Publication

• Title: Personalized automatic management of tracheal cuff pressure and subglottic secretions drainage to prevent pneumonia in critically ill intubated patients. The MICROINHALO multicenter randomized controlled trial
• Acronym: MICROINHALO
• Year: 2026
• Journal published in: Intensive Care Medicine
• Citation: De Pascale G, Cutuli SL, Vargas M, Cortegiani A, Dalfino L, Greco M, et al. Personalized automatic management of tracheal cuff pressure and subglottic secretions drainage to prevent pneumonia in critically ill intubated patients. The MICROINHALO multicenter randomized controlled trial. Intensive Care Med. 2026. Epub 2026 Jun 2.

Context & Rationale

• Background

  • Ventilator-associated pneumonia (VAP) remains a common ICU complication, but its prevention literature is complicated by diagnostic subjectivity, changing surveillance frameworks, and inconsistent translation from VAP reduction to patient-centred outcomes.
  • Microaspiration of contaminated oropharyngeal and gastric secretions around the endotracheal tube cuff is a plausible mechanistic pathway for VAP.
  • Current prevention guidance supports considering endotracheal tubes with subglottic secretion drainage (SSD) ports for patients expected to require more than 48–72 hours of intubation, while recognising that evidence for reduced ventilation duration, length of stay, or mortality is uncertain.¹
  • Continuous or automatic cuff-pressure control has shown a VAP signal in meta-analysis, but certainty has been limited by lack of blinding, heterogeneity, and potential conflicts of interest.²
  • SSD meta-analyses have generally found lower VAP incidence, but have been less persuasive for hard outcomes such as mortality, ICU length of stay, or hospital length of stay.³
• Research Question/Hypothesis

  • MICROINHALO tested whether a personalised, automated airway-management system combining CO₂-guided cuff-pressure control with automatic SSD, rinsing, and venting would reduce early bacterial tracheobronchial colonisation compared with manual cuff-pressure control and manual SSD.
  • The hypothesis was mechanistic: improved tracheal sealing plus more efficient removal of pooled subglottic secretions should reduce microaspiration, bacterial colonisation, and ultimately VAP.
  • The AnapnoGuard concept had preceding physiological and pilot randomised data suggesting better cuff-pressure control and improved secretion clearance.⁴⁵
• Why This Matters

  • Unlike trials comparing SSD tubes with non-SSD tubes, MICROINHALO compared two active airway-care strategies: automated personalised cuff/SSD management versus manual cuff/SSD management.
  • This is clinically important because many ICUs already use cuff-pressure checks and SSD tubes for patients expected to need prolonged ventilation.
  • A positive result would have supported device-intensive, personalised airway management; a neutral result would argue that better physiology does not necessarily improve early colonisation or patient-centred outcomes.

