Respiratory sounds. Advances beyond the stethoscope

Obstructive Sleep Apnea Screening and Airway Structure Characterization During Wakefulness Using Tracheal Breathing Sounds

Screening for obstructive sleep apnea (OSA) disorder during wakefulness is challenging. In this paper, we present a set of tracheal breathing sounds characteristics with classification power for separating individuals with apnea/hypopnea index (AHI) ≥ 10 (OSA group) from those with AHI ≤ 5 (non-OSA group) during wakefulness. Tracheal breathing sound signals were recorded during wakefulness in supine position; subjects were instructed to have a few deep breaths through their nose, then through their mouth. Study participants were 147 individuals (80 males) referred to overnight polysomnography (PSG) assessment; their AHI scores were collected after their overnight-PSG study was completed. The signals were normalized; then, their power spectra were estimated. After conducting a multi-stage process for feature extraction and selection on a subset of training data, two spectral features showing significant differences between the two groups were selected for classification. These features showed a correlation of 0.42 with AHI. A 2-class support vector machine classifier with a linear kernel was used. Following this an exhaustive leave-two-out cross-validation was performed. The overall accuracies were 83.83 and 83.92% for training and testing datasets, respectively, while the overall sensitivity and specificity of the test datasets were 82.61 and 85.22%, respectively. We also applied the same method for anthropometric information (i.e., age, weight, etc.) as features, and they resulted in an overall accuracy of 77.6 and 76.2% for training and testing datasets, respectively. The results of this study show a superior classification power of respiratory sound features compared to anthropometric features for a quick screening of OSA during wakefulness. The relationship of the sound features and known morphological upper airway structure of OSA subjects are also discussed.

Snoring sound classification from respiratory signal

Snoring is common in the general population and the irregularity could lead to the presence of Obstructive sleep apnea. Diagnosis of OSA could therefore be made by snoring sound analysis. However, there is still a shortage of robust methods to automatically detect snoring sounds without the need to calibrate for every individual. In this paper, a novel method based on neural network is proposed to classify breathing sound episodes from snoring and non-snoring sound segments. Our snore detection algorithm was applied to the tracheal sounds of nine individuals with different OSA severities. On the testing dataset, the classifier achieved a sensitivity and specificity of 95.9% and 97.6% respectively. Our results indicate that using such a method could help to detect snoring sounds with high accuracy which would be useful in the diagnosis of sleep apnea.

Automatic lung tidal volumes estimation from tracheal sounds

This paper presents a method to automatically estimate lung tidal volumes from the acoustic signals generated in the respiratory track. The signal is measured with an acoustic based sensor placed in the suprasternal notch. The method does not require any previous knowledge or modelling of the individual respiratory track, and relies on just one calibration parameter. The proposed algorithm is tested on 316 respiratory phases obtained from 4 volunteers. The subjects were simultaneously wearing a Wright respirometer which was used as a gold standard for comparison. Agreement between the two methods was assessed with Bland-Altman techniques. The results show the potential the technique has, integrated with a small acoustic sensor, for less-intrusive and even remote and/or continuous monitoring of lung tidal volumes.

A Subject-Specific Acoustic Model of the Upper Airway for Snoring Sounds Generation

Monitoring variations in the upper airway narrowing during sleep is invasive and expensive. Since snoring sounds are generated by air turbulence and vibrations of the upper airway due to its narrowing; snoring sounds may be used as a non-invasive technique to assess upper airway narrowing. Our goal was to develop a subject-specific acoustic model of the upper airway to investigate the impacts of upper airway anatomy, e.g. length, wall thickness and cross-sectional area, on snoring sounds features. To have a subject-specific model for snoring generation, we used measurements of the upper airway length, cross-sectional area and wall thickness from every individual to develop the model. To validate the proposed model, in 20 male individuals, intensity and resonant frequencies of modeled snoring sounds were compared with those measured from recorded snoring sounds during sleep. Based on both modeled and measured results, we found the only factor that may positively and significantly contribute to snoring intensity was narrowing in the upper airway. Furthermore, measured resonant frequencies of snoring were inversely correlated with the upper airway length, which is a risk factor for upper airway collapsibility. These results encourage the use of snoring sounds analysis to assess the upper airway anatomy during sleep.

Relationship between computerized wheeze detection and lung function parameters in young infants

OBJECTIVE

Computerized respiratory sound analysis (CORSA) has been validated in the assessment of wheeze in infants, but it is unknown whether automatically detected wheeze is associated with impaired lung function. This study investigated the relationship between wheeze detection and conventional lung function testing (LFT) parameters.

METHODS

CORSA was performed using the PulmoTrack® monitor in 110 infants, of median (interquartile range) postmenstrual age 50 (46-56) weeks and median body weight 4,810 (3,980-5,900) g, recovering from neonatal intensive care. In the same session, LFT was performed, including tidal breathing measurements, occlusion tests, body plethysmography, forced expiratory flow by rapid thoracoabdominal compression, sulfur hexafluoride (SF6 ) multiple breath washout (MBW), and capillary blood gas analysis. Infants were classified as wheezers or non-wheezers using predefined cut-off values for the duration of inspiratory and expiratory wheeze.

RESULTS

Wheezing was detected in 72 (65%) infants, with 43 (39%) having inspiratory and 53 (48%) having expiratory wheezing. Endotracheal mechanical ventilation in the neonatal period for > 24 hr was associated with inspiratory wheeze (P = 0.009). Airway resistance was increased in both inspiratory (P = 0.02) and expiratory (P = 0.004) wheezers and correlated with the duration of expiratory wheeze (r = 0.394, P < 0.001). Expiratory wheezers showed a significant increase in respiratory resistance (P = 0.001), time constant (0.012), and functional residual capacity using SF6 MBW (P = 0.019). There was no association between wheezing and forced expiratory flow or blood gases.

CONCLUSION

CORSA can help identify neonates and young infants with subclinical airway obstruction and may prove useful in the follow-up of high-risk infants.

Wheezing recognition algorithm using recordings of respiratory sounds at the mouth in a pediatric population

BACKGROUND

Respiratory diseases in children are a common reason for physician visits. A diagnostic difficulty arises when parents hear wheezing that is no longer present during the medical consultation. Thus, an outpatient objective tool for recognition of wheezing is of clinical value.

