J Manipulative Physiol Ther 2015 (Nov); 38 (9): 677–691
~ FULL TEXT
Jeffrey J. Hebert, DC, PhD, Norman J. Stomski, PhD, Simon D. French, PhD, MPH, BAppSc(Chiropractic), Sidney M. Rubinstein, DC, PhD
Department of Epidemiology and Biostatistics,
EMGO+ Institute for Health and Care Research,
VU University Medical Center,
Amsterdam, the Netherlands.
Objective: The purpose of this study was to systematically search the literature for studies reporting serious adverse events following lumbopelvic spinal manipulative therapy (SMT) and to describe the case details.
Methods: A systematic search was conducted in PubMed including MEDLINE, EMBASE, CINAHL, and The Cochrane Library up to January 12, 2012, by an experienced reference librarian. Study selection was performed by 2 independent reviewers using predefined criteria. We included cases involving individuals 18 years or older who experienced a serious adverse event following SMT applied to the lumbar spine or pelvis by any type of provider (eg, chiropractic, medical, physical therapy, osteopathic, layperson). A serious adverse event was defined as an untoward occurrence that results in death or is life threatening, requires hospital admission, or results in significant or permanent disability. We included studies published in English, German, Dutch, and Swedish.
Results: A total of 2046 studies were screened, and 41 studies reporting on 77 cases were included. Important case details were frequently unreported, such as descriptions of SMT technique, the pre-SMT presentation of the patient, the specific details of the adverse event, time from SMT to the adverse event, factors contributing to the adverse event, and clinical outcome. Adverse events consisted of cauda equina syndrome (29 cases, 38% of total); lumbar disk herniation (23 cases, 30%); fracture (7 cases, 9%); hematoma or hemorrhagic cyst (6 cases, 8%); or other serious adverse events (12 cases, 16%) such as neurologic or vascular compromise, soft tissue trauma, muscle abscess formation, disrupted fracture healing, and esophageal rupture.
Conclusions: This systematic review describes case details from published articles that describe serious adverse events that have been reported to occur following SMT of the lumbopelvic region. The anecdotal nature of these cases does not allow for causal inferences between SMT and the events identified in this review. Recommendations regarding future case reporting and research aimed at furthering the understanding of the safety profile of SMT are discussed.
Keywords: Cauda Equina Syndrome; Injury; Intervertebral Disc Displacement; Lumbosacral Region; Manipulation; Risk; Spinal.
From the FULL TEXT Article:
Spinal manipulative therapy (SMT) is a form of manual therapy commonly used to provide care for people with low back pain and other disorders of the lumbar spine and pelvis.  The frequency of SMT use by health care providers has increased over the past several decades.  Spinal manipulative therapy is generally recommended by treatment guidelines [3–5] and appears to be a cost-effective therapeutic option for patients with spinal pain. 
Previous prospective analyses of harm following lumbopelvic SMT have primarily reported benign and self-limiting events, such as muscle soreness and local discomfort, but have not observed and, hence, reported the occurrence of serious adverse events. Senstad et al  investigated the outcomes of a large cohort of chiropractic patients (n = 1,058) and reported that when SMT was included in the course of care, an adverse reaction was associated with 25% of SMT treatments and 55% of patients reported at least 1 adverse event. Reactions primarily consisted of short-term local discomfort, headache, and tiredness or radiating discomfort and were classified as mild to moderate in most patients. There were no reports of serious adverse events. In another analysis, Leboeuf-Yde et al  surveyed 66 chiropractors who reported on 625 patients who received 1,858 chiropractic treatments, of which 99% included spinal manipulation. Adverse events were found to be common, benign, and usually self-limiting within 24 hours. In addition, the authors reported that adverse reactions with chiropractic treatment were associated with female sex, treatment at first consultation, and longer pain duration. Neither study, however, reported a serious adverse event associated with SMT.
Recent systematic reviews of randomized controlled trials, cohort studies, and a patient survey have failed to identify a single serious adverse event following lumbopelvic SMT. [9–13] However, the reporting of harms in the primary literature is generally poor,  and case reports of serious adverse events following lumbopelvic SMT have been described. Examples of serious adverse events following lumbopelvic SMT include cauda equina syndrome,  lumbar disk herniation,  fracture,  and spinal hematoma.  Gouveia et al  undertook a systematic review of articles reporting adverse events following chiropractic treatment. Although the authors identified several reports of serious adverse events following lumbopelvic SMT, they implemented a limited search strategy and excluded adverse events following therapy performed by other types of healthcare providers.
