Publication

• Title: Efficacy and safety of a paired sedation and ventilator weaning protocol for mechanically ventilated patients in intensive care (Awakening and Breathing Controlled trial): a randomised controlled trial.
• Acronym: ABC — Awakening and Breathing Controlled.
• Year: 2008; published in the 12 January issue.
• Journal published in: The Lancet.
• Citation: Girard TD, Kress JP, Fuchs BD, Thomason JWW, Schweickert WD, Pun BT, et al. Efficacy and safety of a paired sedation and ventilator weaning protocol for mechanically ventilated patients in intensive care (Awakening and Breathing Controlled trial): a randomised controlled trial. Lancet 2008;371:126-134.

Context & Rationale

• Background

  • Sedation facilitates invasive ventilation and relieves distress, but persistent sedation can obscure neurological recovery and delay recognition that ventilatory support is no longer needed. Decisions about sedation and ventilator liberation were often made separately by different members of the ICU team.
  • Kress and colleagues had shown that daily interruption of sedative infusions shortened ventilation and ICU stay in a single-centre medical ICU trial. Whether this approach added benefit when patients already received systematic breathing trials remained unresolved.¹
  • Daily respiratory screening, spontaneous breathing trials (SBTs) and notification of physicians after a successful trial had previously accelerated ventilator liberation. ABC therefore incorporated an active, protocolised weaning strategy in both groups.²
• Research Question/Hypothesis

  • Would pairing daily spontaneous awakening trials (SATs) with SBTs increase ventilator-free days compared with patient-targeted sedation and daily SBTs alone?
  • The hypothesis concerned coordination of two care processes: reducing sedation sufficiently to assess recovery, then promptly testing the ability to breathe without substantial assistance.
• Why This Matters

  • A patient may have recovered adequate respiratory function while remaining too sedated for clinicians to recognise readiness for extubation. Improving this sequence could shorten exposure to invasive support without requiring a new drug or device.
  • A multicentre trial could test whether the earlier single-centre sedation findings translated into reproducible clinical benefit, while directly measuring self-extubation, reintubation, brain dysfunction and longer-term survival.

