We cannot recommend using hyaluronic acid for patients with symptomatic osteoarthritis of the knee.

Guideline: Treatment of Osteoarthritis of the Knee (2nd edition)
Rationale:
Fourteen studies (three high-strength studies and 11 moderate-strength studies) assessed intraarticular hyaluronic acid (HA) injections. A comparison of the patients in these studies and the ones validating the MCIIs we used to judge clinical significance revealed that they were demographically comparable for WOMAC and VAS pain as well as WOMAC function on the basis of age, baseline pain scores, BMI, weight and gender. Meta-analysis  in meaningfully important difference (MID) units showed that the over effect was less than 0.5 MID units, indicating a low likelihood that an appreciable number of patients achieved clinically important benefits in the outcomes (Guyatt et al.). Although meta-analyses of WOMAC pain, function, and stiffness subscales scores all found statistically significant treatment effects, none of the improvements met the minimum clinically important improvement thresholds. When we differentiated high- versus low- molecular weight viscosupplementation, our analyses did show that most of the statistically significant outcomes were associated with high-molecular cross linked hyaluronic acid but when compared to mid-range molecular weight, statistical significance was not maintained. Treatment comparisons between any weights higher than 750 kDa were not significantly different. The strength of this recommendation was based on lack of efficacy, not on potential harm.
 
The 2008 edition of this guideline where the benefits of viscosupplementation were found to be inconclusive rather than non-affirming used a systematic review from AHRQ that compared Hylan G-F 20 to placebo. Although there was a statistically significant treatment effect associated with the high molecular weight, different pain measurement outcomes (WOMAC and VAS pain) were combined so clinical significance could not be determined. Also, the work group found evidence of publication bias (publicizing of primarily favorable studies). We excluded the AHRQ systematic review because the selection criteria did not match ours. The primary difference was that in the current edition of the guideline clinical efficacy beyond a 4-week treatment period was required for studies to be included. This 2nd edition was based on meta-analyses that combined like measurement instruments, which made it possible to determine that the overall effect of hyaluronic acid did not provide minimum clinically important improvement to patients. Additionally, the AHRQ review included trials of varying research-design quality due in part to variations in sample sizes. In AAOS clinical practice guidelines, evidence of lower strength is excluded when there are at least two higher strength studies evaluating an outcome, and we excluded many of the lower strength studies included in the AHRQ review since they did not meet our selection criterion of at least 30 patients in each treatment group. Noted in the AHRQ review was that “There is evidence consistent with potential publication bias. Pooled results from small trials (<100 patients) showed effects up to twice those of larger trials consistent with selective publication of underpowered positive trials” (page 64).” Future research using clinically relevant outcomes, sub-group analyses, and controls for bias are needed.