Abstract
-
Background
- Time-based criteria for mechanical ventilation (MV) liberation in tracheostomized patients undergoing T-tube trials with ventilator discontinuation are inconsistent. We aimed to define the minimum post-liberation observation time and identify factors associated with successful liberation.
-
Methods
- This single-center retrospective cohort study included adults with tracheostomy who received MV in the intensive care unit and high-care unit of a Japanese hospital from November 14, 2019, to August 31, 2024. We measured maximum endurance time (MET) without MV, MV liberation success, and the shortest MET when success first exceeded 80% (mini-MET). Time to mini-MET was assessed by Kaplan-Meier median survival time (MST), and competing risk analysis treating in-hospital death as a competing event was performed. Predictors were screened by univariate C-index, and survival tree cutoffs were examined using logistic regression sensitivity analyses.
-
Results
- Among 160 patients, MV liberation success increased with MET from 65% to 94% at 168 hours. The MST to achieve the mini-MET, which was identified as 4 hours, was 7 days (95% CI, 5–10 days). Mechanical power normalized by dynamic compliance (MPCdyn) ≤454.72 J/min and albumin >2.3 g/dl were associated with success. MST was 3 days when both were satisfied. In a covariate-adjusted model, low MPCdyn plus high albumin remained associated with success. Replacing MPCdyn with driving pressure yielded similar findings.
-
Conclusions
- Four hours may be a practical minimum post-liberation observation time, whereas 48 hours may be a conservative option. Stratification by MPCdyn or driving pressure plus albumin can support consistent MV liberation decisions after tracheostomy.
-
Key Words: intensive care units; protective factors; time factors; tracheostomy; treatment outcomes; ventilator weaning
INTRODUCTION
Mechanical ventilation (MV) is indispensable for the management of critically ill patients, but nonphysiologic breathing patterns can lead to a variety of complications [1]. Failed or difficult weaning from MV is associated with higher mortality and longer intensive care unit (ICU) length of stay [2]. Therefore, ensuring appropriate early weaning is a key objective in critical care. Based on guidelines and clinical studies, the widely accepted definition of successful liberation from MV or extubation is the ability to avoid reintubation within 48–72 hours after extubation [3,4]. However, when difficult or prolonged weaning is anticipated, tracheostomy is generally selected during ICU stay. Tracheostomy has been reported to reduce respiratory workload by bypassing the upper airway and reducing anatomical dead space and flow resistance, as well as by improving airway clearance and facilitating weaning [5,6]. Furthermore, considering that liberation can be performed with the cannula in place, airway obstruction due to tongue root collapse is less likely, and the relative importance of level of consciousness becomes less significant; this is emphasized in extubation decisions for intubated patients [4,7,8]. For these physiological reasons, the sufficient observation period to confirm successful liberation may be shorter in patients on tracheostomy than in intubated patients. Importantly, a daily spontaneous breathing trial (SBT) is recommended and routinely used to guide liberation, even in patients with tracheostomy. Our study does not propose a time-based framework as an alternative to SBTs; rather, it addresses the practical question of how long patients should be observed after ventilator discontinuation after a successful liberation trial to confirm sustainable liberation success. Identifying this confirmation window may directly guide downstream decisions, such as discontinuation of standby ventilator availability and initiation of transfer/discharge planning, which remain heterogeneous among previous studies and in clinical practice. However, in practice, there is no standardized time-based definition of successful liberation for patients on tracheostomy; previous studies have used observation periods of 3, 5, or 7 days, at hospital discharge, or at the discretion of the attending physician [8-14], all of which are longer time frames than those used for intubated patients. As a result, a proportion of patients are transferred to the general wards while still receiving MV. In Japan, 10.7% of patients who are transferred out of the ICU are still on MV [15]. In other countries, there is institutional variation, but the reported rates are high at 51.8% to 72.0% [16,17]. Compared with the ICU, general wards have fewer staff, less monitoring, limited expertise, and insufficient handover and lack of training on managing tracheostomies and T-tubes [18], which can lead to delays in weaning. Moreover, inconsistent evaluation criteria can hinder clinical decision-making, care standardization, efficient use of medical resources, fair interfacility comparisons of liberation success rates, and development of evidence-based strategies. Therefore, this issue should not be considered confined to the ICU but rather viewed within the context of continuum of care from the ICU to the wards.
Determining if the observation period to assess successful liberation in patients on tracheostomy is actually shorter than the conventionally used criteria would allow decisions on the success or failure of liberation while still in the ICU or soon after tracheostomy and may reduce unnecessary prolonged MV duration and MV-related complications. Furthermore, confirming durable MV liberation before ICU discharge may be associated with fewer ICU readmissions, less burden on general ward staff who are unfamiliar with MV management, and improved healthcare resource utilization. To address these gaps, this study aimed to determine the minimum observation period after ventilator discontinuation to assess successful MV liberation and to identify related factors in patients with a tracheostomy in the ICU and/or high-care unit (HCU).
MATERIALS AND METHODS
This study was conducted in accordance with the ethical principles of the Declaration of Helsinki and implemented after obtaining approval from the Human Research Ethics Committee on Life Sciences and Medicine of the University of the Ryukyus (No. 23-2220-00-00-00). The requirement for individual informed consent was waived by the committee, given the retrospective minimal-risk design of the study and use of deidentified data. In accordance with national guidelines, study information and the opportunity to opt out were provided on the hospital website.
Study Design, Setting, and Participants
This was a single-center retrospective cohort study conducted in Japan on consecutive patients aged ≥18 years who were on tracheostomy and MV management at the ICU and/or HCU from November 14, 2019 to August 31, 2024. The end of follow-up was defined as the last day of hospitalization (discharge, transfer, or death). Exclusion criteria were patients not on tracheostomy, ICU or HCU readmission, patients who had undergone tracheostomy before admission or MV initiation, those who did not meet the weaning initiation criteria, those with missing data on the study outcomes, and those in whom pressure-controlled ventilation (PCV) settings on MV were not used within 24 hours before or after tracheostomy. All variables used for analysis, including ventilator parameters and vital signs, were extracted during documented periods under PCV mode within 24 hours before or after tracheostomy. The flowchart for participant selection is shown in Figure 1.
