J Manipulative Physiol Ther 1999 (Jul); 22 (6): 368–375 ~ FULL TEXT
Rivett DA, Sharples KJ, Milburn PD
School of Physiotherapy,
University of Otago,
Dunedin, New Zealand.
BACKGROUND: Neck manipulation occasionally causes stroke after trauma to the vertebral or internal carotid artery. Premanipulativ e tests involving cervical spine rotation or extension have been recommended to detect patients at risk of neurovascular ischemia. However, the effect of these procedures on extracranial blood flow is not well established, and their validity is thus controversial.
OBJECTIVE: To determine the effect of premanipulative tests involving cervical spine rotation or extension on vertebral artery and internal carotid artery blood flow parameters. DESIGN: Two-group experimental study.
SUBJECTS: Twenty subjects consisting of 16 patients treated with physiotherapy and four volunteers.
METHODS: Subjects were tested with a recommended premanipulative protocol by both an independent physiotherapist and an investigator. One group consisted of 10 subjects with signs or symptoms indicative of neurovascular ischemia on premanipulative testing, with 10 subjects with no signs or symptoms indicative of neurovascular ischemia on premanipulative testing comprising the second group. Hemodynamic measurements for both vertebral and both internal carotid arteries were taken by use of duplex Doppler ultrasonography with color-flow imaging with the subjects in the following positions: neutral, end-range extension, 45 degrees contralateral rotation, end-range contralateral rotation, and combined end-range contralateral rotation/extension.
RESULTS: The reliability of premanipulative testing was supported. Significant changes in flow velocity of the vertebral artery (and to a lesser extent of the internal carotid artery) were shown in end-range positions involving rotation and extension. No meaningful significant differences were found between the two groups.
CONCLUSIONS: Screening procedures that use rotation and extension may be useful tests of the adequacy of collateral circulation. A larger study is needed to determine whether subjects testing positive significantly differ from those testing negative.
From the Full-Text Article:
The results of this investigation have demonstrated that the blood flow of the extracranial arteries is significantly affected by cervical spine positions involving end-range contralateral rotation or extension. Variations in BP and PR are unlikely to account for the changes observed because these measures did not significantly differ from before to after the ultrasound examination, indicating stability of these parameters.
For the VAs there seems to be a trend of decreasing blood flow velocity (as measured with Doppler sampling in the region between the second and third cervical vertebrae) with increasing degrees of rotation. End-range contralateral rotation significantly reduced PS blood flow velocity for both VAs, consistent with the findings of Refshauge.  Refshauge  hypothesized that, at end-range rotation, narrowing of the vessel diameter may reach a critical level whereby blood viscosity becomes a factor leading to slowing of the blood flow. The combination of a relatively small vessel orifice (because of positional narrowing) and red blood cell viscosity may result in a decelerating effect known as viscous friction.  Alternatively, the decrease in flow velocity observed in the rotatory positions may reflect the fact that the site of sampling is upstream of the likely site of vessel narrowing, the atlanto-axial region. Thus the blood flow approaching this region may tend to slow, whereas the flow velocity at the actual site of narrowing will tend to increase to maintain a constant flow volume. 
The RI also decreased in end-range rotation and the combined end-range rotation/extension, suggesting that the resistance encountered by the blood flow is actually reduced. This finding seems to be inconsistent with the expectation of vessel narrowing and associated increased resistance to flow. The RI is based on the premise that diastolic velocity is likely to be reduced to a greater extent by higher resistance than is systolic velocity, leading to a rise in the index. However, the raw data indicate that both the PS and the ED velocities are reduced in these positions, albeit the PS to a proportionally greater degree.
The effect of cervical spine extension on extracranial blood flow has received comparatively little scrutiny in Doppler investigations. [11, 16] It is thus notable that flow velocity in this position generally increased for the VA and the ICA. This change in flow velocity with extension may be indicative of narrowing of the vessel at or proximal to the site of sampling. To the best of our knowledge, changes in ICA flow with cervical extension have not been previously investigated in spite of the fact that the ICA has been implicated in a small percentage of manipulative strokes. [20, 31] It may thus be of clinical note that neck maneuvers involving extension can result in blood flow changes, particularly considering that this artery is often already stenotic because of vascular disease. Refshauge  demonstrated that rotation may affect ICA peak velocity. The results of this study are not inconsistent with her results, but the trend was not always statistically significant, perhaps because of the small sample size in each group.
