Diagnostic performance of lung ultrasound compared to thoracic radiography in non-traumatized dogs and cats with respiratory distress in an emergency setting.

Introduction: Respiratory distress is a medical emergency requiring rapid diagnostic decision-making. This prospective observational study evaluated the efficiency and diagnostic performance of lung ultrasound (LUS) and thoracic radiography (TXR) in dogs and cats presenting with tachypnea. Imaging-based suspected diagnoses were compared with definitive diagnoses established by specialists using additional diagnostic procedures. The influence of observer experience on diagnostic accuracy was assessed. Methods: Client-owned animals with suspected intrathoracic causes of respiratory distress underwent both LUS and TXR within 24 h of presentation, with a maximum interval of 1 h between examinations. Imaging was performed according to standardized protocols. Based on imaging findings, the suspected cause of breathing difficulties was categorized as: no detectable intrathoracic cause, cardiac-related respiratory distress, pulmonary disease, neoplasia, or pleural effusion of unknown etiology. Following stabilization, patients were referred to specialists for establishment of the definitive diagnosis. Concordance between suspected and definitive diagnoses was analyzed separately for each modality to assess diagnostic performance. Selected cases were re-evaluated in a blinded fashion by observers with three different levels of experience. Results: A total of 144 animals (83 dogs and 61 cats) with acute respiratory distress were included. The suspected diagnosis based on LUS was concordant with the definitive diagnosis in 80.6% of cases, while TXR achieved 89.1%. Both modalities were correct in 77.5% of cases, and at least one modality was correct in 92.2%. Diagnostic accuracy of TXR was statistically significantly higher than that of LUS (absolute difference 8.5%). In dogs, the presence of a heart murmur was strongly associated with a cardiac cause of respiratory distress (odds ratio 10.9), whereas this association was not statistically significant in cats. In the blinded re-evaluation, TXR demonstrated higher interobserver agreement (Fleiss' kappa = 0.69). The highest agreement was achieved by a less experienced observer, with a Cohen's kappa of 0.8 for TXR. Conclusion: Both LUS and TXR are valuable imaging modalities for initial classification of respiratory distress, each offering specific advantages depending on the underlying pathology. In most cases, these techniques provide sufficient information to guide initial therapeutic decisions; however, a multimodal approach is recommended for complex presentations.