A literature search was performed in January 2018 using the PubMed database and reference lists from relevant articles. Search parameters included English-language articles published between 1998 and 2018 to ensure that articles were not older than 20 years. Exceptions to this age limitation included seminal articles in the field of physiologic research related to spinal manipulation. The main search terms used were “spinal manipulation” and “high-velocity low-amplitude,” then the “and” Boolean operator followed by the primary search terms of “physiologic response,” “vertebra,” “vertebral movement,” “neurological effect,” and “mechanism.” The resulting studies were screened for relevance to HVLA and the potential mechanism thereof. Articles that were related to other manipulative therapies were excluded. Articles that presented a physiologic explanation without basis in physiologic measurements were also excluded. The reference sections of systematic reviews were also searched and relevant articles were obtained and studied. We used our best judgement to identify high-quality studies that discussed physiologic responses to HVLA. For the purposes of the present review, all treatments, regardless of the operators’ training or terminology used in the identified articles, are referred to as HVLA. 

Results returned nearly 160 articles, of which 148 were found to be relevant to the technique of HVLA. Of these, 69 were used for this review, with the rest excluded based on the criteria listed in the Methods section. 

The oft-cited mechanism of HVLA focuses on an articulatory model, explaining the therapeutic effects as restoring mobility to a joint or correcting a joint's malposition.^1,2 The concept of vertebral misalignment arose, in part, from the distinct spinal mechanics put forth by Harrison Fryette, DO, and those mechanics helped create the basis for both diagnosis and treatment goals when using osteopathic manipulative treatment (OMT), including HVLA.² Despite a longstanding tradition in the teachings of osteopathic medicine, the current literature makes the continued use of such models questionable. Multiple studies^3-6 have failed to confirm Fryette's mechanics (except perhaps for the cervical spine during motion) or to demonstrate that a vertebra can prefer particular cardinal motions such as preferring right sidebending over left sidebending. If the presumed target of treatment is not present, one must reconsider the therapeutic mechanism of that treatment. 

In the case of HVLA, evidence suggests that correcting the asymmetrical movement preferences of vertebrae (or even affecting the vertebrae at all) is unlikely to be a source of therapeutic benefit. For example, multiple studies reveal minor or no movement of osseous structures after thrust is applied; when such movement does occur, it is well within the range of physiologic motion.^7-11 Thus, the current concept that HVLA re-adjusts one or several vertebrae seems improbable. Beyond this concept, studies^12,13 have illustrated that single-segment localization and directionally specific vertebral thrusts do not appear possible. This inability to localize a single vertebral segment during HVLA has been confirmed when the location of HVLA was compared with the area of the spine that underwent cavitation. Cramer et al¹⁴ found that lumbar HVLA was accurate about 50% of the time and that multiple vertebral segments were almost always involved, despite the goal of single-segment localization. 

In part owing to the aforementioned studies, researchers are moving away from an articulatory model and toward a neuromuscular one that focuses less on vertebral movement and localization and more on the response of the surrounding sensory fibers. Research^15-19 indicates that it is the characteristics of the applied force (ie, the velocity and amplitude of the thrust) that are far more germane to physiologic change than the practitioner's ability to localize the target vertebra or apply the thrust in any one direction. However, it should be noted that these studies demonstrate that the characteristic force, as well as the area of the spine to which the force is directed, be relatively precise. Because HVLA substantially affects local musculature and neural segments, it is important that the technique be applied only to areas requiring therapeutic innervation. To our knowledge, the consequences of applying HVLA to a healthy region have not been studied and, in theory, could potentiate neuromuscular dysfunction. 

