PACBACK in JAMA: 1,000 Patients, Four Treatment Arms, and a Hard Lesson for the SMT-Only Model

The Trial

Bronfort and the multi-institutional PACBACK team enrolled adults with acute or subacute low back pain (under 12 weeks of symptoms) at moderate to high risk of progressing to chronic disabling LBP, screened with the STarT Back stratification tool. Enrollment ran from November 2018 through May 2023 across three research clinics at the Universities of Minnesota and Pittsburgh. Final follow-up completed in June 2024.

The 2x2 factorial randomization produced four arms:

(1) Spinal manipulation therapy (n = 201)
(2) Clinician-supported biopsychosocial self-management (n = 305)
(3) Combined supported self-management with spinal manipulation (n = 193)
(4) Guideline-based medical care control (n = 301)

Interventions ran up to 8 weeks. The supported self-management arm involved sessions with chiropractors and physical therapists structured around pain neuroscience education, fear-avoidance counseling, graded activity, motivational interviewing, and structured home practice. The SMT arm received conventional manipulation without the explicit biopsychosocial intervention. Combined received both. Medical care followed standard guideline-based care for the index episode.

Primary outcomes were averaged over the full 12-month follow-up period: monthly Roland-Morris Disability Questionnaire and weekly pain NRS. The 50% reduction proportion was a pre-specified secondary outcome. Statistical analysis ran November 2024 through June 2025. JAMA's own statistical editor (Heagerty) is an author, an unusual quality signal that the analytic methods cleared internal scrutiny.

What They Found

Read the disability table carefully. The two SSM-inclusive arms separate clearly from the two non-SSM arms by approximately one full RMDQ point on average. The non-SSM arms (SMT alone and medical care) are statistically indistinguishable. SMT alone delivered the same disability outcome as not getting active treatment.

On the 50% responder analysis, the gap is even cleaner. SSM-inclusive arms produced 65% to 67% of patients reaching 50% or greater disability reduction at 12 months. Both non-SSM arms produced 54%. That is a 13-point absolute difference in clinical responders, in favor of the educational and self-management package.

Pain intensity is a tougher story for everyone. None of the four arms produced a statistically significant 12-month pain advantage over medical care. Group differences clustered at -0.2 or smaller. This is consistent with what is now known about chronic pain trajectories: pain ratings are noisy, function is sticky, and disability reduction is the more reliable signal of meaningful clinical change.

What This Means Clinically

The findings cannot be dismissed on methodology. The team includes Bronfort, the leading chiropractic researcher of the modern era. Schneider runs the Pittsburgh DC program. Chou wrote the ACP guideline. The funding pathway is clean (NCCIH and NIH). The blinding limitations are inherent to manual therapy trials, not specific to this one. Sample size is among the largest ever for this question. Adherence and completion are excellent. JAMA's own statistical editor is an author. This is not a paper a chiropractic profession can argue away.

What it says is that for the population studied (acute or subacute LBP, high risk of chronicity by STarT Back), spinal manipulation as a standalone intervention does not produce a measurable 12-month disability or pain advantage over usual medical care. The active ingredient in this clinical category is the structured educational and self-management package.

For practicing chiropractors, three honest reactions are appropriate.

First, the SMT-alone arm in this trial is a research design, not a real-world clinical practice. Most chiropractors deliver some advice, some education, some movement coaching alongside the adjustment. Real practice sits closer to the combined arm than to the SMT-alone arm. The PACBACK data, read fairly, do not say that the chiropractic visit is ineffective. They say that the chiropractic visit, stripped of its educational component, is ineffective.

Second, this is a wake-up call to anyone whose practice IS structured as adjustment-only. Brief in-and-out encounters that contain no fear-avoidance work, no pain neuroscience education, no graded activity coaching, are practicing in the SMT-alone column of this trial. The data are telling them that what they are doing is not adding measurable value over usual medical care for this patient population.

Third, the addition of SMT to SSM did not add benefit beyond SSM alone. Read that carefully. The combined arm and the SSM arm produced statistically indistinguishable outcomes. This does not mean SMT is harmful, and it does not mean SMT should be removed from care. It means that for these patients, in this design, the SMT was not the active ingredient. The education was.

The Three-Paper Picture

This is the third paper in a tightly linked evidence cluster covered in this issue. Read together, the three tell a coherent story.

The Forward Review's commentary on the 2026 Cochrane review of SMT for chronic low back pain showed small to medium effects of SMT vs sham, and similar effects to other conservative care, with low to very low certainty. The Cochrane authors closed with a methodological challenge to the field.

The Forward Review's commentary on the Elliott Trial Sequential Analysis quantified that challenge: the SMT-versus-conservative comparison reached statistical sufficiency in 2002 to 2004. The 22 years of trials since have added cost without adding knowledge.

The PACBACK trial, in JAMA, completes the picture. When the field finally ran a high-quality trial with a NEW question (does adding biopsychosocial self-management to SMT help, and how does it compare to medical care?), the answer landed cleanly. The ingredient that moves the needle is the educational and self-management package. The manipulation, as a standalone modality, does not.

None of these three papers are anti-chiropractic. Bronfort, de Zoete, Rubinstein, Schneider. The chiropractic profession's most respected researchers are co-authors across all three. The message they are converging on is internal critique, not external attack: the field needs to move past the SMT-as-monotherapy framework, and patients will be better served once it does.

What This Looks Like in a $40 Walk-In Practice

I run a high-volume, low-overhead walk-in clinic. Brief encounters. Most patients in and out in 15 minutes. The PACBACK biopsychosocial self-management protocol involved 6 to 8 structured sessions of ~30 to 60 minutes each, including motivational interviewing, pain neuroscience education, fear-avoidance work, graded activity, and home practice. That is not deliverable in my visit format.

What IS deliverable in my visit format: a 60-second pain neuroscience explanation when appropriate. A 30-second fear-avoidance reframe when a patient says "I'm afraid to bend." A specific home self-care recommendation tied to the visit. A patient handout. A return-instruction. A simple movement they can practice. None of these are a 60-minute coached session, but cumulatively, across multiple visits, they build the same skeleton.

Practitioners who want to keep the SMT-only model alive need to see the PACBACK data clearly. The trial's biopsychosocial intervention is the ceiling of what is deliverable in 6 to 8 sessions. A 15-minute walk-in cannot match that ceiling. What it can do is build the floor: simple, repeated, brief biopsychosocial elements layered into the existing visit pattern. The PACBACK numbers are a target to aim for, not a reason to despair.