Weaning from external ventricular drainage after non-traumatic subarachnoid hemorrhage: Rapid vs. gradual weaning and predicting closure trial failure. The SEVDVE retrospective multicenter cohort study

Résumé fourni par la source

BACKGROUND: Weaning from external ventricular drainage (EVD) following subarachnoid hemorrhage (SAH) typically requires an EVD closure trial, performed either straightforwardly (rapid weaning) or after gradual elevation of EVD (gradual weaning). We wanted to compare these two methods and build a sore to predict closure trial failure. METHODS: Among adult SAH patients, this multicenter (n = 5) retrospective study, compared rapid and gradual EVD weaning methods, and identified factors associated with EVD closure trial failure through logistic regressions. We developed a score to predict closure trial failure by splitting the dataset into training (2/3) and testing (1/3) sets. RESULTS: Among 1141 patients with an EVD between 01/01/2018 and 12/31/2022, 407 were hospitalized for SAH and had at least one EVD weaning attempt, 249 (61%) underwent gradual and 158 (39%) rapid weaning. Rapid weaning was associated with more failure (72 (46%) vs. 86 (35%), p = 0.044), but shorter length of stay (LOS) in both ICU and hospital. EVD closure trial failure was independently associated with prolonged EVD maintenance (p < 0.001), prolonged ICU (p = 0.001) and hospital LOS (p = 0.05). We developed a failure closure score using the difference in intracranial pressures (from H0 to H3 after closure), time since EVD insertion, and EVD level. The model's area under the receiver operating curve was 0.63 [0.53-0.74], indicating fair discrimination ability. CONCLUSIONS: EVD weaning strategies vary across centres. Rapid weaning was associated with a high risk of closure trial failure, but shorter LOS. EVD closure trial failure was associated with worse outcomes. A simple 3-criteria score could help.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.