Brain death determination sits at the fraught intersection of critical-care medicine, ethics, and organ-donation logistics. Missteps risk either prolonging mechanical support with no prospect of recovery or declaring death in error—a prospect that evokes the terrors of many Edgar Allan Poe writings.
A 15-ICU multicenter validation effort led by Université de Montréal to test CT scans as a brain-death diagnostic tool reports that brainstem CT imaging alone cannot confirm neurological death criteria.
Clinical bedside examinations use the apparent absence of consciousness, brainstem reflexes, and the drive to breathe as the standard in death determination. Sedatives, facial trauma, or metabolic disturbances can cloud that picture. Many centers have turned to computed-tomography flow imaging, betting that a scan can settle what the stethoscope cannot.
In the study, “Computed Tomography Perfusion and Angiography for Death by Neurologic Criteria,” published in JAMA Neurology, researchers conducted a prospective multicenter blinded diagnostic-accuracy cohort study to test whether CT perfusion and CT angiography can reliably confirm brain death.
A cohort of 282 critically ill adults in 15 Canadian intensive-care units underwent contrast-enhanced brain CT perfusion and CT angiography within two hours of a blinded standardized bedside examination.
Imaging covered the whole brain after 40 mL of iodinated contrast. Two independent neuroradiologists, masked to all clinical data, judged qualitative and quantitative perfusion maps and three angiography scoring scales. Clinicians, blinded to imaging, applied current guidelines for brain death testing, including an apnea test, which served as the reference standard.
Qualitative brainstem CT perfusion showed 98.5% sensitivity yet only 74.4% specificity. Qualitative whole-brain CT perfusion achieved 93.6% sensitivity and 92.3% specificity. CT angiography sensitivity ranged from 75.5–87.3%, while specificity held near 90% across scales.
Sensitivity refers to how well a test finds its target, in this case, brain death. In the brainstem CT, it found nearly all of the target cases, missing just 1.5%. Not bad.
Specificity refers to how well a test identifies non-targets, in this case those who are not brain-dead. Here the brainstem CT correctly identified 74.4% of non-cases, meaning that it also identified about 25% of non-brain-dead patients as being brain-dead. Not great.
Lead investigators conclude that neither CT perfusion nor CT angiography met the pre-set validation hurdle of >98 % for both sensitivity and specificity.
This false-positive burden caused by specificity is why diagnostic tests are not given to the general public without previous clinical assessments. For example, if a test for a rare disease that affects one out of 1,000 people is given to 1,000 random individuals, with 98% sensitivity and a high specificity of 98%, it will likely capture one individual with the rare disease and falsely assign a positive result to 20 people without it.
If the test is only given to individuals who meet clinical criteria for having the rare disease, the chance of the test giving an accurate representation increases dramatically. If by clinical selection where 900 out of the 1,000 actually do have the rare disease, it would correctly confirm 882 cases with only two false positives.
In the current validation of determining brain death among a selected group, the test failed as a stand-alone replacement. Imaging can still play an important role, but should serve only as supportive evidence when clinical assessment is incomplete or confounded.