Risk factors for bronchopleural fistula after lobectomy for lung cancer
Highlight box
Key findings
• The incidence rate of bronchopleural fistula (BPF) after undergoing lobectomy for lung cancer was 0.44%.
• All 14 patients who developed BPF were men who had undergone right lower lobectomy.
• A multivariable analysis in the subgroup of men who underwent right lower lobectomy revealed that high serum C-reactive protein levels and a history of gastric cancer surgery were associated with BPF, whereas bronchial stump coverage was inversely associated with BPF.
What is known and what is new?
• Several studies have identified pneumonectomy, presence of residual carcinomatous tissue at the bronchial stump, preoperative irradiation, and diabetes as risk factors for BPF.
• This study revealed male sex, right lower lobectomy, high serum C-reactive protein levels, and history of gastrectomy as factors associated with the development of BPF after lobectomy for lung cancer.
What is the implication, and what should change now?
• Preventive measures, such as bronchial stump coverage, might be more suitable for patients at high risk of BPF.
Introduction
Surgical treatment for lung cancer has become remarkably safe in recent years due to the increase in the number of early-stage lung cancers, the widespread use of minimally invasive surgery, and advances in surgical instruments (1-4). Although the incidence of bronchopleural fistula (BPF) has decreased with the reduction in the number of central-type lung cancers and advances of endoscopic staplers, it still remains a serious complication that can lead to severe pneumonia, empyema, and surgical mortality.
Numerous studies have shown high incidence rates of BPF after pneumonectomy and surgery following neoadjuvant therapy (5-14). However, few reports have thoroughly examined BPF following standard lobectomy without preoperative treatment or bronchoplasty. Predicting the risk of BPF after regular lobectomy is useful in determining indications for preventive measures, such as bronchial stump coverage. The present study examined the incidence and the risk factors of BPF after lobectomy for lung cancer. We present this article in accordance with the STROBE reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-22-1809/rc).
Methods
Ethics statement
The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The study was approved by the Institutional Review Board for Clinical Research of the Cancer Institute Hospital of the Japanese Foundation for Cancer Research on March 22, 2022 (referral No. 2021-GB-117) and individual consent for this retrospective analysis was waived.
Study design
This retrospective, observational study reviewed patients with primary lung cancer who underwent surgical resection between January 2005 and December 2020. The following exclusion criteria were established: patients who underwent preoperative treatment, pneumonectomy, bilobectomy, segmentectomy, bronchoplasty, wedge resection, and segmentectomy.
The primary outcome was the development of BPF within 1 year after surgery confirmed via bronchoscopy, computed tomography, or reoperation. We investigated the association between the development of BPF and the following background factors: age, sex, performance status, smoking history, respiratory function, comorbidities (diabetes mellitus, interstitial pneumonia, ischemic heart disease, cerebrovascular disease, history of malignancy, and past gastric cancer surgery), preoperative blood tests results (serum albumin, C-reactive protein, and carcinoembryonic antigen), affected lobe, surgical approach, method of bronchial closure, extent of lymphadenectomy, histology, pathological nodal metastasis, pathological residual disease at the bronchial stump, and bronchial stump coverage. Furthermore, the subgroup analysis of men who underwent right lower lobectomy was performed to achieve more accurate selection of high-risk patients of BPF.
Surgical procedure
The standard procedure for lung cancer in the study period was lobectomy with lobe-specific selective lymph node dissection (3,15,16), and subcarinal lymphadenectomy was omitted for lung cancer in right upper lobe and left upper division segment. In right lower lobe lung cancer, upper mediastinal lymph node dissection was omitted unless metastasis was found in the hilar and subcarinal lymph nodes. The lobar bronchus was closed using staplers or interrupted suture. Thoracoscopic lobectomy for lung cancer was started in 2008 (17), and almost all lobectomies were performed via thoracoscopic surgery in recent years. However, the extent of lymph node dissection remains similar to that during open thoracotomy. To cover the bronchial stump, a pedicled intercostal muscle flap was used during open thoracotomy, whereas a free pericardial fat pad was used during thoracoscopic surgery.
