Tracheostomy is generally considered when prolonged invasive ventilation is anticipated, weaning is difficult, airway protection is impaired, or ongoing pulmonary toilet is required. Compared with prolonged translaryngeal intubation, tracheostomy may permit lighter sedation, facilitate oral care and phonation with an appropriate speaking valve, and support mobilization. The timing of tracheostomy is clinically important, but patients considered for the procedure are heterogeneous and the duration of future ventilation is difficult to predict. Randomized evidence has not demonstrated a consistent survival advantage for routine early tracheostomy. In the TracMan trial, assignment to tracheostomy within 4 days did not improve 30-day survival, and only 44.9% of patients assigned to the deferred strategy ultimately underwent tracheostomy (1). Similarly, another multicenter randomized trial found that early tracheotomy at 6-8 days did not significantly reduce ventilator-associated pneumonia compared with tracheotomy at 13-

15 days (2). Meta-analyses have suggested that earlier tracheostomy may reduce the duration of mechanical ventilation or ICU length of stay in some settings, but certainty is limited by clinical heterogeneity, crossover, and difficulty identifying patients who will require prolonged ventilation (3,4). Short-term utilization outcomes do not necessarily reflect recovery outcomes that matter to patients. Survivors of ARDS and prolonged mechanical ventilation may experience weakness, reduced mobility, impaired activities of daily living, diminished quality of life, and delayed return to work for months or years after discharge (5-8). Bed rest, delirium, premorbid frailty, illness severity, and ICU-acquired neuromuscular dysfunction contribute to this trajectory (7,9,10). Tracheostomy timing could influence long-term function through sedation exposure, delirium, communication, swallowing, mobilization, and rehabilitation pathways; alternatively, timing may primarily reflect prognosis, neurologic injury, institutional practice, or clinician expectations. Long-term functional evidence specifically related to tracheostomy timing remains limited. In the SETPOINT2 randomized trial of patients with severe stroke, an early strategy did not significantly improve survival without severe disability at 6 months, although the confidence interval did not exclude a clinically important effect (11). Neurologic cohort data also indicate that decannulation status is strongly associated with subsequent function, but these data do not isolate the effect of timing (12). Accordingly, the association between tracheostomy timing and post-ICU disability in a broader critically ill population remains uncertain. We therefore assessed adults who survived hospitalization after undergoing tracheostomy during critical illness. We hypothesized that tracheostomy within 7 days of intubation would be associated with a lower risk of new disability 6 months after hospital discharge than later tracheostomy. The study was designed to emphasize patient-centered functional outcomes, estimate associations rather than causation, and address measured confounding and missing follow-up using prespecified analyses

