314 Hip Problems In Adults: Classification By Cluster Analysis

S.M.A. Bierma-Zeinstra, A.M. Bohnen, R.M.D. Bernsen, J. Ridderikhoff, J.A.N. Verhaar, A. Prins
Dept. of General Practice. Erasmus University Rotterdam, the Netherlands
No universally accepted classification scheme is available for patients with hip problems in primary care. This study explores whether it is possible to obtain a valid classification scheme of hip problems in middle-aged patients seen in primary care with the method of numerical classification.
Methods: Patients (n=224) aged 50 years or older, consulting the general practitioner for pain in the hip region, and referred for X-ray investigation of the hip, underwent a standardised history taking, physical examination (low back, hip and knee), sonographic and radiographic examination, and blood examination. Patients were grouped into clusters with cluster analysis (Ward method) on the basis of similarity or dissimilarity in symptoms. In this cluster analysis variables from the patient’s history and physical examination of the hip region on the (most) painful side were used (65 variables).
Results: The cluster analysis resulted in a classification with nine different clusters. These clusters could be reproduced in the 10 sub-samples and with the alternative clustering method (complete linkage method). Significant relationships of various radiological and sonographic signs with the separate clusters were found; this was also the case for variables of low-back and knee examination. A group of 20 experts could recognise the symptoms in seven of the nine clusters as known syndromes seen in clinical practice. Seven clusters showed symptoms similar to syndromes described in literature, namely: moderate or severe osteoarthritis of the hip, early stage osteoarthritis of the hip, trochanteric's tendinitis, meralgia paraesthetica, low-back problems, a combination of low-back problems and osteoarthritis of the hip, and knee problems in combination with lateral thigh symptoms. Two clusters were more difficult to identify.
We achieved a symptom-based classification of patients with hip problems in general practice, which was reproducible in sub-samples and showed a relationship with radiological and sonographic signs. Most of the clusters in the classification showed similarity with the various hip disorders described in literature and were as such recognised by experts. However, a classification not leading to a discrimination in prognosis or discrimination in effectiveness of specific treatment has no clinical importance. This should be tested in future research.
314 Hip Problems in Adults: Classification by Cluster Analysis