Establishment and validation of a cough prediction model for lung cancer survivors after pulmonary resection based on a systematic review and meta-analysis of 23 cohort studies

Introduction

Postoperative cough (POC) is a frequent complication following pulmonary resection (1,2). However, it has received less attention compared to other postoperative complications (2). Its incidence is estimated to range from 25% to 50% (1,3,4). POC is defined as a new-onset dry cough lasting more than two weeks following lung surgery, provided chest imaging reveals no abnormalities and other potential causes are excluded (2,4). These include postnasal drip syndrome, bronchial asthma, or the use of oral angiotensin-converting enzyme inhibitors (ACEIs) (3,5). The median duration of POC is approximately 180 days (range, 14–720 days), with some cases persisting for two years or more (2,4). Persistent coughing can exacerbate postoperative pain, significantly hinder recovery, and disrupt sleep. Furthermore, it reduces patients’ quality of life, imposing substantial physical and psychological burdens (6,7). In severe cases, POC may lead to complications such as rib fractures, severe vomiting, or pneumothorax (8).

POC is influenced by multiple factors (2,9). Current research indicates that its occurrence is primarily associated with excessive surgical stimulation of the trachea, vagal nerve injury, peritracheal lymph node dissection, and anesthesia (2,3,10-12). However, the factors identified vary across studies, and their findings are sometimes inconsistent. For instance, studies by Lin et al. (9) and Liu et al. (13) suggest an association between female gender and POC, whereas studies by Mu et al. (2) and Lai et al. (14) found no such correlation. These discrepancies may arise from small sample sizes in cohort studies and inconsistent diagnostic criteria for POC, potentially limiting the generalizability of the results.

Consequently, based on a systematic review of pertinent cohort studies, we conducted a meta-analysis to identify risk factors for POC following pulmonary resection. In contrast to the conventional method of employing data from individual cohort studies to develop predictive models, we utilized the pooled results from the meta-analysis, which offers larger sample sizes, to construct a risk prediction model for POC. This approach seeks to provide robust evidence-based support for clinicians in identifying and managing lung cancer survivors at high risk of POC. We present this article in accordance with the TRIPOD (15) and PRISMA reporting checklists (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-572/rc).

Methods

The study aimed to identify key risk factors for POC through a systematic review and meta-analysis and to develop a predictive model based on the meta-analysis findings. The study design consisted of two components: a systematic review with meta-analysis and an observational study. The model was derived from the meta-analysis and subsequently trained and validated using data from a Chinese cohort included in the observational study. This study was conducted in accordance with the ethical principles of the Declaration of Helsinki and its subsequent amendments and its subsequent amendments. The study protocol was registered in PROSPERO (No. CRD42022360462) and reviewed and approved by the Institutional Review Board (IRB) of West China Hospital, Sichuan University (No. 2024-1177). The IRB waived the requirement for individual informed consent due to the retrospective nature of this study, which used anonymized clinical data without compromising patient privacy.

Study populations

Derivation cohort

Our model was derived from a systematic review and meta-analysis. A comprehensive search of databases, including PubMed, Embase, Cochrane Library, Web of Science, WANFANG DATA (www.wanfangdata.com.cn), and CNKI (www.cnki.net), was conducted up to October 2024, employing the following search terms: (cough [Title/Abstract]) AND (risk factor[Title/Abstract]) AND ((((prospective[Title/Abstract]) OR (retrospective[Title/Abstract])) OR (cross-sectional[Title/Abstract])) OR (cohort[Title/Abstract])). In total, 23 cohort studies were included in the analysis (Figure 1A). The quality of all included studies was assessed using the Newcastle-Ottawa Scale, while randomized controlled trials were evaluated using the Cochrane Risk of Bias tool. Detailed descriptions of the literature search strategy, inclusion criteria, quality assessment methods, and data extraction procedures are available in the Appendix 1.

Figure 1 Schematic overview of study methodology. (A) Flowchart of literature search and study selection process. (B) Flow diagram of patient cohort allocation for model development and validation.

Training and validation cohort

A total of 5,570 patients who underwent surgical treatment for lung cancer at the Department of Thoracic Surgery, West China Hospital, Sichuan University, between August 2020 and June 2023, were included in this study. The inclusion criteria were as follows: (I) undergoing pulmonary resection; (II) having complete data with sufficient baseline information for model variables; (III) not having received neoadjuvant or adjuvant therapy; (IV) presenting with a clear diagnosis of POC, characterized by: (i) persistent cough symptoms lasting longer than two weeks following pulmonary resection; (ii) the exclusion of preoperative factors known to cause cough, such as postnasal drip syndrome, chronic obstructive pulmonary disease (COPD), bronchial asthma, and drug-related etiologies; (iii) chest X-ray or CT demonstrating only normal postoperative changes; and (V) without preoperative cough. Cough was assessed using the Leicester Cough Questionnaire in Mandarin Chinese (16). The inclusion and exclusion process is illustrated in Figure 1B.

