knee osteoarthritis therapies

The Annals of Intl Medicine had a recent systematic review (137 studies with 33,243 participants) and network meta-analysis of different therapies for knee osteoarthritis, including acetaminophen, diclofenac, ibuprofen, naproxen, celecoxib, intra-articular (IA) steroids, IA hyaluronic acid, oral placebo and IA placebo (see knee arthritis therapies AIM 2015 in dropbox, or Ann Intern Med. 2015;162:46-54​). a network meta-analysis is a mathematical construct to assess the data from multiple individual head-to-head comparisons and integrate it all into a single comparative analysis

results:

--median age overall was 62, 67% women

--for pain (129 trials, 32,129 participants), all interventions were better than placebo. size effects ranged from 0.63 for most efficacious treatment (IA hyaluronic acid) to 0.61 for IA steroids, 0.52 for diclofenac, 0.44 for ibuprofen, 0.38 for naproxen, 0.33 for celexoxib, 0.29 for IA placebo, and 0.18 for acetaminophen.

--for function (76 trials, 24,059 participants), all were better than oral placebo except IA steroids. naproxen, ibuprofen, diclofenac, and celecoxib were statistically better than acetaminophen. IA hyaluronic acid was statistically superior to IA steroids.

--for stiffness (55 trials, 18267 participants), most of the treatments were not significantly different from each other. naproxen, ibuprofen, diclofenac, and celecoxib were statistically better than acetaminophen.

a few issues regarding the limitations of a network analysis. 

--the most evident limitation is that there is not a common study comparing all of the different management strategies with a common placebo group. and we know that even comparisons of the same treatments in different studies typically lead to differing results. so given that most of the individual trials done in this network analysis included <5 studies, with 9of the 19 therapycomparisons having only 1 or 2 trials, the small differences between many of the results above may well not be statistically or clinically significant. and, of note,  there were very few direct comparisons between IA and oral agents. this is relevant since IA steroids are commonly used, yet the only comparison studies with IA steroids was with IA hyaluronic acid and IA placebo. also, even though 67% of the participants were women, some studies had only/mostly men, so one cannot assume that the overall results necessarily apply to women without looking specifically at the studies in question.

--second, they did not include trials with multiple interventions (and, from my experience, that is the most common clinical scenario: IA steroids with use of acetaminophen or NSAID as needed, alternating acetaminophen with NSAID, etc). 

--third, the assessments tools use in the different analyses (eg, assessment of pain, stiffness, function) were sometimes different in different studies, so the researchers tried to convert these outcomes to more standard scales (eg, WOMAC VAS, or Western Onatario and McMaster Universities OA Index, visual analogue scale), with attendant potential errors in these conversions. 

--their conclusion that "for function, all interventions except IA corticosteroids were significantly superior to oral placebo" is a tad suspect/overstated, since there were no studies (ie, zero) which compared IA steroids to oral placebo. in fact the largest numbers of studies compared celecoxib to oral placebo to, IA hyaluronic acid to diclofenac, and IA hyaluronic acid to naproxen and to placebo. at least in our health center in Boston, most of these large-comparator therapies are almost never used (with the exception of naproxen).

-- my experience with IA steroids for knee OA, which is pretty extensive (probably on the order of 500 injections over many years), is overall impressive. although some people with OA do not respond or only for a few weeks (especially in those who have had multiple prior injections), a very large percentage do get significant reasonably longterm relief (essentially pain-free and full return of function for 3-12 months) and are spared the significant adverse effects of NSAIDS, especially common in the elderly (who are the ones who mostly get knee OA....). i do understand that there are likely placebo effects for IA therapies (as they found above), but in my experience i have seen zero adverse events for knee injections (or any other place i've injected, which probably brings my sample size to around 1000....). so IA steroids seems to me to be mostly very effective (lots of benefits and not apparent risks over several decades).