Abstract
Background
Delayed extubation after anesthesia can lead to adverse clinical outcomes. We aimed to identify the risk factors for delayed extubation in patients undergoing robot-assisted radical prostatectomy and to develop a visualized nomogram prediction model for clinical use.
Methods
A total of 624 patients were enrolled and divided into training group, validation group, and temporal validation group. The training group was utilized to develop a nomogram, whereas the validation group and temporal validation group were used to assess its performance. LASSO regression was employed to refine variables and select predictors, and a nomogram was constructed using multivariate logistic regression. The performance of the model was internally validated using calibration and receiver operating characteristic curves. Additionally, decision curve analysis and clinical impact curves were used to assess the clinical utility of the model.
Results
Patients who underwent robot-assisted radical prostatectomy between January 2022 and April 2024 were included and divided into a training group (n = 389), a validation group (n = 98), and a temporal validation group (n = 137). Logistic regression identified cerebral infarction, pulmonary disease, coronary heart disease, age, and intraoperative hypotension as independent predictors of delayed extubation. A nomogram constructed based on these factors demonstrated excellent predictive performance, with area under the curve values of 0.763 (95% CI: 0.717–0.810) in the training group, 0.811 (95% CI: 0.726–0.897) in the validation group, and 0.769 (95% CI: 0.689–0.848) in the temporal validation group. Across all three groups, the model demonstrated a good fit, as indicated by a non-significant P-value from the Hosmer–Lemeshow test, and the calibration curves indicated a strong alignment between the predictions and actual observations. Furthermore, decision curve analysis and clinical impact curve demonstrated the clinical efficiency and benefits of the prediction model.
Conclusion
This study identified key risk factors for delayed extubation and established an effective predictive nomogram with high discriminative power and clinical applicability for predicting the risk of extubation delay in patients undergoing robot-assisted radical prostatectomy.
Trial registration
The Medical Ethics Committee of Nanjing Drum Tower Hospital granted ethical approval for this research(grant number: 2024–742-01).
Keywords: Anesthesia, Nomogram, Robotic-assisted radical prostatectomy, Prostate cancer, Airway extubation
Introduction
Tracheal extubation at the end of surgery is a crucial step in recovery from general anesthesia. Delayed extubation postoperatively may increase the risk of various respiratory complications, such as laryngeal injury, dysphagia, pneumonia, and glottic stenosis [1–4]. Previous research has substantiated that early removal of tracheal catheters in patients undergoing esophagectomy or cardiac surgery can reduce lung complications and promote hemodynamic stability [5, 6]. Moreover, delayed extubation imposes a heavier workload on medical staff and reduces the turnover rate in the operating room [7]. Therefore, identifying the risk factors associated with delayed extubation and establishing a clinical prediction model are vital. The timely recognition of high-risk individuals and subsequent administration of tailored interventions can effectively reduce the incidence of delayed extubation and prevent perioperative complications.
Prostate cancer significantly affects the male population globally and is the second most prevalent cancer in men [8, 9]. It accounts for 7% of newly diagnosed male cancers worldwide and 15% in developed regions. Furthermore, the annual number of newly diagnosed cases exceeds 1.2 million, with over 350,000 deaths attributable to prostate cancer annually [10, 11]. These findings suggest that prostate cancer stands as a major contributor to cancer-specific deaths in men.
Radical resection is a primary and often gold-standard treatment for prostate cancer. Among radical resection techniques, robot-assisted surgery is an emerging minimally invasive approach. Compared with traditional laparoscopic radical prostatectomy, it effectively reduces operative time, surgical trauma, intraoperative blood loss, and postoperative complications, making it widely used in clinical practice [12, 13]. The incidence of prostate cancer strongly correlates with advancing age, with over 85% of new diagnoses occurring in individuals over 60 years of age [14]. Advanced age is positively correlated with a higher incidence of postoperative extubation delay [15]. Nonetheless, predictive factors specific to delayed extubation following robot-assisted radical prostatectomy remain uninvestigated. Therefore, a retrospective analysis was conducted on the clinical data of patients who underwent robot-assisted radical prostatectomy, with the objectives of identifying risk factors for delayed extubation, developing a predictive nomogram, and internally validating the model to aid in clinical prognostication.
Materials and methods
Study design and patients
This single-center retrospective study was approved by the Medical Ethics Committee of Nanjing Drum Tower Hospital (grant number: 2024–742-01). We retrospectively collected data from 750 patients who underwent robot-assisted radical prostatectomy between January 2022 and April 2024. After applying exclusion criteria, 624 patients were included in the study to investigate factors associated with delayed extubation. A total of 487 patients (January 2022 to October 2023) were randomly divided into a training cohort (n = 389) and an internal validation cohort (n = 98) at a predefined 4:1 ratio. An additional 137 patients (November 2023 to April 2024) were enrolled as the temporal validation cohort (n = 137).
