Advances in intensive care have improved survival rates; however, many survivors experience new physical and mental health impairments, affecting up to half of patients with prolonged ICU stays.1 This shift has redirected research in critical care toward long-term functional and mental health outcomes.2 Among the most common mental health problems in ICU survivors are anxiety, depression and post-traumatic stress disorder (PTSD).3 Factors such as depth and duration of sedation, length of mechanical ventilation and hospital stay are associated with the development of these conditions.4 From a physical perspective, patients requiring mechanical ventilation for acute respiratory distress syndrome (ARDS) remain at increased risk of functional dependence for up to five years after discharge.5

Although the association between the duration of mechanical ventilation and adverse outcomes is well established, the impact of weaning process—particularly its difficulty and duration—on survivors’ physical and mental health is less well understood. Most prior studies have focused on short-term outcomes, leaving a gap in knowledge regarding medium- and long-term effects. Our group has previously shown that physical decline is associated with the development of mental health impairments in ICU survivors,6 yet the specific relationship between difficult or prolonged weaning and these disabilities remains underexplored. Thus, this study aims to address this gap by examining the association between weaning difficulty and duration and the occurrence of physical and mental health disabilities in survivors of critical illness.

Of the 1616 patients included in the Post-ICU Quality of Life Project, 804 (49.8%) were included in the present analysis (Fig. 1). Baseline characteristics are shown in Table 1. The median age was 64 years, 43.9% (n = 353) were female, and 56.0% (n = 450) had a Charlson Comorbidity Index of two or higher. A prior history of anxiety and depression was reported in 18.5% (n = 148) and 19.7% (n = 157) of patients, respectively. During the acute phase of critical illness, 78.2% (n = 629) received vasopressors, 42.2% (n = 339) had a diagnosis of sepsis, and 32.1% (n = 258) had delirium.

Figure 1.

Study Flowchart.

* Patient was not assessed or found in the ward after five attempts (screening failures).

# No available telephone contact, or absence of a proxy for patients with communication difficulties, or previous inclusion in the study.

Regarding ventilatory weaning characteristics, 61.2% (n = 492) underwent simple weaning, 7.5% (n = 60) difficult weaning and 31.3% (n = 252) prolonged weaning. Compared to the other weaning groups, the simple weaning group had a higher median age, a greater proportion of patients aged over 65 years, higher household income per capita, and a greater frequency of elective surgical admissions. Baseline health status prior to ICU admission - including comorbidities, dementia, depression, and anxiety- was similar across the three weaning groups. The simple weaning group showed lower severity of critical illness across several parameters. Sepsis was less frequent compared with the difficult and prolonged weaning groups (35.4% vs. 53.3% vs. 52.8%), as was the need for renal replacement therapy (9.8% vs. 21.7% vs. 25.8%) and delirium (26.0% vs. 38.3% vs. 42.5%). ARDS was uncommon overall, but occurred less often in the simple weaning group (2.8%) compared with the difficult (13.3%) and prolonged (17.1%) groups. The incidence of ICU-acquired infection differed significantly across all groups, being highest in prolonged weaning (58.3%), followed by difficult (16.7%) and simple weaning (7.3%). Vasopressor use was common in all groups, but more frequent in prolonged weaning (84.9%) than in simple (75.6%) or difficult (71.7%) weaning. Overall, 670 (83.3%), 630 (78.3%), and 573 (71.3%) of patients were alive at 3, 6, and 12 months, respectively.

Outcomes

The occurrence of outcomes by ventilatory weaning type is shown in Table 2 and in Fig. 2 and Fig. 3. The proportion of respondents for Barthel index, HADS and IES-6 were 70.1%, 33.8%, and 34.2%, respectively. At 3 months post-ICU discharge, patients in the prolonged weaning group had lower median Barthel index scores (67.5 vs. 90.0 vs. 85.0; p < 0.001) and a higher occurrence of moderate to severe functional dependence (55.9% vs. 37.6% vs. 44.2%; p < 0.001) compared with those in the simple and difficult weaning groups, respectively. At 6 months, the difficult weaning group had higher median HADS scores for anxiety (8.5 vs. 4.0 vs. 4.5; p = 0.01) and higher occurrences of anxiety (55% vs. 23.6% vs. 21.4%; p = 0.02) and depression (50% vs. 24.7% vs. 17.1%; p = 0.04) compared to the simple weaning and prolonged weaning groups, respectively. There was no statistically significant difference between the ventilator weaning groups in HADS depression scores, IES-6 scores and in the occurrence of PTSD. Early mortality was significantly higher in the prolonged weaning group compared with the simple weaning group (13.1% vs. 7.1%, p = 0.01), with no significant difference between the difficult and prolonged weaning groups. Late mortality did not differ significantly among the three groups.

