Boulder Back Pain Study: What It Really Found

You have probably met the headline number already. Two thirds pain-free, no drugs, no surgery, and the effect still measurable a year later.

It is a real result from a real randomised trial. It is also, in nearly every retelling online, stretched past what the trial actually measured. So here is the study itself: who was in it, what was done to whom, what changed, and where the honest edges are.

What was the Boulder back pain study?

It was a randomised clinical trial run at the University of Colorado Boulder, with 151 adults recruited from the surrounding community. Average age 41. Fifty-four percent women. To qualify you had to be between 21 and 70, have had back pain on at least half the days of the previous six months, and rate it at least 4 out of 10 in a typical week (Ashar et al., 2022).

The people who enrolled had carried this for a long time. Mean pain duration was ten years. Their pain was persistent rather than severe, averaging 4.1 out of 10 at the start, with moderate disability scores.

Everyone had what is called primary chronic back pain, meaning pain with no identifiable cause in the tissue. That is not an unusual corner of the problem. Around 85 percent of chronic back pain falls into this category. People whose leg pain was worse than their back pain were excluded, since that points toward a nerve root rather than the pattern under study.

Participants were then randomly assigned to one of three groups.

• Pain Reprocessing Therapy. One telehealth session with a physician, followed by eight treatment sessions over four weeks. The aim was to help people reconceive their pain as brain activity that is not reporting damage in the back, using a mix of cognitive, body-focused and exposure techniques.
• Open-label placebo. A saline injection into the site of worst back pain, given by a physician who told participants plainly that it was saline. Beforehand they watched two videos explaining that placebos can relieve pain even when you know they are inert.
• Usual care. Carry on with whatever care you were already receiving, and start nothing new until the post-treatment assessment.

The primary outcome was deliberately plain: your average pain over the past week, rated 0 to 10, measured one month after baseline. Everyone was then followed for a year.

What did the trial actually find?

Start with the primary outcome, because that is the number the trial was designed to test. At the end of four weeks, average pain was 1.18 out of 10 in the PRT group, 2.84 in the placebo group and 3.13 in usual care. Everyone had started around 4.

The famous 66 percent is a different cut of the same data. It is the proportion who rated their pain 0 or 1 out of 10, which the researchers treated as pain-free or nearly pain-free. That was 33 of the 50 people assigned to PRT. In the placebo group it was 10 of 51, or 20 percent. In usual care, 5 of 50, or 10 percent.

A year later the gap had narrowed but had not closed: 1.51 for PRT, 2.79 for placebo, 3.00 for usual care. In effect-size terms, Hedges g was −1.14 against placebo and −1.74 against usual care at the end of treatment, easing to −0.70 and −1.05 at one year. Those are large effects by the standards of chronic pain research. Forty-four of the 50 people assigned to PRT completed every session, and no adverse events were reported in that group.

What happened after five years

In 2023 the researchers went back to the same participants, five years after randomisation, and published the results in 2025 (PubMed). This is the part of the story most retellings have not caught up with, and it is the part that matters most if you are deciding whether to try this.

Of the original 151, 113 people (75 percent) responded, spread fairly evenly across the three groups: 38 in the PRT arm, 39 in placebo, 36 in usual care. Among them, 21 of the 38 PRT participants, or 55 percent, were still nearly or completely pain free. In the placebo group it was 10 of 39, or 26 percent. In usual care, 13 of 36, or 36 percent. Average pain intensity was 1.93 for PRT, 3.19 for placebo and 2.60 for usual care.

So the headline finding held. Five years on, the PRT group was still measurably better off, and the difference was not small.

Three details complicate the picture, and you should have all of them.

The placebo advantage did not last. Earlier on this page I made a point of the placebo arm doing well at one year, because that is real and usually gets dropped. At five years it is no longer true: there was no statistically significant difference between the placebo group and usual care on any outcome measured. The open-label placebo produced a genuine improvement that faded over years. PRT’s did not.

Usual care caught up somewhat. Thirty-six percent of the usual-care group were nearly or completely pain free at five years, higher than placebo. Given enough time, a meaningful number of people with chronic back pain improve without any particular intervention. Any honest account has to include that.

PRT is not a treatment for mood or sleep. At five years the PRT group was better on pain intensity, pain interference, disability, depression and anger. It was not significantly better on sleep, anxiety, positive mood, pain catastrophising or how much control people felt they had over their pain. If you are hoping this will fix your sleep or your anxiety, the evidence does not support that.

