Non-steroidal anti-inflammatory drugs (NSAIDs) are referred to as the backbone of osteoarthritis pain management. Bruno R da Costa et al. conducted a meta-analysis approach that aims at determining the efficacy of various preparations and doses of NSAIDs on osteoarthritis pain in a network meta-analysis.

The meta-analysis considered RCT's that have undergone a comparison between the following interventions: NSAIDs, paracetamol, or placebo, for the treatment of osteoarthritis pain. A systematic search for relevant articles was conducted from Cochrane Central Register of Controlled Trials (CENTRAL) and reference lists for trials published between Jan 1, 1980, and Feb 24, 2015, with at least 100 patients per group.  The primary and secondary outcomes included were pain and physical function which was obtained in duplicate for up to seven-time points after the start of treatment. Extension of Multivariable Bayesian random effects models was utilised for comparison between multiple mixed therapies with a random effect at the level of trials. A random walk of the first order was used to describe various follow-up outcome data within a trial for the primary analysis. Preparations that employs different total daily dose were taken into account separately. The potential dose-response relationship was assessed by using a preparation-specific covariate assuming linearity on log relative dose.

A total of 8973 manuscripts were identified from search, among which 76 were randomised controlled trials including 58 451 patients. Twenty-three nodes involving seven different NSAIDs or paracetamol with the specific daily dose of administration or placebo were considered. Comparison between placebo and all other preparations shown an improvement in point estimates of pain symptoms, irrespective of dose. A probability of 95% was obtained for differences between six interventions (diclofenac 150 mg/day, etoricoxib 30 mg/day, 60 mg/day, and 90 mg/day, and rofecoxib 25 mg/day and 50 mg/day), and placebo that ranges below or at a prespecified minimum clinically significant effect for reduction in pain (effect size [ES] −0·37). Diclofenac 150 mg/day (ES −0·57, 95% credibility interval [CrI] −0·69 to −0·45) and etoricoxib 60 mg/day (ES −0·58, −0·74 to −0·43) had the highest probability (i.e., 100%) to be the best intervention. These two drugs reached the minimum clinically relevant difference among maximally approved daily doses. An increase in drug dose was obtained by a rise in treatment effects, except for naproxen (p=0.034) for which similar tests for a linear dose effect were significant. There was no evidence obtained for variation in treatment effects over the duration of treatment. The heterogeneity and inconsistency between the trials were low, and the model fit was good. Low risk of bias exists in all trials for blinding of patients. The effect estimates account for methodological quality criteria, and no change was obtained in sensitivity analyses with two additional statistical models.

The study concluded that there was no role of single-agent paracetamol for the treatment of patients with osteoarthritis irrespective of dose. Strong evidence was obtained for the use of diclofenac 150 mg/day, as the most effective NSAID available at present, regarding improving both pain and function. It was inferred that physicians should consider the results of this study and all known safety information regarding the preparation and dose for individual patients, for determining the safety profile of these drugs.