Discussion Forum


JOURNAL CLUB: In-depth analysis of the WEAN SAFE study - The epidemiology of weaning.

By: Davi Mota Alcantara, Médico - 06/16/2023 20:03

In this topic, we will deepen our analysis of the WEAN SAFE study, a large, observational, multicenter, prospective cohort study conducted in 481 Intensive Care Units (ICU) in 50 countries (1).

👉 To see the first post about this article on the Xlung Forum, click here.

👉 To access the article, click here.


Methodology and population:

From October/2017 to June/2018, 10232 patients were evaluated for eligibility and 4363 were excluded, the vast majority for not reaching 48 hours of Mechanical Ventilation (MV). 5869 participants took part in the study.

The studied population consisted of adult patients (older than 16 years) admitted to the ICU with at least more than 48 hours on MV. From then on, the weaning process, in particular, was closely monitored in order to assess the outcomes:

Primary: identify the proportion of patients on MV for at least 2 days who were successfully weaned on day 90.

Secondary: proportion of patients undergoing a ventilator separation attempt; rate and timing of tracheostomy; time intervals between achieving weaning eligibility criteria and the first attempt at ventilator separation; risk factors for delayed initiation of weaning and weaning failure; ICU and hospital mortality; description of the population according to the WIND classification.


🧐 Peculiarities of the study:

One of the major issues of the study is related to the definitions of eligibility for weaning, as defined by the authors. Let's recall them here:

  • Eligibility criteria for weaning (modified from Boles et al)(2):
  • Inspired oxygen fraction (FiO2) less than 50%;
  • Positive pressure at the end of expiration (PEEP) less than 10;
  • Use of no or low dose of vasopressors;
  • Not be under paralyzing agent.

The authors decided not to add the level of consciousness because this would be another extubation criterion and also because sedation management is a potential modifiable factor related to the duration of invasive MV. This was an interesting and valid methodological decision, in our opinion.

⚠️However, the most important limitation is the following: the criteria described above differ from those of the main sources adopted as a reference to classify the patient as potentially suitable for the spontaneous breathing test (SBT), including the same adopted as a reference in the study (2 ). Usually the FiO2 used in practice to assess the eligibility of the patient's spontaneous breathing test is 40% or less and the PEEP is 8 or less or 5 or less.

This means that the definitions used in the study added an important bias, tending to artificially increase the number of patients with delayed weaning.

The authors were asked about the topic (3) and responded, but did not explain why they chose this definition (4), instead of adhering to the most universally used one.

An important question is whether the definition was chosen before or after obtaining the study data (ie, was it the result of a post-hoc analysis?), but this is not clear in the text of the article. Furthermore, such a definition (or any other regarding eligibility for weaning) is not present in the study registration (5) or in its original protocol.

An interesting way to address this question would be to look again at the data on delay in weaning, but using the usual definition if they have such data.

As the doubt remains, we recommend caution in relation to the results related to delay in weaning, since the numbers found tend to be more distant from established conventional clinical practice or even from the truth that we seek to obtain and may not reflect the scenario of your ICU.

Exactly for this reason, we consider little the results related to delay in weaning, as the biases involved may distance us from reality.


✅ Analysis of the results:

62% of patients were men; 66% had at least 1 relevant comorbidity.

77% underwent at least 1 ventilator separation attempt and 65% were successfully weaned. That is, most patients (84%) who underwent at least 1 separation attempt were successful. Even so, of these, 11% died (either in the ICU or after discharge from the ICU).

Patient mortality was high. 32% died in the ICU and, of these, 64% died without any weaning attempts, 32% underwent at least 1 attempt, but were unsuccessful, and 5% were successful in weaning. In all, 38.3% of the patients died in the hospital and, of these, 17% had been discharged from the ICU.

Of course, this average may not reflect what you might expect in the epidemiology of your ICU, especially as this is a heterogeneous study, the average of which includes patients from elective surgeries, trauma, victims of cardiac arrest, etc. Therefore, we suggest that our subscribers evaluate the article itself to verify the numbers corresponding to the type of patient in their day-to-day practice.

Regarding eligibility for weaning, almost 92% of patients met the eligibility criteria defined by the authors and the median time for this was 3 days. It is important to remember that such criteria have a high probability of not reflecting your clinical practice, both due to the high FiO2 and PEEP values and the absence of other important parameters. That is: it is possible that the proportion of patients who actually met eligibility criteria for an ERT is significantly lower.

Of patients who underwent at least 1 separation attempt, the median time from eligibility criteria to the first attempt was 1 day (IQR 0 - 4), with 22% having a delay of 5 days or more. The first attempt occurred at a median of 5 days (IQR 4 - 8) after intubation. 16% of these patients had failed weaning, with 12.2% dying during the process and the remainder being transferred.

65% of the patients had a short weaning, 10% had an intermediate weaning and 10% had a long weaning. 20% of patients underwent tracheostomy at some point in the study.