Table 2.

Long-term outcomes after ICU discharge.

Outcome	Total (n = 804)	Simple Weaning (n = 492)	Difficult Weaning (n = 60)	Prolonged Weaning (n = 252)	Adjusted p-value
Physical dependence1 at 3 months
Respondents – no. (%)	564 (70.1)	351 (71.3)	43 (71.7)	170 (67.5)	0.50
Barthel Index score – median (IQR)	85 (30−95)	90 (50−100)A	85 (40−95)AB	67.5 (20−90)B	<0.001
Moderate to severe functional dependence – no (%)	246 (43.6)	132 (37.6)A	19 (44.2)AB	95 (55.9)B	<0.001
Anxiety2 and depression3 at 6 months
Respondents – no. (%)	272 (33.8)	182 (37.0)	20 (33.3)	70 (27.8)	0.104
Anxiety symptoms – median (IQR)	5 (2−8)	4 (2−7)A	8.5 (3−11.5)B	4.5 (2−7)A	0.01
Anxiety symptoms (HADSa >7) – no (%)	69 (25.4)	43 (23.6)A	11 (55.0)B	15 (21.4)A	0.02
Symptoms of depression – median (IQR)	5 (1.8−7)	5 (2−7)	7.5 (3.8−10)	3 (1−6)	0.14
Symptoms of depression (HADSd >7) – no (%)	67/272 (24.6)	45/182 (24.7)A	10 (50.0)B	12 (17.1)A	0.04
Post-traumatic stress disorder2 at 6 months
Respondents – no (%)	275 (34.2)	184 (37.4)	20 (33.3)	71 (28.2)	0.43
IES-6 score – median (IQR)	4 (1−8)	4 (1−8)	6 (4.5−13.2)	4 (1−8)	0.08
IES-6 score > 9 – no (%)	50 (18.2)	35 (19.0)	6 (30.0)	9 (12.7)	0.29
Post-ICU mortality
Early (0–30 days after discharge) – no. (%)	72/804 (9.0)	35/492 (7.1)A	4/60 (6.7)AB	33/252 (13.1)B	0.01
Late (>30 days after discharge) – no. (%)	159/725 (21.9)	99/452 (21.9)	10/56 (17.9)	50/217 (23.0)	0.41

IES-6, impact event scale-6; IQR, interquartile range; HADS, Hospital Scale of Anxiety and Depression Symptoms; ICU, Intensive Care Unit.

Groups sharing the same superscript letter (A, B, C, etc.) do not differ significantly from each other in post hoc pairwise comparisons. Groups with different letters differ significantly at p < 0.05.

1

Analyses adjusted by age, gender, Charlson comorbidity index, moderate to severe functional dependence 1 month before ICU admission, and risk of death at ICU admission.

2

Analyses adjusted by age, gender, Charlson comorbidity index, moderate to severe functional dependence 1 month before ICU admission, risk of death at ICU admission, and history of anxiety.

3

Analyses adjusted by age, gender, Charlson comorbidity index, moderate to severe functional dependence 1 month before ICU admission, risk of death at ICU admission, and history of depression.

Figure 2.

Prevalence of physical dependence at 3 months, anxiety, depression and post-traumatic stress disorder (PTSD) at 6 months for each type of weaning.

Figure 3.

Survival curves according to type of weaning.

The main findings of our study indicate that prolonged weaning was associated with early post-ICU mortality and higher prevalence of moderate to severe physical dependence at 3 months, whereas difficult weaning was associated with a higher prevalence of anxiety and depression at 6 months. While associations persisted after adjustment for age, sex, comorbidity burden, prior functional status, and risk of death at ICU admission, residual confounding remains possible.

Within 30 days of ICU discharge, approximately 13% of patients in the prolonged weaning group had died—nearly twice the rate observed in the simple weaning group. At 3 months, more than half of these patients remained moderately to severely dependent in activities of daily living, with median Barthel Index scores markedly lower than those in the other groups. These results align with prior evidence linking prolonged ventilation and greater illness severity to post-ICU mortality and disability,17–21 but they also suggest that the weaning phase itself may be a critical determinant. The excess early mortality in the prolonged weaning group likely reflects a combination of greater baseline severity, higher rates of ICU-acquired complications, and sustained physiological stress during liberation from mechanical ventilation. Notably, late mortality did not differ significantly among groups, indicating that the vulnerability associated with prolonged weaning was concentrated in the immediate post-ICU period.

