Aortic dissection: chest x-ray: look for widening of the aortic knob, mediastinum or descending aorta.

Clinical bottom line (level 4)

  1. Radiologists guessed at most features on a chest x-ray for a patient with suspected aortic dissection.
  2. The features they agreed best on were widening of the aortic knob, mediastinum or descending aorta.
  3. A positive chest x-ray made aortic dissection more likely, whilst a negative one made it less likely. Neither could diagnose or rule out aortic dissection.
Jagannath et al: American Journal of Roentgenology 1986; 147: 1123-1126
Expires December 2004

The study

Setting: university hospital, USA

72 patients (aged range 14 to 87 years; mean 63, ?% male) 36 with angiographically proven aortic dissection, and 36 controls with hypertension and/ or atherosclerotic cardiovascular disease

Independent blinded reference standard, applied in all patients from a consecutive inappropriate spectrum.
Diagnostic test: chest x-ray (majority were PA or AP portable). Five blinded radiologists were told that the patients had "suspected aortic dissection". They looked for a number of set signs and asked to rank as definitely absent, probably absent, possibly absent, possibly present, probably present or definitely present. radiologists also made a final conclusion for chest x-ray.
  • Stepwise multiple logistic regression was used to construct sensitivity and specificity for all the features.

The evidence


diagnostic test number of patients sensitivity for

(95% CI)		 specificity for

(95% CI)		 LR+ LR-
logistic regression on features 81%
(% to %) 		75%
(% to %) 		3.2 0.25
radiologist conclusion 81%
(% to %) 		89%
(% to %) 		7.3 0.21
total

widening aortic knob: K=0.33
widening of mediastinum: K=0.28
widening of descending aorta: K=0.28
pleural effusion: K=0.27
tracheal shift: K=0.27
widening of ascending aorta: K=0.26
parasternal shadow widening: K=0.25
blurring of aortic knob: K=0.23
displacement of intimal calcification >6 mm: K=0.13
localising lump on aortic arch distal to great vessels: K=0.09
overall: K=0.24

Comments

  1. How does this compare with real life? Other trials suggest pre-test probability is ~50-60%, so the study represents the mix reasonably well.

Citation

  1. Jagannath AS, Sos TA, Lockhart SH, et al: Aortic dissection: A statistical analysis of the usefulness of plain chest radiographic findings. American Journal of Roentgenology 1986; 147: 1123-1126
Contributor: Chris Ball and Clare Wotton, December 2000
Reviewer:

Clinical Question.
Patient suspected aortic dissection
Intervention or Exposure chest x-ray
Outcome radiologists agreement and diagnosis