Study design and setting This illustrative retrospective cohort was designed as a single-center study at ICU, LRH, MTI. Adult ICU admissions during 26 June 2024 to 26 June 2026 were considered for eligibility. For manuscript structure only, institutional approval and a waiver of informed consent for a de-identified retrospective review are assumed; the final manuscript must replace this statement with the actual ethics approval information from the institution. Reporting was guided by the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement (13). Participants Eligible patients were aged 18 years or older, underwent invasive ventilation through an endotracheal tube, received a tracheostomy during the index hospitalization, and survived to hospital discharge. For patients with more than one eligible admission, only the first admission was included. Exclusion criteria were a pre-existing tracheostomy; tracheostomy performed for an immediately reversible upper-airway procedure without prolonged critical illness; inability to determine the primary outcome; a timing decision made primarily for comfort-focused rather than disease-directed treatment; or incomplete records of tracheostomy timing. Missing covariates were addressed using multiple imputation, and missing outcomes were addressed using inverse-probability-of-observation weighting. Exposure Exposure was defined as the elapsed time from initial endotracheal intubation to tracheostomy. Early tracheostomy was defined as placement within 7 days of intubation, and late tracheostomy as placement after day 7. The 7-day threshold was selected a priori because it is clinically interpretable and lies within the 4-10-day early-tracheostomy windows used in previous trials. Timing was also modeled as a continuous variable using restricted cubic splines and evaluated using alternative cut points of 5 and 10 days. The procedural technique was not used to define exposure. Outcomes The primary outcome was new disability 6 months after hospital discharge, defined as an increase of at least 10 points in the 12-item WHODAS 2.0 score relative to pre-illness function. WHODAS 2.0 scores range from 0 (no disability) to 100 (maximum disability) (14). The instrument was administered by telephone to the patient or, when necessary, a proxy, by trained assessors who were blinded to exposure status. Secondary outcomes included the continuous WHODAS 2.0 score, Barthel Index, independent ambulation, residence at home, return to work among those employed before illness, EQ-5D-5L utility, tracheostomy decannulation by 6 months, and the composite of death or new disability. The Barthel Index was scored on the conventional 0-100 scale, with higher scores indicating greater independence (15). Covariates and data sources Age, sex, body mass index, comorbidities, baseline frailty, pre-illness function, admission source, medical versus surgical admission, neurologic diagnosis, illness-severity scores, organ support, and indication for tracheostomy were obtained from electronic health records and follow-up records. Primary adjustment variables were selected a priori and restricted to factors known at or near the time of intubation: age, sex, pre-illness WHODAS 2.0 score, Clinical Frailty Scale, Charlson Comorbidity Index, admission category, neurologic diagnosis, APACHE II score, day-1 SOFA score, and study year. Sedation exposure, delirium, corticosteroid use, neuromuscular blockade, duration of mechanical ventilation, and ICU length of stay were considered potential mediators or post-exposure variables and were therefore excluded from the primary adjustment model. Statisticalanalysis Continuous variables were summarized as mean (standard deviation) or median (interquartile range), as appropriate, and categorical variables as count (percentage). Covariate balance was assessed using standardized mean differences. The primary association was estimated using modified Poisson regression with robust variance and reported as an adjusted risk ratio (aRR) with a 95% confidence interval. A propensity score for early tracheostomy was estimated using prespecified covariates; stabilized inverse-probability weights were truncated at the 1st and 99th percentiles. An absolute standardized mean difference of less than 0.10 after weighting was considered acceptable balance. Missing covariates were imputed using chained equations across 20 imputed datasets. No data-dependent variable selection was performed. Sensitivity analyses included: (1) redefining early tracheostomy using 5-day and 10-day thresholds; (2) modeling timing continuously; (3) propensity-score overlap weighting; (4) conventional multivariable adjustment; (5) excluding patients with primary neurologic injury and, separately, patients admitted with coronavirus disease 2019; (6) evaluating the composite of death or new disability in the full eligible tracheostomy cohort; and (7) complete-case analysis. Exploratory mediation models evaluated whether delirium duration, timing of mobilization, or duration of mechanical ventilation might account for part of the observed association; these analyses were not intended to support causal claims. All tests were two-sided with alpha=.05. The illustrative analyses were specified in R version 4.3.1; executable code was not provided with the source manuscript.

Cohort derivation and follow-up

A total of 1,146 mechanically ventilated adults were evaluated, as shown in Figure 1. Of these, 894 were excluded: 713 did not undergo tracheostomy, 129 died before hospital discharge, 18 had a pre-existing tracheostomy, 11 had missing timing data, and 23 were excluded for other prespecified reasons. The analytic cohort therefore included 252 hospital survivors, of whom 118 (46.8%) underwent tracheostomy within 7 days and 134 (53.2%) after day 7. The median time from intubation to tracheostomy was 9 days (IQR, 6-13) overall, 5 days (IQR, 4-6) in the early group, and 12 days (IQR, 10-16) in the late group. Primary outcome data were available for 229 of 252 participants (90.9%) at 6 months: 108 in the early group and 121 in the late group. Participants without outcome data had greater baseline frailty than those with follow-up; the largest standardized difference was observed for the Clinical Frailty Scale (0.24). Proxy responses were used for 63 assessments (27.5%). Median follow-up was 183 days (IQR, 176-195) after hospital discharge. Median index-hospital length of stay was 34 days (IQR, 24-51); 78 participants (31.0%) were discharged directly home, 92 (36.5%) to inpatient rehabilitation, and 82 (32.5%) to a long-term acute-care or skilled-nursing facility. At discharge, 86 participants (34.1%) remained cannulated and 31 (12.3%) required ongoing ventilatory support.

Figure 1: Cohort selection and 6-month follow-up

Note: Counts and effect estimates in this figure are simulated for drafting and require verification against the source dataset.

Baseline and clinical characteristics

Baseline characteristics are shown in Table 1. The cohort had a median age of 60 years (IQR, 48-69); 107 participants (42.5%) were women; and the median day-1 APACHE II score was 21 (IQR, 16-26). The main admission categories were medical non-neurologic (45%), neurologic (28%), and surgical (27%). Before weighting, early-tracheostomy recipients had lower illness severity, less baseline frailty, and fewer neurologic diagnoses. The largest absolute standardized differences involved the Clinical Frailty Scale, APACHE II score, and Charlson Comorbidity Index. After weighting, all prespecified baseline covariates had absolute standardized mean differences below 0.10.

