2019 American College of Rheumatology/Arthritis Foundation Guideline for the Management of Osteoarthritis of the Hand, Hip, and Knee

This is not a single experiment but a clinical practice guideline. The authors systematically reviewed the entire published literature on osteoarthritis (OA) management to answer 36 specific clinical questions. For each question, they compared an intervention against a control (placebo, no treatment, or another active treatment) and rated the quality of the evidence from randomized controlled trials (RCTs), observational studies, and systematic reviews.

The interventions tested fall into six categories:

**Educational and behavioral:** Self-management programs, self-efficacy training, weight loss programs, cognitive behavioral therapy (CBT)

**Physical and exercise:** Land-based exercise, aquatic exercise, balance training, tai chi, yoga, walking programs

**Mechanical and assistive:** Cane use, hand orthoses (splints), knee braces (tibiofemoral and patellofemoral), kinesiotaping, footwear modifications

**Mind-body and complementary:** Acupuncture, thermal modalities (heat/cold), radiofrequency ablation

**Topical pharmacologic:** Topical NSAIDs (diclofenac, ketoprofen), topical capsaicin, lidocaine patches

**Oral pharmacologic:** Oral NSAIDs (ibuprofen, naproxen, celecoxib, etc.), acetaminophen, duloxetine, tramadol, chondroitin sulfate, glucosamine, intraarticular glucocorticoid injections, hyaluronic acid injections

The primary outcome was pain reduction. Secondary outcomes included physical function, quality of life, joint stiffness, and adverse events (side effects, toxicity, withdrawal due to side effects).

The guideline synthesizes evidence from hundreds of studies. The target population is adults with symptomatic osteoarthritis of the hand, hip, or knee. Specific populations varied by study:

**Knee OA:** Typically adults aged 50–80, with radiographic Kellgren-Lawrence grade 2–4 (moderate to severe), body mass index (BMI) ranging from 25–40 kg/m², and pain on most days for at least 3 months

**Hip OA:** Adults aged 45–80, with radiographic hip OA, often with BMI >25

**Hand OA:** Adults aged 40–75, with symptomatic OA of the carpometacarpal (CMC) joint of the thumb or interphalangeal joints

**Exclusions common across studies:** Inflammatory arthritis (rheumatoid, psoriatic), gout, recent joint surgery, corticosteroid injection within 3 months, severe comorbidities (heart failure, renal disease), pregnancy, allergy to study drugs

Sample sizes in individual RCTs ranged from 30 to over 1,000 participants. The guideline's voting panel included 14 rheumatologists, 1 internist, 2 physical therapists, 1 occupational therapist, and 2 patients with OA.

The guideline used standardized outcome measures from the included studies:

**Pain:** Visual Analog Scale (VAS, 0–100 mm, 0 = no pain, 100 = worst imaginable pain), Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain subscale (0–20, lower = less pain), Knee Injury and Osteoarthritis Outcome Score (KOOS) pain subscale

**Physical function:** WOMAC function subscale (0–68), Health Assessment Questionnaire (HAQ), Timed Up and Go test, 6-minute walk test

**Global assessment:** Patient Global Assessment (PGA, 0–100 mm VAS), physician global assessment

**Quality of life:** Short Form-36 (SF-36), EuroQol-5D (EQ-5D)

**Adverse events:** Self-reported side effects, serious adverse events (hospitalization, death), withdrawal due to adverse events, laboratory abnormalities (liver enzymes, kidney function for NSAIDs)

**For hand OA specifically:** Australian/Canadian Osteoarthritis Hand Index (AUSCAN, 0–100), grip strength (dynamometer), pinch strength

The quality of evidence was rated using GRADE methodology: high, moderate, low, or very low. A "strong" recommendation means the panel was confident the benefits clearly outweigh harms (or vice versa). A "conditional" recommendation means the balance is less certain, and individual patient preferences matter more.

**Study design:** This is a clinical practice guideline developed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) methodology. It is not a single study but a systematic synthesis of existing evidence.

**Process:**

1. A Literature Review Team (LRT) conducted a systematic literature search of MEDLINE, Embase, and the Cochrane Library from January 2012 to August 2018 (updating the 2012 guideline). They identified 2,187 abstracts, of which 192 full-text articles were reviewed.

2. For each of the 36 clinical questions, the LRT extracted data on benefits and harms, rated the quality of evidence (high/moderate/low/very low), and produced evidence summaries.

3. A Voting Panel (19 members including patients) reviewed the evidence summaries and voted on recommendations. Consensus required ≥70% agreement.

