Pilot trial finds exploratory brain-connectivity differences after genuine chiropractic care

A single session of genuine instrument-assisted chiropractic care was associated with greater changes than sham care in resting-state functional connectivity across a broad set of cortical brain regions in a small quasi-experimental pilot trial. The differences appeared immediately after the intervention in the delta frequency band and one week later in the theta band, within a distributed system the researchers call the “extended triple-network.” No between-group difference was detected in heart rate variability, a measure associated with parasympathetic nervous system activity. Because the study included only 30 adults, used nonrandomized assignment and was not designed to establish clinical efficacy, its neural findings are exploratory rather than evidence that spinal manipulation improves brain function, mental health or physical health.

The trial, conducted at Life University’s Center for Chiropractic Research in Marietta, Georgia, enrolled 15 participants who received genuine care and 15 who received a sham procedure. Participants were allocated alternately rather than randomly, creating a single-blind design in which the study’s blinding assessment indicated successful blinding. Genuine care involved high-velocity, low-amplitude thrusts delivered with an Activator instrument at sites indicated by the chiropractor. The sham procedure reproduced the assessment, positioning and instrument contact, but the instrument was discharged parallel to, rather than touching, the skin. Resting-state electroencephalography and electrocardiography were recorded before treatment, immediately afterward and one week later. The trial was prospectively registered as ClinicalTrials.gov NCT05249530.

Investigators used 64-channel scalp EEG recordings, projected source activity with a linearly constrained minimum variance beamformer and estimated functional connectivity using the multivariate interaction measure, which captures undirected phase relationships between regional signals. Source projection used the Desikan–Killiany 68-region cortical atlas, while the predefined network analyses focused on 36 bilateral regions of interest drawn from that atlas. Connectivity was examined in five frequency ranges: delta from 1 to 4 hertz, theta from 4 to 8 hertz, alpha from 8 to 13 hertz, beta from 13 to 30 hertz and gamma from 30 to 40 hertz. A network-based statistic identified clusters of connections that differed more between groups than expected under permutation testing. The selected regions represented the medial frontoparietal, dorsolateral frontoparietal and mid-cingulo-insular networks, which together incorporate the default mode, central executive, salience, dorsal attention and ventral attention systems.

Immediately after treatment, the analysis found one significant delta-band cluster containing 28 nodes and 52 edges, with a family-wise error-corrected p-value below 0.001. Every edge in that cluster reflected a greater increase from baseline in the genuine-care group than in the sham group. At the one-week assessment, a separate theta-band cluster contained 35 nodes and 109 edges and also reached a family-wise error-corrected p-value below 0.001. Of those 109 edges, 108 reflected larger increases from baseline after genuine care; one connection between the left and right middle temporal regions showed a marginally smaller increase in the genuine group, with a difference of −0.013134 in the multivariate interaction measure. No significant clusters emerged in the alpha, beta or gamma bands at either post-baseline time point.

The network-based result does not mean that every individual brain connection inside either cluster was independently significant. The family-wise correction applied to each cluster as a whole, while its connected components summarize the spatial pattern of supra-threshold differences. Network-specific interpretations are also constrained by the atlas and region assignments used. The study employed the Desikan–Killiany 68-region cortical atlas, but no published mapping assigns every parcel unambiguously to a network. Assignments therefore involved author judgment, and several parcels span more than one functional system. The insula was not divided into anterior and posterior portions, while the superior frontal, inferior parietal and supramarginal regions included territories associated with different networks. All analyzed regions were cortical; no subcortical structures were included, and source localization is generally less reliable for medial than lateral cortical regions.

The autonomic measurements did not show a corresponding between-group signal. The researchers analyzed root mean square of successive differences, a time-domain heart rate variability measure commonly used as an indicator of cardiac-related parasympathetic activity, with a baseline-adjusted mixed model for repeated measurements. The estimated difference between genuine and sham care in change from baseline was 1.99 milliseconds immediately after treatment, with a 95% confidence interval from −3.89 to 7.87 milliseconds. At one week, the estimated difference was 1.49 milliseconds, with a 95% confidence interval of −4.45 to 7.44 milliseconds. These intervals included zero, and the analyses were exploratory and not powered to test efficacy. The study therefore found an exploratory neural association without a detectable difference in heart rate variability, not a demonstrated pathway linking spinal adjustment to brain-network change.

