Automated Weaning: Smarter Approach or Just Clinician Recalcitrance?

Abstract & Commentary

Director of Quality Assurance, Respiratory Care Services, San Francisco General Hospital

Financial Disclosure: Mr. Kallet reports no financial relationships relevant to this field of study.

SYNOPSIS: In this study, closed-loop, titrated pressure support ventilation was associated with a significantly shorter time course until patients passed a spontaneous breathing trial and were successfully extubated compared to a standardized weaning protocol.

SOURCE: Burns KE, et al. Wean earlier and automatically with new technology (The WEAN study): A multicentre, pilot randomized controlled trial. Am J Respir Crit Care Med 2013;187:1203-1211.

This pilot study compared closed-loop pressure support (PS) ventilation to a standard spontaneous breathing trial (SBT) protocol in a multidisciplinary ICU setting. The primary objective was to test clinician adherence to automated weaning vs a clinician-directed weaning protocol, whereas duration of mechanical ventilation and successful extubation were secondary objectives. Eligible study patients required more than 24 hours of mechanical ventilation, and randomization only occurred in those who could tolerate a trial of moderately high PS level (i.e., 15 cm H2O driving pressure) for 60 minutes. Standard criteria for weaning readiness were used. Patients randomized to automated weaning had PS titrated to maintain a "comfort zone" (i.e., respiratory rate 15-34 breaths/min; minimum tidal volume 300 mL; end-tidal carbon dioxide tension < 55 mmHg). This was achieved by closed-loop evaluation of the respiratory pattern every 2-4 minutes with subsequent PS titration in 2-4 cm H2O increments until the patient could achieve the comfort zone at a standardized SBT-PS setting of 7 cm H2O. The standard weaning arm consisted of daily SBTs using traditional settings and modes.

Over a 26-month period, nine participating ICUs enrolled and randomized 97 patients. Baseline characteristics were not different between groups. There was a significantly higher incidence of protocol violations (nearly 2.5 times greater) in the automated weaning group that corresponded with surveys indicating lower acceptance of this approach by both physicians and respiratory care practitioners. Despite these issues, automated weaning was associated with a reduced median time to the first successful SBT (4 vs 1 day, P < 0.0001), time to first extubation trial (4.5 vs 3 days, P = 0.02), and time to first successful extubation (5 vs 4 days, P = 0.01), with fewer tracheostomies (35% vs 16%, P = 0.04). Both post-extubation requirements for noninvasive ventilation (14% vs 8.2%, P = 0.5) and extubation failure rate (25.6% vs 18.4%, P = 0.4) were not different.


What is most striking about this study is that despite passing an SBT at a median of 1 day, automated weaning patients who (theoretically) should have been extubated at that point remained intubated for an additional 2 days. And patients in the protocolized weaning arm required 3 days to pass an SBT. This finding is particularly salient because large, randomized, controlled trials uniformly reported 75-80% of critically ill patients (with similar durations of mechanical ventilation) are able to pass their initial SBT.1 This finding coincided with a relatively high incidence of protocol noncompliance in both study arms (47 total days; average off-protocol duration of approximately 18-23 hours).

Moreover, poor screening and enrollment performances (7 and 0.4 patients per month, per ICU, respectively) suggest insufficient support for investigators/study coordinators. The success of the ARDSNet studies was in part due to robust National Institutes of Health financial support, which allowed for multiple daily interactions between study personnel and clinicians that facilitated protocol adherence.2 Regardless of whether algorithms are written or programmed into a ventilator mode, practitioners are reluctant to "trust" anything other than their own judgment. This study underscores the continued "culture war" between traditional clinician-centered vs algorithm-driven practice that likely will impede the pace of progress in improving patient outcomes. Understandably, the high-stakes nature of critical care leads to an unavoidable and often unrecognized emotional bias among those entrusted with safeguarding patient well-being. However, there must be a concerted daily effort among clinicians to examine whether their reaction to either protocolized or automated approaches may be unduly influenced by subjective impressions.

In addition, the higher success of automated weaning suggests that more subtle aspects of weaning may be important in determining outcomes. Transient intolerance of an SBT may occur because of abrupt increases in work, particularly when respiratory muscle weakness or anxiety is present. Perception of loaded breathing (and hence dyspnea) is determined largely by the magnitude in muscle load change relative to muscle strength.3 Thus, SBTs may be aborted too quickly in some patients who might need marginally more time to adapt to load changes. An unanticipated impact of SBTs is that respiratory care practitioners now frequently have multiple patients who require daily (or twice daily) testing that increasingly is being coordinated with daily sedation interruptions. This increases the demand placed on clinicians to complete their assigned tasks and may inadvertently tax their ability (or patience) to fully assess their patients’ responses to SBTs. Therefore, a gradual but constant titration of PS may limit bouts of distress and facilitate weaning in ways that traditional ventilator practices and staffing patterns cannot accommodate.


1. Esteban A, et al. Effect of spontaneous breathing trial duration on outcome of attempts to discontinue mechanical ventilation. Spanish Lung Failure Collaborative Group. Am J Respir Crit Care Med 1999;159:512-518.

2. Kallet RH. What is the legacy of the NIH ARDS Network? Respir Care 2009;54:912-924.

3. Mahler DA, O’Donnell DE, eds. Dyspnea: Mechanisms, Measurement, and Management. 2nd edition. Boca Raton, FL: Taylor & Francis Group; 2005.