Post-gastrectomy refractory hypoxemia — lung-protective reset, proning, source hunt and ECMO referral

Management walkthrough

Define severe ARDS aloud — then exclude the mimics before the label sticks. P/F 96 is Berlin severe disease at 45% mortality — three hypoxemia bands with the 27/32/45 ladder, beating AECC prediction.^[1] With +8 L and a fresh anastomosis, still exclude overload (echo, volume responsiveness), atelectasis, cardiac failure and embolism — postoperative ARDS is common but its mimics dominate early deterioration, and recognition failure is the documented norm.^[20]

Reset the ventilator to the platform every later trial stood on. Drop volumes from 8 mL/kg actual to 6 mL/kg predicted body weight with plateau capped at 30: ARDSNet cut death 31.0% vs 39.8% in 861 patients and added ventilator-free days.^[3] His plateau 32 already breaches the ceiling at excessive volume. Then minimise driving pressure — the best risk variable, +7 cm H2O carrying RR 1.41 — rather than chasing volume or PEEP alone.^[15]

Prone, dry, dexamethasone — and no routine paralysis. He meets PROSEVA severe entry: prone sessions of at least 16 hours cut 28-day death to 16.0% vs 32.8% (HR 0.39) without excess complications — start now, not after a supine night.^[4] With pressors weaning, run FACTT-conservative: +2.5 ventilator-free days without shock or renal cost, with the surgical subgroup gaining ventilator-free and ICU-free days dry.^[12][14] Established severe disease earns dexamethasone 20 mg daily x5 then 10 mg x5 (21% vs 36% at 60 days, balanced harms).^[11] He is synchronous, so no routine cisatracurium — ROSE stopped for futility at 42.5% vs 42.8% with harm signals; blockade is for dyssynchrony only.^[10]

Hunt the anastomosis while ventilating — the abdomen decides postoperative death. Postoperative ARDS death tracks extrapulmonary organ dysfunction and surgical context, not lung numbers — so examine the abdomen, trend drain output and inflammatory markers, image on suspicion (contrast swallow/CT with oral contrast per local protocol), and keep the oesophagojejunal leak high on the list despite currently satisfactory output.^[20] Afebrile with lactate 2.2 is reassuring, not exonerating — trajectories decide, and postoperative ARDS shows more favorable early paths only when complications are caught.^[20]

Draw the ECMO line before it is needed. At P/F 96 on PEEP 14 he is severe but above EOLIA rescue entry (below 50 for 3+ hours, below 80 for 6+ hours, or acidemia with hypercapnia) — so notify the ECMO centre early, optimise (prone, dry, steroids, source), and transfer if thresholds are met: EOLIA was 35% vs 46% (ns) with 28% crossover, meaning rescue ECMO belongs in the plan, not as improvisation.^[5] Disposition is ICU with surgical, intensive-care and (by phone) ECMO input; re-escalation triggers (P/F, plateau, driving pressure, drain, lactate) are written, not assumed.