Research Review by Dr. Shawn Thistle©


Feb. 2007

Study Title:

The validity of manual examination in assessing patients with neck pain


King W, Lau P, Lees R & Bogduk N

Publication Information:

The Spine Journal 2007; 7: 22-26.


For many years, it has been assumed by practicing manual therapists and the institutions that educate us that we can accurately diagnose symptomatic joints in the neck via manual examination. This conviction however, is based on a very limited body of research, namely one study published nearly twenty years ago which only had twenty subjects (Jull G, Bogduk N, and Marland A. Med J Aust 1988).

This study claimed that manual examination had 100% sensitivity and 100% specificity for diagnosing painful facet joints (also referred to here as zygapophyseal joints or Z-joints). Skeptical? You should be, as this study has never been duplicated.

That being said, there is a volume of published literature outlining the role of cervical Z-joints in neck pain conditions, including specific zones of potential pain referral from individual cervical levels. Further, the use of radiologically guided diagnostic blocks for identifying painful Z-joints has been validated, making this technique the gold standard for diagnosing this condition. Existing literature has also identified C2-3 and C5-6 as the most common levels for Z-joint pain.

Physical examination of the cervical spine typically involves active and passive ranges of motion, palpation, muscle testing, and assessment of intersegmental motion (or joint play). Previous research on these assessment methods has revealed concerning shortcomings, including poor interobserver agreement and low reliability. Numerous studies have examined these two factors, but in the absence of validation studies, a reliable test that two different examiners can agree on is not terribly meaningful.

This study aimed to verify the results of the previously mentioned study, and evaluate the validity of manual examination for identifying painful Z-joints in the neck. A consecutive series of 173 patients presenting to a private community practice with neck pain were used in the study. Patients had to have neck pain for at least three months, be willing to undergo diagnostic nerve blocks, and had no response to conservative therapy. Patients were excluded if they had neurological signs of radiculopathy, previous neck surgery, or were unwilling to undergo diagnostic nerve blocks.

Each subject provided a history and underwent a clinical examination by the senior author (I assume that would be Nikolai Bogduk) with particular attention paid to "segmental signs" described previously in Jull's paper (NOTE: the examination procedures were not described in this paper and the original paper is not available in electronic format - see discussion below). If physical signs under study were present in patterns consistent with Z-joint impairment at one or more levels the subject was deemed "clinically positive". Other cervical levels with no signs were deemed "clinically negative".

Previous imaging studies were reviewed after the examination to identify any contraindications for diagnostic blocking, but not to identify possible pain-generating structures. Those patients identified as "clinically positive" were offered cervical Z-joint blocks, while those identified as "clinically negative" were not. The diagnostic blocks were then performed by an interventional radiologist in accordance with guidelines established by the International Spine Intervention Society. In order for a joint to be diagnosed as definitively symptomatic (block positive), the diagnostic block had to provide complete relief of pain (short-acting when 2% lignocaine was used, and long-lasting when 0.5% bupinvacaine was used). Any other response to the diagnostic block was considered "block negative". Levels identified by manual examination were then compared to diagnostic block results.

Pertinent Results:

  • both clinical examination and diagnostic blocks revealed C2-3 and C5-6 as the most common symptomatic levels (this agrees with previous literature)
  • of the 173 neck pain patients in this study, 133 (77%) had positive responses to the diagnostic blocks (for a prevalence of 77% for facet joint pain - keep in mind this is a neck pain patient sample)
  • for the C2-3 level - manual examination had a sensitivity of 0.88, specificity of 0.39, and likelihood ratio of 1.4
  • for the C5-6 level - manual examination had a sensitivity of 0.89, specificity of 0.50, and likelihood ratio of 1.8
  • pooling the results from all levels did not significantly change the overall resultant statistics: sensitivity 0.89, specificity 0.47, and likelihood ratio of 1.7

Conclusions & Practical Application:

This study could not replicate the high values from Jull's original paper, indicating that manual examination for facet joint pain in the neck has high sensitivity, but low specificity and poor likelihood ratios. This combination indicates that manual examination is not a valid method of diagnosing cervical facet joint pain when compared to a criterion standard (diagnostic blocks).

In light of these results it should be remembered that high sensitivity does not imply validity for a test. When an examiner is accustomed to diagnosing certain levels as positive, a high sensitivity is nearly ensured. When this is the case, the true value of a test rests in its specificity. The low specificity found in this study compromises the value of the manual examination, which is also reflected by the low likelihood ratio.

The main drawback of this study is that the manual testing procedures were not described. The authors merely referred to the previous study, which as mentioned was published nearly twenty years ago. I could not find the original paper, but having read most of Jull and Bogduk's work over the years I assume the examination included joint play assessment in different cervical motions (rotation and lateral bending in particular) and perhaps a variation of motion palpation.

The examiners were presumably looking for lack of relative motion, or patient-reported pain. This is just my guess, but as we know the available techniques for assessing joint pain in the neck are rather limited.

The results of this study should not force us to abandon manual examination of the neck. It is only one study, which itself requires replication on a larger sample with specifically described clinical testing. The following points should all be kept in mind as well. First, in our offices, we often have only our hands to work with as diagnostic tools, so it is imperative that we continue studying our methods. Diagnostic blocking is invasive, and not readily available in most locations.

Second, just because we may not be able to accurately identify the facet as a pain-generator does not mean current manual treatments are not effective for neck pain patients. It may just mean that our treatments (soft tissue, mobilization/manipulation, traction, rehab etc.) are affecting different anatomical structures or having generalized effects.

Finally, we all believe we have some degree of competence in identifying painful joints in the cervical spine. The results of this study may indicate that our tests (joint play, motion palpation etc.) are identifying a diagnosable entity in the cervical spine that has yet to be determined.