Empirically derived chronic pain patient subgroups: the utility of multidimensional clustering to identify differential treatment effects

Publisher Website
Google Scholar
Abstract
The current study used multidimensional clustering to delineate empirically subgroups of chronic pain patients and to compare their responses to interdisciplinary pain rehabilitation. A total of 180 chronic pain patients were used as subjects. They were administered the Sickness Impact Profile (SIP), Medical Examination and Diagnostic Information Coding System (MEDICS) and treatment outcome measures including subjective pain intensity, hours standing and walking, medication usage and work status. All subjects then participated in an outpatient interdisciplinary pain rehabilitation program, with 120 being randomly selected and 90 available for follow-up assessment. Multidimensional cluster analyses using SIP and MEDICS data identified 4 replicable subgroups: cluster A — highly dysfunctional with moderate levels of physical pathology; cluster B — moderately dysfunctional with moderate levels of physical pathology; cluster C — highly functional with low levels of physical pathology; and cluster D — highly dysfunctional with low levels of physical pathology. Cluster-A and -D patients showed significantly higher levels of depression, more medication usage, less activity and were less likely to be working at pretreatment. These 2 clusters also showed the largest improvement in subjective pain intensity, medication usage, activity level, and return to work post-treatment. Patients in cluster B exhibited the least amount of improvement across outcome measures and, unlike the other 3 clusters, failed to show any significant improvement in work status at post-treatment. Cluster differences were not primarily a function of age, sex, pain intensity, pain location, pain duration, or depression. It was concluded that useful subgroups of chronic pain patients could be reliably identified through multidimensional clustering. More importantly, these subgroups exhibited differential response to treatment that could be used to guide treatment application and maximize efficiency and effectiveness.