Mechanical Ventilation

The four parameters: Phigh (high pressure) — set to the plateau pressure measured on the prior conventional mode, not exceeding 30 cmH2O. Plow (low/release pressure) — commonly set to 0 cmH2O to maximize expiratory flow (this does not create zero PEEP; the release time is too brief for full lung emptying, so meaningful intrinsic PEEP persists even at a Plow of 0). Thigh (time at high pressure) — typically 4–6 seconds in patients with reasonably preserved lungs; longer Thigh generally improves oxygenation. Tlow (release time) — the most consequential and most poorly standardized parameter, commonly set to terminate the release when expiratory flow has fallen to 50–75% of the peak expiratory flow rate, targeting roughly 10–14 releases per minute.

Titrating for gas exchange: for hypoxemia, increase Phigh and/or Thigh (and FiO2) before reaching for other changes. For hypercapnia, shorten Thigh to increase release frequency, or lighten sedation to allow more spontaneous breathing between releases — spontaneous effort is central to how APRV is intended to work, not incidental to it.

2026 update: a meta-analysis of 9 studies (1,921 patients) found APRV/BiVent improved oxygenation and increased ventilator-free days compared to conventional ventilation, but showed no mortality benefit. Worth noting for CO2 clearance specifically: the brief Tlow release is the only point in the cycle where bulk CO2 clearance happens, and one study found APRV actually improved CO2 clearance at a lower minute ventilation — but in patients with very stiff lungs or high CO2 production, short release times can make CO2 management genuinely difficult. The TCAV protocol (time-controlled adaptive ventilation) refines APRV by personalizing Tlow to the patient's own expiratory flow curve rather than a fixed target — promising physiologically, but unproven in large RCTs.

Volume control vs. pressure control — the great non-debate: no RCT has ever shown a mortality difference between VC and PC. A Cochrane review found no difference in barotrauma, ventilator duration, or ICU stay between the two. A randomized comparison of pressure-control with spontaneous breathing (PC-SV) vs. volume-control assist-control in ARDS patients found no mortality difference between strategies, though PC used less sedation with otherwise similar outcomes. The mode doesn't matter nearly as much as the settings — low tidal volume, low plateau pressure, and appropriate PEEP achieve lung protection in either VC or PC.

SIMV — the mode that won't die, and the data against it: the theoretical appeal of SIMV (mandatory breaths plus preserved spontaneous effort between them) runs into a practical problem: between mandatory breaths, the patient does all the work on unsupported spontaneous breaths unless pressure support is added — potentially exhausting an already sick patient and prolonging weaning.

Ortiz et al.'s propensity-matched secondary analysis compared 350 patients ventilated with SIMV plus pressure support against 1,228 patients on assist-control (1,578 patients total). After adjustment for propensity score, the difference in in-hospital mortality was not significant (OR 1.04) — SIMV offered no measurable advantage over assist-control. SIMV was designed for an earlier era of weaning practice; a structured spontaneous breathing trial protocol now accomplishes the same goal more directly.

PRVC / VC+ / PCV-VG — the hybrid (also called Volume Target, AutoFlow, APVcmv): you set a target tidal volume; the ventilator delivers pressure-controlled breaths but adjusts the pressure level breath-by-breath to hit that VT target — combining decelerating flow with a volume guarantee. A meta-analysis of 17 studies found lower peak pressures and better compliance compared to pure volume-controlled ventilation in the operating room, though no ICU mortality data exists for this specific comparison.

The catch: if the patient takes a large spontaneous breath, the ventilator may interpret this as improved compliance and decrease support accordingly — this can lead to under-support during genuinely high patient demand, precisely when more support might be needed.

NAVA (Neurally Adjusted Ventilatory Assist, Maquet/Getinge) — triggers off neural signal, not flow or pressure: a specialized nasogastric catheter with electrodes detects the electrical activity of the diaphragm (Edi, measured in microvolts). The ventilator delivers pressure proportional to that signal: PIP = NAVA level (cmH2O/µV) × (Edi max − Edi min) + PEEP. Turning up the NAVA level is like turning up a gain knob — the harder the diaphragm signals, the more support the ventilator delivers, instantly.

This is the best-synchrony mode available, since it triggers off neural effort rather than flow or pressure change, and may reduce over-assistance and diaphragm atrophy as a result. A meta-analysis of 7 randomized/crossover trials (693 patients) found NAVA improved weaning success compared with other partial-support modes (OR 1.93), reduced duration of mechanical ventilation (MD −2.63 days), reduced hospital mortality (OR 0.58), and increased ventilator-free days (MD +3.48) relative to comparator modes.