Design & Methods

• Research Question: In critically ill adults undergoing orotracheal intubation with expected invasive mechanical ventilation for more than 48 hours, does personalised automatic tracheal cuff-pressure control plus automatic subglottic secretion drainage reduce Day 3 bacterial tracheobronchial colonisation compared with manual cuff-pressure control plus manual subglottic secretion drainage?
• Study Type: International, multicentre, open-label, cluster-randomised controlled trial conducted in 10 ICUs in Italy and Israel between June 2022 and April 2024.
• Population:
  • Adults aged more than 18 years.
  • Primary intubation using a study endotracheal tube.
  • Expected invasive mechanical ventilation duration more than 48 hours.
  • Key exclusions: invasive mechanical ventilation during the previous 14 days; contraindication to enteral feeding; enrolment in another potentially interfering study; clinical evidence of inhalation before intubation; pregnancy.
  • Screening flow: 1408 patients were scheduled for endotracheal intubation; 837 were not included because expected ventilation duration was 48 hours or less; 275 were excluded by protocol criteria; 5 declined; 21 were excluded for logistic reasons; 270 were randomised.
• Intervention:
  • Automatic management group: AnapnoGuard endotracheal tube, internal diameter 7.5/8.0 mm, with ellipsoidal thin-wall polyurethane cuff, dual suction lines, and an additional venting/CO₂ line.
  • The tube was connected to the AnapnoGuard 100 control unit, which continuously regulated cuff pressure and automatically evacuated subglottic secretions.
  • The system sampled gas from the subglottic space every few minutes, used CO₂ above the cuff as a marker of leak, reduced target cuff pressure by 1 mmHg when leak was absent, and increased pressure when leak was detected.
  • The system combined intermittent suction, rinsing, and venting of the subglottic space through dedicated lumens.
• Comparison:
  • Manual management group: TaperGuard Evac endotracheal tube, internal diameter 7.5/8.0 mm, with a dorsal subglottic suction lumen ending above the cuff.
  • Cuff pressure was manually maintained between 20 and 30 cmH₂O using a portable manometer, monitored at least every 8 hours.
  • Subglottic secretions were manually drained using a 10 mL syringe, with intended frequency of one suction per hour.
  • The actual manual suction exposure was 7031 effective aspirations, with median daily frequency 10 [IQR 8 to 12].
  • Both groups received the same VAP-prevention bundle, including no routine ventilator-circuit change, semi-recumbent positioning, enteral nutrition through a nasogastric tube, stress-ulcer prophylaxis, oral cleaning every 8 hours, sedation and weaning protocols, and no selective digestive decontamination.
• Blinding: Open-label. Treating teams could not be blinded because the endotracheal tubes and control device differed visibly; this matters more for clinically diagnosed VAP and co-interventions than for the microbiological primary endpoint.
• Statistics:
  • Power calculation: 240 analysed patients, 120 per group, were required to detect a 50% relative reduction in Day 3 tracheobronchial colonisation, from 30% in controls to 15% with automatic management, with 80% power at a two-sided 5% significance level; allowing 10% dropout, the investigators planned to enrol 270 patients.
  • Analysis population: modified intention-to-treat, including randomised patients who required mechanical ventilation for more than 48 hours, excluding those who withdrew consent, were subsequently found to have been ventilated during the previous 14 days, or did not reach Day 3 because of extubation or death.
  • No interim analysis was planned.
  • Post-hoc Bayesian sensitivity analyses were performed for the primary endpoint and the two VAP secondary endpoints.
• Follow-Up Period: Primary endpoint at Day 3 after randomisation; VAP surveillance until ICU discharge or Day 28, whichever occurred first; mortality reported through 90 days.

Key Results

This trial was not stopped early. No interim analysis was planned. Of 270 randomised patients, 250 were included in the modified intention-to-treat analysis: 127 in the automatic management group and 123 in the manual management group.

• The primary endpoint was clearly neutral: Day 3 colonisation was 37.0% versus 41.5%, far from the hypothesised 50% relative reduction and with a 95% CI crossing clinically important benefit and no effect.
• The VAP findings were directionally and clinically notable, but they were secondary endpoints in an open-label trial, with no reported multiplicity adjustment and no parallel improvement in ventilator-free days, antibiotic-free days, length of stay, or mortality.
• The intervention achieved strong physiological separation: cuff-pressure out-of-range values were 10.2% versus 24.5%, and median daily SSD volume was 25 versus 10.5 mL/day.

