In reply: Mansfield, Vitry and Wright take issue with our statement that the celecoxib studies have not shown an increased risk of thrombosis, but provide no data to support their claims, while Cleland and James highlight differences in COX-2 selectivity as a potential explanation for differences in the cardiovascular safety of coxibs. Recently published clinical data confirm an increased risk of cardiovascular events with rofecoxib but not celecoxib, 1 which is consistent with in-vivo studies suggesting that celecoxib but not rofecoxib improves endothelial function,2,3 as well as a significantly lower incidence of oedema and hypertension with celecoxib compared with rofecoxib.4 Nevertheless, we reiterate that it remains incumbent on drug manufacturers and regulatory authorities to demonstrate cardiovascular safety for all new and existing coxibs, including celecoxib.

The published coronary artery bypass graft surgery randomised trial referred to by Cleland and James did not report a significant excess of adverse cardiovascular events with valdecoxib,5 nor did two recently published meta-analyses.6,7 However, unpublished data from a second coronary artery bypass graft surgery trial, as well as meta-analyses presented at the American Heart Association meeting in New Orleans in November 2004, indicate that valdecoxib compared with placebo significantly increases the risk of adverse cardiovascular events.8 This is reflected in the recently revised US prescribing information for valdecoxib. 9

Manev and Manev propose enhanced 5-lipoxygenase activity as a mechanism for increased cardiovascular risk in patients treated with COX-2-selective inhibitors. We agree that this important hypothesis merits further study.

Day and Graham emphasise paracetamol as first-line drug treatment in the management of osteoarthritis, referring to recently published American, European and Australian guidelines to support their position. We do not dispute the effectiveness of paracetamol to reduce chronic pain. However, data from the 2004 systematic review quoted in our editorial10 demonstrate that nonsteroidal anti-inflammatory drugs are better than paracetamol for pain relief and are often preferred by patients, despite a higher incidence of adverse effects. Only one of the 10 randomised controlled trials included in this systematic review followed patients beyond 3 months; this study reported the primary efficacy outcome only during the first 6 weeks and was not powered for safety.