Cough Frequency as a Digital Biomarker

Cough frequency is the count of cough events over a recording period, conventionally expressed as coughs per hour and reported separately for the whole day and for waking hours. It is measured by an ambulatory monitor that records ambient sound continuously and applies an algorithm to identify cough sounds, which removes the need for the person to notice, remember or report their own coughing.

That matters because cough is a symptom people are demonstrably bad at quantifying. Severity questionnaires capture how much cough bothers someone, which is a different quantity from how often they cough, and the two correlate only moderately. Objective counting separates the two, allowing a trial to show that a treatment reduced cough events, reduced the distress they cause, or did one without the other.

The standard approach uses a body worn recorder with a contact or free field microphone worn for twenty four hours. Recorded audio is processed by a cough detection algorithm, and in validation work a proportion of the recording is manually counted by trained listeners as the reference standard. The Leicester Cough Monitor was the first system to demonstrate that this could be done semi automatically with acceptable agreement against human counting, and later systems have built on the same design.

Because the method depends on continuous audio capture, it carries data handling obligations that movement based measures do not. Protocols should state what audio is retained, for how long, and whether speech is discarded at the point of processing. Participants are consenting to a microphone running in their home and workplace, and that should be presented plainly rather than buried in a device description.

The dominant use is in refractory and unexplained chronic cough, where objective twenty four hour cough frequency has served as the primary endpoint in large randomised trials, including the phase three programme for the P2X3 receptor antagonist gefapixant. Meaningful change thresholds for objective cough frequency have been estimated, which gives trials a basis for judging whether an observed reduction matters.

In chronic respiratory disease the application is different. Domiciliary cough monitoring has been studied for predicting exacerbations in COPD, where a rise in night time cough precedes other signs of deterioration. Across both uses, cough frequency is reported alongside a validated cough specific instrument such as the Leicester Cough Questionnaire, because reducing the count without reducing the burden is an incomplete result and regulators expect both perspectives.

Automated cough detection is imperfect and performance varies with background noise, speech, and how the microphone is positioned. Reported agreement figures usually come from controlled validation samples rather than from unrestricted daily life. Cough frequency is also highly variable between days within the same person, which means short recordings give unstable estimates and studies generally need repeated twenty four hour periods.

The measurement is intrusive in a way that step counting is not, and some participants decline or modify their behaviour when a microphone is running. Finally, dedicated hardware is required. No consumer wearable in this library counts coughs, and treating a phone based cough app as equivalent to a validated ambulatory monitor is not defensible.

• Birring SS, et al. Development of a symptom specific health status measure for patients with chronic cough: Leicester Cough Questionnaire (LCQ). Thorax. 2003. pubmed.ncbi.nlm.nih.gov
• Birring SS, et al. The Leicester Cough Monitor: preliminary validation of an automated cough detection system in chronic cough. Eur Respir J. 2008. pubmed.ncbi.nlm.nih.gov
• McGarvey LP, et al. Efficacy and safety of gefapixant, a P2X3 receptor antagonist, in refractory chronic cough and unexplained chronic cough. Lancet. 2022. pubmed.ncbi.nlm.nih.gov
• Schelfhout J, et al. Validation and meaningful change thresholds for an objective cough frequency measurement in chronic cough. Lung. 2022. pubmed.ncbi.nlm.nih.gov
• Crooks MG, et al. Domiciliary cough monitoring for the prediction of COPD exacerbations. Lung. 2021. pubmed.ncbi.nlm.nih.gov