Abstract
Background
Traditional predictors of weaning outcomes primarily focus on acute illness severity and physiological parameters, while geriatric vulnerability domains such as frailty, functional dependence, and nutritional risk are often overlooked. Evidence regarding the incremental prognostic value of these domains for predicting weaning failure in critically ill older patients remains limited. This study aimed to evaluate the association between pre-admission frailty, functional status, comorbidity burden, and acute organ dysfunction with weaning failure in critically ill patients aged ≥ 65 years, and to compare their prognostic contribution with traditional severity scores.
Methods
This study is a secondary analysis of a national, multicenter, prospective observational cohort conducted across adult ICUs in Türkiye. Consecutive ICU patients aged ≥ 65 years who required invasive mechanical ventilation for more than 24 h were included. Multivariable logistic regression was used to identify factors independently associated with weaning failure. Weaning failure was defined as the need for reintubation within 7 days after extubation, death during the weaning process, or persistent requirement for invasive mechanical ventilation at day 90. Model discrimination was assessed using the area under the receiver operating characteristic curve (AUC).
Results
A total of 647 critically ill older patients were included in the study. Weaning failure occurred in 347 patients (53.6%). There was no significant difference in age between patients with weaning failure and those successfully weaned. Patients with weaning failure had significantly higher frailty scores, greater comorbidity burden, more severe organ dysfunction, higher nutritional risk, and worse functional dependency. In the final multivariable logistic regression model, higher Clinical Frailty Scale (CFS) (aOR = 1.26 per point; 95% CI = 1.16–1.38) and higher Sequential Organ Failure Assessment (SOFA) score (aOR = 1.12 per point; 95% CI = 1.06–1.19) were independently associated with weaning failure.
Conclusions
In critically ill older patients, frailty and early organ dysfunction were the factors most strongly associated with weaning failure, whereas chronological age alone showed limited prognostic value, supporting a shift from age-based to vulnerability-based risk assessment.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12877-026-07545-0.
Keywords: Weaning failure, Older patients, Frailty, Mechanical ventilation, Intensive care unit, Prognosis
Introduction
An increasing number of older adults are being admitted to intensive care units (ICUs) and receiving invasive mechanical ventilation (IMV). In this population, liberation from mechanical ventilation represents a critical milestone, particularly given their reduced physiological reserve and heterogeneous recovery trajectories. Consequently, disruptions in the weaning process have a substantial impact on clinical outcomes and healthcare resource utilization [1–3].
Weaning from IMV is a critical step in the ICU course, as prolonged weaning or failure is associated with increased complications, mortality, and resource use. These risks are particularly pronounced in older adults, in whom weaning represents a demanding physiological stress test requiring adequate cardiopulmonary and neuromuscular reserve. Although the independent effect of age remains debated, difficult weaning appears to be more frequent in older patients [1–5].
In clinical practice, prediction of weaning outcomes has traditionally focused on acute illness severity and organ dysfunction. However, in older adults, outcomes may be more strongly influenced by multidimensional vulnerability, particularly frailty, which reflects reduced physiological reserve and may impact successful liberation from mechanical ventilation [1–3].
Evidence remains limited regarding the prognostic value of frailty, functional status, and nutritional risk in predicting weaning failure in older critically ill patients, particularly when evaluated alongside acute organ dysfunction and conventional severity scores within the same framework. Given inter-center and cross-country variation in case mix, care processes, and resources, setting-specific data are needed to refine risk stratification and support individualized weaning decisions. Therefore, we aimed to identify factors independently associated with weaning failure and to compare the incremental predictive value of frailty, acute illness severity, and functional status beyond basic demographics.
Methods
Study design and participants
This study represents a secondary analysis of a multicentre, prospective, observational cohort that enrolled critically ill older adults (≥ 65 years) admitted to participating intensive care units (ICUs) during a predefined one-month study period. Study participation was announced by the Turkish Thoracic Society through an online call for collaboration. The parent study was designed and coordinated by the Turkish Thoracic Society Respiratory Failure and Intensive Care Working Group and was officially supported by the Turkish Thoracic Society.
During the 1-month inclusion period (24 March 2025 to 22 April 2025), all consecutive eligible patients were prospectively recruited from 55 adult mixed medical–surgical ICUs across 27 provinces in Türkiye, including university hospitals, training and research hospitals, and state hospitals. Each participating ICU received standardized case report forms (CRFs) with predefined clinical variables. Patients were followed until ICU discharge or death. Completed CRFs were submitted via e-mail and centrally reviewed to verify completeness, accuracy, and internal consistency prior to analysis.
Inclusion and exclusion criteria
This secondary analysis included patients from the parent multicentre prospective cohort who (1) were aged ≥ 65 years, (2) were admitted to one of the participating ICUs during the predefined 1-month enrolment period, (3) received IMV for > 24 h, and (4) underwent at least one separation attempt (SA) from IMV. Patients were eligible for the comparative analysis only if a weaning outcome (weaning success vs. weaning failure) was available in the standardized case report forms.
Data collection at ICU admission
At ICU admission, trained investigators prospectively recorded predefined clinical variables using standardized case report forms (CRFs). An English-translated version of the CRF is provided as Supplementary Material (Supplementary File 1).
Admission-level data included demographic characteristics, pre-ICU living status/residence (nursing home, own home, or living with children), and recent healthcare exposure (hospitalization and ICU admission within the preceding 6 months). Pre-existing comorbidities were recorded systematically.
