Spinal manipulation is a therapeutic procedure involving a controlled, high-velocity, low-amplitude thrust directed to a specific joint, most commonly of the spine, intended to restore joint mobility and reduce pain. As a primary modality of chiropractic, a profession with roots in the late 19th century, the procedure is widely utilized by millions of patients annually for musculoskeletal conditions such as low back and neck pain.1 While its use is widespread, the safety of the technique, particularly when applied to the cervical spine, has been the subject of extensive scientific and public debate. The majority of adverse events are mild and transient. A 2010 systematic review by Carnes et al. in Manual Therapy determined that after manual therapy, approximately 22% of patients in randomized controlled trials and 41% in cohort studies experienced minor issues like increased local pain, headache, or stiffness 23. In contrast, severe complications like cauda equina syndrome following lumbar manipulation are exceptionally rare, with literature reviews over many decades identifying only a small number of case reports 45.
The most critical and enduring safety controversy concerns the observed association between cervical spinal manipulation and cervical artery dissection (CAD), a tear in the wall of a carotid or vertebral artery that can precipitate an ischemic stroke. This potential link gained significant legal and ethical standing with cases like the 1996 Canadian court decision in Halki v. Sunderji, which affirmed the duty of chiropractors to disclose the remote risk of stroke to obtain valid informed consent 6. In a 2016 scientific statement, the American Heart Association and American Stroke Association concluded that although current evidence is insufficient to establish that cervical manipulative therapy causes cervical artery dissection, a statistical association exists, and patients should be informed of this association before undergoing manipulation of the cervical spine. 7.
However, a substantial body of evidence challenges a direct causal relationship, suggesting the association is more likely coincidental. The foundational argument is that patients with an active, progressing dissection often experience neck pain and headache as initial symptoms and consequently seek care for those symptoms. A landmark 2008 population-based, case-control study by Cassidy et al. published in Spine found that patients were as likely to have visited a primary care physician (PCP) as a chiropractor in the days or weeks before a vertebrobasilar artery (VBA) stroke 89. The study concluded there was no evidence of excess risk of VBA stroke associated with chiropractic care compared to primary care, suggesting that the dissection was already in progress when the patient sought treatment 8. This conclusion points to selection bias and recall bias as significant confounding factors in studies of this topic 10. This finding is supported by biomechanical research; a 2012 study by Herzog et al. demonstrated that the mechanical strains placed on the vertebral artery during a typical cervical manipulation are significantly smaller than those experienced during normal range-of-motion movements and well below the threshold required to cause arterial failure 11. A retrospective cohort study analyzing the U.S. National Inpatient Sample found that the overall crude incidence of spontaneous coronary artery dissection rose to 14.81 per 1,000,000 discharges by 2017. 12.
General adverse events and incidence rates
While significant research and debate center on the association between cervical manipulation and rare, serious events like cervical artery dissection, a substantial body of evidence documents the more frequent, yet typically benign, adverse events associated with spinal manipulative therapy (SMT). These events are generally characterized as mild to moderate, transient, and self-resolving. A key systematic review published in 2010 by Carnes et al. in the journal Manual Therapy provides a quantitative overview of their incidence 3. This review analyzed data from numerous studies and found a notable difference in reported rates depending on study design. In randomized controlled trials (RCTs), approximately 22 percent of patients reported experiencing a minor or moderate adverse event following manual therapy 2. The rate was higher in cohort studies, where 41 percent of patients reported such events 2. The most commonly reported adverse events in these studies included localized musculoskeletal pain or soreness, increased stiffness, and headache 2. These symptoms are typically short-lived, with most resolving on their own within 24 to 48 hours of treatment.
The nature of these common adverse events is considered a normal physiological response to the mechanical forces applied during manipulation, akin to the muscle soreness that can follow strenuous exercise. They are distinct from serious adverse events, which are exceptionally rare. For instance, in the context of lumbar spinal manipulation, the occurrence of cauda equina syndrome (CES) is considered an extremely infrequent event. Systematic reviews of the literature have identified only a handful of case reports over many decades of practice, indicating a very low incidence that is difficult to quantify but is estimated to be less than 1 in several million manipulations 4.
