Association of weaning preparedness with extubation outcome of mechanically ventilated patients in medical intensive care units: a retrospective analysis

Introduction

Weaning from mechanical ventilation (MV) is an essential task in the care of patients with acute respiratory failure. Despite advances in assessing the feasibility of extubation, the rate of reintubation remains high at 10–19%, with a high mortality rate in those who fail extubation of 25–50% (Thille, Cortés-Puch & Esteban, 2013). The failure rate at the first attempt to wean can be as high as 50% (Peñuelas et al., 2011) and even up to 70% for those ventilated for more than 2 days (Tonnelier et al., 2011). Previous studies have reported the application of various assessments and predictors to increase successful weaning and extubation. Factors associated with extubation failure include age, the primary reason for intubation, neurological dysfunction, cough efficacy and amount of secretion (Tonnelier et al., 2011). Extubation failure may also be caused by hidden factors such as delirium (Ely et al., 2004) and intensive care unit (ICU)-acquired weakness (Garnacho-Montero et al., 2005).

Spontaneous breathing trials are currently the fastest approach to wean patients (MacIntyre et al., 2001), and it has been reported to be the most accurate predictor of extubation failure (Yang & Tobin, 1991). A single daily spontaneous breathing trial lasting from 30 min to 2 h has been recommended as the primary approach to weaning (Cook et al., 2000; MacIntyre et al., 2001). However, the probability of successful extubation in patients with acute respiratory failure may be related to a variety of factors not limited to traditional “weaning parameters”, and a clinical composite scoring method is lacking. A possible approach is to assess the preparedness of weaning by using a checklist-based bundle approach, such as the WEANSNOW checklist, which has been recommended to assess the degree of preparedness of a patient to begin breathing spontaneously (Hasani & Grbolar, 2008; Witt, 2008; Frances, 2010; Godara et al., 2014; Johnson & Abraham, 2015; Mendez, 2018). This evaluation method involves the assessment of eight items. However, the WEANSNOW checklist has not previously been investigated with regards to predicting the success of liberation from MV. Therefore, this study aimed to assess the prognostic ability of a composite score based on the WEANSNOW checklist for extubation failure in intubated and mechanically ventilated patients in a medical intensive care unit.

Materials and Methods

Results

Participants

Figure 1 shows the flow diagram of the patients included in this study. A total of 289 patients admitted were supported with mechanical ventilation during the study period and 205 of them underwent at least one attempt of extubation at the ICU, including 138 (67.3%) who succeeded in the first attempt of extubation and 67 (32.7%) who failed. For the failure group, 22 were re-intubated without receiving noninvasive ventilation, whereas 38 received NIV, but 28 were later reintubated and 10 were liberated from NIV upon transferred out the ICU; seven patients died because of withholding reintubation or noninvasive ventilation.

Regarding the performance of the ICU in this study, there were 321 admitted patients during the study period, including 289 receiving MV. Of these 289 patients, 148 (51.2%) were transferred out of the ICU after successful liberation from MV, 91 (31.5%) died and 50 (17.3%) were transferred out of the ICU under a ventilator-dependent state.

Table 2 compares the demographic and clinical characteristics between the two groups. Compared with the failure group, the success group had a significantly shorter duration of MV before the weaning attempt (11.2 ± 11.6 vs. 31.7 ± 26.2 days, p < 0.001), more with congestive heart failure (42.0% vs. 25.4%, p = 0.020), and had different distribution of the types of acute respiratory failure (p = 0.037) (Table 2).

Table 2:

Comparisons of demographic and clinical characteristics of the patients with successful and failed weaning attempts.

Variable	Total patients (n = 205)	Weaning success (n = 138)	Weaning failure (n = 67)	p-Value
Age, years	68.5 ± 14.0 (24–97)	68.1 ± 13.7 (34–93)	69.1 ± 14.8 (24–97)	0.637
Male, n (%)	124 (60.5%)	85 (61.6%)	39 (58.2%)	0.642
Duration of intubation, days	17.9 ± 20.1 (2–108)	11.2 ± 11.6 (2–67)	31.7 ± 26.2 (2–108)	<0.001
Main comorbidity, n (%)
Congestive heart failure	75 (36.6%)	58 (42.0%)	17 (25.4%)	0.020
Chronic obstructive pulmonary disease	43 (21.0%)	28 (20.3%)	15 (22.4%)	0.729
Chronic kidney disease	69 (33.7%)	45 (32.6%)	24 (35.8%)	0.648
Hypertension	95 (46.3%)	65 (47.1%)	30 (44.8%)	0.754
Diabetes	52 (25.4%)	35 (25.4%)	17 (25.4%)	0.999
Malignancy	92 (44.9%)	57 (41.3%)	35 (52.2%)	0.140
Type of respiratory failure, n (%)
Type I	57 (27.8%)	41 (29.7%)	16 (23.9%)	0.037
Type II	86 (42.0%)	64 (46.4%)	22 (32.8%)
Type III	9 (4.4%)	4 (2.9%)	5 (7.5%)
Type IV	53 (25.9%)	29 (21.0%)	24 (35.8%)
APACHE II score on admission	21.3 ± 8.1	21.4 ± 8.1	21.1 ± 8.1	0.860