Baseline physiological and laboratory measurements obtained on the day of ICU admission were recorded to characterize acute derangement and to support severity scoring. These included arterial blood gas parameters, renal/metabolic indices, liver function tests, serum proteins, electrolytes and hematological variables.
Assessment of comorbidity burden, acute severity, nutritional risk, and pre-ICU function
Pre-existing medical conditions were systematically documented for each participant, and overall comorbidity load was summarized using the Charlson Comorbidity Index (CCI) [6]. The severity of acute illness at the time of ICU admission was evaluated with the Acute Physiology and Chronic Health Evaluation II (APACHE II) and the Sequential Organ Failure Assessment (SOFA) scores, both computed from clinical and laboratory data obtained on the admission day [7, 8]. Nutritional vulnerability was screened with the modified Nutrition Risk in the Critically Ill (mNUTRIC) score; patients were categorized as having high nutritional risk when mNUTRIC ≥ 5 and low nutritional risk when mNUTRIC < 5 [9]. Baseline functional capacity prior to critical illness was defined as the patient’s usual status before ICU hospitalization. To capture this, the Clinical Frailty Scale (CFS), the Katz Index of Activities of Daily Living (ADL), and the FRAIL questionnaire were administered, using information obtained from the patient whenever possible, supplemented by interviews with family proxies/next of kin and review of available medical records. Frailty was assessed using the Clinical Frailty Scale (CFS), based on a previously translated, culturally adapted, and validated Turkish version (version 2.0) [10]. Assessments were performed prospectively by trained investigators using standardized case report forms, and a CFS score ≥ 5 was used to define frailty [11]. Frailty was also evaluated with the FRAIL instrument, which assesses five domains—Fatigue, Resistance, Ambulation, Illness burden, and Loss of weight—with FRAIL ≥ 3 defining frailty [12]. Functional dependence was quantified using the Katz ADL index. Patients with Katz ADL ≤ 4 were classified as functionally dependent, as scores of 4 or lower indicate at least moderate functional impairment in older adults [13, 14]. All pre-ICU functional assessments were referenced to the patient’s baseline condition approximately 2–4 weeks before ICU admission, based on patient/proxy report and, when feasible, corroborated using outpatient documentation and nursing-home records.
Assessment of weaning and classification of weaning outcomes
In patients receiving IMV, the weaning process was monitored on a daily basis, and the onset of weaning was defined—consistent with the WIND definition—as the first “separation attempt (SA)” aimed at discontinuing ventilatory support [5].
A separation attempt (SA) was defined as follows: in intubated patients, an SA corresponded to a spontaneous breathing trial (SBT) (with or without subsequent extubation) or direct extubation (planned or unplanned) without a formally documented SBT [5].
Patients who died or were transferred to another ICU before the first SA were classified as the “No SA” group, in line with the WIND approach describing patients who did not enter the weaning process (“no weaning/no SA”) [5].
Successful weaning (successful separation), based on WIND criteria, was defined in intubated patients as the absence of death or reintubation within 7 days after extubation (even if post-extubation noninvasive ventilation was used), or ICU discharge without IMV within 7 days [5].
Weaning failure was defined as (i) the need for reintubation within 7 days after extubation, (ii) death during the weaning process before meeting success criteria, or (iii) as a modified follow-up endpoint in our study, persistent requirement for IMV at day 90 [2, 5]. This extended endpoint was included to capture prolonged ventilator dependence and delayed weaning failure, which may not be adequately reflected by the standard 7-day WIND definition, particularly in critically ill older patients with reduced physiological reserve and slower recovery trajectories. Similar extended outcome definitions, including day-90 ventilator dependence, have been used in prior studies and secondary analyses of the WEAN SAFE cohort.
Data collection during ICU follow-up
Patients were prospectively followed throughout their ICU stay until discharge or death, and daily clinical data were recorded using standardized case report forms (CRFs). Primary clinical outcomes were assessed at ICU discharge, while 30-day mortality was recorded as a secondary outcome. For the assessment of weaning outcomes, a day-90 endpoint was used to capture prolonged ventilator dependence.
During ICU follow-up, key information related to the clinical course and weaning process was collected. This included data on respiratory support (including the need for invasive mechanical ventilation and extubation events), as well as the use of organ support therapies such as vasopressors/inotropes and renal replacement therapy. Nutritional support modalities were also recorded. In addition, major ICU complications were prospectively documented, including events such as delirium, acute kidney injury, ventilator-associated pneumonia, and other clinically relevant complications. Key outcomes, including ICU length of stay and 30-day mortality, were also recorded.
Statistical analysis
All study data were extracted from the hospital electronic database, cleaned, and organized in Microsoft Excel before analysis. Statistical analyses were performed using SPSS for Windows, version 31.0.1.0 (IBM Corp., Armonk, NY, USA). All tests were two-sided, and a p value < 0.05 was considered statistically significant. Effect estimates are reported with 95% confidence intervals (95% CI).
Continuous variables were assessed for distribution using the Kolmogorov–Smirnov test and visual inspection (histograms). Normally distributed variables are presented as mean ± standard deviation (SD), whereas non-normally distributed variables are summarized as median (interquartile range, IQR). Categorical variables are presented as counts (n) and percentages (%). Between-group comparisons were performed using Student’s t-test for normally distributed continuous variables and the Mann–Whitney U test for non-normally distributed continuous variables. Categorical variables were compared using the chi-square (χ²) test; Fisher’s exact test was used when expected cell counts were small.
