Dexamethasone in Unresolving ARDS: A Trial Enrolled at Hour 24, a Patient at Day 6
A single patient, six days into ARDS that hasn't resolved. The disagreement isn't about the physiology of anti-inflammatory therapy — it's about whether a trial that enrolled within a day of diagnosis still describes a patient this far past it.
L.F. has taught seventh-grade science for over twenty years and had her first seizure in fifteen years at home six days ago, unwitnessed — her husband found her afterward with vomit on her clothes and assumed at first she'd simply been sick. Her epilepsy has been controlled on levetiracetam since her twenties, and this was her first breakthrough seizure in over a decade — no missed doses either she or her husband can identify, no clear trigger beyond a bad flu the week before. By the time EMS arrived she was hypoxic and confused, and aspiration pneumonitis was the working diagnosis before the day was out; a bronchoscopy on ICU day 1 confirmed food particulate in her lower airways, consistent with aspiration during the seizure itself rather than a primary pneumonia. It progressed to moderate-severe ARDS by her second ICU day, and six days in, it has not meaningfully turned around: her PaO2/FiO2 ratio has sat at roughly 142 for the past three days, essentially flat rather than trending either way, and her chest imaging still shows the same bilateral infiltrate pattern it did on day 3.
The complication is that her sputum culture from day 3 is still growing a mixed gram-negative organism, and her procalcitonin, while trending down, hasn't normalized — her fever curve has stayed low-grade rather than resolving cleanly, another reason her infection status reads as improving rather than settled. Her infection isn't clearly controlled, only clearly improving. That timing matters more than it would in a cleaner case. DEXA-ARDS (Villar, 2020), the trial behind the current push toward dexamethasone in moderate-severe ARDS, did not enroll at the moment of diagnosis: it reassessed patients at 24 hours after ARDS onset and randomized only those still meeting moderate-severe criteria then, deliberately screening out the ones who improve on their own inside the first day. Its ventilator-free-day and 60-day mortality benefit belongs to that population — ARDS established, but one day old, not six. It is worth being exact about what that leaves. DEXA-ARDS also excluded anyone already receiving corticosteroids, and reported nothing at all about patients whose infection was still open at randomization. On the question actually in front of the team, the trial is not negative — it is silent, and silence is a different thing to reason from than a null result.
Day 6, deciding whether the window already closed
I'd start dexamethasone now regardless of the day count. Her ARDS is still severe by every measure that matters — flat PaO2/FiO2, ongoing high ventilatory support — and the physiologic rationale for glucocorticoid-mediated anti-inflammatory effect doesn't expire on a calendar. Waiting longer only means more accumulated lung injury before treatment starts.
The physiology argument doesn't answer the actual question, which is whether DEXA-ARDS tells us anything about a patient at day 6. Villar's trial randomized at the 24-hour mark, after confirming the ARDS was established rather than resolving — she's three days past that point, with a culture still growing organism and a procalcitonin that hasn't cleared. Starting a systemic glucocorticoid on top of an infection we can't yet call controlled is a real risk the trial's early-enrollment population never carried.
I'm not disputing that her ARDS is still severe — I'm disputing that ‘still severe’ is the same claim DEXA-ARDS actually tested. The trial's benefit was measured in patients started at day one, and it excluded anyone already on corticosteroids outright — it never characterized a patient whose infection was still open the way hers is.
I'd separate the two questions you're actually arguing rather than treat ‘steroids or not’ as one binary. The infection-control question and the anti-inflammatory-timing question don't have to be resolved together tonight. Repeat cultures and a procalcitonin trend are already pending at 48 hours — get that answer first. If infection control is confirmed, start dexamethasone immediately rather than waiting further; if it isn't, the delay was never really about DEXA-ARDS's enrollment window in the first place, it was about not adding an immunosuppressive drug on top of an infection we can't yet call contained.
Agreed: hold dexamethasone until the pending 48-hour culture and procalcitonin recheck return, with an explicit pre-commitment that if infection control is confirmed at that point, dexamethasone starts immediately rather than being delayed further on the day-count question alone.
Not agreed:
Whether DEXA-ARDS's mortality benefit genuinely applies this many days into unresolving ARDS even once infection is confirmed controlled. The critical care physician holds that ongoing severe ARDS physiology is the operative fact regardless of enrollment-window mismatch; the pulmonologist holds that a trial population and a day-6 unresolving patient are different enough claims that the benefit shouldn't be assumed to transfer, infection status aside. The 48-hour recheck resolves the infection question, not this one.