The Leicester Cough Monitor: preliminary validation of an automated cough detection system in chronic cough

Transcription

1 Er Respir J 2008; 31: DOI: / CopyrightßERS Jornals Ltd 2008 The Leicester Cogh Monitor: preliminary validation of an atomated cogh detection system in chronic cogh S.S. Birring*, T. Fleming*, S. Matos #, A.A. Raj ", D.H. Evans # and I.D. Pavord " ABSTRACT: Chronic cogh is a common condition that presents to both primary and secondary care. Assessment and management are hampered by the absence of well-validated otcome measres. The present stdy comprises the validation of the Leicester Cogh Monitor (LCM), an atomated sond-based amblatory cogh monitor. Cogh freqency was measred with the LCM and compared with coghs and other sonds conted manally over 2 h of a 6-h recording by two observers in nine patients with chronic cogh in order to determine the sensitivity and specificity of the LCM. Atomated cogh freqency was also compared with manal conts from one observer in 15 patients with chronic cogh and eight healthy sbjects. All sbjects nderwent 6-h recordings. A sbgrop consisting of six control and five patients with stable chronic cogh nderwent repeat atomated measrements o3 months apart. A frther 50 patients with chronic cogh nderwent 24-h atomated cogh monitoring. The LCM had a sensitivity and specificity of 91 and 99%, respectively, for detecting cogh and a false-positive rate of 2.5 events?h -1. Mean SEM atomated cogh conts?patient?h -1 was 48 9 in patients with chronic cogh and 2 1 in the control grop (mean difference 46 conts?patient?h -1 ; 95% confidence interval (CI) 20 71). The atomated cogh conts were repeatable (intra-sbject SD 11.4 coghs?patient?h -1 ; intra-class correlation coefficient 0.9). The cogh freqency in patients ndergoing 24-h atomated monitoring was 19 coghs?patient?h -1 ; daytime (08:00 22:00 h) cogh freqency was significantly greater than overnight cogh freqency (25 verss 10 coghs?patient?h -1 ; mean difference 15 coghs?patient?h -1, 95% CI 8 22). The Leicester Cogh Monitor is a valid and reliable tool that can be sed to assess 24-h cogh freqency in patients with cogh. It shold be a sefl tool to assess patients with cogh in clinical trials and longitdinal stdies. AFFILIATIONS *Dept of Respiratory Medicine, King s College Hospital, London, # Dept of Medical Physics, Leicester Royal Infirmary, and " Institte for Lng Health, Dept of Respiratory Medicine, Glenfield Hospital, Leicester, UK. CORRESPONDENCE I.D. Pavord Dept of Respiratory Medicine Glenfield Hospital Leicester LE3 9QP UK Fax: ian.pavord@hl-tr.nhs.k Received: May Accepted after revision: December STATEMENT OF INTEREST None declared KEYWORDS: Chronic cogh, cogh conts, cogh freqency, cogh monitor, Leicester Cogh Monitor Chronic cogh is a common reason for referral to respiratory physicians. The assessment of patients with chronic cogh is often based on the anatomical diagnostic protocol, which is a systematic evalation based on the nderstanding that most cases are de to disease of the pper respiratory tract where cogh receptors are most plentifl [1]. The most common conditions implicated in casing chronic cogh in nonsmokers are asthma, gastrooesophageal reflx and rhinitis, or a combination of these [2]. The identification of an important contribtion by the aforementioned conditions is largely based on the evalation of treatment trials [3]. However, there are few well-validated otcome measres to assess cogh severity and treatment efficacy. Cogh visal analoge scores, diary score cards, qality-of-life qestionnaires, coghreflex sensitivity measrement and cogh monitors have been proposed as potential tools to assess cogh severity [2]. The sbjective natre of symptom scores and qality-of-life qestionnaires [4, 5] and the poor specificity of coghreflex sensitivity measrement [6] to identify patients with chronic cogh have led to a renewed interest in the development of atomated amblatory cogh monitors [7 15]. BIRRING et al. [7] have previosly shown that there are marked differences in cogh freqency between patients with chronic cogh and healthy sbjects and that these measrements are repeatable, and have sggested that cogh freqency Eropean Respiratory Jornal Print ISSN Online ISSN c EUROPEAN RESPIRATORY JOURNAL VOLUME 31 NUMBER

