The analysis compared the risk of MI, cardiovascular events, cardiovascular deaths and stroke in 4,422 patients taking either celecoxib or a placebo.
It showed the drug more than doubled the risk of MI, but did not affect stroke or other cardiovascular risks.
A second analysis of 12,780 patients compared celecoxib with placebo, diclofenac, ibuprofen and paracetamol. It found a similar but smaller increased risk of MI associated with celecoxib.
The paper's authors concluded that the whole class of COX-2 inhibitors increased the risk of MI.
Dr Kamran Abbasi, editor of the Journal of the Royal Society of Medicine, where the research was published, said fears about the drug's safety were justified.
'Given the popularity of celecoxib in the treatment of arthritis, drug regulators must undertake a risk assessment based on these findings,' he said.
The European Medicines Agency reviewed the safety of all COX-2 inhibitors just over a year ago after Vioxx was withdrawn when it was found to cause serious thrombotic events.
It concluded GPs should be cautious about giving any COX-2 inhibitor to patients with cardiovascular risk factors.
Dr Mike Kirby, Hertfordshire GP and member of the Primary Care Cardiovascular Society, said: 'We need to be careful when prescribing these drugs to people with a history of cardiovascular disease.'
He added that the drugs were effective and without risk in healthy people.
'This new research is important but won't change the guidance,' he said.
JRSM 2006; 99: 132-40.