Ventilator Weaning and Discontinuation Practices for Critically Ill Patients

Key Points

Question

In critically ill patients who receive invasive mechanical ventilation, how is invasive mechanical ventilation discontinued and do discontinuation practices differ internationally?

Findings

In this prospective observational study that included 1868 patients from 142 intensive care units in Canada, Europe, the US, India, the UK, and Australia/New Zealand from November 2013 to December 2016, 22.7% of patients underwent direct extubation, 49.8% underwent an initial spontaneous breathing trial (of which 81.8% had successful extubation), 8.0% had a direct tracheostomy, and 19.5% died before a weaning attempt. There was notable variation in several aspects of mechanical ventilation weaning practices.

Meaning

Mechanical ventilation weaning practices varied internationally, with nearly 50% of patients undergoing an initial spontaneous breathing trial.

Abstract

Importance

Although most critically ill patients receive invasive mechanical ventilation (IMV), few studies have characterized how IMV is discontinued in practice.

Objective

To describe practice variation in IMV discontinuation internationally, associations between initial discontinuation events and outcomes, and factors associated with the use of select discontinuation strategies and failed initial spontaneous breathing trials (SBTs).

Design, Setting, and Participants

Prospective, multinational, observational study of critically ill adults who received IMV for at least 24 hours from 142 intensive care units (ICUs) in 19 countries within 6 regions (27 in Canada, 23 in India, 22 in the UK, 26 in Europe, 21 in Australia/New Zealand, and 23 in the US).

Exposures

Receiving IMV.

Main Outcomes and Measures

Primary analyses characterized types of initial IMV discontinuation events (extubation, SBT, or tracheostomy) and associations with clinical outcomes (including duration of ventilation, ICU and hospital mortality, and ICU and hospital length of stay). Secondary analyses examined the associations between SBT outcome and SBT timing and clinical outcomes.

Results

Among 1868 patients (median [interquartile range] age, 61.8 [48.9-73.1] years; 1173 [62.8%] men) 424 (22.7%) underwent direct extubation, 930 (49.8%) had an initial SBT (761 [81.8%] successful), 150 (8.0%) underwent direct tracheostomy, and 364 (19.5%) died before a weaning attempt. Across regions, there was variation in the use of written directives to guide care, daily screening, SBT techniques, ventilator modes, and the roles played by clinicians involved in weaning. Compared with initial direct extubation, patients who had an initial SBT had higher ICU mortality (20 [4.7%] vs 96 [10.3%]; absolute difference, 5.6% [95% CI, 2.6%-8.6%]), longer duration of ventilation (median of 2.9 vs 4.1 days; absolute difference, 1.2 days [95% CI, 0.7-1.6]), and longer ICU stay (median of 6.7 vs 8.1 days; absolute difference, 1.4 days [95% CI, 0.8-2.4]). Patients whose initial SBT failed (vs passed) had higher ICU mortality (29 [17.2%] vs 67 [8.8%]; absolute difference, 8.4% [95% CI, 2.0%-14.7%]), longer duration of ventilation (median of 6.1 vs 3.5 days; absolute difference, 2.6 days [95% CI, 1.6-3.6]), and longer ICU stay (median of 10.6 vs 7.7 days; absolute difference, 2.8 days [95% CI, 1.1-5.2]). Compared with patients who underwent early initial SBTs, patients who underwent late initial SBTs (>2.3 days after intubation) had longer duration of ventilation (median of 2.1 vs 6.1 days; absolute difference, 4.0 days [95% CI, 3.7-4.5]), longer ICU stay (median of 5.9 vs 10.8 days; absolute difference, 4.9 days [95% CI, 4.0-6.3]), and longer hospital stay (median of 14.3 vs 22.8 days; absolute difference, 8.5 days [95% CI, 6.0-11.0]).

Conclusions and Relevance

In this observational study of invasive mechanical ventilation discontinuation in 142 ICUs in Canada, India, the UK, Europe, Australia/New Zealand, and the US from 2013 to 2016, weaning practices varied internationally.

Trial Registration

ClinicalTrials.gov Identifier: NCT03955874

This critical care epidemiology study describes practice variation in invasive mechanical ventilation weaning and discontinuation practices, associations between initial discontinuation events and outcomes, and factors associated with use of select discontinuation strategies and failed initial spontaneous breathing trials among critically ill patients in ICUs in 19 countries.

Introduction

Invasive mechanical ventilation (IMV) is a defining feature of critical illness. In epidemiologic studies conducted in the US between 2002 and 2010, the incidence of IMV ranged from 270 to 314 cases per 100 000 population,^1,2 and this incidence is expected to increase in the future.³ IMV increases the complexity and costs of health care.⁴ Cumulative exposure to IMV has been associated with potentially harmful cointerventions (eg, sedation), increased morbidity (eg, ventilator-related complications [such as pneumonia]),⁵ and long-term functional sequelae and cognitive impairment.⁶ Thus, limiting the duration of IMV has been identified as a priority area for research.⁷

IMV can be discontinued abruptly by direct extubation or gradually through 1 or more spontaneous breathing trials (SBTs) or tracheostomy collar trials. Randomized trials have evaluated the use of screening protocols (vs usual care) to identify patients who are ready to undergo an SBT,⁸ different techniques to conduct SBTs,⁹ duration of an SBT,¹⁰ and strategies to eventually extubate patients whose initial SBT failed.¹¹ Discontinuation involves the skills of diverse clinicians whose roles differ across settings and may be influenced by factors related to patients, institutions, and care processes. An international survey of intensivists’ weaning practices identified significant variation in screening frequency, ventilator modes, SBT techniques, written directives to guide weaning, and the roles of available personnel.¹² Although several large-scale observational studies of IMV use have been conducted,^13,14,15 none have focused exclusively on IMV discontinuation.

The objectives of this prospective, multinational, observational study of critically ill adults who were intubated and receiving mechanical ventilation were to describe ventilator weaning and discontinuation practices, investigate associations between discontinuation strategies and outcomes, and identify factors associated with the use of select discontinuation strategies and initial SBT failures.

