Research Review by Dr. Shawn Thistle©


Nov. 2006

Study Title:

Predictors for the immediate responders to cervical manipulation in patients with neck pain


Tseng Y et al.

Publication Information:

Manual Therapy 2006; 11: 306-315.


Neck pain is a very common condition that has a large impact on society. Neck pain can be associated with trauma, postural strain, headaches, and numerous other complaints. Although not as prevalent as low back pain, patients with neck pain endure not only pain, but functional limitation, time off work, and in some cases, long-term disability.

Physical or manual therapy is becoming increasingly popular for treating neck pain. Spinal manipulation (or spinal manipulative therapy - SMT) is one of the most commonly utilized modalities, and is generally thought to be beneficial for treating this condition. However, the current state of the evidence for SMT in the treatment of neck pain is equivocal. Pain relief following SMT has been documented in many studies, while others have questioned the benefit.

When studying manual therapy in general, and SMT specifically, inherent difficulties in research design and implementation are ample and cumbersome to say the least. It is not as simple as the traditional pharmaceutical model of using drug A to treat condition B with complete patient and doctor blinding. Manual interventions involve a personal interaction and are difficult to standardize.

Further, when studying spinal pain, identifying and attaining a homogenous patient group is next to impossible. Herein lie the challenges in studying spinal manipulation.

Recently, a group of researchers (Anthony DeLitto, Julie Fritz, Timothy Flynn etc.) has developed and validated a clinical prediction rule for low back pain. Essentially, they have identified a simple cluster of findings that when positive, indicate that a certain patient is likely to respond favourably to a certain intervention (spinal manipulation is one of the possible treatment modalities in this model).

This is the basis of most clinical prediction rules, which have been created for other conditions. If you are not familiar with this ongoing low back pain project, I highly recommend reviewing the relevant studies. This is the type of research that is practical and helpful in a clinical environment, without requiring drastic changes to your practice habits.

This study, conducted in Taiwan, is (to my knowledge) the first attempt to develop a similar clinical prediction rule for the cervical spine. Specifically, this study attempted to identify predictors of immediate response to neck manipulation in patients with neck pain.

One hundred patients with neck pain referred by their physician to physical therapy for treatment of neck pain were recruited into this study.

Patients had one of the following diagnoses:
  • cervical spondylosis with or without radiculopathy
  • herniated disc of the cervical spine
  • Myofascial pain syndrome
  • cervicogenic headache
Exclusion criteria included:
  • a diagnosis of vertebrobasilar insufficiency
  • progressive neurological deficit
  • severe osteoporosis
  • history of neck fracture or surgery
  • psychological disorders
  • systemic disease
  • or other contraindications to spinal manipulation
Each patient was examined by an independent assessor who was unaware of the treatment group the subjects belonged to.

Examination procedures included the following to identify hypomobile segments to be manipulated:
  • cervical ROM
  • cervical compression and distraction
  • cervical segmental side-gliding (essentially motion palpation/joint challenge)
Treatment included one session involving standardized cervical manipulation performed by one of two experienced clinicians (one a chiropractor/physiotherapist with 13 years experience and the other a physiotherapist with postgraduate training in manipulation with 6 years experience). The manipulation described in the study would be familiar to most chiropractors as a rotary cervical. Each identified hypomobile segment was manipulated with only one thrust.

After treatment, each patient was evaluated on the criteria listed below. In order to be deemed "successful", only one of the following had to be reported positive (explained with each item):
  1. Pain Rating - 11-point numeric scale - a positive result was an improvement in pre-test score of > 50%
  2. Perceived Improvement - 15-point Likert scale from worst grade of -7 to best grade of +7 - a positive result was at least a +4 on the scale
  3. Satisfaction Level - ranging from "very satisfied" to "very unsatisfied" - in order to be assessed as positive, the response had to be "very satisfied" (5 options total)
Patients who did not meet any of the above criteria were considered as non-responders to treatment. Significance level for each factor above was set to p < 0.10 - intentionally high so as not to miss any potential factors.

Pertinent Results:

  • 34 males and 66 females, mean age 46 participated in this study
  • according to the criteria for judging a successful treatment - 60% of were classified as responders (no significant difference was noted between the success rates of the two treating clinicians)
  • six factors demonstrated a significant relation to a successful treatment outcome in the regression analysis: initial NDI score of < 11.5, bilateral pain pattern, not performing sedentary work > 5 hours/day, feeling better while moving the neck, without feeling worse while extending the neck, diagnosis of spondylosis without radiculopathy
  • no subjects were positive for all six factors, but 40 of the 60 responders had 3 or 4 of those factors
  • conversely, of subjects with 2 or fewer of the six factors, 28 of 40 were in the non-responder group

Conclusions & Practical Application:

I chose to review this study because I feel it represents a small but important step in a long crucial process. Developing a clinical prediction rule for neck pain is an important step for manual medicine research, and should be addressed as such. With a "smallish" sample size of 100 subjects, this study suggests six factors that (at least in this study) appear helpful in predicting which patients will respond favourably to SMT for neck pain.

It should be remembered that only one treatment was given in this study. This made for a simpler study design, but may limit the external validity of these findings to everyday practice. These findings cannot be applied to any multimodal treatment plan that many of us employ.

This is something that will be altered in future studies on this topic if the evidence base is to be strengthened.

Another potential problem with this study is the subjective nature of the scoring for a "positive" on the three outcome criteria. The authors acknowledge this weakness, and defend their choice by stating that dramatic improvements were set as the levels at which a "positive" is recorded. Lastly (and obviously), the follow-up period was not long enough to elucidate any long-lasting effects of SMT. All measures were re-done directly after the treatment in this study.

This is a time for important feedback and information however, as the authors state that patient satisfaction with treatment is one of the main factors that dictates adherence with a treatment program.

One of the factors associated with positive outcome in this study may be surprising to those in practice. Having no pain with cervical extension was associated with a positive treatment response. Cervical extension, and extension with rotation (+/- compression) are often used to identify facet involvement and hence direct the application of SMT. This doesn't necessarily mean they won't respond to manipulation, rather more than one treatment may be needed. Another factor that should be pointed out is the rather low NDI score of < 11.5 - this indicates a relatively minor level of pain or discomfort.

In this study, patients having 3 of the 6 factors listed above were significantly more likely to respond well immediately after SMT for neck pain. Further study is required to further develop and validate this model, but a workable framework has been described in this study.