To evaluate the incremental discrimination of frailty, comorbidity, nutritional risk, and illness severity to the prediction of weaning failure, hierarchical logistic regression models were constructed. A base model including age, sex, and pre-ICU living status was first fitted. Candidate clinical scores (CCI, CFS, Katz ADL, FRAIL, mNUTRIC, APACHE II, and SOFA) were then added individually to the base model to quantify incremental changes in discrimination. The final multivariable model was built using clinically relevant baseline variables and predictors demonstrating incremental discriminative value. In addition, variables were not selected solely based on statistical significance but also on their clinical plausibility and relevance to the weaning process. Adjusted odds ratios (aORs) with corresponding 95% confidence intervals (CIs) were reported for all variables included in the final multivariable model.
Model discrimination was assessed using receiver operating characteristic (ROC) curves and the area under the ROC curve (AUC) with 95% CIs. Model calibration was evaluated using the Hosmer–Lemeshow goodness-of-fit test for the standard logistic regression model. Calibration for the mixed-effects model was not formally assessed, as standard calibration tests such as the Hosmer–Lemeshow test are not directly applicable to multilevel models. Model performance was instead evaluated using classification accuracy and pseudo-R² measures. Analyses were conducted using complete cases for variables included in each model. There were no missing data for frailty and functional variables (CFS, FRAIL, and Katz ADL); therefore, all eligible patients were included in the analysis.
To account for potential clustering across participating ICUs, an additional multilevel analysis was performed using a generalized linear mixed model with a random intercept for center. This approach allowed adjustment for center-level variability in weaning practices. The model included the same covariates as the final multivariable model, and adjusted effect estimates with corresponding 95% confidence intervals were reported.
Results
As shown in Fig. 1, of the 1,529 patients in the parent cohort, 918 received invasive mechanical ventilation (IMV) for > 24 h. Among these, 647 patients underwent at least one separation attempt and were included in the comparative analysis. Of these, 53.6% were classified as weaning failure and 46.4% as weaning success according to the outcomes recorded in the CRFs.
Fig. 1.
Flow diagram of patient selection from the parent cohort and classification according to weaning outcome. IMV: invasive mechanical ventilation
Baseline characteristics of the study population
Baseline characteristics are presented in Table 1. Age, age-group distribution, and sex were similar between groups.
Table 1.
Baseline characteristics of the study population by weaning status
| All patients (n = 647) |
Weaning failure (n = 347) |
Weaning success (n = 300) |
p value | |
|---|---|---|---|---|
| Age, years | 76 [70–82] | 77 [71–83] | 76 [70–82] | 0.088 |
| Age group, n (%) | ||||
| 65–74 | 268 (41.4%) | 132 (38.0%) | 136 (45.3%) | 0.081 |
| 75–84 | 260 (40.2%) | 153 (44.1%) | 107 (35.7%) | |
| ≥ 85 | 119 (18.4%) | 62 (17.9%) | 57 (19.0%) | |
| Sex, n (%) | ||||
| Male | 360 (55.6%) | 188 (54.2%) | 172 (57.3%) | 0.421 |
| Pre-ICU residence, n (%) | ||||
| Nursing home | 42 (6.5%) | 31 (8.9%) | 11 (3.7%) | 0.014 |
| Own home | 376 (58.1%) | 190 (54.8%) | 186 (62.0%) | |
| Living with children | 229 (35.4%) | 126 (36.3%) | 103 (34.3%) | |
| Hospitalization within the last 6 months, n (%) | ||||
| Yes | 368 (56.9%) | 209 (60.2%) | 159 (53.0%) | 0.064 |
| ICU admission within the last 6 months, n (%) | ||||
| Yes | 173 (26.7%) | 99 (28.5%) | 74 (24.7%) | 0.268 |
| Comorbidities, n (%) | ||||
| Any comorbidity | 618 (95.5%) | 328 (94.5%) | 290 (96.7%) | 0.189 |
| Multimorbidity (≥ 2 comorbidities) | 529 (81.8%) | 284 (81.8%) | 245(81.6%) | 0.953 |
| DM | 267 (41.3%) | 147 (42.4%) | 120 (40.0%) | 0.543 |
| HT | 440 (68.0%) | 230 (66.3%) | 210 (70.0%) | 0.312 |
| CAD | 255 (39.4%) | 130 (37.5%) | 125 (41.7%) | 0.275 |
| CHF | 214 (33.1%) | 111 (32.0%) | 103 (34.3%) | 0.527 |
| CKD | 135 (20.9%) | 75 (21.6%) | 60 (20.0%) | 0.614 |
| Alzheimer disease | 129 (19.9%) | 84 (24.2%) | 45 (15.0%) | 0.003 |
| CeVD | 128 (19.8%) | 81 (23.3%) | 47 (15.7%) | 0.015 |
| COPD | 188 (29.1%) | 95 (27.4%) | 93 (31.0%) | 0.312 |
| Hematological malignancy† | 13 (2.0%) | 7 (2.0%) | 6 (2.0%) | 0.992 |
| Solid organ malignancy | 129 (19.9%) | 74 (21.3%) | 55 (18.3%) | 0.342 |
| Arrhythmia‡ | 48 (7.4%) | 25 (7.2%) | 23 (7.7%) | 0.814 |
Abbreviations; DM Diabetes Mellitus, HT Hypertension, COPD Chronic Obstructive Pulmonary Disease, CHF Congestive Heart Failure, CAD Coronary Artery Disease, CKD Chronic Kidney Disease, CeVD Cerebrovascular Disease
† Hematological malignancy: 1 missing value (N of valid cases = 646). ‡ Arrhythmia: 1 missing value (N of valid cases = 646; weaning success valid n = 299)
Pre-ICU residence differed by weaning status: nursing-home residency was more frequent among patients with weaning failure, whereas other living arrangements were comparable. A history of recent hospitalization was common and showed a non-significant trend toward being higher in the weaning failure group.