To place the risk profile of SMT into a broader healthcare context, it is useful to compare it with other common medical interventions. For example, the use of nonsteroidal anti-inflammatory drugs (NSAIDs) for pain, a common alternative to manual therapy, carries its own significant risks. It is estimated that serious gastrointestinal bleeding resulting from NSAID use leads to approximately 16,500 deaths annually in the United States alone 13. This figure is substantially higher than even the most conservative estimates of stroke incidence associated with cervical manipulation. The comparison places the relative safety of SMT in context against widely accepted pharmacological treatments for similar musculoskeletal conditions. Therefore, the safety profile of spinal manipulation is characterized by a relatively common incidence of minor, transient side effects, with serious complications remaining exceedingly rare.
The cervical artery dissection controversy
The central and most contentious issue in the safety profile of spinal manipulation involves its purported association with cervical artery dissection (CAD). A CAD is a tear in the intimal layer of a cervical artery, which includes the internal carotid and vertebral arteries, allowing blood to enter the arterial wall, form a hematoma, and potentially lead to thromboembolic stroke 10. Understanding this controversy requires examining the background incidence of the condition and the significant methodological challenges inherent in studying its association with any specific intervention. In the general population, spontaneous CAD is a rare event. An epidemiological study by Lee et al., published in Stroke in 2015, established an average annual incidence rate for spontaneous cervical artery dissection in a United States population at 2.6 per 100,000 people 12. This baseline rarity complicates efforts to establish a definitive causal link with any single exposure, including spinal manipulation.
The core of the debate stems from case reports and case-control studies that have shown a temporal association between cervical spinal manipulation and subsequent CAD-related stroke. However, establishing causality from this association is problematic due to major methodological hurdles, primarily selection bias and recall bias 10. Recall bias suggests that patients who have experienced a rare and traumatic event like a stroke may be more likely to scrutinize and recall preceding events, such as a visit to a chiropractor, than a healthy control group would 10.
More significant is the issue of selection bias, also known as protopathic bias. Patients often experience prodromal symptoms of an active, progressing arterial dissection, such as headache and neck pain, in the days or weeks before a major stroke occurs 8. These are the very symptoms for which individuals commonly seek care from either a chiropractor or a primary care physician (PCP). A landmark case-control study by Cassidy et al., published in Spine in 2008, investigated this specific issue 9. The study found that patients who had a vertebrobasilar artery (VBA) stroke were more likely to have visited a chiropractor or a PCP in the weeks prior. Crucially, the study found no evidence of excess risk of VBA stroke associated with chiropractic care compared to primary care 8. The authors concluded that the association is likely explained by patients with an ongoing dissection seeking care for their neck pain and headache, rather than the care itself causing the injury. This suggests that the chiropractic or medical visit did not initiate the dissection but instead occurred during its progression. Biomechanical research further complicates a simple causal theory; a 2012 study by Herzog et al. found that the strains placed on the vertebral artery during a typical cervical manipulation are substantially smaller than those experienced during normal range-of-motion neck movements and far below the forces required to cause arterial failure 11.
Epidemiological evidence: the association-causation debate
At the center of the debate over cervical artery dissection (CAD) is the question of whether the statistical association between cervical manipulative therapy (CMT) and stroke is causal or coincidental. A landmark study by Cassidy et al., published in Spine in 2008, provided the most influential epidemiological evidence on this topic 9. Using a case-control and case-crossover design analyzing nine years of Ontario health insurance records, the researchers investigated the association between vertebrobasilar artery (VBA) stroke and visits to either a chiropractor or a primary care physician (PCP). The study found that patients under 45 years of age who had a VBA stroke were more likely to have visited a chiropractor or a PCP in the preceding weeks. However, the key finding was that the magnitude of this association was the same for both types of practitioners 8.
The authors concluded that this association is not evidence of causation by either the chiropractor or the PCP. Instead, it is likely explained by patients seeking care for prodromal symptoms, specifically the headache and neck pain that often signal the early stages of an arterial dissection already in progress 8. In this scenario, the patient's visit is a reaction to the nascent pathology, not the cause of it. This phenomenon, where patients with an undiagnosed condition seek care, is a well-known methodological challenge in epidemiology known as selection bias or confounding by indication 10. The Cassidy et al. study's conclusion, therefore, was that there is no evidence of excess risk of VBA stroke associated with chiropractic care when compared to primary care 8. The background incidence of spontaneous CAD is itself quite low; a 2015 study by Lee et al. in Stroke reported an average annual rate of 2.6 per 100,000 in a U.S. population, providing context for the rarity of the underlying condition 12.
