Twenty-five studies (with seven contributing data to subgroup analyses only) were reportedly included in the review (6,831,341 participants): eight cohort studies (2,322,718 participants); 14 nested or population-based case-control studies (4,493,391 participants); and three hospital field case-control studies (15,232 participants). All 25 studies were rated as having comparable study groups, and 19 studies reported good selection and definition of participants. One cohort study was reported as showing immortal time bias in the reporting of the outcome, and two field case-control studies may have misclassified exposure.
Compared with nonuse or past use of NSAIDs, statistically significantly higher risks of acute myocardial infarction were found with meloxicam (RR 1.25, 95% CI 1.04 to 1.49; three studies; Ι²=0%), diclofenac (RR 1.38, 95% CI 1.26 to 1.52; 11 studies; Ι²=61%), indometacin (RR 1.40, 95% CI 1.21 to 1.62; four studies; Ι²=0%), etodolac (RR 1.55, 95% CI 1.16 to 2.06; three studies; Ι²=0%), rofecoxib (RR 1.34, 95% CI 1.22 to 1.48; 17 studies; Ι²=61%), and etoricoxib (RR 1.97, 95% 1.35 to 2.89; three studies; Ι²=0%). Slightly higher risks of acute myocardial infarction were also found with naproxen (17 studies), ibuprofen (13 studies), and celecoxib (18 studies) in comparison with nonuse or past use of NSAIDs, however these differences were statistically non-significant. Statistically significant heterogeneity was shown with all of these analyses (Ι² range: 67 to 90%).
Compared with non-use or past use of NSAIDs, a statistically significant lower risk of acute myocardial infarction was observed with the use of naproxen in patients who were new users at the index date (five studies). Statistically significant higher risks were observed with the use of ibuprofen, diclofenac, or rofecoxib in all/some of the study subgroup populations (further details reported in the review). Aside from naproxen, higher risk was linked to higher doses of NSAIDs (as defined in individual studies) overall and in patients with prior coronary heart disease. Both high and low doses of diclofenac and rofecoxib were associated with higher risk, with a dose-response relationship for rofecoxib. Other results were fully reported in the paper. No evidence of publication bias was found.