You have taken something. Two hours have passed, then four, and the pain is exactly where it was. The first thing that comes to mind is that the dose was too low. The second, worse, is that the problem must therefore be serious.
In the vast majority of cases it is neither one nor the other. It is that painkillers, on musculoskeletal pain, do far less than everyone takes for granted, and this has been written in black and white in the studies for years, only nobody says it at the pharmacy.
This article does not tell you what to take or what to stop: the drug is prescribed by a doctor and the decision stays there. It tells you the other half, the part that is our job: what it means, for movement, that it is not working.
First of all: the signs that have nothing to do with the drug
There is a minority of situations in which the right question is not which painkiller, but how quickly to get seen. See a doctor, or go to the emergency room, if alongside the pain even one of these signs appears:
- Loss of bladder or bowel control, or sudden difficulty passing urine.
- Loss of sensation in the saddle area: inner thighs, perineum, genital and anal area.
- Weakness in a leg or an arm that worsens over hours or days, a foot that gives way, a grip that lets go.
- Fever or shivering along with the pain, especially with an active infection.
- Unintended weight loss, pain that gives no respite even lying down and wakes you at night, a history of cancer.
- A significant trauma, or even a modest one if you have osteoporosis or have been taking corticosteroids for some time.
These signs are there to decide when to get seen, not to self-diagnose. And none of them is recognised by how well the pill worked.
The number nobody tells you
Let us start with the uncomfortable part, with the real figures. They all concern back pain, because that is where the research is strongest: we do not automatically extend them to the knee or the shoulder.
The 2020 Cochrane review of anti-inflammatories for acute back pain pooled 32 randomised trials and 5,356 people. Compared with placebo, the difference in pain intensity was 7.29 points on a scale from 0 to 100 (moderate-quality evidence), and 2.02 points out of 24 for disability (high-quality evidence). The authors' wording is literal: the size of these effects is small and probably not clinically relevant.
Seven points out of a hundred. Not zero, but if you expected the pain to disappear, seven points out of a hundred is exactly the feeling of "it did nothing for me".
For paracetamol the picture is starker. The 2023 Cochrane overview, which brought together seven reviews covering 103 studies and 22,238 participants, reports high-quality evidence that in acute back pain there is no difference from placebo: 0.49 points out of 100, that is, nothing. The general conclusion of that overview is the most honest thing written on the subject: we found no high or moderate quality evidence that any of the drugs studied gives a large or medium effect on pain, either in acute or in chronic back pain.
In 2025 a meta-analysis in BMJ Evidence-Based Medicine widened the field to 301 placebo-controlled studies and 56 different treatments, surgery excluded. The result: roughly one in ten proved effective, with small analgesic effects. Among those found not effective for acute back pain are paracetamol and cortisone injections.
So: "it is not working" is not a malfunction. Statistically, it is one of the most common outcomes.
Three different reasons, and they lead to three different paths
1. It is doing what it can, and that is little
This is the case of the numbers above. The drug has worked as well as it is able to, and that amount is not enough to cover your pain. Nothing is broken in you: the molecule has a ceiling, and your pain sits above that ceiling.
What changes: you stop looking for the right pill and start looking for what moves the pain. Which position switches it on, which switches it off, what happens the day after you walk for half an hour. That information is worth more than a change of active ingredient, and it is the raw material of any sensible programme.
2. The pain is not (mainly) inflammatory
An anti-inflammatory acts on inflammation. If your pain is mechanical (it arrives with load, with a position held for a long time, with a specific movement, and changes when you change position), then the drug's target is not your target.
The typical sign: the drug dulls the background ache, but the movement that hurts keeps hurting just as much. This is very useful information, and it should be passed on to the physiotherapist in these words.
3. The system has become hypersensitive
In 2017 the IASP, the international association for the study of pain, adopted a third way of describing pain alongside the nociceptive and the neuropathic: nociplastic pain, that is, pain arising from altered nociception that cannot be fully explained either by tissue damage or by nerve lesion. The criteria proposed for the musculoskeletal system are four: pain lasting more than three months, a regional or widespread distribution rather than a localised one, the inability to explain it fully by the other two mechanisms, and clinical signs of hypersensitivity in the area.
