J Manipulative Physiol Ther 2000 (Feb); 23 (2): 134–138 ~ FULL TEXT
Howard Vernon, DC
Canadian Memorial Chiropractic College,
Toronto, Ontario. email@example.com
BACKGROUND: The number of studies that have investigated the direct analgesic effect of a spinal manipulation on spinal or referred pain is small, making knowledge of this crucial aspect of manipulation sparse. This paper reviews a set of studies that measure the immediate effect of manipulation on pain or pain-related phenomena in the spinal and peripheral soft tissues.
METHODS: The literature was accessed through MEDLINE. Key words used were "manipulation," "pain," and "chiropractic." This search was complemented by citation reviews of important research and chapters on the topic. Only studies that directly measured the effect of at least a single spinal manipulation on pain (eg, tenderness, biochemical assay, referred pain) were selected. The selected studies were reviewed descriptively; no systematic assessment of their quality was conducted.
RESULTS: The electronic search yielded 738 citations. Six hundred and forty-two were relevant to chiropractic. Of these, most were clinically descriptive articles about diagnostic and therapeutic procedures or case management. Most of the remaining articles were clinical trial reports or letters to the editor. Only 5 studies were selected according to the established criteria. Thus less than 1% of the indexed literature on chiropractic, manipulation, and pain involved studies that explored the mechanism of the putative effect of spinal manipulation on pain mechanisms. Six other studies were retrieved from citation reviews. These 11 studies were reviewed in order publication.
CONCLUSION: Few studies have investigated the effects of spinal manipulation on pain directly. If the theory of manipulation exerting its therapeutic effects posits that the sensory input created by the intervention results in some form of inhibition of pain, then the results of these studies are largely consistent with one another and with this theory. This review has highlighted the deficiencies in the extant studies and many remaining questions. Only more high-quality research will permit a full elucidation of the hypoalgesic effects of spinal manipulation.
From the Full-Text Article:
The electronic search yielded 738 citations. Six hundred and forty-two were relevant to chiropractic, and the remainder concerned orthopedic manipulation of the peripheral joints or other unrelated topics. Of these 642, most were clinically descriptive articles describing diagnostic and therapeutic procedures or case management. Most of the remainder were clinical trial reports or letters to the editor. Only 5 studies [10, 11, 12, 13, 14] were selected according to the established criteria. Less than 1% of the indexed literature on chiropractic/manipulation and pain involved studies that explored the mechanism of the putative effect of spinal manipulation on pain mechanisms. Six other studies were retrieved from citation reviews. These 11 studies were reviewed in order of publication.
The first report of the immediate pain-relieving effects of a spinal manipulation was a randomized controlled clinical trial by Glover et al.  In this trial, subjects with low-back pain were randomly assigned to receive a single rotary lumbar manipulation or a control treatment consisting of detuned short wave therapy. Glover et al  measured pin-prick pain on the skin of the paraspinal region adjacent to the painful spinal segment of the subjects. The size of this painful cutaneous receptive field was used as the outcome parameter. Patients receiving a manipulation had a statistically significantly greater reduction in the size of the cutaneous receptive field 15 minutes after the intervention compared with the control subjects.
Before the intervention, mechanical spinal pain had produced a disturbance of sensory processing, sensitizing dorsal horn neurons (DHNs). This sensitization resulted in a lowered threshold of excitation that could be monitored as the receptive zone in which relatively innocuous stimuli applied to the skin were perceived as painful. Furthermore, a manipulation appeared to exert some sort of ameliorative effect, reducing the sensitivity in this zone.
Denslow et al  demonstrated lowered thresholds of para-spinal muscular activity to painful stimuli (pressure algometry) applied to the interspinous tissues. This led Korr  to propose the “central facilitated state,” linking spinal pathomechanics with (at least) segmental neurologic dysfunction.
In current neurophysiologic parlance, the painful mechanical dysfunction is associated with and perhaps induces a state of “central sensitization,” resulting in an expanded receptive field to cutaneous stimuli. Manipulation is thought to produce an inhibitory effect that reduces the sensitization of the segmental DHNs. This effect is thought to be mediated by the mechanoreceptors of the spinal joint, [18, 19] which, although activated by the stretch of the manipulative procedure, produce afferent discharges that exert an inhibitory effect on the DHNs. The behavioral correlate of this phenomenon is a reduction in the cutaneous receptive field. 
