SEMIN PERINATOL 48 (2024) 151890

Contents lists available at ScienceDirect

Seminars in Perinatology
journal homepage: www.elsevier.com/locate/semperi

Weaning from mechanical ventilation and assessment of

extubation readiness
Guilherme Sant’Anna a, *, Wissam Shalish b
a
Professor of Pediatrics, Division of Neonatology, Montreal Children’s Hospital Departments of Pediatrics and Experimental Medicine, Senior Scientist of the Research
Institute of the McGill University Health Center, McGill University Health Center, 1001 Boulevard Decarie, Room B05.2711, Montreal, Quebec H4A3J1, Canada
b
Assistant Professor of Pediatrics, Division of Neonatology, Montreal Children’s Hospital Departments of Pediatrics and Experimental Medicine, Junior Scientist of FRQS,
McGill University Health Center, Montreal, Quebec, Canada

A R T I C L E I N F O A B S T R A C T

Keywords: Tremendous advancements in neonatal respiratory care have contributed to the improved survival of extremely
None preterm infants (gestational age ≤ 28 weeks). While mechanical ventilation is often considered one of the most
important breakthroughs in neonatology, it is also associated with numerous short and long-term complications.
For those reasons, clinical research has focused on strategies to avoid or reduce exposure to mechanical venti­
lation. Nonetheless, in the extreme preterm population, 70–100% of infants born 22–28 weeks of gestation are
exposed to mechanical ventilation, with nearly 50% being ventilated for ≥ 3 weeks. As contemporary practices
have shifted towards selectively reserving mechanical ventilation for those patients, mechanical ventilation
weaning and extubation remain a priority yet offer a heightened challenge for clinicians. In this review, we will
summarize the evidence for different strategies to expedite weaning and assess extubation readiness in preterm
infants, with a particular focus on extremely preterm infants.

Introduction with significant clinical instability in the form of multiple episodes of

hypoxemia and hypercapnia.10 Evidently, both premature extubation
In 1958, respiratory distress was successfully treated for the first time and unnecessarily prolonged MV are detrimental. Unfortunately, there is
with mechanical ventilation (MV) on four term neonates in Sweden.1 A a notable paucity of strong data to guide clinicians on the ideal ways to
few years later, MV significantly decreased neonatal mortality from 80% wean MV or determine extubation readiness11 especially in the
to around 50% in infants born with gestational age (GA) between 32 and micro-preemies (22–24 weeks). In this article, the available literature is
36 weeks.2,3 Since then, respiratory care has progressed tremendously reviewed and sensible evidence-based recommendations for efficient
with the widespread and successful use of new ventilators, medications, MV weaning and extubation are outlined.
interventions, and different modes of non-invasive respiratory support.4
Importantly, survival of the most extreme preterm population also Part 1 – expediting weaning from mechanical ventilation
increased substantially, with MV playing a paramount role.5 Indeed,
data from large cohort studies and randomized trials have shown that for Ventilatory support and weaning
infants born with GA between 22 and 25 weeks, 85–89% require MV
during the first days of age and IN almost all THEM, during hospitali­ MV weaning is usually achieved by the gradual reduction of support
zation in the neonatal intensive care unit (NICU).6–8 until it reaches the controversial “minimal ventilatory settings”. Sub­
Although MV offers essential support while the respiratory system stantial debate continues regarding how to accomplish this goal, but
matures or recovers from acute failure, it is associated with many short some general evidence-based guidelines can be articulated.
and long-term complications, including neurodevelopmental impair­
ment.9 For that reason, the focus has been on expediting weaning from Pressure-controlled ventilation
MV and extubation as fast as possible. However, extubation success rates There is a sound evidence-base and rationale for using synchronized
are lowest in the most immature infants and failure has been associated modes that support every spontaneous breath in preterm infants such as

* Corresponding author.
E-mail address: guilherme.santanna@mcgill.ca (G. Sant’Anna).

https://doi.org/10.1016/j.semperi.2024.151890

Available online 23 March 2024

0146-0005/© 2024 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
G. Sant’Anna and W. Shalish Seminars in Perinatology 48 (2024) 151890

Table 1
Ventilatory settings to consider extubation in infants born with BW < 1250g and ≤ 2 weeks of age.
Conventional Ventilation
SIMV/PS: PIP 12-15 cm H2O, PEEP ≤ 6 cm H2O, Rate = 20 bpm, FiO2 ≤ 0.30
AC/PSV: MAP 6-8 cm H2O and FiO2 ≤ 0.30

Volume-targeted ventilation
Tidal volume 4.0 to 5.0 mL/kg (5 to 6 mL/kg if <700 g or >2 weeks of age) and FiO2 ≤ 0.30
High-Frequency Oscillatory Ventilation
MAP 6-8 cmH2O and FiO2 ≤ 0.30

High-Frequency Jet Ventilation

PIP ≤ 14 cm H2O, MAP 6-8 cm H2O and FiO2 ≤ 0.30
These are the minimum values that should trigger an extubation attempt. Many larger, more vigorous infants can be extubated from higher settings. Older infants are often able to be
extubated from higher pressures and tidal volumes. Abbreviations: AC, assist control; bpm, breaths per minute; BW, Birth weight; FiO2, fraction of inspired oxygen; MAP, mean
airway pressure; PEEP, positive end-expiratory pressure; PIP, peak inflation pressure; PSV, pressure support ventilation; SIMV, synchronized intermittent mechanical ventilation.

