Sub-grouping patients with non-specific low back pain based on cluster analysis of discriminatory clinical items.

OBJECTIVE To identify potential subgroups amongst patients with non-specific low back pain based on a consensus list of potentially discriminatory examination items. DESIGN Exploratory study. PARTICIPANTS A convenience sample of 106 patients with non-specific low back pain (43 males, 63 females, mean age 36 years, standard deviation 15.9 years) and 7 physiotherapists. METHODS Based on 3 focus groups and a two-round Delphi involving 23 health professionals and a random stratified sample of 150 physiotherapists, respectively, a comprehensive examination list comprising the most "discriminatory" items was compiled. Following reliability analysis, the most reliable clinical items were assessed with a sample of patients with non-specific low back pain. K-means cluster analysis was conducted for 2-, 3- and 4-cluster options to explore for meaningful homogenous subgroups. RESULTS The most clinically meaningful cluster was a two-subgroup option, comprising a small group (n = 24) with more severe clinical presentation (i.e. more widespread pain, functional and sleeping problems, other symptoms, increased investigations undertaken, more severe clinical signs, etc.) and a larger less dysfunctional group (n = 80). CONCLUSION A number of potentially discriminatory clinical items were identified by health professionals and sub-classified, based on a sample of patients with non-specific low back pain, into two subgroups. However, further work is needed to validate this classification process.