PT - JOURNAL ARTICLE AU - Janwillem W.H. Kocks AU - Mark Weatherall AU - Esther I. Metting AU - Roland A. Riemersma AU - James Fingleton AU - Thys van der Molen AU - Richard Beasley TI - Phenotyping airways disease by cluster analysis in primary care: 6 distinct clusters identified DP - 2014 Sep 01 TA - European Respiratory Journal PG - 1977 VI - 44 IP - Suppl 58 4099 - http://erj.ersjournals.com/content/44/Suppl_58/1977.short 4100 - http://erj.ersjournals.com/content/44/Suppl_58/1977.full SO - Eur Respir J2014 Sep 01; 44 AB - Current obstructive airways disease classification does not sufficiently reflect disease patterns. Cluster analysis is one of the promising approaches to develop a new taxonomy. The majority of current phenotyping studies focus on severe asthma or COPD.Aim To identify phenotypes in a broad spectrum of obstructive airways disease in a primary care population.Methods 952/9225 cases with full data on 13 variables reflecting physiological,lung function,laboratory and questionnaire data from a structured primary care Asthma/COPD service were used to identify clusters using hierarchical clustering. Optimal number of clusters was established by silhouette stats and clinical judgement. Decision rules developed were used to allocate the remaining.Results The optimal number of clusters was 6. 5424 cases had sufficient data to be allocated by the allocation rules based, in order of importance,on smoke exposure,FEV1%pred,ACQ,Age of onset,hyperactivity,bronchitis score,CCQ functional status,FEV1/FVC ratio,CCQ mental status. The clusters identified in order of increasing smoke exposure are:A-Overweight,non smoking,normal lung function,uncertain diagnosis (15% of patients);B-Younger onset allergic asthma(39%);C-Younger onset allergic asthmatic smokers with bronchitis(15%);D-Adult onset,high symptomatic asthma(6%); E-Smoking Non allergic asthma/COPD overlap with obesity and eosinophilia(9%);and F-late onset smoking COPD(17%).Conclusion Six distinct groups could be identified in this primary care population using cluster analysis.Implications These six groups may represent distinct phenotypes if differences in disease progression and treatment responses are shown.Funding: UMCG, co-sponsor: GSK NL.