Hierarchical clustering in evaluating inflammatory upper airway phenotypes; increased symptoms in adults with allergic multimorbidity

Inflammatory upper airway diseases cause significant morbidity. They include phenotypes with different treatment; allergic or non-allergic rhinitis (AR, nAR), and chronic rhinosinusitis with or without nasal polyps (CRSwNP, CRSsNP). In clinical practice, these phenotypes are often difficult to distinguish and may overlap. To evaluate if hierarchical clustering can be used to distinguish these phenotypes based on the presence of nasal polyps, off-seasonal allergic symptoms, and self-reported background characteristics - e.g. atopic dermatitis (AD); and to further analyse the obtained clusters. We studied a random sample of 74 CRS (chronic rhinosinusitis) patients, and a control group of 80 subjects without CRS with/without AR (tertiary hospitals, 2006-2012). All underwent interview and nasal examination, and filled a questionnaire. Variables regarding demographics, off-seasonal symptoms, and clinical findings were collected. Hierarchical clustering was performed, the obtained clusters were cross-tabulated and analysed. Four clusters were identified; 1: "Severe symptoms and CRSwNP" (n = 29), 2: "Asymptomatic AR and controls" (n = 39), 3: "Moderate symptoms and CRSsNP" (n = 36), and 4: "Symptomatic and AD" (n = 50). Cluster 1 had most sinonasal symptoms, cluster 3 had a high prevalence of facial pain. The presence of AR did not distinguish CRS groups. Of the AR subjects, 51 % belonged to cluster 4, where AR with off-seasonal airway symptoms and AD predominated. Hierarchical clustering can be used to distinguish inflammatory upper airway disease phenotypes. The AR phenotype was subdivided by the presence of AD. Adult AR+ AD patients could benefit from active clinical care of the upper airways also off-season.