Statistical analysis

Meta-analysis

The pooled odds ratios (ORs) and 95% confidence intervals (CIs) were calculated using Stata 15.0 (StataCorp, College Station, TX, USA) via the inverse variance-weighted method. Statistical significance was defined as a P value <0.05. Heterogeneity was assessed using the Higgins I² statistic. An I² value exceeding 50% or a P value below 0.05 indicated significant heterogeneity among the included studies, in which case a random-effects model was applied for pooling ORs; otherwise, a fixed-effects model was utilized. Sensitivity analysis was performed using the leave-one-out method to evaluate the robustness of the pooled effect size. In the event that the sensitivity analysis results were inconsistent with the overall pooled results (95% CI), a secondary analysis was conducted employing the trim-and-fill method. Egger’s test was used to evaluate publication bias, with a P value <0.05 considered indicative of significant publication bias. Considering the potential influence of video-assisted thoracoscopic surgery (VATS) and thoracotomy on POC, a subgroup analysis was performed to assess the risk factors for POC in patients undergoing VATS.

Model derivation

A simplified classification scoring system was developed based on the methodologies proposed by Sullivan et al. (17) and Jiang et al. (18). The model incorporated factors that demonstrated statistical significance in the systematic review and meta-analysis. To ensure its clinical applicability and robustness in predicting POC, the included risk factors were required to meet the following criteria: (I) statistical significance in both the meta-analysis and sensitivity analysis; (II) inclusion in at least three pooled studies; (III) preoperative cough was excluded to avoid its potential confounding effect on POC; and (IV) variables with known statistical or clinical collinearity—such as mediastinal lymph node dissection and subcarinal lymph node dissection—were not included simultaneously in the model. Data for the model were derived from the pooled sensitivity analysis results of either the overall analysis or a subgroup analysis, contingent on the degree of heterogeneity among the original studies (18). The β-coefficients for each risk factor were calculated based on their corresponding ORs and 95% CIs. These β-coefficients were then multiplied by 10 and rounded to one decimal place (values between 0 and 0.3 were rounded to 0, 0.4 and 0.6 to 0.5, 0.7 and 0.9 to 1.0). Finally, the score for each factor in the model was determined based on the results of the systematic review and meta-analysis. The total score was calculated as the sum of the scores for all included risk factors. A difference was considered statistically significant at P<0.05.

Model training and validation

To further validate the model’s generalizability and explore its clinical application, the study cohort (n=5,570) was randomly divided into a training cohort (n=3,899) and a validation cohort (n=1,671) using a 7:3 ratio. Guided by the derivation model, the total score for each patient was calculated using baseline data from the training cohort and validation cohort. Subsequently, logistic regression was employed to assess the relationship between total scores and the incidence of POC. The area under the curve (AUC) of the receiver operating characteristic (ROC) curve was utilized to evaluate the model’s discriminatory ability. A calibration curve was employed to analyze the difference between the model’s predicted and the observed values. Decision curve analysis (DCA) and clinical impact curves (CIC) were utilized to assess the model’s clinical net benefit and the corresponding threshold ranges. Subsequently, a decision tree algorithm was applied to determine the optimal cut-off values, categorizing the study population into five risk groups: low, low-middle, middle, high-middle, and high. The low-risk group served as the reference, and the corresponding ORs and 95% CIs for the remaining groups were calculated. All analyses were performed using R version 4.4.1 (R Foundation for Statistical Computing, Vienna, Austria) and SPSS version 26.0 (IBM Corp., USA).

Results

An initial search identified 2,402 studies. Following the removal of 155 duplicate records, 2,106 irrelevant studies were excluded based on the inclusion criteria during title and abstract screening. Subsequently, 22 additional studies were excluded for not being cohort studies, and 97 more were excluded after full-text review. Ultimately, 23 studies (1-3,9,11,19-36) encompassing 5,360 patients were included in the analysis. The study selection process is depicted in Figure 1A.

The included studies comprised 21 retrospective studies, 1 prospective study, and 1 randomized controlled trial. The mean age of patients ranged from 57.88 to 68.44 years, with 2,662 (49.7%) being male. A total of 4,109 patients (76.7%) underwent VATS, and 1,824 patients (33.8%) experienced POC. All study populations were from Asia (China and Japan). Based on the Newcastle-Ottawa Scale (NOS), the median quality score of the included studies was 7 (range, 6–8). Detailed baseline characteristics and NOS scores of the included studies are presented in Table S1. The definitions of POC employed in each study are outlined in Table S2.

