Comparison between intra-articular ozone and placebo in the treatment of knee osteoarthritis: A randomized, double-blinded, placebo-controlled study

Lopes de Jesus CC, Dos Santos FC, de Jesus LMOB, Monteiro I, Sant'Ana MSSC, Trevisani VFM · PLoS One

98 patients aged 60 to 85 with Kellgren-Lawrence grade II-III knee osteoarthritis were randomised to weekly intra-articular ozone at 20 micrograms per millilitre, 10 mL per injection, or to 10 mL of air on the same schedule, for 8 consecutive weeks, and followed to 16 weeks. At 8 weeks the ozone group was ahead on pain (VAS mean difference 2.16, 95 percent CI 0.42 to 3.89, P < 0.003), on the Lequesne index (4.05, 95 percent CI 1.10 to 7.00), on the Geriatric Pain Measure (18.94, 95 percent CI 3.43 to 34.44), on all three WOMAC subscales, and on every SF-36 domain measured. The Timed Up and Go test showed no difference between the groups. The separation on WOMAC pain was still present at 16 weeks, eight weeks after the last injection (median of differences 14.999, 95 percent CI 0.000 to 25.000, P = 0.005), while WOMAC stiffness had converged by then. Adverse events were three puncture accidents, two of them in the placebo group.

Population

Adults with symptomatic knee osteoarthritis

Intervention

Intra-articular oxygen-ozone at 20 micrograms per millilitre, weekly for 8 weeks

Comparator

Placebo injection on the same schedule

Limitations

The comparator was an injection of air, not saline. Allocation was not balanced and the authors say why: group size was not controlled during assortment, ending at 63 ozone patients to 35 placebo, a ratio of 1.75 to 1 that they acknowledge costs statistical efficiency. The analysis was per protocol, excluding the two ozone-group withdrawals as never treated. The groups differed at baseline on marital status (P = 0.044) and on schooling. The authors name their own two principal limitations: no imaging was performed, so nothing in this trial speaks to cartilage or structure, and both the treatment period and the follow-up were short. A 2024 erratum corrects Table 4, in which the placebo and treatment columns had been swapped. Recruitment ran from November 2010 to March 2015 across three centres in Sao Paulo. The authors declare no competing interests; the retrieved article carries no funding statement.