Programme no. 263-P
Professional Development
Inter-observer variation in categorizing lung sounds. A comparison between experts, lung specialists and general practitioners.
Juan Carlos Aviles Solis*1, Hasse Melbye2, Peder Halvorsen3
1Institutt for Samfunnsmedisin,Universitet i Tromsø,Tromsø,Norway, 2Institutt for Samfunnsmedisin,Universitet i Tromsø,Tromsø,Norway, 3Institutt for Samfunnsmedisin,Universitet i Tromsø,Tromsø,Norway
* = Presenting author

To measure the variability in the categorization of lung sounds between different observers Measure the variability between observers in different professional circumstances.


The lung auscultation has been an important clinical skill in the everyday practice and the stethoscope itself is a widely recognized symbol of a doctor. Nevertheless, the auscultation and its prognostic value for the diagnosis of lung diseases has been challenged in recent times by new technology that offers information that is considered to be more precise. At present time, some clinicians look at lung auscultation as an unnecessary routine.

The new technology is expensive and requires specialized personnel; this limits the availability in developing and underdeveloped countries. Stethoscopes on the contrary, are cheap instruments and widely available in the world. The proper use of the stethoscope backed up with good evidence-based information could help to the early detection of lung diseases.

The high variability among observers has been one of the reasons that rest diagnostic value to lung auscultation. The cause for the high variation relies partly in the differences in the classification and terminology used by different doctors.

Results: 18 individuals have so far responded the survey. Preliminary analysis shows that the interobserver agreement in the recording of lung sounds vary between groups of doctors. Best agreement has so far been found for Wheezes in a group of experts with ICC= ,712 (95% CI ,634 to ,779; P=,000) and crackles in a group of General practitioners with ICC= ,636 (95% CI ,556 to ,711; P=,000). The complete analysis will be presented at the congress.
Material/Methods: A sample of 20 patients with high probability for presence of abnormal lung sounds was investigated and recordings from 6 different places on the thorax were obtained. These recordings were sent to general practitioners, pulmonologists, and researchers in the field of lung sounds from Canada and several European countries. These professionals categorized each sound according to a survey designed at our research unit. The results were processed for the analysis of inter observer variability using the value of the Intraclass Correlation Coefficient (ICC).

Understanding the variation could help us finding ways to improve it, so the reliability of lung auscultation can improve. In this study, we willll attempt to characterize it and understand it.

Points for discussion:

To determine if the variation comes from experience or professional differences could help us to design interventions that improve the reliability of this ancient technique.