Ventilator Liberation and Tracheostomy

In depth

The one-page summary above is what the PDF prints. These sections go further.

Why weaning fails: load, capacity, heart and brain

Liberation works when respiratory muscle capacity exceeds the load, the heart tolerates the switch to negative intrathoracic pressure, and the brain drives breathing and protects the airway. Load rises with secretions, bronchospasm, auto-PEEP, stiff lungs and the high minute ventilation of fever, sepsis and acidosis. Capacity falls with sedatives, low phosphate, potassium or magnesium, ICU-acquired weakness and disuse: in organ donors after 18–69 hours of controlled ventilation, diaphragm fibers were 53–57% smaller than in controls (Levine 2008).

An overloaded pump breathes fast and shallow, so dead space takes a larger share of each breath and PaCO₂ rises late. The tube adds little: after a passed 2-hour T-piece trial, work of breathing was the same through the tube as after extubation (1.72 vs 1.63 J/L), the tube accounting for about 11% (Straus 1998).

Negative intrathoracic pressure raises venous return and left ventricular afterload while breathing work and a catecholamine surge raise oxygen demand. A stiff, ischemic or failing ventricle answers with weaning-induced pulmonary edema, behind a third or more of weaning failures (Routsi 2019). Clues during a failing SBT: PAOP ≥18 mm Hg, plasma protein or hemoglobin up over 5%, BNP up over 12%, E/A over 0.95 with E/e′ over 8.5. Treat with diuresis, plus nitrates for hypertension or ischemia; BNP-guided diuresis shortened time to extubation (Mekontso Dessap 2012).

The pathway: daily screen, SAT, SBT and protocols

A daily screen by nurses and therapists followed by a 2-hour SBT cut ventilation from 6 to 4.5 days and complications from 41% to 20% (Ely 1996). The ABC trial (Girard 2008, n = 336) added a spontaneous awakening trial before each SBT: 14.7 vs 11.6 ventilator-free days, ICU stay 9.1 vs 12.9 days, 1-year death hazard ratio 0.68. Self-extubation rose (16 vs 6) but reintubation after it did not (5 vs 3).

Since then, AARC 2024 suggests the RSBI is not needed before an SBT, PS of 8 or less or none, screening and SBT before noon, and no rise in FiO₂ during the trial. KSCCM 2024 accepts PS or T-piece, leaves the RSBI to the clinician, supports inspiratory muscle training and HFNC after extubation, and advises against early tracheostomy to speed weaning.

Thresholds vary: ARDSNet screens at FiO₂ ≤0.40 with PEEP ≤8 (or ≤0.50 with PEEP ≤5), SBP ≥90 without vasopressors and no paralytic; Wake Up and Breathe at SpO₂ ≥88%, FiO₂ ≤0.50, PEEP ≤7.5, no vasopressors or ischemia. A failed SAT (agitation, pain, RR over 35, SpO₂ under 88%, distress, arrhythmia) restarts sedation at half dose, with a new screen in 24 hours.

Running the SBT and reading the numbers

In high-risk patients, Thille 2022 (n = 969) found PS 8 without PEEP and T-piece equal for ventilator-free days (27 each); more were extubated within 24 h after PS (77.7% vs 72.2%), with similar reintubation (14.9% vs 13.6%). Thirty- and 120-minute trials performed alike (Esteban 1999).

ARDSNet stops the trial for SpO₂ under 90%, RR over 35, Vt under 4 mL/kg PBW, pH under 7.30 or two distress signs (HR over 120% of baseline, accessory muscles, paradox, sweating, marked dyspnea); Wake Up and Breathe for SpO₂ under 88%, RR over 35 or under 8, distress, mental status change or arrhythmia. AARC 2024 asks whether 90% is too strict.

The RSBI, measured during unassisted breathing, was the best predictor in Yang and Tobin 1991 (64-patient test set): under 105 had sensitivity 0.97 and specificity 0.64 (maximal inspiratory pressure specificity 0.11); a 2022 meta-analysis cited by AARC gives 0.83 and 0.58. As a gate before the SBT it prolonged weaning (3 vs 2 days) without fewer failures (Tanios 2006).

After a failed SBT, return to full support and look for the reason: sedation or delirium, fluid overload or ischemia, secretions, bronchospasm or auto-PEEP, low phosphate, potassium or magnesium, metabolic alkalosis, weakness. Hypercapnic COPD patients who fail can be extubated straight to NIV in experienced units (ERS/ATS 2017, conditional).

Airway protection, the cuff leak test and steroids

Passing an SBT does not show that secretions can be cleared or that the upper airway will stay open. Among 88 patients who passed (Salam 2004), cough peak flow ≤60 L/min raised failure about fivefold, secretions over 2.5 mL/h threefold and failing four simple commands fourfold; with all three every extubation failed, with none 3%.