assisted control (AC). Short-term studies have described: (a) smaller and physiological dead space (due to heterogeneous lung aeration) require a
less variable tidal volume (VT), (b) less tachypnea and fluctuations in larger VT 17,18 Under most circumstances, VT should not be reduced
blood pressure, and (c) quicker weaning from MV with AC when below 4 to 5 mL/ kg during the weaning process to avoid excessive work
compared with SIMV. 12,13 During SIMV, spontaneous breaths above the of breathing (see Table 1). The objective is to have adequate respiratory
set rate are not supported, causing uneven VT and possibly higher work drive with acceptable blood gases.
of breathing. This situation is particularly true during weaning when the
number of unsupported breaths increases. This is most important in High-frequency ventilation (HFOV)
extremely low birth weight infants with narrow endotracheal tubes HFOV support is weaned largely based on empiric, rather than
(ETT), because resistance to flow is inversely proportional to the fourth experimental data. In general, both pressure amplitude and mean
power of the radius.13 The high airway resistance, weak muscle, and the airway pressure (Paw) are progressively decreased based on improve­
excessively compliant chest wall often lead to small and ineffective VT. ments in ventilation and oxygenation, respectively. Frequency is not
Although the number of mechanical inflations with AC is larger, these reduced as a means of reducing support. With standard HFOV the
inflations should be less injurious, given that the VT is about 33% delivered VT increases as frequency decreases, so that reducing venti­
smaller.12 Another misunderstanding is that supporting every breath lator frequency has the opposite effect from that on conventional
results in lack of respiratory muscle training. However, in awake, ventilation (this may no longer be true when modern HFOV devices
spontaneously breathing infants, VT is the result of the combined capable of HFOV+VG are used thereby maintaining stable VThf).
inspiratory effort of the patient (negative intrapleural pressure on Weaning from HFJV more closely parallels conventional ventilation,
inspiration) and the positive pressure generated by the ventilator. This with stepwise reduction in peak and mean pressures. Frequency may be
combined effort results in the transpulmonary pressure, which, along reduced slightly but the primary means of reducing support is reduction
with compliance of the respiratory system, determines the VT. During in pressure amplitude, largely by reducing PIP. With both modalities of
weaning, as ventilator peak inflation pressure (PIP) is decreased, the HFV, as pressure amplitude continues to be reduced, the infant should
infant gradually takes over the work of breathing and trains their res­ take over progressively more of the respiratory effort, with visible
piratory muscles. When the ventilator pressure falls to a low level no spontaneous VT, and when the settings are judged to be sufficiently low,
higher than what is needed to overcome resistance of the ETT and cir­ extubation should be considered. Although many clinicians are more
cuit, the infant should be considered ‘ready’ for extubation. The com­ comfortable changing from high frequency to conventional modes,
bination of SIMV + pressure support (PS) was shown to result in more extubation directly from both jet and oscillatory ventilation is possible
rapid weaning than SIMV alone14 and to result in significantly lower and desirable, as this practice has been shown to be associated with
work of breathing.15 However, this mode results in added complexity shorter duration of MV and lower incidence of BPD when compared to
and there is a paucity of information regarding how to optimally use this switching the infants to conventional MV prior to extubation.19,20
combination. Ventilator settings at which extubation may be considered
in infants 2 weeks of age or younger are provided in Table 1. Prone position during ventilation
A Cochrane review showed that prone position slightly improved
Volume-targeted ventilation oxygenation in neonates undergoing mechanical ventilation with a low
By using tidal volume as the primary control variable, the inflation to moderate grade of evidence. However, there was no evidence to
pressure is automatically reduced in response to improved lung suggest that particular body positions during MV are effective in pro­
compliance or increased respiratory effort of the infant. This occurs ducing sustained and clinically relevant improvements.21
continuously and in real time, accelerating the weaning process. Indeed,
volume guarantee and volume-controlled ventilation significantly
reduce MV duration when compared with pressure-limited ventila­ Adjunctive therapies and weaning
tion.16 The optimal VT value to accomplish weaning is not well estab­
lished. The VT must be low enough to allow the infant to have a Permissive hypercapnia
respiratory drive (i.e., not be alkalotic), but not so low that it results in Accommodating higher levels of PaCO2 levels during ventilation to
excessive work of breathing. With appropriate VT settings, reduction in facilitate weaning and earlier extubation has been investigated exten­
support occurs automatically with decreases in the pressure needed to sively.22 Safe limits of hypercapnia both during the first few days when
achieve the targeted VT. Hence, it is counterproductive to reduce VT the risk of intraventricular hemorrhage is higher and later during the
below normal physiologic value as it will increase the work of breathing chronic phase of BPD have not been definitively established.23,24
and potentially lead to fatigue, setting up the infant for failure of extu­ Nevertheless, it became common practice in neonatal intensive care to
bation. The appropriate VT for optimal weaning must be individualized; accept higher levels of PaCO2 if pH ≥ 7.20 or 7.25 on the premise that
infants that remain ventilator dependent for extended periods because of trying to achieve perfectly “normal” blood gases would require
increased anatomic dead space (“acquired tracheomegaly”) or increased ventilatory parameters, and hence more ventilator-induced
lung injury. Yet, further trials are needed to evaluate optimal PaCO2