Design & Methods

• Research Question:
  • In mechanically ventilated adults, does a daily SAT followed immediately, when appropriate, by an SBT improve the number of days alive and breathing without assistance through day 28 compared with usual sedation plus the same SBT protocol?
• Study Type:
  • Investigator-initiated, multicentre, parallel-group, 1:1 randomised trial in medical ICUs at four large US medical centres; recruitment ran from October 2003 to March 2006.
  • Recruiting centres were Saint Thomas Hospital, University of Chicago Hospitals, Hospital of the University of Pennsylvania and Penn Presbyterian Medical Center. Vanderbilt coordinated the trial; it was not an additional recruiting centre.
  • Allocation used computer-generated permuted blocks stratified by centre, prepared by a Vanderbilt biostatistician. Assignments were concealed on folded paper in consecutively numbered, sealed, opaque envelopes opened after consent.
  • Funding was from the Saint Thomas Foundation, National Institutes of Health and academic, geriatric and Veterans Affairs research programmes. Sponsors had no role in design, analysis, interpretation or writing; the corresponding author had full data access and responsibility for submission.
  • E. Wesley Ely disclosed grant support or honoraria from Pfizer, Hospira, Lilly and Aspect Medical. The other authors declared no conflicts of interest.
  • An independent data and safety monitoring board reviewed adverse events after 30 and 100 enrolments. No interim efficacy analyses were undertaken.
• Population:
  • Adults aged ≥18 years who required mechanical ventilation for ≥12 hours. Patients receiving full ventilatory support and those already undergoing weaning were eligible.
  • Exclusions were admission after cardiopulmonary arrest, continuous ventilation for ≥2 weeks, imminent death, withdrawal of life support, profound neurological deficits such as a large stroke or severe dementia, and enrolment in another trial.
  • Surgical patients were not enrolled; the investigators highlighted their potential requirement for continuous analgesia. Continuous sedative infusion was not explicitly required by the stated entry criteria.
  • Of 1,658 patients considered for enrolment, 336 were randomised, 168 per group. Immediate withdrawal of permission for data collection in one intervention patient left 167 versus 168 for analysis.
  • Median APACHE II scores were 26 versus 26.5; sepsis/acute respiratory distress syndrome accounted for 79/167 (47%) versus 87/168 (52%) of admission diagnoses. These categories were combined in the baseline table.
• Intervention:
  • Beginning the morning after enrolment, nurses, respiratory therapists or study personnel applied a daily SAT safety screen, performed an SAT when safe, and proceeded immediately to the SBT protocol after a successful SAT.
  • SAT screening excluded patients receiving sedation for active seizures or alcohol withdrawal, escalating sedation for ongoing agitation, neuromuscular blockade, evidence of active myocardial ischaemia during the preceding 24 hours, or raised intracranial pressure.
  • All sedatives and analgesics being used for sedation were interrupted. Analgesia needed for active pain was continued; the intervention did not mandate withdrawal of necessary pain relief.
  • An SAT was successful if the patient opened their eyes to verbal stimulation or tolerated interruption for ≥4 hours without failure. Full orientation or the ability to follow several commands was not required.
  • SAT failure comprised sustained anxiety, agitation or pain; respiratory rate >35/min for ≥5 minutes; oxygen saturation <88% for ≥5 minutes; an acute cardiac dysrhythmia; or at least two signs of respiratory distress. These included tachycardia, bradycardia, accessory muscle use, abdominal paradox, diaphoresis or marked dyspnoea.
  • After failure, sedatives were restarted at half the previous dose and titrated to comfort, with reassessment the following morning. A passed SAT triggered an SBT safety screen rather than automatic extubation.
• Comparison:
  • Patient-targeted sedation plus a daily SBT protocol, beginning the morning after enrolment. Clinicians could interrupt sedation when they judged this appropriate; interruption before the SBT was not mandatory.
  • In both groups, physicians and nurses selected sedatives, analgesics and doses according to the arousal and comfort considered appropriate for each patient. Centres used validated sedation scales, but no uniform light-sedation target was imposed.
  • The shared SBT safety screen required oxygen saturation ≥88% on inspired oxygen ≤0.50 and PEEP ≤8 cm H₂O, some spontaneous inspiratory effort within five minutes, no agitation, no myocardial ischaemia within 24 hours, and no raised intracranial pressure.
  • Significant vasoactive support precluded an SBT: dopamine or dobutamine ≥5 µg/kg/min, noradrenaline ≥2 µg/min, or any vasopressin or milrinone. The noradrenaline threshold was an absolute dose, not a weight-adjusted dose.
  • Eligible patients underwent a 120-minute SBT using a T-piece, CPAP of 5 cm H₂O, or pressure support <7 cm H₂O. Inspired oxygen and PEEP were not increased during the trial.
  • SBT failure criteria included respiratory rate >35/min or <8/min for ≥5 minutes, saturation <88% for ≥5 minutes, abrupt mental-status change, acute arrhythmia, or at least two signs of respiratory distress. Previous ventilator settings were restored after failure.
  • After a successful SBT, the treating physician was notified verbally. Extubation remained a clinical decision; study personnel did not decide whether to remove the endotracheal tube.
• Blinding:
  • Patients, bedside clinicians and study personnel were unblinded. Sedation interruption and the patient’s level of arousal were readily apparent.
  • Daily neurological assessments used the Richmond Agitation–Sedation Scale (RASS) and Confusion Assessment Method for the ICU (CAM-ICU). Mortality was objectively ascertainable, whereas extubation and discharge involved clinician decisions.
• Statistics:
  • Power calculation: 334 patients were required to detect a 25% increase in mean ventilator-free days, from 12.9 to 16.1 days, assuming SD 10.4 days, with 80% power (β=0.20) and two-sided α=0.05.
  • The primary outcome was ventilator-free days during the 28 days from enrolment. Unassisted breathing had to last ≥48 consecutive hours; any death within 28 days resulted in an allocation of zero ventilator-free days.
  • The analysis was described as intention-to-treat, but exclusion of the intervention patient who immediately withdrew made it a modified intention-to-treat analysis of 335 patients.
  • Continuous outcomes, including the primary endpoint, were compared using the Wilcoxon–Mann–Whitney test. A 2,000-sample bootstrap provided the 95% confidence interval for the difference in mean ventilator-free days; categorical outcomes used χ² tests.
  • Kaplan–Meier analyses and log-rank tests assessed discharge and survival. Death was treated as censoring in discharge analyses; the one-year mortality hazard ratio came from unadjusted Cox regression, with no detected violation of proportional hazards.
  • A centre-by-treatment interaction was examined for the primary outcome. Adjustment for multiple secondary comparisons was not reported; an independent biostatistician reanalysed the final dataset and verified the results.
• Follow-Up Period:
  • The primary observation period was 28 days from enrolment. Secondary outcomes included time to ICU and hospital discharge, 28-day mortality, one-year survival, and days with coma or delirium.
  • Coma was RASS −4 or −5; delirium required a positive CAM-ICU assessment while not comatose.
  • Vital status was followed to one year using hospital records, telephone calls, visits and the Social Security Death Master File. Patients lost to follow-up were censored at their last known contact.

Key Results

This trial was not stopped early. It completed recruitment with 336 randomised patients, exceeding the planned 334; 167 intervention patients and 168 controls were analysed. The monitoring board conducted two safety reviews but no interim efficacy analyses.