Terrett and Vernon  cited this study as the basis for their 1984 report on the effect of a spinal manipulation on the cutaneous pain threshold of normal subjects. In their study, 53 normal male subjects were examined to find at least one spinal segment from T3 to T10 that was painful to manual pressure on the interspinous tissues. Fifty of these subjects demonstrated at least one such segment. This is a similar finding to that of Denslow et al,  who found thoracic segments with lowered pressure pain thresholds in their “normal” subjects.
An electrotherapy device was used to measure the cutaneous pain threshold adjacent to the test segment. Two small pads placed on either side of the spine at the test level delivered an electrical current. This current was perceived as prickling at subthreshold levels and painful at threshold levels. Pain threshold measures were obtained 5 minutes before and 30 seconds, 2 minutes, 5 minutes, and 10 minutes after the interventions. Half the subjects were randomly allocated to receive a manipulation of the segment, whereas the other half received only an oscillatory palpatory maneuver without thrust.
Terrett and Vernon20 reported an immediate and statistically significant rise in cutaneous pain threshold in the subjects receiving manipulation versus subjects in the control group. This increase remained through the entire period after treatment, reaching 130% of baseline levels by 10 minutes after intervention. In the control subjects, a minor, brief increase was followed by a quick return to baseline levels. Once again, the manipulation appears to have produced an inhibitory effect on the DHNs originally responding to the cutaneous stimulation produced by the test instrumentation. Because the control group received the same level of manual contact and joint movement as the manipulated subjects, the critical therapeutic ingredient was the cavitation produced by the thrusting of the manipulation.
These first 2 studies established a basic model of before through after intervention investigations of subjective pain reports measured objectively by pain thresholds. Several additional studies in the 1990s followed the same model. A second distinct approach to studying the pain-relieving or analgesic effects of manipulation involves the measurement of biochemical substances thought to participate in the physiology of antinociception. Prompted by studies of endorphin release produced by acupuncture and transcutaneous electrical nerve stimulation, Vernon et al  sought to measure plasma ß-endorphin levels in normal subjects after a manipulation.
Twenty-one normal male subjects were randomly allocated to receive a cervical spinal manipulation, an oscillatory mobilization (as a control for manual contact, pressure, and rotation), or no treatment (resting control). Plasma ß-endorphin levels were assayed from blood draws obtained 15 minutes and 5 minutes before intervention and 30 seconds, 5 minutes, and 20 minutes after intervention. All experiments were conducted between 9:00 and 12:00 because endorphin levels are known to decrease gradually during this time of the day. Heart rate, blood pressure, and self-rated current anxiety state were measured at the same time intervals. Only the group receiving a manipulation showed a small but statistically significant increase in ß-endorphin levels 5 minutes after treatment. In all 3 groups, heart rate, blood pressure, and anxiety levels all decreased gradually, indicating that the significant increase in endorphins was not stress-induced.
Although this finding appears to indicate that, at least in normal subjects, a manipulation might activate the endogenous antinociceptive system subserved by plasma ß-endorphins, this must be interpreted cautiously. This is because plasma ß-endorphin is derived solely from pituitary gland secretion and acts as a neurohormone in the blood at many various systemic receptor sites. Thus a mechanism of central hypothalamic activation is necessary to explain this finding. This mechanism is distinct from those involving spinal opioids, such as enkephalin or dynorphin, which may be involved in the more segmentally organized analgesia typically seen in clinical practice.
These results point to one of the important and currently unresolved controversies about manipulation-induced analgesia: Is its mode of action confined to spinal levels or is there a more complex, supraspinal mechanism (Vincenzino et al)  The limited data support either or both these mechanisms.
Two subsequent studies [10, 23] attempted to replicate the findings of the Vernon et al21 study about plasma endorphin levels after manipulation. Unfortunately, in the former study, assay levels were not sensitive enough to detect even baseline endorphin levels; therefore the absence of response reported by Christian et al10 should not be construed as a negative finding. In the case of Sanders et al,  the manipulative procedure was performed in the lumbar spine. Because a spino-hypothalamic pathway is necessary to produce pituitary-derived plasma endorphin, their study was inherently flawed.