METHOD

We developed a wheezing recognition algorithm from recorded respiratory sounds with a Smartphone placed near the mouth. A total of 186 recordings were obtained in a pediatric emergency department, mostly in toddlers (mean age 20 months). After exclusion of recordings with artefacts and those with a single clinical operator auscultation, 95 recordings with the agreement of two operators on auscultation diagnosis (27 with wheezing and 68 without) were subjected to a two phase algorithm (signal analysis and pattern classifier using machine learning algorithms) to classify records.

RESULTS

The best performance (71.4% sensitivity and 88.9% specificity) was observed with a Support Vector Machine-based algorithm. We further tested the algorithm over a set of 39 recordings having a single operator and found a fair agreement (kappa=0.28, CI95% [0.12, 0.45]) between the algorithm and the operator.

CONCLUSIONS

The main advantage of such an algorithm is its use in contact-free sound recording, thus valuable in the pediatric population.

Regional Lung Sound Asynchrony in Chronic Obstructive Pulmonary Disease Patients

<b><i>Background:</i></b> Regional lung sound distribution in chronic obstructive pulmonary disease (COPD) is reported to be asynchronous. Mathematical analyses using vibration response imaging (VRI), such as left and right lung asynchrony (gap index; GI) and regional lung asynchrony (asynchrony score; AS), are useful measures to evaluate lung sound asynchrony. <b><i>Objectives:</i></b> The aim of this study was to investigate the association of lung sound asynchrony with pulmonary functions and emphysematous lesions in COPD patients. <b><i>Methods:</i></b> VRI recordings and pulmonary function tests were performed in 46 stable male COPD patients and in 40 healthy male smokers. Lung sound asynchrony was evaluated using GI, AS of the left and right lung (AS L-R), and AS of the upper and lower lung (AS U-L). In 38 patients, computed tomography taken within 6 months was available and analyzed. <b><i>Results:</i></b> AS L-R and AS U-L were significantly higher in COPD patients than in healthy smokers, with no significant difference in GI. There were no significant correlations with either AS and pulmonary functions, excluding a negative correlation between AS U-L and diffusion capacity. Although there were no significant correlations between both AS and severity of emphysema, significant positive correlations were observed between heterogeneity of emphysematous lesions and AS L-R (&#x03C1; = 0.38, p < 0.05) or AS U-L (&#x03C1; = 0.51, p < 0.005). <b><i>Conclusions:</i></b> Regional lung sounds are distributed more asynchronously in COPD patients than in healthy smokers, which correlates with the heterogeneous distribution of emphysematous lesions.

Towards the standardisation of lung sound nomenclature

Auscultation of the lung remains an essential part of physical examination even though its limitations, particularly with regard to communicating subjective findings, are well recognised. The European Respiratory Society (ERS) Task Force on Respiratory Sounds was established to build a reference collection of audiovisual recordings of lung sounds that should aid in the standardisation of nomenclature. Five centres contributed recordings from paediatric and adult subjects. Based on pre-defined quality criteria, 20 of these recordings were selected to form the initial reference collection. All recordings were assessed by six observers and their agreement on classification, using currently recommended nomenclature, was noted for each case. Acoustical analysis was added as supplementary information. The audiovisual recordings and related data can be accessed online in the ERS e-learning resources. The Task Force also investigated the current nomenclature to describe lung sounds in 29 languages in 33 European countries. Recommendations for terminology in this report take into account the results from this survey.

Effects of a respiratory physiotherapy session in patients with LRTI: a pre/post-test study

INTRODUCTION

The role of respiratory physiotherapy (RP) in lower respiratory tract infections (LRTI) has been questioned. However, studies have focused on hospitalised patients, and the presence/absence of an underlying disease has been neglected.

OBJECTIVES

To assess the effects of a RP session in community patients with LRTI and to explore the differences between patients with pneumonia (restrictive disease - AR) and those with exacerbations of an obstructive disease (AO).

METHODS

A pre/post-test study was conducted. A RP session was applied to patients with LRTI and crackles, wheezes, dyspnoea, perception of sputum and oxygen saturation were collected pre/post session. Comparisons were performed using paired t-tests or Wilcoxon tests.

RESULTS

Thirty patients (14 males, 55.23 ± 17.78 years) with pneumonia (AR, n = 12), exacerbations of chronic obstructive pulmonary disease, acute bronchitis and asthma (AO, n = 18) were enrolled. After treatment, the total sample presented lower wheeze rates at trachea (P = 0.02; r = -0.54) and less sputum (P = 0.01; r = -0.47). AR patients presented a decrease in the number of crackles (P < 0.05; 0.30 < d_z  < 0.26) and number and rate of wheezes at chest locations (P < 0.05; -0.56 < r < -0.48). AO patients showed an increase in the number of crackles (P < 0.05; 0.20 <d_z <0.31), wheeze frequency (P = 0.03; r = -0.27) and dyspnoea (P = 0.04; r = -0.55); and a decrease in the number of wheezes at trachea (P = 0.02; r = -0.54).

CONCLUSIONS

RP seems effective in reducing wheezes and perception of sputum in patients with LRTI. However, when considering AR and AO diseases separately, further changes in respiratory sounds and dyspnoea emerged. This highlights the importance of considering subgroups of patients with LRTI to develop RP evidence-base practice.

Computerized respiratory sounds: a comparison between patients with stable and exacerbated COPD

INTRODUCTION

Diagnosis of acute exacerbations of chronic obstructive pulmonary disease (AECOPD) is often challenging as it relies on patients' clinical presentation. Computerized respiratory sounds (CRS), namely crackles and wheezes, may have the potential to contribute for the objective diagnosis/monitoring of an AECOPD.

OBJECTIVES

This study explored if CRS differ during stable and exacerbation periods in patients with COPD.

METHODS

13 patients with stable COPD and 14 with AECOPD were enrolled. CRS were recorded simultaneously at trachea, anterior, lateral and posterior chest locations using seven stethoscopes. Airflow (0.4-0.6l/s) was recorded with a pneumotachograph. Breathing phases were detected using airflow signals; crackles and wheezes with validated algorithms.

RESULTS

At trachea, anterior and lateral chest, no significant differences were found between the two groups in the number of inspiratory/expiratory crackles or inspiratory wheeze occupation rate. At posterior chest, the number of crackles (median 2.97-3.17 vs. 0.83-1.2, P < 0.001) and wheeze occupation rate (median 3.28%-3.8% vs. 1.12%-1.77%, P = 0.014-0.016) during both inspiration and expiration were significantly higher in patients with AECOPD than in stable patients. During expiration, wheeze occupation rate was also significantly higher in patients with AECOPD at trachea (median 3.12% vs. 0.79%, P < 0.001) and anterior chest (median 3.55% vs. 1.28%, P < 0.001).

