Research Review by Dr. Michael Haneline©


June 2008

Study Title:

Predictors for immediate and global responses to chiropractic manipulation of the cervical spine


Thiel H & Bolton J

Authors’ Affiliations:
Anglo-European College of Chiropractic, Bournemouth, United Kingdom

Publication Information:

Journal of Manipulative and Physiological Therapeutics 2008; 31:172-183.


The underlying cause of nonspecific neck pain, like the name implies, is generally unknown. There are subgroups of patients with this condition, however, that may respond to manipulation differently. If those patients who are most likely to respond to manipulation can be identified prior the commencement of care, clinical outcomes can potentially be improved. This can be accomplished by recognizing which subgroup a given patient belongs to, and the likelihood for successful treatment in that subgroup. Clinical prediction rules have been developed for manipulation of the neck and lower back by identifying these subgroups, and the variables that were used to classify the patients.

Knowing the potential for a good outcome in a particular patient would be important information for a clinician to have. Equally important would be knowledge of which patients are more likely to have a bad outcome following manipulation. Treatment could then be modified in an effort to avoid an adverse event.

The purpose of this study was to develop clinical prediction rules that could be used to identify patients who would be expected to benefit from cervical spine manipulation (CSM), as well as those who would be expected to get worse. Thus, a prospective cohort study that involved 377 chiropractors and 19,722 patients was conducted to observe patient and clinical variables and their degree of association with good outcomes and/or adverse effects following CSM.

British chiropractors collected data from their patients in a practice-based research setting using standardized forms that dealt with certain details about treatment and outcomes. Patients were not required to have a specific condition for inclusion in the study, but they were required to have received at least one CSM.

Predictor variables:
Treating chiropractors gathered information on:
  • patients' gender, age, and presenting symptoms;
  • the number and types of manipulative procedures;
  • the areas of the cervical spine treated; and
  • whether the patient had received CSM within the previous 4 weeks.
Outcome variables:
Details were recorded regarding improvement or worsening of presenting symptoms immediately following the initial treatment session and global improvement at their next visit up to 7 days later. The following descriptors were used by patients to characterize their condition in global terms:
  1. “much better and a noticeable change that has made a real difference”
  2. “slightly better but no noticeable or worthwhile difference”
  3. “no change or worse”
Descriptors 2 and 3 were collapsed into a “no global improvement” group for the analysis.

The data analysis involved the use of X2 and odds ratio (OR) statistics to identify the influence of the various predictor variables on the outcomes. Positive likelihood ratios (PLRs) were also calculated for the prediction models based on the number of symptoms present, which were then used to determine how much each prediction model could raise the probability of obtaining the outcome.

Pertinent results of this study:
  1. 31.9% (377/1183) of the target population of chiropractors participated in the study
  2. 19,722 patients were included, involving 28,807 treatment sessions where CSM was performed
  3. data were gathered on 13,873 follow-up visits recording global improvement
  4. pain or stiffness in the neck, shoulder or arm were the most common initial symptoms
  5. 69.7% (20,083/28,807) of patients reported immediate improvement in presenting symptoms following CSM
  6. 14 variables were identified as being significant predictors of immediate improvement, although only 6 remained after stepwise multiple logistic regression analysis:
    • “neck pain”
    • “shoulder, arm pain”
    • “reduced neck, shoulder, arm movement, stiffness”
    • “headache”
    • “upper, mid back pain”
    • “none or one presenting symptom”
  7. The presence of any 4 of these 6 predictors raised the probability for immediate improvement following CSM from 70% to nearly 95%
  8. 4.4% (1269/28,807) of the treatment sessions were associated with worsening of presenting symptoms
  9. 18 variables emerged as significant predictors of immediate worsening, although only 6 remained after stepwise multiple logistic regression analysis:
    • “neck pain”
    • “shoulder, arm pain”
    • “headache”
    • “numbness, tingling upper limbs”
    • “upper, mid back pain”
    • “fainting, dizziness, light-headedness”
  10. The presence of any 4 of these 6 predictors raised the post-treatment probability for immediate worsening following CSM from 4.4% to about 12%
  11. Global improvement was recorded in 56.0% (7775/13,873) of follow-up treatment visits; however, only “neck pain” and the use of the “toggle” method of CSM significantly predicted global improvement
  12. The presence of either of these predictors raised the probability of reporting global improvement only very slightly, from 56% to 60%
Limitations to the study:
  • data may have been inaccurate, since it was collected in routine clinical practices
  • reporting bias by the chiropractors and patients
  • selection bias
  • patients may have been treated concurrently by other health care providers
  • chiropractors may not have recruited patients consecutively, as instructed
  • the chiropractors were aware of the study’s purpose

Conclusions & Practical Application:

This study reported that about 70% patients in the overall group responded to CSM and that this rate improved in the presence of certain predictor variables. The strongest predictor for immediate improvement after CSM was the presenting symptom of “stiffness.” Several other presenting symptoms, including “neck pain,” “shoulder, arm pain,” “headache,” and “upper, mid back pain” were also significant predictors for immediate improvement.

The overall 70% immediate improvement rate that was reported could be improved to 85% in the presence of at least 2 of the identified predictors and it could be raised to 95% if 4 predictor variables were present. For some reason, however, the probability dropped to 60% if 5 of the predictor variables were present. This is possibly related to the observation that multiple symptoms in a patient are typically related to an unfavorable prognosis, and there might be a critical number of presenting symptoms or a combination thereof that would lower the probability for improvement.

“Fainting, dizziness, light-headedness” and “numbness, tingling upper limbs” were the strongest predictors for immediate worsening following CSM. 4.4% of the overall group experienced immediate worsening, which increased to 12% when 4 predictor variables were present.

None of the predictor variables for global improvement following CSM were very meaningful, even though “neck pain” and the use of the “toggle” method of CSM significantly predicted global improvement. This is because the resulting PLR was only about 1, which does not change the probability of the condition being present from pretest to post-test.

CSM practitioners may want to consider that patients with certain presenting symptoms outlined in this study may be more likely to respond favorably to CSM, while those with other presenting symptoms may tend to respond unfavorably. Further research is required to clarify and replicate the findings in this study.