Classification of shoulder complaints in general practice by means of cluster analysis

Abstract

Objective: To determine if a classification of shoulder complaints in general practice can be made with a cluster analysis of variables of medical history and physical examination. Method: One hundred one patients with shoulder complaints were examined upon inclusion (week 0) and after 2 weeks. Eleven variables of the medical histories and 19 variables of the physical examinations were used for the analysis. Results: The analyses of week 0 and week 2 reveal three stable clusters: one cluster with almost half of the patients who hardly had limitations in the range of scapulohumeral mobility (ROM), one cluster with a small number of patients with a short history of complaints and a limitation of scapulohumeral mobility in all directions (7 and 6, respectively), and a third cluster containing the rest of the patients, with 30% to 50% limitation in ROM. The degree of limitation in ROM decreased after 2 weeks. Comparison of the patients in the clusters of week 0 and week 2 revealed a shift of patients within the clusters, although the number of patients in the clusters remained almost constant. Conclusion: Only three stable clusters could be identified, which questions the suitability of more detailed classifications, such as described by Cyriax or the Dutch Guidelines for Shoulder Complaints, for the diagnosis of patients with shoulder complaints in general practice.