20, 11, 77
The quantitative counterpart to Chapter 11's 'prognosis is not indication', and the anchor of Chapter 20 §20.6. Every prognostic model treats higher pH and lower lactate as favourable; this analysis found the opposite variables identified the patients ECPR helped most, because a profoundly acidotic patient does very badly without the circuit and merely badly with it. When refusing a patient, know whether you are making a prognostic claim ('they will probably die anyway' — well supported) or a treatment-effect claim ('ECPR would not help them' — contradicted here).
Rule-based selection: the entry criteria of the ARREST, PRAGUE-OHCA and INCEPTION trials, and the hospitals' own protocols
DOI 10.1161/JAHA.125.047815
ECPR allocated by a gradient-boosted causal forest model estimating each patient's INDIVIDUALISED TREATMENT EFFECT — the predicted difference ECPR makes to that patient — rather than their predicted survival
Retrospective, single health system, one paper, no external validation, abstract only. Treatment effect is a modelled counterfactual, not an observable quantity in an individual; the '20.0% survival without ECPR' in the top decile is an estimate. The confidence interval on the headline effect runs 1.56% to 57.58% — it excludes zero and very little else. THE SPECIFIC MODEL IS WEAK; THE STRUCTURAL POINT IS NOT — an observational literature that only measures outcome among the cannulated cannot in principle say whom to cannulate.
Overall survival was similar with and without ECPR: 11.1% vs 12.8% (standardised mean difference 0.054). In the TOP 10% BY PREDICTED BENEFIT, survival to discharge was 50.0% with ECPR versus 20.0% without — absolute observed treatment effect 30.0% (95% CI 1.56–57.58, P = 0.042). The model identified subgroups with higher observed benefit than those selected by any rule-based criteria tested. FACTORS ASSOCIATED WITH A LARGER TREATMENT EFFECT: LOWER serum pH, LOWER PaCO2, HIGHER lactate, younger age, and bystander CPR.
Adult, non-traumatic, EMS-attended out-of-hospital cardiac arrests from 4 tertiary centres in Taiwan, 2016–2024. After propensity-score matching on shockable rhythm and witnessed arrest: 1,953 matched patients, 977 of whom received ECPR.