2
G. Sant’Anna and W. Shalish Seminars in Perinatology 48 (2024) 151890

targets for high-risk infants. rates of PNC administration in infants with GA <28 weeks.35 Because
available data are still conflicting clinicians must use their own clinical
Permissive hypoxemia judgment for each patient to decide when to start, which steroids to use
and how much to give, balancing the adverse effects of prolonged MV
The Neonatal Oxygenation Prospective Meta-analysis (NeOProM) and BPD against the potential adverse effects of treatments.
Collaboration has conducted a prospective meta-analysis, using indi­
vidual patient data of several trials. The primary outcome was a com­ Fluids and nutritional support
posite of death or major disability at 18–24 months postmenstrual age.25 Some attention has been paid to the importance of fluids and nutri­
The available evidence pointed toward avoidance of very low (83–87%) tional support in facilitating extubation. In a secondary analysis of a
and very high SpO2 ranges (95%–98%) by keeping SpO2 between 90, 91 randomized clinical trial, peak fluid balance (FB) was associated with
and 95%. the need for MV on postnatal day 14 and severe BPD or death.36 For
every 10% increase in peak FB there was 103% increased odds of
Caffeine receiving MV at postnatal day 14 (adjusted OR, 2.03; 95% CI,
The use of caffeine in preterm infants has been investigated in several 1.64–2.51). In another study, higher cumulative FB during the first 10
small studies and one large randomized controlled trial (RCT), the days of life was associated with higher odds of death/BPD in infants born
caffeine for apnea of prematurity (CAP) trial.26 In the latter, infants born <29 weeks.37 Both studies proposed that monitoring of postnatal FB
with weights between 500 and 1250 g and randomized to caffeine may be an appropriate non-invasive strategy to favor MV weaning and
received the drug within 10 days of age at the discretion of the survival without BPD. Nutritional support should be pursued while the
responsible physician. Fifty three percent of the patients in the caffeine infant is intubated, as a means of optimizing strength and endurance
group received the medication at a median age of 3 days while stable on after extubation. However, maintaining infants on MV to facilitate
low MV settings: 33% to facilitate the removal of the ETT and 20% to weight gain is an outdated practice that should be abandoned. If an
prevent postextubation apnea. A secondary analysis showed that infants infant is on low ventilatory support, they should be extubated as pro­
assigned to caffeine were ventilated for a shorter period, received less longed MV increases the chances of lung injury via several mechanisms,
continuous positive airway pressure (CPAP) and oxygen supplementa­ regardless of the weight. In a severely ill and unstable infant on high
tion, and had less BPD.26 Another secondary analysis observed that ventilatory settings after the first week of age, waiting for weight gain
earlier administration was associated with faster weaning.27 Thus, the instead of resorting to PNC therapy may end up exposing these infants to
administration of caffeine seems to be safe and effective for ventilator high O2 concentrations (direct oxygen toxicity to the lungs), multiple
weaning and is currently considered standard of care for extremely episodes of hypoxemia and acidosis (brain injury), and further inflam­
preterm infants during the peri-extubation period. However, extrapo­ mation, all of which can be detrimental to optimal growth. Moreover, as
lations to include clinical conditions, indications, or dosages not studied body growth is a high O2-consuming metabolic process the presence of
in the CAP trial should be avoided. For instance, caffeine initiation multiple hypoxemia episodes impairs optimal growth.38 Ultimately,
immediately after birth or within a few hours of age, with the objective careful strategic consideration for type of milk intake and fortification
of improving any short-term or long-term outcome, has not been judi­ strategies offers an opportunity to optimize the balance between fluids
ciously studied and is therefore not justified based on the best available and macronutrients. Indeed, preterm infants with evolving lung disease
evidence. often experience disproportionate weight gain without corresponding
linear growth, and increased BMI z-scores up to 36 weeks of CGA, which
Postnatal corticosteroids were associated with higher odds of BPD despite similar caloric intake,
’The use of systemic postnatal corticosteroids (PNCs) for MV wean­ suggesting that growth and nutritional targets should be an individu­
ing and extubation, either dexamethasone or hydrocortisone, has been alised moving target as one strategy may not fit all.39
the subject of several reviews.28,29 A careful assessment should be made,
restricting PNCs use to situations in which the risk of adverse outcome Diuretics
related to prolonged ventilator dependence is high. Two meta-regression The effects of diuretics on duration of ventilatory support and oxygen
analyses have indicated that the risk/benefit ratio favors the use of PNCs administration, length of hospital stay, potential complications, sur­
in the more immature and severely ill infants who remain on MV after 1 vival, and long-term outcomes have not been investigated in appropri­
to 2 weeks of age.30,31 The hydrocortisone for BPD trial used a total dose ately designed RCTs.40 The apparent benefits of diuretics need to be
of 24.5 mg/kg tapered over 10 days (equivalent to ~0.98 mg/kg balanced against the known adverse effects of these drugs, especially in
dexamethasone) in infants < 30 weeks of GA intubated for at least 7 days view of equivocal evidence of benefit and known adverse event profile
by 14–28 days of age.32 The probability of extubation by the end of the when using chronic diuretics.
treatment was significantly higher in the hydrocortisone group
compared to placebo (44.7% vs 33.6%), with a 3-day reduction in the Protocols for MV
median duration of MV before 36 weeks’ postmenstrual age. Rates of An observational study from Canada showed that implementation of
survival without moderate-to-severe BPD at 36 weeks’ CGA were similar a weaning protocol for the management of infants with birth weight ≤
between groups (16.6% vs 13.2%; adjusted RR, 1.27; 95% CI, 1250g by using objective criteria for weaning, extubation, and reintu­
0.93–1.74). The Dexamethasone a Randomised Trial (DART) study bation, resulted in earlier extubation and an overall reduction of extu­
administered a cumulative dexamethasone dose of 0.89 mg/kg also bation failure and duration of MV.41 Well-designed clinical trials in the
tapered over 10 days and demonstrated a significant improvement in preterm population are lacking and would benefit the practice of
rates of successful extubation (60.0% vs 11.7%; RR, 5.1; 95% CI, neonatology.
2.0–13.3; NNT, 2; 95% CI: 1–3) which was achieved at a median age of 7
days of therapy.33 A recent systematic review and network Micropreemies
meta-analysis including 59 RCTs indicated that moderately early initi­ With the remarkable improvement of neonatal intensive care, many
ation of systemic dexamethasone between 8 and 14 days of postnatal infants born between 22 to 24-weeks of GA are now surviving.5 These
age, with a medium cumulative dose of 2 to 4 mg/kg, appeared to be the infants comprise a very special group with the highest risk of MV
most successful intervention in reducing the risk of death or BPD.34 It is weaning and extubation failure due to cardiorespiratory (CR) instability.
important to understand that the risk profile of infants currently Recently, three centers have published successful experiences with the
receiving MV may be even higher given the increased successful use of management of micropreemies by using comprehensive bundles,
non-invasive respiratory support therapies, explaining the increasing specialized teams, and different aspects of care, including respiratory

3
G. Sant’Anna and W. Shalish Seminars in Perinatology 48 (2024) 151890

Fig. 1. Age at extubation, extubation outcomes, and death or bronchopulmonary dysplasia in extremely preterm infants. The top panel presents the extubation
failure rate (defined as reintubation within 7 days of extubation) for infants extubated at different postnatal ages. The bottom panel presents, for each extubation
group, the observed death/BPD rate, and the expected probability of death/BPD as computed using a web-based BPD outcome estimator which was developed and
validated by the Eunice Kennedy Shriver National Institute of Child Health and Development (NICHD), while assuming that all infants are of “White” race/ethnicity.

