AARC Clinical Practice Guideline: Spontaneous Breathing Trials for Liberation From Adult Mechanical Ventilation

Background.

Yang and Tobin¹⁴ introduced the RSBI, which calculates the ratio of breathing frequency (f) to tidal volume (V_T) during a short (< 5 min) period without mechanical ventilatory support. The authors reported that RSBI with a cutoff of < 105 more accurately predicted liberation success than integrative compliance, f, oxygenation, and pressure index and its individual components (f and V_T). Meade et al¹⁵ identified 65 studies that included ventilator liberation in which pooled likelihood ratios predicted lower probability of successful extubation when f > 38 breaths/min and an RSBI > 100 breaths/min/L.¹⁵ Subsequently, RSBI gained favor. However, an international survey of critical care physicians in 2018 reported that 51% did not use the RSBI.¹⁶

Summary of the evidence.

In a 2006 RCT, Tanios et al¹⁷ compared 2 ventilator liberation protocols—one of which included the RSBI—to determine the effect of an RSBI on time to liberation from mechanical ventilation. Subjects (n = 304) were randomized into 2 groups. All subjects were managed with an SBT protocol, but in one group, an RSBI < 105 breaths/min/L was required before an SBT was conducted. The authors found that including RSBI to determine readiness for an SBT prolonged time on the ventilator by, on average, an additional day.

A 2022 systematic review and meta-analysis evaluated the usefulness of RSBI in predicting successful extubation.¹⁸ The observational studies included a heterogeneous ICU population of subjects with COPD, postoperative subjects, and those with primary neurological issues. All pooled measurements displayed significant heterogeneity. After reviewing 48 studies including 10,946 subjects, they reported that an RSBI of < 105 has moderate sensitivity (0.83 [95% CI 0.78–0.87], moderate certainty) and poor specificity (0.58 [95% CI 0.49–0.66]) for predicting extubation success. An RSBI < 80 had a sensitivity of 0.84 (95% CI 0.75–0.90, low certainty) and a specificity of 0.62 (95% CI 0.53–0.70, low certainty). A subgroup analyses also evaluated measurement technique (T-piece vs CPAP vs pressure support; no significant effects) and timing of RSBI measurement relative to time of SBT (no differences found).¹⁸ The authors concluded that, as a stand-alone test, the RSBI has moderate sensitivity and poor specificity for predicting extubation success.

Recommendation.

We suggest that an RSBI is not needed to determine readiness for an SBT (conditional recommendation; moderate certainty).

Justification and implementation.

Based on systematic reviews of Trivedi et al¹⁸ and Meade et al,¹⁵ there is not strong support for the use of RSBI to predict readiness for an SBT. Moreover, the results of Tanios et al¹⁷ suggest that the use of RSBI might unnecessarily delay an SBT. A simple screen is likely sufficient to identify readiness for an SBT: evidence for some reversal of the underlying cause for respiratory failure, adequate oxygenation, hemodynamic stability, and the capability to initiate an inspiratory effort.⁷ That said, criteria to determine whether a patient is tolerating an ongoing SBT should include assessment (especially changes) of the respiratory pattern, hemodynamics, gas exchange, and patient comfort.

It is important again to make the distinction between liberation and extubation.¹⁹ Tobin¹⁹ argues that the RSBI is intended to predict liberation rather than extubation. Indeed, successful completion of an SBT does not necessarily lead to extubation.⁶ The meta-analysis of Trivedi et al¹⁸ focused heavily on RSBI as a predictor of successful extubation rather than a predictor of successful liberation. A successful SBT is just one factor identifying extubation readiness; additional factors predicting extubation success include airway protection and ability to clear secretions. Trivedi et al¹⁸ suggests that RSBI might be useful as a screening tool for an SBT specifically in patients with intermediate pretest probability. However, this will need to be confirmed in clinical studies before a recommendation can be made.

Future research opportunities.

Future studies should focus on the potential benefit of RSBI in focused subject populations, such as subjects with intermediate pretest probability, as suggested by Trivedi et al.¹⁸ There is interest in the potential for noninvasive imaging techniques, such as diaphragmatic ultrasound^20,21 and electrical impedance tomography,²² as predictors of ventilator liberation. The benefit of these, either alone or in combination with RSBI, is yet to be determined.