Real limitations: it requires a special, comparatively expensive Edi catheter, is only available on Maquet/Getinge platforms, and titrating the NAVA level takes practice — set too high and you risk over-assistance, too low and the patient does the work alone. Watch the Edi trend: if Edi falls as NAVA level rises, the ventilator is unloading the diaphragm appropriately; if Edi stays high despite escalating support, the patient still needs more.

PAV+ (Proportional Assist Ventilation Plus, Puritan Bennett/Medtronic) — amplifies the patient's own effort by a fixed percentage: the ventilator continuously measures respiratory system compliance and resistance in real time (via brief end-inspiratory occlusions roughly every 10–15 breaths, applying the equation of motion), then delivers pressure assistance proportional to the patient's own instantaneous inspiratory effort. A 70% gain setting means the ventilator does 70% of the calculated work.

Earlier meta-analyses suggested better patient-ventilator synchrony and weaning success compared to pressure support. The PROMIZING trial (Bosma et al., NEJM 2025) — 573 patients across 23 centers in 7 countries, the largest RCT of PAV+ to date — found no significant difference between PAV+ and PSV in time to successful liberation from mechanical ventilation, reintubation rate, ventilator-free days, or 90-day mortality (29.6% vs. 26.6%). Patients on PAV+ required less sedation, though this was not a prespecified outcome. Clinicians in the trial frequently reported unfamiliarity with PAV+, and often switched patients to PSV as a result — a real-world adoption barrier that may partly explain the gap between the physiologic promise and the trial result.

Bottom line: PAV+ works as designed and monitors patient effort well at the bedside, but does not clearly outperform PSV when weaning is already protocolized and structured — which most of the benefit in earlier, smaller trials may have reflected rather than the mode itself.

ASV / IntelliVent-ASV (Adaptive Support Ventilation, Hamilton Medical) — the autopilot: you enter the patient's weight and a target percentage of predicted minute ventilation; the ventilator uses the Otis equation to calculate the tidal volume/rate combination that minimizes calculated work of breathing, then auto-adjusts pressure to reach it. IntelliVent-ASV extends this further, also auto-adjusting FiO2 and PEEP based on SpO2 and EtCO2.

Compared with SIMV specifically, ASV has shown shorter mechanical ventilation duration and lower peak airway pressures across several studies and a pooled meta-analysis (11 studies, 627 patients). Compared with PSV specifically, meta-analyses have generally found no significant difference in weaning outcomes — the advantage over SIMV doesn't necessarily translate into an advantage over already-good manual weaning practice.

The real critique: ASV has a "black box" feel — it can be genuinely difficult to know exactly what the ventilator is doing and why at any given moment, and it may not be well suited to complex ARDS physiology where the underlying assumptions behind the Otis equation optimization break down.

The physiology behind "beaking": once alveoli are near their elastic limit, additional pressure buys essentially no additional volume — the loop's upper limb flattens into a shape classically compared to a penguin's beak or a duck's bill. This is the direct visual correlate of volutrauma/barotrauma risk building in real time.

Bronchodilator response, visualized: in obstructive disease, the pre-bronchodilator expiratory limb shows the classic scooped-out pattern. After an effective bronchodilator, that scoop often straightens into a more linear line from peak expiratory flow down to end-exhalation — a direct, waveform-level readout of treatment response, faster than waiting on a repeat exam.

Detecting overdistension before it's obvious elsewhere: a beaking P-V loop can flag volutrauma risk before plateau pressure alone tells the full story, especially in a patient whose compliance is changing breath to breath.

Titrating PEEP and recruitment maneuvers: comparing pre- and post-recruitment P-V loops — leftward shift with improved compliance is the visual confirmation a maneuver worked, rather than inferring it from oxygenation alone.

Catching auto-PEEP and air trapping: an F-V loop that doesn't return to baseline, or a P-V loop with a rightward creep breath over breath, both flag dynamic hyperinflation — worth a direct look whenever a patient is tachypneic on the vent or hemodynamics are drifting for no clear reason.

Detecting circuit leaks: a P-V loop that fails to close — a visible break rather than a continuous trace — indicates a leak in the circuit or around the cuff, worth checking before troubleshooting anything more complex.

Normal values (adult): EtCO2 35–45 mmHg. PaCO2 typically exceeds EtCO2 by 2–5 mmHg (the a-ET gradient) due to dead-space ventilation — a gradient that widens meaningfully in conditions that increase dead space, like PE or severe COPD.

