Need help now? Support now: crisis lines by country.
Questions about pain · Mathematical Institute for Research of Pain
Why do painkillers stop working?
There are several different reasons, and they call for different conversations. The body can adapt to some medicines; the condition itself can change; with opioids, some people may become more sensitive to pain, though the evidence in patients is mixed; frequent use of headache medicines can lead to more headaches; and the natural ups and downs of pain change how relief looks over time.
Life is short, and the Art long; the occasion fleeting; experience fallacious, and judgment difficult.
The short answer
When a medicine that used to help seems to help less, it is natural to assume it has worn out. That is one possibility, called tolerance, but it is not the only one. The pain may have changed. The medicine may, in some cases, be adding to the pain. And starting a medicine during an unusually bad spell can make its early effect look larger than it is, so it later seems weaker even though it works as well as before.
These explanations point in different directions, so working out which fits is a job for the person and their clinician together. This page does not suggest starting, stopping or changing any medicine.
What is known
Tolerance. Tolerance means that a higher dose is needed to get the same effect. A review in The Lancet by Colvin, Bull and Hales (2019) describes opioid tolerance in exactly these terms. In a preliminary study, Chu, Clark and Angst (2006) tested six people with chronic low back pain before and one month after they started oral morphine. In a cold-water test, a short-acting opioid given in the laboratory relieved the pain less than before in all six, a sign of tolerance, and all six had become more sensitive to the cold pain. In a heat-pain test they showed no increase in sensitivity, and tolerance could not be measured in that test. The authors called the study preliminary. The same group then ran a double-blind randomised trial in which 103 people with chronic low back pain completed a month of either slow-release morphine or a placebo (Chu and colleagues, 2012). Those on morphine developed tolerance to the laboratory opioid, on average a 42% reduction in its potency, but did not develop increased pain sensitivity, and their back pain improved more than with placebo.
Opioid-induced hyperalgesia. Colvin and colleagues define this as a paradoxical increase in pain with opioid administration; in research it is usually measured as increased sensitivity in laboratory pain tests. A systematic review by Higgins, Smith and Matthews (2019) brought together 26 studies with 2,706 participants. People who had taken opioids for a long time tolerated heat and cold pain less well in tests, but there was no difference with electrical stimuli and no difference in the level at which pain was first felt. The effect was clearer in people with opioid use disorder than in people being treated for pain, and in groups given drugs that also block NMDA receptors, mostly people taking methadone; the authors called for better control of other factors that affect pain judgements. How often it matters for a patient’s everyday pain is not known.
Medication-overuse headache. The International Classification of Headache Disorders (ICHD-3) defines this as headache on 15 or more days a month in someone who already has a headache disorder, developing as a consequence of regular overuse of acute or symptomatic headache medication for more than three months. Overuse means use on 10 or more, or 15 or more, days a month, depending on the medicine. The classification says it usually, but not invariably, resolves after the overuse is stopped. This diagnosis applies to headache only; it does not say the same happens with pain elsewhere in the body. Any change to headache medicines is a decision to make with a clinician.
Regression to the mean. People often start a new medicine when their pain is unusually bad. Pain that is unusually high tends to be followed by pain closer to the person’s usual level, whatever is done, a pattern called regression to the mean (Barnett, van der Pols and Dobson, 2005). Some of the early improvement can therefore come from that return, which makes the medicine look stronger at first than it is. Later, compared with that first impression, it can seem to fade, even if its own effect has not changed. Englund and Turkiewicz (2023) applied the entry criteria typical of knee osteoarthritis trials to people in a large population study, with no trial treatment given, and found that regression to the mean alone produced an improvement of about 1 point on a 0 to 10 pain scale.
Expectation. Bingel and colleagues (2011) gave healthy volunteers the same infusion of the opioid remifentanil during heat pain and changed only what they were told. Compared with when they did not expect relief, expecting relief about doubled the pain relief volunteers reported; when they were led to expect that the drug had been stopped and their pain would get worse, its pain-relieving effect disappeared, although the same infusion was still running. The study was in healthy volunteers in the laboratory, but it shows that what a person expects can change the effect of the same drug at the same dose.
A changing condition. Pain that rises over months may reflect the condition itself changing, or a new source of pain, rather than anything about the medicine; researchers who study tolerance treat advancing disease as a separate explanation (Chu and colleagues, 2012).