The largest and most durable differences at five years were not in the body at all. They were in how dangerous people judged movement to be, and in whether they understood their pain as a brain process rather than as damage. That is the mechanism the therapy targets, and it is the thing that stayed changed.

Two cautions on the five-year data. A quarter of the original participants did not respond, and people who drop out of follow-up studies are not a random sample. And the groups by then were small — 38, 39 and 36 people. The direction is clear; the precision is not.

What the brain scans showed, and what they did not

Participants were scanned with functional MRI before and after the four weeks, while pressure was applied to the back to evoke pain under the scanner.

Afterwards, the PRT group showed reduced responses to that evoked pain in the anterior midcingulate cortex and anterior prefrontal cortex compared with placebo, and in the anterior insula compared with usual care. Resting connectivity between prefrontal regions, the insula and the sensory cortex increased. These are regions involved in how the brain evaluates threat and salience, not regions that detect tissue damage, which is the point the researchers were making.

Two things need saying alongside that. The imaging effects were modest in size, and some did not survive correction across the whole brain. The authors say so themselves.

The second correction matters more, because it is repeated almost everywhere. The scans were taken before and after the four weeks, not at one year. When you read that the brain changes lasted a year, that is not what was measured. Pain was tracked for a year. Brain activity was not rescanned.

There is one finding buried in the analysis that is arguably more useful than the imaging. What tracked with improvement was a shift in one specific belief: that pain means injury, and that movement is dangerous. Changes in pain catastrophising, and changes in beliefs about emotion, did not explain the results at all. If the mechanism holds, it is about perceived danger rather than distress in general. Neither age nor sex changed how well the treatment worked.

What the study does not establish

This is the part that gets left out, so here it is in full. Most of it comes from the authors’ own limitations section.

• One condition. Primary chronic back pain. Not fibromyalgia, not migraine, not nerve pain, not chronic pain in general. Extending the result to other conditions is a guess, not a finding.
• One site, one expert team. Everything happened in Boulder, and the clinicians delivering the treatment were experts in the model. Whether it works equally well elsewhere, delivered by someone less practised in it, was not tested here.
• A narrow sample. Participants answered an advertisement. They were relatively well educated, with roughly three quarters holding a college degree and not one reporting high school as their highest level. They were relatively active. Almost none reported Hispanic ethnicity. Their pain was long-standing but low to moderate, so severe, disabling back pain is simply not represented.
• A self-reported main outcome. Pain has to be self-reported, and that is fine, there is no blood test for it. But people in the PRT group knew they had received an active treatment, and knowing that moves self-reported scores.
• No head-to-head comparison. PRT was measured against an injection and against changing nothing. It has not been tested against the other established non-drug treatments for back pain, so “better than the alternatives” is not something this trial can tell you.

None of that makes the result less real. A trial can be genuinely strong and still be the beginning of the evidence rather than the end of it. Saying so plainly is more useful to you than another page rounding 66 up to “most people.”

What this means if you have chronic back pain

The honest summary is short. One well-run trial produced large, durable improvements in one clearly defined condition, with a plausible mechanism behind it and no reported harm. That is a good reason to take the approach seriously. It is not a reason to expect the same outcome, and anyone promising you 66 percent is quoting a group average as if it were a prediction about you.

The question that comes first is not whether the therapy works. It is whether your pain is the kind the trial studied. Pain driven by an active structural, inflammatory or infectious process needs that cause found and treated, and no amount of retraining substitutes for that. Sorting one from the other is a medical assessment. It is not something to settle from a study summary, and it is the main reason a physician-led version of this work is different from an app.

If you want the wider picture of the approach rather than the trial behind it, start with Pain Reprocessing Therapy. If your pain has been repeatedly waved off and you want that part addressed directly, read what happens when pain is not believed. If your diagnosis is fibromyalgia rather than back pain, the evidence there is different and is covered in PRT for fibromyalgia.

What working with Dr. Vaid looks like

You see Dr. Vaid herself at every visit. Savera is a solo practice, which is the point of it. She is board certified in Internal Medicine, Infectious Disease and Addiction Medicine, with advanced training in Pain Reprocessing Therapy, and she delivers it herself as part of medical care rather than referring it out. She consults in English, Punjabi and Hindi.

A first appointment covers your pain history, what has already been tried, and what has been ruled out, because deciding whether this approach fits your pain comes before any technique. Savera is out of network by design, so visits stay unhurried. HSA and FSA funds are accepted, and cost is discussed openly on the first call.