Identification of preliminary prognostic indicators for back rehabilitation in patients with nonspecific chronic low back pain : a retrospective cohort study
Faculty of Medicine and Health Sciences
Publication type
Hagerstown, Md ,
Human medicine
Source (journal)
Spine. - Hagerstown, Md
(2016) , p. 1-9
Target language
English (eng)
Full text (Publishers DOI)
University of Antwerp
Study design. Retrospective Cohort Objective. Our aim was to identify prognostic indicators for success after a back rehabilitation program (BR) in patients with nonspecific chronic low back pain (CLBP). Summary of background data. Exercise therapy is recommended for patients with nonspecific CLBP. Consensus on the type of exercises is lacking, largely due to heterogeneity in the studied patient samples. The identification of subgroups through the identification of prognostic indicators is therefore needed. To our knowledge, no specific prognostic indicators for BR are described in the literature. Methods. We retrospectively analyzed the patient files of 49 nonspecific CLBP patients who followed a BR. Patients were selected based on predefined in- and exclusion criteria. All underwent 43 therapy sessions, two times per week. Primary outcome measure and dependent variable was the change in Oswestry Disability Questionnaire (ODI) score. Potential predictive variables were tested for association with the primary outcome and consequently entered in a logistic regression model. Results. In this study, the post hoc calculated power was 91%. Based on the change in ODI scores, 24 patients were considered as therapy success (8 points or 50% improvement on change in ODI score) and 25 as therapy failure. Univariate and multiple regression analysis revealed only one significant prognostic indicator: higher scores on the physical function subscale of the SF36 (PF-SF36) corresponded with high risk of therapy failure (odds ratio of 0.791 (95%CI = 0.662-0.945); sensitivity of 0.79 and specificity of 0.68). Conclusions. Potentially, the preset exercises of the BR in this study design were not appropriate for the identified subgroup. The results of this study should be replicated in a RCT design that conforms to the necessary methodological steps in the identification of prognostic indicators and clinical prediction rules (CPRs).