Outcome	SAT plus SBT	Usual sedation plus SBT	Effect	p value / 95% CI	Notes
Ventilator-free days to day 28	Mean 14.7 days	Mean 11.6 days	Mean difference 3.1 days	95% CI 0.7 to 5.6; P=0.02	Primary outcome; deaths within 28 days assigned zero days.
Time to ICU discharge	9.1 days (5.1–17.8)	12.9 days (6.0–24.2)	Not reported	P=0.01	Kaplan–Meier median (IQR); deaths censored.
Time to hospital discharge	14.9 days (8.9–26.8)	19.2 days (10.3–not estimable)	Not reported	P=0.04	Kaplan–Meier median (IQR); >25% of controls remained in hospital at day 28.
Death by day 28	47/167 (28%)	58/168 (35%)	Not reported	P=0.21	No statistically conclusive difference at this time point.
Death by one year	74/167 (44%)	97/168 (58%)	HR 0.68	95% CI 0.50 to 0.92; P=0.01	Prespecified secondary survival outcome; HR is not a risk ratio.
Days with coma	2 (0–4)	3 (1–7)	Not reported	P=0.002	Median (IQR); RASS −4 or −5.
Days with delirium	2 (0–5)	2 (0–6)	Not reported	P=0.50	Median (IQR); assessed when not comatose.
Any self-extubation	16/167 (10%)	6/168 (4%)	Absolute increase 6.0 percentage points	95% CI 0.6 to 11.8; P=0.03	Important safety signal; five intervention events occurred during or within 12 hours of an SAT.
Self-extubation requiring reintubation	5/167 (3%)	3/168 (2%)	Published difference 1.2 percentage points	Published 95% CI −5.2 to 2.5; P=0.47	Within 48 hours; printed effect and interval retained. Contrast direction warrants clarification.
Any reintubation	23/167 (14%)	21/168 (13%)	Published difference 1.3 percentage points	Published 95% CI −8.6 to 6.1; P=0.73	Within 48 hours; printed effect and interval retained. Absence of significance does not establish equivalent safety.
Tracheostomy	21/167 (13%)	34/168 (20%)	Absolute reduction 7.6 percentage points	95% CI −0.3 to 15.6; P=0.06	No tracheostomies were present at enrolment.
Dysrhythmia during an SBT	15/603 trials (3%)	9/948 trials (1%)	Absolute increase 1.6 percentage points	95% CI 0.3 to 3.2; P=0.02	Trial-level events, not patient-level incidence; no reported sequelae beyond stopping the SBT.

• The primary finding was a gain of 3.1 mean ventilator-free days, with a confidence interval spanning a modest to substantial benefit. This composite reflects both survival and freedom from ventilation; it is not a 3.1-day reduction in ventilation for every patient.
• One-year mortality favoured SAT plus SBT, whereas the 28-day comparison was inconclusive. The published one-year number needed to treat was 7.4 (95% CI 4.2 to 35.5), but this comes from a secondary outcome in a trial powered for ventilator-free days.
• Self-extubation increased from 6 to 16 patients. The trial did not establish equivalence for reintubation, despite the absence of a statistically significant difference in that outcome.

