Contribution to Cluster Analysis in Chronic Obstructive Airway Diseases

Chronic obstructive pulmonary disease (COPD) is a complex, multidimensional and heterogeneous disease. A common objective within this large datasets study is the classification of observations into homogenous groups to identify clinical phenotypes. A retrospective study was conducted on 178 COPD patients in stable state recruited from ambulatory care at University hospital of Liege. In this study, the patients were described by more than 70 mixed sets of variables. The rate of missingness ranged from 0% to 25% and 73% of patients presented at least one missing value. The presesnt study attempts to introduce a new framework for cluster analysis combining multiple imputation and variable reduction. The challenge of missing values was solved by multiple imputation. Factor analysis for mixed data (FAMD) was applied on quantitative and qualitative variables for creating new lower dimensional components. The number of clusters in each imputed dataset was determined using hierarchical clustering, finally K-means was applied for assigning clusters to patients. In the consensus clustering step, final result was achieved by fitting mixed multivariate multinomial model. Two different clusters, which shared similar smoking history were derived. Cluster 1 included men who have received more treatment and have higher symptoms in airway obstruction (n=70) while cluster 2 comprised women who were lower airway and neutrophilic systemic inflammation (n=108). Two clusters had a low rate of bacterial colonization (5%), a low median FeNO value (