Chapter 23 — Neurological Outcomes and Prognostication

101 kB. 13 sections, 38 callouts, 11 tables, 1 mermaid flowchart. Three controversies: enolase on the circuit (23.4), the electroencephalogram's two jobs (23.6), when prediction may begin (23.9). Central claim: every specificity figure in this field was generated in systems that withdraw treatment on the basis of the test being evaluated; the two no-withdrawal cohorts (ProNeCA 2019, Gotze 2026) reproduced the rule but not the certainty. Five numerical/contradiction flags printed uncorrected - the ECPR score's goodness-of-fit indicating poor calibration presented as support; its validation point estimate on its own lower bound; the three-way enolase contradiction; the detection gradient (16.5% registry vs 68% autopsy; 3% vs 63% seizures); and both decreased and increased grey-white differentiation indicating poor outcome. Structural finding: the largest ECPR imaging study (SAVE-J II, 1,146 patients) produces an AUC of 0.628 and argues against its own instrument. All external evidence from structured abstracts; every seeded database row Verified = No. 12 evidence rows and 1 guideline row added. Citation audit run against the tracker after five wrong chapter cross-references were found and repaired in published Chapters 20, 21 and 22.