Specifically, a lot of attention has been paid to the response of type Ia afferents (sensory fibers arising from muscle spindles) in the paraspinal musculature to HVLA thrusts. It has been well documented, even outside the context of manual therapy, that muscle spindles are far more responsive to smaller amplitude stretch than larger, more physiologic stretch, such as that seen with movement.²⁰ Muscle spindles, then, are an ideal target of a technique such as HVLA. In the setting of simulated HVLA using an adjustable machine on anesthetized cats, fast, short thrusts elicited a greater response from group Ia afferents than larger or slower impulses, and this response was much higher than that found with normal physiologic motion.^15-19 These experiments also elicited responses from Golgi tendons and possibly from Pacinian corpuscles; however, the response of these fibers paled compared with the magnitude of the response from the type Ia afferents.²¹ 

The effects of HVLA on muscle spindles has been further elucidated in human studies that were unable to directly measure single spindles but instead measured muscle spindle activity. One consequence of discharging type Ia afferents is the effect it would have on the α motor neuron pool. Measuring this effect is indirect in nature and requires some creativity. One method is via electromyography (EMG), which can look at the response of muscles after their type Ia afferents have been electrically stimulated, known as the Hoffmann reflex (H-reflex). The H-reflex more directly represents the activity of the α motor neuron pool, as directly stimulating the type Ia afferents bypasses the muscle spindles. Stated differently, the afferent portion is induced with an electrical stimulation that bypasses muscle spindles and directly results in action potentials travelling along the afferent fibers to the α motor neuron pool. The efferent portion results from action potentials from the α motor neuron pool travelling along efferent fibers until reaching the neuromuscular junction and producing a twitch response.^22,23 In theory, by measuring the H-reflex before and after HVLA, one can measure the effect HVLA has on the α motor neuron pool.²³ Through this technique, HVLA was shown to attenuate the α motor neuron pool.²³ However, further research into the H-reflex method has since suggested that it may be a less-than-ideal metric because presynaptic factors could skew results via heteronymous Ia afferents and provide inaccurate data.²³ 

This study²⁴ concluded that a possibly superior method of more directly measuring the α motor neuron pool was through transcranial magnetic stimulation. It was argued that such stimulation would more directly study the corticospinal inputs on α motor neuron pool excitability. With this approach, HVLA was shown to actually facilitate the α motor neuron pool.²⁴ The simultaneous attenuation of the H-reflex and facilitation of the α motor neuron pool is not without precedent. The “vibration paradox” accomplishes the same feat when muscles are subjected to a vibratory force, and likely does so via the same neuromuscular processes.^25,26 Nevertheless, the effect of HVLA on the α motor neuron pool remains moot, and more research is needed. 

Whatever effect HVLA has on the α motor neuron pool, it has not been demonstrated by lower velocity techniques such as soft-tissue massage.²⁷ The higher velocity applied in HVLA may provide insight into its therapeutic mechanism, as well as emphasize the proper application of the technique. For example, evidence suggests that stabilizing the neuronal gain of the motor neuron pool via proprioceptive feedback would help alleviate aberrant and uncoordinated muscle function.²⁴ Furthermore, it has been theorized that such uncoordinated muscle function may lead to buckling of a vertebral unit, even without trauma.²⁸ This buckling is likely due to a failure in the vertebral unit's ability to oppose such buckling behavior through the proper timing in the recruitment of its attached supportive muscles.²⁸ If, as suggested by multiple studies,^24,27 HVLA could potentially reset this timing by stabilizing the α motor neuron pool, it would explain HVLA's long-term benefits, even in the setting of relatively short-term local effects. Beyond the hypothetical, studies have shown that dysfunctional muscle behavior is corrected after HVLA. For example, Tunnell²⁹ found a decrease in multifidus tonus in patients with low back pain, whereas Koppenhauer et al³⁰ found an increase in multifidus contraction force after HVLA.³⁰ Similarly, de Almeida et al³¹ noted that HVLA was able to increase pelvic floor tonus in women. Herzog et al³² found a decrease in these EMG responses after HVLA. Although the effect of HVLA on the local musculature has not been consistent, it again appears related to a normalization of previously dysfunctional behavior. These data suggest that rather than a single effect on muscle tonus, HVLA may stabilize the α motor neuron pool to correct for aberrant postural and proprioceptive behavior. 