Methods

Objectives

The primary objectives of this study were to characterize variation in discontinuation practices among patients who received IMV for at least 24 hours, with regard to written directives, daily screening, preferred methods of ventilator support, SBT techniques, and sedation and mobilization practices, and to describe the association between the initial strategy used to discontinue IMV (direct extubation, initial SBT, or direct tracheostomy) and clinical outcomes (duration of ventilation; ICU and hospital mortality; ICU and hospital length of stay; and the percentage of patients who were reintubated, readmitted to the ICU, receiving mechanical ventilation at day 28, and in the ICU at day 28). Secondary objectives included describing the associations between initial SBT outcome (failure vs success), SBT timing (early vs late initial SBT), SBT techniques, and humidification strategies and clinical outcomes. We also described factors associated with selection of alternative discontinuation strategies. In a tertiary objective, we described factors associated with initial SBT failure. The full protocol for this study was published previously.¹⁶ Ethics approval was sought by site investigators at all participating sites. Seventeen ICUs required prospective written or verbal consent for participation. The institutional regulatory policy at 1 site enabled collection of data that were deidentified and protected by prior consent or privacy standards. The remaining ICUs were permitted to screen patients for eligibility and collect deidentified data for patients who met study eligibility criteria without written or oral consent.

Patient Population and Participating Centers

Research personnel prospectively followed up all newly admitted patients who received IMV for at least 24 hours.¹⁶ Patients were excluded if they were transferred to participating ICUs without a documented intubation time, underwent a tracheotomy at ICU admission, were already using ventilator settings compatible with SBT (T-piece or continuous positive airway pressure [CPAP] ≤5 cm H₂O or pressure support [PS] ≤8 cm H₂O with or without positive end-expiratory pressure [PEEP] or automatic tube compensation or equivalent) at ICU admission, were at participating ICUs for 24 hours or more, were readmitted during the study period, or participated in other studies with explicit weaning protocols.

A region was defined as a country or collection of countries represented by a critical care society.¹⁶ We aimed to include 150 ICUs with similar representation (approximately 25 ICUs) from each region.¹⁶ We used a multimodal strategy (information cards enclosed in a previously administered, postal, international weaning survey¹⁶ and correspondence with regional collaborators, critical care societies, and site investigators) to identify participating ICUs and permitted flexible start dates to allow for variable approval processes and clinical coverage models.

Data Collection

Data were collected from November 2013 to December 2016. In each ICU, we collected data from at least 10 discontinuation events and all deaths before a weaning attempt. Patients were followed up until successful extubation or disconnection (tracheostomy), death, or ICU discharge or transfer or until day 60 for patients with ventilator dependence.

Outcomes

Data regarding written directives, daily screening, preferred methods of ventilator support, SBT techniques, and sedation and mobilization practices pertaining to IMV and discontinuation were collected. Types of initial IMV discontinuation events were categorized as direct extubation, SBT, or tracheostomy. We defined an SBT as a focused assessment of the patient's capacity to breathe spontaneously with any technique (eg, T-piece, CPAP, PS with minimal assistance) with the goal of discontinuing IMV. We defined successful extubation (disconnection) as the time when unsupported spontaneous breathing began and noninvasive ventilation or IMV was not used for more than 48 hours after extubation or disconnection. Clinical outcomes were duration of ventilation; ICU and hospital mortality; ICU and hospital length of stay; and the percentage of patients who were reintubated, readmitted to the ICU, receiving mechanical ventilation at day 28, and in the ICU at day 28. Duration of mechanical ventilation summarizes the time to successful extubation and is reported for both patients who died during hospitalization and for those who did not.

Research personnel identified eligible patients using a screening log; collated data to characterize patients, discontinuation events, care processes, and clinical outcomes; and entered data into Medidata RAVE (Medidata Solutions). Two investigators (K.B. and L.R.) reviewed all data forms, transmitted queries to clarify discontinuation events and illogical data, and adjudicated outcomes, when required, in collaboration with a third reviewer (M.M. or D.C.).

Statistical Analysis

Data were planned to be collected on at least 10 discontinuation events other than death in each of 150 participating ICUs. We sought to achieve 225 initial SBT failures, anticipating that at least 50% of discontinuation events would involve an initial SBT (expected range, 750-1050 SBTs) and that 70% (n = 525) would be successful.¹⁶

We used descriptive statistics to summarize binary (number and percentage) and continuous (mean and SD and median and interquartile range [IQR], when appropriate) variables. We expressed differences in clinical outcomes (SBT outcome and timing) using the absolute difference (AD) with 95% CIs with Yates continuity correction for binary outcomes and median (IQR) with bootstrapped 95% CIs for continuous outcomes. Analyses of mechanical ventilation duration and length of stay were conducted overall as well as limited to patients who died during hospitalization and those who did not.

To examine associations between discontinuation strategies and SBT outcome and timing, we accounted for clustering of ICUs within hospitals. We used the median time to initial SBT to characterize early (≤2.3 days) vs late (>2.3 days) SBTs. P values were adjusted for clustering of ICUs within hospitals by using a mixed-model framework and were corrected for multiple testing using the Holmes-Bonferroni method.¹⁷ Continuous outcomes with skewed distributions were log-transformed. Residuals were assessed to ensure the normality assumption was met.

A generalized linear mixed model with a logit link function was used to identify baseline characteristics and factors that developed between ICU admission and discontinuation events related to the use of specific discontinuation strategies (eMethods in the Supplement). Clustering was accounted for by including a random effect for hospital. To identify factors associated with initial SBT failure (vs success), we constructed 3 generalized linear mixed models with a logit link including patient variables (age, SBT duration, MV duration, Sedation-Agitation Scale [SAS] scores,¹⁸ modified Sequential Organ Failure Assessment [SOFA] respiratory score,^19,20 SBT method, and primary diagnosis; model 1); patient and site variables (screening, hospital type, hospital profit status, written document for SBT conduct and adjusting IMV support, respiratory therapist presence, and region; model 2); and patient, site, and practitioner variables (clinician type and years in ICU; model 3). We used a likelihood ratio test to compare models. Model 2 had the best balance of variables based on discrimination (C statistic) and calibration (calibration slope and calibration-in-the-large) after performing internal bootstrap validation.^21,22 We assumed that missing variables in regression analyses (eg, SOFA score at ICU admission, SOFA score before discontinuation event) were missing at random. These variables were imputed using multiple imputation by chained equations.²³ In a post hoc analysis, we examined the association between different SAS scores¹⁸ (classified as low [SAS score, 1-2; indicates more sedated], middle [SAS score, 3-5], and high [SAS score, 6-7; indicates less sedated]) on initial SBT failure rates. We considered 2-sided P values <.05 to indicate statistical significance. Analyses were performed using R software, versions 3.5.0 and 4.0.2 (R Foundation for Statistical Computing).