4. Recommendations were classified as "strong" (we recommend) or "conditional" (we suggest) for or against each intervention.

**What this design can prove:**

The guideline provides the best available synthesis of evidence for OA management as of 2019.

Strong recommendations are based on high-quality evidence from multiple RCTs with consistent results.

The inclusion of patient representatives ensures recommendations reflect real-world values and preferences.

**What this design cannot prove:**

It cannot prove causality for any single intervention — that depends on the underlying RCTs.

It cannot account for individual patient variability (n-of-1 effects). A "conditional" recommendation does not mean the intervention is ineffective for you.

It cannot compare interventions head-to-head in a controlled way (no direct network meta-analysis was performed for most comparisons).

The guideline is limited by the quality of the underlying studies. Many OA trials have high dropout rates, short durations (4–12 weeks), and lack blinding for physical interventions.

**Major methodological weaknesses:**

Many physical therapy and exercise studies cannot be blinded (participants know they are exercising), introducing performance bias.

Placebo effects in OA are large (often 30–50% pain reduction in placebo groups), making it hard to detect true treatment effects.

The guideline did not perform a formal network meta-analysis to rank interventions by effect size.

Industry funding was present for many pharmacologic trials, though the guideline attempted to account for this.

The guideline is now several years old (published 2020, literature search through 2018). New evidence on glucosamine, CBD, and biologic agents is not included.

**Strong recommendations (benefits clearly outweigh harms):**

**Exercise (land-based):** Strong recommendation for knee and hip OA. Meta-analyses show exercise reduces pain by 0.49 standard deviations (moderate effect) compared to no exercise. This translates to approximately 12–15 points on a 0–100 pain scale. Effect size is similar for aerobic and resistance training.

**Weight loss:** Strong recommendation for knee and hip OA in overweight/obese patients (BMI ≥25). A 5–10% body weight loss reduces pain by 0.2–0.5 standard deviations. For a 90 kg person, losing 4.5–9 kg produces clinically meaningful pain reduction.

**Self-efficacy and self-management programs:** Strong recommendation. Programs like the Arthritis Self-Management Program (ASMP) reduce pain by 0.1–0.2 standard deviations (small effect) and improve self-efficacy scores by 0.3–0.4 standard deviations.

**Tai chi:** Strong recommendation for knee OA. A meta-analysis of 7 RCTs (n=348) found tai chi reduces pain by 0.5 standard deviations (moderate effect) and improves physical function by 0.6 standard deviations compared to no treatment or usual care.

**Cane use:** Strong recommendation for knee and hip OA. One RCT (n=64) found cane use reduced pain by 1.1 points on a 0–10 scale (95% CI 0.4–1.8) and improved function.

**Hand orthoses (splints) for first CMC joint OA:** Strong recommendation. A systematic review of 6 RCTs found thumb orthoses reduce pain by 1.5 points on a 0–10 scale (95% CI 0.8–2.2) and improve grip strength by 2–4 kg.

**Tibiofemoral knee brace:** Strong recommendation for tibiofemoral knee OA. One RCT (n=119) found a valgus brace reduced pain by 1.8 points on a 0–10 scale (95% CI 0.9–2.7) compared to no brace.

**Topical NSAIDs for knee OA:** Strong recommendation. Meta-analysis of 13 RCTs (n=3,200) found topical diclofenac reduces pain by 0.4 standard deviations (moderate effect) compared to placebo gel. Number needed to treat (NNT) for 50% pain relief = 6 (95% CI 4–10).

**Oral NSAIDs:** Strong recommendation for knee, hip, and hand OA. Meta-analysis of 74 RCTs (n=58,556) found oral NSAIDs reduce pain by 0.3–0.5 standard deviations (moderate effect). NNT for 50% pain relief = 4–8 depending on the specific NSAID. However, gastrointestinal adverse events (dyspepsia, ulcers) occur in 10–20% of users, and cardiovascular risk increases with long-term use (relative risk 1.3–1.5 for myocardial infarction with high-dose diclofenac or ibuprofen).

**Intraarticular glucocorticoid injections for knee OA:** Strong recommendation. Meta-analysis of 27 RCTs (n=1,500) found corticosteroid injections reduce pain by 0.4–0.6 standard deviations at 2–4 weeks post-injection. Effect diminishes by 8–12 weeks. No evidence of long-term cartilage damage with limited use (≤4 injections per year).

**Conditional recommendations (benefits likely outweigh harms, but less certainty):**

**Balance exercises:** Conditional for knee and hip OA. Limited evidence from 3 small RCTs (n=120) shows balance training reduces pain by 0.3 standard deviations and improves fall risk.