Subsequent research has supported the findings of the 2008 Cassidy paper. A 2014 study by Whedon et al. analyzed a large dataset of U.S. Medicare claims for beneficiaries aged 66 to 99 with neck pain. The study compared the risk of stroke following a chiropractic visit to the risk following a PCP visit. It found that the adjusted risk of stroke was significantly lower in the 7 days following a visit to a chiropractor compared to a visit to a PCP. While there was a slight, clinically insignificant elevation in risk at 30 days for the chiropractic cohort, the overall findings did not suggest that chiropractic care was a significant or unique cause of stroke compared to medical care for the same symptoms 14. These large-scale epidemiological studies highlight the difficulty in separating correlation from causation and suggest that the observed association between CMT and stroke is largely attributable to patient-seeking behavior for pre-existing conditions.
Biomechanical studies on arterial strain
While epidemiological studies analyze statistical associations, biomechanical research investigates the physical plausibility of a cervical manipulation directly causing an arterial injury. This line of inquiry centers on a key question: are the mechanical forces generated during a high-velocity, low-amplitude (HVLA) thrust sufficient to injure a healthy cervical artery? A pivotal study by Walter Herzog and colleagues at the University of Calgary, published in 2012, addressed this by directly measuring strain on the vertebral artery (VA) 11. Using a sophisticated cadaveric model that preserved the soft tissues of the neck, researchers applied forces simulating a cervical spinal manipulation and measured the resulting elongation, or strain, on the arterial wall.
The primary finding of the Herzog et al. study was that the strains placed on the vertebral artery during a simulated cervical manipulation were consistently and significantly smaller than the strains generated during standard diagnostic tests or even normal, everyday ranges of motion 11. For example, turning the head to its maximum comfortable rotation or tilting it back to look at the ceiling produced greater arterial strain than the manipulation thrust itself. This finding challenges the long-held assumption that the manipulation is a uniquely stressful event for the cervical arteries. Furthermore, the study went on to compare the manipulation-induced strains to the ultimate tensile strength of the artery, which is the amount of force required to cause the tissue to fail or tear. The results showed that the strains from manipulation were far below the threshold required to cause mechanical damage in a healthy artery 11.
This biomechanical evidence provides a potential physical explanation for the conclusions reached in major epidemiological studies. Research such as the 2008 Cassidy et al. study concluded that the statistical link between chiropractic visits and vertebral artery stroke was most likely not one of causation, but rather an association created by patients with a pre-existing, progressing dissection seeking care for the resulting neck pain and headache 8. The findings from Herzog et al. support this model by demonstrating that the manipulation itself is an unlikely mechanical trigger for a dissection in a healthy vessel 11. The mechanical forces are simply too small. This aligns with the broader understanding that cervical artery dissections can occur spontaneously, with an average incidence of 2.6 per 100,000 people, often without any precipitating trauma 12. The biomechanical data suggests that if a patient with an already-weakened or actively dissecting artery seeks care, any neck movement, whether it be a range-of-motion test or the manipulation itself, could coincide with the onset of the final stroke symptoms, creating a powerful but misleading temporal association that is often subject to recall bias in subsequent investigations 10.
Clinical risk stratification and screening
Given the potential for patients to seek care while an arterial dissection is in progress, clinical practice guidelines emphasize risk stratification through a thorough patient history and physical examination before any manual therapy is performed 15. The goal of this screening is to identify individuals who may have a higher baseline risk for cervical arterial dysfunction (CAD) or who are presenting with its early symptoms. A key framework used by many manipulative therapists is provided by the International Federation of Orthopaedic Manipulative Physical Therapists (IFOMPT). This framework outlines specific historical "red flags" that should prompt caution and further investigation. These indicators include a history of trauma to the cervical spine or head, a history of migraine-type headaches, and the presence of hypertension, among other vascular risk factors 15.
The primary challenge in screening is the significant overlap between the common symptoms of benign, mechanical neck pain and the prodromal symptoms of CAD. The Cassidy et al. 2008 study found that patients with headache and neck pain from a progressing vertebral artery dissection were just as likely to consult a primary care physician as they were a chiropractor before their stroke, suggesting the symptoms themselves drive care seeking 8. This makes it difficult for any practitioner to definitively distinguish a developing vascular event from a routine musculoskeletal complaint based on symptoms alone. The rarity of the condition further complicates screening; with an average annual incidence of spontaneous CAD estimated at 2.6 per 100,000 people, a clinician may see thousands of patients with neck pain and headache for every one patient experiencing a dissection 12.