Statistical analysis
Continuous variables were expressed as the mean ± standard deviation. Variables were compared and analyzed using Student’s t-test, Welch’s method, or χ2-test. The factors associated with the development of BPF were analyzed by performing multivariable logistic regression and least absolute shrinkage and selection operator (LASSO) regression analyses (18) using covariates with P<0.1 in univariable analysis. LASSO regression is one of the regularized regression models and minimizes the residual sum of squares subject to the sum of the absolute value of the coefficients being less than a constant. LASSO regression is a method that allows the selection of only those explanatory variables that influence the objective variable. The statistical analyses were performed in the R software environment (version 4.0.3; the R Foundation for Statistical Computing, Vienna, Austria). LASSO regression was performed using the “glmnet” package. P<0.05 indicated statistical significance.
Results
A flowchart of the patient selection process is provided in Figure 1. Among the 4,200 patients who underwent resection for lung cancer, 1,020 patients were excluded based on the exclusion criteria, and the remaining 3,180 patients were included in the study. There were no patients with incomplete data. Of 195 (6.1%) patients who underwent bronchial stump coverage, intercostal muscle flaps and pericardial fat pads were used in 137 and 58 patients, respectively. The incidence of BPF was 14 (0.44%) in 3,180 patients who underwent lobectomy without bronchoplasty nor preoperative treatment. The median interval from surgery to BPF onset was 21 days (range, 10–287 days). Eight patients (57%) underwent open window thoracostomy. Two of the 14 patients died of BPF within 6 months after its development (mortality rate, 14%). The characteristics of patients with BPF and without BPF are shown in Table 1. All 14 patients who developed BPF were men and had undergone right lower lobectomy. Other factors significantly associated with the development of BPF were older age, heavy smoking, obstructive ventilatory failure, interstitial pneumonia, history of malignancy, history of gastric cancer surgery, low serum albumin levels, and histology. The bronchi were closed using staplers in 99% of the cases. Mediastinal lymphadenectomy was omitted in 21% of the cases due to old age, comorbidities, noninvasive cancer, and so on. Multivariable logistic regression analysis using eleven variables revealed that having a history of gastric cancer surgery [odds ratio =7.02, 95% confidence interval (CI): 1.25–39.6, P=0.027] and high serum C-reactive protein levels (odds ratio =1.61, 95% CI: 1.02–2.55, P=0.040) were significantly associated with the development of BPF (Table 2). The odds ratios for male sex and right lower lobectomy were infinite.
Table 1
| Variables | BPF (n=14) | No BPF (n=3,166) | P |
|---|---|---|---|
| Age (years) | 72±11 | 67±10 | 0.042 |
| Sex | <0.001 | ||
| Male | 14 [100] | 1,693 [53] | |
| Female | 0 [0] | 1,473 [47] | |
| Performance status | 0.347 | ||
| 0 | 13 [93] | 3,070 [97] | |
| 1–3 | 1 [7] | 96 [3] | |
| Smoking history, pack-years | 54±32 | 24±31 | <0.001 |
| %VC | 112±21 | 114±16 | 0.632 |
| FEV1.0% | 68±11 | 74±9 | 0.019 |
| Comorbidities | |||
| Diabetes mellitus | 3 [21] | 426 [13] | 0.421 |
| Interstitial pneumonia | 3 [21] | 116 [4] | 0.014 |
| Ischemic heart disease | 1 [7] | 137 [4] | 0.463 |
| Cerebrovascular disease | 0 [0] | 115 [4] | 1.000 |
| History of malignancy | 8 [57] | 851 [27] | 0.029 |
| Gastric cancer surgery | 4 [29] | 93 [3] | 0.001 |
| Preoperative blood tests | |||
| Albumin (g/dL) | 4.0±0.6 | 4.2±0.3 | 0.036 |
| C-reactive protein (mg/dL) | 0.9±1.4 | 0.3±1.2 | 0.062 |
| Carcinoembryonic antigen (ng/mL) | 4.4±4.6 | 9.2±192.0 | 0.924 |
| Affected lobe | <0.001 | ||
| Right upper | 0 [0] | 1,146 [36] | |
| Right middle | 0 [0] | 248 [8] | |
| Right lower | 14 [100] | 588 [19] | |
| Left upper | 0 [0] | 688 [22] | |
| Left lower | 0 [0] | 496 [16] | |
| Thoracoscopic surgery | 11 [79] | 2,008 [63] | 0.281 |
| Stapler closure | 14 [100] | 3,123 [99] | 0.471 |
| Mediastinal lymphadenectomy | 11 [79] | 2,500 [79] | 1.000 |
| Histology | 0.001 | ||
| Adenocarcinoma | 6 [43] | 2,539 [80] | |
| Squamous cell carcinoma | 7 [50] | 342 [11] | |
| Others | 1 [7] | 285 [9] | |
| Nodal metastasis | 4 [29] | 514 [16] | 0.265 |
| Residual disease at the bronchial stump | 0 [0] | 4 [0.1] | 1.000 |
| Bronchial stump coverage | 1 [7] | 194 [6] | 0.588 |
Data are presented as mean ± standard deviation or number [frequency]. BPF, bronchopleural fistula; %VC, percent vital capacity; FEV1.0%, forced expiratory volume in one second percent.
