Supplements trail the drug for knee arthritis pain

Knee osteoarthritis is a large and growing burden, and the mainstay drug treatment carries long-term safety questions that push people towards adjuvant options. Probiotics, metformin, krill oil and glucosamine with chondroitin have each produced encouraging results in individual trials, but they have never been compared with one another directly. Without head-to-head data, anyone choosing between them is effectively guessing which, if any, does more. This analysis attempted to place all five interventions on a single comparative footing using indirect evidence.

The authors searched PubMed, Web of Science, Embase and the Cochrane Library from inception through February 2026 for randomised controlled trials of the five interventions in knee osteoarthritis. Twenty-three RCTs met the inclusion criteria. A random-effects network meta-analysis was run, with continuous outcomes synthesised as mean differences and dichotomous safety endpoints as risk ratios. Treatments were ranked by surface under the cumulative ranking curve (SUCRA), and the quality of evidence for each comparison was graded using the GRADE framework. Outcomes covered visual analogue scale pain, WOMAC total score and its pain and physical function subscales, plus safety.

Celecoxib ranked highest for analgesic efficacy on the visual analogue scale, with a mean difference of -0.48 (95% CI -0.78 to -0.17) and a SUCRA of 0.93, and it also led on WOMAC total score (SUCRA 0.912), followed by glucosamine with chondroitin at 0.756. Probiotics (0.66) and metformin (0.63) showed potential efficacy but sat lower in the ordering. On the WOMAC subscales, celecoxib was associated with the largest improvements in pain and physical function, while krill oil ranked highest on the pain subscale alone at 0.99. For safety, probiotics and metformin had the most favourable profiles, with SUCRA values of 0.885 and 0.724. No significant network inconsistency or publication bias was detected.

This is the first attempt to put these five options into one comparative network, and it broadly confirms that the established anti-inflammatory drug remains the strongest performer on pain and function. What it cannot do is settle the adjuvant question: the authors themselves flag that certainty for most of those comparisons was low or very low, driven by severe imprecision and the complete absence of direct head-to-head trials. Krill oil topping a single subscale while ranking unremarkably elsewhere is exactly the kind of unstable result that small networks produce. The open question is whether any of the adjuvants holds up when tested against each other rather than against placebo.

This is a reason to keep expectations modest about supplement options for knee osteoarthritis, and to treat a high ranking as a hypothesis rather than a result. Rankings built from indirect comparisons can shuffle substantially once real head-to-head trials arrive, particularly when the underlying certainty is graded low or very low. The safety ordering is a separate consideration from the efficacy ordering, and the two did not line up here. Anyone weighing these options is doing so with genuinely incomplete evidence, which is itself worth knowing.