Research Review by Dr. Shawn Thistle©


July 2007

Study Title:

Subgrouping patients with low back pain: Evolution of a classification approach to physical therapy


Fritz JM, Cleland JA, Childs JD

Publication Information:

Journal of Orthopaedic & Sports Physical Therapy 2007; 37(6): 290-302.


Low back pain (LBP) is one of the most common reasons people seek healthcare from manual medicine providers and family physicians. Despite its prevalence, LBP remains challenging for all health care providers, a persistent "black box" of musculoskeletal medicine. We know there is some type of dysfunction causing pain, but we are not always sure exactly what that is. In fact, the literature indicates that an exact pathological diagnosis can be achieved in only 10% of cases. Combine this with the questionable relevance of some common imaging findings, and it can be very confusing.

Over 1000 randomized clinical trials exist in the literature, investigating conservative and surgical interventions for LBP. To date, this expansive volume of research has been unable to offer conclusive support for most treatment interventions. One logical explanation for this trend relates to study designs incorporating broad inclusion criteria, resulting in heterogeneous patient samples.

It has been advocated that the power of research efforts could be improved by attempting to match subgroups of LBP patients to appropriate treatment interventions they would be most likely to benefit from. This makes sense, as clinical experience dictates that not all patients will respond to one treatment method or approach. This is where a Clinical Prediction Rule (CPR) can help by assisting clinicians to identify clusters of factors (clinical or historical) that can predict patient response to a treatment intervention (ex. manipulation, stabilization exercise etc.).

A CPR does exist for LBP. A substantial volume of published theoretical and experimental research on this topic has been completed by a group of physical therapists (including the three authors of this paper). The existing CPR can now be used for patients with LBP to guide clinical decision making. This paper reviews the development and content of this CPR.

Overview of the Classification System

The original classification system was first published in 1995 (Delitto A et al. Phys Ther 1995; 75: 470-485), and was intended for patients with acute LBP or an acute exacerbation of pre-existing LBP causing substantial pain and limitations in daily activities. Basically, after screening patients for medical red flags, the model used information gathered from the history and examination to place each patient into one of four treatment categories: manipulation, stabilization, specific exercise (flexion, extension, and lateral shift patterns), and traction. These factors vary slightly for each group, but some overlap exists, which makes clinical application more efficient.

The treatment groups were initially proposed based on expert opinion, and existing literature on common treatment approaches for LBP. In recent years, numerous studies have added to, and adapted the content of the initial system, resulting in the current CPR (the paper discusses this evolution, but the focus here will be on the practical application based on current evidence). To illustrate the utility of this approach, the research to date on this CPR has demonstrated (references below):
  • the CPR to have high inter-rater reliability (Fritz & George 2000)
  • the CPR to be reliable (Fritz et al. 2006)
  • using the CPR improves clinical outcomes (Brennan et al. 2006)
  • this CPR outperforms existing Clinical Practice Guidelines (Fritz et al. 2003)
  • the CPR can effectively identify those likely to benefit from manipulation (Flynn et al. 2002)
  • there are potential risks of NOT performing manipulation on appropriate patients identified using the CPR (Childs JD et al. 2006)
Content of the Clinical Prediction Rule

The following are revised and updated signs and symptoms associated with each of the four treatment groups. A small description of each category follows:

(1) Manipulation
  • no symptoms distal to the knee
  • recent onset of symptoms (<16 days)
  • low FABQ (Fear Avoidance Beliefs Questionnaire) score (<19)
  • hypomobility of the lumbar spine
  • hip internal rotation > 35° for at least one hip
When 4 of 5 of these factors are present, patients are highly likely to improve with manipulation (positive likelihood ratio of 24!), while the presence of 2 or fewer factors are predictive of failure with manipulation (negative likelihood ratio of 0.009). Factors against this approach include symptoms below the knee, increasing episode frequency (suggests need for stabilization), peripheralization with motion testing (suggests need for specific exercise), and no pain with spring testing.

