Classification schemes take narrower approach to low back pain

Studies support matching interventions to subgroups of patients who are most likely to benefit

Published in the July 2007 issue of BioMechanics

By: Jordana Bieze Foster

Call it the paradox of back pain management. The vast majority of low back pain cases remain inexplicable, with as many as 80% of patients exhibiting symptoms that cannot be linked to identifiable pathology based on imaging examinations or other diagnostic tests. Yet without knowing the exact cause of the symptoms, practitioners are increasingly able to reduce pain and improve function in patients with low back pain by matching treatments to particular symptom characteristics.

For decades, clinicians and researchers alike have been frustrated by two defining elements of the low back pain population. First is that a large percentage have symptoms of unknown etiology. Second is the formidable degree of heterogeneity within the population, such that a single intervention applied to a group of nonspecific low back pain patients might be effective in some patients but not in the group overall.

A small contingent of researchers, however, saw this diversity as less of a frustration than an opportunity. The effectiveness of various interventions could be maximized, they reasoned, by classifying nonspecific low back pain patients into subgroups based on symptom characteristics and matching each subgroup to a different intervention. Although researchers continue to tease out the best means of defining subgroups and matching them to the most appropriate intervention, the results of randomized trials suggest that this classification-based approach to chronic low back pain therapy is more effective than therapy based on existing clinical practice guidelines.

"The initial treatment phase really represents a window of opportunity in terms of long-term disability," said Gerard Brennan, PT, PhD, director of clinical quality and outcomes at Intermountain Health Care in Salt Lake City, UT. "If we can get the right treatments to the right patients, maybe we can cut down on that disability."

Back pain matchmakers

Brennan and colleagues classified 123 patients with nonspecific low back pain of less than 90 days duration into one of three subgroups, and randomized them to receive treatment that either was or was not matched to their classification. This was based on the presence of symptoms distal to the buttocks, and on whether or not the symptoms centralize or move proximally toward the lumbar spine, in conjunction with flexion or extension movement. (Those who centralize with extension movement, for example, were assigned to an exercise program that emphasizes strengthening of the extensor muscles.)

All patients were assigned to receive either specific exercise (extension- or flexion-oriented), manipulation, or stabilization exercise twice a week for four weeks. Those who demonstrated significant improvement, as measured by the Oswestry Disability Index (ODI) score, progressed from the initial classification-based treatment to a multimodal exercise program.

At four weeks, mean improvement in ODI score was 6.5 points greater in patients whose treatments were matched to their classification than in unmatched patients; at one year, the improvement in the matched patients was 8.3 points greater. Both between-group differences were statistically significant. Also notable, however, was that the researchers found no significant interactions between treatment and time or between classification and time, suggesting that the interaction of all three factors is what determined the positive outcome in the matched group.

"This outcome depended on matching the right treatment to the right subgroup over time," said Brennan, who presented the results in February at the American Physical Therapy Association's annual Combined Sections Meeting.

The findings, which were also published in the March 15, 2006, issue of Spine, are consistent with those of a 2003 study from the University of Pittsburgh in which classification-based management outperformed conventional care in 78 patients with occupational low back pain of less than three weeks' duration. After four weeks, patients randomized to classification-based treatment had significantly greater improvement in ODI scores and the physical component of the SF-36 questionnaire than patients treated according to evidence-based guidelines established by the Agency for Health Care Policy and Research. Matched patients were also significantly more likely to have returned to work at the four-week follow-up.

The statistically significant between-group differences were not maintained at one year, but researchers did document a trend toward significantly reduced ODI scores in the matched group, as described in the study published in the July 1, 2003, issue of Spine.

New rules

As encouraging as the results of the two randomized trials are, researchers believe that further refinements to the classification process can improve outcomes even more. Two clinical prediction rules (CPRs), one for stabilization exercise and one for manipulation-based treatment, have been published since Brennan and colleagues initiated their study and were not incorporated into that protocol. A multicenter study group headed by researchers at the University of Maryland in Baltimore developed a CPR for stabilization exercise based on the treatment responses of 54 patients with nonradicular low back pain. The patients received eight weeks of training focused on the multifidus/erector spinae, transverse abdominus, and oblique abdominal muscles.

