases. The same exclusion criteria were applied to controls as to cases (i.e. recorded history on the GPRD of less than 3 years, and/or circulatory or metabolic diseases predisposing for AMI recorded >60 days before the index date).

Exposure definition

For each case and control the exposure history to NSAIDs (i.e. acemetacin, diclofenac, diflunisal, etodolac, fenbufen, fenoprofen, flurbiprofen, ibuprofen, indomethacin, ketoprofen, mefenamic acid, nabumetone, naproxen,  piroxicam,  sulindac,  tenoxicam  and  tiaprofenic acid) was assessed. A patient was defined as ‘current user’ if the supply of the last prescription for an NSAID prior to the index date ended at or after the index date. Subjects were defined as ‘recent users’ when the supply ended between 1 and 29 days and as ‘past users’ when the supply ended 30 or more days prior to the index date; ‘nonusers’ were defined as patients who had no prescription for an NSAID in the medical record before the index date.

Statistical analysis

We conducted a matched analysis (conditional logistic regression model) using the software program SAS, Version  6.12  (SAS  Institute  Inc.,  Cary,  NC).  Relative risk estimates (odds ratios, OR) are presented with 95% confidence intervals (95% CI).

For each case and control, the potential confounders body mass index (BMI) (<25, 25–29.9, ≥30kgm−2, unknown) and smoking status (never, ex, current, unknown) were assessed from the patient profiles. We also assessed aspirin use and longer-term use of hormone replacement therapy and adjusted the final multivariate models for these covariates. By doing stratified regression analyses, we further evaluated potential effect modification by age (<65 years, ≥65 years) and sex.


We included 3315 cases with AMI and 13 139 controls in the analysis. Table 1 shows the characteristics of cases and controls including age, sex, smoking status and body mass index. Overall, cases were predominantly male, and the majority (57%) was under the age of 65. Current smoking status and high BMI (≥30) were substantially more prevalent in cases than controls. Current use of aspirin at the index date (OR 0.6, 95% CI 0.4, 1.0) and longer-term use (≥10 prescriptions prior to the index date) of hormone replacement therapy in women (OR 0.6, 95% CI 0.4, 0.9) were also associated with altered risk estimates for AMI.

Table 1.  Characteristics of cases and controls in relation to risk of developing acute myocardial infarction (AMI).
Number (%)
= 3315)
Number (%)

(= 13139)
odds ratio
(95% CI)
  • *

    adjusted for all covariates in the table in the same multivariate model

  • autopsy finding (patient did not reach hospital alive)

Age (years)
 <40 91 (2.8)367 (2.8) 
 40–49 417 (12.6)1656 (12.6) 
 50–59 830 (25.0)3314 (25.2) 
 60–691227 (37.0)4832 (36.8) 
 70–75 750 (22.6)2970 (22.6) 
 Male2452 (74.0)9715 (73.9) 
 Female 863 (26.0)3424 (26.1) 
Smoking status
 Non1079 (32.6)6204 (47.2)1.0 (Referent)
 Current1100 (33.2)2574 (19.6)2.7 (2.4–2.9)
 Ex 376 (11.3)1353 (10.3)1.6 (1.4–1.9)
 Unknown 760 (22.9)3008 (22.9)1.5 (1.3–1.7)
Body mass index (kgm−2)
 <25 885 (26.7)4240 (32.3)1.0 (Referent)
 25–29.91100 (33.2)4004 (30.5)1.4 (1.3–1.6)
 ≥30 387 (11.7)1208 (9.2)1.7 (1.4–1.9)
 Unknown 943 (28.4)3687 (28.0)1.2 (1.1–1.4)
Died from AMI   467 (14.1) 

The relative risk estimates (odds ratios, OR) of developing a first-time AMI in relation to current, recent past or past exposure to NSAIDs are shown in Table 2. Overall, current exposure to NSAIDs was not associated with a reduced risk of AMI compared with the reference group of nonusers (adjusted OR 1.17; 95% CI 0.99, 1.37). Stratification by duration of exposure did not yield materially different results in current users. Long-term NSAID use (≥30 prescriptions) yielded an adjusted relative risk estimate of 1.21 (95% CI 0.94, 1.55). Stratification by dose indicated that patients currently exposed to high NSAID doses may even have a slightly increased risk of AMI (adjusted OR 1.29; 95% CI 1.05, 1.58).