Internal Validity

• Randomisation and Allocation:
  • Computer-generated allocation stratified by centre and sequential opaque envelopes provided reasonable concealment. The block size was not reported.
  • Consent preceded allocation, reducing the opportunity to select patients after discovering treatment assignment. The envelope procedure was less independently controlled than a central electronic service, but no breach was described.
• Dropouts and Post-randomisation Exclusions:
  • One of 168 intervention assignments was excluded before protocol initiation or any data collection because the surrogate withdrew consent. This small departure from full intention-to-treat is transparent.
  • Seven controls discontinued the protocol: three withdrew but permitted outcome collection, and four transferred for surgery. All 168 remained in the analysis; these seven discontinuations should not be described as seven missing outcomes.
  • The trial profile records two controls lost to follow-up, in addition to the immediately withdrawn intervention patient. Survival analysis censored at last contact; the exact consequences depend on when follow-up ceased.
• Performance and Detection Bias:
  • Identical SBT criteria and notification procedures reduced differences in respiratory assessment between groups. Extubation nevertheless depended on unblinded clinicians, and awareness of allocation could affect the speed of acting on a successful SBT.
  • RASS and CAM-ICU standardised neurological assessment, but assessors were not masked. Mortality is less susceptible to detection bias, although treatment knowledge can still influence care before death.
• Baseline Characteristics:
  • Median age was 60 versus 64 years, APACHE II 26 versus 26.5, and SOFA 9 versus 8. Median first-day RASS was −4 in both groups; 94/167 (56%) versus 87/168 (52%) were comatose.
  • The cohort was sufficiently ill and deeply sedated to offer substantial scope for improving recognition of recovery. Similar severity scores do not eliminate chance prognostic imbalance in a trial of this size.
  • Pre-enrolment propofol exposure was higher in the intervention group: median 5,102 versus 3,248 mg; P=0.02. The investigators found no association with outcomes, but supporting estimates were not presented.
• Timing and Heterogeneity:
  • Median admission-to-enrolment time was 2.2 days in both groups; protocol delivery began the next morning. ABC addressed ongoing ICU sedation and liberation, rather than randomising the first sedative dose at intubation.
  • Four centres and a range of medical diagnoses improved replication across settings. No centre-by-treatment interaction was detected for ventilator-free days; numerical interaction estimates were not reported. This does not demonstrate equal effects in every diagnosis or severity subgroup.
• Protocol Adherence and Dose:
  • An SAT was delivered to 150/167 intervention patients (90%). Seventeen never passed the SAT safety screen; withholding the intervention under those circumstances was part of the assigned strategy.
  • Of 1,140 SAT safety screens, 939 were passed and 895 SATs were performed: 95% delivery after a passed screen. There were 837 successful SATs and 58 failures; agitation was the commonest reason for failing the preliminary screen.
  • An SBT was performed in 136/167 (81%) versus 146/168 (87%); P=0.17. The 31 versus 22 patients who never passed an SBT safety screen were appropriately retained in the assigned groups.
  • The tested dose was a daily opportunity to remove unnecessary sedation, conditioned on safety, with half-dose restart after failure. The trial did not compare alternative interruption frequencies or establish an optimal duration for every patient.
• Separation of the Variable of Interest:
  • Sedatives were held before an SBT in 150/167 (90%) versus 52/168 (31%). The control strategy therefore included discretionary sedation holds, despite zero protocol-defined SATs in its treatment table.
  • Median cumulative post-enrolment benzodiazepine exposure was 20 versus 39 mg in lorazepam equivalents; P=0.02. Median average daily exposure was 2 versus 3 mg; P=0.12.
  • Median cumulative propofol exposure was 8,950 versus 8,380 mg; P=0.90. Cumulative opioid exposure was 2,662 versus 3,700 µg in fentanyl equivalents; P=0.07.
  • At the first successful SBT, median RASS was −1 versus −2.5; P=0.0001. Among patients who passed an SBT, same-day extubation at the first successful trial occurred in 59/109 (54%) versus 49/124 (40%); published difference 14.6 percentage points, 95% CI 1.0 to 26.0; P=0.03.
• Key Delivery Aspects and Co-interventions:
  • Bedside nurses and respiratory therapists could deliver the intervention, with study-personnel involvement and round-the-clock investigator support. Staffing intensity and the additional time required were not quantified.
  • Analgesia for active pain continued during 132/895 SATs (15%). This records opioid continuation during SATs; it does not show that all other trials involved untreated pain.
  • Haloperidol was administered to 31 patients (18%) versus 45 (27%); P=0.07. Other clinical care and sedative selection remained discretionary; the intervention was a care strategy rather than an isolated pharmacological comparison.
• Outcome Assessment and Statistical Rigour:
  • The primary endpoint, its 28-day horizon and the treatment of death were specified a priori. The planned sample was achieved, with no efficacy-driven early stopping.
  • Rank-based testing accommodated the unusual distribution of ventilator-free days, while bootstrapping quantified uncertainty around the mean difference. These are complementary but different summaries of the outcome distribution.
  • Independent statistical verification strengthened confidence in computation. The main limitations for inference are the influence of unblinded decisions on liberation, the treatment of death in discharge analyses, and multiple secondary comparisons, discussed below.

Conclusion on Internal Validity: Moderate overall. Concealed randomisation, substantial protocol separation, minimal exclusion and a prespecified primary endpoint support a credible benefit of the assigned care strategy; unblinded liberation decisions and less secure secondary inference limit confidence in its mechanism and survival effect.

External Validity

• Population Representativeness:
  • The trial included seriously ill adults with common medical ICU diagnoses, including sepsis/ARDS, cardiac disease and obstructive airway disease. Community and university settings, and open and closed ICU arrangements, were represented.
  • Applicability is narrower for surgical patients, profound neurological injury, post-cardiac-arrest care, prolonged ventilation and patients approaching withdrawal of life support, all absent or excluded.
  • The 1,322 exclusions included 324 surrogate or physician refusals and 306 patients unable to provide consent. Recruitment therefore depended substantially on consent availability and clinician willingness, beyond physiological eligibility.
• Applicability to Contemporary Practice:
  • The strongest direct applicability is to units where ongoing sedation delays assessment of extubation readiness despite routine SBTs. Baseline RASS −4 in both groups indicates considerable scope for improvement in the original setting.
  • An ICU already maintaining comfortable, responsive patients throughout the day may have less scope for an additional scheduled interruption to change outcomes. The size of the original benefit should not be transferred automatically to such practice.
  • The operational principle is portable to different healthcare systems, but safe delivery requires pain assessment, airway observation, coordinated staffing and prompt action after a successful SBT.
  • Patients requiring deep sedation or neuromuscular blockade for a specific indication need reassessment of that indication and SAT eligibility. Inclusion of some patients with ARDS does not establish safety of interruption during every phase of severe respiratory failure.