support.42 The objectives of respiratory care often included elective prevent catastrophic outcomes amongst patients that require reintuba­
intubation immediately after birth with specific guidelines focused on tion.45 In contrast, the field of neonatology has only recently started to
providing gentle ventilation. In examining the existing data from three understand better the cost of reintubation. Based on the recent literature
NICUs the approach to weaning and extubation varied widely, with the in extremely preterm infants, there is a significant association between
timing of successful extubation ranging from 29 to 36 weeks’ of CGA. It reintubation and increased risks of respiratory morbidities and death or
is important to note that these data come from the retrospective expe­ BPD.11 This is most likely attributable to the fact that a reintubation adds
rience of these centers where respiratory management is just one part of an average of 10–12 days on MV. Moreover, there may be some addi­
the intervention bundle. Therefore, it is difficult to derive any recom­ tional independent risk amongst infants who require 3 or more reintu­
mendations concerning optimal weaning and extubation readiness in bations and infants reintubated within the immediate 24–48 h period
this vulnerable population at the moment. following extubation. In the latter case, a few studies have described rare
but severe adverse events that occurred in infants otherwise deemed
Part 2 – assessment of extubation readiness ready for extubation who then died unexpectedly within hours of
extubation from pulmonary hemorrhage, high grade intraventricular
Dilemmas in the decision to extubate hemorrhage, or fulminant necrotizing enterocolitis.11 Furthermore,
reintubation may lead to hemodynamic instability and/or airway injury
In mechanically ventilated preterm infants, the cost of prolonged MV due to the difficulty of the technique itself, and may lead to frequent and
is clear: each additional week of MV increases the risk of ventilator- prolonged apneas and intermittent hypoxias, which have been inde­
induced lung injury, ventilator-associated pneumonia, airway injury, pendently associated with increased risks of BPD, retinopathy of pre­
bronchopulmonary dysplasia (BPD), and death or long-term neuro­ maturity, and neurodevelopmental impairment.46–48 Putting it all
developmental impairment.9,43,44 For those reasons, clinicians generally together, it is understandable that clinicians caring for extremely pre­
agree that extubation should be attempted as early as possible. However, term infants weigh the costs of MV against the costs of reintubation in
amongst infants with birth weight < 1250g, overall reintubation rates different ways, and thus will have different inclinations as to when and
are high, reaching 50% for all causes and nearly 30% for how to extubate.
respiratory-related reintubations that could not be explained by new
diagnoses such as infection or necrotizing enterocolitis.10 These rein­
tubation rates are higher than reported in children (5–10%) and adults When to extubate
(10–20%), thereby reflecting the complexity and high-risk nature of this
population. In the adult critical care literature, mortality rates are higher Clinical perspectives on the optimal timing of extubation in
amongst reintubated patients (25–50%), and every effort is made to extremely preterm infants have changed considerably over the years. In
the 1990s, it was standard practice to keep infants intubated for

4
G. Sant’Anna and W. Shalish Seminars in Perinatology 48 (2024) 151890

Table 2
Available tools for the assessment of extubation readiness and their respective pros and cons
Clinical predictors
Pros: Simple clinical variables that are readily available at the bedside.
Cons: Only modest accuracies for the prediction of extubation readiness.
Future directions: requires larger and more balanced datasets, and integration with richer time-series data from electronic medical records, bedside monitors, and/or ventilators.

Physiological predictors
Spontaneous breathing trials
Pros: Relatively simple bedside assessment that can be implemented by clinicians, nurses, and/or respiratory therapists serially.
Cons: Low balanced accuracy due to limited ability to identify extubation failures, exposes infants to some clinical instability, and prone to variable test conduct and subjective
interpretations.
Future directions: targeted evaluation in the highest-risk patients, use of pressure support during the trial, and integration of more objective measurements during the trial.
Lung function tests
Pros: Measurements of minute ventilation, lung mechanics, and respiratory muscle strength are all in theory important as they may play a role in the outcome of extubation.
Cons: require invasive techniques or extensive data extraction from the ventilator, difficult to determine the optimal level of ventilatory support during the test that best matches
post-extubation conditions.
Future directions: further explore the clinical utility of less invasive techniques, such as lung ultrasound, to assess lung aeration and lung edema as surrogate measures of lung
function.
Cardiorespiratory behavior
Pros: Assessments of heart rate variability and breathing patterns have demonstrated improved balanced accuracy for the prediction of extubation readiness, especially when
combined with other tests (e.g., clinical predictors or spontaneous breathing trials).
Cons: still requires instrumentation and/or significant information technology resources to acquire, store, and analyze the data.
Future directions: design, develop, and test less invasive/sophisticated methods for enhanced cardiorespiratory monitoring (ex: wearable biotechnology), and embrace the use of
more complex machine learning methods into future predictors through collaborations between clinicians, biomedical engineers, and computer scientists.

