Seventy-Six Trials, Twelve Thousand Patients, and the Cochrane Verdict on Spinal Manipulation

The Study

de Zoete and colleagues conducted a Cochrane systematic review and meta-analysis updating their team's 2011 review of the same question. They searched CENTRAL, MEDLINE, Embase, and two trial registers through October 2024, with no language restriction. They included randomized controlled trials of adults with chronic low back pain (defined as 12+ weeks of symptoms) where spinal manipulation or mobilisation was compared to sham SMT or placebo, no treatment, or other conservative interventions. Trials limited to sciatica were excluded.

The critical outcomes were pain, functional status, and adverse events at one month after intervention. Pain was converted to a 100-point scale across all studies. Risk of bias was assessed with the original Cochrane RoB 1 tool. Certainty of evidence was graded using the GRADE approach.

76 RCTs met inclusion criteria, with 11,866 total participants. Two-thirds of the included trials (50 of 76) were new since the 2011 review, reflecting the volume of SMT research conducted in the past 14 years. Most trials were in middle-aged adults, mixed sex, in high-income countries (53 trials) or middle-income countries (23 trials). The three primary comparisons broke down as 17 trials versus sham/placebo (2,021 participants), 4 trials versus no treatment (435 participants), and 43 trials versus other conservative interventions (8,291 participants).

What They Found

The pattern is consistent with what most clinicians who have practiced manipulation for any length of time would predict. SMT does outperform a fake version of itself by a small but statistically detectable margin. SMT clearly outperforms doing nothing at all. SMT and other conservative interventions, including exercise, medication, and education, perform similarly when both are delivered as active treatments. None of the differences are clinically dramatic. None of them disappear, either.

Heterogeneity across studies was substantial. The I-squared statistic, which measures the share of variability in effect estimates that comes from real differences between trials rather than chance, sat between 71% and 94% across the comparisons. That is a high number. It tells us the trials were not measuring the same thing, even when they reported the same outcome label. Different SMT techniques. Different doses. Different patient subgroups. Different control conditions calling themselves the same name.

On safety, the adverse event picture is reassuring within its limitations. Less than half of trials reported on adverse events at all, and only 12 reported them systematically. Of those that did, the reported adverse events were the familiar profile: muscle soreness, stiffness, transient flare. No serious complications were registered across the entire dataset of 76 trials and approximately 12,000 patients.

What This Means Clinically

In clinical practice, the question I get from patients about SMT for chronic low back pain is rarely "is there evidence." The question is "is it going to help me, and is it safe." This Cochrane review tightens the answers to both questions, but it does not transform them.

For patients who have not yet tried any active treatment, SMT will almost certainly help more than waiting. The medium-effect estimate against no treatment, with a 14-point pain reduction on a 100-point scale, is the largest signal in the review. That is a meaningful clinical difference for someone who has been suffering for 12 weeks or more and has been told to "just see if it goes away."

For patients who have already tried other conservative interventions, the answer is more nuanced. Adding SMT to a regimen that already includes structured exercise, NSAIDs, and education will probably not move the pain needle much. It may help with functional status, the SMD of -0.25 corresponds to a small but real effect. The honest framing for these patients is that SMT is one valid option among several, not the missing piece that will resolve a stubborn case.

For the patient who has never had a manual therapy visit and is hesitant, the safety data should be reassuring. Across nearly 12,000 trial participants, no serious complications were reported. The risk profile is dominated by transient muscle soreness, which most patients absorb without difficulty.

The Heterogeneity Problem Is the Story

The most informative part of this review for a practicing clinician is not the headline effect sizes. It is the I-squared statistics. When 90% of the variability in pain outcomes comes from differences between trials rather than within them, what you are looking at is not a single intervention being studied. You are looking at a category of interventions being lumped together under a single name.

Diversified, Gonstead, Activator, and motion-palpation-driven mobilisation are not the same intervention. Twelve sessions over six weeks is not the same dose as four sessions over two weeks. A patient with discogenic pain is not the same case as a patient with facet-mediated pain or sacroiliac joint involvement. When all of these get pooled into "SMT versus other," the meta-analytic answer becomes structurally fuzzy. The Cochrane authors are clear about this. Their methodological challenge to the field is the right one: stop running underpowered, heterogeneous, poorly-blinded trials that fail to disentangle technique from dose from patient phenotype. Run trials that match a specific technique to a specific patient profile, with adequate sham control and adequate sample size.

What I Tell Patients

I tell my patients three things, and the Cochrane update reinforces all of them. First, the evidence supports trying manipulation if you have not done so, especially if you have been doing nothing or if other approaches have plateaued. Second, the evidence does not support manipulation as categorically superior to other active conservative care. Combination care, where adjustments are paired with movement work and self-management coaching, almost always outperforms any single modality in my clinic. Third, the safety record across thousands of trial participants is clean for the kind of low-velocity work I do, and the rare serious complications described in case reports almost always occur in the cervical spine with high-velocity techniques, not in the lumbar spine.

For deeper context on functional recovery patterns, our recent commentary on Shannon et al. 2026 mediator analysis shows that early functional improvement at six weeks carries most of the long-term benefit. That mediator finding fits well with the Cochrane data here. SMT works through the same pathway whether it is delivered to military or civilian patients, whether the comparator is sham or active care: reduce mechanical pain, restore early function, let everything downstream reorganize.