Although the incidence of serious adverse events following lumbopelvic SMT is thought to be extremely low,  such events represent a potentially important source of morbidity. Improved knowledge of SMT risk with respect to serious adverse events has potential to inform clinical decision making, and understanding the circumstances surrounding such events would serve as a first step in this process. Therefore, the purposes of this study were to systematically search the literature for cases reporting serious adverse events following lumbopelvic SMT, to describe the case details, and to offer recommendations regarding future case reporting.
Types of Studies and Participants
Case reports, case series, and studies using other designs to report original, individual case details were included. Identified cases involved individuals 18 years or older who experienced a serious adverse event following SMT applied to the lumbar spine or pelvis. We included studies published in English, German, Dutch, and Swedish.
Types of Interventions
The intervention of interest was SMT applied to the lumbar spine or pelvis. The terminology reported in the literature to describe SMT and other manual therapy procedures has been described as problematic,  with some calling for a clear distinction between spinal mobilization and spinal manipulation. [22, 23] Consequently, we considered SMT to include both spinal manipulation and spinal mobilization, although we attempted to distinguish between these 2 approaches when extracting data. We operationally defined spinal manipulation as a therapeutic procedure involving the use of a high-velocity, low-amplitude thrust, whereas spinal mobilization was defined as a nonthrust therapeutic procedure involving low-velocity passive joint movements. Cases involving SMT applied while the patient was under anesthesia were excluded.
Types of Outcome Measures
The outcomes of interest in this systematic review were serious adverse events. No widely adopted definition of serious adverse event exists in the rehabilitation literature. We defined a serious adverse event as an untoward occurrence that results in death or is life threatening, requires hospital admission, or results in significant or permanent disability.  Examples of serious adverse events resulting from lumbopelvic SMT could include disk herniation, cauda equina syndrome, fracture, dislocation, or hematoma/hemorrhagic cyst.
Search Methods for Identification of Studies
A comprehensive search was conducted by an experienced reference librarian in the following databases from inception to January 12, 2012: PubMed including MEDLINE, EMBASE, CINAHL (via EBSCO), and The Cochrane Library (via Wiley). Search terms included MeSH terms in PubMed, EMtree terms in EMBASE, and CINAHL Headings in CINAHL as well as free text terms. We used free text terms only in The Cochrane Library. The search syntax developed for PubMed is presented in Table 1. No time or language restrictions were applied. The reference lists of included studies were also examined.
Selection of Studies
Two review authors independently examined the title and abstract of studies identified by the search strategy and excluded those studies not meeting the selection criteria. Next, the full text of reports thought to fulfill the selection criteria were retrieved and assessed. Disagreements between the review authors regarding study inclusion were resolved by consensus, and if necessary, a third review author was consulted. All review authors were experienced in conducting systematic reviews.
We extracted the following information, when available: clinician related (discipline, country of origin, years in practice) and participant related (age, sex) as well as clinical case details (indication for manual therapy, manual therapy technique, adverse event description including time from SMT to the adverse event, contributing factors, and clinical outcome). When the country of origin was not identified, we reported the primary author's country as identified by their affiliation. Given that there are no widely accepted criteria for judging the quality of adverse event reporting and the current studies' objective of describing case details, we did not assess the risk of bias in the included studies.
Selection of Studies for Inclusion
Figure 1 describes the study selection process.  The literature search generated a total of 2,512 references: 1,083 in PubMed, 1,224 in EMBASE.com, 70 in CINAHL, and 135 in The Cochrane Library. An additional 29 references were identified by searching the references of selected articles. After removing duplicate references, 2046 records remained. Following the title and abstract screen, 117 articles were obtained for full-text review, and 41 studies [15–18, 26–62] involving 77 cases were included.
Description of Cases
Table 2 describes the case details for all included studies. Reporting was frequently incomplete as evidenced by the suboptimal case details of the included studies. This was most notable with reporting related to the descriptions of SMT technique, the pre-SMT presentation of the patient, the specific details of the adverse event, time from SMT to the adverse event, factors contributing to the adverse event, and clinical outcome.