Background.

In 1997, Esteban et al²³ conducted an RCT comparing SBTs conducted with T-piece versus SBTs with 7 cm H₂O PSV. The percentage of subjects failing the SBT was significantly higher when a T-piece was used (22% vs 14%, P = .03). However, the percentage of subjects who remained extubated after 48 h was not different between the 2 groups (63% T-piece, 70% PSV, P = .14). Although the authors concluded that SBTs with either PSV or T-piece are suitable, this study led to widespread acceptance of low-level PSV during SBTs. In a prospective, multinational, observational study, Burns et al²⁴ found that initial SBTs most often used PSV with PEEP (49.1%) or T-piece (25.4%) and less frequently CPAP (10.8%) or PSV without PEEP (9.5%). SBTs with PSV and PEEP were commonly used in North America, whereas T-piece was more commonly used in Europe. The previously published ACCP/ATS CPG made a conditional (weak) recommendation for use of PSV for the initial SBT.¹⁰

Summary of the evidence.

The evidence related to this PICO consists of 5 systematic reviews (Table 4)^25-29 and 4 RCTs.^23,30-32 Two of the systematic reviews used study-level meta-analysis, and two used network meta-analysis methodology. In aggregate, the systematic reviews support a small benefit for a fixed-level PSV or a variable level of PSV (tube compensation) in terms of SBT success and extubation success.

In a recent multi-center RCT, Subirà et al³⁰ randomized subjects to undergo a 2-h T-piece SBT (n = 578) or a 30-min SBT with 8 cm H₂O PSV (n = 557). Successful extubation occurred in 473 subjects (82%) in the PSV group and 428 subjects (74%) in the T-piece group (P = .001). Complicating the interpretation of this RCT was different durations of the SBTs. Thus, it is unclear whether the improved outcomes were attributable to mode (PSV or T-piece), duration (30 min or 120 min), or both.

Thille et al³¹ conducted a multi-center RCT (n = 969) to determine whether SBT with PSV, using 8 cm H₂O and zero PEEP, results in a shorter time to successful extubation than SBT with T-piece. At day 28, the median number of ventilator-free days was 27 in both groups (P = .31). Though 5.5% more (95% CI 0.01–10.90) of the subjects in the SBT with PSV group were extubated within 7 d of randomization, this was a secondary end point, so the authors concluded that SBTs performed with PSV did not result in significantly more ventilator-free days at day 28 than SBTs performed with a T-piece.

Indirect evidence from a different population may inform this PICO question. Jubran et al³³ conducted an RCT (n = 316) in tracheotomized subjects transferred to a single long-term acute care hospital for liberation from prolonged ventilation. Unassisted breathing through a tracheostomy, compared to weaning with a gradual reduction in PSV, resulted in shorter liberation time.

Recommendation.

We suggest that SBTs can be conducted with or without low-level PSV (≤ 8 cm H₂O) (conditional recommendation, moderate certainty).

Justification and implementation.

From our review of the literature, there is not compelling evidence supporting a large benefit of conducting SBTs with either PSV or T-piece when comparing one to the other. At the time of this writing, a multi-center RCT, the Frequency of Screening and SBT Technique (FAST) trial (Clinical Trials.gov ID: NCT02969226), is currently ongoing and relates directly to this topic.³⁴ FAST is examining the effects of SBT screening frequency (once vs at least twice daily) and SBT technique (PSV + PEEP vs T-piece) on time to successful extubation. The target enrollment is 760 critically ill adults. The results of this trial will likely inform whether PSV should be used during an SBT. It is important to appreciate that PSV in this context is the setting of PSV at a low level during SBT, presumably to decrease the work imposed by the endotracheal tube. This differs from PSV weaning, where the level of PSV is gradually reduced over time.