Laryngeal edema follows 5–54% of extubations and stridor 1.5–26.3%, with reintubation for it in 1.1–10.5% (Pluijms 2015). Risk factors (ATS/CHEST 2017): traumatic intubation, intubation over 6 days, a large tube, female sex, reintubation after unplanned extubation.

1. Volume assist-control; suction the mouth and above the cuff.
2. Note the delivered Vt, then deflate the cuff fully.
3. Average the 3 lowest exhaled Vt of 6 breaths; leak = delivered minus that.
4. Under 110 mL or about 10–25% of Vt fails; reinflate to 20–30 cm H₂O if waiting.

A leak under 110 mL had a positive predictive value of 0.80 and a negative predictive value of 0.98 for stridor (Miller and Cole 1996); across nine studies sensitivity was 0.56 and specificity 0.92 (Ochoa 2009). A good leak reassures; a poor one overcalls, since pooled secretions or a large tube can shrink the leak without edema, so ATS/CHEST treats a failed test with steroids rather than waiting.

Support after extubation, and when NIV harms

Low risk in Hernández 2016 meant age under 65, APACHE II under 12, BMI under 30, simple weaning, good secretion control, at most one comorbidity, and no heart failure, moderate to severe COPD, airway problem or prolonged ventilation. ATS/CHEST 2017 strongly recommends immediate preventive NIV at high risk; ERS/ATS 2017 makes it conditional; ERS 2022 prefers NIV over HFNC at high risk and HFNC over oxygen at low risk. HIGH-WEAN gave NIV at least 12 hours a day for 48 hours.

Rescue is different: in Esteban 2004, NIV for failure within 48 hours of extubation delayed reintubation (median 12 h vs 2.5 h) and raised ICU deaths, and ERS/ATS 2017 advises against it. NIV also needs an awake patient who protects the airway; obtundation, copious secretions, vomiting, upper airway obstruction or shock call for reintubation.

Extubation failure, stridor and swallowing

Extubation failure is reintubation within 48–72 hours; NIV and HFNC trials often count 7 days because support can postpone it. It affects about 10–20%, and mortality is higher: 43% vs 12% in hospital (Epstein 1997) and 32.6% vs 6.6% in the ICU (Dadam 2024). In 74 reintubated patients (Epstein 1998), mortality was 53% for non-airway vs 17% for airway causes, and 51% vs 24% when reintubation came after vs within 12 hours.

Timing hints at the cause: laryngeal edema, aspiration and retained secretions usually show within hours; fatigue, weaning-induced pulmonary edema, pneumonia and encephalopathy build over a day or two.

• Stridor: sit upright, humidified oxygen, keep the patient calm, call for airway help early.
• Nebulized epinephrine (optimal adult dose unknown; 1 mg in 5 mL suggested) plus IV methylprednisolone 20–40 mg or dexamethasone 5 mg for 24–48 h (Pluijms 2015).
• Heliox does not shrink edema; it can lower work of breathing but must not delay reintubation.
• NIV does not help stridor and delays reintubation.
• Respiratory distress: reintubate without delay, smaller tube, difficult-airway plan.

Post-extubation dysphagia is common and often silent: 12.4% overall and 18.3% of emergency admissions in DYnAMICS, persisting to ICU discharge in over 80%, with 9.2% excess 90-day mortality (Zuercher 2019). Screen before oral intake, ideally by trained ICU nurses within hours of extubation, and refer abnormal results to speech-language pathology.

Prolonged weaning and tracheostomy timing

The 2007 consensus calls weaning simple (extubated after the first SBT), difficult (up to 3 SBTs or 7 days) or prolonged (beyond that). WIND (Béduneau 2017), counting from the first separation attempt, found 57% finished within 24 hours, 10% took 1–7 days and 9% longer, while 24% never had an attempt; mortality rose from 6% to 17% to 29% across the three weaning groups.

A tracheostomy is better tolerated than a translaryngeal tube, so sedation can lighten; it allows mouth care, speech and eating and spares the larynx. Its risks are bleeding, stomal infection, displacement, stenosis and, rarely, tracheo-innominate fistula.

The stoma forms in about 4 days after open surgery and 7–10 days after a percutaneous procedure (ICS 2014), so a tube that falls out early cannot simply be pushed back. Trach collar sessions beat stepwise pressure support weaning (15 vs 19 days, Jubran 2013), and cuff deflation off the ventilator sped weaning and cut respiratory infections (20% vs 36%, Hernández 2013). SKR/SRLF 2026 adopts both and, unlike KSCCM 2024, advises against threshold inspiratory muscle training.

