Sensitivity, specificity, and interobserver variability of survey thoracic radiography for the detection of heart base masses in dogs

OBJECTIVE To determine the sensitivity, specificity, and interobserver variability of survey thoracic radiography (STR) for the detection of heart base masses (HBMs) in dogs. DESIGN Retrospective case-control study. ANIMALS 30 dogs with an HBM and 120 breed-matched control dogs (60 healthy dogs and 60 dogs with heart disease and no HBM). PROCEDURES In a blinded manner, 2 observers (designated as A and B) evaluated STR views from each dog for a mass-like opacity cranial to the heart, tracheal deviation, cardiomegaly, findings suggestive of pericardial effusion or right-sided congestive heart failure, and soft tissue opacities suggestive of pulmonary metastases. Investigators subsequently provided a final interpretation of each dog's HBM status (definitely affected, equivocal, or definitely not affected). RESULTS Considering equivocal interpretation as negative or positive for an HBM, the sensitivity of STR for diagnosis of an HBM was 40.0% (95% confidence interval CI, 22.5% to 57.5%) and 56.7% (95% CI, 38.9% to 74.4%), respectively, for observer A and 63% (95% CI, 46.1% to 80.6%) and 80.0% (95% CI, 65.7% to 94.3%), respectively, for observer B. The corresponding specificity was 96.7% (95% CI, 93.5% to 99.9%) and 92.5% (95% CI, 87.8% to 97.2%), respectively, for observer A and 99.2% (95% CI, 97.5% to 100%) and 92.5% (95% CI, 87.8% to 97.2%), respectively, for observer B. The presence of a mass-like opacity cranial to the heart or tracheal deviation, or both, was significantly associated with a true diagnosis of HBM. CONCLUSIONS AND CLINICAL RELEVANCE Results indicated that STR is a highly specific but not a highly sensitive predictor of HBM in dogs.