Differential features of chronic cough according to etiology and the simple decision tree for predicting causes

Abstract

Finding etiology of chronic cough is an essential part of treatment. Although guidelines include many laboratory tests for diagnosis, these are not possible in many primary care centers. We aimed to identify the characteristics and the differences associated with its cause to develop a clinical prediction model. Adult subjects with chronic cough who completed both Korean version of the Leicester Cough Questionnaire (K-LCQ) and COugh Assessment Test (COAT) were enrolled. Clinical characteristics of each etiology were compared using features included in questionnaires. Decision tree models were built to classify the causes. A total of 246 subjects were included for analysis. Subjects with asthma including cough variant asthma (CVA) suffered from more severe cough in physical and psychological domains. Subjects with eosinophilic bronchitis (EB) presented less severe cough in physical domain. Those with gastro-esophageal reflux disease (GERD) displayed less severe cough in all 3 domains. In logistic regression, voice hoarseness was an independent feature of upper airway cough syndrome (UACS), whereas female sex, tiredness, and hypersensitivity to irritants were predictors of asthma/CVA; less hoarseness was a significant feature of EB, and feeling fed-up and hoarseness were less common characteristics of GERD. The decision tree was built to classify the causes and the accuracy was relatively high for both K-LCQ and COAT, except for UACS. Voice hoarseness, degree of tiredness, hypersensitivity to irritants and feeling fed-up are important features in determining the etiologies. The decision tree may further assists classifying the causes of chronic cough.