Results

Patient and Site Characteristics

Of 1868 participants (median [IQR] age, 61.8 [48.9-73.1] years; 1173 [62.8%] men), 852 (45.6%) had acute respiratory failure and 1093 (58.5%) were receiving IMV at the time of ICU admission (Figure; Table 1). Of the 142 participating sites in 19 countries within 6 regions (27 in Canada [n = 340 {18.2%}], 23 in India [n = 327 {17.5%}], 22 in the UK [n = 311 {16.6%}], 26 in Europe [n = 351 {18.8%}], 21 in Australia/New Zealand [n = 263 {14.1%}], and 23 in the US [n = 276 {14.8%}]), most were medical/surgical (57 [40.1%]) or multidisciplinary (33 [23.2%]) ICUs, and 114 (80%) had a university affiliation (Table 2). There was a median of 13 patients per ICU.^11,15

Figure. Patient Flow in a Study of Ventilator Weaning and Discontinuation Practices for Critically Ill Patients.

^aFive participants met 2 exclusion criteria (3 had no intubation time and had a tracheotomy at ICU admission, 1 had a tracheotomy at ICU admission and was readmitted to the ICU, and 1 had no intubation time and was already receiving a spontaneous breathing trial setting).

^bData for 1 participant who died before undergoing a discontinuation event were missing.

Table 1. Patient Characteristics in a Study of Ventilator Weaning and Discontinuation Practices in Critically Ill Patients.

Characteristic	No. (%)
	Total (n = 1867 in 142 ICUs)a, b	Death before discontinuation attempt (n = 363 in 112 ICUs)a	Direct extubation (n = 424 in 105 ICUs)	Direct tracheostomy (n = 150 in 78 ICUs)	Initial spontaneous breathing trial (n = 930 in 129 ICUs)
Sex
Men	1173 (62.8)	220 (60.6)	272 (64.2)	102 (68.0)	579 (62.3)
Women	694 (37.2)	143 (39.4)	152 (35.8)	48 (32.0)	351 (37.7)
Age at admission, median (IQR), y	61.8 (48.9-73.1)	64.8 (53.1-75.1)	58.7 (46.3-71.3)	57.5 (42.8-68.4)	62.5 (49.3-73.3)
SOFA score at ICU admission, median (IQR)c	5 (3-7)	6 (3-8)	5.0 (3-7)	4 (2-7)	5 (3-7)
Type of admission
Medical	1299 (69.6)	297 (81.8)	261 (61.6)	103 (68.7)	638 (68.6)
Emergency surgical	367 (19.7)	51 (14.0)	97 (22.9)	37 (24.7)	182 (19.6)
Elective surgical	201 (10.8)	15 (4.1)	66 (15.6)	10 (6.7)	110 (11.8)
Reason for intubationd
Decreased level of consciousness	455 (24.4)	100 (27.5)	105 (24.8)	44 (29.3)	206 (22.2)
Operative	415 (22.2)	39 (10.7)	125 (29.5)	22 (14.7)	229 (24.6)
Hypoxemia alone	320 (17.1)	64 (17.6)	61 (14.4)	31 (20.7)	164 (17.6)
Hypercarbia and hypoxemia	183 (9.8)	36 (9.9)	29 (6.8)	16 (10.7)	102 (11.0)
Cardiac arrest	167 (8.9)	88 (24.2)	25 (5.9)	9 (6.0)	45 (4.8)
Airway patency	162 (8.7)	12 (3.3)	50 (11.8)	20 (13.3)	80 (8.6)
Other	89 (4.8)	17 (4.7)e	19 (4.5)f	2 (1.3)g	51 (5.5)h
Hypercarbia alone	55 (2.9)	5 (1.4)	10 (2.4)	2 (1.3)	38 (4.1)
Secretions	21 (1.1)	2 (0.6)	0	4 (2.7)	15 (1.6)
Primary diagnosis for receiving invasive mechanical ventilationd
Acute respiratory failure	852 (45.6)	209 (57.6)	159 (37.5)	65 (43.3)	419 (45.1)
Neurologic/coma (nonsurgical)	407 (21.8)	87 (24.0)	94 (22.2)	45 (30.0)	181 (19.5)
Postoperative (elective or emergent) admission	387 (20.7)	34 (9.4)	121 (28.5)	26 (17.3)	206 (22.2)
Acute on chronic respiratory failure	112 (6.0)	18 (5.0)	21 (5.0)	8 (5.3)	65 (7.0)
Other	92 (4.9)	14 (3.9)	28 (6.6)	3 (2.0)	47 (5.1)
Neuromuscular disorder	17 (0.9)	1 (0.3)	1 (0.2)	3 (2.0)	12 (1.3)
Ventilator support at ICU admission
Invasive ventilation	1093 (58.5)	199 (54.8)	296 (69.8)	72 (48.0)	526 (56.6)
No invasive or noninvasive support	670 (35.9)	146 (40.2)	106 (25.0)	65 (43.3)	353 (38.0)
Bilevel noninvasive ventilation	81 (4.3)	13 (3.6)	16 (3.8)	8 (5.3)	44 (4.7)
Continuous positive airway pressure	20 (1.1)	5 (1.4)	6 (1.4)	5 (3.3)	4 (0.4)
Otheri	3 (0.2)	0	0	0	3 (0.3)
Home use of noninvasive ventilation	46 (2.5)	8 (2.2)	12 (2.8)	0	26 (2.8)

Abbreviations: ICU, intensive care unit; IQR, interquartile range; SOFA, Sequential Organ Failure Assessment.

^aData from 1 participant who died before a discontinuation event are missing.

^bThe median (IQR) number of patients per ICU was 13 (11-15).

^cComputed by summing organ scores (range, 0-4) for each of the 5 organ systems (respiratory, coagulation, bilirubin, cardiovascular, renal). Neurologic score was not included given concerns regarding accuracy and errors in measurement.²⁰ Patients who were receiving dialysis (peritoneal, intermittent, or continuous) were assigned the highest renal score. The total score range was 0 to 20, with higher scores indicating higher illness severity.

^dReason for intubation included common indication for securing airway (eg, decreased level of consciousness). For primary reason for receiving IMV, specific diagnoses were captured (eg, subarachnoid hemorrhage).

^eCardiac failure (n = 1), combative (n = 1), hemodynamic instability (n = 1), hypotensive/septic (n = 2), hypoxia/metabolic acidosis/exhaustion (n = 1), mixed metabolic and respiratory acidosis (n = 1), pending respiratory failure (n = 1), pulmonary embolism (n = 1), respiratory distress (n = 4), respiratory distress/septic shock (n = 1), respiratory failure (n = 1), and work of breathing (n = 2).

