Interventions for osteoarthritis pain


A systematic review with network meta-analysis



The study published in Osteoarthritis and Cartilage Open (2022), titled "Interventions for osteoarthritis pain: A systematic review with network meta-analysis of existing Cochrane reviews," provides a massive-scale synthesis of the available evidence for treating osteoarthritis (OA) pain. Conducted by Smedslund and colleagues, this research aims to move beyond simple "A vs. B" comparisons to create a comprehensive map of all treatments—pharmacological, surgical, and lifestyle-based—within a single statistical framework.

1. Introduction and Objectives

Osteoarthritis is a leading cause of global disability, characterized by the breakdown of joint cartilage and bone. While many clinical guidelines exist (e.g., ACR, OARSI, EULAR), they are often inconsistent in their recommendations for non-pharmacological and pharmacological treatments. This creates a "choice paradox" for clinicians and patients: although many treatments exist, it is unclear which is definitively "best."

The authors set out to perform a network meta-analysis (NMA), a statistical technique that allows for the comparison of multiple treatments simultaneously, even if they have never been compared head-to-head in a trial, by using a common comparator (usually placebo or "usual care"). The primary goal was to rank treatments by their effectiveness in reducing pain, identify research gaps, and assess the overall quality of evidence.

2. Methodology

The researchers utilized the Cochrane Library as their primary source, as it represents the "gold standard" for systematic reviews. They supplemented this with data from Epistemonikos to ensure recent trials were included.

  • Study Selection: They included 35 systematic reviews, which encompassed 445 randomized controlled trials (RCTs).

  • Participants: The data covered patients with knee, hip, or hand OA.

  • Interventions: They identified 153 unique treatments or combinations. To make the data manageable, they grouped these into 17 broad categories, including:

  • Pharmacological: Oral/topical NSAIDs, corticosteroids, analgesics (e.g., acetaminophen), opioids, and "slow-acting drugs" (SYSADOAs like glucosamine).

  • Non-Pharmacological: Exercise (land-based and aquatic), mind-body exercise (e.g., Yoga, Tai Chi), diet/weight loss, and orthotics (braces/insoles).

  • Invasive/Other: Surgery, regenerative medicine (e.g., PRP), passive treatments (e.g., massage, TENS), and herbal therapies.

  • Statistical Approach: They used a frequentist NMA approach. The primary outcome was pain, measured via Standardized Mean Differences (SMD). A predefined "Minimal Clinically Important Difference" (MCID) of 0.469 was used to determine if a treatment’s effect was meaningful for patients.

  • Quality Assessment: The authors used CINeMA (Confidence In the results of Network Meta-Analysis) software to evaluate the quality of the evidence based on factors like risk of bias, indirectness, and incoherence.

3. Key Results

A. The Landscape of Evidence

The study revealed a stark imbalance in research:

  • Knee OA was the most studied (491 comparisons).
  • Hip OA had significantly fewer studies.
  • Hand OA was severely neglected, with only nine comparisons available in the network.

B. Treatment Effectiveness

While many treatments showed statistically significant improvements over controls, only six categories exceeded the MCID (the threshold for clinical relevance):

  1. Corticosteroids (Injections): Showed the strongest short-term effect.
  2. Herbs: Various oral and topical herbal therapies showed moderate-to-high effects.
  3. Mind and Body Exercises: (e.g., Yoga, Tai Chi) performed remarkably well.
  4. Orthotics: Braces and specialized footwear showed meaningful pain reduction.
  5. Passive Treatments: Including manual therapies and TENS.
  6. Regenerative Medicine: Such as Platelet-Rich Plasma (PRP), though much of this evidence was derived from indirect comparisons.

Interestingly, "Diet/weight loss" and "Surgery" showed effect sizes close to zero in this specific meta-analysis. The authors noted this might be due to the heterogeneous nature of surgical trials and the difficulty of blinding in weight-loss studies.

C. Short-term vs. Long-term Effects

The study highlighted a major "decay" in treatment efficacy over time:

  • Corticosteroids: Had a large short-term effect (SMD -0.97) but became statistically insignificant after 12 weeks (SMD -0.36).
  • NSAIDs: Followed a similar pattern, with effects dropping from -0.4 to -0.06 as the duration increased.
  • Exercise: Remained one of the most stable interventions, maintaining small-to-moderate effects (SMD -0.38) across both direct and indirect analyses.

D. The "Confidence" Problem

The most critical finding regarding the results was the extremely low confidence in the data. Out of 136 treatment comparisons:

  • 0% were rated as "High Confidence."
  • Only 6 (4.4%) were rated as "Moderate Confidence."
  • The vast majority (86%) were rated as "Very Low Confidence."

4. Discussion: Why the Network Failed to Rank

The authors originally intended to rank the 153 treatments from most to least effective. However, they concluded that producing a ranking would be misleading.

The primary reason was incoherence and heterogeneity. In a network meta-analysis, the "indirect" evidence (calculating A vs. C by using B as a bridge) should ideally match the "direct" evidence (actually testing A vs. C). In this study, these values often disagreed. This occurred because:

  • Control Groups: What one study calls "placebo," another might call "usual care" or "education." These are not the same, but the NMA often treats them as a single "control" node.
  • Population Diversity: Trials included everyone from elite athletes with minor knee pain to elderly patients awaiting total joint replacement.
  • Reporting Bias: Many older trials lacked the rigorous reporting standards of modern Cochrane reviews.

5. Conclusions and Clinical Implications

The study concludes that while the medical community wants a clear hierarchy of OA treatments, the current state of clinical research makes it impossible to provide one with certainty.

Key Takeaways:

  • Exercise is the Anchor: While its effect size is smaller than a fresh steroid shot, it is the most consistent and durable intervention across all analyses.
  • The "Core" Treatments are Valid: Diet, exercise, and education remain the foundation, even if the "diet" data in this NMA was statistically weak (likely due to trial design).
  • Short-term vs. Long-term: Patients should be warned that pharmacological "fixes" like NSAIDs and steroids have a high failure rate over long durations.
  • The Need for Standardization: Future OA trials must better define their "control" groups and include more diverse joint locations (specifically hand and hip).

Final Summary: This study is a "wake-up call" for evidence-based medicine in rheumatology. It proves that despite having hundreds of trials and thousands of patients, our "confidence" in which treatment is truly superior remains "very low." For now, treatment remains a highly individualized process of trial and error, prioritizing low-risk, consistent interventions like exercise over short-lived pharmacological "bursts."

Link: https://pmc.ncbi.nlm.nih.gov/articles/PMC9718209/