High velocity, low amplitude is known to produce local and segmental effects, such as increased range of motion,^33,34 improved muscle strength,^34-37 decreased pain both locally and segmentally,^34,39-41 and changes in cutaneous blood flow,^34,42 temperature,^34,43 and skin conductance.^34,44 However, even more so than the shift from an articulatory process to a neuromuscular one, research has pushed the notion that HVLA works primarily via a centralized process rather than just through the aforementioned local effects. One indication of this centralized effect arises from studies that found HVLA not only increased the pressure pain threshold (PPT) in areas related to the segment treated (eg, the upper extremity after cervical or upper thoracic HVLA) but also in more distant areas (eg, increased PPT in the toes after upper thoracic HVLA).³⁸ Pressure pain threshold represents the minimum force applied that induces pain; therefore, a global increase in PPT suggests that there is more to the mechanism of HVLA than just a localized, segment-specific neuromuscular milieu. 

One step toward defining HVLA's centralized impact started with studies that mapped out the areas of the cerebrum affected by HVLA. Using somatosensory evoked potential recordings, a measure of brain electrical activity in response to touch, researchers demonstrated that HVLA preferentially affected the N20 and N30 cortical areas.^45-49 The control during these measurements ruled out passive stretch, touch, and vestibular changes through head movement as the source for the N20 and N30 activation. Specifically, the experiments found that HVLA decreased the amplitudes of N20 and N30 areas for an average of 20 minutes.^45-49 

The implications of HLVA altering the N20 and N30 areas are multifaceted. Research indicates that activity in the primary somatosensory cortex increases with chronic pain and that reduction of this activity correspondingly relieves the pain.⁵⁰ N20 represents the arrival of afferent information to the primary somatosensory cortex. Therefore, it can be hypothesized that the reduction of the amplitude of N20 may help in this goal of reducing primary somatosensory cortex activity.⁴⁹ The N20 and N30 areas are also responsible for somatosensory integration—specifically, dual somatosensory integration.⁴⁶ It is this role that may be more salient to HVLA's capacity to treat a patient's dysfunction and pain. 

The brain is constantly bombarded by somatosensory information and requires a gating or filter mechanism to make sense of it all. A study⁴⁹ of muscle-centric pathologies such as dystonia illustrates that muscular disease may be a consequence of the brain no longer being able to properly integrate somatosensory information. A failure of the brain's somatosensory integration mechanism has also been implicated in chronic pain conditions, such as chronic low back pain.^51,52,53 Studies illustrate that the brain translates poor sensorimotor integration as pain, as was demonstrated when a healthy population was given unexpected visual feedback to limb movement. This experiment not only elicited pain in otherwise healthy individuals but also exacerbated preexisting pain symptoms in patients with fibromyalgia. Furthermore, Brumagne et al⁵⁴ revealed that introducing vibration into the multifidus muscle prevented healthy individuals from accurately repositioning their spine, and a follow-up study⁵⁵ found that it was a common attribute among those with low back pain.^54,55 

Whether the brain is misinterpreting unfiltered somatosensory information such as pain, or whether poor somatosensory integration leads to aberrant body mechanics that induce pain, patients with chronic pain also have poor dual somatosensory integration.⁴⁹ Of note, research has shown that HVLA improves somatosensory integration.^45-49 The end result is decreased pain⁴⁹ and improved joint function and proprioception^56,57 after HVLA. 

This aberrant somatosensory integration is theorized to revolve around the same cortical mechanisms that allow for adaptive neuroplasticity. Just as one can learn to ride a bike after an injury via this mechanism the body can “learn” dysfunctional pain patterns via the same processes.^48,58 Correcting this maladaptive neuroplasticity may be the key to how HVLA can help patients with chronic pain “unlearn” dysfunctional pain patterns, as has been clinically demonstrated.^59,60 

Normalizing somatosensory integration provides a feasible mechanism of action for the therapeutic benefits of HVLA. While these alterations are temporary in nature, their overall results are more long term. Normalizing somatosensory integration leads to more accurate motor control, which in turn improves function. This effect has been demonstrated in several studies wherein HVLA equalized standing postural weight distribution,⁶¹ improved symmetry during gait,⁶² and corrected disordered feed-forward activation of the abdominal muscles.⁶³ Eliciting appropriate joint movement by normalizing somatosensory integration provides a possible explanation for HVLA's ability to correct dysfunction over the long term.^48,59,60 