Overall comorbidity burden was high and largely similar between groups. However, Alzheimer’s disease and cerebrovascular disease were more frequent in patients with weaning failure (Table 1).
Baseline laboratory parameters are summarized in Supplementary Table S1. Most variables were comparable between groups, although total bilirubin and sodium levels differed significantly.
Premorbid condition and severity of illness scores
Baseline clinical scores differed significantly between groups (Table 2). Patients with weaning failure had higher comorbidity burden, greater frailty, and more frequent functional dependence. Nutritional risk was also higher in this group. In addition, organ dysfunction severity was greater among patients with weaning failure, whereas APACHE II showed only a non-significant trend.
Table 2.
Baseline clinical scores and frailty/nutritional risk categories according to weaning outcome
| All patients (n = 647) |
Weaning failure (n = 347) |
Weaning success (n = 300) |
p value | |
|---|---|---|---|---|
| CCI | 7 [5–9] | 7 [5–9] | 6 [5–8] | 0.018 * |
| CFS | 6 [4–7] | 7 [5–8] | 6 [4–7] | < 0.001 * |
| CFS group, n (%) | ||||
| CFS ≥ 5 | 468 (72.3) | 272 (78.4) | 196 (65.3) | < 0.001 ** |
| CFS < 5 | 179 (27.7) | 75 (21.6) | 104 (34.7) | |
| Katz ADL scale | 2 [1–4] | 2 [1–4] | 3 [1–5] | < 0.001 * |
| Katz ADL scale group, n (%) | ||||
| Katz ADL scale ≤ 4 | 504 (77.9) | 292 (84.1) | 212 (70.7) | < 0.001 ** |
| Katz ADL scale > 4 | 143 (22.1) | 55 (15.9) | 88 (29.3) | |
| FRAIL scale | 3 [3–4] | 4 [3–4] | 3 [2–4] | < 0.001 * |
| FRAIL scale group, n (%) | ||||
| FRAIL scale ≥ 3 | 513 (79.3) | 299 (86.2) | 214 (71.3) | < 0.001 ** |
| FRAIL scale < 3 | 134 (20.7) | 48 (13.8) | 86 (28.7) | |
| mNUTRIC score | 7 [6–8] | 7 [6–8] | 6 [5–8] | 0.001 * |
| mNUTRIC score group, n (%) | ||||
| mNUTRIC score ≥ 5 | 573 (88.6) | 321 (92.5) | 252 (84.0) | 0.001 ** |
| mNUTRIC score < 5 | 74 (11.4) | 26 (7.5) | 48 (16.0) | |
| SOFA score | 8 [6–10] | 8 [6–10] | 7 [5–10] | < 0.001 * |
| APACHE-II score | 25 [20–30] | 25 [21–30] | 24 [19–30] | 0.056* |
Data are presented as median [interquartile range (25th–75th percentile)] or n (%)
CCI Charlson Comorbidity Index, CFS Clinical Frailty Scale, FRAIL scale Fatigue, Resistance, Ambulation, Illness and Loss of weight scale, Katz ADL scale, Katz Activities of Daily Living scale, mNUTRIC score modified Nutrition Risk in Critically Ill Score, SOFA score Sequential Organ Failure Assessment, APACHE-II Acute Physiology and Chronic Health Evaluation II score
* p values were calculated using the Mann–Whitney U test
** p values were calculated using Pearson’s chi-square test
Statistical significance was defined as p < 0.05
Critical care interventions: organ support and adjunctive therapies
Significant between-group differences were observed in organ support and adjunctive therapies during ICU stay (Supplementary Table S2). The distribution of respiratory support at ICU admission was comparable between groups.
Albumin replacement and blood product use differed significantly between groups, including red blood cell, platelet, and plasma transfusions. Nutritional support strategies also varied significantly. Furthermore, requirements for renal replacement therapy and vasopressor use were higher in the weaning failure group.
Complications during ICU stay
The occurrence of complications during ICU stay was more frequent in the weaning failure group (Supplementary Table S3). Infectious complications—including ventilator-associated pneumonia, catheter-related infections, and urinary tract infections—were significantly more common in this group. Acute kidney injury was also more frequent, whereas delirium, gastrointestinal bleeding, and pneumothorax did not differ between groups.
Factors associated with weaning failure
Hierarchical logistic regression models were developed to evaluate the incremental prognostic value of clinical variables (Fig. 2). The base model (age, sex, pre-ICU living status) showed limited discriminative ability.
Fig. 2.
Receiver operating characteristic (ROC) curves of predictive models for weaning failure. The ROC curves compare the discriminative performance of the base model, including age, sex, and pre-ICU living status, with the final model incorporating the Clinical Frailty Scale (CFS) and SOFA score in addition to baseline variables. The base model demonstrated limited discrimination (AUC = 0.57), whereas the inclusion of frailty and organ dysfunction substantially improved model performance. The final model achieved the highest area under the curve (AUC = 0.67), indicating superior overall discrimination for prediction of weaning failure
Among individual predictors, frailty (CFS) provided the greatest improvement in model discrimination, followed by organ dysfunction severity (SOFA), while other variables contributed more modestly. The best model performance was achieved by combining frailty and organ dysfunction (Supplementary Table S4).
