J Manipulative Physiol Ther 2003 (Jan); 26 (1): 4852 ~ FULL TEXT
Licht PB, Christensen HW, Hoilund-Carlsen PF
Department of Clinical Physiology and Nuclear Medicine,
Odense University Hospital,
Objective: Concern about cerebrovascular accidents after cervical manipulation is common. We report a case of cerebrovascular infarction without sequelae.
Clinical Features: A 39-year-old man with nonspecific neck pain was treated by his general practioner with cervical manipulation.
Intervention and Outcome: This immediately elicited severe headache and neurologic symptoms that disappeared completely within 3 months despite permanent signs of a complete left-sided cerebellar infarction on computed tomography and magnetic resonance imaging. At 7-year follow-up the patient was fully employed, and repeated magnetic resonance imaging still showed infarction of the left cerebellar hemisphere. However, the patient remained completely free of neurologic symptoms, and color duplex ultrasonography showed normal cervical vessels, including patent vertebral arteries.
Conclusion: It appears that the risk of cerebrovascular accidents after cervical manipulation is low, considering the enormous number of treatments given each year, and very much lower than the risk of serious complications associated with generally accepted surgery. Provided there is a solid indication for cervical manipulation, we believe that the risk involved is acceptably low and that the fear of serious complications is greatly exaggerated.
From the Full-Text Article:
Neck pain is a frequent condition and is costly in terms of treatment, individual suffering, and time lost from work.  Neck pain and headache are the next most common reasons for providing SMT.  Surveys of patient satisfaction favor chiropractic treatment compared with physiotherapy and treatment by general medical practitioners,  but the scientific evidence supporting the use of SMT is scarce. Clinical trials, reviews, or meta-analyses have generally been supportive of lumbar manipulation for low back pain, but data on the efficacy of spinal manipulation of the cervical spine are more limited.  In a review article, Shekelle and Coulter  identified 9 controlled clinical trials of cervical manipulation of patients with neck pain or headache: 5 trials for subacute or chronic neck pain, 3 trials for muscle tension-type headache, and 1 for migraine headache. Although the quality of these trials was claimed to vary widely, a multidisciplinary expert panel did agree, based on these data and the panel's own clinical judgement, that cervical spine manipulation is an appropriate therapy for selected patients with neck pain and headache. Vernon et al  systematically reviewed randomized clinical trials of complementary/alternative therapies in the treatment of tension-type and cervicogenic headache. Six out of the 24 identified trials used SMT. The authors conclude that no studies exist that used an exclusive placebo or sham-controlled group, but that 3 of the 4 high-quality studies report a benefit of SMT. 
In another meta-analysis of 24 randomized controlled trials of various kinds of conservative management of neck disorders, Aker et al  found that SMT in combination with other treatments was of benefit but stressed that the lack of evidence for many of the standard approaches to neck pain used in health care today was most evident. More recently, a Danish randomized controlled trial of 119 patients found no clinical difference between intensive training of the cervical musculature, chiropractic treatment, and physiotherapy in patients with chronic neck pain; all 3 groups showed meaningful improvement after 4 and 12 months, but it was unknown whether this was the result of treatment or simply a result of time.  Another recent Danish randomized controlled trial of patients with cervicogenic headache found a significant effect for cervical manipulation compared with soft tissue therapy and laser treatment.  This was not the case in a later randomized controlled trial of patients with episodic tension-type headache in which the same authors underlined the importance of establishing an accurate diagnosis when selecting headache patients for spinal manipulation. 
Concern about cerebrovascular accidents after cervical manipulation is widespread. [10-12] According to a recent editorial in the New England Journal of Medicine, cervical manipulation arouses far more concern about safety than the use of lumbar manipulation.  Many physicians frankly discourage the use of cervical manipulation for the treatment of midline neck pain because of fear of an unacceptably high risk/benefit ratio.  Fortunately, serious complications are in fact rare, whereas minor complaints such as local discomfort or headache are quite frequent. 
Blaine,  in 1925, was the first to publish findings of a complication after cervical manipulation and described forward displacement of the atlas on the axis. Other complications from cervical manipulation include myelopathies and radiculopathies, as well as spinal cord injury, but the most common of serious complications are cerebrovascular accidents in the vertebrobasilar circulation. They have been recognized since 1934,  and numerous reports have emphasized the susceptibility of the vertebral artery to trauma. The exact incidence of such accidents is not known. However, considering that spinal manipulative procedures are used approximately 250 million times every year in the United States alone,  the rate of accidents appears to be low, with only 183 cases reported in the literature until 1996.  Estimations vary from 1 in 400,000  to 1 in 3,000,000 manipulations.  Recent evidence suggests, however, that several cases remain unreported  and that serious complications may follow a hitherto unrecognized substantial proportion of spinal manipulative therapies. 
In perspective, serious cerebrovascular accidents after coronary bypass surgery occur in 1.3%,  and
risk assessments have shown that cervical manipulation is several hundred times safer than nonsteroid anti-inflammatory drugs for the treatment of neck pain, because the latter may have serious gastrointestinal side effects.  The rate of in-hospital errors is also much higher. A review of more than 30,000 patient records in 51 New York hospitals showed that iatrogenic disabling injury occured in 3.7% of cases and that 13% of these were fatal.  Similarly,
major suction lipectomy is considered a reasonably safe procedure with a low complication rate, despite a reported rate of fatal complications of 1 per 7500 operations. 
The present case adds to the list of CVAs after cervical manipulation. However, it also illustrates that such complications may leave the patient completely asymptomatic, allowing a normal life despite a severe pathologic condition demonstrated by MR. The outcome of many cases reported by Terrett1  was not available, but in 27 out of 183 patients there was a complete recovery, and 15 had an almost complete recovery. Conversely, 33 out of the 183 patients had a fatal CVA after cervical manipulation. Therefore, the treatment is certainly not without serious risk. However, the fear of CVAs seems greatly exaggerated, considering the low number of reported cases compared to the amount of treatment given and in view of the higher rate of complications with many generally accepted treatments. It is tempting to speculate that the widespread fear of cervical manipulation within the medical profession is more a political than a factual issue.