2 LCM: A NOVEL COUGH MONITOR S.S. BIRRING ET AL. measrement is potentially sefl in the assessment of patients with chronic cogh. Crrently available cogh monitors are limited by difficlty in achieving nrestricted amblatory measrement in the patients own environment, an inability to perform 24-h recording and a lack of atomated cogh detection systems. The aim of the present stdy was to develop an atomated cogh monitor (the Leicester Cogh Monitor (LCM)) capable of recording cogh sonds for 24 h. The present stdy shows the validation of the LCM and preliminary findings of 6- and 24-h recordings in patients with chronic cogh. METHODS Sbjects A total of 15 consective patients with an isolated chronic cogh (.3 weeks dration) were recrited from a specialised cogh clinic. The clinic receives referrals from primary and secondary care largely confined to a poplation of 970,000 within Leicestershire, UK. The cases of cogh in patients with chronic cogh were: cogh variant asthma (n54); gastrooesophageal reflx (n53); eosinophilic bronchitis (n51); idiopathic (n53); post-viral (n51); bronchiectasis (n51); chronic obstrctive plmonary disease (n51); and chronic bronchitis (n51). Nine ot of these patients were randomly selected for the first stage of validation (cogh variant asthma (n54), eosinophilic bronchitis (n51), idiopathic (n52), postviral (n51), bronchiectasis (n51)) and all patients were inclded in the second validation stage. Investigations were carried ot according to a standardised algorithm [16]. The protocol for investigation and treatment and the criteria for accepting diagnosis were as previosly described [16]. Eight controls were recrited from healthy volnteers responding to local advertising. Control sbjects were asymptomatic, nonsmokers and had normal spirometry and a concentration of methacholine reqired to case a 20% decrease in forced expiratory volme in one second (FEV1) of.16mg?ml -1.No patients had received corticosteroids or other specific treatment for the condition casing cogh for o6 weeks prior to the stdy. A randomly selected sbgrop of six control sbjects and five patients with a stable chronic cogh and stable treatment reqirements (three with cogh variant asthma, one with gastro-oesophageal reflx-associated cogh and one with idiopathic chronic cogh) participated in cogh freqency repeatability stdies 3 6 months after the first, at the same time of day in order to avoid possible bias from dirnal variations. A total of 50 frther consective patients with chronic cogh were recrited in order to evalate 24-h recordings with the LCM (idiopathic cogh (n526), asthma (n58), eosinophilic bronchitis (n52), rhinitis (n52), sarcoidosis (n52), gastro-oesophageal reflx (n53), bronchiectasis (n52), chronic obstrctive plmonary disease (n52), enlarged tonsils (n52) and obstrctive sleep apnoea (n51)). All sbjects gave fll informed written consent to participate. The protocol for the present stdy was approved by the Leicestershire Research Ethics Committee. Cogh monitor The LCM (fig. 1) is a digital amblatory cogh monitor that records sond continosly from a free-field microphone necklace (Sennheiser MKE 2-5; Sennheiser electronic GmbH & Co. KG, Woedemark, Germany) onto a digital sond recorder (dimensions mm; iriver ifp-799; iriver Erope GmbH, Eschborn, Germany) at a sampling freqency of 16 khz and with an encoding bit rate of 64 kbit?s -1. The cogh monitor was attached at 09:00 h in all sbjects and retrned 6 24 h later. Sbjects were told that the LCM was a new investigative tool being developed to assess the natre of the cogh and were encoraged to resme their normal activity in their sal environment. When the recording was complete, data stored on the recorder was downloaded onto a compter, where it was analysed by an atomated cogh detection algorithm (the Leicester Cogh Algorithm). A general otline of the Leicester Cogh Algorithm has been described previosly [8, 17]. Briefly, the detection algorithm is based on Hidden Markov Models, a statistical method that can be sed to characterise the spectral properties of a timevarying pattern. The cogh detection algorithm was implemented based on the keyword-spotting approach, as defined in speech recognition, in which the objective is to detect the occrrence of a particlar set of keywords in a seqence of continos speech. Continos amblatory recordings in patients with chronic cogh were sed to train statistical models of the characteristics of cogh sonds and of the adio backgrond. Dring the detection process, the recorded adio signal was divided into contigos 10-s segments to be analysed by the Hidden Markov Models-based algorithm. Each 10-s adio segment was recognised by the detection algorithm as a seqence of variable-length adio sections, each in trn classified either as backgrond adio or as a possible cogh sond, depending on its statistics [8]. A second algorithm phase then sed brief operator inpt to facilitate the atomated algorithm in order to eliminate sonds that Recorder FIGURE 1. The Leicester Cogh Monitor. Microphone 1014 VOLUME 31 NUMBER 5 EUROPEAN RESPIRATORY JOURNAL