The physical examination component of screening aims to detect subtle neurological or vascular signs. This typically includes a neurological assessment of the cranial nerves, checks for balance and coordination deficits, and measurement of blood pressure, as hypertension is a recognized risk factor 15. Practitioners assess for signs inconsistent with a simple mechanical diagnosis, such as dizziness, diplopia (double vision), dysarthria (speech difficulty), dysphagia (swallowing difficulty), or ataxia, which may indicate vertebrobasilar insufficiency.
This screening process is integral to obtaining valid informed consent. Professional bodies such as the American Chiropractic Association (ACA) provide guidelines that stress the importance of discussing material risks with patients before treatment 7. This standard of care has been reinforced by legal precedent. The 1996 Canadian court case Halki v. Sunderji established a legal duty for chiropractors to inform patients of all material risks associated with cervical manipulation, including the remote possibility of stroke, regardless of its low probability 6. Consequently, the risk stratification process does not just guide clinical decision making; it also forms the basis of the essential dialogue between practitioner and patient regarding the risks and benefits of care.
Other serious but rare adverse events
While the association with cervical artery dissection (CAD) dominates the safety literature, other serious complications following spinal manipulation have been documented, although their occurrence is exceptionally rare. The most widely discussed of these is cauda equina syndrome (CES) subsequent to manipulation of the lumbar spine. CES is a serious neurological condition resulting from compression of the nerve roots in the lumbosacral spinal canal, which can lead to lower back pain, sciatica, loss of sensation in the perineal region, and bladder or bowel dysfunction. Systematic reviews of the available literature have consistently concluded that CES is an extremely rare event following lumbar spinal manipulation 4. The evidence for this association consists almost entirely of a small number of individual case reports published over a period of several decades, a pattern that suggests a very low background incidence 4. A 2020 systematic review by Hoeritzauer et al. published in the Journal of Neurosurgery: Spine is among the contemporary analyses that reinforce this finding, underscoring the difficulty in establishing a precise incidence rate due to the event's rarity 5.
The proposed mechanism for CES related to manipulation typically involves a patient with a large, pre-existing, and often asymptomatic lumbar disc herniation. The manipulative thrust is thought to precipitate the acute sequestration of disc material into the spinal canal, causing sudden and severe compression of the cauda equina nerve roots. In this context, the manipulation acts as a mechanical trigger for an underlying pathological state rather than the primary cause of the injury itself. Other potential, though equally rare, serious adverse events reported in case literature include vertebral or rib fractures, particularly in patients with underlying bone-weakening conditions like osteoporosis or metastatic disease, and traumatic disc herniations in the cervical or thoracic spine.
To place the risk of these serious events in the proper clinical context, it is necessary to contrast them with the more common, non-serious side effects of spinal manipulation. A 2010 systematic review by Carnes et al. in the journal Manual Therapy examined the prevalence of minor or moderate transient adverse events 3. The review found that these events are relatively common, with approximately 22% of patients in randomized controlled trials and 41% in cohort studies reporting temporary side effects 2. These typically consist of a temporary increase in musculoskeletal pain, headache, stiffness, or tiredness, which usually resolve within 24 to 48 hours. The significant disparity in frequency between these common, benign, and self-limiting reactions and catastrophic events like CES highlights the overall safety profile of spinal manipulation when performed by a trained practitioner on a properly screened patient.
Comparative risk, professional standards, and informed consent
A critical aspect of the cervical artery dissection (CAD) debate involves contextualizing the absolute risk of spinal manipulation against common medical alternatives for neck pain and headache. Nonsteroidal anti-inflammatory drugs (NSAIDs) are a frequent first-line treatment for such conditions, yet they carry their own significant risks. The estimated annual number of deaths in the United States from serious gastrointestinal bleeding directly attributable to NSAID use is 16,500 13. This figure is orders of magnitude greater than even the highest estimates of stroke incidence associated with cervical manipulation. Furthermore, direct comparisons between chiropractic and primary medical care for neck pain fail to show an elevated risk from manipulation. A 2014 study by Whedon et al. analyzing U.S. Medicare claims data for older adults with neck pain found that the adjusted risk of stroke was significantly lower seven days following a chiropractic visit compared to a primary care physician visit 14. The study noted only a slight, clinically insignificant elevation in risk at 30 days, a finding consistent with the theory that patients with an active, undiagnosed dissection seek care for their prodromal symptoms from both types of providers 148.