In 1984, Payson and Holloway  used a novel model to investigate the neurochemical basis of manipulation-induced hypoalgesia. Twelve male subjects with low-back pain were included in a double-blind crossover trial of manipulation with and without an injection of naloxone hydrochloride before the treatment. Naloxone hydrochloride is known to antagonize opiate receptor sites and therefore block the action of endorphins or enkephalins. It was hypothesized that in the naloxone hydrochloride blockade group there would be significantly less pain relief than in the trial involving injection of control vehicle before manipulation. Strategies to measure pain included a VAS score every 15 minutes after treatment for 2 hours and the McGill Pain Questionnaire at 15 and 120 minutes after treatment. In addition, forward trunk flexion was measured at 15 and 120 minutes. From 75 minutes after treatment to 120 minutes after treatment, there was a statistically significant reduction in pain in subjects receiving the naloxone hydrochloride before treatment. No significant changes were found in the McGill Pain Questionnaire or forward-flexion values. This result is the opposite of what would be expected and was explained by the fact that naloxone hydrochloride acts on peripheral receptors to block inflammation and to reduce ischemia. The research contended that the source of the low-back pain likely involved muscles that were in spasm and therefore in a state of ischemia. The naloxone hydrochloride may have acted to counter this effect and actually induced pain relief versus blocking it by its action on spinal opioid receptors.
To my knowledge, no other study has investigated spinal-related opioids or other inhibitory neurotransmitters in response to manipulation. This represents a huge gap in the understanding of the basic neurophysiologic mechanisms of manipulation-induced analgesia.
From the early 1990s, there are several reports involving the use of the pressure algometer initially devised by Fischer.  Vernon  was the first to report the improvement in paraspinal pressure pain threshold (PPT) levels after manipulation. Six tender muscle spots were measured bilaterally in a subject with chronic right-sided neck and scapular pain. The right side muscle values were all significantly lower than those on the left and were lower than the normal cut-off value of 3.5 kg/cm2 established by Fischer.  After a cervicoscapular manipulation, PPT levels rose by an average of 45%, whereas the patient's pain score dropped from 6 to 1/10 on a 10-centimeter VAS.
In 1992, Vernon et al  reported on 9 subjects with chronic neck pain. Baseline PPT values were obtained bilaterally around the painful segment (fixation) for a total of 4 measured sites. Five subjects were randomly assigned to receive a rotary manipulation, and 4 subjects received the same sort of oscillatory mobilization that had been used in the endorphin study.  In the group receiving manipulation, PPT levels at 5 minutes after treatment rose at all 4 sites (ie, bilaterally) an average of 45%, whereas in the control group there was no increase. This difference was statistically significant at all 4 points.
Once again, these results point to an inhibitory mechanism evoked by the manipulation compared with the non-thrusting, manual procedure. These studies [12, 25] revealed for the first time a reduction in tenderness in the deep somatic tissues, which are of greatest interest to chiropractors. This phenomenon likely represents another form of reduced receptive field activation of DHNs or reduced central sensitization.
In 1996, Cote et al  measured PPT values at 3 standard lumbopelvic muscular or ligamentous points in 30 subjects with mechanical low-back pain. Subjects were randomly allocated to receive either a thrusting rotary sacroiliac joint manipulation or a sham (nonthrusting) procedure. No significant differences in the changes in PPT levels were reported between the groups, representing the first “negative” report of manipulation-induced analgesia. However, these authors used a technique that may have minimized the potential changes induced by manipulation. They measured 5 standardized points, regardless of whether these points were actually tender in the subjects. Two different strategies may have improved the results of the study. Initially, manual palpation should have been conducted to identify the actual tender spot within the vicinity of the standardized point, increasing the range of potential change after treatment. In each subject, the standardized point with the lowest PPT value could have been distinguished as that subject's test value. Instead, the values of all subjects' measurements at each point were averaged, including points that may not have been tender. A ceiling effect may have been created by the inclusion of higher PPT values in this study's calculations, leading to negligible differences between the 2 treatment groups. This problem should be addressed in future studies that use deep-tissue tenderness as a pain analogue.
In the recent past, Vincenzino et al [14, 22] have used the model of lateral epicondylitis as a clinical target for the effects of cervical manipulation. They reported on a variety of pain-related parameters, including the upper limb tension test (similar to the straight leg raising test, a test of painful neural tension in the upper limb); pressure pain threshold to pressure algometry over the lateral extensor tendon insertion; pain on grip strength testing (ie, the point at which grip becomes painful); and several VAS scores for current and 24-hour pain and 24-hour function of the upper limb. Fifteen subjects received 3 different experimental procedures in random order:
(1) lateral glide manipulation of C5–C6,
(2) sham mobilization, and
(3) no treatment. The latter 2 procedures, as with the Vernon et al  study, set 2 controls: resting quietly with no treatment and manual contact with the perception of treatment. Most subjects in groups 1 and 2 were unable to distinguish the real from the sham treatment.
