Twenty-two studies (n=2,949) were included.
The mean drop-out rate, where reported, was 12.4%. Seventeen studies were sponsored by industry. The sample size ranged from 24 to 408. Seven RCTs provided data on an intention-to-treat basis.
The forest plot showed two outlier studies with an ES greater than 1.5. Both used the highest molecular weight hyaluronic acid. The other study using this formulation showed a non-statistically significant ES.
Statistically significant heterogeneity was detected (P<0.001) for the meta-analysis of all studies. The meta-analysis of the 3 RCTs (n=252) using the highest molecular weight hyaluronic acid showed statistically significant heterogeneity (P<0.001). After removing these 3 RCTs, heterogeneity was no longer statistically significant (P=0.58).
Overall, the studies found a small effect for hyaluronic acid compared with placebo (ES 0.32, 95% CI: 0.17, 0.47, P<0.001). After removal of the 3 RCTs using the highest molecular weight hyaluronic acid, the effect decreased (ES 0.19, 95% CI: 0.10, 0.27, P<0.001).
There was evidence of publication bias (asymmetrical funnel plot and Egger test P=0.07).
Unpublished studies showed a non significant ES for hyaluronic acid compared with placebo (ES 0.07, 95% CI: -0.15, 0.28).