The sensitivity of the system falls from 91 to 86% when 6-h validation was performed. BIRRING et al. [1] state that the reduction in sensitivity is most likely due to a more robust measure of true cough frequency used for the 2-h recordings. An alternative explanation is that the 2-h validation is poorly representative of the performance of the device over longer time periods. The data is presented for 24-h recordings on 50 subjects. So where are the data on sensitivity and specificity validated over a full 24 h? Previous publications have manually counted cough over 24-h periods in larger numbers of patients [2]. Furthermore, the title of the paper describes the Leicester Cough Monitor as an automated cough detection system. In fact, operator input is required to classify sounds that might have been wrongly identified as cough events in the first phase. Therefore, the system would be more accurately described as semi-automated. This is now the third publication on the Leicester Cough Monitor without validation beyond 6 h, yet the authors invite others to contact them to use the monitor based on this data. We ask S.S. Birring and co-workers to state candidly whether this device has or has not been validated over 24 h. If it has, can they please provide the missing data on 24-h validation so that we can assess the true accuracy of the Leicester Cough Monitor? K. McGuinness*, A. Morice # , A. Woodcock and J. Smith* *Respiratory Research Group, University of Manchester, Manchester, # Cardiovascular and Respiratory Studies, University of Hull, Hull, UK. STATEMENT OF INTEREST Statements for interest for all authors can be found at www.erj.ersjournals.com/misc/statements.shtml REFERENCES 1 Birring SS, Fleming T, Matos S, Raj AA, Evans DH, Pavord ID. The Leicester Cough Monitor: preliminary validation of an automated cough detection system in chronic cough. Eur Respir J 2008; 31: 1013–1018. 2 Decalmer S, Webster D, Kelsall A, McGuinness K, Woodcock A, Smith JA. Chronic cough: how do cough reflex sensitivity and subjective assessments correlate with objective cough counts during ambulatory monitoring? Thorax 2007; 62: 329–334. DOI: 10.1183/09031936.00052008 From the authors: K. MCGUINNESS and colleagues request further details regard- ing the validation of the Leicester Cough Monitor (LCM) over 24 h. This was not possible during our study because of limitations in battery life [1]. Therefore, we compared auto- mated and manual cough counts during the day, at a time when the frequency of cough and other sounds was highest and the demands on the system were most stringent. Since our paper has been published [1], we are very grateful to Professor A. Monce for providing us with the opportunity to independently validate 24-h recordings obtained from his institution. Our findings are consistent with earlier reports suggesting that the LCM is an accurate measure of cough frequency [1–4]. Seven patients with stable unexplained chronic cough (six females, mean age 51 yrs) were randomly recruited from Castle Hill Hospital (Hull, UK). Cough frequency measured by automated LCM analysis was compared with that measured manually by one observer. All patients underwent 24-h continuous ambulatory sound recording in their own environ- ment with a Sony Minidisc (MZ-RH1) recorder (Sony Corporation, Tokyo, Japan) and an external cardioid stereo microphone (SP-CMC-2A; The Sound Professionals, Hainesport, NJ, USA) [5]. An experienced independent observer from Castle Hill Hospital, blinded to the automated cough analysis, measured cough counts manually by audio and visual inspec- tion. Unfiltered sound recordings were also analysed using the automated LCM by an observer blinded to the manual cough counts [1]. Coughs were identified as single events whether they occurred as isolated events or in a cluster. All patients gave written consent and the local ethics committee approved the study. There was no significant difference between manual and LCM automated cough counts (mean¡SEM 23¡7 versus 24¡6 coughs?patient -1 ?h -1 ; mean difference51; p50.5) and there was a highly significant correlation between manual and automated coughs?patient -1 ?h -1 (r50.97, p,0.001). The intra-class correlation coefficient of agreement between manual and auto- mated cough counts was also highly significant (0.98, p,0.001; fig. 1). We were unable to determine sensitivity analysis or to carry out an hour-by-hour analysis because of the difficulty in synchronising time stamping of detected events between the two cough recording systems. However, the close agreement between the numbers of coughs detected in the 24-h recordings across a wide range of cough frequencies increases confidence that the LCM has high sensitivity and specificity for detecting coughs and provides further validation of our system against independently acquired and counted recording. Automated cough detection systems are potentially an important advance in an important and grossly under                                   FIGURE 1. Bland-Altman plot comparing cough frequency measured by an automated cough monitor (Leicester Cough Monitor) and manually by an observer. ???????????: mean difference between manual and automated cough counts, –––: 95% limits of agreement. 530 VOLUME 32 NUMBER 2 EUROPEAN RESPIRATORY JOURNAL