Measurements were taken before and after intervention. The measurements showed statistically significant increases in the manipulation group compared with the group receiving the sham treatment and the control group as follows (mean percent improvement): upper limb tension test (43%), pressure pain threshold (26%), and pain-free grip (29%). The pain VAS improved an average of 1.9 cm.
Vincenzino et al  explained the effects of manipulation as “a non-noxious sensory input at the cervical spine which resulted in a reduction of elbow pain that outlasted the duration of its application.” They compared this stimulus-induced analgesia with that of acupuncture, transcutaneous electrical nerve stimulation, and ice massage. They theorized that one site of action for this phenomenon might be the dorsal periaqueductal grey nucleus that, when stimulated, would evoke a descending pain inhibitory system. This mechanism is apparently not naloxone-reversible and is therefore described as a nonopioid mechanism. In 1998, Vincenzino et al  added another parameter to this model: sympathetic activity at the elbow and hand measured as skin temperature, skin conductance, and blood flow in both regions. After the intervention, only the manipulation group showed both analgesia and autonomic changes suggestive of sympatho-excitation (eg, decreased temperature, blood flow, and skin conductance). This sympatho-excitation was credited to the dorsal periaqueductal grey nucleus mechanism because they cited evidence from animal studies that this stimulation results in analgesia and increased sympathetic outflow. This increase in sympathetic activity is in contrast to previous findings [26, 27, 28] that demonstrate decreased sympathetic activity in the upper limbs after manipulation. This latter finding is more congruent with the Korr's theory that sympathicotonia is induced by spinal dysfunction and ameliorated by manipulation. [17, 18]
There are few studies directly investigating the effects of spinal manipulation on pain. If the theory of manipulation exerting its therapeutic effects posits that the sensory input created by the intervention results in some form of inhibition of the pain, then the results of these few studies are largely consistent with one another and with this theory. These results are almost all a product of psychophysical experiments that involve conscious human beings relating their pain to investigators using objective means of measuring that experience or sensation. The studies of endorphins are the exception. The actual mechanism subserving this inhibitory effect is currently unknown. The questions that must be answered include:
Is the central generator of these effects at the level of the spinal cord or in more rostral nuclei, such as in the brainstem?
If the spinal cord is the “site of action,” are the inhibitory effects confined to more or less a segmental distribution?
If the inhibitory effects are controlled by higher centers, is there a more systemic distribution of their effects?
In either case, what are the neurochemical mechanisms involved?
Are spinal inhibitory neurotransmitters, such as ?-aminobutyric acid or enkephalin, involved or are more centrally acting neuro-transmitters, such as serotonin, noradrenaline, or endorphin involved? and
Does the inhibitory effect depend on the cavitation phenomenon or can it be elicited by noncavitating procedures?
The studies by Vernon et al [12, 20, 21] and by Vincenzino et al [14, 22] show a clear difference between thrusting manipulations and other manual procedures (such as oscillatory mobilization) in the magnitude of hypoalgesia produced after manipulation. No study in this category has examined these effects for longer than 15 minutes. No study has examined the cumulative effects of several manipulations to determine if there is a dose-dependent response.
Clinical experience may shed some light on these questions. Gillette  argued that the forces involved in a thrusting manipulation were large enough to be capable of stimulating low- and high-threshold receptors (ie, nonnoxious and noxious transmission). Why is it that the hallmark of a successful manipulation is the lack of evoked pain? All the subjects in the studies cited related that the manipulative procedure provided was not painful. How can a physical procedure, which is theoretically capable of evoking a pain response, not be painful? Is instantaneous inhibition a part of the procedure?
Furthermore, what can be learned from common clinical experience and the clinical trial literature? After receiving a manipulation, patients neither indicate that they feel no pain in their bodies, nor do they indicate that they feel mildly euphoric in a way that would signal that some central analgesic state had been induced (such as with low-frequency acupuncture). In addition, patients do not request spinal manipulation as a form of central anesthesia before surgery although acupuncture or meditation may produce such a state. In fact, more has been written recently about patients receiving anesthetic before some manipulative procedures (manipulation under anaesthesia) to treat chronic pain conditions. All this suggests that the effect of manipulation is probably not targeted to a center located too high in the nervous system; it is more likely that this effect is targeted segmentally. Similar experiences suggest that if a patient has pain at a particular site in the spine and a successful manipulation is rendered there, then pain relief occurs at that point. Perhaps this argument will focus future investigations at some putative spinal inhibitory mechanism such as ?-aminobutyric acid—induced inhibition of DHNs.
More studies conducted carefully and skillfully that expand on this theoretic and empiric basis are needed before the mechanism of manipulation-induced hypoalgesia can be confirmed.