The following is the summary of “Predicting neurological outcome in adult patients with cardiac arrest: systematic review and meta-analysis of prediction model performance” published in the December 2022 issue of Critical care by Amacher, et al.

This systematic review and meta-analysis aimed to compare the accuracy of 3 prediction models for the prognosis of neurological outcome after cardiac arrest: 2 developed after the event (out-of-hospital cardiac arrest, OHCA, and cardiac arrest hospital prognosis, CAHP), and one developed before the event (good outcome following attempted resuscitation, GO-FAR). From November 2006 through December 2021, we systematically searched Embase, Medline, and the Web of Science Core Collection, and researchers tracked the citations of high-scoring articles in both databases. 

The search yielded 1’021 records, 25 of which were used in the analysis (with a total of 124’168 patients). Calibration (total observed vs. predicted [O:E] ratio) and C-statistics were combined in a random-effects meta-analysis. The OHCA and CAHP scores both had good discriminatory performance (summary C-statistics of 0.83 [95% CI 0.81-0.85], 16 cohorts and 0.84 [95% CI 0.82-0.87], 14 cohorts, respectively), while the GO-FAR score had an acceptable discriminatory performance (summary C-statistic of 0.78 [95% CI 0.72-0.84], 5 cohorts). Both the OHCA (total O:E ratio: 0.78 [95% CI 0.67-0.92], 9 cohorts) and CAHP score (total O:E ratio: 0.78 [95% CI 0.72-0.84], 9 cohorts) were shown to have satisfactory overall calibration, albeit with an overestimation of poor outcome. 

Calibration of the GO-FAR score was poor overall, leading to an underestimate of the favorable result (total O:E ratio: 1.62 [95% CI 1.28-2.04], 5 cohorts). Prognostic accuracy of 2 post-arrest ratings for predicting neurological outcomes after cardiac arrest was high, which could aid in early talks about goals-of-care and therapeutic planning in the intensive care unit. The predictive accuracy of a pre-arrest score was adequate, suggesting it could be useful in code status discussions.