One on-ventilator approach to conducting SBTs is to set PSV to 0 cm H₂O and PEEP to 0 cm H₂O. Many clinicians consider this approach to be a reasonable surrogate for a T-piece trial. PSV 0/PEEP 0 allows use of the monitoring and alarms present on the ventilator. Moreover, in the event of a failed SBT, return to ventilatory support can be quickly re-established. Gacouin et al³⁵ conducted a single-center prospective observational study comparing PSV 0/PEEP 0 versus T-piece SBTs on re-intubation rates. The re-intubation rate at day 7 was 14.6% with PSV 0/PEEP 0 and 17.5% with T-piece (P = .40). However, because effort is required to trigger ventilator gas delivery, PSV 0/PEEP 0 might more closely resemble PSV than T-piece. A bench study by Sameed et al³⁶ reported that modern ventilators, in an effort to maintain a constant circuit pressure, may apply some positive pressure when set for PSV 0.

A related question concerns the use of CPAP during an SBT. To our knowledge, this has not been studied. It is worth noting that the RCTs by both Subirà³⁰ and Thille³¹ set PEEP to zero in the PSV arm of the RCT. In a physiologic meta-analysis, Sklar³⁷ reports that a CPAP of 0 cm H₂O and T-piece more accurately reflect the physiologic conditions after extubation compared to PSV. Conceptually, the addition of PEEP might improve triggering in obstructive diseases with auto-PEEP, but no trial focused only on PEEP exists.

There is one circumstance in which low-level PSV for the SBT makes clinical sense. In patients who will be extubated to NIV, such as those at risk for extubation failure (eg, COPD),¹⁰ it is reasonable that the SBT be performed on the settings that will be used postextubation. Anecdotally, common postextubation NIV settings are PSV of 5–8 cm H₂O and PEEP 5 cm H₂O.

Future research opportunities.

In the general population of mechanically ventilated patients, our recommendation is that either PSV or T-piece is acceptable for an SBT. But there might be populations for whom one or the other approach is better. In patients with cardiac failure, for example, there is the potential for acute cardiogenic pulmonary edema when positive pressure is removed. In patients with COPD and auto-PEEP, the removal of positive pressure might result in increased effort to breathe. For patients intubated with a small endotracheal tube, an SBT without PSV might result in excessive inspiratory muscle load. Whether these subgroups of patients might benefit from a PSV SBT is worthy of study.

Background.

In 1996, Ely et al⁴ published an RCT that became a primer for evaluating the readiness of patients for ventilator liberation. RTs conducted SBT screening between 6:30–7:00 am. Subjects were progressed to the SBT if they successfully passed all 5 screening criteria. Compared with usual practice (physician-directed liberation), the RT-driven protocol resulted in significantly fewer number of days from the time the subject had a successful screening test to the liberation from mechanical ventilation. Following this RCT, there was widespread uptake of this protocol, including daily morning screenings and SBTs. With little evidence to oppose these practices, there is less impetus for teams to alter or expand the process. However, there is much variation in the implementation of the liberation process among critical care areas.

Summary of the evidence.

Our literature review revealed a paucity of studies to further our understanding of this question. Only a few articles mentioned time of day when the SBT was performed, and we found no RCTs that directly assessed impact of SBT time of day. In the study by Ely et al,⁴ subjects were screened each morning between 6:30–7:00 am by an RT. In the Tanios et al study,^5,17 subjects were assessed every morning.

A corollary to this PICO is the frequency at which assessments of readiness and SBTs should be conducted. Esteban et al² reported that once-daily SBTs led to earlier ventilator liberation than SIMV or PSV weaning, but their RCT found no significant difference in speed of liberation between once-daily SBTs and multiple SBTs per day. In an observational study, Patel et al³⁸ compared RSBI measurements in the morning versus in the evening. The precise time of the 2 measurements was not specified in the protocol, except to say that they should be separated by > 4 h. They found no significant difference in RSBI measured in the morning versus the evening.

Recommendation.

We suggest a standardized approach to assessment and, if appropriate, completion of an SBT before noon each day (conditional recommendation, very low certainty).

Justification and implementation.

Local ICU culture influences the timing of SBTs. Based on expert opinion and practices used (but not compared) in previous RCTs, morning screening followed by an SBT when appropriate should be adopted as a standard where possible, with other patient care routines coordinated around that timing. This requires coordination between nurses adjusting sedation and RTs performing SBTs.^5,39 Because multidisciplinary rounds typically occur in the morning, it may be more likely that a decision regarding liberation and extubation will occur following a successful SBT in the morning.