Internal Validity

• Randomisation and Allocation:
  • Cluster randomisation by predetermined clusters of 9 consecutive patients, stratified by centre, was pragmatic for urgent intubation.
  • Cluster allocation also creates potential predictability because sites knew which study tubes were available during each cluster period.
  • Concealment of future cluster assignments was not described.
  • The design reduces feasibility barriers but is less robust than individual concealed randomisation for selection bias.
• Dropouts and Post-Randomisation Exclusions:
  • 270 patients were randomised equally: 135 automatic management and 135 manual management.
  • 20 randomised patients were excluded from the analysis: automatic management excluded 6 deaths before Day 3 and 2 extubations before Day 3; manual management excluded 6 deaths before Day 3, 3 extubations before Day 3, and 3 screening failures.
  • The analysed population was therefore 127 versus 123.
  • This exclusion strategy makes the primary estimand conditional on surviving and remaining intubated long enough to reach the Day 3 culture.
  • The protocol anticipated dropout, but the published modified intention-to-treat definition was narrower than a conventional all-randomised intention-to-treat analysis.
• Performance and Detection Bias:
  • Blinding of clinicians was not feasible.
  • The primary endpoint was microbiological and therefore less subjective than VAP, although clinical decisions around sampling and diagnostic work-up may still be influenced by allocation awareness.
  • Clinically diagnosed VAP is vulnerable to detection bias because radiographic interpretation, fever/leucocyte thresholds, purulence, and the decision to investigate are partly clinician-dependent.
  • Among 46 clinically diagnosed VAP cases, 31 underwent tracheal aspirate and 15 underwent bronchoalveolar lavage for diagnosis, increasing concern about overdiagnosis compared with a uniformly bronchoscopic strategy.
• Protocol Adherence:
  • The trial demonstrated excellent separation in the delivery variables that define the intervention.
  • Automatic management reduced out-of-range cuff-pressure measurements from 24.5% to 10.2%.
  • Automatic management increased median daily SSD volume from 10.5 [6 to 17] to 25 [8 to 41] mL/day.
  • No cross-over after randomisation was reported.
• Baseline Characteristics:
  • Groups were broadly similar in age, sex, BMI, Charlson index, severity scores, oxygenation, and ventilation mode.
  • Median age was 66 [55.5 to 75] versus 65 [55 to 72.5] years.
  • Median SAPS II was 44 [34 to 53] versus 46 [38 to 58].
  • Median SOFA score was 7 [5 to 10] versus 8 [5 to 12].
  • Baseline infection burden was high: suspected pneumonia at enrolment occurred in 68/127 (53.5%) versus 59/123 (47.9%); positive Day 0 tracheal aspirate at any count occurred in 67/127 (52.8%) versus 58/123 (47.1%); ongoing antibiotics were used in 94/127 (74.0%) versus 96/123 (78.1%).
  • These features make the primary endpoint biologically noisy because many patients already had pneumonia, colonisation, and antimicrobial exposure before the intervention could exert its full effect.
• Heterogeneity:
  • The trial included a clinically heterogeneous ICU population with respiratory failure, neurological failure, cardiovascular failure, sepsis, ARDS, immunosuppression, and suspected pneumonia.
  • Septic shock was present in 34/127 (26.8%) versus 26/123 (21.1%).
  • ARDS was present in 24/127 (18.9%) versus 18/123 (14.6%).
  • Heterogeneity improves pragmatic relevance but dilutes a mechanistic microaspiration-prevention effect, particularly when early cultures are already positive.
• Timing:
  • The primary culture was collected at 72 hours, while median VAP onset was 6 to 7 days.
  • This temporal mismatch weakens Day 3 colonisation as a surrogate for later clinical VAP.
  • The protocol allowed connection of the automatic system within 24 hours, so the earliest peri-intubation microaspiration window may not have been fully modified in all patients; actual connection-time distribution was not reported.
• Dose:
  • The automatic arm delivered a higher cuff-pressure “dose”: 27.1 [26.4 to 28.3] versus 24.7 [22.1 to 27.1] cmH₂O.
  • All 264 automatic-arm out-of-range pressure values were above 30 cmH₂O and were automatically set to seal the trachea.
  • Post-extubation stridor was uncommon and similar, 2.4% versus 1.6%, but systematic laryngoscopy or tracheal mucosal injury assessment was not reported.
• Separation of the Variable of Interest:
  • Separation was strong for cuff-pressure control: 264/2592 (10.2%) versus 541/2205 (24.5%) out-of-range values.
  • Separation was strong for secretion drainage: median daily SSD volume 25 [8 to 41] versus 10.5 [6 to 17] mL/day.
  • The intervention, however, bundled several changes simultaneously: different tube design, two suction ports, venting/rinsing, CO₂-guided automatic cuff control, and higher achieved cuff pressure.
  • The trial therefore cannot isolate whether the VAP signal, if real, came from cuff-pressure control, increased SSD, rinsing/venting, tube geometry, or the combination.
• Adjunctive Therapy Use:
  • VAP-prevention bundle elements were standardised across groups.
  • Antibiotic exposure was very common at baseline: ongoing antibiotics 74.0% versus 78.1%.
  • Among patients with baseline pneumonia and MDR bacteria, appropriate antibiotic therapy was 9/14 (64%) versus 9/17 (53%).
  • High antimicrobial exposure could reduce culture yield, alter airway flora, and attenuate differences in colonisation.
• Outcome Assessment:
  • The primary outcome was objectively quantified as tracheal aspirate bacterial burden greater than 10³ CFU/mL without clinical and radiological signs of VAP.
  • The chosen threshold and timing are mechanistically reasonable but not directly patient-centred.
  • VAP outcomes are clinically important but were secondary and open to ascertainment bias.
  • Microaspiration biomarkers were measured only in 10 patients per group at the coordinating centre, limiting interpretability.
• Statistical Rigor:
  • The analysed sample exceeded the 240-patient requirement for the primary endpoint.
  • The observed control event rate was 41.5%, higher than the 30% assumed in the power calculation, but the observed effect size was much smaller than planned.
  • The cluster-randomised design was not clearly matched by a cluster-adjusted analysis, mixed-effects model, generalised estimating equation, or reported intracluster correlation coefficient.
  • Failure to account for clustering can understate uncertainty when outcomes are correlated within cluster periods or centres.
  • Secondary VAP p values were not adjusted for multiplicity; microbiologically confirmed VAP had a borderline frequentist CI, OR 0.47; 95% CI 0.23 to 0.97; P=0.039.
  • The Bayesian analyses were post-hoc and exploratory: primary endpoint posterior probability of benefit was only approximately 72–79%, whereas clinically diagnosed and microbiologically confirmed VAP had posterior probabilities of benefit of approximately 99% and 98%, respectively.

