Introduction and objectives Cough frequency is known to correlate well with perceived cough severity and is commonly used as an endpoint to evaluate the efficacy of new anti-tussive agents. However, little is known about the distribution of cough and its impact on cough severity. The Weibull distribution is a probability distribution model often used to predict time between events, as applied in COPD exacerbations. The shape parameter describes if coughs are more or less likely to occur over time. We investigated if cough temporal distribution is modulated by an effective anti-tussive agent (morphine) and how this distribution influences perceived cough severity.

Methods Cough recordings, made using the VITALOJAK™ cough monitoring system were analysed from 22 chronic cough patients (18 female, mean age 61.7 years, mean cough duration 14 years) who had participated in a previously reported randomised controlled trial of low dose morphine sulphate. Coughs were identified and the cough intervals recorded and fit to an exponential (random) and Weibull distribution. The goodness of fit was checked with the Akaike information criterion (AIC). The shape parameter was calculated and using generalised estimating equations compared for morphine versus placebo. Relationships between cough severity visual analogue scale (VAS), cough frequency and the shape parameter were also explored.

Results The AIC indicated that cough distribution better fitted the Weibull equation than the exponential in 38/41 (93%) recordings. The Weibull shape parameter was significantly reduced by morphine compared with placebo (geometric mean 0.418 versus 0.548, p=0.04), suggesting more coughs occurred early, with a fall in the rate of subsequent coughs over time with morphine. Furthermore, cough severity VAS was significantly predicted by a combination of cough frequency (p<0.001), the shape parameter (p=0.044) and the interaction between these variables (p=0.005).

Conclusions In chronic cough patients, low dose morphine, in addition to reducing cough frequency may also reduce the temporal clustering of cough events. This may be important as temporal clustering appears to be a component of perceived cough severity, independent of cough frequency.