Deaths after chiropractic spinal manipulation

Article 1
Article section 1 of 3.  Next

By Douglas J Lanska MD MS MSPH

Article under review:

Ernst E. Deaths after chiropractic: a review of published cases. Int J Clin Pract 2010;64(8):1162-5.

Background: Chiropractic spinal manipulation has been associated with multiple serious neurologic sequelae, including brainstem and cerebellar stroke (including cases of Wallenberg syndrome and locked-in syndrome), cervical or thoracic spinal cord injury, cauda equina syndrome, radiculopathies, brachial plexopathies, and phrenic nerve injury. Other reported complications include dislocation of the atlas, vertebral fracture, vertebral artery pseudoaneurysm, spinal epidural hematoma, tracheal rupture, cardiac arrest, and various soft tissue injuries.

Chiropractic spinal manipulation is a risk factor for vertebral artery dissection and vertebrobasilar stroke, particularly among younger individuals (Rothwell et al 2001; Smith et al 2003). In a nested case-control study utilizing a prospective stroke registry at 2 academic medical centers, Smith and colleagues found that vertebral artery dissections were independently associated with spinal manipulation within the preceding 30 days for patients under the age of 60 (odds ratio 6.6) and with neck or head pain preceding stroke or transient ischemic attack (odds ratio 3.8). In a population-based case-control study in Ontario, Canada, Rothwell and colleagues found that, for those aged less than 45 years, vertebrobasilar stroke cases were 5 times more likely than controls to have undergone chiropractic treatment in the week prior to stroke onset (Rothwell et al 2001). In addition, for those aged less than 45 years, cases were 5 times more likely to have had at least 3 chiropractic visits with a cervical diagnosis in the month prior to stroke onset. In the study by Rothwell and colleagues, no significant associations were identified in those at least 45 years of age.

Previous literature syntheses have identified a high frequency of death among such cases of vertebrobasilar stroke following chiropractic spinal manipulation. Terrett reported 28 deaths out of 113 cases (25%) in the English- and non-English-language literature from 1934 to 1987 (Terrett 1990). Assendelft and colleagues reviewed published case reports through 1993 and found 165 cases of vertebrobasilar complications attributed to chiropractic spinal manipulation, of which 18% died (Assendelft et al 1996). Hurwitz and colleagues reported 21 deaths among 118 cases (18%) in the English-language literature (Hurwitz et al 1996).

Purpose: To summarize literature reports of chiropractic spinal manipulation followed by death.

Methods: Systematic review of published case reports, using 4 electronic databases (Medline, Embase, AMED, and Chochrane Library), with no restrictions on date or language of publication. Additional sources included review of bibliographies in identified articles and consultation with several subject matter experts.

Results: Twenty-six fatalities were identified since 1934 (abstracted from 23 separate articles). The fatalities were summarized in a table showing the source, year of publication, age and gender if known, type of recorded vascular accident, and the time interval between chiropractic spinal manipulation and death.

The age of death was given for 20 cases for whom the range of age at death was 25 to 60 years, with a mean of 38.5 years and a median of 35.5 years. Fifteen (75%) were under age 45, and 19 (95%) were under age 60. These cut-offs were chosen by the reviewer to correspond to the age cut-offs used in the epidemiological studies by Smith and colleagues (Smith et al 2003) and by Rothwell and colleagues (Rothwell et al 2001) referred to above. Ernst used a cut-off of 40 years and found that 14 (70%) were below the age of 40.

Gender was available for 22 cases, of which 13 (59%) were women.

Assessment of the type of vascular accident was limited by available details in the reports, with some articles reporting the vessels thrombosed or dissected, others the resulting areas of infarction, and still others both. In most cases for which this information was available (16 of 22; 73%), there was a documented vertebral artery dissection (n=5), vertebrobasilar thrombosis (n=5), and/or posterior circulation ischemic stroke (n=7). Other identified pathologies included single cases of death following a tear in the lateral sinus, a right hemisphere stroke, an intraspinal hemorrhage with spinal cord compression, a cerebellar hemorrhage, and an interventricular hemorrhage. In 4 cases, the type of vascular accident was unclear (n=1) or not provided (n=3).

The interval from chiropractic spinal manipulation to death was available in 19 cases for whom it ranged from 1 hour to 58 days (median of 24 hours). In 10 of 19 cases (53%), the interval from treatment to death was within 24 hours.

Conclusions and commentary: Strokes following chiropractic spinal manipulation are typically posterior circulation strokes associated with vertebral artery dissections (Terrett 1987; 1990; Hurwitz et al 1996; Smith et al 2003; Williams and Biller 2003). The vertebral artery is most commonly injured by the high-velocity, short-lever thrusts on the upper spine with associated rotation during chiropractic spinal manipulation. Injury typically occurs at the level of the atlanto-axial joint where the vertebral artery abruptly changes direction from a vertical to a horizontal orientation. Rotating and extending the neck during chiropractic spinal manipulation stretches the extracranial vertebral arteries at this location, producing a shearing force that can cause intimal tears, dissection, thrombus formation, and artery-to-artery embolism (Lee et al 1995; Assendelft et al 1996; Rothwell et al 2001).

The cases identified by Ernst are only those that were published, and as the author commented "many more might have remained unpublished." Indeed, the author identified several reports where additional deaths were mentioned, but for which insufficient information was available.

The population from which such cases occurred (ie, those that were treated with this procedure) is also unknown. Therefore, it is not possible from these data to calculate mortality rates.