Biobehavioral predictors of outcomes following intervention for low back pain.
Clinical relevance. Although highly prevalent, low back pain typically has no clear patho-anatomic etiology. The lack of correlation between physical impairment and loss of function associated with back pain has led many to consider other sources such as biobehavioral factors in its etiology. Biobehavioral factors (BBF) can be described as a set of psychological, environmental, and psychophysiological processes that attenuate or exacerbate the discrepancy between pathology, pain, and impairment. Identifying the role of BBF in predicting patient outcome could enhance selection of efficient treatment options and improve the accuracy of prognostic judgments. Objective. The purpose of this study was to describe the relationship of BBF to reports of function and disability, as measured by the Roland Morris Score (RM). Methods. An analysis of data from an intervention study including 239 subjects was performed to assess the correlation of pre-intervention measures of BBF to measures of RM scores obtained prior to intervention, at discharge, and 30 and 180 days following discharge. A 30-item instrument represented BBF. Bivariate correlations were obtained between BBF and the RM scores. An exploratory factor analysis was performed to determine the presence of subscales in the 30-item instrument. Bivariate correlations were then performed comparing the means of subscales to follow-up RM scores. Results. When compared to pre-intervention RM, 25 of the 30 items were significantly correlated (r = -.45 to .36), at discharge 22 of 30 were correlated (r = -.33 to .38), at 30 days 12 of 30 were correlated (r = -.22 to .35) and at 180 day follow-up, 9 of 30 items were correlated (r = -.351 to .291). Factor analysis revealed 7 initial subscales: activity-limitation (AL), health and well being (HW), fear of pain (FP), willingness to exercise (WE), pain intensity (PI), psychological distress (PD), and job intensity (JI). At discharge, scores from all pre-intervention subscales except WE and JI were significantly, but weakly, correlated with the RM score (r = .19--.39). At 30 days all pre-intervention subscales except AL and JI were significantly, but weakly, correlated with RM (15--.36). At 180 days all pre-intervention subscales except AL, FP, PD and JI were significantly, but weakly, correlated with RM (.19--.31). Conclusion. Although several items were correlated with outcomes, no individual items within the questionnaire were strong predictors of RM. The Health and Well-Being and Pain Intensity subscales were significant, but weal to moderate predictors of outcome (r = .21--32 and .31--39), respectively. In our sample patient population, outcome as represented by RM scores was not strongly correlated to BBF.