CONCLUSION

Crackles and wheezes are more frequent in patients with AECOPD than in stable patients, particularly at posterior chest. These findings suggest that these CRS can contribute to the objective diagnosis/monitoring of AECOPD, which is especially valuable considering that they can be obtained by integrating computerized techniques with pulmonary auscultation, a noninvasive method that is a component of patients' physical examination.

Estimation of Respiratory Rates Using the Built-in Microphone of a Smartphone or Headset

This paper proposes accurate respiratory rate estimation using nasal breath sound recordings from a smartphone. Specifically, the proposed method detects nasal airflow using a built-in smartphone microphone or a headset microphone placed underneath the nose. In addition, we also examined if tracheal breath sounds recorded by the built-in microphone of a smartphone placed on the paralaryngeal space can also be used to estimate different respiratory rates ranging from as low as 6 breaths/min to as high as 90 breaths/min. The true breathing rates were measured using inductance plethysmography bands placed around the chest and the abdomen of the subject. Inspiration and expiration were detected by averaging the power of nasal breath sounds. We investigated the suitability of using the smartphone-acquired breath sounds for respiratory rate estimation using two different spectral analyses of the sound envelope signals: The Welch periodogram and the autoregressive spectrum. To evaluate the performance of the proposed methods, data were collected from ten healthy subjects. For the breathing range studied (6-90 breaths/min), experimental results showed that our approach achieves an excellent performance accuracy for the nasal sound as the median errors were less than 1% for all breathing ranges. The tracheal sound, however, resulted in poor estimates of the respiratory rates using either spectral method. For both nasal and tracheal sounds, significant estimation outliers resulted for high breathing rates when subjects had nasal congestion, which often resulted in the doubling of the respiratory rates. Finally, we show that respiratory rates from the nasal sound can be accurately estimated even if a smartphone's microphone is as far as 30 cm away from the nose.

Development of a new homecare sleep monitor using body sounds and motion tracking

This paper presents the development of a sleep monitor to provide a comfortable way of detecting sleep-related breathing disorders like the obstructive sleep apnea syndrome (OSAS). OSAS is traditionally diagnosed using polysomnography, which requires a whole night stay at the sleep laboratory of a hospital with multiple electrodes and sensors attached to the patient’s body. However, body sound and motion tracking also provide extensive information about sleep course. A unique recording device offering a good body sound extraction, noise suppression and a small size is developed. Using this device a reliable detection of breathing and heart beat is possible. In addition sleeping positions and the activity of the patient will be evaluated using an inertial measurement unit (IMU). The device is easy to set up and offers the possibility to use it independently at home.

Initial experiments have shown that volunteers were able to set up the device on their own. Furthermore several overnight recordings revealed the capability to monitor breathing, heart rate, sleeping position as well as movements of the patient.

Auscultation of the respiratory system

Auscultation of the lung is an important part of the respiratory examination and is helpful in diagnosing various respiratory disorders. Auscultation assesses airflow through the trachea-bronchial tree. It is important to distinguish normal respiratory sounds from abnormal ones for example crackles, wheezes, and pleural rub in order to make correct diagnosis. It is necessary to understand the underlying pathophysiology of various lung sounds generation for better understanding of disease processes. Bedside teaching should be strengthened in order to avoid erosion in this age old procedure in the era of technological explosion.

Using K-Nearest Neighbor Classification to Diagnose Abnormal Lung Sounds

A reported 30% of people worldwide have abnormal lung sounds, including crackles, rhonchi, and wheezes. To date, the traditional stethoscope remains the most popular tool used by physicians to diagnose such abnormal lung sounds, however, many problems arise with the use of a stethoscope, including the effects of environmental noise, the inability to record and store lung sounds for follow-up or tracking, and the physician's subjective diagnostic experience. This study has developed a digital stethoscope to help physicians overcome these problems when diagnosing abnormal lung sounds. In this digital system, mel-frequency cepstral coefficients (MFCCs) were used to extract the features of lung sounds, and then the K-means algorithm was used for feature clustering, to reduce the amount of data for computation. Finally, the K-nearest neighbor method was used to classify the lung sounds. The proposed system can also be used for home care: if the percentage of abnormal lung sound frames is > 30% of the whole test signal, the system can automatically warn the user to visit a physician for diagnosis. We also used bend sensors together with an amplification circuit, Bluetooth, and a microcontroller to implement a respiration detector. The respiratory signal extracted by the bend sensors can be transmitted to the computer via Bluetooth to calculate the respiratory cycle, for real-time assessment. If an abnormal status is detected, the device will warn the user automatically. Experimental results indicated that the error in respiratory cycles between measured and actual values was only 6.8%, illustrating the potential of our detector for home care applications.

Pneumothorax effects on pulmonary acoustic transmission

Pneumothorax (PTX) is an abnormal accumulation of air between the lung and the chest wall. It is a relatively common and potentially life-threatening condition encountered in patients who are critically ill or have experienced trauma. Auscultatory signs of PTX include decreased breath sounds during the physical examination. The objective of this exploratory study was to investigate the changes in sound transmission in the thorax due to PTX in humans. Nineteen human subjects who underwent video-assisted thoracic surgery, during which lung collapse is a normal part of the surgery, participated in the study. After subjects were intubated and mechanically ventilated, sounds were introduced into their airways via an endotracheal tube. Sounds were then measured over the chest surface before and after lung collapse. PTX caused small changes in acoustic transmission for frequencies below 400 Hz. A larger decrease in sound transmission was observed from 400 to 600 Hz, possibly due to the stronger acoustic transmission blocking of the pleural air. At frequencies above 1 kHz, the sound waves became weaker and so did their changes with PTX. The study elucidated some of the possible mechanisms of sound propagation changes with PTX. Sound transmission measurement was able to distinguish between baseline and PTX states in this small patient group. Future studies are needed to evaluate this technique in a wider population.

Tidal volume estimation using the blanket fractal dimension of the tracheal sounds acquired by smartphone

In this paper, we propose the use of blanket fractal dimension (BFD) to estimate the tidal volume from smartphone-acquired tracheal sounds. We collected tracheal sounds with a Samsung Galaxy S4 smartphone, from five (N = 5) healthy volunteers. Each volunteer performed the experiment six times; first to obtain linear and exponential fitting models, and then to fit new data onto the existing models. Thus, the total number of recordings was 30. The estimated volumes were compared to the true values, obtained with a Respitrace system, which was considered as a reference. Since Shannon entropy (SE) is frequently used as a feature in tracheal sound analyses, we estimated the tidal volume from the same sounds by using SE as well. The evaluation of the performed estimation, using BFD and SE methods, was quantified by the normalized root-mean-squared error (NRMSE). The results show that the BFD outperformed the SE (at least twice smaller NRMSE was obtained). The smallest NRMSE error of 15.877% ± 9.246% (mean ± standard deviation) was obtained with the BFD and exponential model. In addition, it was shown that the fitting curves calculated during the first day of experiments could be successfully used for at least the five following days.