The investigators found that a positive outcome was four times more likely in patients who met three of the following criteria: age less than 40, average straight-leg-raise range of motion of more than 91 degrees , aberrant movements through sagittal-plane lumbar ROM, and a positive prone instability test.

Perhaps more important, they also found that a negative outcome was 18.8 times more likely in patients who met three of these four criteria: negative prone instability test, absence of aberrant movements through sagittal-plane ROM, absence of hypermobility of the lumbar spine, and a score of nine or higher on the Fear Avoidance Beliefs Questionnaire (FABQ). (For more on the variable of fear in back pain, see sidebar, page 52.) The CPR was published in the September 2005 issue of the Archives of Physical Medicine & Rehabilitation.

More recently, researchers from the U.S. Army-Baylor University in San Antonio performed a detailed arthrokinematic analysis of 20 patients with low back pain who met at least two of the CPR's four positive predictive criteria and did not meet two or more of its negative predictive criteria. Using digital fluoroscopic video to assess the classified subjects and a group of 20 healthy controls, the researchers identified 10 arthrokinematic variables that differentiated the two groups. The resulting kinematic model was found to be most accurate when four or more of the variables were present.

Although patient classification based on digital fluoroscopic assessment is not clinically practical, the findings offer some insight into the mechanisms that may be contributing to low back pain in this subgroup of patients: a combination of altered segmental structural integrity, segmental stiffness, and altered neuromuscular control during lumbar spine movement. The results were published in the March issue of Physical Therapy.

More than manipulation

Another multicenter effort, this one headed by the San Antonio group, resulted in the 2002 publication of a CPR for lumbopelvic manipulation. The 71-patient study, published in the December 15, 2002, issue of Spine, found that patients who met four of five predictive criteria were 24 times more likely to experience at least 50% improvement in disability scores within one week of receiving no more than two manipulation sessions that also included ROM exercise. The variables found to be predictive of outcome were symptom duration (less than 16 days), an FABQ subscale score less than 19, lumbar hypomobility, unilateral hip internal ROM of at least 35 degrees , and lack of symptoms distal to the knee.

A subsequent retrospective analysis of data for 141 patients, led by researchers from the University of Utah in Salt Lake City, found that meeting both of the two most clinically practical criteria-symptom duration and no symptoms distal to the knee-resulted in a positive likelihood ratio of 7.2 (using the same definition of a positive outcome as in the original CPR study). That study was published in the online journal BioMed Central Family Practice on July 14, 2005.

Later studies involving a manipulation classification group, including Brennan's study, have combined the two sessions of manipulation with a three-week strengthening and stabilization exercise program.

"Clinical evidence continues to show that it's the combination of manipulation and an active exercise program that's best, and better than manipulation alone," said MAJ John D. Childs, PhD, PT, assistant professor and director of research at U.S. Army-Baylor University. "There seems to be improvement immediately postmanipulation, which seems to suggest that the manipulation is acting to improve muscle function. It may be like using ctrl-alt-delete on a computer. If the muscles don't seem to be working correctly and you provide a quick thrust through the pelvis, it may 'reset' the muscles and allow the patient to complete a more active exercise program."

Although the manipulation CPR was developed using a lumbopelvic thrust technique, researchers are currently investigating whether alternative manual therapy techniques might be equally effective. A multicenter trial, headed by investigators from Franklin Pierce College in Concord, NH, will randomize 240 patients who satisfy the manipulation CPR to receive lumbopelvic thrust manipulation, lumbar neutral gap manipulation, or a nonthrust mobilization technique for two visits, followed by three weeks of active exercise.