If this is the picture, drugs designed for inflammation tend to do little by design. This does not mean the pain is imaginary: it means the alarm system has been set too low, and that is what is worked on.
And when the drug is the right move?
This is the part usually missing from articles written by physiotherapists, and it is what makes the rest credible.
In 2022 a randomised trial published in JAMA Network Open followed 100 people with long-standing Achilles tendinopathy. All of them did exercise; half also received an ultrasound-guided cortisone injection, half a placebo injection. At six months the cortisone group was 17.7 points better on the VISA-A score (confidence interval from 8.4 to 27.0), with no serious adverse events and no worsening at two years.
Seventeen points is a lot. But look closely at how the study was built: both groups did exercise. The cortisone did not replace rehabilitation, it made it possible. This is the right model for thinking about a drug within a musculoskeletal pathway: it opens a window, and something has to go through that window.
On the opposite side, the same 2023 Cochrane overview reports, for muscle relaxants in acute back pain, moderate-quality evidence of a small effect on pain, and moderate-quality evidence that they increase the risk of side effects. They help a little and cost something: it is a clinical decision, not an automatic one.
What not to do when it is not working
- Do not raise the dose on your own. The effect does not grow in proportion, the risk does. And the news that it did not work is more useful to the doctor than a secretly doubled dose.
- Do not stack different drugs hoping one will hit. If an anti-inflammatory did not work, the problem is not solved by adding a second one of the same kind.
- Do not take it to get through training or a shift. It is the most common mistake among athletes and people doing physical work: without the signal you do not know where you are, and the bill arrives the next day or the week after.
- Do not conclude that nothing works. It is the worst trap, because it leads to prolonged rest. In the same 2025 meta-analysis in which almost everything proved ineffective, exercise is one of the five treatments found effective for chronic back pain. The effect is small there too, and it should be said: nobody here is selling you a magic wand. But there is a difference that averages do not capture: the pill changes the afternoon, exercise changes the direction.
The three things to say, and to whom
When the drug has not worked, these three pieces of information really change the plan, for the doctor and the physiotherapist alike:
- What you took, how much and for how many days, and what happened: nothing, little, or relief that wore off before the next dose.
- Whether the movement that hurts keeps hurting just as much even when the background ache goes down. It is the sign that we are looking at a mechanical problem.
- How long it has lasted, and whether it is localised or widespread. Beyond three months and widespread, the reasoning changes, and so does the kind of work that makes sense to propose.
If you want someone to look at your case with this framework, start with the questionnaire: the first consultation lasts twenty minutes, is by video call and is free. We do not prescribe medicines (that is not our job and we do not pretend otherwise), but if the consultation shows the path is different, we will tell you and point you in the right direction. And if your case needs hands-on work, the physiotherapist who follows you will say so and refer you to one of our people near you.
Sources
- van der Gaag WH et al., Non-steroidal anti-inflammatory drugs for acute low back pain, Cochrane Database of Systematic Reviews, 2020
- Cashin AG et al., Pharmacological treatments for low back pain in adults: an overview of Cochrane Reviews, Cochrane Database of Systematic Reviews, 2023
- Cashin AG et al., Analgesic effects of non-surgical and non-interventional treatments for low back pain: a systematic review and meta-analysis of placebo-controlled randomised trials, BMJ Evidence-Based Medicine, 2025
- Johannsen F et al., Effect of Ultrasonography-Guided Corticosteroid Injection vs Placebo Added to Exercise Therapy for Achilles Tendinopathy: A Randomized Clinical Trial, JAMA Network Open, 2022
- Hayden JA et al., Exercise therapy for chronic low back pain, Cochrane Database of Systematic Reviews, 2021
- Kosek E, The concept of nociplastic pain: where to from here?, Pain, 2024
Last source check: September 2026.
