Development and internal validation of a clinical nomogram for predicting major postoperative pulmonary complications following radical cystectomy: a retrospective cohort study

Introduction

Bladder cancer represents a significant global health burden, ranking as one of the most common malignancies worldwide. Recent epidemiological data from 2020 estimated approximately 573,000 new cases and 213,000 deaths globally, with projections indicating a substantial increase in this burden by 2040, primarily driven by population aging (1). For patients diagnosed with muscle-invasive bladder cancer (MIBC), radical cystectomy (RC) remains the cornerstone of curative-intent therapy and is established as the standard of care in international guidelines (2). This extensive surgical procedure, however, is associated with a high rate of postoperative morbidity and mortality, challenging patient recovery and straining healthcare resources.

Among the most frequent and severe sequelae following RC are postoperative pulmonary complications (PPCs), which encompass a range of adverse respiratory events including pneumonia, atelectasis, and respiratory failure. The incidence of PPCs in the post-RC patient population is alarmingly high, with studies reporting rates between 13% and 17% (3,4). The onset of a PPC has profound negative implications for patient prognosis, being directly linked to a significant prolongation of hospital stays, increased healthcare costs, and a stark rise in 30-day mortality rates (3). This underscores a critical need for effective risk stratification to identify patients who would benefit most from targeted perioperative preventative strategies.

In response, several general risk prediction models, such as the ARISCAT score, have been developed to forecast the likelihood of PPCs in the broader surgical population (5). While valuable in other contexts, the predictive accuracy of these generic models has been found low when applied to the specific, high-risk cohort undergoing RC. These models often demonstrate poor discrimination and calibration in RC patients, largely because they fail to account for the unique pathophysiological challenges inherent to this procedure (6). For instance, the extensive intestinal manipulation required for urinary diversion in RC is a major contributor to postoperative ileus, a condition that in turn significantly elevates PPC risk through mechanisms like diaphragmatic dysfunction and increased risk of aspiration—a critical variable not captured by general risk models (3). This highlights a distinct research gap and a pressing clinical need for a more specialized predictive tool.

To address this limitation, our study aims to develop and validate a novel predictive model in the form of a nomogram. Nomograms offer distinct advantages as clinical predictive tools, translating complex regression models into a simple, graphical interface that provides a user-friendly, individualized risk assessment. This enhances clinical utility by offering high interpretability and readability, thereby facilitating shared decision-making and the implementation of personalized patient care (7,8). The purpose of this study was therefore to develop and internally validate a dynamic perioperative nomogram that integrates preoperative patient characteristics with intraoperative information to estimate the risk of major pulmonary complications following RC for bladder cancer. We present this article in accordance with the TRIPOD reporting checklist (available at https://tau.amegroups.com/article/view/10.21037/tau-2026-0459/rc).

Methods

Study design

This single-center retrospective cohort study was conducted at the Department of Urology of our Hospital, a tertiary teaching hospital, to develop and internally validate a nomogram predicting the risk of major PPCs in patients undergoing RC for bladder cancer. Clinical data of consecutive patients who underwent RC between January 2020 and June 2025 were retrospectively retrieved from the institution’s electronic medical record (EMR) system. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The research protocol was reviewed and approved by the Institutional Review Board (IRB) of The Third Affiliated Hospital of Chongqing Medical University. The requirement for informed consent was waived due to the retrospective nature of the study.

Study population

To rigorously determine study eligibility, potential participants were evaluated against specific criteria. Inclusion criteria comprised all of the following: (I) Histopathologically confirmed diagnosis of bladder urothelial carcinoma requiring RC with curative intent, in accordance with the European Association of Urology (EAU) guidelines (9); (II) age ≥18 years at the time of surgery; (III) Underwent elective open or robot-assisted RC; (IV) Complete perioperative baseline and 30-day postoperative follow-up data. Exclusion criteria encompassed any of the following: (I) palliative, salvage, or emergency RC, which inherently alters the baseline risk profile; (II) pre-existing severe pulmonary diseases before surgery, such as interstitial pneumonia, pulmonary fibrosis, active pulmonary tuberculosis, or end-stage chronic obstructive pulmonary disease (COPD) requiring long-term home oxygen therapy, which severely confound the assessment of incident PPCs (10); (III) presence of other concomitant primary malignancies that could independently affect short-term survival or surgical extent; (IV) preoperative active respiratory tract infection within 2 weeks prior to surgery; (V) missing data exceeding 20% for core perioperative variables; (VI) lost to follow-up or death from non-medical causes within 30 days postoperatively.

