Accuracy of radiographic diagnosis of pneumoperitoneum secondary to gastrointestinal perforation in dogs and cats

Abstract

Background: Radiography is commonly included in the diagnostic investigation of animals with suspected gastrointestinal perforation and resultant pneumoperitoneum. This study aimed to (1) identify the accuracy of radiographic diagnosis of pneumoperitoneum and (2) determine if observer experience affected accuracy. Methods: This was a retrospective case-controlled study evaluating abdominal radiographs of dogs and cats with surgical confirmation of gastrointestinal perforation or peritonitis without pneumoperitoneum. Radiographs were reviewed by a radiologist, an imaging resident, a general practitioner and a veterinary student. Results: Radiographs from 60 dogs and eight cats were evaluated; 34 animals had confirmed perforation. Accuracy was fair to excellent for all observers: student 61.8%, general practitioner 70.6%, resident 85.3% and radiologist 83.8%. There was no significant difference in sensitivity between observers (all 70.6%-85.3%); however, there was a significant difference in specificity between both the resident (91.2%) and radiologist (91.2%) compared to the student (52.9%) (p < 0.001) and the general practitioner (55.9%) (p = 0.002). Overall, there was slight to substantial agreement between observers (κ = 0.28-0.73), with the highest being between the radiologist and resident. Conclusion: Abdominal radiography, when interpreted by veterinarians with differing experience, has variable accuracy for diagnosis of pneumoperitoneum for animals in this study. Further studies are needed to determine if these results are applicable for other populations of animals with pneumoperitoneum.