The two groups did not significantly differ in any meaningful measure, suggesting that the magnitude of any hemodynamic difference is probably small and of doubtful clinical significance. However, this study was not intended to answer the question of whether changes in position will have a greater effect on blood flow in people with positive clinical testing results than in those with negative results. This is currently being investigated in a further study with a larger sample size. It is possible that subjects with false-positive results were included in the positive group, such as those whose response to premanipulative testing may have been due to stressing of the upper cervical spine musculature or joints, resulting in cervical or reflex vertigo.  However, the generalizability of the findings to the clinical situation is supported because all subjects in the positive group were referred from physiotherapy practices and because of the excellent test response agreement between the independent physiotherapists and the investigator.
The negative group reported no signs or symptoms of neurovascular insufficiency during the ultrasound examination, and yet there was demonstrable change in their extracranial blood flow. In fact, one subject experienced total occlusion of the left VA in end-range rotation and also combined end-range rotation/extension on repeated testing. This case is described in detail elsewhere  ; however, it serves to illustrate that the VA may be markedly stressed and effectively ligated and yet the clinical test result is negative. It can be argued that this situation represents a case of false negativity because the vessel is clearly subjected to external forces related to the test position and is therefore at risk of trauma. What the test does indicate is the adequacy of the collateral circulation in preventing an ischemic event if the blood flow of one VA were critically reduced. On the other hand, it is likely that the negative test response does not indicate the real risk of the patient having a stroke. If vessel damage ensued from neck manipulation, then the state of the collateral circulation is of questionable value if an embolus were to dislodge from the site of trauma and enter the intracranial circulation. This risk may be present to a varying extent in lesser degrees of rotation as evidenced by the changes in blood flow at 45 degrees rotation in this investigation and in Refshauge's study. 
Thus, from a clinical perspective, a negative result to premanipulative testing does not guarantee that the proposed manipulative procedure is entirely free of risk. Such a response could only be construed as indicating the likelihood of the patient experiencing a stroke in the event of local pathology of an extracranial vessel (for example vasospasm or intimal dissection) and provided the pathology does not project cranially. It can therefore be argued that the patient at risk of manipulative stroke is often undetectable in spite of the use of premanipulative tests. There is further evidence to support this suggestion because two such incidents occurred even though premanipulative testing was performed according to the APA protocol.  It is therefore desirable to avoid manipulative maneuvers that use rotation or extension, especially those involving end-range positions. There is some evidence that lateral flexion may have minimal effect on the VA [15-17] and that the risks associated with neck manipulation may be reduced by use of procedures principally involving this movement.
This preliminary study has highlighted the need for further investigation with a larger sample to determine more conclusively whether there are differences in changes in blood flow in the testing positions between subjects in the negative and positive groups. As such, the absence of meaningful differences found in this study must be regarded with caution. It would also be desirable to sample at the atlanto-axial region of the VA if this were consistently feasible and involved an acceptable margin of error, because this is the region most vulnerable to the stresses associated with cervical spine rotation. It would also be more relevant to the interpretation of the hemodyamic findings to solely measure the vessel lumen rather than including the vessel wall. [11, 14]
The results of this pilot study provide preliminary evidence that changes in neck position influence blood flow in the extracranial arteries. In particular, the vertebral artery is subjected to forces in positions involving end-range contralateral rotation that are sufficient to significantly reduce blood flow velocity. This finding supports the contention that the premanipulative tests may be of value in assessing the adequacy of the collateral circulation. That no meaningful differences were found between subjects testing positive and those testing negative probably indicates that a larger trial is needed before conclusions can be drawn about the validity of the tests. The reliability of the premanipulative protocol in categorizing subjects as positive or negative was supported.