Open the lung, or use it small?

In ARDS, the lung is a patchwork of collapsed and ventilated regions. The portion that can actually be ventilated is far smaller than a healthy lung — the concept known as the "baby lung".

Two ideas emerged from this. One: use the shrunken lung carefully, keeping it small. The other: open the collapsed regions to expand the usable lung.

The first was established in 2000. The second was tested for 17 years and failed at every turn. Then in 2015, we learned why the first one had been working.

Act I: Using it small, and living

ARDSNet (ARMA, 2000) compared lung-protective ventilation — tidal volume 6 mL/kg (predicted body weight), plateau pressure ≤30 — against the conventional 12 mL/kg. In-hospital mortality was 31.0% vs 39.8% (P=0.007, NNT approximately 11). Ventilator-free days also increased.

Simply by changing the ventilator settings, 39.8% became 31.0%. An NNT of 11. Few interventions in critical care are that efficient.

This is how "lung-protective ventilation" became standard of care. And it immediately raised the next question — what, then, should we do about PEEP?

Why does the ventilator damage the lung?

Why does reducing tidal volume improve survival? Turn the question around: how does the ventilator injure the lung in the first place?

There are four pathways. Volutrauma, in which excessive tidal volume overstretches alveoli. Barotrauma, in which high transpulmonary pressure ruptures alveoli. Atelectrauma, in which repeated collapse and reopening shears the tissue. And biotrauma, in which local injury triggers systemic inflammation via cytokines.

What ARDSNet reduced was mainly volutrauma. So what about the remaining atelectrauma? Keep the collapsed alveoli open. That is PEEP, and that is the starting point of the open-lung hypothesis.

The problem is that raising PEEP increases barotrauma. Reduce the third mechanism, and the second one grows. What happened in Act II was precisely this tug-of-war. That is why ART saw more pneumothoraces, and why OSCILLATE saw haemodynamic collapse.

With this map in hand, it also becomes clear why driving pressure — the protagonist of Act III — works. ΔP is a single number that folds together both volume and pressure.

Act II: Every attempt to open the lung failed

If low tidal volume is preserved while collapsed alveoli are opened and kept open, could more lives be saved? This was the open-lung hypothesis. The logic was elegant, and physiology backed it.

ALVEOLI (2004) compared high PEEP with low PEEP: oxygenation improved, but in-hospital mortality showed no difference (24.9% vs 27.5%). EXPRESS tested "maximal recruitment" — raising PEEP up to a plateau pressure of 28–30 — and found mortality of 39.0% vs 35.4%, not significant (though ventilator-free days and use of rescue therapy improved). LOVS, testing an open-lung strategy, found 36.4% vs 40.4%, RR 0.90, P=0.19. In all three trials, mortality did not move.

The decisive result was ART (2017). The group receiving aggressive recruitment plus PEEP titration had higher 28-day mortality — 55.3% vs 49.3% (HR 1.20) — and more pneumothorax. The intervention meant to open the lung caused harm instead.

The same thing happened with high-frequency oscillatory ventilation (HFOV) — a technique that, by delivering extremely small tidal volumes while keeping the lung open, seemed to be lung-protective ventilation in its ideal form. OSCILLATE (2013) was stopped early after in-hospital mortality reached 47% vs 35% (RR 1.33, P=0.005) in the HFOV group. High mean airway pressure compressed haemodynamics, driving up vasopressor use, sedation, and paralysis. OSCAR, published in the same issue, found no difference in 30-day mortality (41.7% vs 41.1%). Two trials, side by side, removed HFOV from adult ARDS care.

What 17 years established was not that the idea of "opening" the lung was itself wrong. It was that the price paid to open it — pressure, haemodynamics — exceeded whatever was gained.