Outcomes
Primary outcomes included exploratory derivation of the minimum observation time required to determine successful liberation from MV and the time-varying liberation success rate. The latter was primarily summarized using Kaplan-Meier estimates and additionally evaluated using a competing risk framework for the minimum maximum endurance time (mini-MET) threshold. Minimum observation time was operationalized using the MET, defined as the longest continuous MV-free interval during hospitalization. To avoid conceptual confusion, MET was used to operationalize the postliberation observation window (i.e., duration that the patient remained continuously off MV after discontinuation), whereas final liberation success was judged independently at the hospital endpoint (discharge, transfer, or death) using prespecified physiologic criteria. Therefore, MET was not treated as the endpoint being predicted but as a time-based construct to define the minimum confirmation window (mini-MET) for downstream decision-making. For outcome adjudication at the hospital endpoint (discharge, transfer, or death), the gold standard for liberation success was defined as no use of MV at the end of follow-up and absence of physiologic respiratory abnormalities while off MV. Liberation failure was defined as use of MV at the end of follow-up, or the presence of physiologic respiratory abnormalities without MV use. Physiologic respiratory abnormalities that occurred while off MV were defined a priori using any of the following routine objective criteria for weaning discontinuation at the study institution (Supplementary Figure 1): (1) inability to maintain oxygenation target despite oxygen supplementation at ≥6 L/min; (2) transient SpO2 <80%; (3) newly developed arrhythmia or exacerbation of preexisting arrhythmia temporally associated with respiratory deterioration; (4) end-tidal carbon dioxide (EtCO2) ≥45 Torr (for type II respiratory failure and interpreted according to individualized clinical targets); and (5) escalation of respiratory support (e.g., high-flow oxygen, noninvasive ventilation) or MV reinitiation due to respiratory deterioration. The most recent available data on vital signs/monitoring, arterial blood gas (ABG) analysis, and/ or EtCO2 and clinician notes while the patient was off MV within 24 hours of the hospital endpoint were reviewed to evaluate whether any of the aforementioned criteria were met. Assessments were based on available ventilator records, routinely documented vital signs/monitoring data, ABG analysis/EtCO2 results, and contemporaneous clinical notes. Among the prespecified MET thresholds (no minimum threshold, 4, 12, 24, 48, 72, or 168 hours), the shortest MET threshold at which the liberation success rate first exceeded 80% was defined as the operational threshold for judging liberation success (mini-MET). The 80% threshold was chosen based on previously reported liberation success rates of 46.0% to 73.5% among patients with tracheostomy [10,13,19-21]. The secondary outcome included clinical factors associated with achieving mini-MET.
Data Items and Acquisition
Patient information was retrospectively collected from electronic medical records. As a rule, each item was obtained on the day of the tracheostomy; when data on mechanical power (MP) [22] were unavailable on the day itself, values within 24 hours before or after tracheostomy were used. To avoid the influence of intraoperative medications, such as sedatives and neuromuscular blockers, data within 3 hours after tracheostomy were excluded. Items collected were background variables, such as age, sex, primary disease, and medical history; severity indices, such as the Acute Physiology and Chronic Health Evaluation (APACHE) II score; vital signs, such as heart rate, systolic blood pressure, and respiratory rate (RR); weaning-related data, such as MET, total MV duration, and weaning duration; respiratory indices and MV settings, such as driving pressure (DP), positive end-expiratory pressure (PEEP), MP, tidal volume (VT), and water balance; laboratory data, such as albumin and creatinine; and clinical outcomes, such as in-hospital mortality, liberation success or failure, and length of hospital stay. MP for PCV was calculated using the simplified formula of Becher et al. [23]:
MPPCV=0.098×RR×VT×(DP+PEEP) [J/min] [23]
and was standardized using dynamic compliance (Cdyn) [24], which was calculated as VT/inspiratory pressure (PI), or predicted body weight (PBW) [13,24] as shown below.
MP normalized by dynamic compliance (MPCdyn)=MPPCV/Cdyn
MPPBW=MPPCV/PBW
As an alternative normalization, MP was also standardized by MPPBW and included as a candidate variable in the univariate screening step.
PBW was calculated as follows [24,25]:
50+0.91×(height [cm]−152.4) for men and
45.5+0.91×(height [cm]−152.4) for women
In this study, PI equaled the DP under PCV mode settings. Therefore, DP was operationalized as the inspiratory pressure above PEEP under PCV without requiring plateau pressure measurements. Furthermore, a standardized MP using only the elastance component was also calculated, referring to Syed et al. [26], as follows:
MP El=0.098×RR×VT×DP [J/min]
Weaning Protocol
Weaning from MV was gradually implemented in a stepwise manner according to the institutional protocol, which was developed using the recommendations of Boles et al. [3]. To ensure consistency among clinicians, weaning and outcome assessment was conducted by a multidisciplinary respiratory support team (RST) comprising ICU physicians, pulmonologists, clinical engineers, and nurses certified in respiratory therapy and/or critical care, in addition to the attending physician. Each patient was evaluated by at least two RST members using a standardized checklist based on the discontinuation criteria described in the weaning protocol. Discrepancies were resolved by multidisciplinary discussion and consensus.
After tracheostomy, the baseline setting was PCV mode, which was decreased in a stepwise manner according to the patient’s condition. When criteria were met, the mode was switched to pressure support ventilation (PSV), which comprised continuous positive airway pressure and pressure support. The level of PEEP was appropriately adjusted according to the optimal oxygenation level to prevent alveolar collapse and reduce work of breathing, as indicated in the Acute Respiratory Distress Syndrome Network [27]. Thereafter, the pressure support was gradually decreased; when all the criteria for a liberation trial were met, MV was discontinued and switched to T-tube. After ventilator discontinuation, withdrawal was continued unless discontinuation criteria were met. If any weaning discontinuation criterion was met, the protocol mandated reversion to more supportive settings, and weaning was resumed after at least 24 hours.
