Despite the prevalence of knee osteoarthritis (OA), few studies compare treatments head-to-head. Now, a network meta-analysis compares 137 reports on the effectiveness of such treatments and reveals that acetaminophen, the most widely used over-the-counter treatment, does not provide a clinically significant reduction in pain.
“Except for celecoxib, all active interventions were significantly better than acetaminophen. Intra-articular placebo was significantly better than oral placebo, and active [intra-articular] treatments were more effective than active oral treatments,” write Raveendhara R. Bannuru, MD, from the Center for Treatment Comparison and Integrative Analysis, Division of Rheumatology, Tufts Medical Center, Boston, Massachusetts, and colleagues. “Most treatments did not significantly differ in function and stiffness outcomes, in part because of the limited number of trials investigating many of the studied interventions.”
The authors published their analysis in the January 6 issue of the Annals of Internal Medicine.
In their analysis, the researchers included randomized trials that compared two or more treatments for clinically or radiologically diagnosed symptomatic primary knee OA. The studies tested the oral drugs acetaminophen, celecoxib, diclofenac, ibuprofen, and naproxen; intraarticular injections of corticosteroids or hyaluronic acid; and oral and injected placebo.
The researchers analyzed data on pain, function, or stiffness from MEDLINE, EMBASE, Web of Science, Google Scholar, Cochrane Central Register of Controlled Trials, conference proceedings, the US Food and Drug Administration registry, ClinicalTrials.gov, product inserts, and pharmaceutical company websites. The studies, published between 1980 and 2014, included 33,243 participants. Of the 137 studies included, 129 (32,129 participants) assessed pain, 76 (24,059 participants) analyzed function, and 55 (18,267 participants) considered stiffness.
All treatments relieved some knee OA pain after 3 months compared with oral placebo. Hyaluronic acid was best, with an effect size of 0.63 (95% credible interval, 0.39 – 0.88). The least effective, acetaminophen, had an effect size of 0.18 (95% credible interval, 0.04 – 0.33), which did not meet the prespecified boundary for clinical significance.
Only slightly more effective than acetaminophen was celecoxib, which was surprising because past studies had demonstrated a greater effect for the cox-2 inhibitor.
For relieving pain, injections were more effective than oral treatments, and placebo injections were more effective than oral nonsteroidal anti-inflammatory drugs. However, the researchers point out that the apparent superiority of intraarticular treatments may not reflect a placebo effect but, instead, relief from injecting any fluid into the joint space. “[A]ll placebos are not created equal,” write Lisa A. Mandl, MD, MPH, from the Hospital for Special Surgery, Weill Cornell Medical School, New York City, and Elena Losina, PhD, from Harvard Medical School, Boston, in an accompanying editorial.
For stiffness, naproxen, ibuprofen, diclofenac, and celecoxib were more effective than oral placebo and acetaminophen, and injected hyaluronic acid was better than injected placebo. However, injected placebo was not significantly better than oral placebo.
For function, all interventions except injected corticosteroids were better than oral placebo. Naproxen, ibuprofen, diclofenac, and celecoxib were more effective than acetaminophen. Hyaluronic acid was better than injected placebo or injected corticosteroids. The two types of placebo performed equivalently.
Limitations of the study include the short-term data, possible publication bias, and indirect comparisons.
Dr Mandl and Dr Losina write that discordant study findings on treatment efficacy may reflect heterogeneity of pain etiology in knee OA. Inflammation, nociceptive pain, and central sensitization contribute to different degrees in different individuals. They write, “A clear understanding of the role of placebo, pain pathophysiology, and patient preferences should be key factors facilitating shared decision making in treating patients with knee OA.”
Support for this research was provided by the Agency for Healthcare Research and Quality. One coauthor reports receiving grants from Croma and Flexion Therapeutics and personal fees from Bioventus. The other authors and editorialists have disclosed no relevant financial relationships.
Ann Intern Med. 2015;162:46-54, 71-72. Article full text, Editorial full text