RRS Education Research Reviews DATABASE

Research Review By Dr. Michael Haneline©


Download MP3

Date Posted:

March 2017

Study Title:

Risk of Carotid Stroke after Chiropractic Care: A Population-Based Case-Crossover Study


Cassidy J, Boyle E, Côté P, Hogg-Johnson S, Bondy S, Haldeman S.

Author's Affiliations:

Department of Sports Science and Clinical Biomechanics, Faculty of Health, University of Southern Denmark, Odense, Denmark; Division of Health Care and Outcomes Research, Krembil Research Institute, University Health Network, Toronto, Canada; Divisions of Epidemiology & Biostatistics, Dalla Lana School of Public Health, University of Toronto, Canada; Faculty of Health Sciences, University of Ontario Institute of Technology and UOIT-CMCC Centre for Disability Prevention and Rehabilitation, Oshawa, Ontario, Canada; Institute for Work and Health, Toronto, Canada; Department of Neurology, University of California, Irvine, California.

Publication Information:

Journal of Stroke and Cerebrovascular Diseases 2017; 26(4):842-850.

Background Information:

Many people with neck pain and headache seek care from doctors of chiropractic (DCs) and medical doctors. There is evidence that cervical spine manipulation (CSM), which is used by many chiropractors and other manual therapists, can improve neck pain and some forms of headache. However, there is concern that high-velocity, low-amplitude manipulation could harm the cervical arteries (internal carotid and/or vertebral), leading to dissection and subsequent stroke.

Some medical neurologists are so concerned about CSM-related strokes that they advise against chiropractic neck manipulation. A consensus statement from the American Heart and Stroke Associations on CSM-related cervical artery dissections recommended that patients be informed that there is a statistical association before undergoing CSM.

Much of the opinion surrounding CSM-related stroke is based on case reports, which, while capable of raising concerns and hypotheses about risk, are not well-suited to quantify statistical risk and determine causation. Studies that include a control group would be necessary, but there are several problems in performing a study of this nature. Firstly, internal carotid artery dissection is very rare, occurring in an estimated 1.72 persons per 100,000 population annually (1). In the USA, a disease considered rare when it affects fewer than 200,000 (about 1 in 1,500) persons in America at a given time (2). Secondly, most dissections are diagnosed in patients who are hospitalized for stroke.

No cases of stroke have been reported as an adverse event in published trials that have investigated CSM, though these trials have been too small to detect such a rare event.

The case–control study is the best study design to investigate rare events and 5 studies on chiropractic care in relation to cervical artery dissection have been published to date. Strong associations were found between chiropractic care and vertebrobasilar artery (VBA) stroke in 3 of the studies, whereas 1 study found no association. Another study investigated cervical artery dissections that affected either the internal carotid or vertebral artery, or both, but no association was found between CSM and carotid artery dissection strokes.

This study investigated the associations between chiropractic care and carotid artery-related stroke as compared to primary care physician (PCP) exposures in the same analyses. The researchers hypothesized that if the risk of carotid stroke increased with chiropractic care, the strength of association would exceed the association between PCP visits and stroke.

A second hypothesis was that protopathic bias was likely if the association between healthcare visits and carotid stroke increased when the analyses were limited to visits provided for neck pain and headache-related diagnoses. Protopathic bias can happen when an exposure, such as CSM, is provided early in the course of a disease before it is diagnosed; for instance, when a patient presents for care with dissection-related neck pain or headache prior to the dissection causing ischemic symptoms. Protopathic bias in a case-control study may falsely lead one to conclude that the exposure (CMS) caused the final outcome (stroke) (4).

Pertinent Results:

A total of 15,523 carotid artery stroke cases were included in this study – 61.4% were male and the average age was 69.9 years. The clear majority stroke cases were older, with only 214 (1.4%) of the cases occurring in patients less than 45 years of age.

For the entire cohort, during the 14-day period preceding their strokes:
  • 5,433 cases (35%) received only PCP services,
  • 186 cases (1.2%) received only DC services, and
  • 116 cases (0.7%) received both services.
Overall, there were few cases exposed to chiropractic care prior to their strokes. Only 6 cases who were less than 45 years of age saw a DC within 14 days of their strokes, compared to 70 who saw a PCP (remember, there weren’t many of these younger cases). In this group of cases under 45, there was an association with stroke for both the PCP and DC groups; however, the difference between the groups was not significant. There were no younger cases who saw both a DC and PCP within 14 days of their carotid artery stroke.

For stroke cases 45 years of age and older, there was no positive association with chiropractic care, or with a combination of DC/PCP care. However, there was a weak association with PCP care in the group of older persons.

When only visits that involved neck pain and/or headache were included in the analyses, the ORs increased considerably for all DC and PCP exposures in cases who were less than 45 years of age, except for the exposure period of 3 days before the stroke. To illustrate, in those under 45 years of age, the odds ratios (ORs) of the association between having a DC and PCP visit (all services) within 14 days of the stroke were:
  • DC only: 3.14 (95% CI = 0.81-12.11)
  • PCP only: 1.97 (95% CI = 1.36-2.84)
By comparison, when considering only visits involving neck pain and/or headache, the corresponding 14-day odds ratios increased significantly for this younger age group, as follows:
  • DC only: 7.41 (95%CI = 1.39-39.67) and
  • PCP only: 6.56 (95%CI = 2.59-16.59).
It is worth noting that the 14-day ORs for patients over the age of 45 for visits only concerning neck pain and/or headache were:
  • DC only: 0.75 (95% CI = 0.60-0.94)
  • PCP only: 1.86 (95% CI = 1.62-2.12)

Clinical Application & Conclusions:

This study showed that there was an increased association between DC and primary care physician (PCP) visits and subsequent hospitalization for carotid artery strokes in cases who were less than 45 years of age. However, there was statistically no difference between the associations for DC and PCP visits.

