Diagnostic Accuracy of Plain Radiography to Identify Synovial Penetration in Horses with Traumatic Limb Wounds

Background: Radiography is commonly used in the assessment of limb wounds. The diagnostic value of plain radiography for diagnosing synovial penetration in limb wounds near synovial structures has not been determined. 
Objectives: To assess the sensitivity, specificity, and positive and negative predictive values of plain radiography to identify synovial penetration in horses with traumatic limb wounds. 
Study design: Retrospective case study. 
Methods: Case records of horses presented to Ghent University between 2011 and 2021 with limb wounds near synovial structures were categorised in groups with and without synovial penetration using synovial fluid cytology and surgical findings. Cases with incomplete information were excluded. Plain radiographic images were evaluated by a radiologist blinded to clinical information. Sensitivity, specificity, positive and negative predictive values were calculated comparing plain radiography to the gold standard of synovial fluid cytology and surgical findings in horses with limb wounds near synovial structures. 
Results: One hundred and forty-one horses with and 75 cases without synovial involvement were included. Plain radiography had a sensitivity of 61% (95%CI: 52.4-69.1%), a specificity of 81% (95%CI: 70.7-89.4%), a positive predictive value of 86% (95%CI: 79.0-90.9%) and a negative predictive value of 53% (95%CI: 46.8-58.3%). 
Main limitations: Sample size did not allow comparing different synovial structures. Results can be influenced by individual case characteristics. The number of available radiographic views differed between cases, and radiographs with a metallic probe placed in the wound were not always available. 
Conclusions: Based on moderate sensitivity and low negative predictive value, care should be taken before concluding that synovial involvement is unlikely based on the absence of relevant findings on plain radiography alone. The high specificity and high positive predictive value indicate that plain radiography is most reliable when radiographic findings suggest synovial penetration. Therefore, plain radiography should be interpreted cautiously during clinical decision making. 
Ethical animal research: Research ethics committee oversight not required: retrospective review of medical records. 
Informed consent: Not stated. 
Competing interests: None declared. 
Sources of funding: Not applicable.

[...]