Weaning initiation was defined as the time point when management was switched to PSV mode after tracheostomy. In this study, the starting point for MET and mini-MET measurement was defined as the start of the SBT, rather than the time of tracheostomy. The SBT was defined as ventilator discontinuation and switching to T-tube. MET was measured from this time point as the longest continuous MV-free interval. Weaning days were counted from weaning initiation to the final day of MV use. Details of the protocol are shown in Supplementary Figure 1. Comprehensive support during weaning included proactive nutritional therapy and physiotherapy; the use of sedatives was avoided. The MV device used was the Puritan Bennett 980 (PB980) from Medtronic in the ICU and/or HCU or the Dräger Savina300 in the general wards.
Statistical Analysis
Overall patient background characteristics were summarized using descriptive statistics. To identify the primary outcome, we calculated the rate of successful liberation at each MET evaluation point and identified the mini-MET at which the success rate first exceeded 80%. For time-to-event analysis, achievement of mini-MET (1: achieved, 0: censored) and number of days to achievement were integrated for each patient. For patients who did not reach mini-MET, the analysis time was the length of stay. For predictor analyses, time was indexed from weaning initiation (PSV start), whereas MET and mini-MET were measured from SBT initiation (ventilator discontinuation and switching to T-tube). Time-to-event outcomes were summarized using the Kaplan-Meier method, and the median time to mini-MET achievement and its 95% CI were calculated. For the primary threshold, defined as the shortest MET threshold at which the liberation success rate first exceeded 80% (mini-MET=4 hours), a competing risk analysis was performed with in-hospital death treated as a competing event. Patients who were discharged alive, including those who were transferred to another hospital, long-term care facilities, or home, were treated as censored at the time of discharge. The cumulative incidence function (CIF) for achieving mini-MET was estimated using the Aalen-Johansen estimator. Pointwise 95% CIs for the CIF were obtained by nonparametric bootstrap analysis (2,000 resamples).
We focused the competing risk analysis on the 4-hour threshold because MET thresholds represent a sequential process (e.g., achieving 12 hours implies prior achievement of 4 hours); therefore, a single competing-risk time-to-event framework was not directly applicable to all thresholds simultaneously.
For the primary analysis of the secondary outcome, candidate factors were first screened using the univariate C-index. Specifically, a conditional inference tree (C-tree) was fitted with a maximum depth of 1 and a minimum node size of 20. The discriminative ability of the obtained predictions was assessed using Harrell’s C statistic, with the time scale being the number of days since weaning initiation and the event being the achievement of mini-MET. Variables with C >0.5 were considered initial candidates and visualized using bar graphs. Subsequently, using these candidate variables, branching rules and cutoff values were identified with a survival tree. For each terminal node, Kaplan-Meier estimates and median survival time (MST) are displayed in the figure legend for comparison.
An additional exploratory analysis was planned to assess whether DP could serve as a clinical surrogate for MP, if MP emerged as a relevant feature in the primary analysis. Finally, to examine the internal consistency of the identified features, a sensitivity analysis was conducted using logistic regression. For each variable identified in the first and second survival tree branches, three groups were created based on cutoff values: a reference group (first branch variable >cutoff), a ×1 group (first branch variable ≤cutoff and second branch variable ≤cutoff), and a ×2 group (first branch variable ≤cutoff and second branch variable >cutoff). Mini-MET was set as the dependent variable, with the ×1 and ×2 groups represented as dummy independent variables. Odds ratio (OR), 95% CIs (Wald), and P-values were estimated. Statistical analyses were performed using R version 4.4.1 (R Foundation for Statistical Computing), and the significance level was set at P<0.05. Missing values for all variables were handled without imputation.
RESULTS
Patient Characteristics
Data completeness was 100%. Median age was 74 years (interquartile range, 61.0–81.0 years); 70.6% were men, 66.3% were from surgical departments, and 82.5% were emergency admissions to the ICU and/or HCU. Of 160 cases analyzed, 104 (65.0%) had successful MV liberation, and 38 (23.8%) died in the hospital. Mean APACHE II score was 25.0±7.0 at admission and 21.8±5.2 at tracheostomy (delta=−3.2±7.0). Clinical and demographic details are provided in Table 1.
Primary Outcome
The rate of successful MV liberation gradually increased as the MET increased: 65% at no minimum threshold, 83% at 4 hours, 87% at 12 hours, 87% at 24 hours, 90% at 48 hours, 91% at 72 hours, and 94% at 168 hours (Figure 2, Supplementary Table 1). Based on our definition above, mini-MET was determined to be 4 hours. The MST to mini-MET achievement was 7 days (95% CI, 5–10 days) (Supplementary Figure 2).
In a competing risk analysis treating in-hospital death as a competing event, the CIF of achieving mini-MET (4 hours) was 55.0% (95% CI, 46.9%–63.1%) at 168 hours (7 days) and 65.6% (95% CI, 58.1%–72.5%) at 336 hours (14 days). These estimates were lower than the corresponding Kaplan-Meier–based estimates, indicating that mini-MET attainment was overestimated when competing mortality was not taken into account.
Secondary Outcome
In the initial screening, univariate C-index was the highest for various MP values, particularly MPCdyn. MPPBW also showed a discriminative signal in the univariate screening, but MPCdyn was carried forward because it showed the strongest discriminative performance. Other candidates included PaO2/FiO2 ratio, albumin, water balance, emergency surgery, emergency ICU and/or HCU admission, age, elective surgery as the reason for admission, cardiac arrest with return of spontaneous circulation prior to weaning initiation, and white blood cell (WBC) count (Figure 3).
Based on the candidate variables identified in the univariate screening (Figure 3), we performed a survival tree analysis to determine branching rules and cutoffs. For the first and second branches of the survival tree, the variables identified were an MPCdyn of 454.72 J/min (P<0.001) and albumin of 2.3 g/dl (P=0.007) (Figure 4). MST was 27 days for those who did not meet any of the two features, 7 days for those with low MPCdyn+low albumin, and 3 days when both were satisfied (Supplementary Figure 3, Figure 5). For clinical implementation, stratification by DP and albumin yielded results comparable to those obtained using MPCdyn. MST was 15 days for those who did not meet any of the two features, 7 days for those with low DP+low albumin, and 3 days for those with low DP+high albumin (Supplementary Figures 4-6).
