Table 2
Author, date and countryPatient groupStudy type (level of evidence)OutcomesKey resultsStudy weaknesses
Hartnell GG et al, 1993, UKChest radiographs from 18 patient with aortic dissection, and 25 patients with acute myocardial infarctionCase control studySensitivity of chest radiographs interpretation88.9%Unclear how MI chest radiographs were obtained
Chest radiographs reviewed by consultant radiologists1st observer72.2%Only two radiologists interpreted the radiographs. There was a large interobserver variability. It would have been more informative if a larger number of radiologists had been used.
Sensitivity observer 2Radiologists knew that around half the patients had a dissection and half an MI
This study does not use A&E doctors
Vu KH et al, 1994, Australia42 patients with diagnosis of aortic dissectionRetrospective cohortThe following results are calculated using 19 patients with dissection (16 confirmed by surgery/necropsy, but also includes three patients who died without necropsy)52.6%Gold standard not applied to all patients— only 16 had confirmation of dissection at surgery or on postmortem examination. No information given regarding the diagnostic imaging for the other 26.
Sensitivity wide mediastinum47%Retrospective review open to bias
Sensitivity dilated aortic arch17%Chest radiographs reported by a senior radiologist—not A&E doctors
Sensitivity displacement of calcified aortic plaques5%Unclear how patients were identified retrospectively
Tracheal deviation15%
Pleural effusion21%
Hennessy TG et al, 1996, Ireland55 consecutive patients referred with suspected thoracic dissection to a cardiology departmentRetrospective cohortSensitivity wide mediastinum65.7%Retrospective review using case notes.
35 of 55 had a dissecting thoracic aneurysmNormal chest radiograph14 of 35 normalUnclear who reported chest radiographs
Other abnormality7 of 35 showed another abnormalityOnly includes those referred to cardiology department. It is unclear which investigations each patient had undergone before referral
Not clear whether gold standard applied to all patients
Klompas M, 2002, USAPatients with clinically suspected aortic dissection or confirmed dissectionLiterature review including prospective and retrospective cohorts in 21 studiesAbnormal aortic contourSensitivity 61% (CI 56 to 84)Search terms not clearly stated. It would not be possible to repeat this search from the information given
Pleural effusionSensitivity 16% (CI 12 to 21)Vast majority of papers are retrospective reviews of patients with known dissections, unblinded and heavily open to bias
Displaced intimal calcificationSensitivity 9% (CI 6 to 13)Not all studies looked at the same chest radiograph findings
Wide mediastinumSensitivity 64% (CI 44 to 80)
Abnormal chest radiographSensitivity 90% (CI 87 to 92)