Table 2
| Variables | Odds ratio | 95% CI | P |
|---|---|---|---|
| Male sex | ∞ | – | – |
| Right lower lobectomy | ∞ | – | – |
| Gastric cancer surgery | 7.02 | 1.25–39.6 | 0.027 |
| C-reactive protein | 1.61 | 1.02–2.55 | 0.040 |
| History of malignancy | 2.09 | 0.53–8.29 | 0.294 |
| Age | 1.01 | 0.93–1.10 | 0.830 |
| Smoking history | 1.00 | 0.99–1.00 | 0.530 |
| FEV1.0% | 0.96 | 0.90–1.02 | 0.203 |
| Interstitial pneumonia | 1.81 | 0.43–7.55 | 0.417 |
| Albumin | 2.03 | 0.26–15.7 | 0.497 |
| Not adenocarcinoma | 1.19 | 0.32–4.35 | 0.794 |
The symbol ∞ represents infinity. CI, confidence interval; FEV1.0%, forced expiratory volume in one second percent.
Table 3 shows the characteristics of patients with BPF and without BPF in a subgroup limited to men who underwent right lower lobectomy. The following three factors were significantly associated with the development of BPF in univariable analysis: history of malignancy, history of gastric cancer surgery, and high serum C-reactive protein. The risk for BPF was not significantly different between patients with and without subcarinal lymphadenectomy (4.5% vs. 3.6%, P=0.72). The incidence of BPF was lower in patients who underwent bronchial stump coverage than in those who did not, but the difference was not significant (1.0% vs. 5.7%, P=0.072). The multivariable logistic regression analysis revealed that high serum C-reactive protein levels (odds ratio =1.83, 95% CI: 1.17–2.87, P=0.009) and a history of gastric cancer surgery (odds ratio =5.52, 95% CI: 1.06–28.8, P=0.042) were significantly associated with the development of BPF in men who underwent right lower lobectomy (Table 4). Conversely, bronchial stump coverage was inversely associated with the development of BPF (odds ratio =0.06, 95% CI: 0.00–0.77, P=0.031). Table 5 presents the estimates obtained by logistic regression analysis and LASSO regression analysis. Three variables were selected using LASSO regression with λ = 0.02 that were significantly associated with BPF in the logistic regression analysis.
Table 3
| Variables | BPF (n=14) | No BPF (n=316) | P |
|---|---|---|---|
| Age (years) | 72±11 | 68±10 | 0.135 |
| Performance status | 0.384 | ||
| 0 | 13 [93] | 306 [97] | |
| 1–3 | 1 [7] | 10 [3] | |
| Smoking history, pack-years | 54±32 | 41±33 | 0.152 |
| %VC | 112±21 | 112±16 | 0.928 |
| FEV1.0% | 68±11 | 72±10 | 0.162 |
| Comorbidities | |||
| Diabetes mellitus | 3 [21] | 68 [22] | 1.000 |
| Interstitial pneumonia | 3 [21] | 38 [12] | 0.396 |
| Ischemic heart disease | 1 [7] | 18 [6] | 0.572 |
| Cerebrovascular disease | 0 [0] | 12 [4] | 1.000 |
| History of malignancy | 8 [57] | 84 [27] | 0.029 |
| Gastric cancer surgery | 4 [29] | 16 [5] | 0.007 |
| Preoperative blood tests | |||
| Albumin (g/dL) | 4.0±0.6 | 4.2±0.3 | 0.143 |
| C-reactive protein (mg/dL) | 0.9±1.4 | 0.3±0.8 | 0.005 |
| Carcinoembryonic antigen (ng/mL) | 4.4±4.6 | 6.2±20 | 0.744 |
| Thoracoscopic surgery | 11 [79] | 209 [66] | 0.400 |
| Mediastinal lymphadenectomy | 11 [79] | 235 [74] | 0.830 |
| Stapler closure | 14 [100] | 311 [98] | 0.520 |
| Histology | 0.076 | ||
| Adenocarcinoma | 6 [43] | 210 [67] | |
| Squamous cell carcinoma | 7 [50] | 73 [23] | |
| Others | 1 [7] | 33 [10] | |
| Nodal metastasis | 4 [29] | 49 [16] | 0.253 |
| Residual disease at the bronchial stump | 0 [0] | 0 [0] | 1.000 |
| Bronchial stump coverage | 1 [7] | 102 [32] | 0.072 |
Data are presented as mean ± standard deviation or number (frequency). BPF, bronchopleural fistula; %VC, percent vital capacity; FEV1.0%, forced expiratory volume in one second percent.