3 S.S. BIRRING ET AL. LCM: A NOVEL COUGH MONITOR might have been wrongly classified as cogh events in the first phase. For this, the operator is asked to classify, as cogh or otherwise, a small fraction of the sonds detected in the first phase as possible cogh sonds (the second phase takes 5 min to carry ot for a 24-h recording and,50 sonds are classified). The information is then sed to create statistical models that are adapted to the characteristics of the cogh sonds for that particlar recording and the remaining sonds that were not shown to the operator are classified sing these models. Cogh was defined as a characteristic explosive sond. The algorithm identifies coghs as single events whether they occr as isolated events or in a clster (i.e. attempts were made to determine how many coghs occrred in paroxysms). Validation Stage 1 The first stage of validation compared atomated cogh conts against those identified by manal sond analysis of the first and forth recorded hors of nine randomly selected patients with chronic cogh. Manal analysis of sond recordings consisted of three blinded observer conts (observer one twice and observer two once) and cogh or noncogh sonds were positively identified when all three conts were in agreement based on sond and visal inspection of the acostic trace. Each cogh sond was identified separately, whether it occred singlarly or in a clster or epoch of coghs. Intraand interobserver variability in cogh conts was established from the two blinded analyses performed by observer one and by comparing the mean of observer one s conts for these periods with conts obtained by observer two. The recordings were analysed twice sing the Leicester Cogh Algorithm to assess intra-recording repeatability. In order to classify noncogh sonds and determine whether particlar sonds were wrongly classified by the atomated system, observer one listened to all 6 h of the nine patients recordings and classified all recognisable sonds. The reslts of this analysis were compared with the atomated classification. Stage 2 The second stage of validation was extended to all recordings and compared atomated cogh conts against coghs identified manally by observer one, who analysed the entire 6-h recording. A frther 50 patients with chronic cogh nderwent 24-h atomated cogh freqency measrement. Atomated cogh freqency was compared for repeatability stdies. Analysis Sbject characteristics were described sing descriptive statistics and expressed as mean SEM for parametric data and median for non-parametric data. Cogh freqency was expressed as individal coghs?patient?h -1 for the dration of the recording. The validity of the LCM was presented as sensitivity, specificity and false-positive rate of the atomated algorithm for detecting coghs as measred by observer manal analysis. Intra- and interobserver variability of manal cogh conts and repeatability data was assessed as intra-class correlation coefficients and intra-sbject SD. RESULTS The sbject characteristics are shown in table 1. Validation stage 1 (First and forth recorded hor) Mean cogh conts were 39 coghs?patient -1?h -1 by atomated analysis compared with 43 coghs?patient -1?h -1 by manal analysis (mean difference -4 coghs?patient -1?h -1, 95% confidence interval (CI) -6 13; p50.4). The intra- and interobserver intra-class correlation coefficients for manal analysis of sond recordings (between observers one and two) were 0.99 and 0.98, respectively (both p,0.001). The intra- and interobserver (i.e. between mean of observer one and observer two) intraclass correlation coefficients for manal analysis of sond recordings were 0.99 and 0.98, respectively (both p,0.001). The intra-class correlation coefficient between atomated and manal observer conts was 0.9 (p,0.001; fig. 2a). The accracy of manal and atomated cogh conts appeared similar in recordings containing paroxysms and those with isolated coghs. The Leicester Cogh Algorithm had a sensitivity and a specificity of 91% and 99%, respectively, for detecting cogh sonds and a median false positive rate of 2.5 events?patient -1?h -1 against the gold standard of coghs detected manally by observer one twice and observer two once. There was no evidence that any particlar sond was more likely to be classified as a false positive (fig. 3). Validation stage 2 (6-h recordings) Mean SEM atomated cogh conts were 48 9 coghs? patient -1?h -1 in patients with chronic cogh and 2 1 coghs? patient -1?h -1 in control