In the final multivariable model (Table 3), both CFS and SOFA were independently associated with weaning failure. In contrast, chronological age, sex, and pre-ICU living status were not independently associated. Model calibration was acceptable.
Table 3.
Final multivariable logistic regression model for weaning failure
| Variable | B | SE | Wald | Adjusted OR | 95% CI | p value |
|---|---|---|---|---|---|---|
| Age (per year) | 0.001 | 0.011 | 0.016 | 1.00 | 0.98–1.02 | 0.899 |
| Male sex | 0.081 | 0.172 | 0.220 | 1.08 | 0.77–1.52 | 0.639 |
| Clinical Frailty Scale (per point) | 0.233 | 0.044 | 27.777 | 1.26 | 1.16–1.38 | < 0.001 |
| SOFA score (per point) | 0.113 | 0.030 | 14.212 | 1.12 | 1.06–1.19 | < 0.001 |
| Pre-ICU living status (overall) | — | — | 2.942 | 0.230 | — | — |
| Living status (category 1) | −0.596 | 0.382 | 2.428 | 0.551 | 0.260–1.166 | 0.119 |
| Living status (category 2) | −0.668 | 0.390 | 2.931 | 0.513 | 0.238–1.102 | 0.087 |
Model performance: Area under the ROC curve (AUC) = 0.67 (95% CI 0.63–0.71); Hosmer–Lemeshow goodness-of-fit test p = 0.084
Odds ratios (ORs) are adjusted for all variables included in the model
A p-value < 0.05 was considered statistically significant
The mixed-effects logistic regression model accounting for clustering across 55 centers demonstrated that higher frailty and greater organ dysfunction were independently associated with weaning failure. Specifically, each one-point increase in the Clinical Frailty Scale was associated with a 27.9% increase in the odds of weaning failure (aOR 1.279, 95% CI 1.166–1.404; p < 0.001), while each one-point increase in SOFA score increased the odds by 10.6% (aOR 1.106, 95% CI 1.037–1.178; p = 0.002). Age, sex, and pre-ICU living status were not significantly associated with weaning failure (Table 4).
Table 4.
Multivariable mixed-effects logistic regression analysis for weaning failure
| Variable | Adjusted OR | 95% CI | p value |
|---|---|---|---|
| Age | 0.980 | 0.975–1.021 | 0.871 |
| Sex (male) | 0.880 | 0.616–1.256 | 0.480 |
| Clinical Frailty Scale (CFS) | 1.279 | 1.166–1.404 | < 0.001 |
| SOFA score | 1.106 | 1.037–1.178 | 0.002 |
| Pre-ICU living status | Not significant | 0.714 | |
A generalized linear mixed-effects model with a random intercept for center (55 centers) was used. Weaning failure was coded as the outcome (1 = failure). Model discrimination and fit were evaluated using classification accuracy and pseudo-R² indices.
The model demonstrated acceptable performance, with an overall classification accuracy of 70.8%. The marginal and conditional R² values were 0.084 and 0.165, respectively, indicating a modest contribution of fixed effects and additional variance explained by center-level clustering. The intraclass correlation coefficient (ICC ≈ 0.08) suggested a limited but non-negligible center effect.
Discussion
In this national multicenter cohort of critically ill older adults receiving IMV, weaning failure was common, occurring in more than half of patients. We found that pre-admission frailty and acute organ dysfunction were independently associated with weaning failure, whereas chronological age did not retain independent prognostic value after adjustment. These findings suggest that in older ICU patients, unsuccessful liberation from mechanical ventilation is driven more by the interaction between physiological reserve and acute illness burden than by age alone.
The global shift toward an aging population highlights the growing need to better understand weaning in critically ill older adults, in whom prolonged ventilation is linked to worse outcomes [2, 15]. Variation in weaning practices across ICUs further supports the need for updated, patient-centered guidance that incorporates individualized risk assessment [16].
Compared with large international cohorts such as WEAN SAFE, the rates of weaning failure and mortality observed in our study appear higher. These differences likely reflect variations in patient characteristics and case-mix rather than fundamental differences in underlying associations. Our cohort specifically included older critically ill patients (≥ 65 years) with a high prevalence of frailty, comorbidity, and acute organ dysfunction. In contrast, WEAN SAFE enrolled a broader and more heterogeneous ICU population with a lower proportion of high-risk older and frail individuals. In addition, differences in healthcare systems, ICU organization, and local weaning practices across countries may have contributed to the observed discrepancies. Therefore, our findings should be interpreted within the context of a higher-risk population and are intended to provide complementary, population-specific insights rather than direct quantitative comparisons across studies.
While chronological age has traditionally been considered a primary risk factor for weaning failure, emerging evidence suggests that its predictive power may be limited when considered in isolation, particularly when more comprehensive assessments of physiological reserve and functional status are available [2, 17, 18]. Although the model demonstrated independent associations, its discriminative performance was modest (AUC: 0.67), reflecting the multifactorial and dynamic nature of the weaning process. This level of discrimination may limit its use for individualized bedside prediction. However, the primary aim of this study was not to develop a predictive model, but to evaluate the independent and incremental associations of frailty and acute illness severity with weaning outcomes. The same review reports that although increasing age may appear associated with lower success rates in some cohorts [19].