DOI: 10.7717/peerj.8973/table-2

Note:

APACHE, acute physiology and chronic health evaluation.

Discussion

In this study, we developed a composite assessment score for the preparedness of weaning to study its association with the outcomes of weaning in patients with acute respiratory failure receiving MV. The results showed that the composite WS, intended to be used in a “bundle” score, was associated with weaning success when the score was 0 independently of the commonly used rapid shallow breathing index.

Successful extubation is traditionally predicted based on physiological assessments and calculated parameters such as the rapid shallow breathing index (Yang & Tobin, 1991). Other prediction models include APACHE II score (Afessa, Hogans & Murphy, 1999), CROP index (Yang & Tobin, 1991), relative inspiratory effort (Kline et al., 1987), alveolar-arterial oxygen tension gradient + blood urea nitrogen + gender score (A + B + G) (Scheinhorn et al., 1995), new weaning index based on ventilatory endurance and the efficiency of gas exchange (Jabour et al., 1991), and Integrated Weaning Index (Nemer et al., 2009). Several studies have reported that commonly used weaning indices have suboptimal prediction power for successful liberation from MV (MacIntyre et al., 2001; Kacmarek, Stoller & Heuer, 2016). These indices are commonly implemented in a clinical setting, are performed by respiratory therapists and focus more on ventilation function and less on airway condition. Previous reports have suggested that a substantial proportion of patients who fail extubation after a successful spontaneous breathing trial should be defined as “airway failure”, especially those who are reintubated within 48 h after extubation. We also used 48 h to define extubation success in our analysis. We found that most of the patients who were prepared for extubation had a WS of 0, suggesting that the clinicians were also confident about the preparedness for extubation. Therefore, we did not perform the receiver operating characteristic curve analysis to compare the accuracy with other weaning indices.

We also focused on components other than direct measurements of respiratory mechanics, thereby emphasizing the importance of adequately maintained natural airway function. The effect of an adequately sized artificial airway on the work of breathing is controversial (Straus et al., 1998). However, endotracheal tubes, which are partially obstructed by airway secretions, may not show abnormal findings by measuring weaning parameters, and may, therefore, also significantly increase the work of breathing after extubation, potentially resulting in airway failure requiring re-intubation. The size of the endotracheal tube can also affect the results of weaning trials. In critically ill patients, a decrease in internal diameter and an increase in minute ventilation increase the work of breathing. Adults have a critical increase in workload when the tube has an internal diameter of less than 7 mm (Sharar, 1995). Biofilm buildup on the inside of the endotracheal tube can cause dramatic increases in airway resistance, especially among infants and children (Mietto et al., 2014). Furthermore, the presence of upper airway obstructions or laryngeal edema may be detected by a cuff leak test showing diminished gas leak around the endotracheal tube with positive pressure breaths (Wratney & Cheifetz, 2007). Arterial gas analysis, nutritional impairment (McClave et al., 2009) and neuromuscular impairment are considered to affect ventilation function related to endurance, which may not be recognized by measuring weaning parameters alone. Finally, we found that wakefulness was significantly different between the success and failure groups, suggesting that this component contributes to both airway failure and ventilation failure, ultimately resulting in re-intubation.

The WEANSNOW bundle may assist in optimizing the preparedness of weaning-extubation and the WS may be a useful assessment tool to understand the condition of patients with acute respiratory failure more thoroughly. Although for each component, the majority of patients passed the criterion, our data also showed that only 48% had a total score of zero. Therefore, we would suggest that a bundled assessment might still be needed to remind the clinician about the risk of extubation failure. All of the clinical conditions related to the individual components of the WS can be improved before weaning-extubation if the patient is at high risk of failure. Multivariate regression analysis (Table 4) showed that the other variables used to predict the success of weaning-extubation, such as comorbidities, the reason for intubation, age and sex, are non-modifiable and therefore only serve as predictors. Even though the components of the WS can be improved before attempting extubation, patients who are already intubated with an endotracheal tube with a smaller internal diameter or those with chronic pulmonary and airway diseases such as chronic obstructive pulmonary disease that reduce the WS may not benefit from the WS assessment. Otherwise, the implementation of this bundle assessment to both predict and guide patient care might be considered.