The dangerous confusion: a flat capnography line does not automatically mean esophageal intubation. If the tube was confirmed correctly placed moments earlier, an abrupt flatline is a circulatory arrest until proven otherwise — chest compressions restore some waveform even in a correctly placed tube, so a flat line during ongoing CPR (rather than at a pulse check) is a different problem than a flat line right after intubation.

Traumatic brain injury: target EtCO2 35–40 mmHg. Avoid hyperventilation outside of brief, deliberate use for acute herniation with Cushing's reflex — cerebral vasoconstriction from over-ventilation can worsen ischemia in an already injured brain.

Post-ROSC: target EtCO2 35–45 mmHg — avoid both hyper- and hypoventilation in the immediate post-arrest period, when cerebral autoregulation is impaired.

Sepsis / DKA / other compensatory tachypnea: match the ventilator to what the patient's own respiratory drive is already doing rather than "normalizing" a compensatory hyperventilation — a metabolic acidosis being compensated by a low PaCO2 needs that low PaCO2 maintained, not corrected.

Procedural sedation: capnography detects hypoventilation before it becomes hypoxemia — often the earliest available warning sign, arriving before the pulse oximeter changes at all, especially with supplemental oxygen running.

1. Calculate predicted body weight (PBW) — see the exact formulas in section 10.

2. Select any assist-control mode (volume or pressure).

3. Set initial tidal volume to 8 mL/kg PBW.

4. Reduce VT by 1 mL/kg at intervals ≤2 hours until VT = 6 mL/kg PBW.

5. Set initial rate to approximate baseline minute ventilation, not to exceed 35/min.

6. Adjust VT and rate to achieve pH and plateau pressure goals below.

Prone positioning, the practical protocol (PROSEVA-derived): sessions of at least 16 consecutive hours per day, continued as long as an oxygenation benefit persists after returning supine to reassess. Contraindications include hemodynamic instability requiring active resuscitation, unstable spine, recent abdominal surgery, and pregnancy (relative). Watch specifically for facial/airway pressure injury, ETT and line dislodgement during turns, and corneal injury — logistics-driven complications, best managed with a dedicated turning protocol and adequate staff.

The pattern across all six trials: the interventions that sound most aggressive or most physiologically elegant (maximal recruitment, HFOV) either caused harm or showed no benefit, while the more individualized or conservative approaches (esophageal manometry-guided PEEP, severity-stratified higher PEEP, early ECMO in genuinely refractory cases) show a more nuanced, patient-specific signal. Routine high-frequency oscillation and aggressive stepwise recruitment to very high pressures are not supported by current evidence as default ARDS strategies, regardless of how they're described in older textbook chapters.

The fix isn't always more sedation. A patient fighting the ventilator, treated reflexively with escalating sedation, often ends up on more sedation, more delirium, and a longer ICU stay without the underlying dyssynchrony ever being addressed. Before reaching for another dose, try: adjusting trigger sensitivity; switching to a pressure mode for better flow matching (section 2); increasing the inspiratory flow rate if in volume control; or simply asking the respiratory therapist what they're already seeing at the bedside — often the fastest diagnostic step of all.

The core strategy — controlled hypoventilation with permissive hypercapnia:

Tidal volume 5–8 mL/kg (lower than typical, to limit the volume that must be exhaled). Respiratory rate 10–14/min (deliberately low — this is the opposite instinct from most hypoxemic respiratory failure). High inspiratory flow rate (60–100 L/min) to shorten inspiratory time and maximize the time available for exhalation. Permissive hypercapnia, generally tolerated as long as pH remains >7.2, occasionally to extreme degrees in case reports (PaCO2 >200 mmHg, pH <7.0) without apparent harm when oxygenation is maintained.

Permissive hypercapnia to the patient's own baseline: a COPD patient with a chronic PaCO2 of 60 mmHg does not need to be ventilated to 40 — pushing toward a "normal" number in a chronically compensated patient risks post-hypercapnic alkalosis and doesn't reflect their actual physiologic target.

Matching extrinsic PEEP to intrinsic auto-PEEP: in a patient with significant auto-PEEP from dynamic hyperinflation, setting extrinsic (ventilator) PEEP to roughly 70–85% of the measured intrinsic PEEP eases the work of triggering a breath without meaningfully worsening hyperinflation — a counterintuitive move (adding PEEP to a hyperinflated lung) that specifically targets the trigger threshold, not the total lung volume.