sbjects (mean difference 46 coghs? patient -1?h -1, 95% CI coghs?patient -1?h -1 ; p,0.001; fig. 4). There were no significant differences in cogh freqency between diagnostic grops. The cogh analysis took 2 h to complete, comprising 5 min for data download, 105 min for compter atomated analysis (an operator was not reqired to be present dring phase 1) and 10 min for operator inpt (phase 2) and printing reslts. The intra-class correlation coefficient between atomated and observer conts was 0.93 (p,0.001; fig. 2b). The LCM had sensitivity and specificity of 86 and 99%, respectively, for detecting cogh sonds and a median false-positive rate of 1.0 events?patient -1?h -1. The atomated cogh conts were repeatable in the 11 sbjects who nderwent repeatability testing (intra-sbject SD 11.4 coghs? patient -1?h -1, intra-class correlation coefficient 0.9; fig. 5). TABLE 1 Sbject characteristics Control Chronic cogh Sbjects (male) n 8 (0) 15 (5) Age yrs Cogh dration yrs 5 2 FEV1 % pred FEV1/FVC % Data are expressed as mean SEM, nless otherwise stated; FEV1: forced expiratory volme in one second; pred: predicted; FVC: forced vital capacity. c EUROPEAN RESPIRATORY JOURNAL VOLUME 31 NUMBER

4 LCM: A NOVEL COUGH MONITOR S.S. BIRRING ET AL. a) Difference between atomated and observer cogh conts patient -1 h -1 b) Difference between atomated and observer cogh conts patient -1 h The cogh freqency in patients ndergoing 24-h monitoring was 19 coghs?patient -1?h -1 ; daytime (08:00 22:00 h) cogh freqency was significantly greater than overnight cogh freqency (25 verss 10 coghs?patient -1?h -1 ; mean difference 15 coghs?patient -1?h -1, 95% CI 8 22 coghs?patient -1?h -1 ;p,0.001; fig. 6). DISCUSSION The LCM is a lightweight 24-h atomated amblatory cogh monitor that is easy to se and measres cogh in the sbjects own environment. The present stdy has shown that it is a valid and reliable tool for objectively measring cogh freqency. The high sensitivity and specificity for the detection of cogh sonds is comparable to other rotine diagnostic clinical tools and sperior to that reported for other more cmbersome cogh detection systems. Preliminary data is presented in the crrent stdy showing that the cogh freqency measred with the LCM is repeatable over o3 months, a period relevant to the dration of treatment trials that form an important part of the assessment of patients with chronic cogh. Repeatability was marginally better than that of recordings analysed manally [7]. The present data Mean of atomated and observer coghs patient-1 h-1 FIGURE 2. Bland-Altman plot of atomated verss manal observer cogh conts?patient -1?h -1. a) Validation stage 1 at which first and forth recorded hor (n59) were analysed. Each hor is depicted individally. b) Validation stage 2 at which 6-h recordings (n523) were analysed in their entirety. The complete cogh detection algorithm (phases 1 and 2) was tested in each validation stage. : mean difference;??????: 95% limits of agreement (26 intra-sbject SD); e: control sbjects; : chronic cogh patients sonds: 1834 cogh noncogh Phase accepted sonds: 1573 cogh 970 noncogh # Phase accepted sonds: 1528 cogh 152 noncogh rejected sonds: 261 cogh noncogh 863 rejected sonds: 45 cogh 818 noncogh FIGURE 3. False positives characterised manally that were detected by the atomated Leicester Cogh Monitor in nine 6-h recordings of patients with chronic cogh. Atomated analysis of sond recording is performed dring phase 1 and atomated analysis following operator inpt is performed dring phase 2. The sensitivity for cogh detection is slightly lower than that from the gold standard validation stage 1, since the comparator was a 6-h manal conting by observer one only. # : noncogh sonds were: speech (n5351), implsive noise (n5222) throat clearing (n5119), environmental noise (n5111), lagh (n580), other person coghing (n531), incomplete coghs (n522), child talking/shoting (n521), sneeze (n56), telephone ringing (n53), brp (n52) and dog barking (n52). " : noncogh sonds were: speech (n5293), implsive noise (n5214), throat clearing (n596), environmental noise (n587), lagh (n567), other person coghing (n515), incomplete coghs (n518), child talking/shoting (n517), sneeze (n55), telephone ringing (n53), brp (n52) and dog barking (n51). + : noncogh sonds were: speech (n558), implsive noise (n58), throat clearing (n523), environmental noise (n524), lagh (n513), other person coghing (n516), incomplete coghs (n54), child talking/shoting (n54), sneeze (n51) and dog barking (n51). sggest that the LCM might be a particlarly sefl otcome measre in assessing patients with cogh and measring the response to therapy in the clinic and in clinical trials. A limitation of the present stdy is that evalation of cogh freqency was based on 6-h daytime cogh recordings owing -1 h -1 Coghs patient Control sbjects *** Chronic cogh patients FIGURE 4. Mean SEM atomated cogh conts?patient -1?h -1 in control sbjects and chronic cogh patients (6-h recordings). ***: p, VOLUME 31 NUMBER 5 EUROPEAN RESPIRATORY JOURNAL