Moreover, the comorbidities, which are common in older adults, may complicate weaning by increasing susceptibility to complications and prolonging recovery [20]. Comorbidities are more prevalent in older age groups and contribute significantly to the risk of weaning failure [21]. For example, studies have indicated that diabetes and prolonged hospitalization before endotracheal intubation are independent factors influencing weaning failure from IMV in older patients with severe COVID-19 [22]. However, in our national multicenter cohort of critically ill older adults, the overall comorbidity burden was high; nevertheless, the prevalence of common comorbidities did not differ significantly between the weaning failure and weaning success groups.
Severity scores also require careful interpretation. The literature reports mixed findings for APACHE II. In an older COVID-19 cohort, APACHE II has been reported as an independent risk factor for weaning failure, and in an older sample with type II respiratory failure, APACHE II has likewise been identified among the independent risk factors [22, 23]. However, because APACHE II is calculated within the first 24 h, it may not reflect clinical status at the time weaning begins [17]. In our study, APACHE II scores showed only a borderline difference between groups (p = 0.056), and the associated increase in discriminative performance was relatively limited. When APACHE II was added to the model, its incremental contribution to discriminative performance was more modest; within the final framework, the more decisive signal appeared to be the combination of physiological reserve and acute burden. Practically, this suggests that although a “high APACHE II” score is a warning signal for clinicians, it should not, on its own, drive weaning decisions in older patients; rather, it becomes more meaningful when interpreted alongside frailty and organ dysfunction. In addition, we deliberately used admission SOFA as a baseline measure of illness severity, reflecting early organ dysfunction at ICU entry. However, dynamic measures such as SOFA at the time of weaning or delta-SOFA may better capture the trajectory of organ dysfunction and its relationship with weaning outcomes. These time-dependent variables were not consistently available across all participating centers and were therefore not included in the present analysis. Future studies incorporating longitudinal assessments of organ dysfunction may further improve the prediction of weaning outcomes.
The literature also supports the role of SOFA in the weaning trajectory. In the WIND study, patients with a short weaning process had lower SOFA scores at ICU admission, and lower SOFA was shown to be associated with a shorter weaning duration [5]. On the other hand, in large datasets examining the relationship between frailty and weaning outcomes, the inclusion of SOFA in models around the time of the first separation attempt suggests that these two variables capture different yet complementary dimensions rather than measuring the same construct [2].
Frailty is a multidimensional state of vulnerability that reflects loss of “reserve” across multiple physiological systems rather than impairment of a single organ system, and a reduced ability to maintain homeostasis following exposure to stressors such as critical illness, sedation, immobilization, or infection. Accordingly, its impact becomes particularly evident in critically ill older patients, in whom weaning functions as a “load–capacity” stress test [19]. When reserve is limited, age-related reductions in cardiopulmonary capacity, respiratory muscle strength/endurance, and secretion management make a load–capacity imbalance more likely during weaning [24]. Weaning failure is therefore increasingly viewed as multifactorial—driven by the interplay of respiratory and cardiac load, neuromuscular adequacy, critical illness–related weakness, neuropsychological factors, and metabolic/endocrine disturbances—of which frailty is a pragmatic summary marker of “total reserve” rather than single-organ dysfunction [19, 20]. Differences in case-mix, patient selection, and healthcare system characteristics may partly explain the relatively high prevalence of frailty observed in our cohort.
In our study, CFS remained independently associated with weaning failure after adjustment for SOFA. This suggests that frailty provides prognostic information beyond acute organ dysfunction, reflecting baseline reserve and recovery capacity. Consistently, a secondary analysis of the WEAN SAFE study found frailty to be independently associated with delayed weaning initiation and weaning failure, with a consistent effect across models [2]. While these findings are consistent with prior evidence, it is important to note that the WEAN SAFE secondary analysis focused on very old patients (≥ 80 years), whereas our study includes a broader population of critically ill older adults aged ≥ 65 years, thereby extending the applicability of these observations to a wider clinical population. In addition, by integrating multiple domains of vulnerability—including frailty, functional status, comorbidity burden, and nutritional risk—within a single analytical framework, our study provides a more comprehensive assessment of their incremental prognostic value in predicting weaning outcomes.
From a clinical perspective, the combined assessment of frailty (CFS) and organ dysfunction (SOFA) may provide a practical framework for bedside risk stratification in critically ill older patients undergoing weaning. Patients with high frailty and elevated SOFA scores may be at increased risk of weaning failure and may benefit from a more cautious and individualized weaning strategy, including closer monitoring, optimization of reversible factors, and multidisciplinary support. Conversely, patients with lower frailty and organ dysfunction may be considered for earlier and more proactive weaning attempts. Although these findings should not be used in isolation to guide clinical decisions, they support a more integrated and patient-centered approach to weaning in older ICU populations.
Strengths of this study include the national, multicenter prospective design with consecutive enrolment across diverse adult ICUs in Türkiye, supported by standardized CRFs and centralized data checks. We assessed baseline geriatric vulnerability domains (frailty, functional dependence, nutritional risk) alongside conventional severity scores and used hierarchical modeling to quantify their incremental prognostic value, identifying an admission-level combination (CFS + SOFA) with the best discrimination for predicting weaning failure.