Tracheostomy tubes, daily care, speaking valves and decannulation

• Cuff 20–30 cm H₂O by manometer (SKR/SRLF 2026; ICS 2014: ≤25) each shift and after moves; a leak at full pressure suggests displacement or too small a tube.
• Humidify always: HME for scant secretions, heated humidifier for thick ones.
• Inner cannula off the ventilator: clean at least 8-hourly, 4-hourly if secretions are heavy (ICS 2014).
• Stoma: daily saline or water, no antiseptics, foam dressing, no routine sutures or tube changes (SKR/SRLF 2026).

Suction only for secretions (AARC 2022): preoxygenate, catheter under half the lumen, 15 seconds or less, shallow to a premeasured depth, no routine saline; adult pressure under 200 mm Hg (many units use 80–120). A catheter that will not pass means a blocked inner cannula or a displaced tube (ICS 2014).

A one-way speaking valve opens on inspiration and closes on expiration, sending exhaled gas through the cords. It needs the cuff fully deflated or a cuffless tube, room for air around the tube (outer diameter at most about three-quarters of the trachea), an awake patient who manages secretions, and exhalation possible with the hub briefly occluded. With the cuff up the patient cannot exhale: asphyxia, barotrauma, falling cardiac output (ICS 2014).

Early cuff deflation with an in-line valve restored voice a median 11 days sooner (Freeman-Sanderson 2016, n = 30), and routine valve use cut time to speech from 18 to 9 days without longer ventilation (Sutt 2015). Dysphagia affects up to 93% of tracheostomized patients and causes 64% of decannulation failures, so SKR/SRLF 2026 advises early swallowing work by ICU staff rather than waiting for a specialist.

Deciding decannulation by suctioning frequency on continuous high flow, instead of a 24-hour capping trial, cut time to decannulation from 13 to 6 days without more failures (Hernández 2020, n = 330). SKR/SRLF 2026 adopts it: no more than 2 suctions per 8 hours over 24 hours, a 5-minute occlusion test with the cuff down, and no severe dysphagia.

Tracheostomy and laryngectomy emergencies

Obstruction and displacement are the commonest tracheostomy problems (ICS 2014). Partial displacement into pretracheal tissue is the more dangerous because it is subtle: the suction catheter will not pass, capnography fades, pressures rise or volumes fall, surgical emphysema appears, or the patient vocalizes with the cuff up. Capnography is mandatory for ventilator-dependent patients, and without experienced help, removing a partly displaced tube may be safer than pushing it back (ICS 2014).

NTSP (McGrath 2012) starts from the bedhead sign. Green is a tracheostomy with a potentially patent upper airway: oxygen to face and stoma, and the mouth stays a route. A cuffed tube in the trachea blocks bag-mask ventilation from above until the cuff is deflated or the tube removed (ICS 2014). Then cover the stoma and bag by mouth, or put a pediatric mask or supraglottic airway on the stoma; an expert intubates orally with an uncut tube past the stoma or places a small tracheostomy or 6.0 cuffed tube in the stoma.

Red is a laryngectomy: the trachea ends at the neck and the mouth no longer reaches the lungs. Oxygen, rescue ventilation and intubation all go through the stoma; when unsure, give oxygen to face and stoma. NTSP advises against blind reinsertion before day 7, especially after a percutaneous tracheostomy: a false passage makes ventilation impossible.

Early bleeding usually comes from the stoma edges and settles with pressure and correction of coagulopathy. Tracheo-innominate fistula complicates about 0.1–1% of tracheostomies, peaks around days 7–14 (Grant 2006) and often follows a small sentinel bleed: hyperinflate the cuff and leave it up, press a finger through the stoma against the back of the manubrium, secure the airway below the bleed and call surgery.

Exam pearls

• SAT before SBT daily; failed SAT restarts sedation at half dose; the pair lowered 1-year mortality (ABC).
• SBT 30–120 min on PS 5–8, CPAP or T-piece, FiO₂ unchanged; stop for RR over 35, SpO₂ under 88–90%, a 20% HR or BP change, or distress.
• RSBI = RR ÷ Vt in liters; under 105 favors success; sensitive, not specific, not required (AARC 2024).
• Passing the SBT is not enough: cough, secretions and mental status decide airway protection.
• Failed cuff leak test with readiness: steroids at least 4 h before extubation, no retest.
• High risk (over 65, heart failure, COPD, hypercapnia, obesity): extubate to NIV; low risk: HFNC.
• NIV does not rescue established post-extubation failure (Esteban 2004): reintubate.
• Stridor: upright, nebulized epinephrine, IV steroid; worsening: early reintubation, smaller tube.
• Early tracheostomy did not lower mortality (TracMan); first change after the stoma forms.
• Speaking valve only with the cuff fully deflated; laryngectomy: oxygenate through the stoma only.