Conclusion on External Validity: Moderate for eligible medical ICU patients when sedation impedes liberation, with more limited direct applicability to excluded populations and contemporary units already achieving sustained light sedation. The coordination principle is more generalisable than the trial’s exact effect size or every historical safety threshold.

Strengths & Limitations

• Strengths:
  • A clinically important process question tested against an active SBT comparator, rather than leaving ventilator weaning entirely unstructured.
  • Concealed randomisation, centre stratification, completion of the planned sample and high delivery after successful safety screening.
  • Explicit SAT and SBT algorithms, including pain preservation, safety exclusions, failure criteria and a defined restart strategy.
  • A patient-important primary endpoint, detailed treatment-delivery data, prospective adverse-event surveillance and survival follow-up to one year.
• Limitations:
  • Modest sample size for mortality and uncommon harms, with one post-randomisation exclusion and some incomplete long-term follow-up.
  • Unblinded clinicians and assessors, discretionary extubation, and a control sedation strategy without a uniform light-sedation target.
  • No factorial separation of awakening, temporal coordination, medication reduction and the additional attention associated with protocol delivery.
  • Limited assessment of contemporaneous pain, distress, patient experience and staff workload; restricted enrolment outside medical ICUs.
  • Secondary-outcome multiplicity and competing death complicate interpretation of survival, discharge and neurological outcomes.

Interpretation & Why It Matters

• Clinical significance

  • ABC showed that liberation can be delayed by the organisation of care as well as by persisting respiratory failure. Pairing awakening with respiratory assessment increased mean ventilator-free days from 11.6 to 14.7.
  • The practical implication is to make sedation requirements and readiness for ventilator liberation explicit daily decisions, with communication between those managing sedation and those assessing breathing.
• Likely mechanism

  • Time to the first passed SBT was similar: median 3.8 versus 3.9 days; P=0.49. Intervention patients were more awake when they passed and more often extubated that day.
  • This pattern supports the interpretation that awakening helped clinicians recognise and act on combined neurological and respiratory readiness. It does not demonstrate faster resolution of the underlying lung disease or isolate a single biological mechanism.
• Meaning of the outcomes

  • Fewer coma days and earlier liberation are clinically coherent benefits. Unchanged delirium duration does not establish delirium prevention, and the one-year mortality finding should remain distinct from the primary result.
  • The increase in self-extubation means that successful implementation must preserve comfort and airway safety. A sedation hold is a monitored assessment with explicit stopping rules.