There are several limitations to this study. The author decided each criterion for the individual components based on the clinician experience-based consensus as a lack of evidence-based generation of the WEANSNOW criteria. Although the result of multivariate regression showed the significance of WEANSNOW as a bundled assessment, the individual component might need further evaluation in future studies employing prospective inclusion of a sufficient number of patients of different characteristics and physiologic functions, which is beyond the scope of this retrospective study. Also, as our searching showed, most of the articles describing WEANSNOW appear to based on expert experiences and opinions provided in book chapters as a guide to the process in professional practices regarding ventilator weaning and extubation. Nevertheless, the lack of clinical evidence was also one of our original motivations to analyze the usability of this checklist-based assessment. Further evidence-based evaluation of the practice is necessary. Some patient characteristics were difference between the success and failure groups, such as duration of intubation, congestive heart failure as a comorbidity and the type of respiratory failure as a reason to intubate. Despite that these characteristics were included in the multi-variate analysis, further prospective randomized studies are needed to address these potential factors, such as stratification of patients with different co-morbidity, duration of intubation and type of respiratory failure. This study only included intubated patients in the medical ICUs of a single medical center. The patients underwent weaning and extubation mainly based on the decisions of the physician and the nurses followed standardized nursing care procedures. Our findings may not be generalizable to ICUs that have implemented weaning-extubation processes by respiratory therapists, or those that use similar checklist-based assessments and management strategies before proceeding with weaning-extubation. Besides, typical patients treated at the surgical ICU are mechanically ventilated for a brief period after surgery; therefore, the WS needs to be validated in the non-medical population of ICU patients. The design of this study designated those with post-extubation NIV use as failure group (Rothaar & Epstein, 2003). Recently, the potential of earlier extubation with immediate respiratory support using noninvasive ventilation has been suggested (Yeung et al., 2018). As the ICUs in this study did not routinely apply NIV for extubated patients unless there was evidence of worsening respiratory conditions requiring MV support, our study design excluded those patients who required either invasive or non-invasive ventilation within 2 days after extubation. Our study findings might not be generalized to the ICUs that routinely apply NIV proactively after extubation. However, it would be an interesting question whether or not patients with higher WEANSNOW scores can be successfully extubated with NIV as a step down therapy, and prospective studies are needed. Furthermore, regarding the performance of this ICU, only 51.2% were transferred out of the ICU after successful liberation from MV. These were medical ICUs that most of the patients have impaired gas exchange rather than merely post-procedural liberation of MV. The practice of physicians may have especially affected our determination of extubation outcomes as we defined successful extubation as unassisted spontaneous breathing maintained for at least 48 h. Therefore, the contribution of ICU performance to WEANSNOW would also need further investigation. This was a retrospective analysis, and the WS was based on information retrieved from the electronic medical records of the hospital. The information needed to decide whether or not the operational criteria of the WEANSNOW bundle checklist had been fulfilled did not necessarily assure actual optimal preparation in that regard. We included all of the patients who experienced an actual extubation attempt; therefore, the study allowed more than one spontaneous breathing trial before extubation, despite that only the assessment data closest to the time of extubation would be used. Our inclusion of only those patients who experience extubation would preclude those who did not undergo extubation regardless of the scoring of WEANSNOW. An auditing process may be more suitable for the WS. This study did not include a prospective investigation of predictability using a validation cohort of patients with the same inclusion-exclusion criteria. Physicians based on their clinical judgment made the decisions for extubation and re-intubation; however, their decision on whether or not to extubate may have also been based on clinical judgment related to the WEANSNOW components, and therefore the patients with a high probability of extubation failure may have remained intubated until their clinical condition improved. The post-extubation management may not have been standardized so that the decision to re-intubate may also have been affected by the clinicians. For the understanding of the impact of this pre-extubation practice with bundle checklist such as WEANSNOW, a standardized post-extubation care process till ICU discharge or re-institution of MV might be needed.

Conclusions

In conclusion, assessing the pre-extubation status of intubated patients using the WS might provide valuable insights into extubation failure in patients in a medical ICU for acute respiratory failure. As the individual components can be improved by active management, this bundle-based approach may also provide goals to optimize the patients’ condition before extubation to achieve better outcomes regarding liberation from MV. Further prospective studies are warranted to elucidate the practice of assessing weaning preparedness.

Supplemental Information