Several feasibility results supported proceeding to a larger trial. All 30 enrolled participants returned for the one-week assessment, and participants’ success in identifying their assigned group was consistent with effective blinding: the Bang Blinding Index was 0.20 in both arms, with 95% confidence intervals from −0.30 to 0.70. Of 90 EEG acquisitions, 82% remained suitable for analysis after preprocessing, compared with 96% of 90 ECG acquisitions. Enrollment reached 30 of 91 eligible adults, or 33%. The investigators classified retention and ECG data quality as feasible without modification, while EEG quality and enrollment required amendments. Possible EEG refinements include better preparation to reduce movement and muscle artifacts, standardized electrode application and more rigorous real-time quality control during acquisition.

The design also faces limitations that prevent strong causal or clinical conclusions. Alternating assignment was intended to balance the number and locations of contacts between groups, but without randomization or allocation concealment it could not eliminate selection bias or balance unmeasured prognostic factors. Although measured characteristics were similar at baseline, the sample was too small to characterize such factors reliably. The sham design controlled for whether a high-velocity impulse was delivered, but it could not distinguish the effect of treating a clinically indicated vertebral site from the effect of applying any such thrust at an equivalent site. The researchers said a third arm receiving a thrust at a non-indicated location would be needed to separate those possibilities. The study also included generally healthy adults and did not measure clinical symptoms or patient-reported outcomes, so it could not establish whether connectivity changes correspond to meaningful health benefits.

Arousal is another possible explanation that the pilot could not exclude. Delta and theta activity can vary with drowsiness, and the eyes-closed EEG recordings were acquired after participants had lain prone for treatment, conditions that may encourage reduced alertness. Session timing was held approximately constant within each participant but was not standardized across participants, and investigators calculated no formal vigilance indices or absolute band-power measures. Consequently, the pattern could reflect differences in wakefulness rather than a unique effect of the adjustment. Statistical caution is also warranted because investigators examined 10 combinations of five frequency bands and two post-baseline times. Network-based correction controlled family-wise error within each comparison but not across comparisons, creating a risk of false-positive findings. Both major clusters retained significance under stricter connection-forming thresholds, although the apparent breadth of the delta cluster contracted substantially as the threshold tightened.

The findings are therefore best understood as a signal for further testing rather than a new clinical capability. The study provides evidence that a sham-controlled EEG protocol can be implemented with high follow-up retention and a blinding assessment compatible with successful blinding, while also identifying where the procedure needs refinement. Its broader interest lies in the possibility that a mechanical spinal intervention may be associated with measurable changes in large-scale cortical organization even when autonomic measures do not differ. That hypothesis remains speculative, particularly for emotions, cognition and memory, because the pilot did not test those outcomes and its network assignments were partly judgment-based. The authors are planning a fully powered randomized controlled trial to assess genuine versus sham adjustments in adults; confirmation, stable effect-size estimates and clinical relevance would still require larger and more rigorous studies, including standardized vigilance monitoring, clearer network parcellation and endpoints that can connect brain measures to meaningful changes in health.

Subject of Research: Exploratory resting-state brain connectivity and heart rate variability after genuine versus sham chiropractic adjustment in adults.

Article Title: Between-group differences in resting-state functional connectivity within the brain’s extended triple-network following genuine and sham chiropractic adjustments: a quasi-experimental pilot trial

Article References: Sullivan, S., Sliwka, M., Drake, E., & Perez, T. M. (2026). Between-group differences in resting-state functional connectivity within the brain’s extended triple-network following genuine and sham chiropractic adjustments: a quasi-experimental pilot trial. Frontiers in Human Neuroscience, 20, Article 1984309. https://doi.org/10.3389/fnhum.2026.1984309

Image Credits: AI Generated

DOI: 10.3389/fnhum.2026.1984309

Keywords: chiropractic adjustment, resting-state connectivity, electroencephalography, sham-controlled pilot, extended triple-network, heart rate variability, nonrandomized trial, brain networks, Between-group, differences, resting-state, functional