Controversies & Other Evidence

• Was the Control Sedation Strategy an Adequate Comparator?
  • Brochard’s editorial questioned performing SBTs without first requiring sedation reduction. In ABC, clinicians could hold sedation in controls, but were not required to do so; patients could therefore pass respiratory testing while remaining too sedated for extubation.³
  • This is a legitimate limitation of the treatment contrast. An active SBT comparator strengthened ABC, but it did not make the control arm equivalent to a modern protocol that reliably maintains light sedation and assesses neurological readiness alongside breathing.
  • The control strategy reflected participating centres’ usual patient-targeted sedation rather than an explicitly prescribed deep-sedation policy. Describing it as deliberately harmful care goes beyond the evidence; describing it as optimised contemporary sedation also overstates what was tested.
  • SLEAP subsequently randomised 430 patients in 16 ICUs to protocolised light sedation with or without daily interruption. Median time to successful extubation was seven days in both groups: HR 1.08; 95% CI 0.86 to 1.35; P=0.52. Interruption increased drug exposure and nursing workload without improving the primary outcome.⁴
  • The inference from these contrasting trials is that the incremental value of interruption depends on the sedation strategy it changes. SLEAP does not invalidate ABC’s original comparison, but it weakens a universal claim that interruption necessarily adds benefit to effective light sedation.⁴
• Does More Failed Breathing Testing Explain the Control Outcomes?
  • The editorial raised concern that early SBTs under continuing sedation could impose repeated physiological stress, and that knowledge of failed trials might discourage clinicians from extubating later.³
  • Controls had 456 failed SBTs versus 284 in the intervention group, but they also underwent 948 versus 603 SBTs. The published per-trial failure rates were 48% versus 47%, with similar proportions passing and P=0.70 for the pass comparison.
  • Larger cumulative failure counts therefore do not establish greater failure probability during an individual trial. Longer exposure to ventilation creates more opportunities for repeated testing, and multiple trials within one patient are not independent observations.
  • The similar time to first successful SBT, together with more wakefulness and more same-day extubation after awakening, supports delayed recognition or action as an explanation. Repeated-test stress as a cause of excess late mortality remains unproven.
• What Can Be Attributed to Awakening Itself?
  • ABC randomised a sequence of care, not awakening in isolation. Sedation withdrawal, altered arousal, communication, prompt respiratory assessment and clinician response all formed part of the intervention.
  • Unblinded extubation decisions can be both a route through which the strategy works and a source of performance bias. Earlier appropriate action after a successful SBT is clinically useful, but the design cannot separate this from expectations generated by allocation.
  • Lower cumulative benzodiazepine exposure, 20 versus 39 mg, does not prove that drug reduction alone mediated benefit. A shorter treatment course can itself reduce cumulative exposure; average daily doses were not conclusively different, and cumulative propofol exposure was similar.
  • A factorial design or a comparator with equally achieved arousal and equally prompt extubation assessment would be needed to distinguish these components more cleanly. ABC did not perform that comparison.
• How Persuasive Is the One-Year Survival Benefit?
  • One-year survival was a prespecified secondary endpoint, not a post hoc discovery. Its HR of 0.68 (95% CI 0.50 to 0.92) deserves attention, particularly because ascertainment of death is comparatively objective.
  • However, the trial was powered for ventilator-free days, evaluated several secondary outcomes without reported multiplicity adjustment, and produced a wide confidence interval around the number needed to treat. The size of the apparent survival benefit is consequently less secure than the existence of a primary-outcome benefit.
  • The 28-day mortality comparison, 28% versus 35% with P=0.21, neither proves absence of an early effect nor confirms that benefit arose only after hospital discharge. Differences in statistical significance across time points are not evidence that the treatment effects themselves differ.
  • Avoiding prolonged sedation and invasive support offers plausible pathways to better survival, but specific mediators and causes of late death were not established. Brochard highlighted the unresolved explanation for the substantial longer-term mortality separation.³
  • Chance prognostic imbalance remains possible: controls were older by their group medians, while intervention patients had greater prior propofol exposure. Neither observation proves that the survival result is confounded, but the published unadjusted estimate should not be treated as immune to small-trial uncertainty.
• Ventilator-Free Days, Discharge and Brain Dysfunction Measure Different Things:
  • Assigning zero ventilator-free days to every death within 28 days gives appropriate weight to mortality but combines two distinct outcomes. The mean gain cannot identify how much arose from earlier liberation and how much from improved survival.
  • Deaths were censored in the Kaplan–Meier discharge analyses, although death prevents discharge alive. These estimates describe a hypothetical discharge process after removing competing deaths; they should not be interpreted as ordinary observed lengths of stay or as direct estimates of days alive outside hospital.
  • Coma fell from a median three to two days, while delirium remained two days in both groups. Comatose patients cannot be assessed for delirium using CAM-ICU, so more time awake also changes the opportunity to detect delirium.
  • Consequently, ABC supports less deep unresponsiveness under the assigned strategy, but does not establish prevention of delirium or permanent neurological injury. Shorter ICU observation can also influence the number of recorded days with brain dysfunction.
• Safety, Analgesia and Patient Experience:
  • The increase in self-extubation was statistically supported: absolute increase 6.0 percentage points, 95% CI 0.6 to 11.8. Only five intervention self-extubations occurred during or within 12 hours of an SAT, but this does not exclude a contribution from the broader change in arousal.
