6.1 million people just redrew what “normal pain” means
A 118-country study of 6.1 million people gives pain its first age- and sex-adjusted reference curves.
By The Weekend · · 5 min read
A 43-year-old woman reporting a pain score of 6 out of 10 is not having the same experience as a 43-year-old man reporting the same number — and until now, nobody had the data to say by how much. A new analysis in Nature Medicine, built from 6.1 million people across 118 countries, produces the first global reference curves for pain across the lifespan, the pain equivalent of the growth charts pediatricians have used for decades.
A growth chart, but for hurting
Growth charts work because a single number — height at age 7 — means nothing on its own. It only becomes useful next to a curve built from millions of other 7-year-olds. Pain scores have never had that curve. A "7 out of 10" reported in a clinic in São Paulo, a trial in Nairobi, or an ER in Ohio has been treated as the same 7, regardless of the patient's age, sex, or where in the world they live. The Nature Medicine team built reference bands the same way pediatric growth charts are built: by pooling enormous numbers of individual measurements, then fitting curves that show what's typical at each age, for each sex, and letting a clinician see where a given patient sits against that backdrop rather than against an assumed flat average.
The scale of the pooling is what makes the curves usable rather than illustrative. Prior pain research has leaned on hospital cohorts in the thousands or tens of thousands — enough to publish, not enough to say what "normal" looks like for a 12-year-old in one country versus a 70-year-old in another. At 6.1 million individuals and 118 countries, this dataset crosses from describing a sample into describing a population, which is the actual threshold a reference curve needs to clear before anyone should use it to judge an individual patient.
Who gets measured differently once there's a curve to measure against
The practical shift lands first on two groups: older adults and women, both of whom have been systematically under- or over-read by flat pain scales for years. Chronic pain prevalence rises sharply with age in most national surveys, but clinicians have had no way to know whether a given older patient's reported pain is unusually severe for their age bracket or simply average — the scale didn't bend with the population, so every elevated score got treated with the same urgency regardless of context. A reference curve changes that calculus directly: a pain report that sits at the 95th percentile for a 75-year-old carries different clinical weight than the same raw number sitting at the 50th percentile for that age group, even though the number on the chart looks identical.
Sex differences are the second, more contested, beneficiary. Women have long reported higher pain scores than men across comparable conditions — a gap that's been read by different observers as evidence of greater physiological sensitivity, under-treatment bias, or measurement artifact, often depending on the observer's priors rather than the data. A global, age-adjusted reference curve doesn't resolve which explanation is correct, but it does something more useful in the near term: it lets a clinician see whether a specific patient's score is unusual for her demographic, rather than unusual relative to a generic "patient" that was implicitly modeled on neither sex in particular.
The percentile is only as honest as what feeds it
Here is the constraint the headline number obscures: a reference curve is a mirror of whoever got measured, and pain data skews hard toward people who are already inside a health system — hospital patients, trial participants, clinic attendees. People in chronic pain who never see a doctor, who manage with over-the-counter drugs, or who live in regions with thin healthcare infrastructure are structurally undercounted in any dataset built from clinical records, no matter how many countries it spans. A curve built predominantly from people who sought care will describe "pain among people who sought care," which is a narrower and more selected population than "pain," full stop.
This matters because the 118-country spread, while genuinely unusual for a pain study, almost certainly does not mean uniform sampling density — a dataset this size is near-certain to lean heavily on countries with large, digitized health records (strong representation from high-income health systems) and more thinly on countries where pain is tracked informally or not at all. A reference curve pooled this way isn't wrong, but its confidence intervals should be read as tighter where the data is thick and looser where it's thin, and the paper's per-country sample sizes — not published in the available summary — are the detail that would tell a clinician in, say, a lower-income country how much to trust the curve for their own patients.
What a clinician actually does differently on Monday morning
The use case is narrower and more useful than "better pain diagnosis" — it's percentile-based triage. A clinician currently has no standardized way to say "this pain report is unusual for this patient's age and sex" versus "this is within the expected range" — the scale has always been flat, 0 to 10, regardless of who's holding the pen. With a reference curve, a 10-year-old reporting a pain score that sits above the 90th percentile for their age band is a different clinical signal than the same raw score from a 65-year-old, for whom that score might sit closer to the median. That distinction is exactly the kind of thing growth charts already do for height and weight in pediatrics — flagging not an absolute threshold but a deviation from what's typical for the group — and it's the reason growth charts became a universal clinical tool rather than a research curiosity.
Whether this pain curve achieves the same adoption depends on something the paper can't settle by itself: whether health systems build it into the software that already runs intake forms and electronic records. A reference curve that exists only in a journal supplement changes nothing; one that gets embedded into the dropdown menu a nurse uses at triage changes a great deal, because it reframes every pain score entered from that point forward as relative rather than absolute. The falsifiable test over the next two to three years is simple: track whether any major electronic health record vendor — Epic, Cerner, or their national equivalents — incorporates age- and sex-adjusted percentiles into standard pain assessment screens. If none do, the curve will have been a genuine scientific advance that never became a clinical one, which is a more common outcome in medicine than anyone likes to admit.