^fAcute kidney injury (n = 1), acute psychosis/aggression (n = 1), agitation/combative (n = 6), aspiration (n = 1), hypocarbia/tachypnea (n = 1), hypotension (n = 2), inhalational burn (n = 1), respiratory arrest (n = 3), respiratory fatigue (n = 1), pending respiratory arrest (n = 1), and ventricular tachycardia (n = 1).

^gArrhythmia/severe hypotension (n = 1) and neuromuscular weakness (n = 1).

^hAcute lung injury (n = 1), acute pulmonary edema (n = 1), agitation (n = 3), agitation/hypotensive/tachycardiac (n = 1), cardiogenic shock/tamponade (n = 5), decreased level of consciousness/hypoxemia (n = 1), dyspnea/agitation (n = 1), exhaustion (n = 1), procedure (n = 5), hemodynamic instability (n = 3), pending respiratory failure (n = 1), metabolic acidosis (n = 1), respiratory distress (n = 7), respiratory fatigue/failure (n = 3), transport (n = 1), seizure management/shunt failure (n = 2), septic shock (n = 2), work of breathing (n = 10), work of breathing /decreased oxygen saturation (n = 1), and work of breathing/ metabolic acidosis (n = 1).

ⁱHigh-flow nasal cannulae (n = 2) and bag mask ventilation (n = 1).

Table 2. Characteristics of Participating ICUs in a Study of Ventilator Weaning and Discontinuation Practices for Critically Ill Patients (N = 142).

Characteristic	No. (%)
Type of hospital
University-affiliated with physicians in training	114 (80.3)
Community hospital with physicians in training	21 (14.8)
Community hospital without physicians in training	7 (4.9)
Type of ICUs
Medical/surgical	57 (40.1)
Multidisciplinary	33 (23.2)
Medical/surgical and other	22 (15.5)
Medical	10 (7.0)
Othera	7 (4.9)
Cardiac or cardiovascular surgery	5 (3.5)
Neurosurgical	5 (3.5)
Surgical	3 (2.1)
ICU features
Medical directorb	134 (94.4)
Expert in mechanical ventilatorc	68 (47.9)
Closed-model ICUd	102 (71.8)
Open-model ICUe	40 (28.2)
Available personnel in ICUs
Intensivist	142 (100.0)
Nursef	140 (98.6)
Physician in training	129 (90.8)
Physiotherapist	129 (90.8)
Respiratory therapist	84 (59.2)
Most responsible physician (nonintensivist)	59 (41.5)
Other nonphysiciang	46 (32.4)
Other physicianh	32 (22.5)
Kinesiotherapist	16 (11.3)
ICU bed capacity, median (IQR)
Physical beds	18 (13.0-22.8)
Ventilator-capable beds	16 (12.0-22.0)
Non–ventilator-capable beds	0
ICUs with high-dependency beds, No. (%)i	42 (29.6)
High-dependency bedsj	6 (4.0-10.0)
ICU admission volume, median (IQR)
Admissions per month	90 (60.0-126.0)
Patients admitted during study period	86 (52.3-149.8)
Written directives, No./total No. (%)
Daily screening for spontaneous breathing trial readiness	64/138 (46.4)
Conduct of spontaneous breathing trials	58/135 (43.0)
Adjusting support during weaning	48/136 (35.3)
Sedation administration	83/135 (61.5)
Sedation titrated to sedation scales	115/139 (82.7)
Daily sedation interruption	96/137 (70.1)
Delirium managementk	74/132 (56.1)
Early mobilizationl	58/131 (55.7)

Abbreviations: ICU, intensive care unit; IQR, interquartile range.

^aMultidisciplinary/stepdown (n = 1), neurology/neurosurgery (n = 2), surgical/trauma (n = 1), trauma/neurosurgery (n = 1), cardiothoracic (n = 1), and neurosciences (n = 1).

^bAn individual appointed to oversee day-to-day operations in this ICU.

^cLocal expert/opinion leader (recognized locally, nationally, or internationally) in mechanical ventilation.

^dIn closed-model ICUs, after admission, patients were usually (but not always) transferred to the care of an intensivist assigned to the ICU on a full-time basis. Generally, patients were admitted to the ICU only after approval/evaluation by the intensivist. For periods typically ranging from 1 week to 1 month at a time, the intensivist’s main clinical duties consist of caring for patients in the ICU.

^eIn open-model ICUs, patients were admitted under the care of an internist, family physician, surgeon, or other primary attending of record, with intensivists providing expertise through consultation. Intensivists may play a primary role in the treatment of some patients, but only within the agreement of the attending of record.

^fOne ICU in the UK and 1 in Europe did not indicate that nurses were available in their ICUs.

^gThirty-two ICUs reported having 1 or more other physicians available in their ICUs.

^hForty-six ICUs reported having 1 or more other nonphysicians available in their ICUs.

ⁱHigh-dependency beds were defined as “level 2” or “stepdown” beds.

^jAmong ICUs that reported having high-dependency beds.

^kDelirium management includes techniques such as repeat reorientation, sleep protocols, use of eyeglasses and hearing aids, correction of dehydration, and other interventions designed to reduce delirium in the ICU.

^lEarly mobilization is an approach to improving functional recovery by which patients in the ICU are brought out of bed to a seated position and then to standing and walking as early as possible in the course of care, including while they are still receiving mechanical ventilation.

Primary Outcomes

Regional Variation in Discontinuing IMV

Use of Written Directives During Weaning

The percentage of ICUs with written directives for screening for (range, 4.5%-82.6%; P < .001) and conducting (range, 9.1%-78.3%; P < .001) SBTs varied significantly across regions. Written directives to screen for SBT readiness were present in more than half of the participating ICUs in Canada (15 [55.6%]), India (16 [69.6%]), and the US (19 [82.6%]). The percentage of ICUs with written directives for adjusting ventilator support did not vary significantly across regions (range, 9.5%-60.9%; P = .16). However, written directives for SBT conduct and for adjusting ventilator support were present in less than half of the participating ICUs in 5 regions (13 [48.1%] and 11 [40.7%] in Canada, 11 [47.8%] and 9 [39.1%] in India, 2 [9.1%] and 4 [18.2%] in the UK, 12 [46.2%] and 8 [30.8%] in Europe, and 2 [9.5%] and 2 [9.5%] in Australia/New Zealand). Although the percentages of ICUs with written directives for administering sedation (range, 40.9%-78.3%) and interrupting sedation (range, 23.8%-91.3%) were not statistically significantly different, the percentage of ICUs with directives for early mobilization (range, 14.3%-65.2%; P < .001) varied significantly across regions (eTable 1 in the Supplement).