The findings should be interpreted in light of several limitations. First, this study represents a secondary analysis of a prospective multicenter cohort with a predefined one-month enrolment period. While the parent study was prospectively designed and nationally representative, the restricted inclusion window limits temporal representativeness and precludes definitive prognostic inference. Accordingly, the present findings should be interpreted as exploratory and hypothesis-generating rather than as a validated prediction model. Temporal clustering of admissions and seasonal variation in respiratory disease burden or ICU case-mix may have influenced both exposures and outcomes, and residual confounding related to time-dependent factors cannot be excluded. Second, the study population comprised a heterogeneous mix of medical and surgical ICU patients. These groups differ substantially with respect to indications for mechanical ventilation, sedation exposure, perioperative physiology, and expected recovery trajectories. In surgical patients, ventilatory dependence may be transient and procedure-related, whereas in medical patients it more often reflects persistent respiratory or systemic failure. Such heterogeneity may have attenuated condition-specific weaning signals and reduced the precision of effect estimates for individual diagnostic subgroups. However, this approach was intentionally chosen to reflect real-world ICU practice and to enhance external validity in a national cohort of older critically ill patients. At the same time, this heterogeneity may also be considered a strength, as it reflects real-world ICU practice and enhances the generalizability of our findings across diverse clinical settings. Third, despite the multicenter design, potential center-level effects could not be fully accounted for. Inter-unit variability in sedation practices, ventilator management, staffing patterns, thresholds for separation attempts, and local weaning protocols may have influenced both the timing and outcome of weaning. In addition, detailed patient-level data on sedation practices and sedation depth were not systematically collected, as the present analysis focused primarily on baseline clinical predictors. This represents an important limitation, as sedation may influence both readiness for weaning and the risk of weaning failure. Although all participating ICUs used standardized case report forms and prospectively collected data, residual confounding related to unmeasured center-specific practices remains possible and may have contributed to variability in observed outcomes. Fourth, an additional limitation of this study is the exclusion of patients who did not undergo a separation attempt (No SA group), including those who died before the initiation of the weaning process. This approach may have introduced a potential selection bias by excluding the most critically ill patients, thereby potentially underestimating the overall burden and determinants of weaning failure. However, our analysis specifically focused on the weaning process among patients who underwent at least one separation attempt, consistent with prior studies such as the WEAN SAFE study. Nevertheless, our findings should be interpreted within the context of patients who survived long enough to enter the weaning phase and may not be fully generalizable to the entire population of mechanically ventilated older ICU patients.
Conclusion
In critically ill older patients, frailty and acute organ dysfunction emerge as complementary factors associated with weaning failure, whereas chronological age alone provides limited prognostic information. Rather than offering a definitive prediction model, our findings highlight the importance of integrating baseline physiological reserve with acute illness severity when interpreting weaning trajectories in older adults. Future studies with longer enrolment periods and center-level modeling are needed to determine whether frailty-informed weaning strategies can improve patient-centered outcomes in this growing population.
Supplementary Information
Acknowledgements
None declared.
Declaration of generative AI and AI-assisted technologies in the writing process
The authors declare that no generative AI or AI-assisted technologies were used in the writing, analysis, or preparation of this manuscript.
Authors’ contributions
Umut Sabri Kasapoglu contributed to the conception, design, data collection and/or processing, analysis and/or interpretation, writing, literature review and critical review.Ozlem Yazicioglu Mocin contributed to the conception, supervision, design, writing, critical review and fundings.Eylem Tuncay contributed to the conception, supervision, design, writing, critical review, fundings, analysis and/or interpretation.Sinem Gungor contributed to the conception, design, data collection and/or processing, critical review and fundings, analysis and/or interpretation, writing and supervision.Huseyin Arikan contributed to the design, data collection and/or processing, analysis and/or interpretation, writing, literature review.Ozlem Ediboglu contributed to the design, data collection and/or processing, analysis and/or interpretation, writing, literature review, critical review and supervision.Nazli Huma Teke contributed to the design, data collection and/or processing, analysis and/or interpretation, writing, literature review.Rest of the study group contributed to the data collection and/or processing, analysis and/or interpretation and conception.All authors read and approved the final manuscript.
Funding
This manuscript did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Data availability
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.
Declarations
Ethics approval and consent to participate
The parent study was approved by the Marmara University School of Medicine, Non-Drug and Non-Medical Device Research Ethics Committee (approval date: 02 July 2024; decision no: 09.2024.486).
This manuscript presents a secondary analysis of an already approved multicenter prospective cohort. No additional recruitment or changes to routine care or data collection were undertaken; we only evaluated an additional endpoint (ventilator weaning/extubation failure) within the existing dataset. The ethics committee was notified of this added endpoint and confirmed that no new approval was required. The study was conducted in accordance with the principles of the Declaration of Helsinki and applicable national regulations.