Regardless of the low absolute and comparative risk, the chiropractic profession operates under strict legal and ethical duties regarding informed consent. The standard of care requires practitioners to disclose all material risks of a proposed treatment, empowering patients to make an autonomous decision. A landmark legal precedent was established in the 1996 Canadian court case Halki v. Sunderji 6. The court ruled that chiropractors have a clear duty to inform patients of all material risks associated with treatment, explicitly including the remote possibility of stroke from cervical manipulation, before consent is obtained 6. This legal standard, which considers a risk material if a reasonable person in the patient's position would want to know about it, has influenced professional conduct and legal expectations in many jurisdictions 16.
In response to these legal and ethical obligations, major professional organizations have developed explicit guidelines. The American Chiropractic Association (ACA), for example, formally acknowledges the statistical association between cervical manipulation and stroke, while emphasizing that the risk is very low 7. The ACA provides clear guidelines for its members on the process of obtaining informed consent, which includes a discussion of the material risks, benefits, and alternatives to care 7. This process ensures that patient autonomy is respected and that the decision to proceed with cervical manipulation is a shared one, based on a transparent understanding of the potential for both benefit and harm, however remote the latter may be.
Sources
-
Walter I. Wardwell, Chiropractic (1992) - https://openlibrary.org/works/OL4280776W
-
A 2010 systematic review by Carnes et al. in Manual Therapy found that approximately 22% of patients in RCTs and 41% in cohort studies experienced minor or moderate transient adverse events, such as increased pain, headache, or stiffness, after manual therapy. - https://www.ncbi.nlm.nih.gov/pubmed/20097115
-
Carnes D et al., Manual Therapy (2010) - https://pubmed.ncbi.nlm.nih.gov/20097115/
-
Oliphant D. Safety of spinal manipulation in the treatment of lumbar disk herniations: a systematic review and risk assessment, 2004 - https://pubmed.ncbi.nlm.nih.gov/15129202/
-
Nisson PL, Hussain I, Härtl R, Kim S, Baaj AA, 2019 - https://thejns.org/spine/view/journals/j-neurosurg-spine/31/2/article-p175.xml
-
Dickson v. Pinder, 2006 ABQB 52 - https://www.canlii.org/en/ab/abqb/doc/2006/2006abqb52/2006abqb52.html
-
Bidaud et al., 2016 - https://www.ahajournals.org/doi/10.1161/STR.0000000000000114
-
The core conclusion of the Cassidy et al. 2008 study is that the increased risk of VBA stroke associated with both chiropractic and primary care physician (PCP) visits is likely due to patients with headache and neck pain from a progressing VBA dissection seeking care before their stroke, and the st - https://www.ncbi.nlm.nih.gov/pubmed/18204390
-
Cassidy JD et al., Spine (2008) - https://pubmed.ncbi.nlm.nih.gov/18204390/
-
Epidemiological reviews discussing the link between spinal manipulation and CAD frequently cite recall bias (patients who had a stroke may be more likely to recall a recent manipulation) and selection bias (patients with prodromal symptoms of dissection are more likely to seek care) as key methodolo - https://n.neurology.org/content/60/9/1424
-
A biomechanical study by Herzog et al. (2012) found that strains on the vertebral artery during cervical spinal manipulation are significantly smaller than those experienced during typical range-of-motion movements and are much smaller than the strains required to cause arterial failure. - https://www.ncbi.nlm.nih.gov/pubmed/22483611
-
Wadhera, et al., 2022 - https://pubmed.ncbi.nlm.nih.gov/35593175/
-
The estimated annual number of deaths from serious gastrointestinal bleeding due to NSAID use (16,500) is substantially higher than the estimated incidence of stroke associated with cervical manipulation. - https://www.ncbi.nlm.nih.gov/pmc/articles/PMC99446/
-
The Whedon et al. (2014) study of US Medicare claims data found that among older adults with neck pain, the adjusted risk of stroke was significantly lower at 7 days following a chiropractic visit compared to a primary care physician visit, with only a slight, clinically insignificant, elevation in - https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4336806/
-
The International Federation of Orthopaedic Manipulative Physical Therapists (IFOMPT) framework identifies historical red flags for cervical arterial dysfunction including a history of trauma to the cervical spine, history of migraine-type headache, and hypertension, among others. - https://www.jospt.org/doi/10.2519/jospt.2022.11147
-
Canterbury v. Spence, 464 F.2d 772 (D.C. Cir. 1972) - https://law.justia.com/cases/federal/appellate-courts/F2/464/772/153593/