Although that study is about a year from completion, a 60-patient short-term study from the San Antonio group suggests that the effects of the two manipulation techniques are not significantly different. In the study, presented at the APTA's Combined Sections Meeting, patients who satisfied the manipulation CPR were randomized to receive either lumbopelvic or lumbar neutral gap manipulation; patients in both groups experienced statistically significant reductions in pain and disability at 48 hours, but neither technique was associated with significantly more benefit than the other.

"People will say manipulation is very specific, but evidence shows the benefit is more general," Childs said. "It may be that the key is just to get the patient going, not necessarily the specific technique that you use."

Changing definitions

Efforts are also being made to refine the definition of patients who are most likely to benefit from specific exercise. Researchers from San Antonio are finding that additional classification criteria may be necessary to more effectively target patients for whom specific exercise is most appropriate. In 48 patients with low back pain and distal symptoms that centralized with extension movement, those randomized to receive an eight-session extension-oriented treatment approach (EOTA) experienced more significantly improved changes in disability than those randomized to perform stabilization exercises at one week, one month, and six months from baseline. However, no significant between-group differences were seen for change in pain or healthcare resource utilization, suggesting that a narrowing of the criteria for EOTA may be appropriate.

"The subgroup definition may not be optimal," said CPT David Browder, DPT, a staff physical therapist at Lackland Air Force Base in San Antonio, who presented the findings at the Combined Sections Meeting. The study is in press with Physical Therapy.

The researchers also found that the five EOTA patients with a history of lumbar surgery (more than six months from the time of the study) experienced changes in disability that were significantly less than EOTA patients with no history of surgery.

"Patients with a history of spinal surgery may benefit more from a different intervention," Browder said.

The effectiveness of centralization-based classification schemes may also depend on a patient's age. In a 355-patient study in which the mean age was 59, researchers from CentraState Medical Center in Freehold, NJ, found that the prevalence of centralization was lower than expected in patients over the age of 64. Comorbidities and mobility limitations in older patients may preclude centralization of symptoms, but this does not necessarily mean they will not benefit from an intervention based on directional preference.

"Future studies for the elderly may look at different factors," said Mark Werneke, MSPT, a clinician in the spine rehabilitation department at CentraState, who presented his group's findings at the Combined Sections Meeting. "It may not be centralization, maybe change in pain intensity based on positioning or mobilization, or directional preference in absence of centralization."

Jordana Bieze Foster is a freelance writer based in Massachusetts and a former editor of BioMechanics.

Back Pain And The Fear Factor

Fear itself may not be the only thing that patients with back pain have to fear, but research suggests fear avoidance is a variable that influences outcomes-and one that biomechanically focused practitioners may be likely to overlook.

In a study of 355 patients with neck and back pain, investigators from CentraState Medical Center in Freehold, NJ, found that elevated score on the Fear Avoidance Beliefs Questionnaire (FABQ) at the time of discharge was one of two variables (lack of pain centralization being the other) that were associated with lower levels of functional status and higher levels of pain intensity at discharge.

This may not surprise practitioners familiar with classification-based treatment, as the clinical prediction rules for both spinal manipulation and stabilization exercise include FABQ score as a criterion. But Mark Werneke, MSPT, a clinician in the spine rehabilitation department at CentraState, who presented his group's findings at the APTA's Combined Sections Meeting, suggests that many clinicians infrequently consider or manage patients' fear as part of their practice.

"We're trained in the biomedical model, which assumes a linear relationship between pain and pathoanatomical concepts, in contrast to a biopsychosocial model, which accepts the biomedical model but expands it," Werneke said. "You need to look at things besides the physical exam."

Werneke and others have had success treating patients' fear using a cognitive-behavioral approach developed by Dutch researchers, which emphasizes problem solving and graded-activity training that progresses from least-feared activities to most-feared. If a patient's fear remains high despite this intervention, then other psychosocial variables that can affect pain and function should be considered, Werneke said. These include occupational factors, pending litigation, cultural issues, and potential financial gain. -JBF Copyright 2008 Jordana Foster – 24 Kirkland Dr, Stow, MA – Email: – Fax: (815) 346-5239