Blunt Trauma: What Is Behind the Widened Mediastinum on Chest X-Ray (CXR)?

Abstract

BackgroundIt is commonly taught that a widened mediastinum (WM) on chest X-ray (CXR) is a marker for aortic injury (AI). We sought to describe the epidemiology of injuries for all patients with WM and compare their CXR to those of patients with confirmed AI. MethodsAdults (age ≥ 18) sustaining blunt traumatic injuries from January 2017 to June 2017 with both CXR (supine, anterior-posterior) and chest CT were included. We excluded those whose CT preceded CXR and those with missing data. Basic demographics, injury characteristics, mediastinal width (MW), mediastinal-to-thoracic width ratio (MTR), and all thoracic imaging findings were analyzed. MW > 8 cm was considered WM. We also queried our registry for all AI patients over a 4-year period. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of WM on CXR for AI were calculated. Multivariate logistic regression was performed to identify factors associated with positive traumatic findings, controlling for body mass index (BMI), sex, high-energy mechanism, MTR, and mediastinal width. ResultsOf 749 included subjects, 502 (67%) had an MW > 8 cm: mean age was 48 ± 20 y, 381 (76%) were men, and BMI was 28 ± 5 kg/m2. Mechanism of injury was motor vehicle crash in 335 (67%); fall in 113 (23%); assault in 31 (6%); other (jet-ski accidents, etc.) in 17 (3%), and unknown in 6 (1%). Only 128 (26%) of patients with WM had positive findings on CT, with the most common [80 (16%)] being nontraumatic findings (pericardial infusion, lymph nodes, etc.), followed by hemomediastinum/pneumomediastinum [32 (6%)], sternal fractures [18 (4%)], multiple findings [15 (3%)], and vertebral fractures [6 (1%)]. Only 2 (1%) had AI. The sensitivity was 100%, specificity was 33%, PPV was 0.4%, NPV was 100%, and accuracy was 33%. From 2013 to 2017, 38 patients had AI: mean age was 46 ± 19 y, 26 (68%) were men, and BMI was 28 ± 4 kg/m2. Motor vehicle crash was the most common mechanism (89%), followed by “other” trauma mechanism (5%), fall (3%), and assault (3%). On univariate analysis, compared with all patients with WM, patients with AI had significantly greater MW (9.5 [8.8-10.4] versus 10.2 [9.1-11.1]; P = 0.042) and MTR (0.31 [0.28-0.34] versus 0.32 [0.31-0.37]; P = 0.001), although the actual differences were not clinically significant. The regression analysis did not identify any factors associated with traumatic CXR findings. ConclusionsMost bluntly injured adults have a WM, and the majority have either no findings or nontraumatic findings. The PPV of a WM for AI is <1%. WM on supine CXR is nonspecific and inaccurate for diagnosing traumatic injuries, especially AI.