Data collection and outcome definition

The primary outcome of the study was the incidence of major PPCs occurring within 30 days following RC. In accordance with the Prospective Evaluation of a Risk Score for Postoperative Pulmonary Complications in Europe (PERISCOPE) study criteria and the European Perioperative Clinical Outcome (EPCO) guidelines (10), PPCs were constituted by the occurrence of any of the following four adverse events: pneumonia (defined by new or progressive pulmonary infiltrates on chest imaging combined with at least two clinical signs including body temperature >38 °C, leukocyte count >12,000/µL, or purulent airway secretions), respiratory failure (necessitating unplanned endotracheal intubation or continuous mechanical ventilation for more than 48 hours postoperatively), acute respiratory distress syndrome [diagnosed strictly according to the Berlin Definition (11)], or clinically significant atelectasis (confirmed by radiological evidence and requiring therapeutic interventions such as bronchoscopy). To ensure rigorous and unbiased outcome classification, all suspected PPC cases underwent a standardized dual-review adjudication process. Two independent senior attending pulmonologists, who were completely blinded to the patients’ baseline characteristics and perioperative predictor variables, evaluated the electronic medical records to confirm the diagnosis. In instances of diagnostic discrepancy or ambiguity, a third independent chief physician of pulmonary and critical care medicine provided the final binding adjudication through consensus discussion.

Concurrently, a comprehensive array of candidate predictor variables was extracted. Demographic and baseline physical status variables included age, sex, body mass index (BMI), smoking history (quantified as daily smoking of >1 cigarette for >1 year), alcohol consumption history, and the American Society of Anesthesiologists (ASA) physical status classification (12). Comorbid conditions were quantified using the Charlson Comorbidity Index (CCI) (13) and explicitly included hypertension, diabetes mellitus, coronary artery disease, preoperative anemia (defined as hemoglobin <120 g/L for males and <110 g/L for females), and COPD, which was objectively graded based on preoperative spirometry utilizing the Global Initiative for Chronic Obstructive Lung Disease (GOLD) criteria (14). Preoperative laboratory and ancillary testing variables, restricted to a window of 14 days prior to surgery, encompassed serum albumin, derived inflammatory indices including the neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR), and the forced expiratory volume in one second percentage of predicted value (FEV1%pred) derived from baseline pulmonary function tests, if multiple measurements were available within this window, the value closest to the date of surgery was used. Furthermore, granular surgery-related variables were extracted from operative and anesthesia logs, comprising the surgical approach (open versus robot-assisted), total operative time, estimated intraoperative blood loss, the requirement for intraoperative red blood cell transfusion, and the specific method of urinary diversion (ileal conduit versus orthotopic neobladder). Recognizing the complexity of RC and potential confounders often questioned during peer review, we additionally extracted data on the administration of neoadjuvant chemotherapy (NAC) and intraoperative net fluid balance (including crystalloids and colloids), as these factors are recognized to significantly modulate perioperative physiologic reserve, fluid overload risk, and systemic immunity.

Intraoperative net fluid balance was calculated at the end of surgery from the recorded volume of administered crystalloids, colloids, and blood products minus measured urine output and blood loss. Because this variable was not available before surgery, the final nomogram was designed as a dynamic perioperative risk-updating tool to be applied near the completion of surgery or during immediate postoperative handover, rather than as a purely preoperative prediction model.

