Which Patients with COPD Would Benefit from Cough Monitoring?

Background: Cough monitoring for exacerbation detection is optimally effective if used for the appropriate cohort of chronic obstructive pulmonary disease (COPD) patients, i.e., increased cough during exacerbation and prodrome is a prerequisite enabling (early) detection. Methods: A post hoc analysis of data from a validation study on an alert system for exacerbation detection based on nighttime cough was used to study if patient data were predictive for the increased cough during exacerbation and for cough count distribution. The quantitative effect on the performance of the alert system when using patient stratification was studied as well. Results: Patient data were not predictive for robust cough statistics: neither the nighttime cough count median nor the interquartile range were found to have statistically relevant correlation with the available patient data. Patients with and without increased cough during exacerbation did show differences in their characteristics. Using patient age and symptom questionnaire data, a classifier based on a logistic regression model was parametrised having an accuracy of 85% in predicting presence or absence of increased cough during exacerbation. Using the classifier for patient stratification, the performance of the exacerbation alert system increased with sensitivity going from 59 to 76%. Conclusions: The post hoc analysis suggests that patient data can be used to stratify COPD patients for cough monitoring.