In sensitivity analysis using logistic regression, groups were created based on the MPCdyn and albumin cutoff values identified in the survival tree, with mini-MET achievement as the dependent variable. Compared with the reference group (high MPCdyn+low albumin), the ×1 group (low MPCdyn+low albumin) had an OR for mini-MET achievement of 3.46 (95% CI [Wald], 1.38–8.64; P=0.00795) and the ×2 group (low MPCdyn+high albumin) had an OR of 56.89 (95% CI [Wald], 7.33–441.80; P=0.000111). These findings indicated that stratifying groups by MPCdyn and albumin was significantly associated with mini-MET achievement. On additional covariate-adjusted logistic regression that included APACHE II score, water balance, WBC count, and creatinine as variables, both groups remained significantly associated with mini-MET achievement. Compared with the reference group, the adjusted OR was 2.69 (95% CI, 1.01–7.14; P=0.047) for the ×1 group and 24.74 (95% CI, 5.21–117.57; P<0.001) for the ×2 group. In this adjusted model, APACHE II score was independently associated with mini-MET achievement, whereas water balance, WBC count, and creatinine were not.
DISCUSSION
The clinical purpose of mini-MET is to provide an operational definition of the postliberation confirmation window, which can guide downstream disposition-related decisions, such as discontinuation of standby ventilator availability and initiation of transfer/discharge planning, rather than to replace SBT-based liberation practice. Time-based criteria for MV liberation after tracheostomy remain inconsistent, and prolonged observation is common. In this context, our results showed that liberation success first reached the prespecified 80% success rate threshold at 4 hours (mini-MET) and increased to approximately 90% at 48 hours. The MST to the 4-hour threshold was 7 days, and stratification by MPCdyn or DP together with albumin delineated a cohort that reached this threshold earlier (approximately 3 days). Figure 3 shows the results of univariate screening, whereas the final two-feature framework (MPCdyn/DP and albumin) was derived from the subsequent tree-based analysis (Figures 4 and 5).
Importantly, after accounting for the competing risk of in-hospital death, the cumulative incidence of achieving mini-MET (4 hours) was lower than the corresponding Kaplan-Meier–based estimates, indicating that Kaplan–Meier analysis may overestimate absolute mini-MET attainment when competing mortality is not taken into account. Therefore, the 4-hour mini-MET threshold should be interpreted cautiously as an exploratory and practical postliberation confirmation window rather than as a definitively established cutoff. These estimates should be interpreted as probabilities of reaching the mini-MET threshold over time in the presence of competing death rather than as a direct measure of final liberation success.
Taken together, a minimum time-based criterion with a conservative option combined with MPCdyn/DP plus albumin-based stratification may provide an exploratory and reproducible framework that can be operationalized during ICU stay and that can support transparent resource-conscious decision-making, thereby potentially facilitating more consistent evaluations across settings.
The previously reported criteria for successful MV liberation in tracheostomized patients have been inconsistently defined as 3–7 days after ventilator discontinuation, at discharge, or at the physician’s discretion [8-14], and the practice of requiring prolonged observation periods persists [3]. Tracheostomy is known to reduce respiratory workload by decreasing anatomical dead space and flow resistance through upper airway bypass, improve secretion clearance, and facilitate weaning [5,6]. These physiological factors suggest that, in terms of the durability and sustainability of MV liberation, a short observation period may be clinically reasonable in patients on tracheostomy compared with the conventional 48–72 hours used to judge weaning success in intubated patients [3,4,28]. Given this physiological and clinical rationale, positioning mini-MET as a minimum conservative option is coherent with bedside decision needs. Methodologically, use of MPCdyn as the first branch was well justified, given that normalization by Cdyn has been shown to improve predictive accuracy [24]. DP is considered a simpler and appropriate alternative to MPCdyn to enhance clinical applicability, because it is a key component that strongly influences MPCdyn [22]. Combining MPCdyn with albumin, which can affect respiratory function and lung water content as a marker of nutrition, inflammation, and edema [29,30], was useful in identifying groups that were more likely to achieve mini-MET. Importantly, in a covariate-adjusted logistic regression including APACHE II score, water balance, WBC count, and creatinine, albumin-related stratification remained significantly associated with mini-MET achievement. This supported the potential utility of albumin-related stratification in our dataset, although causality cannot be inferred from this observational analysis. Consistency between the survival tree and logistic regression sensitivity analysis, together with the similar effects observed when substituting DP for MPCdyn, provided supportive internal consistency, although these analyses do not constitute independent validation. From a clinical standpoint, these findings suggest that a shorter postliberation confirmation window may be reasonable in selected patients after tracheostomy, and that traditional practices of prolonged observation warrant re-evaluation. As a practical note, in patients who meet the criteria but fail the initial liberation trial, repeated aggressive trials during the first 3 days may be considered, given that the MST was 3 days in the stratified cohort. Conversely, in patients who do not meet the criteria, early initiation of supportive interventions, such as fluid management, nutritional optimization, respiratory muscle training, and secretion management, should be considered before reattempts. By avoiding unnecessary continuation of MV, this approach aligns with strategies that facilitate earlier mobilization, mitigate ventilator-induced/ventilator-associated lung injury and ICU-acquired weakness, and may lead to downstream gains in activities of daily living and quality of life after discharge [1,28,31-35].
This study was not designed to evaluate cost reduction and economic implications; therefore, we refrain from drawing causal inferences about cost-effectiveness. Nevertheless, unnecessary MV continuation plausibly increases costs, and shortening observation when appropriate may contribute to more efficient resource allocation. For context, previous studies have reported representative daily costs of $5,811 for patients receiving MV and $2,880 for those not requiring MV [36].
