Mixed Treatment Comparisons for Nonsurgical Treatment of Knee Osteoarthritis: A Network Meta-analysis.

Introduction: Knee osteoarthritis (KOA) is a significant health problem with lifetime risk of development estimated to be 45%. Effective nonsurgical treatments are needed for the management of symptoms.

Methods: We designed a network meta-analysis to determine clinically relevant effectiveness of nonsteroidal anti-inflammatory drugs, acetaminophen, intra-articular (IA) corticosteroids, IA platelet-rich plasma, and IA hyaluronic acid compared with each other as well as with oral and IA placebos. We used PubMed, EMBASE, and Cochrane Central Register of Controlled Trials to perform a systematic search of KOA treatments with no date limits and last search on October 7, 2015. Article inclusion criteria considered the following: target population, randomized controlled study design, English language, human subjects, treatments and outcomes of interest, ≥30 patients per group, and consistent follow-up. Using the best available evidence, two abstractors independently extracted pain and function data at or near the most common follow-up time.

Results: For pain, all active treatments showed significance over oral placebo, with IA corticosteroids having the largest magnitude of effect and significant difference only over IA placebo. For function, no IA treatments showed significance compared with either placebo, and naproxen was the only treatment showing clinical significance compared with oral placebo. Cumulative probabilities showed naproxen to be the most effective individual treatment, and when combined with IA corticosteroids, it is the most probable to improve pain and function.

Discussion: Naproxen ranked most effective among conservative treatments of KOA and should be considered when treating pain and function because of its relative safety and low cost. The best available evidence was analyzed, but there were instances of inconsistency in the design and duration among articles, potentially affecting uniform data inclusion.