5 S.S. BIRRING ET AL. LCM: A NOVEL COUGH MONITOR Difference between tests coghs patient -1 h Coghs patient Mean of two tests coghs patient-1 h-1 FIGURE 5. Bland-Altman plot of atomated cogh conts?patient -1?h -1 repeated over 3 6 months in chronic cogh patients and control sbjects. : mean difference;???????: 95% limits of agreement (26 intra-sbject SD). e: control sbjects; : chronic cogh patients. to limited battery life (6 8 h) at the inception of the stdy. Advances in battery technology, since then, have allowed the extension of recordings to o24 h. The atomated system has allowed these recordings to be analysed relatively qickly and accrately and shold facilitate the investigation of potential dirnal variations in cogh freqency and the effects of aggravators of potential cogh, sch as environmental polltion, cigarette smoking and gastro-oesophageal reflx. In the present stdy, a range of sonds inclding speech, throat clearing and environmental noises cased false-positive detected coghs. There was little evidence that any of these sonds cased particlar difficlties with detection, nor did cogh paroxysms appear to present problems for accrate manal and atomated cogh conts. However, greater experience with the monitor may identify sonds or cogh paroxysms that present particlar problems for the algorithm to classify and allow frther refinement of the algorithm. The present stdy involved small nmbers of sbjects and it will be important to stdy a larger poplation of control sbjects and patients with well-defined respiratory disease, before and after treatment, and in different environments, to flly validate the cogh monitor. The present preliminary work sggests that sch a stdy will be feasible. Cogh freqency was stable throgh the day and was significantly redced overnight compared with daytimes in accord with previos data, sggesting a dirnal variation in cogh freqency [5, 9, 13]. Frther work is reqired to determine the validity and the short- and longer-term repeatability of 24-h cogh recordings. A limitation of the present stdy is that only 2 h per recording were sed to compare atomated cogh conts with those obtained from manal conting for the validation stdy. Manal cogh conting is very time consming and laborios so only the first and forth hors of each recording were manally analysed for consistency. Each recording was manally conted three times in order to obtain a more robst measre of the tre cogh freqency. The LCM had a high sensitivity and specificity for detecting cogh against this gold 0 09:00 11:00 13:00 15:00 17:00 19:00 standard. This was confirmed in the second part of the validation stdy where cogh conts derived from atomated analysis of 6-h recordings were compared with cogh conts derived from a single manal observer. The sensitivity of the cogh algorithm was slightly lower with 6-h recordings compared with 2-h recordings. This is most probably becase of the more robst measre of tre cogh freqency sed for the 2-h recordings, compared with single observer manal conts sed for the 6-h recordings. A potential criticism of cogh conts derived from adio recordings is that they might not accrately reflect the tre cogh rate since it is not possible to visalise the act of coghing. However, a recent stdy [18] compared manal cogh conts from adio with video recordings and fond them to be very similar. That stdy conclded that manal cogh conts from adio recordings shold be regarded as the gold standard to validate cogh monitors since adio recordings had sperior sond qality to that from video recordings. The LCM qantifies cogh freqency as single episodes of cogh rather than epochs or clsters of coghs and cogh seconds (seconds containing cogh), as sed by others [12]. The present athors believe that single cogh episodes are a more meaningfl measre and are easier to interpret by physicians and patients. Cogh freqency, rather than intensity, was measred since cogh events are less inflenced by microphone position and mffling of sonds by covering the moth dring the act of coghing. Frthermore, cogh intensity determined with sond analysis lacks responsiveness compared with cogh freqency in clinical trials of antitssive drgs [10]. Cogh intensity