Evaluation of COPD is based on only 2 parameters,  which may not reflect phenotypic heterogeneity, makes it difficult to evaluate. Our objective is to apply a hierarchical classification to a heterogeneous data, set of 142 Algerian patients with COPD diagnosed according to the GOLD recommendations, in order to group them into more homogeneous subgroups. A principal component analysis (PCA) was performed using 9 variables selected for their relevance and impact on disease prognosis: age, cumulative amount of tobacco consumed, forced expiratory volume in 1 seconde FEV1 (% predicted), exacerbations, nutritional status (BMI), dyspnea (mMRC), quality of life (SF-36), anxiety and depression (HAD scale). Patient classification was performed using cluster analysis based on PCA-transformed data.

142 subjects were analysed, 98% were male, (age, 67 years +/-10), 9%, 48%, 25% and 16% were classified in (GOLD) stages 1, 2, 3 and 4, respectively. PCA showed that 4 indépendant components accounted for 67% of variance.

Cluster analysis applied to these main components has identified three different phénotypes, two of which are clearly opposed, "the Young with predominant Respiratory disease" and "the older with predominant extra Respiratory disease". Subjects with comparable airflow limitation (FEV1) belonged to different phenotypes and had marked differences in age, symptoms, comorbidities.

Cluster analysis is an original statistical method for homogeneous classification of COPD patients and is used to identify global profiles taking into account several parameters whose impact on prognosis is clearly established, such as age and especially co-morbidities.