Practices in Screening for SBTs

Patients were not screened for SBTs daily in participating ICUs in 4 regions (5 [18.5%] in Canada, 12 [54.5%] in the UK, 3 [11.5%] in Europe, and 13 [61.9%] in Australia/New Zealand). Patients were screened once daily to undergo an SBT in more than half of the ICUs in Canada (18 [66.7%]), India (17 [73.9%]), and the US (19 [82.6%]) and in less than half of the participating ICUs in Europe (11 [42.3%]), the UK (6 [27.3%]), and Australia/New Zealand (3 [14.3%]) (P < .001). Patients were less frequently screened twice daily (range, 3.7%-19.2%) or more than twice daily (range, 0%-26.9%).

Practices in Conducting SBTs

Initial SBTs most commonly used PS with PEEP (457 of 930 [49.1%]) or T-piece (236 of 930 [25.4%]) and less frequently applied CPAP (100 of 930 [10.8%]) or PS without PEEP (88 of 930 [9.5%]). SBTs were commonly conducted using PS with PEEP in Canada (85 of 177 [48.0%]), the UK (78 of 132 [59.1%]), Australia/New Zealand (57 of 76 [75.0%]), and the US (166 of 214 [77.6%]), and T-piece SBTs were commonly performed in India (99 of 181 [54.7%]) and Europe (77 of 150 [51.3%]) (P < .001).

Ventilator Support Before Initial Discontinuation Attempts

Among 424 patients who underwent direct extubation, most patients in ICUs outside of the US were receiving PS mode before extubation (70 of 92 [76.1%] in Canada, 14 of 27 [51.9%] in India, 52 of 69 [75.4%] in the UK, 78 of 103 [75.7%] in Europe, and 101 of 124 [81.5%] in Australia/New Zealand vs 1 of 9 [11.1%] in the US) (P < .001). Similarly, before an initial SBT, most patients in ICUs outside of the US were receiving PS (133 of 177 [75.1%] in Canada, 104 of 181 [57.5%] in India, 61 of 132 in the UK [46.2%], 99 of 149 in Europe [66.4%], and 38 of 76 in Australia [50.0%] vs 15 of 214 [7.0%] in the US) (P < .001). Before direct extubation, patients in ICUs in the US were most commonly receiving assist-control pressure-regulated volume control/volume control plus (3 of 9 [33.3%]), synchronized intermittent mandatory ventilation volume control with or without PS (2 of 9 [22.2%]), or assist-control volume control (2 of 9 [22.2%]). Before initial SBTs, patients in ICUs in the US were most commonly receiving assist-control pressure-regulated volume control/volume control plus (58 of 214 [27.1%]), synchronized intermittent mandatory ventilation volume control with or without PS (20 of 214 [9.3%]), or synchronized intermittent mandatory ventilation pressure-regulated volume control/volume control plus with or without PS (16 of 214 [7.5%]). Among 150 patients who underwent direct tracheostomy, patients across regions were most commonly supported with PS (59 of 150 [39.3%]) and assist-control volume control (25 of 150 [16.7%]) before tracheostomy.

Personnel Involved in Screening for and Conducting SBTs and Adjusting Ventilator Support

Clinicians who were responsible for daily screening (nurses [range, 9.5%-75.0%; P < .001] and respiratory therapists [range, 0%-95.7%; P < .001]) differed significantly across regions, and the clinician who was indicated to be the most responsible for conducting SBTs (P < .001) and adjusting ventilator support (P < .001) also differed significantly across regions (eTable 2 in the Supplement).

Sedation and Mobilization Practices

There was a significant difference in sedation levels immediately before an initial SBT (most patients had SAS scores of 3 [162 of 929 {17.4%}] or 4 [558 of 929 {60.1%}] vs direct extubation (most patients had SAS scores of 4 [266 of 424 {62.7%}] or 5 [54 of 424 {12.7%}]) (P < .001) (eTable 3 and eFigure 1 in the Supplement). Few patients were actively mobilized immediately before discontinuation events (51 of 424 [12.0%] who underwent direct extubation, 99 of 930 [10.6%] who underwent initial SBT, and 12 of 150 [8.0%] who underwent direct tracheostomy) (eTable 4 in the Supplement). There was also a significant difference in mobilization levels (none vs nonmobility vs mobility physiotherapy) immediately before an initial SBT vs direct extubation (P < .001).

Associations Between Initial Discontinuation Events and Clinical Outcomes

Of the 1868 total patients, 424 (22.7%) underwent direct extubation (including 36 [8.5%] unplanned), 930 (49.8%) underwent an initial SBT (761 [81.8%] successful), 150 (8.0%) underwent direct tracheostomy, and 364 (19.5%) died before a weaning attempt (Figure). Within ICUs, there were a median (IQR) of 3 (1-6) direct extubations,^{1,2,3,4,5,6,7,8,9,10,11,12} 8 (5-9) initial SBTs,^{1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17} 1 (1-2) direct tracheostomy,^{1,2,3,4,5,6,7} and 3 (2-4) deaths before weaning attempts.^{1,2,3,4,5,6,7,8,9,10,11} There were no statistically significant differences between men vs women in discontinuation strategies (direct extubation: 272 [64.2%] vs 152 [35.8%]; direct tracheostomy: 102 [68%] vs 48 [32%]; initial SBT: 57 [62.3%] vs 351 [37.7%]; P = .40).

Across discontinuation strategies (initial direct extubation, tracheostomy, and SBT), there were differences in ICU mortality [20 (4.7%) vs 23 (15.3%) vs 96 (10.3%)], reintubation [26 (6.3%) vs 42 (28.4%) vs 67 (7.7%)], and the percentage of patients who were receiving mechanical ventilation [11 (2.6%) vs 38 (26.8%) vs 50 (5.5%)] and in the ICU at day 28 [26 (6.2%) vs 51 (35.7%) vs 86 (9.4%)]. There were differences across discontinuation strategies between initial direct extubation, tracheostomy, and SBT in the median total duration of mechanical ventilation overall (2.9 vs 13.2 vs 4.1 days), in patients who did not die during hospitalization (2.9 vs 12.9 vs 3.7 days), and in patients who died during hospitalization (7.3 vs 16.9 vs 7.6 days); in median duration of ICU stay overall (6.7 vs 19.6 vs 8.1 days), in patients who did not die during hospitalization (6.2 vs 18.9 vs 7.8 days), and in patients who died during hospitalization (17.5 vs 22.2 vs 11.2 days); and in median duration of hospital stay overall (16.9 vs 35.4 vs 18.0 days), in patients who did not die during hospitalization (16.3 vs 35.9 vs 17.9 days), and in patients who died during hospitalization (23.2 vs 33.1 vs 18.1 days) (Table 3; eTables 5 and 6 in the Supplement)

Table 3. Initial Mechanical Ventilation Discontinuation Strategy and Clinical Outcomes Among Critically Ill Patients.