**Yoga:** Conditional for knee OA. One RCT (n=36) found 8 weeks of yoga reduced pain by 1.2 points on a 0–10 scale compared to no treatment. Quality of evidence is low due to small sample size and lack of blinding.

**Cognitive behavioral therapy (CBT):** Conditional for knee and hip OA. Two RCTs (n=250) found CBT combined with exercise reduces pain by 0.2–0.3 standard deviations more than exercise alone. Effect is larger for pain catastrophizing (0.5 standard deviations).

**Kinesiotaping for first CMC OA:** Conditional. One small RCT (n=40) found kinesiotape reduced pain by 1.0 point on a 0–10 scale compared to sham tape at 4 weeks.

**Acupuncture:** Conditional for knee OA. Meta-analysis of 39 RCTs (n=20,827) found acupuncture reduces pain by 0.2–0.3 standard deviations compared to sham acupuncture (needles at non-acupuncture points). The effect is small and may be partially due to placebo.

**Thermal modalities (heat/cold):** Conditional for knee and hand OA. Limited evidence from 4 small RCTs (n=160) shows heat packs reduce pain by 0.3–0.5 standard deviations for 30–60 minutes post-application. Cold packs reduce swelling but have inconsistent effects on pain.

**Radiofrequency ablation (RFA) for knee OA:** Conditional. Two RCTs (n=150) found RFA of the genicular nerves reduces pain by 2–3 points on a 0–10 scale at 3 months compared to sham procedure. Effect lasts 6–12 months. Quality of evidence is moderate.

**Topical capsaicin for knee OA:** Conditional. Meta-analysis of 5 RCTs (n=600) found 0.025% capsaicin cream reduces pain by 0.3 standard deviations compared to placebo cream. Burning sensation at application site occurs in 40–60% of users.

**Acetaminophen:** Conditional for hand, knee, and hip OA. Meta-analysis of 10 RCTs (n=3,500) found acetaminophen (up to 4 g/day) reduces pain by only 0.1–0.2 standard deviations — a small, often clinically insignificant effect. No benefit over placebo for moderate-to-severe OA pain.

**Duloxetine:** Conditional for knee OA. Two RCTs (n=500) found duloxetine 60 mg/day reduces pain by 0.3–0.4 standard deviations at 10–13 weeks compared to placebo. NNT for 50% pain relief = 8 (95% CI 5–15). Side effects (nausea, dry mouth, fatigue) occur in 20–30% of users.

**Tramadol:** Conditional for knee and hip OA. Meta-analysis of 7 RCTs (n=2,000) found tramadol (200–300 mg/day) reduces pain by 0.2–0.3 standard deviations compared to placebo. NNT for 50% pain relief = 10 (95% CI 6–20). Adverse events (nausea, dizziness, constipation) occur in 30–50% of users. Risk of dependence with long-term use.

**Strong recommendations AGAINST:**

Glucosamine (any formulation)

Chondroitin sulfate (for knee and hip — conditional recommendation FOR hand OA)

Hyaluronic acid injections (for knee OA — conditional recommendation against)

Bisphosphonates

Hydroxychloroquine

Methotrexate

To translate these numbers into plain English:

**Exercise** is roughly as effective as taking a standard dose of ibuprofen (400 mg) — about a 15–20% reduction in pain on a 0–100 scale. This is enough that most people notice a difference, but not enough to eliminate pain entirely.

**Weight loss** of 5–10% body weight produces a pain reduction equivalent to about half a standard painkiller dose. For someone weighing 90 kg, losing 5–9 kg would reduce knee pain by about 1–2 points on a 0–10 scale.

**Tai chi** produces a pain reduction similar to exercise — about 15–20% improvement — but with added benefits for balance and mood.

**Topical NSAIDs** (diclofenac gel) reduce knee pain by about 20–25% — roughly equivalent to taking oral ibuprofen but with fewer stomach side effects. You would need to treat 6 people to get one person to experience 50% pain relief.

**Oral NSAIDs** are the most effective pharmacologic option, reducing pain by 25–35% on average. However, 1 in 5 users will experience stomach upset, and long-term use increases heart attack risk by 30–50%.

**Acetaminophen** is barely better than placebo — a 5–10% pain reduction. For moderate-to-severe OA, it is essentially ineffective.

**Corticosteroid injections** provide a strong but temporary effect — 30–40% pain reduction for 2–4 weeks, then wearing off by 8–12 weeks.

**Acupuncture** produces a small effect (10–15% pain reduction) that is only slightly better than sham acupuncture, suggesting a large placebo component.