Data quality control and missing data management

To ensure the integrity and reliability of the retrospective cohort data, a rigorous, multi-tiered data extraction and validation protocol was executed. Initial data extraction from the institutional EMR and anesthesia systems utilized structured queries with automated logic and range checks to prevent chronological errors and physiological impossibilities. Subsequently, a random sample comprising 15% of the total cohort (72 patient records) underwent independent manual abstraction by two blinded clinical researchers to verify the accuracy of key predictor variables and the primary outcome classification, while any cohort eligibility uncertainties were adjudicated by a senior urologist. High inter-rater reliability was confirmed, demonstrating a Cohen’s Kappa coefficient of 0.89 for categorical variables (such as ASA classification and PPCs occurrence) and an intraclass correlation coefficient (ICC) of 0.94 for continuous variables (including operative time and preoperative pulmonary function metrics). Discrepancies identified during this dual-abstraction phase were definitively resolved by a senior urologist through a meticulous review of the original source documents. Addressing the inherent challenge of missing data in retrospective research, variables with a missingness rate between 5% and 20%—specifically preoperative FEV1%pred (missing in 8.4% of cases) and BMI (missing in 5.2% of cases)—were appropriately handled using multiple imputation by chained equations (MICE) under the missing at random (MAR) assumption. This procedure generated 10 complete imputed datasets, ensuring that the statistical uncertainty associated with missingness was properly incorporated. All subsequent statistical modeling, including the least absolute shrinkage and selection operator (LASSO) regression for variable selection and the multivariable logistic regression, was performed across these 10 datasets, with final parameter estimates, odds ratios (ORs), and 95% confidence intervals (CIs) pooled according to Rubin’s rules to yield robust and unbiased predictive models.

Statistical analysis

All statistical procedures were completed with R (version 4.5.0) and SPSS (version 26.0). Statistical significance was defined as a two-sided P<0.05. Baseline characteristics were compared using the pre-imputation dataset to preserve the real-world cohort profile. Depending on the Shapiro-Wilk test results for normality, continuous variables were expressed as mean ± standard deviation (SD) or median [interquartile range (IQR)], and inter-group differences (PPC vs. non-PPC) were examined using the independent t-test or Mann-Whitney U test. Categorical data were described as frequencies (percentages) and compared via the chi-square or Fisher’s exact test.

We did not use univariable screening based on P value thresholds for predictor selection, because this approach may exclude clinically relevant predictors and introduce selection instability. Instead, LASSO regression was applied directly to all prespecified candidate predictors to reduce model complexity and partially limit overfitting. Given the limited number of major PPC events relative to the number of candidate predictors, the resulting model was considered exploratory. The LASSO procedure was performed separately across the 10 multiply imputed datasets. In each dataset, the optimal penalization parameter (λ) was selected using 10-fold cross-validation based on the minimum mean cross-validated error. Variables with non-zero coefficients in at least 5 of the 10 imputed datasets were retained according to a prespecified majority-vote criterion and subsequently entered into a multivariable logistic regression model. Regression coefficients, ORs, and 95% CIs were pooled across the imputed datasets using Rubin’s rules. Univariable logistic regression analyses were performed only for the variables retained by LASSO to describe their unadjusted associations with major PPCs and were not used for predictor selection or model construction. The final model was presented as a nomogram. Internal validation was performed using 1,000 bootstrap resamples to estimate optimism; however, penalization and bootstrap validation were not considered sufficient to eliminate the risk of overfitting associated with the low event count.

Model performance was internally validated via bootstrapping with 1,000 iterations to adjust for optimism. Discrimination was assessed by the C-index (analogous to the AUC), where values >0.70 and >0.80 reflect acceptable and excellent discriminative abilities, respectively. Given the sample size dependency of the Hosmer-Lemeshow test, model calibration was mainly evaluated by visually inspecting the calibration curve’s slope and intercept, and secondarily supported by the Hosmer-Lemeshow test. Decision curve analysis (DCA) was performed to estimate the model-based net benefit across a range of threshold probabilities relative to the treat-all and treat-none strategies. Because the model was evaluated only in the development cohort, DCA was interpreted as exploratory evidence of potential decision usefulness rather than confirmation of clinical effectiveness.