The most commonly reported adverse events were signs and symptoms consistent with cauda equina syndrome (29 cases, 38% of total) and lumbar disk herniation (23 cases, 30% of total). Additional adverse events consisted of fracture (7 cases, 9%), hematoma or hemorrhagic cyst (6 cases, 8%), or other serious adverse events (12 cases, 16%) including neurologic or vascular compromise, soft tissue trauma, muscle abscess formation, disrupted fracture healing, and esophageal rupture.
Most cases were reported from Europe (35, 46%) and North America (32, 42%), with few cases from Australia (7, 9%) and Asia (3, 4%). Of the reported patient demographic data, the mean (SD) patient age was 50.1 (15.9) years, and 41% were female. Of the 61 cases with available data on the patient's pre-SMT presentation, 58 (95%) appeared to have signs or symptoms originating from the lumbopelvic region (eg, low back pain, sciatica), whereas the indication for lumbopelvic SMT in the remainder of cases was for pain in other regions (eg, neck pain, mid back pain) or other complaints (eg, dyspnea).
Of the 50 cases reporting clinician type, 40 (80%) identified the SMT provider as a doctor of chiropractic/chiropractor; 3 (6%), an osteopath; 2 (4%), a medical doctor or physician; and 5 (10%), another type of healthcare provider or nonprofessional.
Approximately half (49%) of the cases reported information regarding the time to onset of the adverse event following SMT; 29 (76%) of these cases occurred within 24 hours of SMT. Sixty-three cases (82% of total) reported information describing treatment for the adverse event, of which 53 (84%) involved some type of surgical intervention. Fifty-three cases (69% of total) reported some information related to the patient's clinical outcome, and of those, 34 (64%) reported favorable clinical outcomes (ie, no major functional impairment) following treatment. No information was available on years of experience of the health care practitioner.
We conducted a comprehensive systematic review of cases describing serious adverse events following lumbopelvic SMT. Of the 77 cases involving a serious adverse event following lumbopelvic SMT, cauda equina syndrome and lumbar disk herniation accounted for the majority. Important details of most cases were frequently missing or poorly reported. These results have clinical and research-related implications.
Given the mechanical nature of SMT, it is intuitive to think that disk herniation and cauda equina syndrome could result from this therapy. This presumption is supported by some cases that we identified in which there appears to be a strong temporal relationship between SMT and the onset of signs and symptoms of these conditions. Based on these cases alone, 1 clinical recommendation could be to avoid the use of SMT when there is potential for disk herniation or cauda equina syndrome. However, it is unreasonable to make such a recommendation based solely on knowledge derived from anecdotal cases. An accurate judgment regarding the value of any therapy requires knowledge of additional domains such as
(1) clinical effectiveness,
(2) cost-effectiveness, and
(3) details of the therapies' risk profile including estimates of adverse event incidence and the magnitude of risk relative to other treatment options.
Recent evidence supports the clinical and cost-effectiveness of lumbopelvic SMT when applied to patients with nonspecific low back pain. [3, 4, 6] However, the existence of a subgroup of patients for whom SMT is most appropriate may be important when considering the therapy's risk profile. Preliminary evidence suggests that SMT's clinical effect may in part be moderated by changes in back muscle function, [63, 64] and recent research suggests that those patients who are most likely to experience improved muscle function  or clinical outcome [66, 67] following SMT can be identified by information from the history and physical examination. Implementing this evidence to inform clinical decision making and directing SMT to those patients most likely to experience benefit has potential to reduce the probability of exposure and harm.
In addition to low back pain, current evidence suggests that SMT is likely to benefit patients with known lumbar disk herniation, [68, 69] with randomized trials reporting SMT to be superior to sham  and equivalent to microdiscectomy.  Conversely, it seems improbable that patients experiencing cauda equina syndrome would benefit from SMT. Cauda equina syndrome is a rare sequela of lumbar disk herniation, with only 0.12% of herniations resulting in this disorder.  Nevertheless, cauda equina syndrome is a serious condition that requires urgent surgical intervention.  Therefore, clinicians should be diligent in screening patients and maintain a high index of suspicion for cauda equina syndrome when patients present with one or more of the following signs or symptoms: (1) bladder and/or bowel dysfunction, (2) reduced sensation in the saddle area, or (3) sexual dysfunction with possible neurologic deficits of the lower limb. 
With respect to serious adverse events, a detailed understanding of SMT's risk profile will be difficult to achieve. Previous attempts to prospectively estimate the incidence of these events have been unsuccessful as no serious events were identified, and therefore, this rate is assumed to be extremely low. [7, 8, 75, 76] Furthermore, incidence estimates derived from case studies are unlikely to be valid representations of the true rate of serious adverse events following SMT and should be interpreted with caution.