The time of assessments for readiness and performance of SBTs should be incorporated into protocols adapted to the ICU culture.^10,40,41 Such protocols should address sedation (SATs), SBTs, and the coordination of SATs with SBTs.¹⁰ Unfortunately, adherence with existing guidelines is low, which causes delays in ventilator liberation.^42,43

The ongoing FAST trial³⁴ is comparing SBT screening frequency (once vs at least twice daily). The intervention calls for an RT to screen subjects between 6:00–8:00 am, and between 1:00–3:00 pm, with additional screens permitted at the clinician’s discretion. The results of this study will inform this PICO question.

Future research opportunities.

Common practice seems to be that SBTs are performed in the morning. However, this is based on very low-level evidence. Thus, additional studies are necessary to determine whether SBTs can be successful when performed at other times of the day. Anderson et al⁴⁴ implemented an automated real-time dashboard that facilitated information transfer regarding patients’ readiness for liberation and alerted clinicians when acceptable parameters were established. This proof-of-concept study demonstrated a faster time to extubation and decrease in ICU length of stay. Additional study is needed to determine whether real-time updates and notifications result in more rapid ventilator liberation rather than protocols tailored to clinician convenience.

Background.

Acceptable oxygenation is a usual criterion assessed before initiating an SBT.⁷ Common criteria to describe adequate oxygenation prior to initiation of SBT are S_pO2 > 90% with an F_IO2 of 0.40 or P_aO2/F_IO2 > 200 mm Hg (Table 5). Adequacy of oxygenation is also evaluated to determine tolerance of an SBT.^{2-5,17,23,30,31,45,46} Commonly, an SBT is terminated with the resumption of ventilator support if S_pO2 falls < 90% (Table 5). The practice in some ICUs is to increase the F_IO2 during an SBT, presumably to mitigate the potential for hypoxemia during the procedure. However, it is unclear whether such practice affects successful determination.

Summary of the evidence.

Our search identified no literature evaluating the effect of an F_IO2 increase on the outcome of an SBT. Thus, we evaluated the F_IO2 used in what we consider 10 seminal RCTs (Table 5).^{2-5,17,23,30,31,45,46} We considered this indirect evidence, in lieu of any RCTs directly addressing the question, to provide guidance and inform practice regarding the setting of F_IO2 during an SBT.

Recommendation.

We suggest that F_IO2 should not be increased during an SBT (conditional recommendation, very low certainty).

Justification and implementation.

In 10 RCTs that were used as indirect evidence, F_IO2 was not increased during the SBT.^{2-5,17,23,30,31,45,46} We recognize that this identifies the practices used in these RCTs without confirming the correctness of this practice. But lacking higher levels of evidence, this suggests that the F_IO2 should not be increased during an SBT.

Of concern, an increase in F_IO2 might lead to an increase in S_pO2, leading to false-positive SBT results. For example, if a patient has a baseline S_pO2 of 94% and desaturates to 88%, the SBT would likely be terminated. However, an increase in F_IO2 could result in an increase in S_pO2 from 94% to 100%, in this case masking the ongoing underlying cause of hypoxemic respiratory failure that otherwise might terminate the SBT. This could result in a false positive declaring an SBT as passed when it would have been classifying a failed SBT had the F_IO2 not been increased. On the other hand, perhaps the threshold of S_pO2 > 90% is too high, creating false negatives.

Future research opportunities.

To the best of our knowledge, the effect of raising F_IO2 during an SBT on important patient outcomes has not been studied. From our search of the literature, it appears that RCTs evaluating SBTs adopted oxygenation criteria used in older trials, dating to the 1990s. Although there have been sufficient studies to support systematic reviews evaluating predictors of SBT success such as the RSBI,¹⁸ there is an absence of high-level evidence to inform the management of oxygenation during an SBT.

The common practice of requiring an S_pO2 > 90% with an F_IO2 of 0.40 prior to SBT initiation may be too conservative. Conducting studies to establish an appropriate level of arterial oxygenation for the initiation of an SBT is important for future consideration. Studies should explore whether using higher F_IO2, especially when employing NIV or high-flow nasal cannula after extubation, leads to earlier extubation. It is also important to determine whether increasing the F_IO2 impacts SBT success or whether the practice of using the same F_IO2 during the SBT as set on the ventilator during the acute phase of mechanical ventilation is optimal.