Reading IMPROVE and PROVHILO together: low tidal volume appears to be the component doing the real work. PEEP and recruitment maneuvers added on top, at least in the routine open-abdominal-surgery population studied in PROVHILO, added hemodynamic cost without a demonstrated complication-rate benefit. This mirrors, in a lower-acuity population, the same pattern from section 6.4: aggressive add-on maneuvers don't consistently outperform doing the fundamentals well.

Core recommendations: assess readiness daily using objective criteria rather than clinical impression alone; conduct the SBT itself as the primary diagnostic test of extubation readiness rather than relying on weaning predictors like RSBI in isolation; use a standardized approach (protocol-driven, often nurse- or respiratory-therapist-initiated) rather than ad hoc physician judgment alone, since standardized approaches shorten ventilator duration in trial data.

The ABCDEF bundle in brief: Assess/prevent/manage pain · Both SAT and SBT, coordinated daily · Choice of sedation/analgesia (favoring lighter, non-benzodiazepine sedation where possible) · Delirium: assess, prevent, manage · Early mobility and exercise · Family engagement and empowerment.

Cuff leak test: in patients at risk of laryngeal edema, deflate the cuff and assess for an audible or measured air leak around the tube before extubating. A reduced or absent leak (a "failed" cuff leak test) flags elevated risk of post-extubation stridor and reintubation.

A meta-analysis found that giving IV corticosteroids at least 4 hours before extubation in patients with a failed cuff leak test reduced post-extubation stridor and reintubation — benefit confined to this high-risk subgroup, not a reason to steroid-treat every extubation.

Prophylactic NIV or HFNC after extubation: in patients at high risk for reintubation (age >65, heart failure as the indication for ventilation, moderate-severe COPD, APACHE II >12 at extubation, BMI >30, difficult/prolonged prior weaning, ≥2 comorbidities, or ventilation >7 days), applying prophylactic respiratory support immediately after extubation — rather than waiting to see if the patient fails — reduces reintubation.

Hernández et al. (JAMA 2016) found HFNC non-inferior to NIV for this purpose in high-risk patients, and generally better tolerated. A separate Hernández trial (JAMA 2016) found HFNC also outperformed conventional oxygen in low-risk patients. A post hoc analysis found patients with ≥4 risk factors did better specifically on NIV (23.9% vs. 45.7% reintubation) — the non-inferiority of HFNC may not hold at the highest end of the risk spectrum.

The IHI Ventilator Bundle, current core elements:

1. Head-of-bed elevation 30–45°. 2. Daily sedation interruption ("sedation vacation") coordinated with daily extubation-readiness/SBT assessment — see section 8.3. 3. Peptic ulcer disease prophylaxis. 4. DVT prophylaxis. 5. Daily oral care.

Correction to the original bundle: daily oral care with chlorhexidine was added as a fifth IHI element in 2010, but a 2022 practice update reversed this specifically for general ICU patients — current guidance is to not use chlorhexidine for routine oral decontamination outside of cardiac surgery. Oral care itself (mechanical cleaning, tooth brushing) remains recommended; the chlorhexidine-specific component does not.

VAE surveillance (CDC framework) — not the same thing as clinical "VAP": current national surveillance runs on a tiered definition rather than the older clinical-pneumonia criteria. VAC (ventilator-associated condition): a sustained rise in FiO2 or PEEP after a period of stability/improvement. IVAC (infection-related VAC): a VAC plus abnormal temperature/WBC plus a new antimicrobial started. PVAP (possible VAP): an IVAC plus microbiologic or histopathologic evidence of pneumonia. Most VAEs are not infectious — atelectasis, pulmonary edema, and ARDS drive the majority of cases, with VILI as the underlying mechanism connecting an oxygenation-requirement spike back to the ventilator itself.

The threshold, and what actually drives it: a 2026 meta-analysis of 34 studies found mechanical power was consistently higher in nonsurvivors (mean difference 1.91 J/min), with each 1 J/min increase independently associated with higher adjusted mortality (adjusted OR 1.04 per J/min). A mechanical power above 17 J/min specifically was associated with substantially greater mortality (OR 1.60, 95% CI 1.34–1.91) — a threshold echoed in several independent single-center cohorts. Despite mechanical power's multi-variable formula, subsequent analyses have found that driving pressure and respiratory rate are the individual variables most predictive of the mortality signal — meaning the extra complexity of the full mechanical power calculation may not be adding much practical value over watching those two numbers directly.