Clinical outcome	No. (%)
	Total (n = 1504 in 142 ICUs)	Direct extubation (n = 424 in 105 ICUs)	Direct tracheostomy (n = 150 in 78 ICUs)	Initial spontaneous breathing trial (n = 930 in 129 ICUs)
Total duration of mechanical ventilation, median (IQR), da	4.1 (2.1-7.9)	2.9 (1.8-5.8)	13.2 (7.3-21.8)	4.1 (2.1-7.5)
Patients who did not die during hospitalization	3.6 (2.0-6.9)	2.9 (1.7-5.3)	12.9 (7.2-20.5)	3.7 (2.0-6.8)
Patients who died during hospitalization	8.9 (5.0-15.8)	7.3 (5.6-14.3)	16.9 (10.7-23.1)	7.6 (4.3-13.1)
ICU mortality	139 (9.2)	20 (4.7)	23 (15.3)	96 (10.3)
Hospital mortality (deaths after ICU discharge)	88 (5.9)	27 (6.4)	6 (4.0)	55 (5.9)
Mechanical ventilation at day 28b	99/1476 (6.7)	11/420 (2.6)	38/142 (26.8)	50/914 (5.5)
In ICU at day 28b	163/1477 (11.0)	26/417 (6.2)	51/143 (35.7)	86/917 (9.4)
ICU length of stay, median (IQR), d	8.1 (4.8-14.9)	6.7 (3.7-10.9)	19.6 (11.1-32.7)	8.1 (5.0-14.7)kl
Patients who did not die before ICU discharge	7.7 (4.6-14.0)	6.2 (3.6-9.8)	18.9 (10.6-30.8)	7.8 (4.8-14.0)
Patients who died before ICU discharge	14.3 (9.1-22.2)	17.5 (14.2-21.6)	22.2 (13.2-41.5)	11.2 (7.9-19.7)
Hospital length of stay, median (IQR), d	18.9 (10.9-34.0)c	16.9 (11.0-31.0)	35.4 (20.7-62.9)	18.0 (10.6-32.1)c
Patients who did not die during hospitalization	18.6 (10.8-33.3)c	16.3 (10.3-29.5)	35.9 (21.1-66.0)	17.9 (10.5-31.8)c
Patients who died before hospital discharged	20.9 (12.4-39.5)	23.2 (15.2-47.1)	33.1 (19.0-51.5)	18.1 (11.1-32.2)
Readmitted to ICU before hospital discharge	79 (5.3)	24 (5.7)	7 (4.7)	48 (5.2)
Reintubated before successful extubation	136 (9.0)	39 (9.2)		97 (10.4)

Abbreviations: ICU, intensive care unit; IQR, interquartile range.

^aTotal duration of mechanical ventilation was calculated up to the time of successful extubation/disconnection in patients who underwent a tracheostomy or died before hospital discharge. This calculation includes data from 1463 total patients (1324 [90.5%] who did not die during hospitalization and 139 [9.5%] who died): 419 underwent direct extubation (399 [95.2%] who did not die and 20 [4.8%] who died), 909 underwent initial spontaneous breathing trials (813 [89.4%] who did not die and 96 [10.6%] who died), and 135 underwent direct tracheostomy (112 [83.0%] who did not die and 23 [17.0%] who died).

^bReceiving mechanical ventilation at day 28 and being in the ICU at day 28 compare responses of yes vs no, with unknown responses excluded.

^cData from 1 patient who did not die during hospitalization were missing.

^dHospital length of stay for patients who died before hospital discharge includes patients who died in the ICU.

Compared with patients who underwent direct extubation, patients who underwent SBT had higher ICU mortality (20 [4.7%] vs 96 [10.3%]; AD, 5.6% [95% CI, 2.6%-8.6%]), longer median duration of mechanical ventilation overall (2.9 vs 4.1 days; AD, 1.2 days [95% CI, 0.7-1.6]) and in those who did not die during hospitalization (2.9 vs 3.7 days; AD, 0.8 days [95% CI, 0.4-1.3]), and longer median ICU stay overall (6.7 vs 8.1 days; AD, 1.4 days [95% CI, 0.8-2.4]) and in patients who did not die during hospitalization (6.2 vs 7.8 days; AD, 1.6 days [95% CI, 0.8-2.6]) (Table 3)

Secondary Outcomes

Association Between Outcome of Initial SBT and Clinical Outcomes

Patients whose initial SBT failed (vs passed) had higher ICU mortality (29 [17.2%] vs 67 [8.8%]; AD, 8.4% [95% CI, 2.0%-14.7%]), were more likely to be still receiving mechanical ventilation at day 28 (19 [11.7%] vs 31 [4.1%]; AD, 7.5% [95% CI, 2.0%-13.0%]) and be in the ICU at day 28 (26 [15.7%] vs 60 [8.0%]; AD, 7.7% [95% CI, 1.4%-13.9%]), and had longer median duration of ventilation (6.1 vs 3.5 days; AD, 2.6 days [95% CI, 1.6-3.6]) and ICU stay (10.6 vs 7.7 days; AD, 2.8 days [95% CI, 1.1-5.2]) (Table 4). Associations of alternative SBT techniques and humidification strategies with clinical outcomes are shown in eTables 7 and 8 in the Supplement.

Table 4. Outcome Based on Success or Failure and Timing of Initial Spontaneous Breathing Trials Among Critically Ill Patients .