This study had several limitations. First, this was a retrospective single-center study, and the defining factors and cutoffs for the mini-MET were derived from this cohort. Therefore, caution is required when generalizing these results to other institutions or settings, because the restriction to PCV mode limits applicability to other ventilator modes, including spontaneous breathing modes [23,37]. Because the cohort was derived from a single institution in Japan, center-specific case mix, tracheostomy timing, and weaning protocols, including local ICU/HCU structure and staffing, may have influenced the results, limiting external validity and generalizability. Second, the primary outcome of MV liberation failure was defined using a composite definition that incorporated clinical judgment regarding the absence of physiological abnormalities. Moreover, because the MET observation points were arbitrarily assigned, subtle changes within long intervals might not have been captured. Nevertheless, for operational reproducibility, the 4-hour mark mark was adopted as an exploratory operational threshold. These should be interpreted as design constraints. Internal validation of the survival tree-derived cutoff values using bootstrapping or cross-validation was not performed; therefore, these thresholds should be regarded as exploratory and external validation is required. Logistic regression analysis was used as a sensitivity analysis to examine internal consistency and should not be interpreted as independent validation of the survival tree-derived cutoffs. The very wide CI around the large OR, particularly for the ×2 group, indicates limited precision and possible instability of subgroup-based estimates; therefore, these findings should be interpreted cautiously. In addition, because spontaneous breathing effort and patient-ventilator asynchrony were not quantified, errors in dynamic compliance (i.e., MPCdyn) measurements cannot be excluded. Moreover, because DP and MPCdyn are time-varying variables, residual measurement variability may remain due to the lack of prospectively standardized measurement timing and conditions. Third, the time-to-event analysis was based on the time required for each patient to reach the prespecified MET threshold. Considering that MET thresholds represent a sequential process (e.g., achieving 12 hours implies prior achievement of 4 hours), a single time-to-event competing-risk framework is not directly applicable to all thresholds simultaneously. Therefore, we focused competing risk analyses on the primary clinically relevant threshold (mini-MET=4 hours) as the main exploratory threshold for postliberation assessment. In addition, recurrent events such as MV reattachment were not modeled within a recurrent-event framework, although MV reattachment and ongoing MV use on the last hospital day were incorporated into the definition of MV liberation failure. Therefore, the time-to-event estimates should be interpreted as the time to initial mini-MET achievement rather than as fully accounting for subsequent recurrent MV use. Fourth, our cohort included both surgical and nonsurgical patients, and such case-mix heterogeneity potentially influenced ventilation practices and liberation trajectories, thereby limiting generalizability. Although weaning was conducted according to standardized institutional protocol across care settings, including general wards, our cohort comprised patients who underwent tracheostomy at the ICU and/or HCU; therefore, residual heterogeneity related to the initial care setting (e.g., baseline severity and monitoring intensity during tracheostomy and early weaning) may have been present. In addition, several clinically relevant factors were not directly measured and may have confounded the observed associations, including respiratory muscle mass/strength and airway secretion burden. Taking these limitations into account, confirmation through external validation and prospective multicenter studies that explicitly address competing events and MV reattachment are desirable. Therefore, the observed relationships should be interpreted as associations rather than causal effects of an intervention.
Based on this single-center retrospective analysis, a short time-based criterion with a conservative fallback may be workable for ICU-based judgments, and MPCdyn/DP plus albumin may assist early stratification. The 4-hour mini-MET threshold may serve as a practical exploratory postliberation confirmation window for clinical decision-making.
KEY MESSAGES
▪ In tracheostomized adult intensive care unit (ICU)/high-care unit patients, the minimum observation time to confirm mechanical ventilation liberation was 4 hours (mini-MET), at which success at the hospital endpoint reached ≥80%; success was 90% at 48 hours, and the median time to reach 4 hours was 7 days.
▪ Stratifying by mechanical power normalized by dynamic compliance (MPCdyn) or driving pressure, together with serum albumin, identified a cohort achieving mini-MET in a median of 3 days.
▪ A simple 4-hour threshold (48 hours as a conservative option) may standardize ICU-based assessments and accelerate decisions; prospective multicenter validation is needed.
NOTES
-
CONFLICT OF INTEREST
No potential conflict of interest relevant to this article was reported.
-
FUNDING
None.
-
ACKNOWLEDGMENTS
We are grateful to the Clinical Engineering Department for training and support in device operation for data extraction, and to the clinical staff and administrators of the ICU, HCU, and general wards at Yuuai Medical Center for their assistance throughout the study. We especially thank all participants and their families for their cooperation.
-
AUTHOR CONTRIBUTIONS
Conceptualization: TN. Data curation: MT. Formal analysis: KS. Methodology: TN, GO. Project administration: TN, KT. Visualization: KS, MT, TN. Writing – original draft: TN. Writing – review & editing: TU, TN. All authors read and agreed to the published version of the manuscript.
SUPPLEMENTARY MATERIALS
Supplementary materials can be found via https://doi.org/10.4266/acc.000338.
Supplementary Figure 1.
Protocol for weaning from a mechanical ventilator. At our institution, ventilator discontinuation after a successful liberation trial is generally continued unless discontinuation criteria are met. In selected patients, an alternative stepwise withdrawal practice may be used, in which ventilator-free time is increased by approximately 2–8 hr/day at the attending physician’s discretion; in this study, patients who were managed with this alternative approach were not included. Catecholamine index was calculated as DOA (μg/kg/min)+DOB (μg/kg/min)+NAd (μg/kg/min)×100. HR: heart rate; SBP: systolic blood pressure; P/F ratio: PaO2/FiO2 ratio; PEEP: positive end expiratory pressure; PCV: pressure control ventilation; DP: driving pressure; VT: tidal volume; ARDS: acute respiratory distress syndrome; PS: pressure support; SBT: spontaneous breathing trial; MV: mechanical ventilation; EtCO2: end-tidal carbon dioxide; DOA: dopamine; DOB: dobutamine; NAd: noradrenaline.
acc-000338-Supplementary-Figure-1.pdf
Supplementary Figure 2.
Median survival time (MST) until achievement of minimum maximum endurance time (mini-MET) (MET=4 hours). The MST for achieving mini-MET was 7 days.
acc-000338-Supplementary-Figure-2.pdf
Supplementary Figure 3.
Comparison of achieved features (MPCdyn) among survival curves. Node 2=MPCdyn ≤454.72 J/min; Node 3=MPCdyn >454.72 J/min. MST: median survival time; mini-MET: minimum maximum endurance time.
acc-000338-Supplementary-Figure-3.pdf
Supplementary Figure 4.
Shortening effect of substituting the top minimum maximum endurance time (mini-MET) feature Mechanical power normalized by dynamic compliance with driving pressure on median survival time. The minimum node was set to 20. Driving pressure (DP) 12 cm H2O (P<0.001) and albumin 2.3 g/dl (P=0.004) were identified as the main factors influencing mini-MET.
acc-000338-Supplementary-Figure-4.pdf
Supplementary Figure 5.
