Treatment of osteoarthritis of the knee: evidence-based guideline, 2nd edition

Jevsevar DS · The Journal of the American Academy of Orthopaedic Surgeons

Recommendation 9 reverses the 2008 edition, in which viscosupplementation had been inconclusive: "We cannot recommend using hyaluronic acid for patients with symptomatic osteoarthritis of the knee", at Strong strength — which the guideline defines as two or more high-strength studies agreeing. The basis is clinical rather than statistical significance. Pooling like measurement instruments made it possible to determine that the overall effect of hyaluronic acid did not provide minimum clinically important improvement to patients, and the work group states that the strength of the recommendation rests on lack of efficacy, not on potential harm. Among the statistically significant outcomes, those associated with high-molecular-weight cross-linked hyaluronic acid did not hold their significance when compared against mid-range molecular weight. The search identified 9,279 citations, recalled 2,222 for full-text review, and carried 218 articles into the recommendations.

Population

AAOS evidence-based guideline, second edition

Intervention

Guideline recommendations for knee OA treatment

Limitations

A guideline, not a trial: it contributes standing rather than evidence, and the 2022 third edition supersedes it. Its verdict turns on selection criteria that differ from the review the 2008 edition relied on — a study needed 30 or more patients in each treatment group and at least 4 weeks of follow-up, and lower-strength evidence was excluded wherever two higher-strength studies assessed the same outcome. On those grounds the work group excluded the AHRQ systematic review the earlier edition had used, and says so. Publication bias is recorded against that earlier basis rather than against its own evidence: the guideline quotes AHRQ's finding that pooled results from trials of fewer than 100 patients showed effects up to twice those of larger trials. No harms analysis was performed for this recommendation. Where a minimum clinically important improvement was not available for this population the work group used thresholds from closely related ones, acknowledging that variability between diseases and subjectivity in what patients count as improvement can cause discrepancies.