Literature Reviews - 1

Summary and Perspective of Recent Literature

Stynes, et al. (2016). Classification of patients with low back-related leg pain: a systematic review. BMC Musculoskeletal Disorders; 17:226

Stynes et al undertook this study to better understand the classification of patients with low back related leg pain. This subgroup is associated with increased levels of disability and higher health costs, but has no consistent and clear definition to identify low back-related leg pain (LBLP) due to nerve root involvement. This review looks at the relevant literature that classifies or subgroups populations with LBLP, and how leg pain due to nerve root involvement is described and diagnosed in the various systems.

A lit search identified 13,358 papers that were eligible. The authors also hand–picked from a review of referenced articles that were cited, producing an additional 21 papers. From this, 50 relevant papers were identified that reported on 22 classification systems. Five themes were identified:

CLINICAL FEATURES

This category included six papers. The Quebec Task Force Classification was the only system that had supporting work on validity and generalizability. Of interest is a study by Werneke et al in 2004 that compared the QTFC system to classifying whether leg pain centralized or peripheralised. Centralizaiton or peripheralisation was superior in predicting outcomes and long term work status. There were no reliability studies identified for the QTFC. Kappa value for physicians reviewing case studies was 0.69.

PATHOANATOMY

This category included six classification systems which tried to identify the patients’ source of pain.  Overall, the systems did not fare well due to lack of supporting work on systems validity and generalizability. All the systems studied recognized the lumbar disc as a source of pain, facet joint was included in five of the systems, four included stenosis, leg pain of radicular origin was recognized as a pain producer in all six systems, but nomenclature and consistency of definition varied widely. One study by Petersen et al, showed a low kappa value of 0.31 with chosen reference standard (Imaging, injections or discography).

TREATMENT BASED APPROACH

This category included four treatment-based classification systems. McKenzie had good support for its validity, reliability and generalizability. Peripheralization was a good predictor of outcome and there was conflicting evidence for inter-rater reliability. Overall, there were few studies included on McKenzie assessment and treatment in this systematic review.

Hall et al described five patterns of low back pain including leg pain, (treatment in this system is very loosely based on the limited use of McKenzie principles). Their terminology included ‘painful discs’, 'worn spinal joints’, ‘pinched nerve’ and ‘bony spurs within the spine’. Further study by Hall demonstrated that patients classified according to their system had more pain relief post treatment, resulting in less treatment days and patients were less likely to use medications. The study had several limitations that would significantly damper any positive conclusions drawn (cohort vs randomized controlled trial, significant baseline differences between groups). Reliability was good with a kappa of 0.6.

Albert et al classified patients with sciatica according to whether their pain centralized, peripheralized or did not change. All patients received exercise and advice. Improvements were seen in both peripheralizers and centralizers which the authors state refutes some of the McKenzie model theory. Interestingly, symptoms centralized in over 90% of the patients with MRI confirmed sequestrated of extruded discs. Essentially this is a study exploring the McKenzie system, but in this review was deemed to represent a separate classification.

SCREENING TOOLS/CLINICAL PREDICTION RULES

This category included three papers. The purpose of the system is to identify clinical features that either guide diagnosis or assist in treatment. Fritz et al. identified a subgroup of patients who benefited from unloading (traction) when the peripheralized with extension. Scholz et al. used statistical methods to identify items in a standardized assessment that would differentiate between LBP patients with and without radicular leg pain. Roach et al. used screening test algorithms, based on a questionnaire to place patients into four predetermined “structure based” classifications (disc, spinal stenosis, disc disease with spinal stenosis and benign low back pain).  The reliability of Roach’s system had kappa values ranging from 0.57-0.91. Scholz’s system identified patients with radicular pain with high sensitivity and specificity. However, there was no published work on applicability of the rules to different populations.

PAIN MECHANISMS

Three pain mechanism systems were identified; Smart et al used statistical analysis to identify discriminatory clusters of signs and symptoms related to each of the categories. Smart’s system classifies into:

There was some support for the discriminative validity of the system. Reliability kappa values were substantial at a Kappa of 0.77.

Schafer et al. described a four-group system:

Reliability for the Shafer system was substantial with a Kappa of 0.72. Some doubts were expressed about the construct validity of this system.

Overall, there was a significant lack of consistency of terminology between all of the systems used. Up to 11 different terms were used for nerve root presentations. Findings from the clinical examinations also showed great inconsistencies between systems, from being very specific to being very vague. The validity of the systems was very poor, particularly in content and construct validity. There were very few systems that had any reliability data. The ability to judge the generalizability of the systems was very limited as well. Only three of the 22 systems focused specifically on LBLP patients. Within all of the systems the need for a more common language to define nerve root involvement and clinical criteria to diagnosis it was very evident.

The authors went on to conclude that the QTFC was the best place to identify leg pain due to nerve root involvement. McKenzie was given a brief mention in the discussion, but was dismissed based on the minimal studies included where McKenzie was not deemed superior to other treatments, even though they stated that its validity, reliability and generalizability had been supported in the literature. Interestingly, McKenzie principles were echoed in three of the five themes identified by Stynes et al. Centralisation and Peripheralisation were mentioned on three separate occasions as good predictors of outcome. Overall though, the McKenzie system scored the highest of any system in any category, as seen in the graph below.

Out of a maximum of seven, the McKenzie system scored 5.5 on criteria based on content validity, face validity, feasibility, construct validity, reliability and generalizability. Two independent reviewers scored the systems.

Since the purpose of the study was to identify how leg pain to nerve root involvement is described and diagnosed in various systems, it seems that language and lack of consistency has gotten the better of us. Using such different language, the diagnostic accuracy of identifying such subgroups on a consistent and reliable basis. What was underscored was the need for more consistent language and nomenclature, so that when clinicians are speaking, we, as a group, are all on the same page. As within the McKenzie system, when a clinician talks about a Derangement a clinical picture emerges that allows for a clinical conversation to ensue. What has emerged from this study is the need for a common language so that clinicians can be part of the bigger conversation that helps to diagnose and treat low back related leg pain. In the long run, McKenzie belongs in that conversation and can be at the forefront because its identification of centralization and peripheralisation and essentially the identification of Derangements have such a big impact on outcomes, as well as performing all the legwork for validity, reliability and generalizability.