  • Reintubation after self-extubation occurred in five versus three patients, and any reintubation in 23 versus 21. Neither comparison was designed to demonstrate non-inferiority, so a non-significant P value cannot justify an unqualified claim of equivalent airway safety.
  • The publication prints positive between-group differences for reintubation alongside confidence intervals with an apparently reversed contrast direction. These values are reproduced as printed in the table; they have not been recalculated or silently corrected.
  • The excess dysrhythmia observations during SBTs, 15/603 versus 9/948 trials, warrant proportionate attention. These were repeated trial-level events, with no reported clinical sequelae beyond stopping the test; multiplicity and within-patient clustering limit the strength of inference.
  • Brochard questioned the limited measurement of pain, distress and unpleasant experiences during interruption. Continuing analgesia for active pain was an explicit safeguard, but a protocol rule is not equivalent to comprehensive measurement of patient comfort.³
  • Anxiety, agitation or pain contributed to 42 failed SATs, and analgesics continued during 15% of SATs. These figures neither demonstrate absence of distress nor establish that analgesia was inappropriate during the remaining trials.
• Long-Term Cognitive and Psychological Follow-Up:
  • A planned single-centre substudy included 180 ABC patients, with a blinded follow-up assessor. Assessments were obtained in 80/99 eligible patients at three months and 63/86 at 12 months after discharge.⁵
  • Cognitive impairment was less frequent at three months, 70% versus 91% (published absolute reduction 20.2 percentage points; 95% CI 1.5 to 36.1; P=0.03), but not at 12 months, 72% versus 70%; P=0.89. Composite cognitive scores were similar.⁵
  • PTSD symptoms occurred in 24% of each group at 12 months; P=0.97. These findings are reassuring but do not prove psychological equivalence: the sample was small, follow-up incomplete, and testing among survivors was affected by differential mortality.⁵
• Subsequent Randomised Evidence and Evidence Synthesis:
  • The directly relevant Cochrane review included nine trials and 1,282 patients. Daily interruption produced an estimated 13% reduction in the geometric mean duration of ventilation, with a 95% CI from a 26% reduction to a 2% increase. ICU mortality was not conclusively reduced: RR 0.96; 95% CI 0.77 to 1.21. Heterogeneous comparators limited a single universal estimate.⁶
  • NONSEDA analysed 700 patients assigned to no sedation or light sedation with daily interruption. Ninety-day mortality was 42.4% versus 37.0%; published difference 5.4 percentage points, 95% CI −2.2 to 12.2; P=0.65. This tested a different contrast and does not establish that eliminating sedation altogether improves outcomes beyond light sedation.⁷
  • A 2024 systematic review of 15 randomised trials involving 4,352 patients found no conclusive mortality advantage from lighter versus deeper sedation: RR 0.94; 95% CI 0.83 to 1.06. Thirteen trials were judged at high risk of bias. This broader comparison does not replicate ABC’s exact protocol, but argues against treating a large survival benefit as an established class effect of lighter sedation.⁸
  • A 2026 network meta-analysis of 39 trials and 11,946 participants estimated 1.84 additional ventilator-free days with daily interruption versus usual care in its merged strategy network (95% CI 0.20 to 3.48). Grouping different usual-care regimens, mixing process and drug comparisons, and reliance on indirect evidence constrain interpretation. It does not demonstrate an incremental benefit over consistently achieved protocolised light sedation.⁹
• From ABC to Broader Liberation Bundles:
  • The ICU Liberation Collaborative studied 15,226 adults in 68 ICUs. Complete ABCDEF bundle performance was associated with lower hospital mortality over the following seven-day observation period: adjusted HR 0.32; 95% CI 0.17 to 0.62. Bundle delivery was observational, so residual confounding and greater feasibility of care in recovering patients remain important; the effect cannot be assigned to SATs alone.¹⁰
  • A 2026 secondary analysis of 566 MIND-USA participants examined ABCDE compliance and longer-term recovery. Among those followed, median compliance was 97%; 304 patients underwent three-month assessment and 251 underwent 12-month assessment. Associations with function and quality of life were inconsistent, and better cognition, mental health or survival was not established. High adherence with little exposure variation and survivor selection limited causal inference.¹¹
• Implementation Remains a Substantial Part of the Intervention:
  • A 2023 study of 1,901 ventilated admissions across 15 ICUs estimated coordinated SAT/SBT delivery within two hours on 21% of eligible days, ranging from 9% to 68% between sites. Interviews identified workflow, role clarity, knowledge and measurement barriers. These process findings show why issuing a protocol alone may not reproduce ABC.¹²
  • A comparison published in 2025 examined ABC intervention patients and 2,083 PILOT patients. SATs were performed after 95% versus 98% of passed safety screens, but screening eligibility and SAT success differed substantially. This supports feasibility in an established system, while the non-randomised comparison across eras cannot establish equivalent clinical effectiveness.¹³
• Contemporary Guidelines and the Remaining Clinical Question:
  • The 2018 PADIS guideline conditionally favoured light over deep sedation and stated that either daily interruption or nursing protocols can achieve light sedation. It did not establish that combining both strategies is necessary for every patient.¹⁴
  • The 2025 focused PADIS update conditionally favours dexmedetomidine over propofol when light sedation or delirium reduction is a leading priority. That recommendation concerns drug selection and does not provide new proof that routine interruption adds benefit to effective light sedation.¹⁵
  • The 2024 AARC guideline supports standardised daily SBT assessment, preferably completing an appropriate trial before noon, and permits SBTs with or without pressure support. These conditional recommendations support organised liberation without making every ABC ventilator setting or screening threshold a universal requirement.¹⁶
  • The remaining question is which patients need interruption to reveal readiness that routine titration has obscured. Implementation should assess achieved arousal, comfort, coordination and timely extubation, as well as whether a sedation hold was documented.

