• Marije van der Hulst, Miriam M R Vollenbroek-Hutten, Karin G M Groothuis-Oudshoorn, and Hermie J Hermens.
• Roessingh Research and Development, University of Twente, Enschede, The Netherlands. m.vanderhulst@rrd.nl
• Clin J Pain. 2008 Jun 1; 24 (5): 421-30.

Objectives(1) To determine if treatment outcome in chronic low back pain can be predicted by a predefined multivariate prognostic model based on consistent predictors from the literature and (2) to explore the value of potentially prognostic factors further.MethodsData were derived from a randomized controlled trial on the effect of a multidisciplinary rehabilitation program for chronic low back pain compared with usual care. The primary outcome measure was the Roland and Morris Disability Questionnaire and secondary outcomes were the Physical and Mental Component Summary Scales, derived from the Short Form Health Survey. Outcomes were expressed as the differences between baseline and follow-up (8 wk and 6 mo) values. A confirmatory and an exploratory model were defined. Baseline predictors included in the confirmatory model were pain intensity, work status, and Multidimensional Pain Inventory subgroup membership. The exploratory model included sick leave, compensation, depression, and fear-avoidance beliefs. Statistical analysis was performed using multiple linear regression analysis.ResultsOne hundred and sixty-three patients participated in the study. More pain was prognostic for more improvement in the rehabilitation group. No value was found for work status or the Multidimensional Pain Inventory subgroups. For the exploratory model, more depression and fear-avoidance beliefs predicted more improvement after rehabilitation. The explained variance ranged from 18.5% to 43.8% depending on the length of follow-up evaluation, the treatment group, and the outcome variable of interest.DiscussionThe results of this study do not support the construction of a clinical prediction model. Future confirmative studies of homogeneous rehabilitation treatments and outcome measures are needed to shed more light on relevant prognostic factors.

18 keywords...

hide…