Comparison of achieved features (driving pressure [DP]) among survival curves. Node 2=DP ≤12 cm H2O; Node 3=DP >12 cm H2O. MST: median survival time; mini-MET: minimum maximum endurance time.
acc-000338-Supplementary-Figure-5.pdf
Supplementary Figure 6.
Comparison of achieved features (driving pressure [DP] and albumin) among survival curves. Node 3=DP ≤12 cm H2O and albumin ≤2.3 g/dl; Node 4=DP ≤12 cm H2O and albumin >2.3 g/dl; Node 5=DP >12 cm H2O. Time origin was defined as the start of the spontaneous breathing trial (ventilator discontinuation and switching to T-tube). MST: median survival time.
acc-000338-Supplementary-Figure-6.pdf
Figure 1.Patient selection flowchart. The flowchart shows the identification, exclusion, and final inclusion of tracheostomized patients receiving mechanical ventilation in the intensive care unit (ICU) and/or high-care unit (HCU). MV: mechanical ventilation; PCV: pressure-controlled ventilation; HR: heart rate; SBP: systolic blood pressure.
Figure 2.Relationship between maximum endurance time (MET) and mechanical ventilator (MV) liberation success. With prolonged MET, MV liberation success rate increased from 65% as the overall base rate, when no minimum MET threshold was applied, to 83% at 4 hours and 87% at 12 hours, then plateaued between 12 and 24 hours before reaching 90% at 48 hours, 91% at 72 hours, and 94% at 168 hours. The vertical dotted line indicates MET (4 hours), defined as the shortest MET threshold at which the success rate first exceeded 80%. Points represent success rates, and error bars represent 95% CIs calculated using the Wilson method. Leftmost point represents the overall cohort success rate when no minimum MET threshold was applied and does not indicate a subgroup with zero ventilator-free time.
Figure 3.Univariate screening of candidates for achieving minimum maximum endurance time (univariate C-index). All variables, except those directly linked with research outcomes, were included as candidate predictors in the analysis. As a result, several top features were related to mechanical power (MP). This figure shows the variables included in the univariate screening step; variables that were carried over to the subsequent tree analysis are shown in Figure 4 and Supplementary Figure 4. Cdyn: dynamic compliance; PBW: predicted body weight; RSBI: rapid shallow breathing index; DP: driving pressure; APACHE: Acute Physiology and Chronic Health Evaluation; RR: respiratory rate; P/F ratio: PaO2/FiO2 ratio; VT: tidal volume; CPA: cardiopulmonary arrest; WBC: white blood cell.
Figure 4.Shortening effect on median survival time when considering the features of minimum maximum endurance time (mini-MET). The minimum node was set to 20. Mechanical power normalized by dynamic compliance (MPCdyn) 454.72 J/min (P<0.001) and albumin 2.3 g/dl (P=0.007) are identified as the main factors influencing mini-MET.
Figure 5.Comparison of achieved features (mechanical power normalized by dynamic compliance [MPCdyn] and albumin) among survival curves. Node 3=MPCdyn ≤454.72 J/min and albumin ≤2.3 g/dl; Node 4=MPCdyn ≤454.72 J/min and albumin >2.3 g/dl; Node 5=MPCdyn >454.72 J/min. Time origin is defined as the start of the spontaneous breathing trial (ventilator discontinuation and switching to T-tube). MET: maximum endurance time; MST: median survival time.
Table 1.Patient characteristics
|
Variable |
Value (n=160) |
|
Clinical characteristics |
|
|
Age (yr) |
74.0 (61.0–81.0) |
|
Sex (male) |
113 (70.6) |
|
BMI (kg/m2) |
23.0 (20.4–25.9) |
|
Obesity (BMI ≥25kg/m2) |
35 (21.9) |
|
Smoking history |
67 (41.9) |
|
Department (Surgery) |
106 (66.3) |
|
Emergency surgery |
62 (38.8) |
|
ICU admission type (emergency) |
132 (82.5) |
|
CPA before weaning |
23 (14.4) |
|
Vital sign |
|
|
APACHE II score at tracheostomy |
21.8±5.2 |
|
Initial APACHE Ⅱ score |
25.0±7.0 |
|
Delta APACHE Ⅱ score |
–3.2±7.0 |
|
GCS |
7.0 (6.0–10.0) |
|
GCS <8 |
86 (53.8) |
|
Heart rate (times/min) |
80.0 (71.0–93.0) |
|
Systolic blood pressure (mm Hg) |
124.0 (106.0–140.0) |
|
Respiratory rate (breaths/min) |
16.0 (14.0–20.0) |
|
Primary disease |
|
|
Pneumonia |
21 (13.1) |
|
Surgical operation |
73 (45.6) |
|
Emergency operation |
62 (38.8) |
|
Sepsis |
19 (11.9) |
|
Non-operated stroke |
9 (5.6) |
|
Other |
38 (23.8) |
|
Medical history |
|
|
Cerebral stroke |
32 (20.0) |
|
Chronic heart failure |
57 (35.6) |
|
Chronic kidney disease |
25 (15.6) |
|
Myoneural disease |
10 (6.3) |
|
Chronic respiratory disease |
17 (10.6) |
|
None |
19 (11.9) |
|
Weaning-related parameter |
|
|
≥2 Failure of extubation evaluation (SBT with cuff-leak test) |
90 (56.3) |
|
Trial history of extubation |
42 (26.3) |
|
History of re-intubation within 48 hr |
38 (23.6) |
|
Indications for tracheostomy |
|
|
In addition to expected intubation for over 2 wk, ≥2 failure of SBT with cuff-leak tests |
119 (74.4) |
|
Conditions that would affect the respiratory center and airway clearance (improvement may be slow or irreversible) |
41 (25.6) |
|
Duration of PC-A/C (PCV) mode before tracheostomy from start of ventilation |
