Switching from controlled to assisted ventilation based on oxygenation thresholds and PEEP: a target trial emulation

Le résumé fourni par la source

Abstract Objective To determine whether PaO 2 /FiO 2 -based dynamic treatment strategies can optimise the timing of switching from controlled to assisted ventilation. Design and setting Target trial emulation using the MIMIC-IV database v3.1 (2008–2022). Patients Adult ICU patients on controlled invasive mechanical ventilation (IMV) for ≥ 48 h with at least one PaO 2 /FiO 2 <300 mmHg. A sensitivity analysis included patients ventilated for ≥ 24 h, excluding post-surgical patients. Interventions We compared strategies defined by switching once PaO 2 /FiO 2 exceeded predefined thresholds (150, 200, or 250 mmHg). Additionally, to account for possible PEEP-dependent effects, the thresholds were evaluated depending on the set PEEP (< 8 vs. ≥8 cmH 2 O), yielding nine strategies. Outcomes were 28-day restricted mean time lost (RMTL) to successful extubation (where a lower RMTL indicates a longer time to extubation) and mortality. Measurements and main results Among 1,635 patients, the strategy of delaying the switch until PaO 2 /FiO 2 >250 mmHg showed a longer time to successful extubation compared to switching at > 150 mmHg or > 200 mmHg (RMTL 17.4 days [95% CI: 15.9–18.6] vs. 19.6 days [95% CI: 18.6–20.9] and 19.8 days [95% CI: 18.5–21.1], respectively). Exploratory PEEP-dependent analyses demonstrated the longest time to successful extubation with ‘>250 mmHg’ for both low and high PEEP and the shortest with ‘>150 mmHg for low PEEP; >200 mmHg for high PEEP’. Among patients ventilated for ≥ 24 h, the 150 mmHg threshold was most favourable, irrespective of PEEP. No mortality differences were observed across strategies. Conclusions Switching as soon as PaO 2 /FiO 2 >150 mmHg is generally favoured over delaying until > 250 mmHg. Patients requiring ≥ 48 h of controlled IMV with PEEP ≥ 8 cmH 2 O may potentially benefit from a slightly higher threshold (> 200 mmHg). Prospective validation is warranted.

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