prolonged periods until they were larger, more mature, and clinically weeks, 26–28 weeks, and 27 weeks, respectively.53–55 In contrast, the
stable on very low ventilatory settings. In the 2000s, with the increased timing of first successful extubation amongst infants born at 22 and 23
adoption of surfactant therapy and non-invasive respiratory support, weeks’ gestation has only been reported by a few centers with expertise
combined with the recognized harms of prolonged MV, the pendulum in caring for this population, ranging from a PMA of 29 to 36 weeks.42
swung in the opposite direction. Indeed, results from international sur­ While these data provide a practical snapshot of current reality across
veys suggested that most clinicians preferred to attempt extubation NICUs, it does not mean that some infants couldn’t be successfully
within the first 24 to 72 hours of life.49 This belief was further extubated sooner. Ideally, predictors capable of promptly identifying
strengthened by studies showing that early extubation was associated infants’ readiness for extubation based on individual risk profile are
with improved respiratory outcomes, even if reintubation was needed to reduce unnecessary MV exposure while maximizing chances
required.50,51 However, those studies were retrospective in nature, of success.
included very few patients less than 25 weeks gestation, and were
conducted at a time where the vast majority were intubated liberally, as How to extubate
prophylactic surfactant was the norm. In contemporary clinical practice,
MV is increasingly used only in the most complex patients born at the Determining readiness for extubation has not proven to be a
limits of viability (between 22- and 25-weeks’ gestation), and it is not as straightforward matter. In practice, extubation decisions typically rely
clear whether early extubation is feasible or whether it imparts favor­ on clinical judgment, based on assessment of ventilatory requirements,
able outcomes even when reintubation is required. The most recent blood gases, and overall clinical stability of the patient. Moreover,
study to tackle this question came from a secondary analysis of the APEX protocols are rarely used in NICUs to streamline the extubation process.
study, a multicenter prospective observational study conducted between As a result, extubation decisions are prone to subjective interpretations,
2013 and 2018 aimed to develop an automated predictor of extubation lead to variable peri-extubation practices and outcomes, and are often
readiness in preterm infants with birth weight less than 1250g under­ associated with inaccurate decisions. To circumvent those concerns, it
going their first planned extubation.52 As part of APEX, a large re­ makes sense to look for more objective assessment tools that can predict
pository of cardiorespiratory (CR) signals and clinical database were the likelihood of extubation success or failure with more accuracy and
generated; the latter was used in secondary studies to better understand consistency. An outline of available assessment tools and their pros and
various issues surrounding extubation, including the timing of extuba­ cons is provided in Table 2.
tion. Out of 250 infants enrolled, 52% were extubated early (within 7 Clinical predictors – While variables such as low gestational age and
days of age) and 37% were extubated late (between postnatal ages weight (at birth and at extubation), and high mean airway pressure
8–35).53 Extubation failure rates, defined as reintubation ≤ 7 days from (MAP) and oxygen requirement at extubation, are significantly associ­
extubation, varied as a function of postnatal age at extubation (Fig. 1, ated with increased risk of extubation failure, results are not consistent
top panel). Infants were further categorized into 4 extubation groups across studies. Indeed, no single cut-off for age, weight, or pre-
(early success, early failure, late success, and late failure) and multi­ extubation MAP and O2 has been shown to accurately separate infants
variate regression models were used to evaluate how the timing and with successful or failed extubations.11 Moreover, large cohort studies
outcome of extubation differentially affected rates of death/BPD. As have attempted to develop predictors of extubation readiness using
expected, infants with early and successful extubation had the lowest either logistic regression or other machine-learning methodologies,
rates of death/BPD. Surprisingly, even though infants with early failed which have resulted in only modest accuracies compared to clinical
extubation had lower expected probabilities of death/BPD compared to judgment alone. One such example, a model which uses 5 clinical var­
infants with late successful extubation as they were larger and more iables to estimate the probability of extubation success in extremely
mature, they had an observed death/BPD rate of 83% compared to 66% preterm infants, demonstrated an area under the receiver operating
in the late success group (Fig. 1, bottom panel). Therefore, early extu­ characteristics curve of 0.77 in the original retrospective single-center
bation was only associated with lower death/BPD rates when successful. cohort study and 0.72 when retrospectively validated in an external
Based on contemporary epidemiological cohort studies, infants born at cohort.56,57 Although the estimator has been made publicly available
24 weeks’, 25 weeks’, and 26, 27weeks’ gestation only achieve their first online, caution is advised in using such a tool until it is prospectively
successful extubation at a median postmenstrual age (PMA) of 27–29 validated, and its clinical usefulness is delineated. Unit-specific practices

5
G. Sant’Anna and W. Shalish Seminars in Perinatology 48 (2024) 151890

Fig. 2. Summary of the existing evidence and recommendations for weaning and extubation of extremely preterm infants

or other unmeasured confounders may substantially alter the applica­ The Automated Prediction of Extubation Readiness (APEX) Study –
bility of such tools to other NICUs. APEX was an international, multicenter prospective observational study
Extubation readiness tests – clinical judgment can be supplemented by that embraced the complexity of extubation through an interdisciplinary
the use of a bedside extubation readiness test, which consists of a short collaboration between clinicians, biomedical engineers, and computer
period of spontaneous breathing on endotracheal continuous positive scientists.52 APEX was a pragmatic study, whereby once the clinician
airway pressure (ETT-CPAP) during which various physiological and/or deemed the infant ready for extubation, CR signals were acquired during
clinical parameters are monitored, to determine whether the patient can a 60-minute period on mechanical ventilation followed by an additional
be extubated or not. Based on a systematic review and meta-analysis, 5-minutes on ETT-CPAP to capture the baby’s spontaneous breathing
over 30 extubation readiness tests have been evaluated in preterm in­ unaltered by mechanical inflations. The infant was extubated right after
fants, including various measurements of pulmonary function, respira­ completion of the data collection and the cardiorespiratory signals were
tory muscle strength, cardio-respiratory behavior, and spontaneous processed to characterize the infants’ cardiorespiratory behavior. Next,
breathing trials (SBT).57 However, significant heterogeneity exists in the machine-learning methods were applied to train and test a classifier
conduct of those tests, as they use variable PEEP level, duration, use of using the best combination of clinical and CR features. Infants were
pressure support and different pass/fail definitions. Moreover, the vast classified into extubation success or failure (reintubation within 72 h
majority of extubation readiness tests have shown an overall low from extubation). The classifier which combined both clinical and CR
balanced accuracy, with excellent sensitivities, meaning that they could data performed better than each one alone, correctly identifying 70% of
validate a successful extubation in those infants already deemed ready successful extubations and 75% of failed extubations. For extubations
by the clinician, but poor specificities, meaning that they add little value that occurred in the first week of life, the APEX classifier correctly
in the identification of extubation failures. identified 70% of successful extubations and 89% of failures. Thus, the
Spontaneous breathing trials – When looking specifically at SBTs, at use of machine-learning methods combining both clinical and CR fea­
least 6 diagnostic accuracy studies totalling more than 600 preterm in­ tures showed a potential to improve the prediction of failed extubations
fants have been published, with median cohort GA ranging between 26 in extremely preterm infants, especially in the first week of life. While
and 33 weeks.58–63 The vast majority of SBTs employed ETT-CPAP levels promising, these methods require significant instrumentation to acquire
of 5 and 6 cm H2O, except for one study that assisted each breath with the data, further fine-tuning to improve the detection of successful
additional pressure support of 10 cm H2O.63 Overall, SBTs showed low extubation, and prospective validation. Fortunately, the field of artificial
balanced accuracies, with high sensitivities (between 81 and 98%) but intelligence continues to grow at a very fast pace and recent advances in
low specificities (anywhere from 0 to 83%). In the largest study, a sec­ technology have seen the advent of new wireless wearable sensors that
ondary analysis from the APEX database, over 40,000 definitions of SBT may facilitate and enhance continuous CR monitoring.65,66 Further­
pass/fail were evaluated, but the SBT with highest balanced accuracy more, the addition of newer non-invasive tools, such as point of care
only predicted extubation outcomes with a sensitivity of 93% and ultrasound for the assessment of lung aeration and edema, may become
specificity of 39%.61 Moreover, the only prospective study to evaluate a useful adjunct in the prediction of extubation readiness but requires
the practice of conducting daily SBTs showed no impact on extubation further evaluation.67 Finally, with the capacity in some NICUs to store
outcomes or MV duration.11 There are a few problems that may explain all patient data derived from the electronic medical records, bedside
why SBTs as performed today do not have the optimal balanced accu­ monitors, and ventilators, integration of such data into future predictors
racy. First, a 3–5 min SBT may be too short to evaluate respiratory drive, may lead clinicians to continuously assess and recognize an infant’s
especially considering that most infants are reintubated due to apnea. potential for extubation, therefore possibly leading to earlier weaning
Second, SBTs using a PEEP of 5 and 6 cm H2O have been shown to and successful extubation.
significantly increase respiratory load, which may not match their
post-extubation conditions.64 Although longer SBTs with added pressure Conclusions
support may help with these shortcomings, their use cannot be recom­
mended until they are evaluated in the highest-risk infants. And lastly, Rapid advances in neonatal respiratory care have led to improved
relying on clinical parameters may be too simplistic and still leave room respiratory outcomes in the vast majority of preterm and term neonates,
for subjective interpretation. Therefore, in this multifaceted extubation through improved ventilatory support strategies and adjuvant therapies
equation, it is likely that more objective, complex, and individualized helping to expedite weaning from the ventilator, along with optimized
assessments of CR behavior could lead to better results. peri-extubation practices and post-extubation provision of non-invasive