The first attempt at separation was an ERT in 66% of patients, direct extubation in 20.5%, and ERT at tracheostomy in 13.5%. Among the weaned patients, 31% did not formally undergo an ERT.

Factors independently associated with delay in the first separation attempt were: a) among demographic factors, frailty, admission due to trauma and admission due to non-traumatic neurological events; cardiac arrest was associated with a lower risk of delay; b) severity of critical illness, assessed using the SOFA score; c) among the potentially modifiable factors, previous use of neuromuscular blockade and the presence of moderate or deep sedation levels on the first day in which the eligibility criteria defined by the authors were met. Sedation levels remained important risk factors even after sensitivity analyses.

Factors independently associated with weaning failure: a) among demographic factors, advanced age, immunocompromised state and frailty; b) among the disease severity factors, general severity, measured using the non-neurological SOFA, cardiac arrest and non-traumatic neurological event as the cause of admission, pre-existing limitations of care and level of respiratory dysfunction (RR, P ratio /F) and ventilatory assistance (Driving pressure, PEEP) at the time of the first separation attempt; c) among the potentially modifiable factors, the presence of deep sedation at the time of the first attempt and the time interval between fulfilling the eligibility criteria for weaning and the first attempt at separation were observed. As a factor associated with successful weaning, heart failure was observed as the cause of admission. Results remained similar after sensitivity analysis, especially sedation at the time of readiness for weaning.

The following figures show the results of the multivariate analysis in Odds ratio and confidence intervals:

Figure 1: Variables associated with delay in weaning


Figure 2: Variables associated with weaning failure


✍ Conclusion:

We consider this an important cohort study, because: a) It was able to assess well the epidemiology of the weaning process in 50 countries; b) It offers us an epidemiological and prognostic mapping for our own practice regarding patients on MV for more than 48 hours in the most varied conditions; c) It offers us hypotheses to be tested in future clinical trials about what can interfere, positively or negatively, in the weaning process, especially in relation to potentially modifiable factors.

👉 We recommend our subscribers to independently assess the results related to specific causes of intubation and conditions similar to those of their patients in everyday life, so that the epidemiological data really have practical value for you.

As this is an observational study, we should not delve into causality conjectures, so it is important to note that the associations found may simply reflect confounding effects. That is, the same elements that lead to the need for deep sedation can also lead to delay (in aspects not measured in the study) in making the first attempt to separate from the ventilator. Likewise, the same unmeasured factors that lead to delay (as defined by the authors) in the initiation of weaning could also be causing prolonged weaning or weaning failure.

✍ As key points of this more in-depth analysis, we highlight the following:

  • Elements associated with possible delay in weaning: frailty, trauma admission, non-traumatic neurological event, disease severity. Regardless of the delay, longer intubation is to be expected in these patients.

  • Elements associated with weaning failure: age, immunocompromise, frailty, cardiac arrest, non-traumatic neurological event, disease severity; cardiac arrest and non-traumatic neurological event as cause of admission; pre-existing care limitations and level of respiratory dysfunction (RR, P/F ratio) and ventilatory assistance (Driving pressure, PEEP) at the time of the first separation attempt. That is: in patients with this profile, a greater probability of prolonged weaning should be expected.

  • Higher sedation at the time of the first separation attempt was associated with possible delay in weaning and weaning failure. We agree with the authors regarding the need for special attention to sedation levels, in order not to unnecessarily prolong ventilatory support in a patient.

🤓 Xlunger, what did you think of this review of ours? If you have any comments to make or are interested in more in-depth analysis of the studies, please comment on this post. It's a pleasure to hear them.

Note: The article was kindly provided by Dr John Laffey, one of the main authors, at the request of our team by email. Here is our thanks.



  1. Pham T, Heunks L, Bellani G, et al. Weaning from mechanical ventilation in intensive care units across 50 countries (WEAN SAFE): a multicenter, prospective, observational cohort study [published correction appears in Lancet Repir Med. 2023 Mar;11(3):e25]. Lancet Respire Med. 2023;11(5):465-476. doi:10.1016/S2213-2600(22)00449-0
  2. Boles JM, Bion J, Connors A, et al. Weaning from mechanical ventilation. Eur Respire J. 2007;29(5):1033-1056. doi:10.1183/09031936.00010206
  3. Cuyas CS, Fernandez RF. WEAN SAFE and the definition of the first separation attempt. Lancet Respire Med. 2023;11(5):e43. doi:10.1016/S2213-2600(23)00051-6
  4. Pham T, Heunks L, Bellani G, Brochard L, Laffey J; WEAN SAFE Investigators. WEAN SAFE and the definition of the first separation attempt - Authors' reply. Lancet Respire Med. 2023;11(5):e44. doi:10.1016/S2213-2600(23)00089-9
  5. https://clinicaltrials.gov/ct2/show/record/NCT03255109

To add an answer on this topic and read the replies...

You must have a valid and active xlung subscription

If you are already a subscriber, please Login at the top of the page, or subscribe now

Subscribe now