ABSTRACT
STUDY DESIGN: A prospective methodological interrater reliability study. OBJECTIVES: To calculate the interrater reliability among clinicians newly trained in a classification system for acute low back pain and to determine the level of agreement at key junctures within the classification algorithm. BACKGROUND: The utility of a classification system for patients with low back pain depends on its reliability and generalizability. To be practical, clinicians must be able to apply the system after a reasonable amount of training. Identifying key points in the classification algorithm where disagreement occurs can lead to better operational definitions. METHODS: Four physical therapists read an article and attended a 1-day training session in the classification system. Randomly paired therapists classified patients referred for treatment of acute low back pain and noted decisions at key junctures in the system algorithm. RESULTS: Forty-five patients were classified. Repeated examinations did not increase the patient's pain (P>.05). For 3 out of the 4 therapists, the interrater reliability showed a kappa value of 0.45. The fourth therapist, excluded from the overall analysis, exhibited a bias towards the immobilization classification. Among the 3 therapists, major disagreement occurred with the determination of symmetry with trunk side bending and the effects of repeated movements. CONCLUSIONS: Three out of 4 clinicians newly trained in the system showed moderate reliability. The reliability was slight when the fourth therapist was included. Refinement of the operational definitions and criteria for determining lumbar capsular patterns are needed. One day of training is probably not adequate for all therapists, especially for those biased towards specific low back pain syndromes.