Conclusion on Internal Validity: Internal validity is moderate. The neutral primary endpoint is credible because the trial achieved its target sample size and objective microbiological measurement, but inference about VAP reduction is less secure because it was secondary, open-label, not multiplicity-adjusted, and potentially affected by diagnostic subjectivity and unadjusted cluster randomisation.

External Validity

• Population Representativeness:
  • The population was clinically relevant for ICUs managing adults expected to require more than 48 hours of invasive ventilation.
  • The trial excluded short ventilation, recent invasive ventilation, pre-intubation inhalation, pregnancy, and contraindication to enteral feeding.
  • Only 270/1408 screened intubation episodes were randomised, so the findings apply to a selected prolonged-ventilation cohort rather than all intubated ICU patients.
  • The cohort was very ill: 28-day mortality was 41.7% versus 43.9%, and 90-day mortality was 46.5% versus 49.6%.
  • Approximately half had suspected pneumonia and positive Day 0 tracheal aspirates, limiting extrapolation to clean postoperative, neurological, or elective surgical ventilation populations.
• Applicability:
  • The trial is most applicable to high-resource ICUs able to stock device-specific endotracheal tubes, maintain an automated cuff/SSD control unit, and train staff in its use.
  • The comparator was not “usual unstructured care”; it was active manual cuff-pressure monitoring plus an SSD tube and intended hourly suction.
  • The trial therefore does not answer whether AnapnoGuard is better than a standard non-SSD endotracheal tube.
  • It also does not establish superiority over other continuous cuff-pressure controllers or other automated SSD systems.
  • Resource-limited systems may find the technology difficult to implement, and cost-effectiveness was not assessed.
  • The 2022 prevention guidance remains a useful external benchmark: SSD may reduce VAP incidence in selected patients expected to need prolonged ventilation, but stronger outcome benefits remain uncertain.¹

Conclusion on External Validity: External validity is moderate for high-resource ICUs using SSD tubes and manual cuff-pressure protocols in prolonged ventilation patients. Generalisability is limited for short-duration ventilation, emergency systems without ready device availability, low-resource settings, and ICUs not already using SSD-capable tubes.

Strengths & Limitations

• Strengths:
  • Prospectively registered international multicentre randomised trial.
  • Pragmatic cluster design addressing the practical difficulty of individual randomisation during urgent intubation.
  • Active comparator reflecting a high standard of manual airway care rather than a weak control.
  • Objective quantitative microbiological primary endpoint.
  • Excellent follow-up after inclusion in the analysis: no missing follow-up data reported.
  • Clear physiological separation for both cuff-pressure control and secretion drainage.
  • Standardised VAP-prevention bundle across groups.
  • Trial reported no sponsor; Fondazione Policlinico A. Gemelli IRCCS and Medinat Srl provided costs for insurance, eCRF, and device availability, with no declared role in design, data collection, interpretation, or publication decision.
• Limitations:
  • Open-label design.
  • Cluster randomisation without a clearly reported cluster-adjusted primary analysis.
  • Modified intention-to-treat analysis excluded 20 randomised patients after allocation.
  • Primary endpoint was a surrogate measured at Day 3, whereas VAP occurred later.
  • High baseline pneumonia, Day 0 colonisation, and antibiotic exposure reduced the biological clarity of an early colonisation endpoint.
  • VAP outcomes were secondary and potentially vulnerable to diagnostic bias.
  • No multiplicity adjustment for secondary endpoints was reported.
  • Combined intervention prevents attribution to one component.
  • Device-related mucosal injury was not comprehensively assessed; post-extubation stridor was reported, but systematic laryngeal or tracheal injury outcomes were not.
  • Cost-effectiveness, nursing/respiratory therapist workload, alarm burden, and implementation feasibility were not assessed.
  • Microaspiration biomarker data were limited to 20 patients in one centre.

