Researchers analyzing previous trials of hyaluronic acid injections for knee osteoarthritis (OA) found that, although measurable benefit was demonstrated, it was too small to be clinically relevant.

Pooled data from 24 large placebo-controlled studies comprising a total of 8,997 participants with knee OA indicated that, relative to placebo injections, hyaluronic acid reduced patients’ pain with a standardized mean difference (SMD) of -0.08 points (95% CI -0.15 to -0.02), corresponding to a reduction of 2.0 mm on a conventional 100-mm visual analogue scale, according to Bruno R. da Costa, PhD, of St. Michael’s Hospital in Toronto, and colleagues.

But a clinically important degree of pain reduction would require an SMD of at least -0.37 points — 9 mm on the 100-mm scale — meaning hyaluronic acid injections (a.k.a. viscosupplementation) provided no meaningful benefit for most patients, they wrote in The BMJ.

“Our findings do not support the broad use of viscosupplementation for the treatment of knee osteoarthritis,” da Costa and colleagues stated.

This conclusion isn’t exactly new, but da Costa’s group stated it more boldly than other reviews. For example, the American Academy of Orthopaedic Surgeons’ (AAOS) website acknowledges that “[t]he most recent research… has not found viscosupplementation to be effective at significantly reducing pain or improving function.”

But the AAOS summary doesn’t warn patients or clinicians away from the shots. It says “some” patients obtain pain relief and “some” do not, and describes the injections as “an option” for patients who “have tried all other nonsurgical treatment methods.”

The 24 included trials were among a total of 169 published over the past several decades. They were selected for having at least 100 participants and pain intensity evaluated with visual analogue scales or the Western Ontario and McMaster Universities Arthritis Index (WOMAC), except for one trial that used the Lequesne index.

Nineteen of the trials also evaluated functional effects, as assessed under the WOMAC or Lequesne systems. For this too, hyaluronic acid injections provided a statistically but not clinical significant benefit, with an SMD of -0.11 points (95% CI -0.18 to -0.05). An SMD of -0.37 was the researchers’ standard for clinical importance, and as with pain, even the 95% confidence interval didn’t stretch that far.

There didn’t seem to be any particular species of injection or trial design for which the injections could be more effective. Only one of the 24 trials came with an SMD exceeding the 0.37-point threshold for pain, but it appeared in 1983, had “high risk/some concerns” about bias in measuring outcomes, and wasn’t published in an English-language journal. Stratifying the trials by details of the injections (number, type, and molecular weight categories, for example), follow-up time, or potential for bias did not show major variations in SMDs, which mostly clustered around -0.1 points.

Also of note, the data did suggest that hyaluronic acid injections are not totally benign. With 15 trials (6,462 participants) reporting adverse events in sufficient detail, da Costa and colleagues calculated that the shots came with significantly elevated risk for serious complications (relative risk 1.49, 95% CI 1.12-1.98). “Overall, 3.7% of patients receiving viscosupplementation and 2.5% receiving placebo experienced a serious adverse event,” the researchers found.

Analysis of trial results by the year they were conducted indicated that it was only the very earliest that pointed to a substantial benefit. After that 1983 study appeared, the next large one came in 1993; with data pooled from both, the SMD already fell short of -0.37 points. By 2004, when nine trials had appeared with a total of about 2,240 participants, it was apparent that the injections were not markedly helpful. Trials appearing since then merely reinforced that conclusion.

In those subsequent trials, the researchers said, more than 12,000 patients were “subjected” to the shots even though it should have already been clear that the treatment wasn’t particularly effective, “which raises ethical concerns,” da Costa and colleagues wrote.

The authors also said it seems that findings from a number of large industry-sponsored trials have never seen daylight. They pointed to three listed on ClinicalTrials.gov, totalling more than 2,000 participants, for which full results have not been published. “At the time of writing, at least 12 other unpublished trials were known to be completed, but their results were not retrievable (>3,000 patients randomised in total),” they added.

Please enable JavaScript to view the

comments powered by Disqus.