(2) Stabilization
  • younger age (<40)
  • greater general flexibility (ex. postpartum, SLR > 91°)
  • ”instability catch” (when returning to neutral from a flexed position), or other aberrant movements during lumbar flexion/extension
  • positive prone instability test
  • for patients who are postpartum: 1) positive posterior pelvic pain provocation, active SLR, and Trendelenberg tests, and 2) pain with palpation of the dorsal sacral ligament or pubic symphysis
Not including the last factor listed above, the presence of 3 of the other 4 factors suggests that a patient would benefit from stabilization rehab (Stu McGill's "Big 3" exercises, etc.). Factors against this approach include a SLR discrepancy of > 10° between legs, and a FABQ score < 9. The strength of these associations is not as strong as those for manipulation, so further research is required specific to this category.

(3) Specific Exercise

Extension (more common)
  • symptoms distal to the buttock
  • symptoms centralize with lumbar extension
  • symptoms peripheralize with lumbar flexion
  • directional preference for extension
  • older age (>50)
  • directional preference for flexion
  • imaging evidence of lumbar spinal stenosis
Lateral Shift (least common)
  • visible frontal plane deviation of the shoulders relative to the pelvis
  • directional preference for lateral translation movements of the pelvis
For this category - centralization is defined as abolishment of pain or paresthesia with a specific position of the spine (flexion or extension normally), and directional preference is defined as a situation where movement in a certain direction improves pain and ROM limitation, while movement in the opposite direction causes signs and symptoms to worsen. This approach is consistent with what most know as the Mechanical Diagnosis and Therapy (McKenzie) method of assessment and treatment. Factors against this approach include having only LBP (no distal symptoms) and having no change in symptoms with any type of movement.

(4) Traction
  • signs and symptoms of nerve root compression
  • no movement centralizes symptoms
There continues to be a lack of evidence supporting the use of traction for patients with LBP. It is within reason however, that a small subset of LBP patients may benefit from this therapy. More research is required to identify specific factors to support this treatment. For now, the focus of the CPR is generally on the first three categories.

Conclusions & Practical Application:

Currently, there is a lack of clarity regarding the treatment of LBP. Systematic reviews and meta-analyses continually paint a dreary picture for LBP treatment, citing equivocal results which require further study. This clinical prediction rule has been well developed and researched, and provides a simple model that can assist in classification of LBP patients. By providing appropriate treatment, we can increase our likelihood of success. Further, using this model may guide future research by providing homogenous patient samples, allowing a more “fair” analysis of certain treatments.

RRS encourages all readers to become familiar with this line of research, as well as the treatment modalities contained within in. Additional references about the CPR are listed below.

Additional References

  • Brennan GP et al. Identifying subgroups of patients with acute/subacute “nonspecific” low back pain: Results of a randomized clinical trial. Spine 2006; 31(6): 623-631.
  • Childs JD, Flynn TW & Fritz JM. A perspective for considering the risks and benefits of spinal manipulation in patients with low back pain. Manual Therapy 2006; 11: 316-320.
  • Cook C et al. Subjective and objective descriptors of clinical lumbar spine instability: A Delphi study. Manual Therapy 2006; 11: 11-21.
  • Flynn T et al. A clinical prediction rule for classifying patients with low back pain who demonstrate short-term improvement with spinal manipulation. Spine 2002; 27(24): 2835-2843.
  • Fritz JM et al. The use of a classification approach to identify subgroups of patients with acute low back pain: Interrater reliability and short-term treatment outcomes. Spine 2000; 25(1): 106-114.
  • Fritz JM et al. Comparison of classification-based physical therapy with therapy based on clinical practice guidelines for patients with acute low back pain. Spine 2003; 28(13): 1363-1372.
  • Fritz JM et al. An examination of the reliability of a classification algorithm for subgrouping patients with low back pain. Spine 2006; 31(1): 77-82.