Table 2.  Risk of first-time acute myocardial infarction associated with current, recent past or past exposure to anti-inflammatory drugs (NSAIDs) stratified by duration of NSAID therapy (expressed as number of prescriptions) and dose.
NSAID use (number
of prescriptions)
(= 3315)
(= 13139)
odds ratio
(95% CI)
  • *

    adjusted for smoking status, body mass index, hormone replacement therapy and aspirin

  • low dose (dose per tablet): acemetacin 60mg, diclofenac <100mg, diflunisal 250mg, etodolac 200mg, fenbufen 300mg, fenoprofen 300mg, flurbiprofen 100mg, ibuprofen ≤200mg, indomethacin ≤75mg, ketoprofen 100mg, mefenamic acid 250mg, nabumetone 500mg, naproxen 275mg, piroxicam 10mg, sulindac 100mg, tenoxicam 20mg and tiaprofenic acid 200mg

  • high dose (dose per tablet): diclofenac 100mg, diflunisal 500mg, etodolac 300mg, fenbufen 450mg, fenoprofen 600mg, flurbiprofen 200mg, ibuprofen >200mg, indomethacin 100mg, ketoprofen 200mg, mefenamic acid 500mg, nabumetone 500mg, naproxen 500mg, piroxicam 20mg, and tiaprofenic acid 300mg

Non-users150262361.0 (Referent)
Current NSAIDs 242 8251.17 (0.99–1.37)
 1–4 34 1111.30 (0.87–1.93)
 5–9 45 1571.10 (0.78–1.54)
 10–19 36 1450.97 (0.66–1.42)
 20–29 38 1191.31 (0.89–1.91)
 30 + 89 2931.21 (0.94–1.55)
Recent past NSAIDs 118 3771.26 (1.01–1.57)
 1–4 25 1050.95 (0.61–1.48)
 5–9 21 950.90 (0.55–1.46)
 10–19 23 771.13 (0.70–1.83)
 20–29 14 441.33 (0.72–2.46)
 30 + 35 562.71 (1.75–4.22)
Past NSAIDs145357011.04 (0.96–1.13)
 1–4 98440021.02 (0.93–1.12)
 5–9 31111901.06 (0.92–1.22)
 10–19 91 3521.00 (0.78–1.28)
 20–29 26 821.26 (0.80–2.01)
 30 + 41 752.33 (1.57–3.46)
Current by dose
 Low dose 98 3671.02 (0.81–1.29)
 High dose 144 4581.29 (1.05–1.58)

We also observed a suggestion of an increased relative risk estimate of first-time AMI in the group of recent past users of NSAIDs (adjusted OR 1.26; 95% CI 1.01, 1.57). The stratification by duration of therapy indicated that this association was based on the subset of cases and controls with ≥30 prescriptions (adjusted OR 2.71; 95% CI 1.75, 4.22). For past users, stratification by duration again showed an increased risk in the subgroup of patients with ≥30 prescriptions (adjusted OR 2.33; 95% CI 1.57, 3.46), but not for those who used less NSAIDs. These results were very similar in different age groups (<65 years of age vs ≥65 years of age) and in both genders.

We additionally explored the risk of AMI associated with current exposure to individual NSAIDs (Table 3). None of the individual agents was associated with a substantially reduced AMI risk. Current naproxen exposure yielded an adjusted OR of 0.68 (95% CI 0.42, 1.13).

Table 3.  Risk of first-time acute myocardial infarction in current NSAID users, stratified by individual agents.
NSAID useCases
(= 3315)
(= 13139)
odds ratio
(95% CI)
  • *

    adjusted for smoking status, body mass index, hormone replacement therapy and aspirin

  • only NSAIDs presented with ≥5 exposed cases

Non-users150262361.0 (Referent)
Individual NSAIDs   
 Ibuprofen 60 2041.17 (0.87–1.58)
 Diclofenac 97 2771.38 (1.08–1.77)
 Piroxicam 10 281.65 (0.78–3.49)
 Fenbufen  7 132.06 (0.80–5.30)
 Ketoprofen 15 481.39 (0.77–2.51)
 Indomethacin 15 561.03 (0.58–1.85)
 Flurbiprofen  8 142.26 (0.93–5.46)
 Naproxen 19 1050.68 (0.42–1.13)


The present study provides evidence that current exposure to NSAIDs in patients ≤75 years of age without a diagnosed prior history of cardiovascular or metabolic disease is not associated with a reduced risk of first-time AMI. Thus, a possible effect of NSAID exposure on platelet aggregation by COX-1 inhibition and/or modification of inflammatory processes does not seem to be of clinical relevance in reducing the AMI risk. The level of COX-1 inhibition by NSAIDs at conventional analgesic dosages may be insufficient to inhibit platelet aggregation in vivo.

Our finding of no association with a decreased AMI risk with NSAID exposure is in accordance with a recent follow-up study using Medicaid data from Tennesse. In this study in a high-risk population of people 50 years of age or older (22% of the patients had a diagnosis of a serious cardiovascular disease in the past year before myocardial infarction and 67% used different cardiovascular drugs in the year prior to AMI indicating some kind of pre-existing cardiovascular risk factor) the authors found no   evidence   for   an   altered   risk   of   serious   coronary heart disease [11]. Multivariate adjusted relative risks for current   or   former   use   of   nonaspirin   NSAIDs   were 1.05 (95% CI 0.97, 1.14) and 1.02 (95% CI 0.97, 1.08), respectively. However, there was some indication of an excess risk in association with high dose ibuprofen use.

Additonally, our finding are quite similar to the results of a recent nested case-control analysis looking at NSAID exposure  and  AMI  in  postmenopausal  women  [10].  As in the study by Garcia Rodriguez et al.[10], the results of    our    study    indicate    that    current    NSAID    use    was not  associated  with  a  reduced  AMI  risk;  additionally, we also found some tendency towards a slightly in-creased AMI risk associated with current use of NSAIDs especially in users of high NSAID doses. Statistically significantly increased risk estimates around 2.5 were unexpectedly found in subjects who used NSAIDs on a long-term basis (i.e. ≥30 prescriptions) but who stopped NSAID use at some point in time before the index date. Based on the available data we cannot tell whether this is a chance finding, the result of some unknown bias or confounding, or a causal relationship. The latter would raise the hypothesis that subjects with chronic inflammation, who used NSAIDs for a long time but stopped the therapy for whatever reason, may be at an increased AMI-risk. If this finding were real, it may mean that chronic NSAID use indeed does reduce the AMI risk by suppression of inflammation; thus, current exposure to NSAIDs in subjects with chronic inflammation (i.e. those who regularly take NSAIDs) would reduce the risk from around 3 to 1, but not below 1, leading to an erroneous conclusion that NSAIDs do not lower the AMI risk. This hypothesis is purely speculative and needs further evaluation in future studies.

However, it has been shown that atherosclerosis is a process with inflammatory features [23] and that COX-2 is expressed in human atherosclerotic lesions [4, 24, 25]. It was hypothesized that products of COX-2 may be important in the pathogenesis of atherosclerosis [24], and selective COX-2 inhibitors may potentially have antiatherogenic effects [26].

A recent analysis of randomized trials suggested that current exposure to rofecoxib, a selective COX-2 inhibitor, may increase the risk of AMI as compared with use of naproxen [26]. In one of these studies, the incidence of AMI was lower in the naproxen group than in the rofecoxib group (0.1%vs 0.4%; relative risk, 0.2; 95% CI 0.1, 0.7) [27]. If this difference were real, COX-2 inhibitors either increase the AMI risk, or naproxen lowers it. In our study, we had no information on the selective COX-2  inhibitors  rofecoxib  and  celecoxib.  However, we  compared  the  AMI  risk  of  current  naproxen  users to nonusers    of    any    NSAIDs;    there    was    a    suggestion    of a reduced AMI-risk in naproxen users (OR 0.68, 95% CI 0.42, 1.13), although not statistically significant. Naproxen has been shown to inhibit thromboxane production particularly strongly (by approximately 95%) and platelet aggregation by 88% in healthy volunteers, an effect that is maintained throughout the dosing interval [9]. In addition, naproxen has a particularly long elimination half-life of approximately 14h [28].

In our study, we were not able to adjust for socioeconomic status or life style habits such as physical activity or dietary information, because this information is not routinely recorded in the GPRD. Since such factors are associated with an altered risk of cardiovascular diseases, they may in theory also be related to NSAID use and thereby potentially confound the association between NSAID use and AMI. Furthermore, we only studied the effect of NSAIDs on the AMI risk in patients without recorded previous cardiovascular or metabolic diseases. This was done because the effect of drugs on the risk of developing a first-time diagnosis of an outcome of interest can best be studied in subjects who are free of clinical risk factors for the disease [29], since preexisting diseases (e.g. hypertension) may both influence the likelihood of using NSAIDs as well as the AMI risk. In addition, it has been shown that use of NSAIDs can elevate blood pressure by about 5 mmHg [30] which might be associated with an increased risk of developing myocardial infarction. This risk elevation may in theory have counterbalanced some risk reduction exerted by NSAIDs, leading to the null result. This theoretically possible pathway – even though speculative – was another reason to exclude a priori subjects with recorded hypertension since we would most likely expect a relevant NSAID effect on blood pressure in subjects with preexisting hypertension.

In summary, we have found evidence that both current use of NSAIDs as well as longer-term use of these drugs does not seem to be associated with a substantially altered risk of developing a first-time diagnosis of AMI in subjects free of recorded clinical risk factors for AMI.

We thank the participating general practitioners for their excellent co-operation.

The Boston Collaborative Drug Surveillance Program is partly supported by grants from AstraZeneca, Berlex Laboratories, Boehringer Ingelheim Pharmaceuticals, Bristol-Myers Squibb Pharmaceutical Research Institute, GlaxoSmithKline, Hoffmann-La Roche, Janssen Research Foundation, and Novartis Farmacéutica. This study was not directly funded by any of these companies.

CRM is recipient of a grant from the Swiss National Science Foundation (grant no. 32–056 751).