Effectiveness of a low back pain classification system.

BACKGROUND CONTEXT: One goal of low back pain (LBP) assessment is to direct clinicians to specific subgroups that benefit from particular treatment approaches.
PURPOSE: To compare outcomes in a nonoperative care setting between patients assessed and treated based on a diagnostic system of LBP classification with patients managed without a classification system. STUDY DESIGN/
SETTING: A prospective double-cohort study. PATIENT SAMPLE: Mechanical LBP cases (n=2110) who started a rehabilitation program at 15 clinics across Canada between February 2006 and August 2007. OUTCOME MEASURES: Subjective global pain rating at discharge; change in reported medication usage; Visual Analog Scale (VAS) pain rating from assessment to discharge; change in perceived function from assessment to discharge based on score change from a modified version of the Low Back Outcome Score; total number of treatment days.
METHODS: The two cohorts were a comparison group (n=754) and a classification group (n=1356). The comparison group consisted of consecutive, consenting patients attending treatment at eight clinics that provided generic, traditional, therapy including modalities and exercise and that emphasized reassurance of likely recovery, encouragement to remain active, and avoidance of bed rest. The classification group consisted of consenting patients attending seven clinics where the staff had been trained to use the LBP classification system. Patients were categorized into one of four classifications, each dictating a separate treatment approach.
RESULTS: For those reporting "no pain" posttreatment, odds ratios for those treated according to a Pattern classification ranged from 2 to 10 times the odds of the comparison group (p<.05). For "no medication use" posttreatment, odds ratios for the classification group ranged from 2 to 4 times the odds of the comparison group (p<.01). Odds ratios of a statistical and minimal clinically important difference (30% increase in functional score) for the classification group ranged from approximately 2 to 4 times the odds of the comparison group (p<.01). For VAS Pain Rating (0-10 scale), the odds ratios of a statistically and clinically significant pain reduction (2 points) for the classification group ranged from 30% to 60% higher than odds for the Comparison Group (p<.05). The comparison group had the highest number of treatment days, statistically significantly greater than for each pattern within the classification group (p<.001).
CONCLUSIONS: LBP is a heterogeneous condition and treatment results may significantly improve when clinically relevant syndromes are determined initially to guide treatment. Classifying LBP based on pattern recognition shows promise to help clarify future clinical trials and surgical referrals.