Table 1: Baseline characteristics by tracheostomy timing

Abbreviations: APACHE, Acute Physiology and Chronic Health Evaluation; COVID-19, coronavirus disease 2019; IQR, interquartile range; SOFA, Sequential Organ Failure Assessment; WHODAS 2.0, World Health Organization Disability Assessment Schedule 2.0.

Primary outcome

New disability at 6 months occurred in 43 of 108 early-tracheostomy recipients (39.8%) and 65 of 121 late recipients (53.7%), corresponding to an unadjusted risk ratio of 0.74 (95% CI, 0.56-0.98). In the prespecified adjusted model, early tracheostomy was associated with a lower risk of new disability (aRR, 0.75; 95% CI, 0.58-0.97; P=.029) (Table 2). The adjusted absolute risk difference was -13.2 percentage points (95% CI, -25.4 to -1.0). Propensity weighting yielded an effective sample size of 221 and a weighted risk ratio of 0.76 (95% CI, 0.59-0.98). The primary model included age, sex, pre-illness WHODAS 2.0 score, Clinical Frailty Scale, Charlson Comorbidity Index, admission category, neurologic diagnosis, APACHE II score, day-1 SOFA score, and study year. Effect estimates compare early (<=7 days) with late (>7 days) tracheostomy.

Table 2: Association between early tracheostomy and new disability at 6 months

Primary model covariates: age, sex, pre-illness WHODAS 2.0, Clinical Frailty Scale, Charlson Comorbidity Index, admission category, neurologic diagnosis, APACHE II score, day-1 SOFA score, and study year. Effect estimates compare early (<=7 days) with late (>7 days) tracheostomy.

Secondary outcomes

At 6 months, median WHODAS 2.0 scores were 25.0 (IQR, 12.5-45.8) in the early group and 37.5 (IQR, 20.8-58.3) in the late group; the adjusted mean difference was -8.6 points (95% CI, -15.9 to -1.3). Functional independence was observed in 60.2% of the early group versus 45.5% of the late group (aRR, 1.29; 95% CI, 1.02-1.64). Table 3 summarizes residence at home, return to work, EQ-5D-5L utility, decannulation, and other secondary outcomes. Seventeen participants (6.7%) died after hospital discharge and before the 6-month assessment. In the illustrative full eligible tracheostomy cohort, the composite of death or new disability occurred in 53.5% of early recipients and 63.2% of late recipients (aRR, 0.84; 95% CI, 0.71-0.99). The denominators for this composite outcome must be verified against the source dataset before publication.

Table 3: Six-month secondary outcomes

Note: The denominators for the composite outcome of death or new disability must be verified against the source dataset before publication.

Sensitivity and exploratory analyses

Results were directionally consistent when early tracheostomy was defined as <=5 days (aRR, 0.78; 95% CI, 0.60-1.01) or <=10 days (aRR, 0.81; 95% CI, 0.64-1.03) (Table 4). No clear evidence of nonlinearity was identified in the continuous timing model (P=.31). Estimates from overlap weighting, landmark analysis, complete-case analysis, and models excluding neurologic or COVID-19 admissions were similar in direction and magnitude, although several confidence intervals included the null. There was no statistically significant evidence of interaction by neurologic diagnosis, baseline frailty, age, or admission category; these analyses were interpreted cautiously because of limited power. Exploratory models did not clearly identify the proportion of the association explained by delirium duration, mobilization timing, or ventilation duration. Model diagnostics were within acceptable limits and no influential observations were identified. Propensity-score overlap was adequate; seven observations had weights greater than 10, and truncation changed the primary estimate from 0.73 to 0.76. Inverse-probability-of-observation weighting produced an adjusted effect estimate of 0.77 (95% CI, 0.59-1.01) in the missing-outcome analysis. A prespecified tipping-point analysis suggested that an outcome-rate difference of at least 18 percentage points among participants lost to follow-up would be required to overturn the main conclusion.