Upper airway resistance changes from inspiration to expiration during wakefulness is a predictor of sleep apnea: A pilot study

Upper airway structure is known to be affected in individuals with obstructive sleep apnea (OSA). In this pilot study, we investigated whether the changes of upper airway resistance (UAR) within a breathing phase, estimated from tracheal breathing sounds and airflow, is a predictor of OSA severity. Tracheal breathing sounds and airflow of 15 individuals in three groups of mild, moderate and severe OSA during wakefulness, recorded in supine position, were used in this study. The grouping of the study subjects were done based on their apnea/hypopnea index (AHI) per hour, prospectively. The breathing sounds and airflow were recorded simultaneously with full overnight polysomnography (PSG) assessment. The sounds of 3 noise-free breathing cycles were extracted and sequestered into inspiratory and expiratory phase segments manually for each study subject. The power spectra of the sound signals of each respiratory phase were calculated in 15ms windows with 50% overlap between adjacent windows. The UAR was then estimated as the ratio of the energy of power spectrum of each window within 100 to 2500 Hz, and its corresponding airflow. Then, the variance of the short-time windows' UAR, Var(UAR) during each respiratory phase was calculated. The ratio of the Var(UAR) in logarithmic scale between inspiration and expiration was found to be significantly different (p<;0.01) between the three OSA groups; also it was found to be highly correlated with AHI. These results are congruent with the hypothesis that the upper airway patency is not kept constant in OSA population, and suggest the change in acoustic UAR during wakefulness maybe considered as a predictor of OSA.

Feature extraction using time-frequency analysis for monophonic-polyphonic wheeze discrimination

The aim of this study is monophonic-polyphonic wheeze episode discrimination rather than the conventional wheeze (versus non-wheeze) episode detection. We used two different methods for feature extraction to discriminate monophonic and polyphonic wheeze episodes. One of the methods is based on frequency analysis and the other is based on time analysis. Frequency analysis based method uses ratios of quartile frequencies to exploit the difference in the power spectrum. Time analysis based method uses mean crossing irregularity to exploit the difference in periodicity in the time domain. Both methods are applied on the data before and after an image processing based preprocessing step. Calculated features are used in classification both individually and in combinations. Support vector machine, k-nearest neighbor and Naive Bayesian classifiers are adopted in leave-one-out scheme. A total of 121 monophonic and 110 polyphonic wheeze episodes are used in the experiments, where the best classification performances are 71.45% for time domain based features, 68.43% for frequency domain based features, and 75.78% for a combination of selected best features.

Chaos to randomness: distinguishing between healthy and non-healthy lung sound behaviour

Lung abnormalities and respiratory diseases increase as side effects of urban life and development. Therefore, understanding lung dynamics and its changes during the presence of abnormalities are critical in order to design more reliable tools for the early diagnosis and screening of lung pathology. The goal of this paper is to indicate the chaotic nature of normal lung sound and its transition to randomness in the presence of lung disease. The latter characteristic could serve as an indicator for evaluating the recovery process for patients suffering from lung disease. To verify this idea, we compared group of healthy and non-healthy subjects and also group of non-healthy subjects before and after treatments. Chaotic and randomness indices applied to lung sound signals which captured by multichannel data acquisition system. Results show that the normal lung displays chaotic dynamics. However, with the increase in lung abnormality, moves toward more random behaviour and away from its original chaotic state. Also, chaotic and randomness indices indicate their abilities to classify healthy and non-healthy lung sounds.

Validation of computerized wheeze detection in young infants during the first months of life

BACKGROUND

Several respiratory diseases are associated with specific respiratory sounds. In contrast to auscultation, computerized lung sound analysis is objective and can be performed continuously over an extended period. Moreover, audio recordings can be stored. Computerized lung sounds have rarely been assessed in neonates during the first year of life. This study was designed to determine and validate optimal cut-off values for computerized wheeze detection, based on the assessment by trained clinicians of stored records of lung sounds, in infants aged <1 year.

METHODS

Lung sounds in 120 sleeping infants, of median (interquartile range) postmenstrual age of 51 (44.5-67.5) weeks, were recorded on 144 test occasions by an automatic wheeze detection device (PulmoTrack®). The records were retrospectively evaluated by three trained clinicians blinded to the results. Optimal cut-off values for the automatically determined relative durations of inspiratory and expiratory wheezing were determined by receiver operating curve analysis, and sensitivity and specificity were calculated.

RESULTS

The optimal cut-off values for the automatically detected durations of inspiratory and expiratory wheezing were 2% and 3%, respectively. These cutoffs had a sensitivity and specificity of 85.7% and 80.7%, respectively, for inspiratory wheezing and 84.6% and 82.5%, respectively, for expiratory wheezing. Inter-observer reliability among the experts was moderate, with a Fleiss' Kappa (95% confidence interval) of 0.59 (0.57-0.62) for inspiratory and 0.54 (0.52 - 0.57) for expiratory wheezing.

CONCLUSION

Computerized wheeze detection is feasible during the first year of life. This method is more objective and can be more readily standardized than subjective auscultation, providing quantitative and noninvasive information about the extent of wheezing.

Can we distinguish pneumonia from wheezy diseases in tachypnoeic children under low-resource conditions? A prospective observational study in four Indian hospitals

BACKGROUND

Acute respiratory infections are the commonest cause of mortality and morbidity in children worldwide. A quarter of all deaths occur in India alone. In order to reduce this disease burden, there is a need for better diagnostic criteria, particularly ones allowing early detection of high-risk children.

METHODS

We enrolled 516 under 5 year olds, in four Indian hospitals, who met WHO age-dependent tachypnoea criteria for pneumonia at presentation. Patients underwent a protocolised examination assessing 29 items, including history, examination, O2 saturation, plus scores for chest X-ray, auscultation and conscious level. Treatment was determined by the emergency room (ER) physician. All children were reviewed at day 4 by a paediatrician and placed into four diagnostic categories: pneumonia, wheezy disease, mixed and non-respiratory.

RESULTS

The majority had wheezy diseases (42.8%). The remainder had pneumonia (35.9%), mixed disease (18.6%) and non-respiratory (2.7%). Best diagnostic predictors for wheezy disease were (auscultation/previous similar episodes) and for pneumonia (auscultation/CXR score). Mortality was 1.6%. Best disease severity predictors were conscious level, weight/age z score and respiratory/pulse rates.

INTERPRETATION

Current tachypnoea-based algorithms significantly overdiagnose pneumonia in children and underdiagnose wheezy diseases. Diagnostic accuracy can be improved by various combinations of clinical variables, but the best single diagnostic predictor is auscultation. Simple criteria can also be defined that reliably detect which tachypnoeic children are at high risk of death or deterioration. Management plans based on these protocols could reduce unnecessary antibiotic use, improve the management of wheezy diseases and reduce mortality by earlier identification of high-risk children.

The correlation between lung sound distribution and pulmonary function in COPD patients

Background

Regional lung sound intensity in chronic obstructive pulmonary disease (COPD) patients is influenced by the severity and distribution of emphysema, obstructed peripheral airways, and altered ribcage and diaphragm configurations and movements due to hyperinflation. Changes in the lung sound distribution accompanied by pulmonary function improvements in COPD patients were observed after bronchodilator inhalation. We investigated the association of lung sound distribution with pulmonary functions, and the effects of emphysematous lesions on this association. These studies were designed to acquire the basic knowledge necessary for the application of lung sound analysis in the physiological evaluation of COPD patients.

Methods

Pulmonary function tests and the percentage of upper- and lower-lung sound intensity (quantitative lung data [QLD]) were evaluated in 47 stable male COPD patients (54 - 82 years of age). In 39 patients, computed tomography taken within 6 months of the study was available and analyzed.

Results

The ratio of lower QLD to upper QLD showed significant positive correlations with FEV₁ %predicted (%FEV₁; ρ = 0.45, p<0.005) and MEF₅₀ %predicted (%MEF₅₀; ρ = 0.46, p<0.005). These correlations were not observed in COPD patients with dominant emphysema (% low attenuation area >40%, n = 20) and were stronger in less emphysematous patients (n = 19, %FEV₁; ρ = 0.64, p<0.005, %MEF₅₀; ρ = 0.71, p<0.001).

Conclusions

In COPD patients, the ratio of lower- to upper-lung sound intensities decreased according to the severity of obstructive changes, although emphysematous lesions considerably affected lung sound distribution.

Physiological acoustic sensing based on accelerometers: a survey for mobile healthcare

This paper reviews the applications of accelerometers on the detection of physiological acoustic signals such as heart sounds, respiratory sounds, and gastrointestinal sounds. These acoustic signals contain a rich reservoir of vital physiological and pathological information. Accelerometer-based systems enable continuous, mobile, low-cost, and unobtrusive monitoring of physiological acoustic signals and thus can play significant roles in the emerging mobile healthcare. In this review, we first briefly explain the operation principle of accelerometers and specifications that are important for mobile healthcare. Applications of accelerometer-based monitoring systems are then presented. Next, we review a variety of accelerometers which have been reported in literatures for physiological acoustic sensing, including both commercial products and research prototypes. Finally, we discuss some challenges and our vision for future development.

Wheeze

Wheeze is a continuous, musical, whistling-like abnormal breath sound that is usually heard in expiration. It is a common respiratory symptom and sign in children and adults; the 1958 British National Cohort study found the cumulative incidence of wheezing illness by age 33 to be 43% ( Strachan et al, 1996 ). This article considers wheeze only in adults.

Joint application of audio spectral envelope and tonality index in an e-asthma monitoring system

This paper presents in detail a recently introduced highly efficient method for automatic detection of asthmatic wheezing in breathing sounds. The fluctuation in the audio spectral envelope (ASE) from the MPEG-7 standard and the value of the tonality index (TI) from the MPEG-2 Audio specification are jointly used as discriminative features for wheezy sounds, while the support vector machine (SVM) with a polynomial kernel serves as a classifier. The advantages of the proposed approach are described in the paper (e.g., detecting weak wheezes, very good ROC characteristics, independence from noise color). Since the method is not computationally complex, it is suitable for remote asthma monitoring using mobile devices (personal medical assistants). The main contribution of this paper consists of presenting all the implementation details concerning the proposed approach for the first time, i.e., the pseudocode of the method and adjusting the values of the ASE and TI parameters after which only one (not two) FFT is required for analysis of a next overlapping signal fragment. The efficiency of the method has also been additionally confirmed by the AdaBoost classifier with a built-in mechanism to feature ranking, as well as a previously performed minimal-redundancy-maximal-relevance test.

A quality assessment of respiratory auscultation material on YouTube

YouTube contains a large volume of medical educational material. This study assessed the quality of respiratory auscultation videos contained in YouTube. Videos were searched for using the terms 'breath sounds', 'respiratory sounds', 'respiratory auscultation' and/or 'lung sounds'. In total, 6,022 videos were located, 36 of which were considered suitable for scoring for video accuracy, comprehensiveness and quality. The average score was 3.32/6 (55.3% ± 1.30). Video score correlated with time-adjusted YouTube metadata: hits per day (0.496, p=0.002) and likes per day (0.534, p=0.001). Video score also correlated with the first search page on which the video was located in the 'breath sounds' and 'lung sounds' searches (-0.571, p=0.001; -0.445, p=0.014, respectively). The quality of videos was variable. Correlation between video score and some metadata values suggests that there is value for their use in judging video quality. However, the large number of videos found and inability to filter these results quickly makes locating educational content difficult.

Tracheal sounds acquisition using smartphones

Tracheal sounds have received a lot of attention for estimating ventilation parameters in a non-invasive way. The aim of this work was to examine the feasibility of extracting accurate airflow, and automating the detection of breath-phase onset and respiratory rates all directly from tracheal sounds acquired from an acoustic microphone connected to a smartphone. We employed the Samsung Galaxy S4 and iPhone 4s smartphones to acquire tracheal sounds from N = 9 healthy volunteers at airflows ranging from 0.5 to 2.5 L/s. We found that the amplitude of the smartphone-acquired sounds was highly correlated with the airflow from a spirometer, and similar to previously-published studies, we found that the increasing tracheal sounds' amplitude as flow increases follows a power law relationship. Acquired tracheal sounds were used for breath-phase onset detection and their onsets differed by only 52 ± 51 ms (mean ± SD) for Galaxy S4, and 51 ± 48 ms for iPhone 4s, when compared to those detected from the reference signal via the spirometer. Moreover, it was found that accurate respiratory rates (RR) can be obtained from tracheal sounds. The correlation index, bias and limits of agreement were r² = 0.9693, 0.11 (-1.41 to 1.63) breaths-per-minute (bpm) for Galaxy S4, and r² = 0.9672, 0.097 (-1.38 to 1.57) bpm for iPhone 4s, when compared to RR estimated from spirometry. Both smartphone devices performed similarly, as no statistically-significant differences were found.

Pulmonary Acoustic Signal Classification Using Autoregressive Coefficients and k-Nearest Neighbor

— Pulmonary acoustic signals provide important information of the condition of the respiratory system. It can be used to assist medical professionals as an alternative diagnosis tool. In this paper, we intend to discriminate between normal (without any pathological condition), Airway Obstruction (AO) pathology and Interstitial lung disease (ILD) pathology using pulmonary acoustic signals. The proposed method filters the heart sounds and other artifacts using a butterworth bandpass filter and windowed to 256 samples per segment. The autoregressive coefficients (AR coefficients) were extracted as features from the pulmonary acoustic signals. The extracted features are distinguished using k-nearest neighbor (k-nn) classifier. The classifier performance is analysed by using confusion matrix technique. A mean classification accuracy of 96.12% was reported for the proposed method. The performance analysis of the knn classifier using confusion matrix revealed that normal, AO and ILD pathology are classified at 94.36%, 95.18% and 94.68% classification accuracy respectively. The analysis reveals that the proposed method performs better in distinguishing between the normal, AO and ILD.Keywords—Respiratorysound,ARcoefficients,k-nearestneighbor,confusionmatrix

A comparative study of the SVM and K-nn machine learning algorithms for the diagnosis of respiratory pathologies using pulmonary acoustic signals

Background

Pulmonary acoustic parameters extracted from recorded respiratory sounds provide valuable information for the detection of respiratory pathologies. The automated analysis of pulmonary acoustic signals can serve as a differential diagnosis tool for medical professionals, a learning tool for medical students, and a self-management tool for patients. In this context, we intend to evaluate and compare the performance of the support vector machine (SVM) and K-nearest neighbour (K-nn) classifiers in diagnosis respiratory pathologies using respiratory sounds from R.A.L.E database.

Results

The pulmonary acoustic signals used in this study were obtained from the R.A.L.E lung sound database. The pulmonary acoustic signals were manually categorised into three different groups, namely normal, airway obstruction pathology, and parenchymal pathology. The mel-frequency cepstral coefficient (MFCC) features were extracted from the pre-processed pulmonary acoustic signals. The MFCC features were analysed by one-way ANOVA and then fed separately into the SVM and K-nn classifiers. The performances of the classifiers were analysed using the confusion matrix technique. The statistical analysis of the MFCC features using one-way ANOVA showed that the extracted MFCC features are significantly different (p < 0.001). The classification accuracies of the SVM and K-nn classifiers were found to be 92.19% and 98.26%, respectively.

Conclusion

Although the data used to train and test the classifiers are limited, the classification accuracies found are satisfactory. The K-nn classifier was better than the SVM classifier for the discrimination of pulmonary acoustic signals from pathological and normal subjects obtained from the RALE database.

Modeling the pharyngeal anatomical effects on breathing resistance and aerodynamically generated sound

The objective of this study was to systematically assess the effects of pharyngeal anatomical details on breathing resistance and acoustic characteristics by means of computational modeling. A physiologically realistic nose-throat airway was reconstructed from medical images. Individual airway anatomy such as the uvula, pharynx, and larynx was then isolated for examination by gradually simplifying this image-based model geometry. Large eddy simulations with the FW-H acoustics model were used to simulate airflows and acoustic sound generation with constant flow inhalations in rigid-walled airway geometries. Results showed that pharyngeal anatomical details exerted a significant impact on breathing resistance and energy distribution of acoustic sound. The uvula constriction induced considerably increased levels of pressure drop and acoustic power in the pharynx, which could start and worsen snoring symptoms. Each source anatomy was observed to generate a unique spectrum with signature peak frequencies and energy distribution. Moreover, severe pharyngeal airway narrowing led to an upward shift of sound energy in the high-frequency range. Results indicated that computational aeroacoustic modeling appeared to be a practical tool to study breathing-related disorders. Specifically, high-frequency acoustic signals might disclose additional clues to the mechanism of apneic snoring and should be included in future acoustic studies.

Assessment of time-frequency representation techniques for thoracic sounds analysis

A step forward in the knowledge about the underlying physiological phenomena of thoracic sounds requires a reliable estimate of their time-frequency behavior that overcomes the disadvantages of the conventional spectrogram. A more detailed time-frequency representation could lead to a better feature extraction for diseases classification and stratification purposes, among others. In this respect, the aim of this study was to look for an omnibus technique to obtain the time-frequency representation (TFR) of thoracic sounds by comparing generic goodness-of-fit criteria in different simulated thoracic sounds scenarios. The performance of ten TFRs for heart, normal tracheal and adventitious lung sounds was assessed using time-frequency patterns obtained by mathematical functions of the thoracic sounds. To find the best TFR performance measures, such as the 2D local (ρ(mean)) and global (ρ) central correlation, the normalized root-mean-square error (NRMSE), the cross-correlation coefficient (ρ(IF)) and the time-frequency resolution (res(TF)) were used. Simulation results pointed out that the Hilbert-Huang spectrum (HHS) had a superior performance as compared with other techniques and then, it can be considered as a reliable TFR for thoracic sounds. Furthermore, the goodness of HHS was assessed using noisy simulated signals. Additionally, HHS was applied to first and second heart sounds taken from a young healthy male subject, to tracheal sound from a middle-age healthy male subject, and to abnormal lung sounds acquired from a male patient with diffuse interstitial pneumonia. It is expected that the results of this research could be used to obtain a better signature of thoracic sounds for pattern recognition purpose, among other tasks.

Low-power wearable respiratory sound sensing

Building upon the findings from the field of automated recognition of respiratory sound patterns, we propose a wearable wireless sensor implementing on-board respiratory sound acquisition and classification, to enable continuous monitoring of symptoms, such as asthmatic wheezing. Low-power consumption of such a sensor is required in order to achieve long autonomy. Considering that the power consumption of its radio is kept minimal if transmitting only upon (rare) occurrences of wheezing, we focus on optimizing the power consumption of the digital signal processor (DSP). Based on a comprehensive review of asthmatic wheeze detection algorithms, we analyze the computational complexity of common features drawn from short-time Fourier transform (STFT) and decision tree classification. Four algorithms were implemented on a low-power TMS320C5505 DSP. Their classification accuracies were evaluated on a dataset of prerecorded respiratory sounds in two operating scenarios of different detection fidelities. The execution times of all algorithms were measured. The best classification accuracy of over 92%, while occupying only 2.6% of the DSP's processing time, is obtained for the algorithm featuring the time-frequency tracking of shapes of crests originating from wheezing, with spectral features modeled using energy.

Detecting unilateral phrenic paralysis by acoustic respiratory analysis

The consequences of phrenic nerve paralysis vary from a considerable reduction in respiratory function to an apparently normal state. Acoustic analysis of lung sound intensity (LSI) could be an indirect non-invasive measurement of respiratory muscle function, comparing activity on the two sides of the thoracic cage. Lung sounds and airflow were recorded in ten males with unilateral phrenic paralysis and ten healthy subjects (5 men/5 women), during progressive increasing airflow maneuvers. Subjects were in sitting position and two acoustic sensors were placed on their back, on the left and right sides. LSI was determined from 1.2 to 2.4 L/s between 70 and 2000 Hz. LSI was significantly greater on the normal (19.3±4.0 dB) than the affected (5.7±3.5 dB) side in all patients (p = 0.0002), differences ranging from 9.9 to 21.3 dB (13.5±3.5 dB). In the healthy subjects, the LSI was similar on both left (15.1±6.3 dB) and right (17.4±5.7 dB) sides (p = 0.2730), differences ranging from 0.4 to 4.6 dB (2.3±1.6 dB). There was a positive linear relationship between the LSI and the airflow, with clear differences between the slope of patients (about 5 dB/L/s) and healthy subjects (about 10 dB/L/s). Furthermore, the LSI from the affected side of patients was close to the background noise level, at low airflows. As the airflow increases, the LSI from the affected side did also increase, but never reached the levels seen in healthy subjects. Moreover, the difference in LSI between healthy and paralyzed sides was higher in patients with lower FEV₁ (%). The acoustic analysis of LSI is a relevant non-invasive technique to assess respiratory function. This method could reinforce the reliability of the diagnosis of unilateral phrenic paralysis, as well as the monitoring of these patients.

Assessment of regional ventilation distribution: comparison of vibration response imaging (VRI) with electrical impedance tomography (EIT)

Background

Vibration response imaging (VRI) is a bedside technology to monitor ventilation by detecting lung sound vibrations. It is currently unknown whether VRI is able to accurately monitor the local distribution of ventilation within the lungs. We therefore compared VRI to electrical impedance tomography (EIT), an established technique used for the assessment of regional ventilation.

Methodology/Principal Findings

Simultaneous EIT and VRI measurements were performed in the healthy and injured lungs (ALI; induced by saline lavage) at different PEEP levels (0, 5, 10, 15 mbar) in nine piglets. Vibration energy amplitude (VEA) by VRI, and amplitudes of relative impedance changes (rel.ΔZ) by EIT, were evaluated in seven regions of interest (ROIs). To assess the distribution of tidal volume (V_T) by VRI and EIT, absolute values were normalized to the V_T obtained by simultaneous spirometry measurements. Redistribution of ventilation by ALI and PEEP was detected by VRI and EIT. The linear correlation between pooled V_T by VEA and rel.ΔZ was R² = 0.96. Bland-Altman analysis showed a bias of −1.07±24.71 ml and limits of agreement of −49.05 to +47.36 ml. Within the different ROIs, correlations of V_T-distribution by EIT and VRI ranged between R² values of 0.29 and 0.96. ALI and PEEP did not alter the agreement of V_T between VRI and EIT.

Conclusions/Significance

Measurements of regional ventilation distribution by VRI are comparable to those obtained by EIT.

Vibration response imaging: protocol for a systematic review

BACKGROUND

The concept of lung sounds conveying information regarding lung physiology has been used extensively in clinical practice, particularly with physical auscultation using a stethoscope. Advances in computer technology have facilitated the construction of dynamic visual images derived from recorded lung sounds. Arguably, the most significant progress in this field was the development of the commercially available vibration response imaging (VRI) (Deep Breeze Ltd, Or-Akiva, Israel). This device provides a non-invasive, dynamic image of both lungs constructed from sounds detected from the lungs using surface sensors. In the literature, VRI has been utilized in a multitude of clinical and research settings. This systematic review aims to address three study questions relating to whether VRI can be used as an evaluative device, whether the images generated can be characterized, and which tools and measures have been used to assess these images.

METHODS/DESIGN

This systematic review will involve implementing search strategies in five online journal databases in order to extract articles relating to the application of VRI. Appropriate articles will be identified against a set of pre-determined eligibility criteria and assessed for methodological quality using a standardized scale. Included articles will have data extracted by the reviewers using a standardized evidence table. A narrative synthesis based on a standardized framework will be conducted, clustering evidence into three main groups; one for each of the study questions. A meta-analysis will be conducted if two or more research articles meet pre-determined criteria that allow quantitative synthesis to take place.

DISCUSSION

This systematic review aims to provide a complete overview of the scope of VRI in the clinical and research settings, as well as to discuss methods to interpret the data obtained from VRI. The systematic review intends to help clinicians to make informed decisions on the clinical applicability of the device, to allow researchers to identify further potential avenues of investigation, and to provide methods for the evaluation and interpretation of dynamic and static images. The publication and registration of this review with PROSPERO provides transparency and accountability, and facilitates the appraisal of the proposed systematic review against the original design.

TRIAL REGISTRATION

PROSPERO registration number: CRD42013003751.

Diagnosis of hidden bronchial obstruction using computer-assessed tracheal forced expiratory noise time

BACKGROUND AND OBJECTIVE

Increased forced expiratory time was first recognized as a marker of obstruction half a century ago. However, the reported diagnostic capabilities of both auscultated forced expiratory time (FET(as)) and spirometric forced expiratory time are contradictory. Computer analysis of respiratory noises provides a precise estimation of acoustic forced expiratory noise time (FET(a)) being the object-measured analogue of FET(as). The aim of this study was to analyse FET(a) diagnostic capabilities in patients with asthma based on the hypothesis that FET(a) could reveal hidden bronchial obstruction.

