Respiratory Research (Aug 2020)

Cluster analysis in 975 patients with current cough identifies a phenotype with several cough triggers, many background disorders, and low quality of life

  • Heikki O. Koskela,
  • Tuomas A. Selander,
  • Anne M. Lätti

DOI
https://doi.org/10.1186/s12931-020-01485-y
Journal volume & issue
Vol. 21, no. 1
pp. 1 – 9

Abstract

Read online

Abstract Background Recognition of disorder phenotypes may help to estimate prognosis and to guide the clinical management. Current cough management guidelines classify patients according to the duration of the cough episode. However, this classification is not based on phenotype analyses. The present study aimed to identify cough phenotypes by clustering. Methods An email survey among employed, working-age subjects identified 975 patients with current cough. All filled in a comprehensive 80-item questionnaire including the Leicester Cough Questionnaire. Phenotypes were identified utilizing K-means partitional clustering. A subgroup filled in a follow-up questionnaire 12 months later to investigate the possible differences in the prognosis between the phenotypes. Results Two clusters were found. The cluster A included 608 patients (62.4% of the population) and the cluster B 367 patients (37.6%). The three most important variables to separate the clusters were the number of the triggers of cough (mean 2.63 (SD 2.22) vs. 6.95 (2.30), respectively, p < 0.001), the number of the cough background disorders (chronic rhinosinusitis, current asthma, gastroesophageal reflux disease, 0.29 (0.50) vs. 1.28 (0.75), respectively, p < 0.001), and the Leicester Cough Questionnaire physical domain (5.33 (0.76) vs. 4.25 (0.84), respectively, p < 0.001). There were significant interrelationships between these three variables (each p < 0.001). Duration of the episode was not among the most important variables to separate the clusters. At 12 months, 27.0% of the patients of the cluster A and 46.1% of the patients of the cluster B suffered from cough that had continued without interruptions from the first survey (p < 0.001). Conclusions Two cough phenotypes could be identified. Cluster A represents phenotype A, which includes the majority of patients and has a tendency to heal by itself. The authors propose that cluster B represents phenotype TBQ (Triggers, Background disorders, Quality of life impairment). Given the poor prognosis of this phenotype, it urges a prompt and comprehensive clinical evaluation regardless of the duration of the cough episode.

Keywords