Interpretation & Why It Matters

• Primary message

  Personalised automatic cuff-pressure control plus automatic SSD did not reduce Day 3 tracheobronchial colonisation compared with a manual cuff-pressure and manual SSD strategy.
• Biological signal

  The device did what it was intended to do mechanically: it reduced unsafe cuff-pressure readings and drained more subglottic secretions.
• Clinical signal

  The lower VAP rates are intriguing because both clinically diagnosed VAP and microbiologically confirmed VAP were reduced by approximately half, but the result should be treated as hypothesis-generating rather than definitive.
• Practice implication

  MICROINHALO does not justify routine adoption of this automated system solely to prevent early colonisation; it supports a larger blinded-adjudication trial powered for VAP and patient-centred outcomes.
• Conceptual implication

  The trial reinforces a recurring ICU prevention theme: improving a plausible physiological pathway may reduce a diagnosis such as VAP without necessarily improving ventilator duration, antibiotic exposure, length of stay, or mortality.

Controversies & Other Evidence

• The primary endpoint may have been misaligned with the clinical event of interest.
  • Day 3 tracheal colonisation was selected as a mechanistic surrogate, but VAP occurred at a median of 6–7 days.
  • Approximately half the cohort had suspected pneumonia and positive tracheal cultures at enrolment, making “prevention of new colonisation” difficult to demonstrate.
  • A later serial-culture strategy might have been more biologically informative, but would also have increased dropout from extubation and death.
• The VAP signal is clinically interesting but not definitive.
  • Clinically diagnosed VAP was lower with automatic management: 12.6% versus 24.4%, OR 0.45; 95% CI 0.23 to 0.87; P=0.016.
  • Microbiologically confirmed VAP was lower: 10.2% versus 19.5%, OR 0.47; 95% CI 0.23 to 0.97; P=0.039.
  • There was no parallel improvement in IMV-free days, antibiotic-free days, ICU-free days, hospital length of stay, or mortality.
  • This discordance is consistent with the broader VAP-prevention literature, where SSD and cuff-control interventions often reduce VAP diagnoses more reliably than patient-centred outcomes.¹³
• Continuous cuff-pressure control remains contested.
  • A meta-analysis of continuous cuff-pressure control found lower VAP risk, OR 0.51; 95% CI 0.31 to 0.82, but judged certainty very low because of bias, heterogeneity, and conflicts of interest.²
  • Earlier physiological RCT evidence showed reductions in gastric microaspiration, tracheal bacterial burden, and VAP with continuous cuff-pressure control.⁶
  • The accompanying editorial critique of the cuff-pressure meta-analysis argued that claims for continuous cuff-pressure control may be overinflated, particularly when trials are small, unblinded, device-dependent, and focused on VAP diagnosis rather than harder outcomes.⁷
• SSD evidence remains favourable for VAP incidence but inconsistent for broader outcomes.
  • The 2020 SSD overview and updated meta-analysis reported reduced VAP incidence, RR 0.56; 95% CI 0.48 to 0.63, but outcome translation remained less certain.³
  • A 2026 systematic review again found reduced VAP incidence with SSD, OR 0.42; 95% CI 0.29 to 0.61, and reduced VAP per 1000 ventilator-days, OR 0.44; 95% CI 0.29 to 0.66, but did not establish a consistent independent effect on ventilation duration or hospital length of stay.⁸
• PreVent 2 is an important counterweight.
  • PreVent 2 randomised 1068 emergency-intubation patients to a polyurethane-cuff ETT with SSD versus standard PVC ETT and found no reduction in possible VAP, IVAC, VAC, mortality, quality of life, cognitive outcomes, or laryngeal injury at 6 months.⁹
  • The accompanying editorial argued against routine faith in subglottic suction and polyurethane cuff technology as the answer to preventing downstream morbidity after emergency intubation.¹⁰
  • MICROINHALO differs from PreVent 2 because it compared automatic personalised management against an active manual SSD/cuff-pressure strategy, not against a standard non-SSD tube.
  • Taken together, the trials suggest that future device trials should prioritise adjudicated infection outcomes, ventilator-associated event frameworks, antibiotic exposure, mucosal safety, long-term airway outcomes, and cost-effectiveness rather than VAP incidence alone.