The increased associations were even higher when the analyses were limited to cases with neck pain and headache-related diagnoses.

Because increased associations between DC and PCP visits and carotid artery stroke were found to be similar, yet PCPs do not typically perform CSM, there must be another explanation than CSM causing these strokes. The authors suggest protopathic bias as an explanation, wherein younger patients with an impending carotid artery stroke seek care for dissection-related head and neck pain prior to developing the stroke. Care provided by either a DC or PCP in this scenario would be coincidental to the stroke and not on the causal pathway.

Headache and neck pain are commonly treated by DCs and are also common symptoms in patients with cervical artery dissection. Carotid artery dissections are also notoriously difficult to diagnose (3), however, and when neurological signs and symptoms are absent, no practical, clinically valid screening tests are available to identify underlying dissections in patients with head or neck pain.

Because the symptoms of internal carotid artery dissection are often indistinguishable from those of musculoskeletal neck pain and headache, coupled with the fact that the condition is difficult to diagnose, clinicians may misdiagnose and consequently provide well-intentioned, yet inappropriate treatment leading to claims of malpractice. Fortunately, internal carotid artery dissection is very rare, so most DCs may never see such a patient in their entire careers.

Study Methods:

This was a population-based, case-crossover study that used data from a Canadian administrative healthcare database. Cases served as their own controls by sampling different control periods that occurred prior to the date of the index stroke. ICD-9-CM codes were used to identify the diagnoses of stroke cases.

The source population included adults 18+ years of age who resided in Ontario, Canada and were eligible to receive health care under the provincial health insurance plan. Services by DCs were covered by the Ontario Health Insurance Plan (OHIP) when this study was conducted.

Inclusion criteria:
  • All incident carotid artery stroke cases discharged from hospitals between April 1, 1993 and March 31, 2002.
  • At least 1 year of healthcare coverage prior to the incident stroke.
Exclusion criteria:
  • Previous hospital admission(s) with a discharge diagnosis of stroke.
  • A concurrent stroke discharge diagnosis.
  • Cases who were in a long-term care facility during the year prior to their strokes.
Ambulatory DC and primary care physician (PCP) fee codes were extracted from OHIP for the year prior to the index stroke. In addition, visits for neck pain or headache-related services were analyzed to investigate for protopathic bias.

Four control periods were randomly chosen during the year prior to the stroke for each case, using a time-stratified approach. The control periods were matched to exposure windows of 1, 3, 7, and 14 days in accord with the hazard period under consideration.

Associations between stroke and healthcare visits were estimated using conditional logistic regression. Discrete models were built for all visits and for neck or headache-related visits for each different hazard period. The 4 levels of exposure variables were:
  1. No exposure;
  2. DC only visit;
  3. PCP only visit; and
  4. Both DC and PCP visits.
DC visits that occurred on the day of the stroke were included in the analyses, whereas PCP visits on the day of the stroke were excluded because cases may have seen a PCP after their strokes, but prior to hospital admission.

Cases who were younger (age < 45 years) were analysed separately from older cases (age ≥ 45 years), as were the 1-, 3-, 7-, and 14-day exposure periods.

Study Strengths / Weaknesses

  • One of the strengths of this study is that the researchers were cognizant of the potential for protopathic bias when investigating this relationship. As stated in the first section of this Review, this type of bias can occur when an exposure, such as CSM, is provided early in the course of a disease before it is diagnosed; for instance, when a patient presents for care with dissection-related neck pain or headache prior to the dissection causing ischemic symptoms. Protopathic bias in a case-control study may falsely lead one to conclude that the exposure (CMS) caused the final outcome (stroke) (4).
  • Another strength is the use of the case-crossover design where cases were compared with themselves within different periods of time. In other words, the cases in this study also served as the controls. As a result, stroke risk factors (e.g., smoking, obesity, genetic susceptibility, etc.) that otherwise would not be captured in health administrative data were compensated for.
  • A study limitation was the use of administrative data which could not accurately determine whether all chiropractic visits resulted in CSM. While more than 80% of patients seeing chiropractors in Ontario receive spinal manipulation, visits may have been misclassified as CSM exposures but were not, in which case an underestimation of the DC–stroke association could result.
  • Other limitations include the exclusion of PCP visits on the same day of the stroke, diagnostic work-up bias in cases that presented to the hospital after chiropractic care, and the results of the study being based on a small proportion of exposed cases.

Additional References:

  1. Lee VH, Brown RD Jr, Mandrekar JN, et al. Incidence and outcome of cervical artery dissection: a population based study. Neurology 2006; 67: 1809-1812.
  2. What is a Rare Disease? [Internet] Rare Disease Day 2016 [cited 30 Jan 2016]. Available from:http://www.rarediseaseday.org/article/what-is-rare-disease-day
  3. Lee WW, Jensen ER. Bilateral internal carotid artery dissection due to trivial trauma. J Emerg Med. 2000;19(1):35–41.
  4. Horwitz RI, Feinstein AR. The problem of "protopathic bias" in case-control studies. Am J Med. 1980 Feb;68(2):255-8.

Contact Tech Support  Contact Dr. Shawn Thistle
RRS Education on Facebook Dr. Shawn Thistle on Twitter Dr. Shawn Thistle on LinkedIn Find RRS Education on Instagram RRS Education (Research Review Service)