What is not yet known
There is no simple test that tells, for one person, which of these explanations is at work, and several can act at once. The human research on tolerance and hyperalgesia uses laboratory pain tests, often in small groups, and it is not clear how closely those results track everyday pain. Most of the evidence on this page concerns opioids and headache medicines; it does not show that every painkiller behaves the same way.
What would tell us
A record helps a clinician tell these apart. Useful things to bring up are when the relief started to fade and how quickly; whether the pain has spread or changed in character; for headaches, how many headache days and how many medication days there are each month, which is how ICHD-3 counts; and whether the medicine was started during an unusually bad spell. Pain that is new, different in kind or steadily rising is worth raising promptly.
A worked example
One common model of how relief depends on dose is
,
where is the most relief the medicine can give and is the dose that gives half of it. With invented numbers in invented dose units, take points and . A dose of 10 gives
points.
Tolerance, in this model, raises . If it doubles to 20, the same dose gives
points.
Regression to the mean looks different. Suppose, again with invented numbers, that someone’s usual level is 6, they start a medicine on a bad day rated 8, and the medicine truly gives 2 points of relief. Their ratings then settle around
.
Measured against the starting day, the medicine seems to give points of relief, twice its true effect. Over the following months ratings stay around 4, which is only points below the usual level, and a bad day can still reach 6. Against that first impression of 4 points, the medicine now seems to give only half as much. In this model the medicine did not stop working; the first comparison was flattered by an unusually bad starting day.
These pages explain reasoning and evidence. Nothing here is medical advice or a forecast about any person.
Sources
- Colvin LA, Bull F, Hales TG. Perioperative opioid analgesia—when is enough too much? A review of opioid-induced tolerance and hyperalgesia. Lancet. 2019;393(10180):1558–1568. doi:10.1016/S0140-6736(19)30430-1
- Chu LF, Clark DJ, Angst MS. Opioid tolerance and hyperalgesia in chronic pain patients after one month of oral morphine therapy: a preliminary prospective study. J Pain. 2006;7(1):43–48. doi:10.1016/j.jpain.2005.08.001
- Chu LF, D’Arcy N, Brady C, et al. Analgesic tolerance without demonstrable opioid-induced hyperalgesia: a double-blinded, randomized, placebo-controlled trial of sustained-release morphine for treatment of chronic nonradicular low-back pain. Pain. 2012;153(8):1583–1592. doi:10.1016/j.pain.2012.02.028
- Higgins C, Smith BH, Matthews K. Evidence of opioid-induced hyperalgesia in clinical populations after chronic opioid exposure: a systematic review and meta-analysis. Br J Anaesth. 2019;122(6):e114–e126. doi:10.1016/j.bja.2018.09.019
- Headache Classification Committee of the International Headache Society. The International Classification of Headache Disorders, 3rd edition (ICHD-3): 8.2 Medication-overuse headache. Source
- Barnett AG, van der Pols JC, Dobson AJ. Regression to the mean: what it is and how to deal with it. Int J Epidemiol. 2005;34(1):215–220. doi:10.1093/ije/dyh299
- Englund M, Turkiewicz A. Pain in clinical trials for knee osteoarthritis: estimation of regression to the mean. Lancet Rheumatol. 2023;5(6):e309–e311. doi:10.1016/s2665-9913(23)00090-5
- Bingel U, Wanigasekera V, Wiech K, et al. The effect of treatment expectation on drug efficacy: imaging the analgesic benefit of the opioid remifentanil. Sci Transl Med. 2011;3(70):70ra14. doi:10.1126/scitranslmed.3001244
Related questions
Pain words: Hyperalgesia · Regression to the mean · Placebo · Nocebo · All words
Related, in Nociflux, the Institute’s ecosystem for living with pain: the visit report and the N-of-1 Lab.
Inside the Institute
The Pain Bias Codex gathers checks on conclusions about pain, in four families. With an invite code you can read this check there beside the others, with its full evidence notes. Pain math explains what the Institute does.
No code yet? Ask for an invite. A person reads every request and writes back by hand with an invite code, so it may take a few days.
Need help now? Support now: crisis lines by country.
Except where a page says otherwise, the Institute’s own text and figures are shared under CC BY-NC 4.0. Pictures and quotations from others keep their own credits and licences.