6
G. Sant’Anna and W. Shalish Seminars in Perinatology 48 (2024) 151890

respiratory support. A summary of the existing evidence and recom­ 23. Ambalavanan N, Carlo WA, Wrage LA, et al. PaCO2 in surfactant, positive pressure,
and oxygenation randomised trial (SUPPORT). Arch Dis Child Fetal Neonatal Ed.
mendations for weaning and extubation are provided in Fig. 2. None­
2015;100(2):F145–F149. Mar.
theless, as we continue to care for an increasingly smaller and more 24. Travers CP, Carlo WA, Nakhmani A, et al. Late permissive hypercapnia and
immature preterm population, the challenges of MV weaning and respiratory stability among very preterm infants: a pilot randomised trial. Arch Dis
extubation inherent in this unique population with the highest risk of Child Fetal Neonatal Ed. 2023;108(5):530–534. Sep.
25. Askie LM, Darlow BA, Finer N, et al. Neonatal oxygenation prospective meta-
morbidities and mortality. Future research needs to focus primarily on analysis (NeOProM) collaboration. association between oxygen saturation targeting
these patients, to identify strategies that can expedite weaning while and death or disability in extremely preterm infants in the neonatal oxygenation
concomitantly reducing the risks of failed extubation. To achieve this prospective meta-analysis collaboration. JAMA. 2018;319(21):2190–2201. Jun 5.
26. Schmidt B, Roberts RS, Davis P, et al. Caffeine therapy for apnea of prematurity.
goal, future studies should embrace complexity, interdisciplinarity, and N Engl J Med. 2006;354(20):2112–2121.
multimodal monitoring and analytycal methods. 27. Davis PG, Schmidt B, Roberts RS, et al. Caffeine for apnea of prematurity trial:
benefits may vary in subgroups. J Pediatr. 2010;156(3):382–387.
28. Doyle LW, Cheong JL, Ehrenkranz RA, Halliday HL. Early (< 8 days) systemic
Disclosure postnatal corticosteroids for prevention of bronchopulmonary dysplasia in preterm
infants. Cochrane Database Syst Rev. 2021;10, CD001146. Oct 21.
Authors do not have any financial and personal relationships to 29. Doyle LW, Cheong JL, Ehrenkranz RA, Halliday HL. Late (>7 days) systemic
postnatal corticosteroids for prevention of bronchopulmonary dysplasia in preterm
disclose. infants. Cochrane Database Syst Rev. 2021;11, CD001145. Nov 11.
30. Doyle LW, Halliday HL, Ehrenkranz RA, Davis PG, Sinclair JC. Impact of postnatal
References systemic corticosteroids on mortality and cerebral palsy in preterm infants: effect
modification by risk for chronic lung disease. Pediatrics. 2005;115(3):655–661. Mar.
31. Doyle LW, Halliday HL, Ehrenkranz RA, Davis PG, Sinclair JC. An update on the
1. Benson F, Celander O, Haglund G, Nilsson L, Paulsen L, Renck L. Positive-pressure
impact of postnatal systemic corticosteroids on mortality and cerebral palsy in
respirator treatment of severe pulmonary insufficiency in the newborn infant; a
preterm infants: effect modification by risk of bronchopulmonary dysplasia.
clinical report. Acta Anaesthesiol Scand. 1958;2(1):37–43.
J Pediatr. 2014;165(6):1258–1260. Dec.
2. Delivoria-Papadopoulos M, Levison H, Swyer PR. Intermittent positive pressure
32. Watterberg KL, Walsh MC, Li L, et al. Hydrocortisone to improve survival without
respiration as a treatment in severe respiratory distress syndrome. Arch Dis Child.
bronchopulmonary dysplasia. N Engl J Med. 2022;386(12):1121–1131. Mar 24.
1965;40(213):474–479. Oct.
33. Doyle LW, Davis PG, Morley CJ, McPhee A, Carlin JB. Low-dose dexamethasone
3. Northway WH, Rosan RC, Porter DY. Pulmonary disease following respirator
facilitates extubation among chronically ventilator-dependent infants: a multicenter,
therapy of hyaline-membrane disease. Bronchopulmonary dysplasia. N Engl J Med.
international, randomized, controlled trial. Pediatrics. 2006;117(1):75–83. Jan.
1967;276(7):357–368. Feb 16.
34. Hay S, Ovelman C, Zupancic JA, et al. Systemic corticosteroids for the prevention of
4. Keszler M. Novel ventilation strategies to reduce adverse pulmonary outcomes. Clin
bronchopulmonary dysplasia, a network meta-analysis. Cochrane Database Syst Rev.
Perinatol. 2022;49(1):219–242. Mar.
2023;8(8), CD013730. Aug 31.
5. Bell EF, Hintz SR, Hansen NI, et al. Mortality, in-hospital morbidity, care practices,
35. Parikh S, Reichman B, Kusuda S, et al. International network for evaluation of
and 2-year outcomes for extremely preterm infants in the US, 2013-2018. JAMA.
outcomes (iNeo) of neonates investigators. Trends, characteristic, and outcomes of
2022;327(3):248–263. Jan 18.
preterm infants who received postnatal corticosteroid: a cohort study from 7 high-
6. Hatch LD, Clark RH, Carlo WA, Stark AR, Ely EW, Patrick SW. Changes in use of
income countries. Neonatology. 2023;120(4):517–526.
respiratory support for preterm infants in the US, 2008-2018. JAMA Pediatr. 2021;
36. Starr MC, Griffin R, Gist KM, et al. Association of fluid balance with short- and long-
175(10):1017–1024. Oct 1.
term respiratory outcomes in extremely premature neonates: a secondary analysis of
7. SUPPORT Study Group of the Eunice Kennedy Shriver NICHD Neonatal Research
a randomized clinical trial. JAMA Netw Open. 2022;5(12), e2248826. Dec 1.
Network Finer NN, Carlo WA, Walsh MC, et al. Early CPAP versus surfactant in
37. Soullane S, Patel S, Claveau M, Wazneh L, Sant’Anna G, Beltempo M. Fluid status in
extremely preterm infants. N Engl J Med. 2010;362(21):1970–1979. May 27.
the first 10 days of life and death/bronchopulmonary dysplasia among preterm
8. Carlo WA. Gentle ventilation: the new evidence from the SUPPORT, COIN, VON,