Table 4
| Variables | Odds ratio | 95% CI | P |
|---|---|---|---|
| C-reactive protein | 1.83 | 1.17–2.87 | 0.009 |
| Gastric cancer surgery | 5.52 | 1.06–28.8 | 0.042 |
| History of malignancy | 2.14 | 0.55–8.38 | 0.274 |
| Not adenocarcinoma | 1.74 | 0.50–6.12 | 0.387 |
| Bronchial stump coverage | 0.06 | 0.00–0.77 | 0.031 |
CI, confidence interval.
Table 5
| Variables | Logistic regression | LASSO regression (λ=0.02) | ||
|---|---|---|---|---|
| Estimates | P | Estimates | ||
| C-reactive protein | 0.60 | 0.009 | 0.24 | |
| Gastric cancer surgery | 1.71 | 0.042 | 1.32 | |
| History of malignancy | 0.76 | 0.274 | – | |
| Not adenocarcinoma | 0.55 | 0.387 | – | |
| Bronchial stump coverage | −2.81 | 0.031 | −0.26 | |
LASSO, least absolute shrinkage and selection operator.
Discussion
The results of the current study revealed that the incidence of BPF after lobectomy for lung cancer was 0.44% and was limited to men who underwent right lower lobectomy. Multivariable analysis in the subgroup limited to men who underwent right lower lobectomy revealed that high serum C-reactive protein levels and past gastric cancer surgery were significantly associated with BPF, whereas bronchial stump coverage was inversely associated with BPF.
BPF, which occurs in 0.6–4.4% of patients undergoing pulmonary resection and has a high mortality rate of 18–71%, has long been a serious threat (5-8,12,14). Algar et al. (19) and Jichen et al. (20) reported particularly high mortality rates in early BPF occurring within 1 month after surgery. In 1992, Asamura et al. (14) reported a 2.1% and 4.7% incidence rate of BPF after lung cancer surgery and pneumonectomy, respectively. The Japanese Joint Committee of Lung Cancer Registry reported that among the 18,973 patients who underwent lung cancer surgery including sublobar resection in 2010, 77 (0.4%) developed BPF within 30 days after surgery (21). In the current study, 0.44% of the patients who underwent lobectomy developed BPF within 1 year after surgery, with two patients dying of BPF within 6 months after BPF onset (mortality rate, 14%). Our study included cases after the year 2005, and both incidence and mortality rates seemed to have decreased compared to reports over the previous century.
Several studies have identified pneumonectomy, residual carcinomatous tissue at the bronchial stump, preoperative irradiation, and diabetes as risk factors for BPF (6,7,12,14). However, to determine the indication for preventive measures against BPF, evaluating the risk of BPF in patients who underwent standard lobectomy without preoperative therapy is important. In the current study, the incidence of BPF was limited to men who underwent right lower lobectomy, suggesting that men and patients who underwent right lower lobectomy were at increased risk of BPF. The multivariable analysis in the subgroup of men who underwent right lower lobectomy revealed that high level of serum C-reactive protein and a history of gastric cancer surgery were significantly associated with BPF, whereas bronchial stump coverage was inversely associated with BPF. Therefore, the bronchial stump coverage should be considered for patients with risk factors such as male sex, right lower lobectomy, high serum C-reactive protein levels, and history of gastrectomy.
