Publication:

Predict cardiac death after withdrawal of life-supporting treatment in the neurocritical patients: a neurological score

Background Accurate prediction the time of death after withdrawal of life-supporting treatment (WLST) in the neurocritical patients is important for donation after cardiac death. We aimed to establish a new neurological scoring system to predict the probability of death within 60 minutes after WLST.

Methods We retrospectively reviewed the clinical data from a cohort of 231 neurocritical patients from June 2016 to July 2018. The patients were divided into training and external validation sets. In the training set, we used univariate and multivariable logistic regression analyses to assess associations between death within 60 minutes after WLST and these variables. Points attributed to each variable were summed to create a predictive score for cardiac death. We assessed performance of the score using ROC analysis. Finally, we validated the predictive ability of the score in an external validation set.

Results Multivariable logistic regression analysis showed that absent corneal reflexes, absent cough reflexes, extensor or absent motor response and anisocoric or bilaterally dilated pupil were associated with death within 60 minutes in training set (all p<0.05). The area under curve for the score was 0.839 (95% CI 0.763-0.916) for prediction of death within 60 minutes. A score of 4-6 had a sensitivity of 95.88% and a specificity of 69.05% to predict death within 60 minutes after WSLT. In external validation set, the prevalence of these predictors was similar with training sample.

Conclusions The neurological score may be useful to predict time of death after WLST in neurocritical patients.