We thank D. Hahn for his interest in our paper 1. We agree that following the natural history of individuals defined by these clusters may help define whether they represent distinct phenotypes, and we plan to do this in a follow-up cohort study of the Wellington Respiratory Survey. We agree that the fourth cluster identified by the “Diana” method may represent a phenotype of airway inflammation characterised by sputum production without marked elevation of exhaled nitric oxide and without marked airflow variability, as has been previously described 2, 3. However, we are cautious in extrapolating the results of the cluster analysis too far. The analysis was of a cross-sectional study and, although, for purposes of understanding, we presented the results in terms of the mean values of individual contributing variables, the cluster technique uses a multivariable distance measure in nine dimensions; thus, the associations between the contributing variables may be more complex than this. Apart from determining the natural history of these clusters in our cohort study, other forms of external validation, such as developing a clinical allocation rule and determining the response of different phenotypes to therapy, will also contribute understanding to the different phenotypes of chronic obstructive pulmonary disease.
Statement of interest
None declared.
- © ERS Journals Ltd