New classification for men with lower urinary tract symptoms: cluster analysis using the International Prostate Symptom Score

OBJECTIVE To determine whether men with lower urinary tract symptoms (LUTS) in various combinations could be classified into characteristic symptom groups using cluster analysis. SUBJECTS AND METHODS Secondary analysis of anonymous data from 10 434 patients enrolled in a postmarketing surveillance study of tamsulosin in Japan was performed. Data were prospectively collected through the central register from men diagnosed with LUTS in 1100 medical institutions between October 2004 and March 2005. Those who had received an a alpha 1-adrenoceptor antagonist within 1 week before the initial visit were excluded. The survey items were International Prostate Symptom Score (IPSS), quality of life (QOL) score, and patient characteristics including age, bodyweight, body mass index, and LUTS-related comorbid conditions. Data on IPSS items were used in a hierarchical cluster analysis (Ward's method). RESULTS Of 10 434 men with LUTS, 9910 were included in the analysis after 5% data trimming (524 patients). Five symptom clusters were identified and the symptom types of each cluster were examined. The largest cluster (27%) consisted of patients with multiple severe symptoms, i.e. complaining of six or more symptoms with a mean score >= 2.8. In contrast, the second smallest cluster (13%) consisted of patients with minimal symptoms, i.e. complaining of essentially one or two symptom with a mean score <= 2.1. The other three clusters were labelled based on their dominant symptoms. The clusters were weak stream (27%), storage symptoms (21%) and voiding symptoms (12%). The storage symptoms group was older, and had more comorbidities. The distribution of the QOL score was different among the clusters, and the percentage of patients who were very dissatisfied was highest in the multiple severe symptoms group and lowest in the minimal symptoms group. CONCLUSIONS Cluster analysis using the IPSS showed that men with LUTS can be classified into five characteristic symptom groups. A new approach to symptom-based classification may be useful to elucidate the pathology of male LUTS and individualize the therapeutic strategy for affected patients, but further studies are needed.