infants. Pediatr Res. 2021;90(2):353–358. Aug.
CURPAP, Colombian Network, and Neocosur Network trials. Early Hum Dev. 2012;
38. Mortola JP. Implications of hypoxic hypometabolism during mammalian
88(2):S81–S83. MaySuppl.
ontogenesis. Respir Physiol Neurobiol. 2004;141(3):345–356. Aug 12.
9. Walsh MC, Morris BH, Wrage LA, et al. Extremely low birthweight neonates with
39. Li Ching Ng L, Patel S, Plourde H, et al. The association between BMI trajectories
protracted ventilation: mortality and 18-month neurodevelopmental outcomes.
and bronchopulmonary dysplasia among very preterm infants. Pediatr Res. 2023;93
J Pediatr. 2005;146(6):798–804. Jun.
(6):1609–1615. May.
10. Shalish W, Kanbar L, Keszler M, et al. Patterns of reintubation in extremely preterm
40. Stewart A, Brion LP, Ambrosio-Perez I. Diuretics acting on the distal renal tubule for
infants: a longitudinal cohort study. Pediatr Res. 2018;83(5):969–975. May.
preterm infants with (or developing) chronic lung disease. Cochrane Database Syst
11. Shalish W, Keszler M, Davis PG, Sant’Anna GM. Decision to extubate extremely
Rev. 2011;(9), CD001817.
preterm infants: art, science or gamble? Arch Dis Child Fetal Neonatal Ed. 2022;107
41. Hermeto F, Bottino MN, Vaillancourt K, Sant’Anna GM. Implementation of a
(1):105–112. Jan.
respiratory therapist-driven protocol for neonatal ventilation: impact on the
12. Hummler H, Gerhardt T, Gonzalez A, et al. Influence of different methods of
premature population. Pediatrics. 2009;123(5):e907–e916. May.
synchronized mechanical ventilation on ventilation, gas exchange, patient effort,
42. Sindelar R, Nakanishi H, Stanford AH, Colaizy TT, Klein JM. Respiratory
and blood pressure fluctuations in premature neonates. Pediatr Pulmonol. 1996;22
management for extremely premature infants born at 22 to 23 weeks of gestation in
(5):305–313.
proactive centers in Sweden, Japan, and USA. Semin Perinatol. 2022;46(1), 151540.
13. Kapasi M, Fujino Y, Kirmse M, et al. Effort and work of breathing in neonates during
43. Miller JD. Pulmonary complications of mechanical ventilation in neonates. Clin
assisted patient-triggered ventilation. Pediatr Crit Care Med. 2001;2(1):9–16.
Perinatol. 2008;35(1):273–281.
14. Reyes ZC, Claure N, Tauscher MK, et al. Randomized, controlled trial comparing
44. Jensen EA, DeMauro SB, Kornhauser M, Aghai ZH, Greenspan JS, Dysart KC. Effects
synchronized intermittent mandatory ventilation and synchronized intermittent
of multiple ventilation courses and duration of mechanical ventilation on respiratory
mandatory ventilation plus pressure support in preterm infants. Pediatrics. 2006;118
outcomes in extremely low-birth-weight infants. JAMA Pediatr. 2015;169(11):
(4):1409–1417.
1011–1017.
15. Patel DS, Sharma A, Prendergast M, et al. Work of breathing and different levels of
45. Thille AW, Richard JC, Brochard L. The decision to extubate in the intensive care
volume-targeted ventilation. Pediatrics. 2009;123(4):e679–e684.
unit. Am J Respir Crit Care Med. 2013;187(12):1294–1302.
16. Wheeler K, Klingenberg C, McCallion N, et al. Volume-targeted versus pressure-
46. Jensen EA, Whyte RK, Schmidt B, et al. Association between intermittent hypoxemia
limited ventilation in the neonate. Cochrane Database Syst Rev. 2010;(11),
and severe bronchopulmonary dysplasia in preterm infants. Am J Respir Crit Care
CD003666.
Med. 2021;204(10):1192–1199.
17. Keszler M, Nassabeh-Montazami S, Abubakar K. Evolution of tidal volume
47. Di Fiore JM, Bloom JN, Orge F, et al. A higher incidence of intermittent hypoxemic
requirement during the first 3 weeks of life in infants < 800 g ventilated with
episodes is associated with severe retinopathy of prematurity. J Pediatr. 2010;157
volume guarantee. Arch Dis Child Fetal Neonatal Ed. 2009;94(4):F279–F282.
(1):69–73.
18. Bhutani VK, Ritchie WG, Shaffer TH. Acquired tracheomegaly in very preterm
48. Poets CF, Roberts RS, Schmidt B, et al. Association between intermittent hypoxemia
neonates. Am J Dis Child Fetal Neonatal Ed. 1986;140(5):449–452.
or bradycardia and late death or disability in extremely preterm infants. JAMA.
19. Clark RH, Gerstmann DR, Null DM, et al. Prospective randomized comparison of
2015;314(6):595–603.
high-frequency oscillatory and conventional ventilation in respiratory distress
49. Al-Mandari H, Shalish W, Dempsey E, Keszler M, Davis PG, Sant’Anna G.
syndrome. Pediatrics. 1992;89(1):5–12.
International survey on periextubation practices in extremely preterm infants. Arch
20. Courtney SE, Durand DJ, Asselin JM, et al. High-frequency oscillatory ventilation
Dis Child Fetal Neonatal Ed. 2015;100(5):F428–F431.
versus conventional mechanical ventilation for very-low-birth-weight infants. N Engl
50. Berger J, Mehta P, Bucholz E, Dziura J, Bhandari V. Impact of early extubation and
J Med. 2002;347(9):643–652.
reintubation on the incidence of bronchopulmonary dysplasia in neonates. Am J
21. Rivas-Fernandez M, Roqué I Figuls M, Diez-Izquierdo A, Escribano J, Balaguer A.
Perinatol. 2014;31(12):1063–1072.
Infant position in neonates receiving mechanical ventilation. Cochrane Database Syst
51. Mukerji A, Razak A, Aggarwal A, et al. Early versus delayed extubation in extremely
Rev. 2016;11(11), CD003668. Nov 7.
preterm neonates: a retrospective cohort study. J Perinatol. 2020;40(1):118–123.
22. Thome UH, Genzel-Boroviczeny O, Bohnhorst B, et al. PHELBI Study group.
Permissive hypercapnia in extremely low birthweight infants (PHELBI): a
randomised controlled multicentre trial. Lancet Respir Med. 2015;3(7):534–543. Jul.