determined by other parameters sch as airflow or chest wall movement is less practical for rotine clinical measrement. The LCM was validated in sbjects with chronic cogh de to a wide range of conditions so it reliably detects coghs with differing characteristics. One of the challenges of developing cogh monitors in the past has been differentiating cogh sonds from throat clearing, sneezing, speech and other cogh-like sonds. The Leicester Cogh Monitor differentiates cogh from other sonds reliably 21:00 Time h FIGURE h amblatory, atomated cogh freqency recordings in 50 chronic cogh patients. Data are presented as mean+sem. 23:00 01:00 03:00 05:00 07:00 09:00 c EUROPEAN RESPIRATORY JOURNAL VOLUME 31 NUMBER

6 LCM: A NOVEL COUGH MONITOR S.S. BIRRING ET AL. as indicated by the high sensitivity and particlarly high specificity for detecting cogh. The Leicester Cogh Monitor represents a potential advance over existing cogh monitors in that it is portable otside a controlled environment, it does not reqire measrement of abdominal electromyography and it can be set to record for 24 h. The 24-h sond recording is discarded after the atomated analysis process, thereby ensring patient privacy. The Leicester Cogh Monitor cold be sed to validate the presence of excessive cogh, to assess cogh severity and to objectively evalate response to therapy. Frther stdies are reqired in order to assess the se of the Leicester Cogh Monitor in clinical practice. ACKNOWLEDGEMENTS The present athors wold like to thank the sbjects who participated in the stdy, C. Long (Dept of Respiratory Physiology, Glenfield Hospital, Leicester, UK) for assistance in data analysis, D.D. Vara (Dept of Respiratory Physiology, Glenfield Hospital) for assistance in the clinical characterisation of some of the patients and the lng fnction nit staff at King s College Hospital, London, UK (C. Wood, L. Morgan, D. Cox and D. Land) and Glenfield Hospital. REFERENCES 1 Irwin RS, Madison JM. The diagnosis and treatment of cogh. N Engl J Med 2000; 343: Irwin RS. Assessing cogh severity and efficacy of therapy in clinical research: ACCP evidence-based clinical practice gidelines. Chest 2006; 129: Sppl. 1, 232S 237S. 3 Morice AH, Fontana GA, Sovijarvi AR, et al. The diagnosis and management of chronic cogh. Er Respir J 2004; 24: Birring SS, Prdon B, Carr AJ, Singh SJ, Morgan MD, Pavord ID. Development of a symptom specific health stats measre for patients with chronic cogh: Leicester Cogh Qestionnaire (LCQ). Thorax 2003; 58: Raj AA, Birring SS. Clinical assessment of chronic cogh severity. Plm Pharmacol Ther 2007; 20: Prdon B, Birring SS, Vara DD, Hall AP, Thompson JP, Pavord ID. Cogh and glottic-stop reflex sensitivity in health and disease. Chest 2005; 127: Birring SS, Matos S, Patel RB, Prdon B, Evans DH, Pavord ID. Cogh freqency, cogh sensitivity and health stats in patients with chronic cogh. Respir Med 2006; 100: Matos S, Birring SS, Pavord ID, Evans DH. Detection of cogh signals in continos adio recordings sing hidden Markov models. IEEE Trans Biomed Eng 2006; 53: Coyle MA, Keenan DB, Henderson LS, et al. Evalation of an amblatory system for the qantification of cogh freqency in patients with chronic obstrctive plmonary disease. Cogh 2005; 1: Pavesi L, Sbbraj S, Porter-Shaw K. Application and validation of a compterized cogh acqisition system for objective monitoring of acte cogh: a meta-analysis. Chest 2001; 120: Chang AB, Newman RG, Phelan PD, Robertson CF. A new se for an old Holter monitor: an amblatory cogh meter. Er Respir J 1997; 10: Smith J, Owen E, Earis J, Woodcock A. Effect of codeine on objective measrement of cogh in chronic obstrctive plmonary disease. J Allergy Clin Immnol 2006; 117: Hs JY, Stone RA, Logan-Sinclair RB, Worsdell M, Bsst CM, Chng KF. Coghing freqency in patients with persistent cogh: assessment sing a 24 hor amblatory recorder. Er Respir J 1994; 7: Barry SJ, Dane AD, Morice AH, Walmsley AD. The atomatic recognition and conting of cogh. Cogh 2006; 2: Pal IM, Wai K, Jewell SJ, Shaffer ML, Varadan VV. Evalation of a new self-contained, amblatory, objective cogh monitor. Cogh 2006; 2: Brightling CE, Ward R, Goh KL, Wardlaw AJ, Pavord ID. Eosinophilic bronchitis is an important case of chronic cogh. Am J Respir Crit Care Med 1999; 160: Matos S, Birring SS, Pavord ID, Evans DH. An atomated system for 24-hor monitoring of cogh freqency: the Leicester cogh monitor. IEEE Trans Biomed Eng 2007; 54: Smith JA, Earis JE, Woodcock AA. Establishing a gold standard for manal cogh conting: video verss digital adio recordings. Cogh 2006; 2: VOLUME 31 NUMBER 5 EUROPEAN RESPIRATORY JOURNAL