It has also been shown that HVLA has a more immediate antihyperalgesic effect.^34,38-41 Research suggests the presence of a descending mechanism of global pain relief involving neuropeptides. For example, in one study,⁶⁴ various neuropeptide blockers were used intrathecally in rodents to determine which neuropeptide might be influenced by HVLA. Blocking both opioid and γ-aminobutyric acid receptors after HVLA did not mitigate the decrease in PPT; however, when serotonin or noradrenaline receptors were blocked, the decrease in hypersensitivity typically seen after HVLA was nearly completely eliminated.⁶⁴ Thus, this study⁶⁴ suggests that HVLA triggers a release of serotonin and noradrenaline that helps mitigate hypersensitivity on a systemic level. 

Whether HVLA is able to systemically decrease pain sensitivity has been debated; however, reconciliation of these discrepancies may lie in the research on HVLA's effect on the periaqueductal gray (PAG) region of the brain. The PAG region can be separated into dorsal (dPAG) and ventral (vPAG) sections. The dPAG appears related to mechanosensitivity and can inhibit mechanical pain via noradrenergics and serotonin. In contrast, the vPAG is responsible for inhibiting thermal pain.^65,66 The discrepancy in the literature mostly surrounds the findings that HVLA decreases sensitivity to pressure throughout the body, as other studies have found that it does not tend to decrease sensitivity to temperature.^41,67 However, studies have shown that HVLA affects the dPAG but does not likely affect the vPAG.⁶⁸ With these findings is mind, a discrepancy in the literature may no longer exist: HVLA does produce a systemic antinociceptive state, but it is primarily mechanical rather than thermal. 

To state that HVLA works via a single mechanism would not accurately reflect the complexity of the musculoskeletal system as a whole. With so many neuromuscular elements occurring simultaneously, all of which must be interpreted and integrated by the central nervous system, it is unlikely that HVLA only works via a single mechanism. However, the existing literature reveals a general pathway is beginning to form. A very specific type of sensory impulse (HVLA) delivered to the relevant paraspinal region will have multiple effects on the type Ia afferents in the paraspinal musculature. This impulse creates a cascade that locally balances the α motor neuron pool of the nearby segments affecting the corresponding musculature. This cascade then ascends to the brain to modulate areas responsible for somatosensory integration, thereby shifting the entire proprioceptive and mechanical forces of the body. Simultaneously, the dPAG is stimulated to produce a descending cascade of noradrenergics and serotonin, resulting in a systemic decrease in sensitivity to mechanical pain. Therefore, the long-term effects of HVLA are likely a combination of improved overall somatosensory integration, a break in the pain-spasm-pain cycle resulting from a temporary state of global hyperalgesia, resolved dysfunctional muscle imbalances present at vertebral segments, and improved global functioning secondary to the correction of aberrant proprioception. The end result of this intricate pathway seems to be a lasting decrease in pain and dysfunction of the musculoskeletal system. 

Although HVLA is still widely taught as an articular technique, research suggests that HVLA's potent and more long-term effects seem less joint-centric and more neuromuscular. This paradigm shift from a joint-centric to a neuromuscular mechanism may reflect the direction osteopathic manipulative medicine is heading as a whole. The original views of Andrew Taylor Still, MD, DO, that his manipulative techniques were properly positioning osseous structures so that the body's fluids⁶⁹ could flow more freely, may need to be updated in favor of a mechanism that instead looks at central nervous system somatosensory integration via neuromuscular afferents, such as the type Ia afferents discussed in this article. 

Despite HVLA having more mechanism-oriented research than perhaps any other OMT technique, more research is needed. We believe that the current literature provides a suitable jump-off point for more specific research into the mechanisms discussed in this article. Using the framework provided by the currently published research may help guide physiologic studies in a more unified direction in hope that the science behind HVLA (and other OMT techniques) may be bolstered and appreciated at the same level as the art behind these techniques. 

We thank Charles McClung, DO, David Beatty, DO, and Jessica Smith-Kelly, DO, for their suggestions and proofreading of this article. We also thank the West Virginia School of Osteopathic Medicine for resources provided.