Cluster analysis classification of SF-36 profiles for patients with spinal pain.

STUDY DESIGN

A k-means cluster analysis of patients with spinal and radicular pain based on the SF-36 Health Survey scales.

OBJECTIVE

The aim was to determine whether spine patients fall into clusters according to self-reported health status as measured by the SF-36 and to determine if clustering is similar across four common diagnostic categories: herniated disc, spinal stenosis, spondylosis, and chronic pain syndrome.

SUMMARY OF BACKGROUND DATA

Cognitive-behavioral classifications of chronic pain patients have previously identified three patient groups described as Dysfunctional, Interpersonally Distressed, and Minimizers/Adaptive Copers. The purpose of these classifications is to facilitate and direct treatment based not only on biomedical diagnosis but also on emotional, social, and behavioral diagnoses. This type of analysis has not been done on the quality-of-life scores of patients with specific spinal diagnoses.

METHODS

Health status data were reviewed from the initial visits of 15,748 spine patients in the National Spine Network database. Based on the eight scales of the SF-36, k-means cluster analysis divided the National Spine Network population into distinct clusters of similar patients. Clustering was performed separately for each clinical diagnosis group.

RESULTS

In all four diagnostic categories, cluster analysis classified patients into three groups. Group 1 had fairly high (relative to the entire sample) scores on all scales and was labeled "Highly Functional." Group 2 had low measures on physical variables but comparatively high scores on the mental scales. These were labeled "Emotional Adapters." Group 3 had low scores on all scales. These patients were labeled "Dysfunctional." Although patients in each diagnostic category fell into one of the three groups, the proportion of patients within each group was quite different among chronic pain patients as compared to the other three diagnostic groups. For example, 29% of herniated disc patients were in the Highly Functional group, whereas only 14% of patients in chronic pain were categorized as Highly Functional. Thirty-three percent of spondylosis patients were classified as Dysfunctional compared with 51% of chronic pain patients.

CONCLUSIONS

Patients with spinal pain fall into three groups according to their profile of scores on the SF-36 Health Survey. It is proposed that such empirical groupings can guide decision-making in selecting the most appropriate therapies.