The prevalence rates of anxiety, depression and PTSD were similar to those reported in prior studies,22–24 which have described anxiety in 24–28% of patients, depression in 16–23%, and PTSD in approximately 16% at 6 months post-ICU discharge. In a systematic review and meta-analysis, Rabiee et al25 also reported suprathreshold in 29–30% of patients across studies. Notably, our study identified a significantly higher burden of anxiety and depression symptoms in the difficult weaning group compared with the other groups. Potential explanations for the finding of a higher occurrence of symptoms of anxiety and depression in the difficult weaning group compared to the prolonged weaning group include: (1) survival bias, as patients with prolonged weaning and more severe mental health impairment may have died before follow-up or been unable to respond; (2) a higher baseline prevalence of potential risk factors for mental health impairment in the difficult weaning group, such as lower household income and, although not statistically significant, a higher proportion of patients with prior history of depression. Accordingly, Davydow et al.26 found rates of psychiatric history of 11−38% in patients prior to ICU admission, and Rabiee et al.25 reviewed 15 studies reporting the prevalence of previous psychiatric problems before hospital admission with rates of 10–54%, and concluded that ICU survivors with comorbid psychopathology before and during hospitalization had a higher prevalence of depressive symptoms after discharge. In interpreting these mental health outcomes, we acknowledge that differences in delirium, ICU-acquired infection, ARDS, and baseline risk of death may partly account for observed associations despite multivariable adjustment.

Our study has several limitations. First, approximately two-thirds of eligible survivors did not provide HADS or IES-6 responses at 6 months. Although non-response was relatively balanced among groups, attrition could bias mental health estimates in either direction (e.g., sicker patients being unable to respond or healthier patients being less motivated); therefore, these results should be interpreted with caution. Second, because weaning strategies followed local protocols instead of a standardized study protocol, unmeasured center-level variation may have introduced heterogeneity in outcomes. Third, we did not assess mental health symptoms at 3 months, which could have provided a more comprehensive understanding of the early psychological impact of prolonged weaning. Similarly, functional status at 6 months was not evaluated using the Barthel Index, limiting our ability to track long-term physical recovery. Fourth, we did not collect SOFA scores, a validated and widely used measure of organ dysfunction. Instead, we relied on clinical proxies, which may not fully capture the severity or extent of organ failure. Fifth, we were unable to perform robust subgroup analyses comparing surgical vs. non-surgical patients or patients with and without ARDS, as the small number of patients in different weaning categories within each subgroup would likely preclude accurate statistical analyses and increase the risk of type II error. Finally, the study’s eligibility criteria excluded a large proportion of patients (e.g., those with an ICU stay of less than 3 days), which limits the generalizability of findings to less severely ill populations. However, our findings are noteworthy given the scarcity of evidence on how weaning duration influences physical dependence, neuropsychiatric disorders, and mortality in ICU survivors.

ICU survivors who experienced prolonged weaning had higher early post-ICU mortality and greater rates of moderate to severe physical functional dependence at three months after discharge, whereas those with difficult weaning showed higher rates of anxiety and depression at six months. Although these associations may be influenced by survival bias, particularly for mental health outcomes in the difficult weaning group, these results highlight the potential role of standardized weaning protocols in reducing the proportion of difficult and prolonged weaning cases, which could, in turn, lessen their impact on long-term functional dependence and neuropsychiatric outcomes. Future research is still needed to correlate other variables with the presence of these neuropsychiatric disorders and to establish with greater certainty the risk factors and their impact on the outcome of ICU survivors, so that preventive interventions can be applied an also provide these patients with the necessary follow-up after discharge.

Bruna B. Valer: Literature search, Study design, Analysis of data, and Manuscript preparation.

Cassiano Teixeira: Data collection, Study design, and Manuscript preparation.

Regis Rosa: Literature search, Data collection, Study design, and Review of manuscript

Daniel Sgarzela: Analysis of data.

Gilberto Friedman: Study design, Analysis of data, Manuscript preparation and Review of manuscript.

We declare non-use of AI.

Brazilian Ministry of Health and Hospital Moinhos de Vento.