Department of Large Animal Internal Medicine, Faculty of Veterinary Medicine, Ghent University, Merelbeke, Belgium.
Sciences cliniques, Faculte de Medecine Veterinaire, Saint-Hyacinthe, Quebec, Canada.
Vet Rec. 2019 Jul 27;185(4):109. doi: 10.1136/vr.105238. Epub 2019 Jul 18.
In practice, veterinary surgeons frequently rely on lung auscultation as a confirmation test for pneumonia. To what extent diagnostic accuracy of lung auscultation varies between different practitioners is currently unknown. In this diagnostic test study, 49 Dutch veterinarians each auscultated between 8 and 10 calves, and communicated whether they would decide to treat the animal with antimicrobials or not. They were not allowed to perform any other aspect of the clinical examination. Their decisions were compared with lung ultrasonography findings. The average sensitivity and specificity of lung auscultation were 0.63 (sd=0.2; range=0.2-1.0) and 0.46 (sd=0.3; range=0.0-1.0), respectively. Of the participants, 8.2 per cent were 100 per cent sensitive, 16.3 per cent were 100 per cent specific, and only 4.0 per cent were perfect. The Krippendorff's alpha was 0.18 (95 per cent confidence interval: -0.01 to 0.38), signifying poor reliability between multiple raters. Regardless of the poor diagnostic accuracy in this study, especially the large variation in a confirmation test between different practitioners could potentially cause professional damage as well as misuse of antimicrobials. This study could be seen as a gentle stimulus to regularly evaluate one's diagnostic skills. Both complementary training and the use of more accurate techniques with less inter-rater variation could improve the situation.
在实践中，兽医经常依赖肺部听诊作为肺炎的确认测试。目前尚不清楚不同从业者之间肺部听诊的诊断准确性差异有多大。在这项诊断测试研究中，49 名荷兰兽医每人听诊了 8 到 10 头小牛，并沟通他们是否会决定用抗生素治疗动物。他们不允许进行临床检查的任何其他方面。他们的决定与肺部超声检查结果进行了比较。肺部听诊的平均敏感度和特异性分别为 0.63（标准差=0.2；范围=0.2-1.0）和 0.46（标准差=0.3；范围=0.0-1.0）。在参与者中，8.2%的人敏感度为 100%，16.3%的人特异性为 100%，只有 4.0%的人是完美的。Krippendorff's alpha 为 0.18（95%置信区间：-0.01 至 0.38），表示多个评分者之间的可靠性较差。尽管本研究中的诊断准确性较差，尤其是不同从业者之间确认测试的差异较大，但这可能会导致专业损害以及抗生素的滥用。这项研究可以被视为对定期评估诊断技能的温和刺激。补充培训和使用更准确、评分者间差异更小的技术可以改善这种情况。
