Placebo responses in US pain trials grew for two decades. Drug responses did not.

Between 1990 and 2013, the placebo arms of US trials of drugs for chronic neuropathic pain improved more and more, while the drug arms stayed where they were [s1]. The gap between them — which is what a trial measures, and what a regulator approves on — shrank for reasons that had nothing to do with the drugs being tested. Anyone reading a pain trial needs to understand this, because it changes what a negative result means.

The finding

Researchers extracted data from published randomised controlled trials of drugs for chronic neuropathic pain across that 24-year window [s1]. Placebo responses increased considerably over the period. Drug responses remained stable. The result was a diminishing treatment advantage — the appearance of drugs getting worse, produced entirely by controls getting better [s1].

The trend was driven by studies conducted in the United States [s1]. In the US, but not elsewhere, trials grew in size and in length over the same period, and both of those changes were associated with larger placebo responses [s1]. This is not a mysterious cultural effect; it is a set of design variables that happened to shift in one country.

The analysis also examined how the two responses evolve within a trial, and found different kinetics: the treatment response develops more quickly and then plateaus, while the placebo response keeps climbing, so that the maximum separation between the arms is reached within four weeks [s1]. A twelve-week trial of an analgesic is therefore not a more rigorous version of a four-week trial. It is a trial whose primary endpoint sits past the point of maximum measurable advantage.

Where you recruit changes the answer

The same phenomenon has been measured in a different disease with a different design. An analysis of 124 randomised trials of targeted therapies in rheumatoid arthritis, covering 14,272 patients, tested whether the socioeconomic profile of the recruiting countries predicted placebo response [s5].

It did. Using per-capita gross national income weighted by recruiting centres as the proxy, placebo response rates fell by 3.7 percentage points per 10,000 international dollars of national income (95% CI −5.61 to −1.80; P < .001) [s5]. The finding was confirmed using the Human Development Index and out-of-pocket health expenditure as alternative metrics, and using geographic data at individual patient level [s5]. Crucially, active treatment responses remained stable across the same gradient — so this is a placebo-arm phenomenon, not a general responsiveness phenomenon [s5].

The authors' interpretation is that recruiting from lower-income countries brings incentives that raise placebo response, such as limited access to therapies outside a trial [s5]. Whatever the mechanism, it means that two trials of the same drug, run to the same protocol in different countries, can produce different verdicts.

Placebo responses are not nothing

There is a habit of reading "placebo response" as "no real effect", and it does not survive contact with the trials that gave people a placebo openly.

An open-label placebo trial randomised 97 adults with persistent low back pain of more than three months' duration, diagnosed by a board-certified pain specialist, to treatment as usual or treatment as usual plus three weeks of pills they were told were placebos; 83 completed [s3]. Reduction on the composite 0–10 pain scale was 1.5 points (95% CI 1.0–2.0) in the open-label placebo group against 0.2 (−0.3 to 0.8) with treatment as usual, at P < 0.001, with moderate to large effect sizes [s3]. Disability improved by 2.9 points (1.7–4.0) against 0.0 (−1.1 to 1.2) [s3]. When the treatment-as-usual group was subsequently given placebo pills as well, they showed reductions of 1.5 (0.8–2.3) in pain and 3.4 (2.2–4.5) in disability [s3].

People who knew they were taking placebos reported meaningfully less pain. That result has an obvious limit, and a separate research group tested it: a three-year follow-up of another open-label placebo trial in chronic low back pain, using records from 89 previously enrolled patients, found no differences in any outcome between the groups with and without previous open-label placebo treatment [s4]. The authors state plainly that their follow-up does not support the assumption that a three-week open-label placebo treatment has long-term effects [s4].

Short-term real, long-term absent, is the fair summary.

Why this matters for reading any pain study

Four practical consequences follow.

An analgesic trial that fails is not necessarily testing a useless drug; it may be testing an active drug against an unusually responsive control arm [s1]. That is a reason for caution in both directions, not a licence to dismiss negative trials.

Trial length is not a neutral parameter in pain research. If treatment advantage peaks within four weeks [s1], then a longer trial systematically reports a smaller effect for the same drug.

Comparisons across trials conducted in different countries or eras are not like-for-like, because the placebo arm is not a fixed reference point [s1] [s5].

And any uncontrolled report of improvement in pain — a case series, a clinic's own outcome data, a before-and-after — is close to uninformative, because the placebo arms of formal trials improve substantially on their own [s1] [s3].

A recent review of placebo responses in analgesic trials frames the field's central difficulty in exactly these terms: the magnitude of the placebo response varies across pain syndromes and with the treatment being tested, it has identifiable predictors, and distinguishing a true therapeutic effect from it is a persistent challenge rather than a solved problem [s2].

What to watch

Whether analgesic trials shorten their primary endpoints toward the four-week window where separation is greatest [s1], and whether the geography of recruitment starts being reported and adjusted for the way other trial characteristics are [s5]. Both are cheap methodological changes with large effects on what gets approved.

This article is informational and is not medical advice.

Sources

• Increasing placebo responses over time in U.S. clinical trials of neuropathic pain — Pain, 2015-08-26
• Placebo Responses in Clinical Trials of Analgesics — Neurologic Clinics, 2025-09-16
• Open-label placebo treatment in chronic low back pain: a randomized controlled trial — Pain, 2016-10-14
• No long-term effects after a 3-week open-label placebo treatment for chronic low back pain: a 3-year follow-up of a randomized controlled trial — Pain, 2022-08-10
• Global recruitment patterns and placebo responses in clinical trials of rheumatoid arthritis — Annals of the Rheumatic Diseases, 2025-08-05