Various factors may contribute to the development of BPF, including technical problems with suturing, staple malformation, ischemia, infection, and nutritional deficiencies. A stapler was used for bronchial closure in 99% of the cases, with no correlation observed between the bronchial closure method and the development of BPF. Satoh et al. (22) reported that male, smoking, diabetes mellitus, subcarinal lymphadenectomy, and pulmonary complications were risk factors for postoperative ischemic bronchitis. Considering the risk of ischemic bronchitis and distribution of lymph node metastasis (15,16,23), lobe-specific selective lymphadenectomy has been our standard procedure for primary lung cancer, which could have possibly preserved blood flow at the bronchial stump, thereby reducing the incidence of BPF. In this study, history of gastric cancer surgery was significantly associated with the development of BPF. Our previous study had revealed that upper gastrointestinal surgery was an independent factor associated with postoperative pulmonary complications after lung cancer surgery (24). Generally, surgery for upper gastrointestinal cancer can have a profound effect on postoperative nutritional status (25). Nutritional status may have some influence on the risk of BPF. Moreover, regurgitation of gastrointestinal contents following upper gastrointestinal surgery may have caused aspiration pneumonia and increased the risk of developing BPF.
The present study has some limitations worth noting. This was a retrospective, observational, and single-institution study. Moreover, the results of this study may not be applicable to institutions with different methods of bronchial closure or extent of lymphadenectomy. As this study reviewed surgeries performed over the span of 16 years, minor changes in the surgical techniques and perioperative management used during the study period may have affected the results. The proportion of patients with a history of gastric cancer surgery (3%) was higher than globally expected. The regional and institutional specificities might have contributed to the high prevalence of gastric cancer in this study cohort. If the study was conducted in a population with low prevalence of gastric cancer, no statistically significant difference might have been found. Furthermore, considering that only 14 relevant cases were recorded, the predictive performance of the multivariable logistic regression analysis might have been hampered. A larger, multi-institutional study should be conducted in the future. Although events per variables of >10 is a common criterion, some studies have shown that events per variables have a weak correlation with predictive accuracy (26,27). In this study, the reliability of the results was ensured by conducting subgroup analyses and comparing the results obtained by LASSO regression analysis with the results of logistic regression analysis.
Conclusions
The current study showed that men and patients who underwent right lower lobectomy were at increased risk of BPF. The risk was even higher when the patient had high serum C-reactive protein or a history of gastric cancer surgery. Bronchial stump coverage might be effective in patients at high risk of BPF.
Acknowledgments
We would like to thank Enago (www.enago.jp) for English language editing.
Funding: None.
Footnote
Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-22-1809/rc
Data Sharing Statement: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-22-1809/dss
Peer Review File: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-22-1809/prf
Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-22-1809/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. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The study was approved by the Institutional Review Board for Clinical Research of the Cancer Institute Hospital of the Japanese Foundation for Cancer Research on March 22, 2022 (referral No. 2021-GB-117) and individual consent for this retrospective analysis was waived.
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
- Yamamichi T, Ichinose J, Iwamoto N, et al. Correlation Between Smoking Status and Short-term Outcome of Thoracoscopic Surgery for Lung Cancer. Ann Thorac Surg 2022;113:459-65. [Crossref] [PubMed]
- Watanabe S, Asamura H, Suzuki K, et al. Recent results of postoperative mortality for surgical resections in lung cancer. Ann Thorac Surg 2004;78:999-1002; discussion 1002-3. [Crossref] [PubMed]
- Ichinose J, Kohno T, Fujimori S, et al. Locoregional control of thoracoscopic lobectomy with selective lymphadenectomy for lung cancer. Ann Thorac Surg 2010;90:235-9. [Crossref] [PubMed]