Table 4: Prespecified sensitivity analyses

Among adult hospital survivors who underwent tracheostomy, placement within 7 days of intubation was associated with a lower risk of new disability at 6 months than later placement. The adjusted risk ratio was 0.75, corresponding to an adjusted absolute risk difference of -13.2 percentage points. Secondary measures, including WHODAS 2.0 and functional independence, showed a similar pattern, and sensitivity analyses were directionally consistent, although some estimates included the null. This study addresses a patient-centered outcome that has been underrepresented in trials of tracheostomy timing. Previous randomized trials have not shown a consistent mortality benefit from an early strategy. In TracMan, early tracheostomy did not improve 30-day survival, and fewer than half of patients assigned to deferred tracheostomy ultimately underwent the procedure (1). Terragni and colleagues likewise found no significant reduction in ventilator-associated pneumonia with early versus later tracheotomy (2). Some meta-analyses have reported reductions in ventilation duration, ICU length of stay, or pneumonia, but no consistent mortality benefit has emerged, and study populations and definitions of timing have varied (3,4). Because all participants in the present analytic cohort ultimately underwent tracheostomy and survived to hospital discharge, these short-term efficacy questions are not directly addressed here. Instead, the analysis focuses on subsequent functional recovery among a clinically selected survivor population. The most appropriate randomized comparison for long-term function is SETPOINT2. For those who had a severe stroke, there was no significant difference in the number who survived or had severe disability at 6 months with an early vs. a late tracheostomy strategy (11). This illustrative cohort is expanded to include a wider spectrum of admission diagnoses, and to measure change in disability prior to the stroke, not a stroke-specific outcome scale. Any apparent benefit may be due to population differences, residual confounding or care pathways, and not reflect the true effect of timing. Due to the nature of the numerical results presented in this draft being simulated, substantive comparison with SETPOINT2 is not appropriate until the results have been compared against the actual data set. There are several mechanisms that are biologically plausible. Early tracheostomy can minimize exposure to sedatives, aid in communication, secretion management and minimize practical mobility barriers. These pathways may help to minimize delirium, immobility, and ICU-acquired weakness. The longitudinal studies indicate that muscle weakness and physical limitation can last for years following critical illness (5,7,9) and delirium has been linked to increased disability following mechanical ventilation (10). Early tracheostomy does not, however, cure organ failure, neurologic injury, pre-existing frailty, or catabolism, and some patients may be at risk of complications of the procedure or of ongoing care with the tracheostomy. Time alone is not likely to be sufficient for functional recovery, however; intensity of rehabilitation, swallowing management, availability of speaking valves, place of discharge, and caregiver support may also play a role. The principal estimand is the survival rate for hospital patients who had a tracheostomy. The conditioning on survival response to a clinically relevant prognostic question for further counseling, but it is not estimate the overall effect of an early-tracheostomy strategy from the moment of the initiation of ventilation. The prognosis and the changing treatment also affect death before discharge and tracheostomy receipt or timing, which can lead to selection and immortal-time bias. As a result of this, the manuscript refrains from the use of causal language and provides supplementary threshold, landmark, weighting and composite-outcome analyses. If time-varying eligibility, treatment strategies, and confounders are reliably measured, then a target-trial emulation starting prior to the timing decision would result in a more robust causal structure. The study design has a number of strengths: The primary outcome was patient-centered and was specifically prespecified and referenced to pre-illness function; baseline disease factors were separated from potential mediators; the results were presented both as relative and absolute effects; and the results were presented under various definitions of timing. Six-month follow-up is also possible and the concurrent use of WHODAS 2.0, independent ambulation, home residence, and return to work provides clinical interpretability. Several limitations remain. First, the timing of the tracheostomy was not randomized so, although residual confounding by prognosis, local scheduling, staffing, approach to the procedure, goals of care and expected duration of ventilation is unlikely to be present, it cannot be ruled out. Second, the pre-illness disability was reconstructed post hoc, and could be subject to recall or proxy error. Third, incomplete 6-month follow-up introduces the possibility of attrition bias, and weighting and imputation are based on assumptions regarding the nature of the missing data. Fourth, the 7-day cutoff was not a biological threshold but rather a convention used in clinical practice, and continuous and alternative-threshold analyses were performed to help minimize reliance on this one cutoff. Fifth, transportability may be restricted as practices may have changed during COVID-19. Sixth, the study might have been underpowered for interactions and uncommon complications. Lastly, functional measures employed are not comprehensive in capturing communication, swallowing, fatigue, cognition, caregiver burden, and patient preferences. Most importantly, all numerical results which are presented in this illustrative draft shall be replaced or verified with the actual cohort before publication. The timing of tracheostomy should continue to be individualized clinically. With patients and/or surrogates, it is important to separate known short term factors (airway, comfort, sedation needs, transfer) from unproven long term factors (independence). Outcomes should include death, disability compared to baseline, cognition, mental health, communication, swallowing, quality of life, caregiver burden, discharge destination and return to usual roles, which should be measured in future studies in patients prior to the tracheostomy decision, and in patients who have been treated with rehabilitation doses. Reassessment at 3, 6 and 12 months may help to differentiate between delayed recovery and ongoing disability.