However, additional knowledge of SMT's risk profile is gained from analyses of comparative safety between SMT and other treatment options. Carnes et al  undertook a meta-analysis of randomized controlled trials, which, in part, examined the relative risk (RR) of mild to moderate adverse events (no major adverse events were identified) when manual therapy was compared with other therapeutic options. Their results demonstrated a similar risk of adverse events occurring from manual therapy as compared with exercise (RR [95% confidence interval], 1.04 [0.83-1.31]) and a lower level of risk with manual therapy (high-velocity manipulation) when compared with drug therapy (RR [95% confidence interval], 0.05 [0.0-0.20]). Similarly, Oliphant  estimated lumbar manipulation to be safer than nonsteroidal anti-inflammatory drugs among patients with lumbar disk herniation.
Recommendations for Future Research and Reporting
Additional research is needed to enhance our knowledge of serious adverse events reported to follow lumbopelvic SMT. Although anecdotal evidence is of limited value, additional case reporting may be beneficial given our limited knowledge of these rare adverse events. However, future case reporting should strive to ensure that a detailed and accurate account of the case is presented. In this literature, one known reporting issue relates to the inaccurate representation of provider type and treatment. Specifically, the terms “chiropractor,” “chiropractic,” and “chiropractic manipulation” are problematic with respect to their misuse in the biomedical literature, where any health care worker who provides SMT is labeled as a chiropractor regardless of their clinical training. [78–82]
Additional instances of suboptimal case reporting identified in this review include the details of the patient's presentation before SMT, SMT technique, time to the adverse event, contributing factors, and long-term clinical outcome. One possible explanation for the lack of detailed reporting is that many cases appear to be authored by clinicians, such as spine surgeons, who are responsible for the care of patients following occurrence of the serious adverse event. In these instances, the clinicians may have little knowledge of or interest in the details of the manual therapy or the patient presentation before SMT. Therefore, seeking additional information directly from the SMT provider has potential to improve the quality of case reporting and further our understanding in this area.
For instance, little is known about the safety profile of different SMT techniques. Although SMT is often considered to be a singular approach, it is a collection of poorly described methods that may differ with respect to their safety profile.  A review of the manual therapy techniques reported in Table 2 reveals that generic terminology such as “back manipulation” frequently represents the extent of reporting. Furthermore, some technique descriptions were unusual and inconsistent with what is taught in modern education programs teaching manual therapy. For example, Livingston  includes the following patient description of a manual therapy treatment received 2 years prior: “He stated that he lay on a table and the chiropractor stood on a high stool and jumped onto his back with both knees.” Although it is possible such unusual descriptions accurately describe the treatment provided, they raise the question of mistaken reporting. Accurate and comprehensive reporting of SMT technique details, obtained directly from the SMT provider, would likely enhance reporting and may serve as a first step toward understanding the comparative safety of these methods.
Parties interested in additional recommendations for publishing adverse event reports should refer to the guidelines endorsed by the International Society for Pharmacoepidemiology and the International Society of Pharmacovigilance.  Although these guidelines were developed for the purpose of enhancing case reports describing suspected adverse effects of drugs and medical products, many of the recommendations are relevant to reporting of adverse events following SMT.
Although case reports make a modest contribution to scientific knowledge, they represent a level of scientific evidence that is inappropriate for making causal inferences. Ultimately, our understanding of lumbopelvic SMT risk will be enhanced through high-quality clinical research. Future research efforts should aim to further explore the potential association between lumbopelvic SMT and the types of adverse events reported here as well as identify accurate estimates of incidence using methodology ensuring low risk of bias and confounding. Ideally, this would involve prospective investigations of large population-based patient cohorts. Previous [7, 8, 82–84] and current85 research efforts have undertaken a prospective approach to examine the adverse events associated with SMT. However, as previously discussed, prospective studies have failed to identify a serious adverse event following lumbopelvic SMT. Accordingly, prospective study designs may not be feasible in this context.
Retrospective study designs such as cohort, case-control, and “case only” approaches86 including case-crossover and self-controlled case series designs offer a more efficient method of assessing the risk of rare events. However, an inherent source of bias in retrospective study designs is their inability to identify the temporal sequence of exposure to a potential risk factor (eg, SMT) and the outcome of interest (eg, serious adverse event). Therefore, with retrospective study designs, it is not possible to know whether a person is free from the outcome of interest at the time of exposure. Incorrect temporal judgments can result in confounding due to protopathic bias.  This form of bias occurs when a treatment for the first symptoms of a disease or injury appear to cause the outcome and is a potential source of confounding in any retrospective study, including the cases identified in this review. For example, Fisher31 reported on a 32-year-old woman who “was given an adjustment to her lower spine.” Before SMT exposure, the patient had severe, bilateral radicular leg pain and progressive left leg weakness. Following the SMT, the patient went on to develop saddle anesthesia and bowel and bladder incontinence leading to a diagnosis of cauda equina syndrome. Because of the retrospective nature of this report, it is unknown whether the patient was already experiencing the early signs of cauda equina syndrome (eg, bilateral lower extremity pain and progressive motor deficit) before SMT exposure.
Researchers seeking to understand the potential relationship between neck manipulation, vertebrobasilar artery dissection, and stroke have encountered a similar challenge. Early case-control studies identified a relationship between vertebrobasilar artery stroke and previous exposure to neck manipulation. [88, 89] Subsequently, Cassidy et al  conducted a case-control and case-crossover study investigating the potential relationship between vertebrobasilar artery stroke and visits to both doctors of chiropractic and primary care physicians. Their results replicated the previous studies in that an association between vertebrobasilar artery stroke and visits to a doctor of chiropractic was identified. However, Cassidy et al also identified an association between vertebrobasilar artery stroke and visits to primary care physicians. Because common symptoms of vertebrobasilar artery dissection include neck pain and headache,91 the authors concluded that the increased risk of vertebrobasilar artery stroke associated with visiting a doctor of chiropractic or primary care physician is likely due to patients with dissection-related headache and neck pain seeking care before the onset of stroke. Therefore, cases of vertebrobasilar artery stroke following neck manipulation may be inappropriately attributed to exposure to neck manipulation when the symptoms of dissection (eg, headache, neck pain) were present before treatment (ie, these associations may result from protopathic bias). A similar research approach investigating lumbopelvic SMT may help elucidate whether this therapy should be included in the causal pathway of the types of adverse events identified in this review.
Finally, the development of surveillance tools allowing the widespread and systematic reporting of adverse events following SMT may provide a valuable contribution to knowledge in this area. Early attempts at developing such approaches have been undertaken,92 and these efforts should be encouraged. Accurate estimates of incidence and the exploration of possible risk factors or predictors of serious adverse events are likely to enhance clinical decision making for health care providers and help patients to make informed health care decisions.
Study Strengths and Limitations
This study has several strengths and limitations. The strengths included that the search process involved a comprehensive examination of 4 databases by an experienced reference librarian, and the selection of studies was undertaken by 2 independent reviewers using predefined criteria.
The limitations of this study included that our methods only considered cases and not other types of study designs. Because of the inherent nature of case reports and other anecdotal reporting, it is not possible to make inferences regarding cause and effect. Therefore, it is not known whether the serious adverse events in cases identified in this review were caused by lumbopelvic SMT or whether the association between therapy and event was incidental (ie, the result of natural history, protopathic bias, or other source of confounding). As well, we did not search all available databases (eg, Index to the Chiropractic Literature or Manual, Alternative and Natural Therapy Index System); therefore, it is possible that not all cases were identified in our searches.
We identified 77 cases involving a serious adverse event following lumbopelvic SMT. Important information such as SMT description, pre-SMT presentation of the patient, and adverse event details were lacking. Cauda equina syndrome and lumbar disk herniation accounted for most adverse events. In addition, we identified cases of spinal fracture, hematoma, hemorrhagic cyst, and other serious adverse events. Two-thirds of cases had a favorable clinical outcome. Additional high-quality research is needed to better estimate the incidence of adverse events associated with lumbopelvic SMT and to elucidate the relationship between this therapy and the types of adverse events reported in this systematic review.
Funding Sources and Potential Conflicts of Interest
No funding sources or conflicts of interest were reported for this study.
The authors thank the following Murdoch University undergraduate students for their assistance with the acquisition and organization of the articles considered in this study: Giulia Bertamini, Rodrigo Guerrero, Ryan Pope, Kim Prater, Jason Sparks, and Josh Summers.
Please refer to Full-Text
Return to ADVERSE EVENTS