The inclusion criteria for this study were as follows: 1) No severe hearing or visual impairment prior to surgery. 2) No dementia, coma, or disturbance of consciousness prior to surgery. 3) Diagnosed with prostate cancer and underwent robot-assisted radical prostatectomy. 4) Extubated in the operating room. The exclusion criteria were as follows: 1) Incomplete clinical data. 2) Patients who died within 24 h after surgery.The study flow diagram is shown in Fig. 1.
Fig. 1.
Flowchart of patient selection, grouping, and model development
Delayed extubation
Extubation time was defined as the interval from the cessation of anesthetic administration to removal of the tracheal tube. The timing of extubation was comprehensively evaluated by a registered anesthesiologist and Post-anesthetic care unit nurses considering the patient's spontaneous breathing efforts, neurological status, motor function, and other relevant clinical indicators for extubation readiness [16]. Ultimately, the decision to remove or retain the tracheal tube was made by a registered anesthesiologist. Delayed extubation was defined as extubation performed more than 1 h after the cessation of anesthesia. This 1-h cutoff is supported by previous studies [17, 18], which demonstrated that extubation delays exceeding 1 h are associated with an increased risk of postoperative complications. Furthermore, this threshold aligns with the routine clinical practice at our institution. This binary classification not only holds clinical relevance but also facilitates predictive modeling by providing a clear outcome variable.
Data collection
We collected patient data after screening for the inclusion and exclusion criteria. Preoperative examination data and basic patient information included preoperative white blood cell count, neutrophil percentage, glucose, creatinine, serum potassium, serum calcium, and blood pressure; ASA grade; age; body mass index; smoking status; anemia (preoperative Hb < 120 g/L); hypoproteinemia (total serum protein < 60 g/L or albumin < 35 g/L); alanine transaminase levels; and presence of hypertension, diabetes, cerebral infarction, pulmonary disease, and coronary heart disease. Intraoperative data included the volume of crystalloid and colloid infusions, intraoperative blood transfusion, urine volume, intraoperative blood loss, intraoperative glucose levels, intraoperative serum potassium levels, and IOH (Currently, there is no universal definition of IOH, and different studies have used either absolute or relative thresholds [19]. In the present study, IOH was categorized into three levels based on previous literature: MAP 70–80% of baseline [20], MAP < 70% of baseline [21], and MAP < 65 mmHg [22]. This multi-threshold grouping is helpful for exploring the influence of different degrees of IOH on the delay of extubation).
Statistical analysis
Continuous variables were summarized according to their distribution: normally distributed data as mean ± standard deviation and compared using the independent samples t-test, while non-normally distributed data as median with interquartile range and compared using the Mann–Whitney U test. Categorical variables were expressed as percentages (%) and compared with the chi-square test or Fisher’s exact test, as appropriate [23]. All variables were subjected to a LASSO regression analysis for variable selection and risk factor identification. The regularization parameter (λ) for LASSO regression was selected via 20-fold cross-validation within the training cohort. To balance model complexity and predictive accuracy, the optimal λ was determined using the “one-standard-error rule”, which selects the simplest model (smallest number of predictors) whose cross-validation error is within one standard error of the minimum error. Variables selected by LASSO were incorporated into a multivariable logistic regression model, and a nomogram was developed based on the results. We assessed the model's discriminatory power using ROC analysis, with performance quantified by AUC. Internal validation of the model was performed using bootstrapping with 500 replicates to assess overfitting and bias-correct the calibration curves. Simultaneously, the Hosmer–Lemeshow test was utilized to evaluate the model's goodness-of-fit.The clinical utility of the model was analyzed using decision curve DCA and CIC. The data analysis was carried out in R (version 3.5.1; R Foundation for Statistical Computing, Vienna, Austria). P-value below 0.05 was considered to indicate statistical significance.
Results
Demographics of the training group
The study site was a tertiary hospital in China. A total of 750 patients undergoing robot-assisted radical prostatectomy between January 2022 and April 2024 were retrospectively collected. Because this was an observational study, 126 patients with incomplete clinical data were excluded from the group. After applying exclusion criteria, 624 patients were included in the study to investigate factors associated with delayed extubation. A total of 487 patients (January 2022 to October 2023) were randomly assigned to training cohort (n = 389) and validation cohort (n = 98) in a 4:1 ratio. An additional 137 patients (November 2023 to April 2024) were enrolled as the temporal validation cohort (n = 137). Of the training group, 204 (52.6%) experienced delayed extubation after surgery.
Table 1 summarizes the characteristics of the patients who participated in the study, comparing those who experienced delayed extubation with those who did not. Patients with delayed extubation were older than those without delayed extubation (P < 0.001) and had a higher incidence of anemia (P = 0.022), cerebral infarction (P = 0.003), pulmonary disease (P < 0.001), and coronary heart disease (P < 0.001). Simultaneously, the incidence of IOH was significantly higher in patients with delayed extubation than in those without (P = 0.004).