Earlier tracheostomy was also associated with a lower risk of new disability at 6 months than later tracheostomy (illustrative aRR, 0.75; 95% CI, 0.58-0.97), which equates to an illustrative adjusted absolute risk difference of -13.2 percentage points per 100 adult hospital survivors who underwent tracheostomy during critical illness. Sensitivities were mostly in the same direction. The timing of tracheostomy is associated with prognosis, clinical evolution and institutional practices and should not be considered as a proof that changing the timing of the tracheostomy will benefit functional recovery. Factors to be taken into account for clinical decisions include: expected duration of ventilation, airway protection, neurologic prognosis, procedural risk, patient preference and access to multidisciplinary rehabilitation. Future prospective studies and target-trial emulations should have a consistent measure of disability, quality of life, communication, swallowing and return to participation, as well as explore potential pathways for how the timing of tracheostomy may impact these outcomes. In this illustrative draft, all numerical results must be checked against the patient-level data set before they can be considered as study results or submitted for publication.

1. Young D, Harrison DA, Cuthbertson BH, Rowan K; TracMan Collaborators. Effect of early vs late tracheostomy placement on survival in patients receiving mechanical ventilation: the TracMan randomized trial. JAMA. 2013;309(20):2121-2129. doi:10.1001/jama.2013.5154.
2. Terragni PP, Antonelli M, Fumagalli R, et al. Early vs late tracheotomy for prevention of pneumonia in mechanically ventilated adult ICU patients: a randomized controlled trial. JAMA. 2010;303(15):1483-1489. doi:10.1001/jama.2010.447.
3. Siempos II, Ntaidou TK, Filippidis FT, Choi AMK. Effect of early versus late or no tracheostomy on mortality and pneumonia of critically ill patients receiving mechanical ventilation: a systematic review and meta-analysis. Lancet Respir Med. 2015;3(2):150-158. doi:10.1016/S2213-2600(15)00007-7.
4. Chorath K, Hoang A, Rajasekaran K, Moreira A. Association of early vs late tracheostomy placement with pneumonia and ventilator days in critically ill patients: a meta-analysis. JAMA Otolaryngol Head Neck Surg. 2021;147(5):450-459. doi:10.1001/jamaoto.2021.0025.
5. Herridge MS, Cheung AM, Tansey CM, et al. One-year outcomes in survivors of the acute respiratory distress syndrome. N Engl J Med. 2003;348(8):683-693. doi:10.1056/NEJMoa022450.
6. Cox CE, Carson SS, Lindquist JH, Olsen MK, Govert JA, Chelluri L; Quality of Life After Mechanical Ventilation in the Aged Study Investigators. Differences in one-year health outcomes and resource utilization by definition of prolonged mechanical ventilation: a prospective cohort study. Crit Care. 2007;11(1):R9. doi:10.1186/cc5667.
7. Fan E, Dowdy DW, Colantuoni E, et al. Physical complications in acute lung injury survivors: a two-year longitudinal prospective study. Crit Care Med. 2014;42(4):849-859. doi:10.1097/CCM.0000000000000040.
8. Chelluri L, Im KA, Belle SH, et al. Long-term mortality and quality of life after prolonged mechanical ventilation. Crit Care Med. 2004;32(1):61-69. doi:10.1097/01.CCM.0000098029.65347.F9.
9. Van Aerde N, Meersseman P, Debaveye Y, et al. Five-year impact of ICU-acquired neuromuscular complications: a prospective, observational study. Intensive Care Med. 2020;46(6):1184-1193. doi:10.1007/s00134-020-05927-5.
10. Brummel NE, Jackson JC, Pandharipande PP, et al. Delirium in the ICU and subsequent long-term disability among survivors of mechanical ventilation. Crit Care Med. 2014;42(2):369-377. doi:10.1097/CCM.0b013e3182a645bd.
11. Bösel J, Niesen WD, Salih F, et al. Effect of early vs standard approach to tracheostomy on functional outcome at 6 months among patients with severe stroke receiving mechanical ventilation: the SETPOINT2 randomized clinical trial. JAMA. 2022;327(19):1899-1909. doi:10.1001/jama.2022.4798.
12. Schneider H, Hertel F, Kuhn M, et al. Decannulation and functional outcome after tracheostomy in patients with severe stroke (DECAST): a prospective observational study. Neurocrit Care. 2017;27(1):26-34. doi:10.1007/s12028-017-0390-y.
13. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet. 2007;370(9596):1453-1457. doi:10.1016/S0140-6736(07)61602-X.
14. Üstün TB, Chatterji S, Kostanjsek N, et al. Developing the World Health Organization Disability Assessment Schedule 2.0. Bull World Health Organ. 2010;88(11):815-823. doi:10.2471/BLT.09.067231.
15. Mahoney FI, Barthel DW. Functional evaluation: the Barthel Index. Md State Med J. 1965;14:61-65.