Current issues of ACP Journal Club are published in Annals of Internal Medicine


Nonsteroidal antiinflammatory drugs did not increase risk for all-cause mortality in osteoarthritis


ACP J Club. 2007 May-Jun;146:77. doi:10.7326/ACPJC-2007-146-3-077

Clinical Impact Ratings

GIM/FP/GP: 5 stars

Cardiology: 6 stars

Rheumatology: 6 stars

Source Citation

Lee TA, Bartle B, Weiss KB. Impact of NSAIDS on mortality and the effect of preexisting coronary artery disease in US veterans. Am J Med. 2007;120:98.e9-16. [PubMed ID: 17208086]



In patients with osteoarthritis, what are the mortality and cardiovascular (CV) and cerebrovascular (CBV) risks associated with long-term nonsteroidal antiinflammatory drug (NSAID) use?


Design: Nested case–control study of a cohort of veterans from the Veterans Health Administration (VHA) health care system with 3-year follow-up.

Setting: U.S. VHA health care system.

Patients: 354 456 patients with osteoarthritis who had ≥ 2 prescriptions from a VHA pharmacy. Patients were divided into those with coronary artery disease (CAD) at baseline (11 912 cases, 135 379 controls, mean age 72 y, 98% men) and those without CAD (16 869 cases, 190 296 controls, mean age 70 y, 97% men). Exclusion criteria included receipt of > 1 type of NSAID during follow-up; history of myocardial infarction, stroke, or cancer; NSAID prescription filled or outcome event within the first 90 days of follow-up; and > 90 inpatient days. Patients with ≥ 1 NSAID dispensed during follow-up were defined as exposed.

Risk factors: Baseline CAD, receipt of any NSAID, cyclooxygenase-2 (COX-2)–selective NSAID, and nonselective NSAID.

Outcomes: All-cause mortality, and CV or CBV events.

Main results

Exposure to any NSAID or nonselective NSAID was associated with decreased risk for death and increased risks for CV or CBV events in patients both with and without CAD (Table). Exposure to a COX-2–selective NSAID was not associated with any effect on mortality or CV or CBV events, except for a decrease in mortality in the no-CAD group (Table). In the no-CAD group, exposure to rofecoxib (OR 1.32, CI 1.04 to 1.67), naproxen (OR 1.18, CI 1.07 to 1.30), ibuprofen (OR 1.11, CI 1.02 to 1.22), diclofenac (OR 1.32, CI 1.08 to 1.62), or etodolac (OR 1.33, CI 1.10 to 1.62) was associated with increased risk for CV or CBV events. In the CAD group, exposure to ibuprofen increased risk for CV or CBV events (OR 1.27, CI 1.15 to 1.42).


In patients with osteoarthritis, nonsteroidal antiinflammatory drugs were associated with decreased risk for all-cause mortality and increased risk for cardiovascular or cerebrovascular events.

Source of funding: No external funding.

For correspondence: Dr. T.A. Lee, Hines VA Hospital, Hines, IL, USA. E-mail

Table. Association between use of nonsteroidal anti-inflammatory drugs (NSAIDs) and risk for all-cause mortality and cardiovascular (CV) or cerebrovascular (CBV) events*

Risk factors Type of NSAID use Adjusted odds ratio (95% CI)†
All-cause mortality CV or CBV event
Patients with CAD Any NSAID 0.79 (0.73 to 0.86) 1.18 (1.11 to 1.27)
Nonselective NSAID 0.77 (0.71 to 0.85) 1.20 (1.12 to 1.29)
COX-2–selective NSAID 0.93 (0.76 to 1.13)‡ 1.08 (0.89 to 1.30)‡
Patients without CAD Any NSAID 0.72 (0.68 to 0.77) 1.14 (1.08 to 1.21)
Nonselective NSAID 0.71 (0.67 to 0.76) 1.14 (1.08 to 1.21)
COX-2–selective NSAID 0.81 (0.68 to 0.97) 1.12 (0.94 to 1.35)‡

*CAD = coronary artery disease; COX-2 = cyclooxygenase-2. CI defined in Glossary.
†Odds ratio adjusted for age, sex, anticoagulant, antiplatelet, aspirin, digoxin, β-blocker, calcium-channel blocker, nitrates, antiarrhythmics, statins, diuretics, angiotensin-converting enzyme inhibitors, angiotensin-receptor blockers, hypoglycemics, angina, chronic heart failure, peripheral vascular disease, obesity, and tobacco or alcohol dependence.
‡Not significant.


The study by Lee and colleagues adds fuel to the fire that selective and nonselective NSAIDs are associated with increased CV events. Do they cause these events? This is a much harder question to answer. Association does not necessarily mean causality. This study used a nested case–control design. Patient selection may have been biased, and issues with compliance in taking the medication are difficult to ascertain. The exclusion of patients who switched NSAIDs raises concerns about generalizability because switching among NSAIDs, including COX-2 NSAIDs, is common in clinical practice.

Despite the association of all NSAIDs with CV and CBV events in Lee and colleagues' study, overall mortality was not increased. This finding begs the still-unanswered question—what benefit might NSAIDs have that could decrease mortality?

At present, COX-2 NSAIDs are thought to carry a higher risk for CV and CBV events than older NSAIDs, and naproxen to carry a lower risk for CV and CBV events than other older NSAIDs (although this remains controversial) (1). A recent randomized trial (2) concluded that etoricoxib, a new unapproved COX-2 NSAID, did not differ in efficacy or in CV harm from diclofenac. However, perhaps naproxen would have been a better comparator. A randomized controlled trial is needed to compare naproxen, celecoxib, and a control group (i.e., patients who would receive only acetaminophen perhaps with an added narcotic if necessary to control breakthrough arthritis pain).

Emilio B. Gonzalez, MD
University of Texas Medical Branch
Galveston, Texas, USA


1. Laupacis A.Review: Selective COX-2 inhibitors increase vascular events more than placebo and naproxen, but not more than other NSAIDs [Comment]. ACP J Club. 2006;145:66. [PubMed ID: 17080978]

2. Cannon CP, Curtis SP, FitzGerald GA, et al. Cardiovascular outcomes with etoricoxib and diclofenac in patients with osteoarthritis and rheumatoid arthritis in the Multinational Etoricoxib and Diclofenac Arthritis Long-term (MEDAL) programme: a randomised comparison. Lancet. 2006;368:1771-81. [PubMed ID: 17113426]