Independent risk factors and prediction model for tracheostomy in ICU patients: a Firth's penalized logistic regression analysis

Background Tracheostomy is critical airway management in intensive care unit (ICU) patients. However, the independent risk factors and prediction models remain unclear. Methods This retrospective multicenter cohort study with single-center external validation analyzed adult ICU patients from the eICU Collaborative Research Database (eICU-CRD) as the derivation cohort and 186 patients from Tianjin First Central Hospital as validation cohort. Firth's penalized logistic regression was adopted to screen independent risk factors for tracheostomy and construct the prediction model. Multicollinearity among candidate variables was evaluated using the variance inflation factor. Model performance was comprehensively evaluated using receiver operating characteristic (ROC) curve analysis, calibration curves, and the Hosmer-Lemeshow goodness-of-fit test. In addition, the association between tracheostomy timing and in-hospital mortality was further explored through adjusted logistic regression models. Results A total of 12,926 patients were included in the derivation cohort [median age 65.00 years (IQR: 54.00–76.00); 56.74% male], of whom 200 (1.5%) underwent tracheostomy. Multivariate Firth's penalized logistic regression identified nine variables independently associated with tracheostomy: ARDS (OR = 3.331, 95% CI: 2.411–4.567), AKI (OR = 2.455, 95% CI: 1.765–3.379), sepsis (OR = 2.393, 95% CI: 1.728–3.284), heart failure (OR = 2.364, 95% CI: 1.625–3.374), COPD (OR = 1.891, 95% CI: 1.362–2.598), serum creatinine (OR = 1.817, 95% CI: 1.708–1.924), higher OASIS score (OR = 1.032, 95% CI: 1.012–1.054), higher platelet count (OR = 1.002, 95% CI: 1.0008–1.0032), and older age (OR = 1.57 per 10 years, 95% CI: 1.56–1.60). The prediction model achieved an apparent AUC of 0.881 (95% CI: 0.865–0.896) in the derivation cohort, with an optimism-corrected AUC of 0.872 from 1,000 bootstrap resamples and an AUC of 0.73 (95% CI: 0.65–0.81) in the external validation cohort (27 tracheostomy events among 186 patients, 14.52%). Further observational analysis showed that late tracheostomy (>7 days) was associated with higher in-hospital mortality compared with early tracheostomy (OR = 2.77, 95% CI: 1.09–7.07, P = 0.03). Conclusion This study identified nine variables independently associated with tracheostomy in ICU patients and developed a prediction model with acceptable discrimination and preliminary evidence of external validity. Late tracheostomy was observationally associated with higher in-hospital mortality, although residual confounding and time-related biases preclude causal inference.

Authors

Institutions

Publication Details

Journal
Frontiers in Medicine
Published
2026-11-19
DOI
https://doi.org/10.3389/fmed.2026.1932723
Primary Topic
Tracheal and airway disorders
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Independent risk factors and prediction model for tracheostomy in ICU patients: a Firth's penalized logistic regression analysis

Pan Yang, Weidong Zhang
Frontiers in Medicine
Tracheal and airway disorders
article

Independent risk factors and prediction model for tracheostomy in ICU patients: a Firth's penalized logistic regression analysis

Pan Yang, Weidong Zhang
article en

Abstract

Background Tracheostomy is critical airway management in intensive care unit (ICU) patients. However, the independent risk factors and prediction models remain unclear. Methods This retrospective multicenter cohort study with single-center external validation analyzed adult ICU patients from the eICU Collaborative Research Database (eICU-CRD) as the derivation cohort and 186 patients from Tianjin First Central Hospital as validation cohort. Firth's penalized logistic regression was adopted to screen independent risk factors for tracheostomy and construct the prediction model. Multicollinearity among candidate variables was evaluated using the variance inflation factor. Model performance was comprehensively evaluated using receiver operating characteristic (ROC) curve analysis, calibration curves, and the Hosmer-Lemeshow goodness-of-fit test. In addition, the association between tracheostomy timing and in-hospital mortality was further explored through adjusted logistic regression models. Results A total of 12,926 patients were included in the derivation cohort [median age 65.00 years (IQR: 54.00–76.00); 56.74% male], of whom 200 (1.5%) underwent tracheostomy. Multivariate Firth's penalized logistic regression identified nine variables independently associated with tracheostomy: ARDS (OR = 3.331, 95% CI: 2.411–4.567), AKI (OR = 2.455, 95% CI: 1.765–3.379), sepsis (OR = 2.393, 95% CI: 1.728–3.284), heart failure (OR = 2.364, 95% CI: 1.625–3.374), COPD (OR = 1.891, 95% CI: 1.362–2.598), serum creatinine (OR = 1.817, 95% CI: 1.708–1.924), higher OASIS score (OR = 1.032, 95% CI: 1.012–1.054), higher platelet count (OR = 1.002, 95% CI: 1.0008–1.0032), and older age (OR = 1.57 per 10 years, 95% CI: 1.56–1.60). The prediction model achieved an apparent AUC of 0.881 (95% CI: 0.865–0.896) in the derivation cohort, with an optimism-corrected AUC of 0.872 from 1,000 bootstrap resamples and an AUC of 0.73 (95% CI: 0.65–0.81) in the external validation cohort (27 tracheostomy events among 186 patients, 14.52%). Further observational analysis showed that late tracheostomy (>7 days) was associated with higher in-hospital mortality compared with early tracheostomy (OR = 2.77, 95% CI: 1.09–7.07, P = 0.03). Conclusion This study identified nine variables independently associated with tracheostomy in ICU patients and developed a prediction model with acceptable discrimination and preliminary evidence of external validity. Late tracheostomy was observationally associated with higher in-hospital mortality, although residual confounding and time-related biases preclude causal inference.

Frontiers in MedicineVol. 13
Nankai University (CN), Tianjin First Center Hospital (CN)
Tracheal and airway disorders
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon