Medication pain relief for osteoarthritis: what the evidence shows

Osteoarthritis is the most common form of arthritis, and for many adults over 50 the day-to-day question is simple: which medicines for the pain are backed by good evidence, and which carry risks worth knowing about? This is a report of what randomized trials, systematic reviews, and clinical guidelines have found. It is not medical guidance, and it does not tell anyone what to take.

The short version from the evidence

Two large rheumatology guidelines, from the American College of Rheumatology with the Arthritis Foundation (Arthritis & Rheumatology, 2020) and from OARSI (Osteoarthritis and Cartilage, 2019), reach a similar order of preference for drug pain relief. Topical NSAIDs come first for knee and hand osteoarthritis. Oral NSAIDs are effective but their use is shaped by a person’s other health conditions. Acetaminophen has a small effect. Duloxetine is a conditional option. Opioids are recommended against.

Topical NSAIDs: recommended first for knee and hand OA

Topical NSAIDs are creams, gels, or patches (diclofenac is the most studied) applied over the joint. The 2019 ACR/Arthritis Foundation guideline gives them a strong recommendation for knee osteoarthritis, and OARSI rated them at its highest evidence level for knee OA (Osteoarthritis and Cartilage, 2019).

The appeal is the balance of benefit and safety. In the largest network meta-analysis of pain drugs for knee and hip OA (192 randomized trials, 102,829 participants), topical diclofenac showed a high probability of a clinically meaningful benefit and the authors wrote it “should be considered as first line” for knee OA because much less drug reaches the bloodstream (BMJ, 2021). A Cochrane review of topical NSAIDs for chronic musculoskeletal pain found they help, though the size is modest: about one extra person in ten gets good pain relief compared with placebo, and the main downside is local skin reactions rather than internal harm (Cochrane, 2016). Most of this evidence is in knee OA, which is why the hand recommendation is weaker in some guidelines.

Oral NSAIDs: real relief, real risks that grow with age

Oral NSAIDs (for example ibuprofen, naproxen, diclofenac, and the COX-2 drugs such as celecoxib and etoricoxib) produce the most pain relief among the drugs tested head-to-head. In the 2021 network meta-analysis, diclofenac 150 mg/day and etoricoxib 60 mg/day ranked as the most effective oral options for pain and function (BMJ, 2021). The catch is that the doses and drugs that relieve the most pain also tend to carry the most risk.

The older-adult risk picture

Three risks matter, and each is more consequential with age or existing illness.

Gastrointestinal: all oral NSAID regimens raised the risk of upper gastrointestinal complications such as bleeding ulcers in a meta-analysis of individual patient data from randomized trials, roughly doubling to quadrupling the rate depending on the drug (The Lancet, 2013).

Cardiovascular: in that same analysis, high-dose diclofenac and high-dose ibuprofen increased the risk of a major vascular event (heart attack, stroke, or vascular death) by about a third. In people at higher baseline risk, that translated to roughly 7 to 8 additional major vascular events per 1,000 patients per year, of which about 2 were fatal. Naproxen carried less vascular risk than the others (The Lancet, 2013).

Kidney: NSAIDs can reduce blood flow to the kidneys. Reviews of NSAID safety describe a raised risk of acute kidney injury in older adults, especially alongside conditions like heart failure, high blood pressure, or diabetes, or with medicines such as diuretics and ACE inhibitors (British Journal of General Practice, 2016). This is why OARSI advises against any oral NSAID for people with cardiovascular comorbidity or frailty, and steers those with stomach risk toward COX-2 drugs or added stomach protection (Osteoarthritis and Cartilage, 2019).

Acetaminophen (paracetamol): a small effect

Acetaminophen has long been offered as a gentle first step, but the trial evidence is underwhelming. A BMJ meta-analysis found its effect on OA pain was too small to be clinically important (BMJ, 2015), and a Cochrane review reached the same verdict: across 10 trials and 3,541 people, paracetamol beat placebo by only about 3 points on a 0 to 100 pain scale, below the threshold usually considered meaningful, and it was linked to more abnormal liver blood tests of uncertain significance (Cochrane, 2019). Guidelines now treat it as a limited option rather than a reliable one.

