Hyaluronic acid (viscosupplementation)

Hyaluronic acid is FDA-cleared, insurer-recognized, and backed by more trials than almost anything else on this site — and the large blinded ones find a pain effect a patient cannot feel. It is the clearest case here of why trial count is not trial support.

The mechanism problem

HA is the glycosaminoglycan that gives synovial fluid its viscoelasticity; osteoarthritic joints have less of it, at lower molecular weight. "Viscosupplementation" proposes topping the joint back up — restored lubrication plus possible anti-inflammatory receptor effects. The long-noted problem: injected HA resides in the joint for hours to days, while the claimed benefits span months. A lubrication mechanism cannot explain that; something biological would have to, and it remains unproven.

What the trials show when you sort them by quality

The literature is huge — size is not the issue. Quality-stratified results are:

Rutjes 2012 (89 trials, 12,667 patients) found an overall pain effect size of −0.37 — but in the 18 large blinded trials it fell to −0.11, clinically irrelevant, and in unpublished trials to −0.03. The apparent benefit lives almost entirely in small, unblinded, published trials.

Pereira 2022 in the BMJ (169 trials, 21,163 patients; main analysis from 24 large placebo-controlled trials) found benefit over placebo of SMD −0.08 — roughly 2 mm on a 100 mm pain scale, below the minimal clinically important difference. The authors note that strong evidence of no clinically relevant benefit has existed since 2009.

The analysis most often cited in support is Bannuru's 2015 network meta-analysis, where HA ranks top for pain against oral placebo. The same analysis shows intra-articular placebo beats oral placebo — much of every injection's apparent benefit is the needle and the ritual, not the contents.

Guidelines followed the data: AAOS's 2013 second edition issued a strong recommendation against HA, and the 2021/22 third edition maintains "not recommended for routine use."

Serious adverse events

HA is generally well tolerated day to day — transient flares, rare pseudoseptic reactions. But both large meta-analyses report elevated serious adverse events versus placebo: RR 1.41 in Rutjes, RR 1.49 in Pereira. The mechanism is unclear and the finding is debated, but a therapy with sub-threshold benefit cannot justify any serious-event excess.

The molecular-weight claim

Proponents argue high-molecular-weight (≥3000 kDa) and cross-linked products outperform low-MW HA. That claim rests on subgroup-level inference by authors with HA-manufacturer relationships, and the 2022 BMJ analysis found no clinically relevant benefit across products. No high-molecular-weight product has yet beaten saline by a clinically important margin in a large blinded trial, which is the comparison the claim would need.

The counterpoints

Blinded trials subtract the contextual effect of the injection ritual, but patients still experience it — people genuinely feel better after HA, just not more than after saline. Some PRP trials treat HA as a credible active comparator, and OARSI and ACR positions are softer than AAOS's. None of that changes the size of the specific effect, which stays below the clinically important threshold.

If you're offered it anyway

Regimens run 1–5 weekly injections, often repeated every 6 months, at hundreds to thousands per course — and insurers increasingly deny coverage following the guideline position. The expectation to bring to that conversation: about 2 mm on a 100 mm pain scale beyond what a saline injection would deliver. HA products reached market as FDA-cleared devices, a lower evidence bar than drugs — a regulatory path worth knowing when weighing the offer. Oral hyaluronic acid is a separate question with separate (and much thinner) evidence — see the supplements entry.