Table 1.
Patients’ characteristics in the training group
| Variable | Total(n = 389) | Delayed extubation | p | |
|---|---|---|---|---|
| No(n = 184) | Yes(n = 205) | |||
| ASA, n (%) | 0.23 | |||
| 2 | 11 (3%) | 8 (4%) | 3 (1%) | |
| 3 | 361 (93%) | 168 (91%) | 193 (94%) | |
| 4 | 17 (4%) | 8 (4%) | 9 (4%) | |
| preoperative white blood cell count (× 109/L) Median (Q1,Q3) | 5.6 (4.6, 6.5) | 5.7 (4.77, 6.8) | 5.5 (4.5, 6.4) | 0.089 |
| preoperative neutrophil percentage (%), Mean ± SD | 59.6 (54.5, 65) | 59.8 (53.8, 64.82) | 59.4 (55.5, 65.1) | 0.774 |
| anemia, n (%) | 0.022 | |||
| No | 343 (88%) | 170 (92%) | 173 (84%) | |
| Yes | 46 (12%) | 14 (8%) | 32 (16%) | |
| hypoproteinemia, n (%) | 0.581 | |||
| No | 331 (85%) | 159 (86%) | 172 (84%) | |
| Yes | 58 (15%) | 25 (14%) | 33 (16%) | |
| Preoperative glucose (mmol/L), Median (Q1,Q3) | 4.78 (4.41, 5.39) | 4.74 (4.45, 5.32) | 4.8 (4.38, 5.56) | 0.945 |
| preoperative creatinine (µmol/L), Median (Q1,Q3) | 70 (62, 78) | 69 (62, 76) | 71 (62.1, 81) | 0.159 |
| Preoperative serum potassium (mmol/L), Median (Q1,Q3) | 3.92 (3.71, 4.15) | 3.92 (3.71, 4.14) | 3.94 (3.71, 4.16) | 0.677 |
| Preoperative serum calcium (mmol/L), Median (Q1,Q3) | 2.34 (2.26, 2.42) | 2.35 (2.29, 2.42) | 2.33 (2.25, 2.43) | 0.129 |
| alanine transaminase (U/L), Median (Q1,Q3) | 17.5 (13, 23.8) | 18.1 (13, 24) | 16.8 (13, 23.7) | 0.714 |
| hypertension, n (%) | 0.034 | |||
| No | 157 (40%) | 85 (46%) | 72 (35%) | |
| Yes | 232 (60%) | 99 (54%) | 133 (65%) | |
| diabetes, n (%) | 0.244 | |||
| No | 315 (81%) | 154 (84%) | 161 (79%) | |
| Yes | 74 (19%) | 30 (16%) | 44 (21%) | |
| cerebral infarction, n (%) | 0.003 | |||
| No | 315 (81%) | 161 (88%) | 154 (75%) | |
| Yes | 74 (19%) | 23 (12%) | 51 (25%) | |
| pulmonary disease, n (%) | < 0.001 | |||
| No | 226 (58%) | 133 (72%) | 93 (45%) | |
| Yes | 163 (42%) | 51 (28%) | 112 (55%) | |
| coronary heart disease, n (%) | < 0.001 | |||
| No | 345 (89%) | 177 (96%) | 168 (82%) | |
| Yes | 44 (11%) | 7 (4%) | 37 (18%) | |
| age, Median (Q1,Q3) | 71 (66, 76) | 69 (63.75, 74) | 73 (68, 78) | < 0.001 |
| BMI (Kg/m2), Median (Q1,Q3) | 24.02 (22.49, 25.71) | 24.3 (22.52, 25.95) | 23.67 (22.49, 25.34) | 0.356 |
| smoking, n (%) | 1 | |||
| No | 274 (70%) | 130 (71%) | 144 (70%) | |
| Yes | 115 (30%) | 54 (29%) | 61 (30%) | |
| crystalloid infusion volume (ml), Median (Q1,Q3) | 1000 (1000, 1500) | 1000 (1000, 1300) | 1000 (1000, 1500) | 0.337 |
| colloids infusion volume (ml), Median (Q1,Q3) | 1000 (500, 1000) | 500 (500, 1000) | 1000 (500, 1000) | 0.051 |
| Intraoperative infusion volume (ml), Median (Q1,Q3) | 2000 (1500, 2000) | 2000 (1500, 2000) | 2000 (1500, 2000) | 0.633 |
| intraoperative blood transfusion (ml), n (%) | 0.063 | |||
| No | 384 (99%) | 184 (100%) | 200 (98%) | |
| Yes | 5 (1%) | 0 (0%) | 5 (2%) | |
| urine volume (ml), Median (Q1,Q3) | 400 (300, 700) | 400 (300, 600) | 400 (250, 800) | 0.74 |
| intraoperative blood loss (ml), Median (Q1,Q3) | 100 (100, 100) | 100 (100, 100) | 100 (100, 100) | 0.855 |
| intraoperative glucose (mmol/L), Median (Q1,Q3) | 5.9 (5.23, 6.73) | 6 (5.2, 6.82) | 5.75 (5.23, 6.7) | 0.473 |
| intraoperative serum potassium (mmol/L), Median (Q1,Q3) | 3.57 (3.37, 3.8) | 3.6 (3.35, 3.8) | 3.55 (3.4, 3.8) | 0.791 |
| intraoperative hypotension, n (%) | 0.004 | |||
| MAP > 80% of baseline | 103 (26%) | 57 (31%) | 46 (22%) | |
| MAP 70%−80% of baseline | 139 (36%) | 70 (38%) | 69 (34%) | |
| MAP < 70% of baseline | 113 (29%) | 50 (27%) | 63 (31%) | |
| MAP < 65 mmHg | 34 (9%) | 7 (4%) | 27 (13%) | |
Model development
This study utilized the LASSO regression model to analyze 27 variables obtained from patients, and 5 variables were identified as significant predictors due to non-zero coefficients. These selected variables were age, cerebral infarction, pulmonary disease, coronary heart disease and IOH (Fig. 2).
Fig. 2.
Feature selection was performed utilizing LASSO logistic regression. A The coefficient profile plot displays the trajectories of the 27 selected variables (B) The optimal tuning parameter (λ) was determined via cross-validation, with the vertical dashed lines representing mean-square error criterion (left dashed line) and the standard error criterion (right dashed line)
The five variables selected from the LASSO regression analysis were used to perform a multivariate logistic regression to build a predictive model for delayed extubation. Multivariate logistic regression model identified the following independent predictors of delayed extubation: age (OR: 1.064, 95% CI: 1.031–1.098, P < 0.001), cerebral infarction (OR: 1.875, 95% CI: 1.035–3.399, P = 0.038), pulmonary disease (OR: 3.213, 95% CI: 2.027–5.092, P < 0.001), coronary heart disease (OR: 4.878, 95% CI: 2.029–11.728, P < 0.001) and IOH (MAP < 65 mmHg) (OR: 3.802, 95% CI: 1.356–10.657, P = 0.011) (Table 2). Based on the multivariate logistic regression, a clinical prediction nomogram was developed (Fig. 3A). Each patient received a points assignment for the presence of each risk factor, with the aggregate score then being transformed into an estimated probability of delayed extubation (Fig. 3B).
Table 2.
Multivariate analysis of the predictive factors in the training group
| Variable | B | OR(95%CI) | P |
|---|---|---|---|
| cerebral infarction | 0.629 | 1.875(1.035–3.399) | 0.038 |
| pulmonary disease | 1.167 | 3.213(2.027–5.092) | < 0.001 |
| coronary heart disease | 1.585 | 4.878(2.029–11.728) | < 0.001 |
| age | 0.062 | 1.064(1.031–1.098) | < 0.001 |
| intraoperative hypotension | |||
| MAP 70%−80% of baseline | 0.198 | 1.219(0.689–2.154) | 0.496 |
| MAP < 70% of baseline | 0.455 | 1.576(0.867–2.863) | 0.135 |
| MAP < 65 mmHg | 1.336 | 3.802(1.356–10.657) | 0.011 |
Fig. 3.
Nomogram used for predicting the risk of delayed extubation in patients undergoing robot-assisted radical prostatic cancer surgery. A Based on multivariable logistic regression results, nomogram was constructed using age, cerebral infarction, pulmonary disease, coronary heart disease and intraoperative hypotension. B Each patient was assigned a point score based on their specific risk factor profile. This cumulative score was then converted into a final probability estimate for delayed extubation
Evaluation and validation of the nomogram model
To evaluate the predictive performance and clinical utility of the model, ROC curve analysis was employed. AUC was 0.763 (95% CI: 0.717–0.810) for the training groups (Fig. 4A), 0.811 (95% CI: 0.726–0.897) for the validation group (Fig. 4B), and 0.769 (95% CI: 0.689–0.848) for the temporal validation group (Fig. 4C). These results indicate that the nomogram model exhibits strong predictive accuracy for delayed extubation. To further validate the clinical utility of the model, key diagnostic metrics were calculated at the optimal cutoff value of 0.5524 (Table 3). In the training, internal validation, and temporal validation cohorts, the sensitivity was 64.88%, 63.27%, and 63.41%, respectively; the specificity was 76.63%, 71.43%, and 67.27%; the PPV was 75.57%, 68.89%, and 74.29%; and the NPV was 66.20%, 66.04%, and 55.22%. These results demonstrate that the model balances the ability to identify high-risk patients and exclude low-risk individuals, providing a reliable reference for clinical decision-making.
Fig. 4.
ROC curves for model evaluation. A-C ROC curves were generated for the training group, validation group, and temporal validation group to assess the model’s predictive performance and clinical utility
Table 3.
Discriminative and calibration performance metrics of the multivariable prediction model
| training group | validation group | temporal validation group | |
|---|---|---|---|
| threshold | 0.5524 | 0.5524 | 0.5524 |
| sensitivity | 64.88% | 63.27% | 63.41% |
| specificity | 76.63% | 71.43% | 67.27% |
| PPV | 75.57% | 68.89% | 74.29% |
| NPV | 66.2% | 66.04% | 55.22% |
PPV Positive Predictive Value, NPV negative predictive value
The Hosmer–Lemeshow test yielded non-significant P-values of 0.707 in the training group, 0.401 in the validation group, and 0.735 in the temporal validation group, indicating a satisfactory goodness of fit for the model. To complement the H–L test, calibration intercept and slope were calculated: (1) Training cohort: Intercept = 0, Slope = 1 (perfect calibration, consistent predicted vs. observed probabilities); (2) Internal validation cohort: Intercept = −0.1181, Slope = 1.0467 (negligible systematic bias, excellent calibration); (3) Temporal validation cohort: Intercept = 0.3134, Slope = 0.7156 (mild overestimation of risk and risk shrinkage, respectively). Near-ideal calibration in the training and internal validation cohorts is clinically expected, as these cohorts share similar baseline characteristics. Deviations in the temporal validation cohort may reflect temporal changes in patient demographics or clinical practice patterns. The magnitude of deviation remains clinically acceptable, and the model can be optimized via shrinkage adjustment for external application.
After 500 bootstrap resampling iterations, the mean absolute error between the simulated and actual curves was 0.011 in the training group (Fig. 5A), 0.047 in the validation group (Fig. 5B), and 0.046 in the temporal validation group (Fig. 5C). These results suggest that the trend trajectories of the two curves were closely aligned and exhibited strong consistency.
Fig. 5.
Calibration curves for model evaluation. A-C Calibration curves in the training group, validation group, and temporal validation group to evaluate the performance of the prediction model. The 45-degree line (dashed) serves as the benchmark for perfect calibration. The model's initial fit within the development cohort is shown by the apparent calibration curve (dotted line), and the bias-corrected curve (solid line) quantifies the expected performance in new populations after applying a bootstrap-based correction for optimism (500 resamples)
DCA demonstrated that the nomogram model yielded higher net benefits compared to the treat-none and treat-all strategies across the training (Fig. 6A), validation (Fig. 6B), and temporal validation (Fig. 6C) groups. Across all groups, the nomogram showed a consistently high net benefit (training: 23%−99%; validation: 21%−89%; temporal validation: 18%−90%). This supports its utility as an effective predictor of delayed extubation risk and a guide for clinical decision-making.
Fig. 6.
DCA for model evaluation. A-C DCA curves in the training group, validation group, and temporal validation group to assess the model's net benefit and clinical applicability across different threshold probabilities. The DCA demonstrates the clinical value of the nomogram. The bold solid blue line denotes the nomogram's predicted risk of delayed extubation, contrasted with two reference scenarios-universal occurrence (solid gray line) and non-occurrence (fine solid black line) of the event
CIC (Fig. 7A-C) was generated by predicting risk stratification among 1000 individuals using bootstrap resampling. The red curve traces the count of patients classified as high-risk for delayed extubation at varying threshold probabilities, whereas the blue curve tracks the concomitant number of true positives. Across all three group, the model demonstrated consistent effectiveness in identifying high-risk patients throughout nearly the entire spectrum of threshold probabilities, indicating robust clinical applicability.
Fig. 7.
CIC for model evaluation. A-C CIC curves in the training group, validation group, and temporal validation group to visually evaluate the potential clinical utility of a prediction model.CIC shows the number of individuals classified as high-risk for delayed extubation (red curve) and the number of true positives (blue curve) at different threshold probabilities
Discussion
Prolonged mechanical ventilation is associated with adverse clinical outcomes, increased morbidity, and extended hospital stays [24]. Prostate cancer is the most common cancer among men, particularly older men [25]. Some studies have indicated that advanced age may lead to delayed extubation [15]; However, to the best of our knowledge, no study has investigated delayed extubation post-anesthesia in patients undergoing robot-assisted radical prostatectomy. Therefore, a preoperative assessment tool is required to predict the risk of delayed extubation in patients undergoing robot-assisted radical prostate cancer surgery.
We developed and validated a predictive model for delayed extubation, which showed good discrimination and calibration for individualized prediction. Our model includes five predictors: age, cerebral infarction, pulmonary disease, coronary heart disease, and IOH.
Consistent with previous studies [1, 26–29], our research also identified age, anemia, coronary heart disease, colloid infusion volume, and intraoperative blood transfusion as risk factors associated with delayed extubation. Although anemia did not reach statistical significance in the multivariate analysis, a significant association was observed in univariate analysis. Similarly, colloid infusion volume showed a suggestive difference in univariate analysis but failed to achieve statistical significance. A trend was also noted for intraoperative blood transfusion; however, likely due to the limited number of positive cases, it did not reach a statistically significant level. Anemia, colloid infusion, and blood transfusion were not significant predictors in the multivariate analysis, potentially due to three factors: (1) Robotic-assisted radical prostatectomy is an elective procedure, and the incidence of preoperative anemia or intraoperative blood transfusion is relatively low, limiting statistical power to detect independent effects; (2) As a retrospective study, colloid infusion volumes were recorded by anesthesiologists intraoperatively, which may have introduced measurement bias (e.g., imprecise documentation of exact milliliters); (3) Confounding factors such as baseline hemodynamics or concurrent fluid therapy may have masked the independent effects of these variables.
Compared with previous studies, our results also identified pulmonary disease, cerebral infarction, and IOH as significant predictors of delayed extubation. Tong et al. [15] reported that a lower FEV1/FVC ratio was an independent risk factor for delayed extubation, while Rasera et al. [30] found that elevated end-tidal carbon dioxide was associated with extubation failure, collectively underscoring the impact of pulmonary function on extubation outcomes. In line with these findings, our study confirmed pulmonary disease as an independent predictor. Chen et al. [31] found that patients with a history of cerebral infarction before surgery had a higher risk of extubation failure, highlighting the role of central nervous system in postoperative extubation. Our analysis similarly showed an association between previous cerebral infarction and delayed extubation. The strong association between a history of cerebral infarction and delayed extubation aligns with the known vulnerability of the post-stroke brain. Such patients often have impaired cerebrovascular autoregulation and increased sensitivity to anesthetic agents and perioperative hemodynamic fluctuations, which may cause perfusion-related brain injury [32]. Intraoperative hypotension, particularly episodes of severe and sustained hypotension (mean arterial pressure < 65 mmHg), may lead to delayed extubation via multiple physiological pathways: (1) Reduced cerebral perfusion may exacerbate subclinical cerebral edema or ischemia, impairing neural recovery [32]; (2) Decreased hepatic and renal perfusion delays the metabolism and excretion of anesthetic drugs, prolonging their duration of action [33, 34]; (3) Hypotension-induced tissue hypoxia may trigger systemic inflammatory responses, further delaying postoperative recovery [35].
The nomogram was developed to enhance the visualization of the model and promote its clinical applicability. The predictive model exhibited strong discriminatory ability, with AUC of 0.763, 0.811, and 0.769 in the training, validation, and temporal validation groups, respectively. Both the Hosmer–Lemeshow test and calibration curves indicated good model consistency across all three groups. Furthermore, the DCA and CIC graphically confirmed a significant net clinical benefit of the nomogram. This finding further validated the model's applicability for clinical decision-making.
The majority of patients undergoing robot-assisted radical prostatectomy are elderly men, who are at higher risk for delayed extubation. However, research specifically addressing this clinical scenario remains scarce. To bridge this gap, we developed a nomogram to predict delayed extubation based on the following predictors: age, history of cerebral infarction, pulmonary disease, coronary heart disease, and IOH.
The developed nomogram has several potential clinical applications. Preoperatively, it can facilitate risk stratification during anesthetic planning, allowing high-risk patients to be allocated to more experienced anesthetists. Intraoperatively, real-time monitoring of a patient’s cumulative risk score can prompt proactive hemodynamic interventions (e.g., fluid resuscitation, vasopressor use) to mitigate modifiable risk factors such as hypotension. Postoperatively, a high predicted risk of delayed extubation can guide PACU resource allocation (e.g., dedicated monitoring beds, increased nurse-to-patient ratios) and discharge planning. Prospective validation is warranted to confirm whether these interventions improve clinical outcomes.
Despite its contributions, this study has several limitations. First, as a single-center investigation, it may be influenced by center-specific biases, and the generalizability of our findings could be limited. Therefore, future multicenter prospective studies are warranted to validate these results. Second, the relatively small number of cases involving intraoperative blood transfusion may have reduced the statistical power to detect significant associations, potentially affecting the reliability of this specific result.
Conclusions
In conclusion, this study identified age, cerebral infarction, pulmonary disease, coronary heart disease, and IOH as independent predictors of delayed extubation following robot-assisted radical prostatectomy. The resulting nomogram provides a visual tool for individualized risk prediction, which may assist clinicians in perioperative planning and decision-making.
Acknowledgements
We extend our sincere gratitude to our colleagues in the Department of Anesthesiology, Nanjing Drum Tower Hospital, for their support of this work.
Animal studies
Not applicable.
Abbreviations
- IOH
Intraoperative hypotension
- MAP
Intraoperative mean arterial pressure
- LASSO
Least absolute shrinkage and selection operator
- ROC
Receiver operating characteristic
- AUC
Area under the curve
- DCA
Decision curve analysis
- CIC
Clinical impact curve
- OR
Odds ratio
- CI
Confidence interval
- FEV1
Forced expiratory volume in 1 s
- FVC
Forced vital capacity
Authors’ contributions
Xiaoping Gu contributed to the study protocol and manuscript preparation. Tantan Fang was responsible for data collection and performed the statistical analysis. Chuanfei Liu assisted in the statistical analysis and was involved in interpreting the results.
Funding
Not applicable.
Data availability
Upon reasonable request, the corresponding author can provide the datasets used and/or analyzed in this study.
Declarations
Ethics approval and consent to participate
This study was approved by the Medical Ethics Committee of Nanjing Drum Tower Hospital with the reference number 2024–742-01, and in compliance with the Helsinki Declaration. Given that this study was conducted using anonymized observational data, individual informed consent for participation was waived by the Committee on Medical Ethics of Nanjing Drum Tower Hospital.
Consent for publication
Not applicable.
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.Zirka HA, John GG, Laura CL, Joanna LM, Eric JH, Mitchell F. Factors that correlate with the decision to delay extubation after multilevel prone spine surgery. J Neurosurg Anesthesiol. 2014;26(2):167–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Joshua C, Toyokazu E, Matthew F. Tracheal resection for post-intubation/post-tracheostomy tracheal stenosis. Thorac Surg Clin. 2025;35(1):61–72. [DOI] [PubMed] [Google Scholar]
- 3.Guidelines for the Management of Adults with Hospital-acquired. Ventilator-associated, and Healthcare-associated Pneumonia. Am J Respir Crit Care Med. 2005;171(4):388–416. [DOI] [PubMed] [Google Scholar]
- 4.Barker J, Martino R, Reichardt B, Hickey EJ, Ralph-Edwards A. Incidence and impact of dysphagia in patients receiving prolonged endotracheal intubation after cardiac surgery. Can J Surg. 2009;52(2):119–24. [PMC free article] [PubMed] [Google Scholar]
- 5.Moon MC, Abdoh A, Hamilton GA, Lindsay WG, Duke PC, Pascoe EA, et al. Safety and efficacy of fast track in patients undergoing coronary artery bypass surgery. J Card Surg. 2001;16(4):319–26. [DOI] [PubMed] [Google Scholar]
- 6.Rosenberg H, Antognini JF, Muldoon S. Testing for malignant hyperthermia. Anesthesiology. 2002;96(1):232–7. [DOI] [PubMed] [Google Scholar]
- 7.Epstein RH, Dexter F, Cajigas I, Mahavadi AK, Shah AH, Abitbol N, et al. Prolonged tracheal extubation time after glioma surgery was associated with lack of familiarity between the anesthesia provider and the operating neurosurgeon. A retrospective, observational study. J Clin Anesthesia. 2020;60(0):118–24. [DOI] [PubMed] [Google Scholar]
- 8.Siegel RL, Miller KD, Jemal A. Cancer statistics, 2018. CA Cancer J Clin. 2018;68:7–30. [DOI] [PubMed] [Google Scholar]
- 9.Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2018;68(6):394–424. [DOI] [PubMed] [Google Scholar]
- 10.Dy GW, Gore JL, Forouzanfar MH, Naghavi M, Fitzmaurice C. Global burden of urologic cancers, 1990–2013. Eur Urol. 2017;71(3):437–46. [DOI] [PubMed] [Google Scholar]
- 11.Wong MC, Goggins WB, Wang HH, Fung FD, Leung C. Global incidence and mortality for prostate cancer: analysis of temporal patterns and trends in 36 countries. Eur Urol. 2016;70(5):862–74. [DOI] [PubMed] [Google Scholar]
- 12.Hughes T, Rai B, Madaan S, Chedgy E, Somani B. The availability, cost, limitations, learning curve and future of robotic systems in urology and prostate cancer surgery. J Clin Med. 2023;12(6):2268. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Patel VR, Coelho RF, Chauhan S, Orvieto MA, Palmer KJ, Rocco B, et al. Continence, potency and oncological outcomes after robotic‐assisted radical prostatectomy: early trifecta results of a high‐volume surgeon. BJU Int. 2010;106(5):696–702. [DOI] [PubMed] [Google Scholar]
- 14.Kimura T, Egawa S. Epidemiology of prostate cancer in Asian countries. Int J Urol. 2018;25(6):524–31. [DOI] [PubMed] [Google Scholar]
- 15.Tong C, Miao Q, Zheng J, Wu J. A novel nomogram for predicting the decision to delayed extubation after thoracoscopic lung cancer surgery. Ann Med. 2023;55(1):800–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Veyckemans F, von Ungern‐Sternberg B. Tracheal extubation in children: planning, technique, and complications. Paediatr Anaesth. 2020;30(3):331–8. [DOI] [PubMed] [Google Scholar]