Consent for publication
Written informed consent was obtained from all participants or their legally authorized representatives as required by the ethics approval of the parent study. For this secondary analysis of de-identified data, additional consent was not required. All data were anonymized prior to analysis, and no information that could identify individual participants is included in this manuscript.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Bruno RR, Wernly B, Bagshaw SM, van den Boogaard M, Darvall JN, De Geer L, de Gopegui Miguelena PR, Heyland DK, Hewitt D, Hope AA, et al. The Clinical Frailty Scale for mortality prediction of old acutely admitted intensive care patients: a meta-analysis of individual patient-level data. Ann Intensive Care. 2023;13(1):37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Laffey CM, Sheerin R, Khazaei O, McNicholas BA, Pham T, Heunks L, Bellani G, Brochard L, Tomescu D, Simpkin AJ, et al. Impact of frailty and older age on weaning from invasive ventilation: a secondary analysis of the WEAN SAFE study. Ann Intensive Care. 2025;15(1):13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Fronczek J, Polok K, de Lange DW, Jung C, Beil M, Rhodes A, Fjolner J, Gorka J, Andersen FH, Artigas A, et al. Relationship between the Clinical Frailty Scale and short-term mortality in patients >/= 80 years old acutely admitted to the ICU: a prospective cohort study. Crit Care. 2021;25(1):231. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Azeredo LM, Nemer SN, Barbas CS, Caldeira JB, Noe R, Guimaraes BL, Caldas CP. The Integrative Weaning Index in Elderly ICU Subjects. Respir Care. 2017;62(3):333–9. [DOI] [PubMed] [Google Scholar]
- 5.Beduneau G, Pham T, Schortgen F, Piquilloud L, Zogheib E, Jonas M, Grelon F, Runge I, Nicolas T, Grange S, et al. Epidemiology of Weaning Outcome according to a New Definition. The WIND Study. Am J Respir Crit Care Med. 2017;195(6):772–83. [DOI] [PubMed] [Google Scholar]
- 6.Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373–83. [DOI] [PubMed] [Google Scholar]
- 7.Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of disease classification system. Crit Care Med. 1985;13(10):818–29. [PubMed] [Google Scholar]
- 8.Vincent JL, Moreno R, Takala J, Willatts S, De Mendonça A, Bruining H, Reinhart CK, Suter PM, Thijs LG. The SOFA (Sepsis-related Organ Failure Assessment) score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis-Related Problems of the European Society of Intensive Care Medicine. Intensive Care Med. 1996;22(7):707–10. [DOI] [PubMed] [Google Scholar]
- 9.Heyland DK, Dhaliwal R, Jiang X, Day AG. Identifying critically ill patients who benefit the most from nutrition therapy: the development and initial validation of a novel risk assessment tool. Crit Care. 2011;15(6):R268. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Asik Z, Kilinc S, Kursun O, Ozen M. Validation of the Clinical Frailty Scale version 2.0 in Turkish older patients. Geriatr Gerontol Int. 2022;22(9):730–5. [DOI] [PubMed] [Google Scholar]
- 11.Rockwood K, Song X, MacKnight C, Bergman H, Hogan DB, McDowell I, Mitnitski A. A global clinical measure of fitness and frailty in elderly people. CMAJ. 2005;173(5):489–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Morley JE, Malmstrom TK, Miller DK. A simple frailty questionnaire (FRAIL) predicts outcomes in middle aged African Americans. J Nutr Health Aging. 2012;16(7):601–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Katz S. Assessing self-maintenance: activities of daily living, mobility, and instrumental activities of daily living. J Am Geriatr Soc. 1983;31(12):721–7. [DOI] [PubMed] [Google Scholar]
- 14.Shelkey M, Wallace M. Katz Index of Independence in Activities of Daily Living (ADL). Director. 2000;8(2):72–3. [PubMed] [Google Scholar]
- 15.Sterr F, Reintke M, Bauernfeind L, Senyol V, Rester C, Metzing S, Palm R. Predictors of weaning failure in ventilated intensive care patients: a systematic evidence map. Crit Care. 2024;28(1):366. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Jaber S, De Jong A. Weaning from mechanical ventilation in intensive care units: a call for new international consensus guidelines. Lancet Respir Med. 2023;11(5):398–400. [DOI] [PubMed] [Google Scholar]
- 17.Baptistella AR, Sarmento FJ, da Silva KR, Baptistella SF, Taglietti M, Zuquello RA, Nunes Filho JR. Predictive factors of weaning from mechanical ventilation and extubation outcome: A systematic review. J Crit Care. 2018;48:56–62. [DOI] [PubMed] [Google Scholar]
- 18.Suraseranivong R, Krairit O, Theerawit P, Sutherasan Y. Association between age-related factors and extubation failure in elderly patients. PLoS ONE. 2018;13(11):e0207628. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Stieff KV, Lim F, Chen L. Factors Influencing Weaning Older Adults From Mechanical Ventilation: An Integrative Review. Crit Care Nurs Q. 2017;40(2):165–77. [DOI] [PubMed] [Google Scholar]
- 20.Ali SH, Ezzelregal HG. Frailty can predict the outcome of weaning from mechanical ventilation. Egypt J Chest Dis Tuberculosis. 2021;70(3):419–26. [Google Scholar]
- 21.Michels-Zetsche JD, Roser E, Ersoz H, Neetz B, Dahlhoff JC, Joves B, Trinkmann F, Hoger P, Kontogianni K, Weissmann C et al. Influence of age on outcomes in prolonged weaning from mechanical ventilation. BMJ Open Respir Res. 2025;12(1):e002730. [DOI] [PMC free article] [PubMed]
- 22.Zhao F, Wang M, Zhou Q, Du Y, Cheng Q, Sun X, Zhang J, Liang Y, Shen N, Sun Y. Analysis of risk factors for weaning failure from mechanical ventilation in critically ill older patients with coronavirus disease 2019. Heliyon. 2024;10(12):e32835. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Ye L, Yuan X, Li Y. Risk factors and predictive model for weaning failure in elderly patients with chronic obstructive pulmonary disease and type II respiratory failure. Am J Transl Res. 2025;17(7):5485–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Su KC, Tsai CC, Chou KT, Lu CC, Liu YY, Chen CS, Wu YC, Lee YC, Perng DW. Spontaneous breathing trial needs to be prolonged in critically ill and older patients requiring mechanical ventilation. J Crit Care. 2012;27(3):e324321–327. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.