- Endo S, Ikeda N, Kondo T, et al. Model of lung cancer surgery risk derived from a Japanese nationwide web-based database of 78 594 patients during 2014-2015. Eur J Cardiothorac Surg 2017;52:1182-9. [Crossref] [PubMed]
- Uramoto H, Hanagiri T. The development of bronchopleural fistula in lung cancer patients after major surgery: 31 years of experience with 19 cases. Anticancer Res 2011;31:619-24. [PubMed]
- Suzuki M, Otsuji M, Baba M, et al. Bronchopleural fistula after lung cancer surgery. Multivariate analysis of risk factors. J Cardiovasc Surg (Torino) 2002;43:263-7. [PubMed]
- Sonobe M, Nakagawa M, Ichinose M, et al. Analysis of risk factors in bronchopleural fistula after pulmonary resection for primary lung cancer. Eur J Cardiothorac Surg 2000;18:519-23. [Crossref] [PubMed]
- Okuda M, Go T, Yokomise H. Risk factor of bronchopleural fistula after general thoracic surgery: review article. Gen Thorac Cardiovasc Surg 2017;65:679-85. [Crossref] [PubMed]
- Li SJ, Zhou XD, Huang J, et al. A systematic review and meta-analysis-does chronic obstructive pulmonary disease predispose to bronchopleural fistula formation in patients undergoing lung cancer surgery? J Thorac Dis 2016;8:1625-38. [Crossref] [PubMed]
- Li SJ, Fan J, Zhou J, et al. Diabetes Mellitus and Risk of Bronchopleural Fistula After Pulmonary Resections: A Meta-Analysis. Ann Thorac Surg 2016;102:328-39. [Crossref] [PubMed]
- Kawaguchi T, Watanabe S, Kawachi R, et al. The impact of residual tumor morphology on prognosis, recurrence, and fistula formation after lung cancer resection. J Thorac Oncol 2008;3:599-603. [Crossref] [PubMed]
- Hu XF, Duan L, Jiang GN, et al. A clinical risk model for the evaluation of bronchopleural fistula in non-small cell lung cancer after pneumonectomy. Ann Thorac Surg 2013;96:419-24. [Crossref] [PubMed]
- d'Amato TA, Ashrafi AS, Schuchert MJ, et al. Risk of pneumonectomy after induction therapy for locally advanced non-small cell lung cancer. Ann Thorac Surg 2009;88:1079-85. [Crossref] [PubMed]
- Asamura H, Naruke T, Tsuchiya R, et al. Bronchopleural fistulas associated with lung cancer operations. Univariate and multivariate analysis of risk factors, management, and outcome. J Thorac Cardiovasc Surg 1992;104:1456-64. [Crossref] [PubMed]
- Kuroda H, Ichinose J, Masago K, et al. Permissible Outcomes of Lobe-Specific Lymph Node Dissection for Elevated Carcinoembryonic Antigen in Non-Small Cell Lung Cancer. Medicina (Kaunas) 2021;57:1365. [Crossref] [PubMed]
- Aokage K, Yoshida J, Ishii G, et al. Subcarinal lymph node in upper lobe non-small cell lung cancer patients: is selective lymph node dissection valid? Lung Cancer 2010;70:163-7. [Crossref] [PubMed]
- Mun M, Ichinose J, Matsuura Y, et al. Video-assisted thoracoscopic surgery lobectomy via confronting upside-down monitor setting. J Vis Surg 2017;3:129. [Crossref] [PubMed]
- Tibshirani R. Regression shrinkage and selection via the lasso. J R Stat Soc Series B Stat Methodol 1996;58:267-88.
- Algar FJ, Alvarez A, Aranda JL, et al. Prediction of early bronchopleural fistula after pneumonectomy: a multivariate analysis. Ann Thorac Surg 2001;72:1662-7. [Crossref] [PubMed]
- Jichen QV, Chen G, Jiang G, et al. Risk factor comparison and clinical analysis of early and late bronchopleural fistula after non-small cell lung cancer surgery. Ann Thorac Surg 2009;88:1589-93. [Crossref] [PubMed]
- Okami J, Shintani Y, Okumura M, et al. Demographics, Safety and Quality, and Prognostic Information in Both the Seventh and Eighth Editions of the TNM Classification in 18,973 Surgical Cases of the Japanese Joint Committee of Lung Cancer Registry Database in 2010. J Thorac Oncol 2019;14:212-22.
- Satoh Y, Okumura S, Nakagawa K, et al. Postoperative ischemic change in bronchial stumps after primary lung cancer resection. Eur J Cardiothorac Surg 2006;30:172-6. [Crossref] [PubMed]
- Ichinose J, Murakawa T, Hino H, et al. Prognostic impact of the current Japanese nodal classification on outcomes in resected non-small cell lung cancer. Chest 2014;146:644-9. [Crossref] [PubMed]
- Yamamichi T, Ichinose J, Tamagawa S, et al. Impact of previous upper gastrointestinal cancer surgery on complications after lobectomy for lung cancer. J Thorac Dis 2022;14:3811-8. [Crossref] [PubMed]
- Baker M, Halliday V, Williams RN, et al. A systematic review of the nutritional consequences of esophagectomy. Clin Nutr 2016;35:987-94. [Crossref] [PubMed]
- van Smeden M, Moons KG, de Groot JA, et al. Sample size for binary logistic prediction models: Beyond events per variable criteria. Stat Methods Med Res 2019;28:2455-74. [Crossref] [PubMed]
- Ogundimu EO, Altman DG, Collins GS. Adequate sample size for developing prediction models is not simply related to events per variable. J Clin Epidemiol 2016;76:175-82. [Crossref] [PubMed]
