The Benefits Outweigh the Risks for Patients Undergoing Chiropractic Care for Neck Pain: A Prospective, Multicenter, Cohort Study
Sidney M. Rubinstein, DC, MS, Charlotte Leboeuf-Yde, DC, MPH, PhD,
Dirk L. Knol, PhD, Tammy E. de Koekkoek, DC, Charles E. Pfeifle, DC,
Maurits W. van Tulder, PhD
Institute for Research in Extramural Medicine (EMGO-Institute),
VU University Medical Center,
Amsterdam, The Netherlands.
OBJECTIVE: This study describes both positive clinical outcomes and adverse events in patients treated for neck pain by a chiropractor.
METHODS: This study was a prospective, multicenter, observational cohort study. Patients with neck pain of any duration who fulfilled the inclusion criteria were recruited in a practice-based study. Data were collected on the patients and from the chiropractors at baseline, the first 3 visits, and at 3 and 12 months. Clinical outcome measures included (1) neck pain in the 24 hours preceding the visit, (2) neck disability, (3) treatment satisfaction, (4) global assessment, and (5) adverse events. Recovery was defined as “completely improved” or “much better” using the global assessment scale. An adverse event was defined as either a new related complaint or a worsening of the presenting or existing complaint by >30% based upon an 11-point numerical rating scale.
RESULTS: In all, 79 chiropractors participated, recruiting 529 subjects, representing 4891 treatment consultations. Follow-up was possible for 90% and 92%, respectively, at 3 and 12 months. Most patients had chronic, recurrent complaints; mild to moderate disability of the neck; and a mild amount of pain at baseline; and two thirds had sought previous care for the presenting complaint in the preceding 6 months. Adverse events after any of the first 3 treatments were reported by 56%, and 13% of the study population reported these events to be severe in intensity. The most common adverse events affected the musculoskeletal system or were pain related, whereas symptoms such as tiredness, dizziness, nausea, or ringing in the ears were uncommon (<8%). Only 5 subjects (1%) reported to be much worse at 12 months. No serious adverse events were recorded during the study period. Of the patients who returned for a fourth visit, approximately half reported to be recovered, whereas approximately two thirds of the cohort were recovered at 3 and 12 months.
CONCLUSION: Adverse events may be common, but are rarely severe in intensity. Most of the patients report recovery, particularly in the long term. Therefore, the benefits of chiropractic care for neck pain seem to outweigh the potential risks.
From the Full-Text Article:
In contrast to clinical trials of prescription medication, researchers in the area of conservative care for musculoskeletal complaints have focused their attention on treatment effectiveness and, to a much lesser degree, on adverse events. This study, consisting of patients treated in a wide variety of chiropractic practices and settings, describes both positive and negative, and short- and long-term clinical outcomes for a relatively large study population with neck pain.
Although adverse events have been described in previous clinical trials of treatment effectiveness with cervical spine manipulation, [4, 34] studies such as these provide limited information on these types of events because of their small sample sizes. Earlier studies on adverse events following spinal manipulation have focused on describing types and patterns of adverse events of the entire spine, [12-14, 29] but did not describe positive outcomes. To our knowledge, only one other study has examined both positive and negative clinical outcomes in patients with neck pain undergoing chiropractic treatments ; however, our study has a larger sample size. In addition, given the study design chosen, it was possible to examine a large number of treatments delivered by a diverse group of chiropractors and was not specifically aimed at the effect of spinal manipulation alone. Therefore, these findings might be more generalizable to clinical practice than those obtained in a single-center, controlled trial setting.
In short, there are 2 major findings. Firstly, in relation to “risks,” despite the fact that more than half of the study population experienced an adverse event, only 1% (5 subjects) of a cohort who had undergone 4891 treatment consultations reported to be much worse at the end of the study period; and there were no serious neurologic complications reported within the study period. Although the number of patients with an intense adverse event seems high, none of these patients were worse or much worse at the end of the study period; therefore, these adverse events should in no way be misconstrued as a measure or indication of harm or be confused with (the lack of) perceived recovery. In addition, only 2 subjects reported to be much worse at 3 months, when most patients are likely to have completed or discontinued care.