METHODS

A group of asthma patients involved 149 males aged 16-25 years. In this group, 71 subjects had spirometry features of bronchial obstruction, meanwhile, the remaining 78 had normal spirometry. A control group involved 77 healthy subjects. Spirometry and forced expiratory tracheal noise recording were sequentially measured for each participant. FET(a) values were estimated by means of a developed computer procedure, including bandpass filtration (200-2000 Hz), waveform envelope calculation with accumulation period of 0.01 s, automated measurement of FET(a) at 0.5% level from the peak amplitude.

RESULTS

Specificity, sensitivity and area under Receiver Operating Characteristic curve of FET(a) and its ratios to squared chest circumference, height, weight were indistinguishable with baseline spirometry index FEV1 /forced vital capacity. Meanwhile, acoustic features of obstruction were revealed in 41%-49% of subgroup of patients with asthma but normal spirometry.

CONCLUSIONS

FET(a) of tracheal noise and its ratio to anthropometric parameters seem to be sensitive and specific tests of hidden bronchial obstruction in young male asthma patients.

A new method to predict values for postoperative lung function and surgical risk of lung resection by quantitative breath sound measurements

OBJECTIVES

We evaluated quantitative acoustic measurements, as a simpler alternative to perfusion scintigraphy, for estimation of predicted postoperative (ppo) lung function after resection surgery in our patient population.

METHODS

Patients with lung cancer, considered as candidates for lung resection, were enrolled in the study. All patients underwent lung function testing and quantitative breath sound testing by vibration response imaging (VRI) on the same day. A subset of patients also had perfusion testing. Forced expiratory volume in 1 second (FEV(1)) and diffusing capacity of the lung for carbon monoxide (DLCO) predictions derived from VRI testing were compared with perfusion values and actual FEV(1) values at 1 month postoperatively.

RESULTS

Fifty-three subjects (40 males; age 66±8 y) participated. There was high correlation between both methods for the calculation of ppoFEV(1)% (R=0.94; n=39) and ppoFEV (L) (R=0.90; n=39). PpoFEV(1) were 58±18% versus 56±20% and 1.69±0.49 L versus 1.62±0.52 L, based on perfusion and VRI methods, respectively. In 92% (36/39) of calculations, the difference between the 2 methods was <10%. High correlations also existed between VRI and perfusion for the calculation of ppoDLCO% (R=0.95; n=37) and ppoDLCO mL/min/mm Hg (R=0.90; n=37). VRI predictions showed good correlation for the 34 patients with actual postoperative lung function (R=0.88 and R=0.80 for FEV(1)% and FEV(1)L, respectively). Accuracy of the VRI to predict surgical risk (<40% cutoff threshold for ppo values) compared with actual postoperative values was 85% (29/34).

CONCLUSIONS

Predictions of postoperative lung function using VRI agree well with radionuclide techniques and actual measured postoperative values. VRI may provide a noninvasive, simpler alternative for estimation of ppo values, particularly when perfusion testing is not readily available.

Vibration response imaging in healthy Japanese subjects

BACKGROUND

Vibration response imaging (VRI) records the intensity and distribution of lung sounds during the respiration cycle. Our objective was to analyze VRI findings in healthy Japanese adults.

METHODS

VRI images of 106 healthy subjects (33.7±9.6 years, 52 male and 54 female), including 67 nonsmokers and 39 asymptomatic smokers, were recorded. The regional intensity of vibrations was assessed using quantitative lung data (QLD), and VRI dynamic images by rater assessment, left and right lung asynchrony (gap index), and regional lung asynchrony (asynchrony score).

RESULTS

A dominance of total left lung QLD was observed in all subjects, and this phenomenon was more prominent in female subjects. However, there was no significant difference between the total left and total right lung QLD in smokers. Rater assessments showed that 81.1% of all subjects had a normal final assessment. Male subjects had a significantly higher percentage of good or normal assessments for all image scores, except dynamic image scoring. The asynchrony score was significantly higher in female subjects. There were no significant differences in these qualitative assessments between non-smokers and smokers.

CONCLUSIONS

Although our QLD results were similar to those of a previous report, there were discrepancies between sexes for the qualitative assessments. A significantly higher number of female subjects had abnormal images as assessed by the raters. Furthermore, significantly higher asynchrony scores were observed in female subjects. The VRI variability in sex may be considered normal among the Japanese population. This study is registered with UMIN-CTR under registration number UMIN000002355.

Computerized analysis of respiratory sounds during COPD exacerbations

Acute exacerbation of chronic obstructive pulmonary disease (AECOPD) is a major event in the natural course of the disease, and is associated with significant mortality and socioeconomic impact. Abnormal respiratory sounds are commonly present in patients with AECOPD. Computerized analysis of these sounds can assist in diagnosis and in evaluation during follow-up. Exploratory data analysis methods were applied to respiratory sounds in these patients when they were hospitalized because of exacerbation. Two different patterns of presentation and evolution of respiratory sounds in AECOPD were found and described from the method of computerized respiratory sound analysis and unsupervised clustering that was devised. Based on the findings of the study, remote monitoring of respiratory sounds may be useful for the detection and/or follow-up of COPD exacerbation.

Variation in family physicians' recording of auscultation abnormalities in patients with acute cough is not explained by case mix. A study from 12 European networks

BACKGROUND

Conflicting data on the diagnostic and prognostic value of auscultation abnormalities may be partly explained by inconsistent use of terminology.

OBJECTIVES

To describe general practitioners use of chest auscultation abnormality terms for patients presenting with acute cough across Europe, and to explore the influence of geographic location and case mix on use of these terms.

METHODS

Clinicians recorded whether 'diminished vesicular breathing', 'wheezes', 'crackles' and 'rhonchi' were present in an observational study of adults with acute cough in 13 networks in 12 European countries. We describe the use of these terms overall and by network, and used multilevel logistic regression to explore variation by network, controlling for patients' gender, age, comorbidities, smoking status and symptoms.

RESULTS

2345 patients were included. Wheeze was the auscultation abnormality most frequently recorded (20.6% overall) with wide variation by network (range: 8.3-30.8%). There was similar variation for other auscultation abnormalities. After controlling for patient characteristics, network was a significant predictor of auscultation abnormalities with odds ratios for location effects ranging from 0.37 to 4.46 for any recorded auscultation abnormality, and from 0.25 to 3.14 for rhonchi.

CONCLUSION

There is important variation in recording chest auscultation abnormalities by general practitioners across Europe, which cannot be explained by differences in patient characteristics. There is a need and opportunity for standardization in the detection and classification of lung sounds.