Results

Patient enrollment and baseline characteristics

Between January 2020 and June 2025, 615 patients undergoing RC were screened. After applying predefined exclusion criteria (Figure 1), a final cohort of 456 patients was analyzed. Within this cohort, 32 (7.0%) developed major PPCs within 30 days postoperatively. Baseline demographic and clinicopathological characteristics (pre-imputation) are detailed in Table 1. Compared to the non-PPC group, patients with PPCs were significantly older (72.4±6.8 vs. 66.5±8.1 years, P<0.001) and had higher rates of smoking (75.0% vs. 51.4%, P=0.01), ASA physical status III–IV (P=0.003), CCI scores (P<0.001), and COPD prevalence/severity (P<0.001). The PPC group also exhibited lower serum albumin (35.2±4.1 vs. 39.8±3.9 g/L, P<0.001) and FEV1%pred (66.5%±11.2% vs. 81.4%±12.5%, P<0.001), alongside higher rates of preoperative anemia (P=0.03) and elevated NLRs (P=0.008). Intraoperatively, the occurrence of PPCs was associated with longer operative times (345.5±52.1 vs. 312.4±48.6 min, P<0.001), greater estimated blood loss (P=0.002), higher red blood cell transfusion rates (P=0.001), and increased net fluid balance (1,650±510 vs. 1,220±430 mL, P<0.001). Other baseline and operative variables, including sex, BMI, comorbidities, NAC, and surgical approaches, showed no significant differences between the two groups (all P>0.05).

Figure 1 Flow diagram of patient enrollment. PPC, postoperative pulmonary complication.

Selection of key predictors via LASSO regression and multivariable analysis

Given the limited number of major PPC events (n=32) relative to the 23 prespecified candidate predictors, LASSO penalized regression was used to reduce model complexity and partially limit overfitting. The LASSO procedure was performed independently across each of the 10 multiply imputed datasets. In each dataset, the penalization parameter (λ) was selected using 10-fold cross-validation based on the minimum mean cross-validated error. Variables with non-zero coefficients in at least 5 of the 10 imputed datasets were retained according to the prespecified majority-vote criterion. This procedure reduced the candidate set to six variables: age, smoking history, ASA physical status III–IV, preoperative serum albumin, FEV1%pred, and intraoperative net fluid balance (Figure 2). These six variables were subsequently included in the multivariable logistic regression model. For descriptive purposes, separate univariable logistic regression analyses were also performed for the retained variables; these analyses were not used for predictor selection or model specification. After pooling the multivariable estimates across the imputed datasets using Rubin’s rules, older age [adjusted odds ratio (aOR) =1.07 per year; 95% CI: 1.02–1.13; P=0.01], smoking history (aOR =2.41; 95% CI: 1.08–5.38; P=0.03), ASA physical status III–IV (aOR =2.65; 95% CI: 1.18–5.95; P=0.02), lower serum albumin (aOR =0.83 per 1 g/L increase; 95% CI: 0.74–0.93; P=0.002), lower FEV1%pred (aOR =0.94 per 1% increase; 95% CI: 0.90–0.98; P=0.003), and higher intraoperative net fluid balance (aOR =1.16 per 100 mL increase; 95% CI: 1.06–1.28; P=0.001) were associated with major PPCs within 30 days after RC (Table 2).

Figure 2 Variable selection using the LASSO regression model. (A) LASSO coefficient profiles of the 23 candidate predictors plotted against the log(λ) sequence. The colored lines represent the six robust predictors ultimately selected by the model (maintaining non-zero coefficients), while the semi-transparent grey lines represent the unselected candidates whose coefficients were shrunk to zero. The vertical dashed line indicates the optimal penalization parameter (λ_min). (B) The 10-fold cross-validation curve. Red dots represent the mean binomial deviance, with vertical error bars indicating the corresponding standard deviations. The vertical dashed line defines the optimal log(λ) value that minimizes the mean cross-validated error, leading to the precise identification of the six critical predictors for the subsequent multivariable analysis. ASA, American Society of Anesthesiologists; FEV1%pred, forced expiratory volume in one second expressed as a percentage of the predicted value; LASSO, least absolute shrinkage and selection operator.

Construction of the dynamic perioperative nomogram

Based on the independent prognostic factors identified through the pooled multivariable logistic regression analysis, a visual nomogram was constructed to facilitate the individualized estimation of the risk for developing major PPCs within 30 days following RC. The predictive nomogram integrated all six robust variables selected via the LASSO consensus: age, smoking history, ASA physical status, preoperative serum albumin level, FEV1%pred, and intraoperative net fluid balance. Within this graphical interface, each clinical predictor is represented by an individual axis and is assigned a specific point value on the uppermost reference scale (ranging from 0 to 100) based on the magnitude of its respective adjusted regression coefficient. The dynamic perioperative nomogram incorporated six variables: age, smoking history, ASA physical status, preoperative serum albumin, FEV1%pred, and intraoperative net fluid balance. The first five variables were available before surgery, whereas net fluid balance was determined near the end of the procedure. Therefore, the intended prediction time point was the completion of surgery or immediate postoperative handover. At this time, the points assigned to the six variables could be summed to estimate the probability of major PPCs within 30 days (Figure 3).

Figure 3 Dynamic perioperative nomogram for estimating the risk of major PPCs after radical cystectomy. The model combines five preoperative variables with intraoperative net fluid balance and is intended for risk estimation near the end of surgery or during immediate postoperative handover. ASA, American Society of Anesthesiologists; FEV1%pred, forced expiratory volume in one second expressed as a percentage of the predicted value; PPCs, postoperative pulmonary complications.

Model performance, internal validation, and DCA

The established nomogram had an apparent C-index of 0.824 (95% CI: 0.758–0.890). After internal validation using 1,000 bootstrap resamples, the optimism-corrected C-index was 0.811, with an estimated optimism of 0.013. The bootstrap-corrected calibration curve indicated reasonable agreement between predicted and observed probabilities within the study cohort, with a calibration slope of 0.96 and an intercept of 0.02. The Hosmer-Lemeshow test showed no evidence of statistically significant lack of fit (χ²=6.45, P=0.60), although this test was considered supplementary to graphical calibration assessment. In DCA, the nomogram showed a higher estimated net benefit than the treat-all and treat-none strategies across threshold probabilities of approximately 4% to 65%. At a threshold probability of 10%, the estimated net benefit was 0.048 (Figure 4).

Figure 4 Validation and clinical utility of the predictive nomogram. (A) The ROC curve demonstrating the model’s apparent discriminative ability. (B) The calibration curve based on 1,000 bootstrap resamples, with the solid line showing actual predictive performance in close alignment with the ideal 45-degree reference line (dotted line). (C) DCA for the nomogram. The wide interval (4% to 65%) where the nomogram curve lies above both the “treat-none” (horizontal solid line, net benefit =0) and “treat-all” (descending solid line) reference lines indicates optimal clinical utility and superior net benefit. AUC, area under the curve; CI, confidence interval; DCA, decision curve analysis; ROC, receiver operating characteristic.

Risk stratification capacity of the predictive model

To further translate the predictive nomogram into a clinically actionable decision-support tool and directly demonstrate its practical risk stratification capacity, the entire study cohort (N=456) was categorized into three distinct prognostic tiers based on the tertiles of their calculated total nomogram scores. Specifically, patients were classified into a low-risk group (total score <85 points, n=152), an intermediate-risk group (total score 85 to 130 points, n=152), and a high-risk group (total score >130 points, n=152). The actual observed incidences of major PPCs within these defined strata were subsequently calculated and compared to assess the clinical validity of the categorizations. The observed occurrence rates of PPCs were 1.3% (2/152) in the low-risk group, 3.9% (6/152) in the intermediate-risk group, and 15.8% (24/152) in the high-risk group. A rigorous statistical evaluation utilizing the Chi-square test for trend revealed a highly significant, stepwise increase in the actual observed PPC incidence across the ascending risk tiers (chi-square for trend =23.45, P<0.001, Table 3).

Discussion

In this study, we developed and internally validated a nomogram for estimating the risk of major PPCs following RC. Major PPCs occurred in 7.0% of the cohort. Six variables were retained in the model: age, smoking history, ASA physical status III–IV, preoperative serum albumin, FEV1%pred, and intraoperative net fluid balance. The nomogram showed preliminary discrimination and calibration within the development cohort. Because intraoperative net fluid balance becomes available only near the completion of surgery, the model should be interpreted as a dynamic perioperative risk-updating tool rather than a purely preoperative prediction model.

Previous studies have assessed morbidity after RC using measures of frailty, comorbidity, nutritional status, systemic inflammation, surgical approach, and urinary diversion. Palumbo et al. found that frailty was associated with higher perioperative complication rates, mortality, length of stay, and hospitalization costs after RC (15). In a multicenter cohort of octogenarian and frail patients, Porreca et al. reported that perioperative outcomes varied according to surgical approach and urinary diversion, with robot-assisted RC combined with ureterocutaneostomy associated with lower blood loss, transfusion requirements, and length of stay than alternative approaches (16). Other studies have evaluated laboratory-based risk measures. Claps et al. found limited predictive reliability for several inflammatory indices, including NLR and PLR, although smoking, anemia, and fibrinogen were associated with major complications (17). The Controlling Nutritional Status (CONUT) score, which integrates serum albumin, lymphocyte count, and cholesterol, has also been associated with perioperative morbidity after RC (17). Collectively, these findings indicate that morbidity after RC is multifactorial and that no single comorbidity, inflammatory, nutritional, or procedural measure is likely to provide sufficient risk discrimination. Our model extends this literature by combining baseline clinical status, nutritional and pulmonary reserve, and an intraoperative exposure to estimate pulmonary-specific rather than overall postoperative morbidity.

Advanced age, smoking history, and ASA physical status III–IV were retained in our model. Older patients have reduced physiological reserve and are more likely to exhibit frailty and multimorbidity, which may increase vulnerability to major surgery (5,15). Smoking impairs mucociliary clearance, promotes airway inflammation, and increases susceptibility to respiratory infection and atelectasis (18). ASA classification reflects the overall burden of systemic disease, although previous studies suggest that conventional comorbidity and performance-status measures alone provide only moderate discrimination for postoperative morbidity (19,20). These variables may therefore be most informative when interpreted together with objective measures of nutritional and pulmonary reserve.

Preoperative serum albumin and FEV1%pred were also retained in the model. A systematic review demonstrated that poor nutritional status is associated with higher complication and mortality rates after RC, while studies using the CONUT score similarly support the relevance of combined nutritional and immune status (2,17). Hypoalbuminemia may reflect malnutrition, systemic inflammation, and reduced physiological reserve. Reduced plasma oncotic pressure may also facilitate pulmonary interstitial fluid accumulation during major surgical stress, particularly when capillary permeability is increased (21). FEV1%pred provides a more direct measure of respiratory reserve. Reduced expiratory flow may impair coughing and secretion clearance and contribute to postoperative atelectasis and pneumonia (22,23). The inclusion of albumin and FEV1%pred therefore adds organ-specific physiological information that is not fully represented by ASA classification or conventional comorbidity indices.

Higher intraoperative net fluid balance was associated with major PPCs. Excessive positive fluid balance may promote pulmonary interstitial edema, particularly in patients with hypoalbuminemia or limited cardiopulmonary reserve (24). Nevertheless, this association should not be interpreted as evidence that a universally restrictive fluid regimen is preferable. In major abdominal surgery, restrictive fluid therapy did not improve disability-free survival and increased the risk of acute kidney injury compared with a more liberal regimen (25). In contrast, an RC-specific randomized trial reported that restrictive deferred hydration combined with preemptive norepinephrine reduced postoperative complications and length of stay (26). ERAS recommendations therefore emphasize individualized goal-directed fluid management and maintenance of euvolemia rather than fixed fluid restriction (27,28). Because net fluid balance is calculated near the end of surgery, the present nomogram cannot prospectively determine an optimal fluid strategy. Instead, this variable updates postoperative pulmonary risk after intraoperative exposure and may help identify patients requiring closer respiratory surveillance during recovery.

The timing of model use is clinically relevant. Age, smoking history, ASA classification, serum albumin, and FEV1%pred characterize baseline susceptibility before surgery, whereas intraoperative net fluid balance updates this estimate near the completion of the procedure. The resulting probability may therefore be relevant to immediate postoperative handover and the planning of respiratory monitoring, airway-clearance measures, mobilization, and escalation of respiratory support. However, these potential applications require prospective evaluation and should not be interpreted as established clinical benefits.

Methodologically, we avoided univariable P value screening and applied LASSO directly to the prespecified candidate predictors. Multiple imputation was used to address missing data, and bootstrap resampling was used to estimate model optimism. Nevertheless, LASSO can only partially reduce overfitting and cannot fully compensate for the limited number of outcome events. DCA suggested possible decision usefulness within the development cohort but does not demonstrate that implementation of the model would improve perioperative management or patient outcomes.

This study has several limitations. First, the single-center retrospective design introduces potential selection and information bias and limits model transportability. PPC incidence and predictor effects may depend on local anesthesia and fluid-management protocols, ERAS adherence, surgical approach distribution, and postoperative respiratory care pathways. Internal bootstrap validation cannot establish performance across different populations and care processes; independent external validation and possible recalibration are therefore required. Second, only 32 major PPC events occurred despite consideration of 23 candidate predictors. The selected variables, regression coefficients, calibration, and discrimination may consequently remain unstable or optimistic, and the model should be regarded as exploratory. Third, the inclusion of intraoperative net fluid balance prevents purely preoperative application. Although a separate preoperative model might be more suitable for prehabilitation or surgical planning, the available event count did not support reliable development and validation of an additional model. Finally, residual confounding remains possible because detailed ventilation parameters, postoperative analgesia, mobilization, respiratory physiotherapy, airway-clearance measures, and variations in surgical technique were not consistently available.

Conclusions

We developed an exploratory dynamic perioperative nomogram that combines five preoperative variables with intraoperative net fluid balance to estimate the risk of major PPCs after RC. Because all predictors become available only near the end of surgery, the model is intended for perioperative risk updating and planning of immediate postoperative surveillance and respiratory care rather than for purely preoperative decision-making. Its performance and clinical usefulness require confirmation in larger external cohorts before implementation.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tau.amegroups.com/article/view/10.21037/tau-2026-0459/rc

Data Sharing Statement: Available at https://tau.amegroups.com/article/view/10.21037/tau-2026-0459/dss

Peer Review File: Available at https://tau.amegroups.com/article/view/10.21037/tau-2026-0459/prf

Funding: None.