Summary

• MICROINHALO randomised 270 critically ill adults expected to need more than 48 hours of invasive ventilation; 250 were analysed.
• The trial did not meet its primary endpoint: Day 3 tracheal colonisation was 37.0% with automatic management versus 41.5% with manual management, OR 0.83; 95% CI 0.50 to 1.38; P=0.52.
• The automated system improved the intended delivery variables: out-of-range cuff pressures were 10.2% versus 24.5%, and median daily SSD volume was 25 versus 10.5 mL/day.
• VAP was lower with automatic management, but this was a secondary signal: clinically diagnosed VAP 12.6% versus 24.4%; microbiologically confirmed VAP 10.2% versus 19.5%.
• No benefit was demonstrated for IMV-free days, antibiotic-free days, hospital length of stay, or mortality.

Overall Takeaway

MICROINHALO is an important mechanistic device trial. It shows that personalised automatic cuff-pressure control plus automatic SSD can improve airway-management physiology and may reduce VAP diagnoses, but it did not reduce the prespecified primary colonisation endpoint or improve patient-centred outcomes.

Bibliography

• 1.Klompas M, Branson R, Cawcutt K, Crist MB, Eichenwald EC, Greene LR, et al. Strategies to prevent ventilator-associated pneumonia, ventilator-associated events, and nonventilator hospital-acquired pneumonia in acute-care hospitals: 2022 update. Infect Control Hosp Epidemiol. 2022;43:687-713.
• 2.Maertens B, Lin F, Chen Y, Rello J, Lathyris D, Blot S. Effectiveness of continuous cuff pressure control in preventing ventilator-associated pneumonia: a systematic review and meta-analysis of randomized controlled trials. Crit Care Med. 2022;50:1430-1439.
• 3.Pozuelo-Carrascosa DP, Herráiz-Adillo Á, Alvarez-Bueno C, Añón JM, Martínez-Vizcaíno V, Cavero-Redondo I. Subglottic secretion drainage for preventing ventilator-associated pneumonia: an overview of systematic reviews and an updated meta-analysis. Eur Respir Rev. 2020;29:190107.
• 4.Efrati S, Bolotin G, Levi L, Zaaroor M, Guralnik L, Weksler N, et al. Optimization of endotracheal tube cuff pressure by monitoring CO2 levels in the subglottic space in mechanically ventilated patients: a randomized controlled trial. Anesth Analg. 2017;125:1309-1315.
• 5.De Pascale G, Pennisi MA, Vallecoccia MS, Bello G, Maviglia R, Montini L, et al. CO2 driven endotracheal tube cuff control in critically ill patients: a randomized controlled study. PLoS One. 2017;12:e0175476.
• 6.Nseir S, Zerimech F, Fournier C, Lubret R, Ramon P, Durocher A, et al. Continuous control of tracheal cuff pressure and microaspiration of gastric contents in critically ill patients. Am J Respir Crit Care Med. 2011;184:1041-1047.
• 7.Bowton DL. Cuff pressure control: are the claims inflated? Crit Care Med. 2022;50:1535-1537.
• 8.Hu H, Tursuniyazi R. The efficacy of subglottic secretion drainage in preventing ventilator-associated pneumonia in the intensive care unit: a systematic review and meta-analysis. Front Med. 2026;13:1758199.
• 9.Treggiari MM, Sharp ES, Ohnuma T, Fajardo E, Aydin A, Akhtar S, et al. Hospital and long-term outcomes for subglottic suction and polyurethane cuff versus standard endotracheal tubes in emergency intubation (PreVent 2): a randomised controlled phase 2 trial. Lancet Respir Med. 2026;14:141-150.
• 10.Klompas M, Branson R. Time to pull the plug on subglottic secretion drainage? Lancet Respir Med. 2026;14:104-105.