Outcome	No. (%)
	Initial spontaneous breathing trial success	Initial spontaneous breathing trial failure	Absolute difference (95% CI)a
Based on success or failure of initial spontaneous breathing trials	n = 761 in 127 ICUs	n = 169 in 78 ICUS
Total duration of mechanical ventilation, median (IQR), db	3.5 (1.9 to 6.8)	6.1 (3.5 to 10.6)	2.6 (1.6 to 3.6)
ICU mortality	67 (8.8)	29 (17.2)	8.4 (2.0 to 14.7)
Hospital mortality (deaths after ICU discharge)	45 (5.9)	10 (5.9)	0.0 (−3.9 to 3.9)
Mechanical ventilation at day 28c	31/751 (4.1)	19/163 (11.7)	7.5 (2.0 to 13.0)
In the ICU at day 28c	60/751 (8.0)	26/166 (15.7)	7.7 (1.4 to 13.9)
ICU length of stay, median (IQR), d	7.7 (4.8 to 13.6)	10.6 (6.4 to 18.6)	2.8 (1.1 to 5.2)
Hospital length of stay, median (IQR), d	17.9 (10.5 to 31.2)	18.9 (10.9 to 34.9)d	1.0 (−3.5 to 5.6,)
Readmitted to ICU before hospital discharge	42 (5.5)	6 (3.6)	−2.0 (−5.6 to 1.6)
Reintubated before successful extubation	82 (10.8)	15 (8.9)	−1.9 (−7.1 to 3.3)
Based on early or late timing of initial spontaneous breathing trialse	n = 467 in 110 ICUs	n = 463 in 123 ICUs
Total duration of mechanical ventilation, median (IQR), df	2.1 (1.6 to 4.0)	6.1 (4.1 to 9.8)	4.0 (3.7 to 4.5)
ICU mortality	39 (8.4)	57 (12.3)	3.9 (−0.2 to 8.1)
Hospital mortality (deaths after ICU discharge)	19 (4.1)	36 (7.8)	3.7 (0.5 to 6.9)
Mechanical ventilation at day 28c	15/461 (3.3)	35/453 (7.7)	4.4 (1.3 to 7.6)
In the ICU at day 28c	29/463 (6.3)	57/454 (12.6)	6.3 (2.3 to 10.3)
ICU length of stay, median (IQR), d	5.9 (3.7 to 10.4)	10.8 (7.0 to 18.1)	4.9 (4.0 to 6.3)
Hospital length of stay, median (IQR), d	14.3 (8.9 to 26.8)	22.8 (12.7 to 38.4)g	8.5 (6.0 to 11.0)
Readmitted to ICU before hospital discharge	19 (4.1)	29 (6.3)	2.2 (−0.9 to 5.3)
Reintubated before successful extubation	44 (9.4)	53 (11.4)	2.0 (−2.1 to 6.2)

Abbreviations: ICU, intensive care unit; IQR, interquartile range.

^aFor absolute differences, 95% CIs were based on bootstrapping of raw data.

^bTotal duration of mechanical ventilation was calculated up to the time of successful extubation/disconnection in patients who underwent tracheostomy or died before hospital discharge and includes data from 751 patients with initial spontaneous breathing trial success and 158 with initial spontaneous breathing trial failure.

^cMechanical ventilation at day 28 and being in the ICU at day 28 compare responses of yes vs no, with other responses excluded.

^dHospital length of stay for initial spontaneous breathing trial failure is missing data from 1 patient (n = 168).

^eEarly vs late initial spontaneous breathing trial timing was dichotomized using the median intubation time (≤2.3 vs >2.3 days).

^fTotal duration of mechanical ventilation was calculated up to the time of successful extubation/disconnection in patients who underwent tracheostomy or death at hospital discharge and includes data from 454 patients who underwent early initial spontaneous breathing trials and 455 who underwent late initial spontaneous breathing trials.

^gHospital length of stay for late initial spontaneous breathing trial is missing data from 1 patient (n = 462).

Compared with patients who underwent an initial SBT earlier (≤2.3 days), those who had an initial SBT after this time were more likely to still be receiving mechanical ventilation at day 28 (15 [3.3%] vs 35 [7.7%]; AD, 4.4% [95% CI, 1.3%-7.6%]) and be in the ICU at day 28 (29 [6.3%] vs 57 [12.6%]; AD, 6.3% [95% CI, 2.3%-10.3%]) and have longer median duration of mechanical ventilation (2.1 vs 6.1 days; AD, 4.0 days [95% CI, 3.7%-4.5%]), ICU stay (5.9 vs 10.8 days; AD, 4.9 days [95% CI, 4.0-6.3]), and hospital stay (14.3 vs 22.8 days; AD, 8.5 days [95% CI, 6.0-11.0]) (Table 4).

Factors Associated With Use of Specific Discontinuation Strategies

In regression analyses, patients who underwent an initial SBT (vs direct extubation) were more likely to be older (odds ratio [OR] per 10-year increase, 1.1 [95% CI, 1.0-1.3]; P = .02) and have solid organ malignancy (OR, 2.6 [95% CI, 1.2-5.2]; P = .01) and hypertension (OR, 1.6 [95% CI, 1.0-2.3]; P = .03). Compared with the US, patients were significantly less likely to undergo an initial SBT vs direct extubation in Australia/New Zealand (OR, 0.01 [95% CI, 0.00-0.05]; P < .001), Canada (OR, 0.03 [95% CI, 0.01-0.17]; P < .001), Europe (OR, 0.03 [95% CI, 0.01-0.18]; P < .001), and the UK (OR, 0.04 [95% CI, 0.01-0.23]; P < .001).

Tertiary Outcomes

Factors Associated With Initial SBT Failure

In regression modeling, 3 factors were identified to be associated with initial SBT failure, including 2 patient-level variables (higher SOFA respiratory score [P = .005] and lower SAS score [P = .02] before an initial SBT attempt) and a site-level variable (not having a protocol to adjust ventilator settings [P = .03]). Missing data in this analysis were less than 1% for most variables, except for SOFA score at ICU admission (279 [30%]) and before discontinuation events (164 [18%]). A post hoc analysis showed that patients with low (more sedated) vs midrange SAS scores were significantly more likely to have a failed initial SBT (OR, 2.4 [95% CI, 1.3-4.5]; P = .006) (eFigure 2 in the Supplement).

Discussion

This international study identified variability in IMV discontinuation practices across geographical regions with regard to the use of protocols, screening for and conducting SBTs, adjustment of ventilator support, and the responsibility of clinicians involved in weaning. The main findings were that nearly 50% of patients underwent an initial SBT, of which more than 80% were successful; an initial SBT (vs direct extubation) was associated with higher ICU mortality and longer duration of ventilation and ICU stay; failing the initial SBT (vs passing the initial SBT) was associated with higher ICU mortality, longer duration of ventilation, longer ICU stay, and greater likelihood of still receiving ventilation and being in the ICU at day 28; and undergoing a later (vs earlier) initial SBT was associated with longer duration of ventilation, longer ICU and hospital stays, and greater likelihood of still receiving ventilation and being in the ICU at day 28.