- 17.Vannucci A, Riordan IR, Prifti K, Sebastiani A, Helsten DL, Lander DP, et al. Prolonged time to extubation after general anaesthesia is associated with early escalation of care. Eur J Anaesthesiol. 2021;38(5):494–504. [DOI] [PubMed] [Google Scholar]
- 18.Mandell MS, Stoner TJ, Barnett R, Shaked A, Bellamy M, Biancofiore G, et al. A multicenter evaluation of safety of early extubation in liver transplant recipients. Liver Transpl. 2007;13(11):1557–63. [DOI] [PubMed] [Google Scholar]
- 19.Brady K, Hogue CW. Intraoperative hypotension and patient outcome: does “one size fit all?” Anesthesiology. 2013;119(3):495–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Hirsch J, DePalma G, Tsai TT, Sands LP, Leung JM. Impact of intraoperative hypotension and blood pressure fluctuations on early postoperative delirium after non-cardiac surgery. Br J Anaesth. 2015;115(3):418–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Bijker JB, Persoon S, Peelen LM, Moons KG, Kalkman CJ. Intraoperative hypotension and perioperative ischemic stroke after general surgery: a nested case-control study. Anesthesiology. 2012;116(3):658–64. [DOI] [PubMed] [Google Scholar]
- 22.Salmasi V, Maheshwari K, Yang D, Mascha EJ, Singh A. Relationship between intraoperative hypotension, defined by either reduction from baseline or absolute thresholds, and acute kidney and myocardial injury after noncardiac surgery: a retrospective cohort analysis. Anesthesiology. 2017;126(1):47–65. [DOI] [PubMed] [Google Scholar]
- 23.Zhang Z. Univariate description and bivariate statistical inference: the first step delving into data. Ann Transl Med. 2016;4(5):91–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Xu S-S, Tian Y, Ma Y-J, Zhou Y-M, Tian Y, Gao R, et al. Development of a prediction score for evaluation of extubation readiness in neurosurgical patients with mechanical ventilation. Anesthesiology. 2023;139(5):614–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Wang LG, Montaño AR, Masillati AM, Jones JA, Barth CW, Combs JR, et al. Nerve visualization using phenoxazine‐based near‐infrared fluorophores to guide prostatectomy. Adv Mater. 2024;36(16):e2304724. [DOI] [PubMed] [Google Scholar]
- 26.Anastasian ZH, Kim M, Heyer EJ, Wang S, Berman MF. Attending handoff is correlated with the decision to delay extubation after surgery. Anesth Analg. 2016;122(3):758–64. [DOI] [PubMed] [Google Scholar]
- 27.Li X, Liu J, Xu Z, Wang Y, Chen L, Bai Y, et al. Early identification of delayed extubation following cardiac surgery: development and validation of a risk prediction model. Front Cardiovasc Med. 2022;9(0):1002768. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Huang X, Tan R, Lin J-W, Li G, Xie J. Development of prediction models to estimate extubation time and midterm recovery time of ophthalmic patients undergoing general anesthesia: a cross-sectional study. BMC Anesthesiol. 2023;23(1):83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Wang R, Liu Y, Li W, Zhang C, Zhang X, Wang F, et al. Analysis of factors affecting smooth extubation in delayed extubation patients following intracranial surgery: an ambispective cohort study. BMC Anesthesiol. 2025;25(1):531. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Rasera CC, Gewehr PM, Domingues AMT. PET(CO2) measurement and feature extraction of capnogram signals for extubation outcomes from mechanical ventilation. Physiol Meas. 2015;36(2):231–42. [DOI] [PubMed] [Google Scholar]
- 31.Chen S, Zhang Y, Che L, Shen L, Huang Y. Risk factors for unplanned reintubation caused by acute airway compromise after general anesthesia: a case-control study. BMC Anesthesiol. 2021;21(1):17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Greene NH, Lee LA. Modern and evolving understanding of cerebral perfusion and autoregulation. Adv Anesth. 2012;30(1):97–129. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Julian Y. Anaesthesia in chronic dialysis patients: a narrative review. World J Crit Care Med. 2025;14(1):100503. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Chetty M, Johnson TN, Polak S, Salem F, Doki K. Physiologically based pharmacokinetic modelling to guide drug delivery in older people. Adv Drug Deliv Rev. 2018;135(0):85–96. [DOI] [PubMed] [Google Scholar]
- 35.Torrez JP, Otsuki DA, Zeferino SP, Sanchez AF, Auler JO Jr, Zeferino S, et al. Vasoplegic syndrome following bypass: a comprehensive review of pathophysiology and proposed treatments. Cureus. 2025;17(1):e78057. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Upon reasonable request, the corresponding author can provide the datasets used and/or analyzed in this study.