Secondly, regarding “benefits,” although many of the subjects had chronic, recurrent neck pain and had undergone prior care for this complaint, many patients experienced benefit from the treatment (based upon diminished pain and disability, the percentage of patients recovered and percentage satisfied with care). Furthermore, many responded relatively quickly to treatment (48% were recovered at the fourth visit); and a significant proportion of patients continued to improve up to 3 months (65%). It is, however, difficult to compare these findings to other studies, especially regarding the rate of recovery and involving other forms of therapy, because both the inclusion criteria and outcome measures may differ. The most similar study is a multicenter study of persistent low back pain treated by chiropractors.  Although the pattern of recovery was different, the percentage of patients who became worse was similarly low in both studies.
The results of this study also confirm earlier work that suggests that adverse events are most prevalent at the beginning of treatment and diminish thereafter in frequency. [13-14] This should have clinical consequences for the practitioner, who might choose to modify his/her treatment approach or limit himself/herself to certain interventions at the start of treatment when the patient is more likely to have a reaction. Another important finding was that some of the same symptoms that are often viewed as a consequence of treatment, such as headache, nausea, dizziness, tiredness, or depression, were present in many subjects at baseline. Furthermore, more than one fifth noted that the symptom was severe in intensity at baseline. This underlies the fact that without a proper reference, there is the real potential to erroneously ascribe previously unreported symptoms to the treatment. In fact, according to our study, many of these concomitant symptoms decreased following the first treatment and continued to improve following the second and third treatments. However, the prevalence of these symptoms returned to their baseline values at 3 and 12 months, suggesting either a short-term positive effect of the treatment or placebo effect.
There are some limitations to this study, however. Firstly, in relation to the data collection, the questionnaires used have not been previously validated, although they were modeled after previous side effects studies. [13-14] Furthermore, given the method of data collection in the clinics (ie, close-ended, self-reported questionnaires), the possibility of response bias cannot be ruled out, meaning it is possible that prompting a patient about the presence or absence of a symptom via the questionnaires might have led to overreporting.
Secondly, the lack of a control group means that it cannot be determined whether the observed adverse events or recovery is a response to the treatment or the result of natural history. Although a control group is obviously desirable, studies such as this one are best designed to describe patterns and changes over time, to investigate the relationship between prognostic factors and outcomes, and to identify subgroups most likely to respond to manipulation for investigation in future clinical trials. Additional reports are forthcoming from this data set.
Thirdly, although this was a prospective study, there is also the potential for recall errors because patients were required to remember and report something about a reaction that took place at prior visit(s). However, almost all (90%) of the second visits had taken place shortly following the first visit, whereas most (79%) of the fourth visits took place within a month of the first visit, so this error is likely to be minimal.
Fourthly, because a convenience sampling of chiropractors was used to collect data, it is possible that more cautious and conservative chiropractors participated. However, a comparison of the practice characteristics of the participating chiropractors with the results of a recent study conducted in the Netherlands,  and with a relatively recent European study,  suggests that participants in this study were sociodemographically similar to their nonparticipating colleagues.
Fifthly, it is possible that those patients deemed by the practitioners likely to have a favorable outcome were more readily recruited. Analysis of recruitment in a sample of 5 of the participating practices revealed that, on average, 78% of the eligible patients were recruited; therefore, recruitment bias was likely to be minimal. Furthermore, practices that saw the highest number of new patients during the recruitment period recruited the fewest eligible patients; therefore, it seems more likely that failure to include individual patients was the result of time constraints in the practices.
Lastly, imaging of the cervical spine was only performed when necessary, at the discretion of the chiropractor. In the Netherlands, few chiropractors have their own radiograph apparatus; and few refer for imaging. Consequently, it is possible that underlying pathology might have been missed by the clinician. However, only one patient was found to have serious pathology; and she was identified at the beginning of the study (based upon history and physical examination) and was excluded from participation. Furthermore, we had a high follow-up rate at the end of the study and no other cases were identified during this period, so this is unlikely to have influenced our results.
Implications for Clinical Practice
Patients respond quickly to care, with the most dramatic improvement occurring in the first 3 treatments. After 3 months, a small percentage will have recurrent symptoms, whereas some will continue to improve; however, most of the patients remain stable. Clinicians should be aware that extended treatment programs might have limited added value when patients do not demonstrate some reasonable improvement by the fourth visit.
Implications for Research
Many symptoms resembling an adverse event were present in nearly all the subjects at baseline and diminished in frequency in the population during the first 3 months. This demonstrates the need to record baseline status for concomitant symptoms to avoid erroneously ascribing their incidence to treatment.
Despite the fact that adverse events following treatment are common, and in some cases severe in intensity, this study shows that the benefits of chiropractic care for neck pain seem to outweigh the potential risks.