Duloxetine: a modest option, mainly in the knee

Duloxetine is an antidepressant that also acts on pain-signaling pathways. Meta-analyses of randomized trials report that 60 to 120 mg/day reduces knee OA pain and improves function over about 10 to 13 weeks compared with placebo, with a modest effect size (Pain Medicine, 2015). The trade-off is tolerability: nausea, constipation, dry mouth, and drowsiness are common reasons people stop. The 2019 ACR/Arthritis Foundation guideline conditionally recommends it, sometimes as an add-on (Arthritis & Rheumatology, 2020).

Opioids: guidelines advise against them

The evidence has moved firmly against opioids for osteoarthritis. In the 2021 network meta-analysis, no opioid preparation reached a convincing pain benefit, while most raised the risk of side effects and dropouts; the authors concluded the benefit “does not outweigh the harm” in OA (BMJ, 2021). The SPACE randomized trial found opioids were not better than non-opioid medicines for pain-related function over 12 months in chronic back or hip/knee pain, and caused more side effects (JAMA, 2018). The 2019 ACR/Arthritis Foundation guideline conditionally recommends against non-tramadol opioids and treats tramadol as a fallback only when other options are not feasible (Arthritis & Rheumatology, 2020).

What the guidelines conclude

The pattern across ACR, OARSI, and the network meta-analysis is consistent. Start topical for knee and hand OA. Use oral NSAIDs where the benefit is judged to outweigh a person’s specific stomach, heart, and kidney risks. Expect little from acetaminophen alone. Consider duloxetine as a modest option. Avoid opioids. Where the evidence is thinner, for example the hand OA data or the exact size of duloxetine’s benefit, the guidelines say so.

This article is evidence reporting, not medical advice. Decisions about any medication, including whether a drug’s benefits outweigh its risks for you, should be made with a qualified clinician who knows your health history.

Common questions

Are topical NSAIDs as effective as pills for osteoarthritis?

For knee osteoarthritis, network meta-analyses and guidelines treat topical NSAIDs (especially diclofenac) as a first-line drug option because they provide meaningful relief with far less drug reaching the bloodstream, which lowers stomach, heart, and kidney risk (BMJ, 2021; Osteoarthritis and Cartilage, 2019). Oral NSAIDs tend to produce somewhat more pain relief on average, but with higher systemic risk. Most of the topical evidence is for the knee rather than hip OA.

Which oral NSAID is safest for the heart and stomach?

No oral NSAID is risk-free. In a large individual-patient meta-analysis, high-dose diclofenac and high-dose ibuprofen raised major vascular events by about a third, while naproxen carried less cardiovascular risk (The Lancet, 2013). COX-2 drugs may reduce stomach bleeding risk but do not remove cardiovascular concerns. The safest choice depends on a person’s own heart, stomach, and kidney profile, which is a decision for a clinician.

Does acetaminophen (paracetamol) work for arthritis pain?

The randomized-trial evidence shows only a very small effect. A Cochrane review found acetaminophen beat placebo by about 3 points on a 0 to 100 pain scale, below the level considered clinically important, and a BMJ meta-analysis reached a similar conclusion (Cochrane, 2019; BMJ, 2015). Guidelines now view it as a limited option rather than a dependable one.

Why do guidelines recommend against opioids for osteoarthritis?

Because the evidence shows the harms outweigh the benefits. The largest network meta-analysis found no opioid reached a convincing pain benefit while most increased side effects (BMJ, 2021), and the SPACE randomized trial found opioids were no better than non-opioid medicines for function over 12 months, with more adverse effects (JAMA, 2018). The 2019 ACR/Arthritis Foundation guideline conditionally recommends against non-tramadol opioids.

Why are NSAIDs riskier for adults over 50?

Age and common conditions raise the stakes. Reviews describe a higher risk of acute kidney injury with NSAIDs in older adults, particularly alongside heart failure, high blood pressure, diabetes, or medicines such as diuretics (British Journal of General Practice, 2016). Cardiovascular and gastrointestinal bleeding risks also climb with age and other illness, which is why OARSI advises against oral NSAIDs in people with cardiovascular comorbidity or frailty (Osteoarthritis and Cartilage, 2019).