Anti-inflammatory drugs: What’s the evidence?


Traditional non-steroidal anti-inflammatory drugs (NSAIDs) include ibuprofen (Advil, Motrin), diclofenac (Voltaren), naproxen (Naprosyn), and others. These drugs are widely available in many dosage forms and most hospitals have several on formulary. NSAIDs are first-line options for many types of pain, but they can cause stomach upset and occasionally gastrointestinal (GI) bleeding.

When cyclooxygenase-2 (COX-2) inhibitors such as celecoxib (Celebrex) became available, they were expected to cause less GI bleeding. Some studies did indeed show less GI bleeding, but some didn’t – and then some studies showed an increased risk of cardiovascular events such as heart attacks and strokes. The COX-2 inhibitor rofecoxib (Vioxx) was removed from the Canadian market in 2004 for this reason. Several conflicting studies led to a COX-2 controversy. Were they safe?


Then studies began to emerge showing that traditional NSAIDs might carry cardiovascular risk as well. How big is this risk? Is it the same for all NSAIDs? With so much conflicting data, what information can a clinician trust, and which drugs should hospitals have on hand for treating mild to moderate pain?

Systematic reviews

This situation demonstrates the value of a systematic review. Systematic reviews of the medical literature capture all studies available on a given topic. As more studies become available on a given topic, we can have more confidence when we see conclusions repeated, or when a more mature data set is presented in one paper with a critical appraisal of all the included studies.

When the design of the studies is similar, the data from different studies can be pooled and re-analyzed together; this is called a meta-analysis. With more data comes more power to detect differences between treatments or to identify rare side effects.

The hierarchy of evidence

For all these reasons, systematic review and meta-analysis sit at the top of the “hierarchy of evidence.” This hierarchy is a way of ranking different types of clinical studies. In general, systematic reviews and meta-analyses are more reliable than a single randomized controlled trial (RCT), which in turn is more reliable than non-randomized studies such as cohort studies and case reports. Quality is important, though; a well-done study at the bottom of the hierarchy may be more reliable than a poorly-done systematic review.


Hallmarks of a high-quality systematic review include: a clearly formulated research question, a structured literature search strategy that others can reproduce, explicit methods for selecting and critically appraising studies, and a clear reproducible description of the methods used to analyze the data.

CADTH evidence review

CADTH recently critically appraised six systematic reviews and meta-analyses on COX-2 and NSAID safety. Two systematic reviews of RCTs reported no differences in cardiovascular or GI outcomes between celecoxib and high dose diclofenac, but one systematic review of non-randomized studies reported a slightly higher cardiovascular risk with diclofenac. For celecoxib vs. ibuprofen, celecoxib was associated with fewer GI complications without any statistically significant differences in major cardiovascular events. For celecoxib vs. naproxen, the risk of cardiovascular events was higher with celecoxib, but there were more GI complications with naproxen.


The bottom line is that:

  • naproxen seems to have lower cardiovascular risk than celecoxib
  • diclofenac and ibuprofen seem to have the same cardiovascular risk as celecoxib
  • celecoxib and diclofenac seem to have lower GI risk than ibuprofen and naproxen
  • clinicians may need to beware of underestimating the risks of these drugs

These results show the value of using systematic reviews, not only for controversial drugs such as COX-2 inhibitors, but also for commonly used drugs such as NSAIDs.

It’s also important to remember that a systematic review has the same limitations as the studies feeding into it. For example, most of the studies in these reviews lasted for three months or less, so by extension, the systematic reviews can only answer questions about short-term use. Systematic reviews and meta-analyses have more power and precision than individual clinical trials, but their quality and relevance will always depend on the quality and relevance of the original studies.