7
G. Sant’Anna and W. Shalish Seminars in Perinatology 48 (2024) 151890

52. Kanbar LJ, Shalish W, Onu CC, et al. Automated prediction of extubation success in 60. Janjindamai W, Pasee S, Thatrimontrichai A. The optimal predictors of readiness for
extremely preterm infants: the APEX multicenter study. Pediatr Res. 2023;93(4): extubation in low birth weight infants. J Med Assoc Thai. 2017;100(4):427–434.
1041–1049. 61. Shalish W, Kanbar L, Kovacs L, et al. Assessment of extubation readiness using
53. Shalish W, Keszler M, Kovacs L, et al. Age at first extubation attempt and death or spontaneous breathing trials in extremely preterm neonates. JAMA Pediatr. 2020;
respiratory morbidities in extremely preterm infants. J Pediatr. 2023;252:124–130. 174(2):178–185.
e3. 62. Khan A, Kumar V, Hussain AS, et al. Accuracy of spontaneous breathing trial using
54. Ohnstad MO, Stensvold HJ, Tvedt CR, Rønnestad AE. Norwegian neonatal network. ET-CPAP in predicting successful extubation of neonates. Cureus. 2021;13(9):
Duration of mechanical ventilation and extubation success among extremely e17711.
premature infants. Neonatology. 2021;118(1):90–97. 63. Li Z, Xue J, Guo XY, et al. Accuracy of the spontaneous breathing trial using a
55. Weisz DE, Yoon E, Dunn M, et al. Duration of and trends in respiratory support combined CPAP + PSV model to predict extubation outcomes in very preterm
among extremely preterm infants. Arch Dis Child Fetal Neonatal Ed. 2021;106(3): infants. BMC Pediatr. 2022;22(1):627.
286–291. 64. Latremouille S, Bhuller M, Rao S, Shalish W, Sant’Anna G. Diaphragmatic activity
56. Gupta D, Greenberg RG, Sharma A, et al. A predictive model for extubation and neural breathing variability during a 5-min endotracheal continuous positive
readiness in extremely preterm infants. J Perinatol. 2019;39(12):1663–1669. airway pressure trial in extremely preterm infants. Pediatr Res. 2021;89(7):
57. Shalish W, Latremouille S, Papenburg J, Sant’Anna GM. Predictors of extubation 1810–1817.
readiness in preterm infants: a systematic review and meta-analysis. Arch Dis Child 65. Chung HU, Rwei AY, Hourlier-Fargette A, et al. Skin-interfaced biosensors for
Fetal Neonatal Ed. 2019;104(1):F89–F97. advanced wireless physiological monitoring in neonatal and pediatric intensive-care
58. Kamlin CO, Davis PG, Morley CJ. Predicting successful extubation of very low units. Nat Med. 2020;26(3):418–429.
birthweight infants. Arch Dis Child Fetal Neonatal Ed. 2006;91(3):F180–F183. 66. Yoo JY, Oh S, Shalish W, et al. Wireless broadband acousto-mechanical sensing
59. Chawla S, Natarajan G, Gelmini M, Kazzi SN. Role of spontaneous breathing trial in system for continuous physiological monitoring. Nat Med. 2023;29(12):3137–3148.
predicting successful extubation in premature infants. Pediatr Pulmonol. 2013;48(5): 67. Mohsen N, Solis-Garcia G, Jasani B, Nasef N, Mohamed A. Accuracy of lung
443–448. ultrasound in predicting extubation failure in neonates: A systematic review and
meta-analysis. Pediatr Pulmonol. 2023. https://doi.org/10.1002/ppul.26598.