12.0 (9.0–17.0) |
|
Sedation within 8 hr before tracheostomy |
59 (36.9) |
|
Days until liberation trial from tracheostomy |
2.0 (1.0–3.0) |
|
SBT failure reason |
|
|
Poor arousal |
95 (59.4) |
|
Poor spontaneous ventilation |
110 (68.8) |
|
Poor gas exchange |
34 (21.3) |
|
Poor self-airway protection |
134 (83.8) |
|
Social issue |
7 (4.4) |
|
SBT failure reason ≥2 items |
125 (78.1) |
|
Clinical outcome |
|
|
Final clinical outcomes |
|
|
In-hospital death |
38 (23.8) |
|
Discharged home |
15 (9.4) |
|
Transfer to chronic care hospital |
105 (65.6) |
|
Transfer to long-term care facility |
2 (1.3) |
|
Stay duration (day) |
|
|
Hospital stays |
53.5 (37.8–82.5) |
|
ICU and/or HCU stays |
17.0 (11.0–26.0) |
|
Ventilatory days (day) |
|
|
Total ventilatory days |
22.0 (12.8–37.0) |
|
Day of weaning after tracheostomy |
12.0 (4.0–25.5) |
|
Day from initiation of ventilation to tracheostomy |
10.0 (7.0–14.0) |
|
Liberation success |
104 (65.0) |
|
Respiratory index |
|
|
MP (J/min) |
13.4 (10.8–17.3) |
|
MPPBW (J/min) |
248.4 (186.1–322.9) |
|
MPCdyn (J/min) |
338.7 (235.2–526.8) |
|
Tidal volume (L) |
0.45 (0.40–0.52) |
|
Minute volume (L/min) |
7.68 (6.30–9.10) |
|
RSBI |
35.6 (29.1–45.0) |
|
P/F ratio |
342.7±109.3 |
|
DP (cm H2O) |
12.0 (10.0–15.0) |
|
PEEP (cm H2O) |
5.0 (5.0–6.0) |
|
Compliance (ml/cm H2O) |
38.3 (29.8–49.8) |
|
Water balance (L) |
0.5 (–2.1 to 3.9) |
|
Laboratory data |
|
|
Albumin (g/dl) |
2.5 (2.1–2.7) |
|
Creatinine (mg/dl) |
0.83 (0.59–1.33) |
|
Hematocrit (%) |
29.1 (26.4–33.1) |
|
WBC (×10³/μl) |
9.9 (7.4–12.9) |
REFERENCES
- 1. Beitler JR, Malhotra A, Thompson BT. Ventilator-induced lung injury. Clin Chest Med 2016;37:633-46.ArticlePubMedPMC
- 2. Jeong BH, Lee KY, Nam J, Ko MG, Na SJ, Suh GY, et al. Validation of a new WIND classification compared to ICC classification for weaning outcome. Ann Intensive Care 2018;8:115.ArticlePubMedPMCPDF
- 3. Boles JM, Bion J, Connors A, Herridge M, Marsh B, Melot C, et al. Weaning from mechanical ventilation. Eur Respir J 2007;29:1033-56.ArticlePubMed
- 4. Wang S, Zhang L, Huang K, Lin Z, Qiao W, Pan S, et al. Predictors of extubation failure in neurocritical patients identified by a systematic review and meta-analysis. PLoS One 2014;9:e112198. ArticlePubMedPMC
- 5. De Leyn P, Bedert L, Delcroix M, Depuydt P, Lauwers G, Sokolov Y, et al. Tracheotomy: clinical review and guidelines. Eur J Cardiothorac Surg 2007;32:412-21.ArticlePubMed
- 6. Vargas M, Sutherasan Y, Antonelli M, Brunetti I, Corcione A, Laffey JG, et al. Tracheostomy procedures in the intensive care unit: an international survey. Crit Care 2015;19:291.ArticlePubMedPMCPDF
- 7. Joint Faculty of Intensive Care Medicine of Ireland (JFICMI). Guidance for the care of the adult patient with a tracheostomy 2023 [Internet]. JFICMI 2023 [cited 2026 Jun 1]. Available from: https://assets.hse.ie/media/documents/Trachy_Care_Pathway__Final_web.pdf
- 8. Lee YC, Wang HC, Hsu CL, Wu HD, Hsu HS, Kuo CD, et al. The importance of tracheostomy to the weaning success in patients with conscious disturbance in the respiratory care center. J Chin Med Assoc 2016;79:72-6.ArticlePubMed
- 9. Hsu CL, Chen KY, Chang CH, Jerng JS, Yu CJ, Yang PC, et al. Timing of tracheostomy as a determinant of weaning success in critically ill patients: a retrospective study. Crit Care 2005;9:R46-52.ArticlePubMedPMCPDF
- 10. Ghiani A, Paderewska J, Walcher S, Neurohr C. Mechanical power normalized to lung-thorax compliance predicts prolonged ventilation weaning failure: a prospective study. BMC Pulm Med 2021;21:202.ArticlePubMedPMCPDF
- 11. Ghiani A, Paderewska J, Sainis A, Crispin A, Walcher S, Neurohr C, et al. Variables predicting weaning outcome in prolonged mechanically ventilated tracheotomized patients: a retrospective study. J Intensive Care 2020;8:19.ArticlePubMedPMCPDF
- 12. Carpene N, Celi A, Carrozzi L, Gabbrielli L, Pellegrini D, Palla A, et al. Predictors of success in weaning from invasive mechanical ventilation (IMV) in patients admitted to an intermediate respiratory care unit: a single-centre experience in Italy. Eur Respir J 2017;50:PA2105.Article
- 13. Nakama T, Umemura T, Hoshino S, Tamashiro M, Satoh K, Sekiguchi H, et al. Mechanical power to predict ventilator liberation in patients with a tracheostomy. Respir Care 2025;70:979-86.ArticlePubMedPMCPDF
- 14. Magnet FS, Bleichroth H, Huttmann SE, Callegari J, Schwarz SB, Schmoor C, et al. Clinical evidence for respiratory insufficiency type II predicts weaning failure in long-term ventilated, tracheotomised patients: a retrospective analysis. J Intensive Care 2018;6:67.ArticlePubMedPMCPDF
- 15. Japanese Society of Intensive Care Medicine (JSICM). Japanese intensive care patient database annual report 2023. version 1.0 [Internet]. JSICM 2025 [cited 2026 Jun 1]. Available from: https://www.jipad.org/images/data/manual/report/jipad_report_2023_English.pdf
- 16. Depuydt P, Oeyen S, De Smet S, De Raedt S, Benoit D, Decruyenaere J, et al. Long-term outcome and health-related quality of life in difficult-to-wean patients with and without ventilator dependency at ICU discharge: a retrospective cohort study. BMC Pulm Med 2016;16:133.ArticlePubMedPMCPDF
