Cough-Based Alert System Predicts COPD Exacerbations

Photo Credit: Cinefootage Visuals

A system that monitored patients’ coughs while they slept showed promising performance for predicting COPD exacerbations, with a lead time of 4 days or more.

A passive cough-monitoring system for patients with COPD showed good performance in exacerbation detection with a low burden to patients, according to a real-world study published in COPD: Journal of Chronic Obstructive Pulmonary Disease.

“Since questionnaire data is often hampered by subjectiveness (due to interpretation of questions and or use of response scales) and by adherence issues (questionnaire fatigue), there is an ongoing drive for automated monitors providing objective measures collected in a patient-friendly way,” wrote Alyn Morice, MD, and colleagues.

The prospective, double-blind study focused on a cohort of 38 patients with stable COPD. Patients were a median 72 years old (range, 57-84 years), with a median forced expiratory volume in 1 second percent (FEV-1%) predicted of 43% (range, 20%-106%).

The device is a stationary, nighttime cough monitor with a microphone. Patients placed it in their bedrooms, and the device monitored cough while they slept.

“It requires no wearables or indeed any other patient action except complying to not turning off the system and, obviously, (regularly) sleeping in [a] designated location,” the researchers said. “Preserving privacy has been a key element of the system from its very start.”

Patients also filled out a daily symptom questionnaire for 12 weeks. If they did not experience an exacerbation during that period, patients were asked to continue being monitored for another 12 weeks.

High Specificity, Few False Alerts

The patients experienced 41 exacerbations over a total of 3,981 days.

“From the data, it is estimated that around 30% of the exacerbations did not give rise to an increase in cough. Though based on a single observation, the alert mechanism flagged 59% of the exacerbations. For the cases with alerts preceding the onset, the associated lead time was 4 days or more,” the researchers found.

The system uses daily cough frequency as a noninvasive marker of disease instability. Over 2,694 days of potential alerts, only 7 additional alerts were generated, translating to a false alert (FA) rate of 1 per 385 days. When numbers were visually validated, this dropped to 1 FA every 674 days. The researchers noted that questionnaire-based systems have comparatively higher false alert rates.

The cough-based approach is limited by its inability to detect exacerbations not associated with increased cough frequency. In addition, the study authors noted that the system requires a 14-day baseline period, which delays its ability to detect early exacerbations. Exacerbation reporting was patient initiated and could have been influenced by the daily symptom questionnaires.

Cough-Based System’s Advantages

Questionnaire-based monitoring requires patients’ active participation, which presents a challenge for long-term adherence. The cough-based alert system, on the other hand, prioritizes low burden and high positive predictive value over maximal sensitivity.

The researchers said their study also demonstrated the cough-based system’s functionality in a real-world, uncontrolled environment. No dedicated absence detection or alert escalation mechanisms were used, yet the system maintained its performance despite occasional data gaps likely due to connectivity or power issues.

“Though based on a single variable only, the cough-based alert system captured more than half of the exacerbations in a passive, free-living scenario,” Dr. Morice and colleagues concluded. “No adherence issues were reported, and patients confirmed the unobtrusive and hassle-free nature of the approach.”