Similar to a large international weaning survey, this study found that most patients were screened once daily to identify SBT candidates, and were less often screened twice daily or more frequently.¹² There was significant regional variation in the presence of written directives to guide care during weaning and the roles played by available clinicians.¹² T-piece SBTs were more common in India and Europe, while PS with PEEP SBTs were more common in Canada, the UK, the US, and Australia/New Zealand. Compared with earlier studies reported by Esteban and colleagues in 2002,¹³ 2008,¹⁴ and 2013,¹⁵ the current study included data from fewer medical/surgical ICUs (68%-77%^13,14,15 vs 40.1%) and observed similar ICU (7-8^13,14,15 vs 8 days) and hospital (16-17^13,14,15 vs 19 days) length of stays, but lower rates of ICU mortality (28%-31%^13,14,15 vs 26.9% [including initial deaths]), reintubation (11%-14%^13,14,15 vs 9%), and tracheostomy (11%-15%^13,14,15 vs 8%). In contrast to their findings, fewer patients in the current study underwent an initial SBT (58%-62%^13,14 vs 49.8%). Among those who underwent SBTs, there was more use of PS with PEEP SBTs than T-piece SBTs. Discordant findings may reflect differences in the study populations, inclusion of geographically disparate ICUs in this study, or temporal changes in care. Similar to the Weaning according to a New Definition (WIND) classification, this study also identified higher mortality and duration of mechanical ventilation with progression from WIND group 1 (first separation attempt to weaning termination ≤24 hours: mortality, 3.6%; mechanical ventilation duration, 2.6 days) to WIND group 2 (first separation attempt to weaning termination within 2-6 days: mortality, 13.1%; duration of mechanical ventilation, 5.2 days) and to WIND group 3 (first separation attempt to weaning termination ≥7 days: mortality, 22.9%; duration of mechanical ventilation, 16.6 days) in patients who received IMV for at least 24 hours.²⁴ Notwithstanding, this study differs from the WIND study in important ways. Approximately half of the patients (1543 of 2729 [56.5%]) in the WIND study received IMV for less than 24 hours, while the current study included 1868 patients who received IMV for at least 24 hours, corresponding to WIND groups 2 and 3 (n = 1183). Although discontinuation attempts (extubation, SBT, tracheostomy) in the current study were similar to separation attempts in the WIND study, the current study collected detailed data at the time of discontinuation attempts and further categorized initial SBT attempts prospectively as successful or failed.

The importance of a daily screen failure was highlighted in a previous retrospective, single-center study (n = 300; 2 ICUs) of a trial that compared daily screening with either T-piece or CPAP SBTs to daily screening with SBTs conducted at the clinician’s discretion.²⁵ In that study, Ely et al²⁵ reported that patients who had a failed daily screen, with or without SBT, were more likely to require IMV for more than 21 days and were less likely to be successfully extubated and alive at hospital discharge. The current study focused on failure of an initial SBT, as opposed to a daily screen with or without an SBT, in a large cohort of predominantly unselected patients (n = 930) in 129 ICUs. Failure of an initial SBT was associated with a broader range of less favorable outcomes, more likely as a marker of illness severity than a cause of poor outcome in this observational study.

Several other findings warrant comment. First, regional variation existed in the use of written directives to guide weaning and in the roles played by clinicians involved in weaning. Second, data from this study can be used to inform decision-makers regarding outcomes associated with specific discontinuation practices (eg, tracheostomy). Third, these data highlight the challenges of generalizing weaning evidence from randomized trials (screening, SBT conduct, use of protocols) into diverse practice settings with variable personnel.^26,27 Fourth, an exploratory post hoc analysis showed that patients with low (more sedated) vs midrange SAS scores were more likely to have a failed initial SBT. These findings highlight an important potential interaction between sedation level and SBT outcome. Compared with prior studies, findings from the current study suggest that clinical outcomes have improved over time, but are less favorable than those reported in clinical trials.^11,28 These findings may reflect the greater potential for selection bias in randomized trials compared with observational studies.²⁹

To our knowledge, this is the first multinational, prospective, observational study focused specifically on describing IMV discontinuation in critically ill patients who received IMV for at least 24 hours. This study has several strengths. By design, similar representation from each region was achieved and allowed us to draw inferences about regional practice variation. Site-level screening logs helped to identify consecutive patients and minimize the potential for selection bias. Near-complete data with minimal missing data were achieved by using a detailed review and query generation process. These data reflect actual practices and complement the findings of a previous cross-sectional survey reflecting intensivists’ stated practices in IMV weaning from ICUs in the same 6 geographic regions.¹²

Limitations

This study had several limitations. First, data were collected on 169 SBT failures, representing 75% of the target sample size.¹⁶ Second, these data are descriptive and analyses are not adjusted for patient characteristics. As such, they highlight associations between discontinuation events and outcomes and cannot infer causality. For example, severely ill patients may have been more likely to undergo an initial SBT (vs direct extubation) or a late initial SBT. Additional data were not available to describe why patients did not undergo an earlier SBT (eg, failure to be screened, did not pass screening criteria earlier, too ill to undergo an earlier SBT). Third, although the eligibility criteria enabled inclusion of a diverse cohort of critically ill patients in whom IMV was discontinued using several different strategies, the use of a different time threshold for inclusion (12 or 48 hours of IMV) and limiting data collection to 10 events at each site may have affected the patient cohort included, the number and type of discontinuation events, and outcomes reported. Fourth, the study findings may not be generalizable to all ICUs and settings because 80% of included ICUs were university-affiliated ICUs with physicians in training. Fifth, these data were collected between 2013 and 2016; whether they reflect current practice is unclear, but the authors are unaware of any data that has suggested that variation in IMV discontinuation has changed. Few large trials¹⁰ or observational studies²⁴ focused on IMV discontinuation have been published since completion of the current study, and the 2017 international weaning guideline was published after data collection for this study was completed.³⁰

Conclusions

In this observational study of invasive mechanical ventilation discontinuation in 142 ICUs in Canada, India, the UK, Europe, Australia/New Zealand, and the US from 2013 to 2016, weaning practices varied internationally.

Supplement.

eMethods

eTable 1. Use of Written Directives to Guide Weaning and Care During Weaning

eTable 2. Practices and Personnel Involved in Discontinuing Mechanical Ventilation in Participating ICUs

eTable 3. Sedation Practices Before Initiation Discontinuation Events Across Region

eTable 4. Mobilization Practices Before Initiation Discontinuation Events Across Regions

eTable 5. Association Between SBT Outcome and Clinical Outcomes Across Regions

eTable 6. Association Between Direct Extubation and Direct Tracheostomy and Clinical Outcomes Across Regions

eTable 7. SBT Techniques and Clinical Outcomes

eTable 8. Humidification Strategy and Clinical Outcomes

eFigure 1. SAS Score Before Direct Extubation and initial SBT

eFigure 2. Post-hoc analysis of the relationship between SAS scores and initial SBT failures

Section Editor: Christopher Seymour, MD, Associate Editor, JAMA (christopher.seymour@jamanetwork.org).