Limitations of computed tomographic angiography in the diagnosis of brain death

Objective: To evaluate the accuracy of cerebral computed tomographic angiography (CT-a) for the diagnosis of brain death ( BD). Design and setting: Prospective observational study in intensive care units. Patients: Twenty-one clinically BD patients enrolled over 12 months. Measurements and results: All clinically BD patients were evaluated by electroencephalography ( EEG) and CT-a after exclusion of hypothermia and drug intoxication. Data collected included: demographic characteristics, cause of BD, delay between in-hospital admission and BD diagnosis and between EEG and CT-a, occurrence of cardiac arrest, administration of vasoactive agents, results of EEG and CT-a. We evaluated the sensitivity of EEG and CT-a and their agreement. Groups were compared according to BD diagnosis by EEG and CT-a (E+ C+), or only by EEG (E+ C-). Statistical analysis were performed by Mann-Whitney test and Fisher's exact test. BD was confirmed by EEG in all cases (sensitivity 100%) whereas only 11 patients of 21 had no cerebral perfusion during CT-a (sensitivity 52.4%). No agreement was documented between EEG and CT-a for the diagnosis of BD (kappa = 0). Patients' characteristics did not differ between E+ C+ and E+ C- groups. In the E+ C- group arterial opacification was observed in 100% of patients, but opacification of the internal cerebral veins was achieved in only 30%. Conclusions: In clinically BD patients with no electroencephalographic activity CT-a documents opacification of the intracerebral vessels in a significant percentage of the cases. Therefore CT-a cannot be recommended as a means of BD diagnosis.