Analysis · Health
Why One Heart Safety Measure Cannot Clear a Painkiller
By AWEI · AI-compiled · Published · Analysis prepared · 1 source · doi.org
A painkiller could show no clear increase in heart attacks yet still raise concern about fatal cardiovascular events. Diclofenac illustrated that distinction in a 2011 analysis by Sven Trelle and colleagues, published in BMJ. Its apparent safety depended on which outcome researchers examined.
The team combined 31 randomised trials involving 116,429 patients and seven non-steroidal anti-inflammatory drugs, or NSAIDs. Many participants had osteoarthritis or rheumatoid arthritis; others entered trials for different conditions. This was a synthesis of existing experiments, with a literature search updated through July 2009.
Their network meta-analysis connected direct comparisons with indirect evidence through shared treatments. That allowed estimates against placebo even without a direct placebo comparison for every drug. The model preserved randomised comparisons within trials, giving stronger grounds for assessing treatment effects than observational associations alone, while relying on comparability across the network.
For diclofenac, the estimated heart attack rate was 0.82 times the placebo rate, but its 95% credibility interval stretched from 0.29 to 2.20: compatible with either benefit or harm. For cardiovascular death, the estimate was 3.98, with an interval from 1.48 to 12.70. These relative rates describe different outcomes; neither gives an individual's probability of an event.
Uncertainty remained substantial. Despite the large patient total, the heart attack analysis contained only 554 events. Some safety data were unavailable, reports disagreed on event counts, and not all trials independently adjudicated events. The researchers could not adequately separate dose effects, and none of the trials tested intermittent use.
Naproxen appeared least harmful for cardiovascular outcomes, but that did not establish overall safety. The useful lesson is how to read a safety claim: check the specific outcome, the uncertainty around its estimate, and whether the studied treatment pattern matches the claim. Failure to detect harm is not proof that harm is absent.
Sources used for this article (1)
Direct links to the publisher reports used to prepare this article.
- Source 1
- Cardiovascular safety of non-steroidal anti-inflammatory drugs: network meta-analysis Sven Trelle, Stephan Reichenbach, Simon Wandel …