Summary

• ABC randomised 336 mechanically ventilated adults in four US medical centres; 335 were analysed. Both groups received daily SBTs, while the intervention added a safety-screened SAT immediately before respiratory assessment.
• Mean ventilator-free days increased from 11.6 to 14.7: difference 3.1 days; 95% CI 0.7 to 5.6; P=0.02. Coma duration and time to discharge also favoured the intervention.
• One-year mortality was 44% versus 58%: HR 0.68; 95% CI 0.50 to 0.92. This prespecified secondary finding is less definitive than the primary result and has not established a universal survival effect of sedation interruption.
• Self-extubation increased from 4% to 10%. Similar observed reintubation counts do not establish equivalent safety; preserving analgesia and monitoring for distress remain integral to the tested approach.
• The enduring contribution is coordination of sedation management with ventilator liberation. Later evidence makes clear that the incremental benefit of daily interruption depends on how effectively the comparator already achieves light sedation.

Overall Takeaway

ABC is a landmark trial because it showed that coordinating awakening with breathing assessment can improve liberation from invasive ventilation. Its strongest evidence is the gain in ventilator-free days over the sedation practice tested, accompanied by increased self-extubation and a promising but less certain secondary survival finding. Modern application should preserve its coordinated, safety-screened approach while recognising that mandatory interruption has not shown consistent additional benefit when light sedation is already reliably achieved.

Bibliography

• 1.Kress JP, Pohlman AS, O’Connor MF, Hall JB. Daily interruption of sedative infusions in critically ill patients undergoing mechanical ventilation. N Engl J Med 2000;342:1471-1477.
• 2.Ely EW, Baker AM, Dunagan DP, Burke HL, Smith AC, Kelly PT, et al. Effect on the duration of mechanical ventilation of identifying patients capable of breathing spontaneously. N Engl J Med 1996;335:1864-1869.
• 3.Brochard L. Sedation in the intensive-care unit: good and bad? Lancet 2008;371:95-97.
• 4.Mehta S, Burry L, Cook D, Fergusson D, Steinberg M, Granton J, et al.; SLEAP Investigators; Canadian Critical Care Trials Group. Daily sedation interruption in mechanically ventilated critically ill patients cared for with a sedation protocol: a randomized controlled trial. JAMA 2012;308:1985-1992.
• 5.Jackson JC, Girard TD, Gordon SM, Thompson JL, Shintani AK, Thomason JWW, et al. Long-term cognitive and psychological outcomes in the Awakening and Breathing Controlled trial. Am J Respir Crit Care Med 2010;182:183-191.
• 6.Burry L, Rose L, McCullagh IJ, Fergusson DA, Ferguson ND, Mehta S. Daily sedation interruption versus no daily sedation interruption for critically ill adult patients requiring invasive mechanical ventilation. Cochrane Database Syst Rev 2014;(7):CD009176.
• 7.Olsen HT, Nedergaard HK, Strøm T, Oxlund J, Wian KA, Ytrebø LM, et al. Nonsedation or light sedation in critically ill, mechanically ventilated patients. N Engl J Med 2020;382:1103-1111.
• 8.Ceric A, Holgersson J, May TL, Skrifvars MB, Hästbacka J, Saxena M, et al. Effect of level of sedation on outcomes in critically ill adult patients: a systematic review of clinical trials with meta-analysis and trial sequential analysis. EClinicalMedicine 2024;71:102569.
• 9.Zhang L, Zhou J, Liu X, Zhou M, Chen W, Jin J, et al. Comparative effectiveness and clinical credibility of nurse-implementable sedation strategies for mechanically ventilated adults in intensive care: a systematic review and network meta-analysis. Front Med (Lausanne) 2026;13:1830570.
• 10.Pun BT, Balas MC, Barnes-Daly MA, Thompson JL, Aldrich JM, Barr J, et al. Caring for critically ill patients with the ABCDEF bundle: results of the ICU Liberation Collaborative in over 15,000 adults. Crit Care Med 2019;47:3-14.
• 11.Boehm LM, Mart MF, LaNoue M, Siao SF, Pun BT, Bowton DL, et al. Association between ABCDE bundle compliance and long-term outcomes: a secondary longitudinal analysis. Intensive Care Med 2026;52:466-475.
• 12.Olsen GH, Gee PM, Wolfe D, Winberg C, Carpenter L, Jones C, et al. Awakening and breathing coordination: a mixed-methods analysis of determinants of implementation. Ann Am Thorac Soc 2023;20:1483-1490.
• 13.Alkhateeb T, Semler MW, Girard TD, Ely EW, Stollings JL. Comparison of SAT and SBT conduct during the ABC trial and PILOT trial. J Intensive Care Med 2025;40:3-9.
• 14.Devlin JW, Skrobik Y, Gélinas C, Needham DM, Slooter AJC, Pandharipande PP, et al. Clinical practice guidelines for the prevention and management of pain, agitation/sedation, delirium, immobility, and sleep disruption in adult patients in the ICU. Crit Care Med 2018;46:e825-e873.
• 15.Lewis K, Balas MC, Stollings JL, McNett M, Girard TD, Chanques G, et al. A focused update to the clinical practice guidelines for the prevention and management of pain, anxiety, agitation/sedation, delirium, immobility, and sleep disruption in adult patients in the ICU. Crit Care Med 2025;53:e711-e727.
• 16.Roberts KJ, Goodfellow LT, Battey-Muse CM, Hoerr CA, Carreon ML, Sorg ME, et al. AARC clinical practice guideline: spontaneous breathing trials for liberation from adult mechanical ventilation. Respir Care 2024;69:891-901.