Conflicts of Interest: Both authors have completed the ICMJE uniform disclosure form (available at https://tau.amegroups.com/article/view/10.21037/tau-2026-0459/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The research protocol was reviewed and approved by the Institutional Review Board (IRB) of The Third Affiliated Hospital of Chongqing Medical University. The requirement for informed consent was waived due to the retrospective nature of the study.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Zhang Y, Rumgay H, Li M, et al. The global landscape of bladder cancer incidence and mortality in 2020 and projections to 2040. J Glob Health 2023;13:04109. [Crossref] [PubMed]
2. Ornaghi PI, Afferi L, Antonelli A, et al. The impact of preoperative nutritional status on post-surgical complication and mortality rates in patients undergoing radical cystectomy for bladder cancer: a systematic review of the literature. World J Urol 2021;39:1045-81. [Crossref] [PubMed]
3. Martinez Carrique S, Crettenand F, Stritt K, et al. Respiratory Complications after Cystectomy with Urinary Diversion: Avoidable Complications or Ineluctable Destiny? J Clin Med 2024;13:1585. [Crossref] [PubMed]
4. Yu J, Hong B, Park JY, et al. Impact of Prognostic Nutritional Index on Postoperative Pulmonary Complications in Radical Cystectomy: A Propensity Score-Matched Analysis. Ann Surg Oncol 2021;28:1859-69. [Crossref] [PubMed]
5. Huang Z, Han Y, Zhuang H, et al. Prediction models for postoperative pulmonary complications: a systematic review and meta-analysis. Br J Anaesth 2025;135:1415-27. [Crossref] [PubMed]
6. Razdan S, Sljivich M, Pfail J, et al. Predicting morbidity and mortality after radical cystectomy using risk calculators: A comprehensive review of the literature. Urol Oncol 2021;39:109-20. [Crossref] [PubMed]
7. Zou J, Luo G, Zhou L, et al. Nomogram for predicting postoperative pulmonary complications in spinal tumor patients. BMC Anesthesiol 2024;24:56. [Crossref] [PubMed]
8. Wei C, Xiang X, Zhou X, et al. Development and validation of an interpretable radiomic nomogram for severe radiation proctitis prediction in postoperative cervical cancer patients. Front Microbiol 2022;13:1090770. [Crossref] [PubMed]
9. Alfred Witjes J, Max Bruins H, Carrión A, et al. European Association of Urology Guidelines on Muscle-invasive and Metastatic Bladder Cancer: Summary of the 2023 Guidelines. Eur Urol 2024;85:17-31. [Crossref] [PubMed]
10. Jammer I, Wickboldt N, Sander M, et al. Standards for definitions and use of outcome measures for clinical effectiveness research in perioperative medicine: European Perioperative Clinical Outcome (EPCO) definitions: a statement from the ESA-ESICM joint taskforce on perioperative outcome measures. Eur J Anaesthesiol 2015;32:88-105. [Crossref] [PubMed]
11. ARDS Definition Task Force. Acute respiratory distress syndrome: the Berlin Definition. JAMA 2012;307:2526-33.
12. Hurwitz EE, Simon M, Vinta SR, et al. Adding Examples to the ASA-Physical Status Classification Improves Correct Assignment to Patients. Anesthesiology 2017;126:614-22. [Crossref] [PubMed]
13. Charlson ME, Pompei P, Ales KL, et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373-83. [Crossref] [PubMed]
14. Agustí A, Celli BR, Criner GJ, et al. Global Initiative for Chronic Obstructive Lung Disease 2023 Report: GOLD Executive Summary. Eur Respir J 2023;61:2300239. [Crossref] [PubMed]
15. Palumbo C, Knipper S, Pecoraro A, et al. Patient frailty predicts worse perioperative outcomes and higher cost after radical cystectomy. Surg Oncol 2020;32:8-13. [Crossref] [PubMed]
16. Porreca A, Marino F, De Marchi D, et al. Robot-Assisted Radical Cystectomy with Ureterocutaneostomy: A Potentially Optimal Solution for Octogenarian and Frail Patients with Bladder Cancer. J Clin Med 2025;14:4898. [Crossref] [PubMed]
17. Claps F, Mir MC, van Rhijn BWG, et al. Impact of the controlling nutritional status (CONUT) score on perioperative morbidity and oncological outcomes in patients with bladder cancer treated with radical cystectomy. Urol Oncol 2023;41:49.e13-22.
18. Lugg ST, Scott A, Parekh D, et al. Cigarette smoke exposure and alveolar macrophages: mechanisms for lung disease. Thorax 2022;77:94-101. [Crossref] [PubMed]
19. Claps F, Rossin G, van Rhijn BWG, et al. The Utility of Inflammatory Serum Markers in the Assessment of Perioperative Morbidity after Radical Cystectomy for Bladder Cancer. Medicina (Kaunas) 2023;59:926. [Crossref] [PubMed]
20. Garg S, Govindaraj V, Dwivedi DP, et al. Postoperative pulmonary complications in patients undergoing upper abdominal surgery: risk factors and predictive models. Monaldi Arch Chest Dis 2025;95: [Crossref]
21. McNeil JB, Jackson KE, Wang C, et al. Linear Association Between Hypoalbuminemia and Increased Risk of Acute Respiratory Distress Syndrome in Critically Ill Adults. Crit Care Explor 2021;3:e0527. [Crossref] [PubMed]
22. Park HJ, Kim SM, Kim HR, et al. The value of preoperative spirometry testing for predicting postoperative risk in upper abdominal and thoracic surgery assessed using big-data analysis. J Thorac Dis 2020;12:4157-67. [Crossref] [PubMed]
23. Fisher BW, Majumdar SR, McAlister FA. Predicting pulmonary complications after nonthoracic surgery: a systematic review of blinded studies. Am J Med 2002;112:219-25. [Crossref] [PubMed]
24. Messina A, Robba C, Calabrò L, et al. Perioperative liberal versus restrictive fluid strategies and postoperative outcomes: a systematic review and meta-analysis of randomized controlled trials in major abdominal elective surgery. Crit Care 2021;25:205. [Crossref] [PubMed]
25. Myles PS, Bellomo R, Corcoran T, et al. Restrictive versus Liberal Fluid Therapy for Major Abdominal Surgery. N Engl J Med 2018;378:2263-74. [Crossref] [PubMed]
26. Wuethrich PY, Burkhard FC, Thalmann GN, et al. Restrictive deferred hydration combined with preemptive norepinephrine infusion during radical cystectomy reduces postoperative complications and hospitalization time: a randomized clinical trial. Anesthesiology 2014;120:365-77. [Crossref] [PubMed]
27. Cerantola Y, Valerio M, Persson B, et al. Guidelines for perioperative care after radical cystectomy for bladder cancer: Enhanced Recovery After Surgery (ERAS) society recommendations. Clin Nutr 2013;32:879-87. [Crossref] [PubMed]
28. Zhu AC, Agarwala A, Bao X. Perioperative Fluid Management in the Enhanced Recovery after Surgery (ERAS) Pathway. Clin Colon Rectal Surg 2019;32:114-20. [Crossref] [PubMed]