- 17. Paul N, Ribet Buse E, Grunow JJ, Schaller SJ, Spies CD, Edel A, et al. Prolonged mechanical ventilation in critically ill patients: six-month mortality, care pathways, and quality of life. Chest 2025;168:106-18.ArticlePubMedPMC
- 18. Jung YJ, Kim Y, Kyoung K, Keum M, Kim T, Ma DS, et al. The effect of systematic approach to tracheostomy care in patients transferred from the surgical intensive care unit to general ward. Acute Crit Care 2018;33:252-9.ArticlePubMedPMCPDF
- 19. Lee JM, Lee SM, Song JH, Kim YS. Clinical outcomes of difficult-to-wean patients with ventilator dependency at intensive care unit discharge. Acute Crit Care 2020;35:156-63.ArticlePubMedPMCPDF
- 20. Tai HP, Lee DL, Chen CF, Huang YT. The effect of tracheostomy delay time on outcome of patients with prolonged mechanical ventilation: a STROBE-compliant retrospective cohort study. Medicine (Baltimore) 2019;98:e16939. ArticlePubMedPMC
- 21. Tseng K, Shieh J, Cheng K, Chiang K, Chiang S, Ko S, et al. Tracheostomy versus endotracheal intubation prior to admission to a respiratory care center: a retrospective analysis. Int J Gerontol 2015;9:151-5.Article
- 22. Gattinoni L, Tonetti T, Cressoni M, Cadringher P, Herrmann P, Moerer O, et al. Ventilator-related causes of lung injury: the mechanical power. Intensive Care Med 2016;42:1567-75.ArticlePubMedPDF
- 23. Becher T, van der Staay M, Schädler D, Frerichs I, Weiler N. Calculation of mechanical power for pressure-controlled ventilation. Intensive Care Med 2019;45:1321-3.ArticlePubMedPDF
- 24. Yan Y, Xie Y, Chen X, Sun Y, Du Z, Wang Y, et al. Mechanical power is associated with weaning outcome in critically ill mechanically ventilated patients. Sci Rep 2022;12:19634.ArticlePubMedPMCPDF
- 25. Zhang Z, Zheng B, Liu N, Ge H, Hong Y. Mechanical power normalized to predicted body weight as a predictor of mortality in patients with acute respiratory distress syndrome. Intensive Care Med 2019;45:856-64.ArticlePubMedPDF
- 26. Syed MK, Selickman J, Evans MD, Dries D, Marini JJ. Elastic power of mechanical ventilation in morbid obesity and severe hypoxemia. Respir Care 2021;66:626-34.ArticlePubMedPMCPDF
- 27. Brower RG, Lanken PN, MacIntyre N, Matthay MA, Morris A, Ancukiewicz M, et al. Higher versus lower positive end-expiratory pressures in patients with the acute respiratory distress syndrome. N Engl J Med 2004;351:327-36.ArticlePubMedPMC
- 28. Girard TD, Alhazzani W, Kress JP, Ouellette DR, Schmidt GA, Truwit JD, et al. An official American Thoracic Society/American College of Chest Physicians clinical practice guideline: liberation from mechanical ventilation in critically ill adults. Rehabilitation protocols, ventilator liberation protocols, and cuff leak tests. Am J Respir Crit Care Med 2017;195:120-33.ArticlePubMedPDF
- 29. Cederholm T, Barazzoni R, Austin P, Ballmer P, Biolo G, Bischoff SC, et al. ESPEN guidelines on definitions and terminology of clinical nutrition. Clin Nutr 2017;36:49-64.ArticlePubMedPMC
- 30. Polito C, Martin GS. Albumin: physiologic and clinical effects on lung function. Minerva Anestesiol 2013;79:1180-6.PubMedPMC
- 31. Schmidt GA, Girard TD, Kress JP, Morris PE, Ouellette DR, Alhazzani W, et al. Official executive summary of an American Thoracic Society/American College of Chest Physicians clinical practice guideline: liberation from mechanical ventilation in critically ill adults. Am J Respir Crit Care Med 2017;195:115-9.ArticlePubMed
- 32. Hickmann CE, Castanares-Zapatero D, Bialais E, Dugernier J, Tordeur A, Colmant L, et al. Teamwork enables high level of early mobilization in critically ill patients. Ann Intensive Care 2016;6:80.ArticlePubMedPMCPDF
- 33. Barr J, Downs B, Ferrell K, Talebian M, Robinson S, Kolodisner L, et al. Improving outcomes in mechanically ventilated adult ICU patients following implementation of the ICU liberation (ABCDEF) bundle across a large healthcare system. Crit Care Explor 2024;6:e1001. ArticlePubMedPMC
- 34. Chen X, Lei X, Xu X, Zhou Y, Huang M. Intensive care unit-acquired weakness in patients with extracorporeal membrane oxygenation support: frequency and clinical characteristics. Front Med (Lausanne) 2022;9:792201.ArticlePubMedPMC
- 35. Busico M, Intile D, Sívori M, Irastorza N, Alvarez AL, Quintana J, et al. Risk factors for worsened quality of life in patients on mechanical ventilation: a prospective multicenter study. Med Intensiva 2016;40:422-30.ArticlePubMed
- 36. Zilberberg MD, Shorr AF. Prolonged acute mechanical ventilation and hospital bed utilization in 2020 in the United States: implications for budgets, plant and personnel planning. BMC Health Serv Res 2008;8:242.ArticlePDF
- 37. Paudel R, Trinkle CA, Waters CM, Robinson LE, Cassity E, Sturgill JL, et al. Mechanical power: a new concept in mechanical ventilation. Am J Med Sci 2021;362:537-45.ArticlePubMedPMC
Citations
Citations to this article as recorded by
KSCCM
, Kenichi Satoh2
PubReader
ePub Link
Cite
XML Download
