Which Functional Outcomes Can be Measured in Low Back Pain Trials and Therapies?: A Prospective 2-Year Factor-, Cluster-, and Reliability-Multicenter Analysis on 42 Variables in 1049 Individuals

Abstract

Study design: Prospective 2-year factor-, cluster-, and reliability-multicenter analysis.

Objective: To provide evidence on the uniqueness and usefulness of a set of potential relevant functional outcomes.

Summary of background data: A very high number of functional outcomes that can be utilized as variables in low back pain (LBP) trials exists.

Methods: Participants (n = 1049) with and without current LBP were included. At 7 visits (baseline, 4 wk, 3 mo, 6 mo, 9 mo, 1 yr, and 2 yr), 42 different functional outcomes were assessed. Two exploratory factor analyses (for baseline-values and for changes scores to 3 mo post-baseline) were calculated (maximum likelihood extraction, varimax factor rotation). The strongest factor-loading outcomes were selected for the following hierarchical cluster analyses (average linkage, Euclidean distance measure). For each cluster, time point, and outcome, reliability analyses were subsequently calculated using intraclass correlation coefficients, standard error of measurements and coefficients of variation.

Results: The factor analysis for the cross-sectional values revealed 9 factors with a cumulative variance explanation of 61.7% and 13 unique ones. The change-score factor analysis revealed nine factors with a total variance explanation of 61.8%, seven outcomes were unique. Ten outcomes were important for both cross-sectional and change-score analyses, 11 were most valuable for cross-sectional and four for the change scores. Patients with pain grades 1 to 3 show comparable patterns (cluster 2). Grade 0 (cluster 1) and grade 4 (cluster 3) are unique and cannot be cumulated with other grades. Most biomechanical outcomes were highly reliable and display low measurement errors.

Conclusion: We found 25 potentially meaningful functional outcomes in the context of objective functional measurements (such as trunk range of motion, dynamic and static balance, strength, and muscle fatigue resistance) and body characteristics. The present framework may help to select appropriate functional outcomes and rate effects beyond the known core set of outcomes.Level of Evidence: 1.