Among the 23 included studies, 17 risk factors for POC were reported: age ≥60 years, female sex, body mass index (BMI) <24 kg/m², smoking history, preoperative cough, COPD, upper lobe surgery, right lung surgery, lobectomy, subcarinal lymph node dissection, mediastinal lymph node dissection, closure of the bronchial stump with a stapler, peritracheal lymph node dissection, postoperative acid reflux, preoperative respiratory training, anesthesia time, and tracheal intubation time (Table S3). The meta-analysis identified 12 risk factors that were statistically associated with POC following pulmonary resection. These included age ≥60 years (OR: 0.461, 95% CI: 0.294–0.721, P=0.001), BMI <24 kg/m² (OR: 1.203, 95% CI: 1.064–1.361, P=0.003), preoperative cough (OR: 2.177, 95% CI: 1.156–4.099, P=0.016), right lung surgery (OR: 2.330, 95% CI: 1.844–2.945, P<0.001), lobectomy (OR: 2.718, 95% CI: 1.666–4.434, P<0.001), subcarinal lymph node dissection (OR: 3.441, 95% CI: 1.713–6.911, P=0.001), mediastinal lymph node dissection (OR: 3.489, 95% CI: 2.066–5.892, P<0.001), closure of the bronchial stump with a stapler (OR: 5.190, 95% CI: 1.788–15.070, P=0.002), peritracheal lymph node dissection (OR: 2.079, 95% CI: 1.196–3.614, P=0.009), postoperative acid reflux (OR: 5.161, 95% CI: 3.066–8.686, P<0.001), preoperative respiratory training (OR: 0.223, 95% CI: 0.135–0.370, P<0.001), and tracheal intubation time ≥172 min (OR: 2.326, 95% CI: 1.302–4.155, P=0.004). These results are depicted in Figure S1. Sensitivity analysis revealed that COPD (OR: 2.379, 95% CI: 1.726–3.033), upper lobe surgery (OR: 3.971, 95% CI: 1.655–6.287), and anesthesia time (OR: 1.095, 95% CI: 1.018–1.171) were also identified as risk factors for POC (Table S4). However, after applying the trim-and-fill method to account for potentially unpublished studies, secondary meta-analysis showed that COPD (OR: 0.885, 95% CI: 0.196–4.739), upper lobe resection (OR: 1.004, 95% CI: 0.482–2.093), and anesthesia time (OR: 1.001, 95% CI: 0.965–1.038) were no longer statistically associated with POC (Table S4).

Given that the inclusion of thoracotomy cases might contribute to heterogeneity, a subgroup analysis was performed focusing exclusively on patients undergoing VATS. Among the 15 risk factors analyzed (Figure S1), the results indicated that age ≥60 years, BMI <24 kg/m², right lung surgery, lobectomy, subcarinal lymph node dissection, postoperative acid reflux, preoperative respiratory training, and tracheal intubation time ≥172 minutes were significantly associated with POC (Table S4).

Model derivation

Ultimately, eight of the 12 identified risk factors were selected to construct the prediction model based on the aforementioned protocol (Figure 2). These risk factors and their corresponding ORs from the sensitivity analyses were as follows: age ≥60 years (OR: 0.400), BMI <24 kg/m² (OR: 1.217), right lung surgery (OR: 2.576), lobectomy (OR: 3.691), subcarinal lymph node dissection (OR: 3.661), peritracheal lymph node dissection (OR: 3.294), postoperative acid reflux (OR: 6.027), and preoperative respiratory training (OR: 0.230). The OR for lobectomy was derived from subgroup sensitivity analysis results, which exhibited lower heterogeneity. Other factors, such as mediastinal lymph node dissection, closure of the bronchial stump with a stapler, and tracheal intubation time ≥172 minutes, were excluded because fewer than three studies reported data on these factors (Table S4). Consequently, a simple POC risk prediction model was developed (Table 1) as follows: age (years, <60 scores 0; ≥60 scores −9), BMI (kg/m², ≥24 scores 0; <24 scores 2), right lung surgery (no scores 0; yes scores 9.5), lobectomy (no scores 0; yes scores 13), subcarinal lymph node dissection (no scores 0; yes scores 13), peritracheal lymph node resection (no scores 0; yes scores 12), postoperative acid reflux (absent scores 0; present scores 18), preoperative respiratory training (no scores 0; yes scores −14.5). The definitions of each risk factor, the number of included original studies, sample size, pooled ORs, β-coefficients, and corresponding risk scores are provided in Table S5. This model is recommended for assessing the risk of POC in Asian patients undergoing pulmonary resection, particularly those undergoing VATS.

Figure 2 Risk factors with corresponding pooled ORs and 95% CIs of sensitivity analysis included in the POC risk prediction model. CI, confidence interval; OR, odds ratio; POC, postoperative cough.

Model training and validation

In the training cohort, 1,475 patients (37.8%) experienced POC. The mean age was 58.77±9.94 years, and 49.5% were female, with a mean BMI of 23.23±3.00 kg/m². In the validation cohort, 634 patients (37.9%) experienced POC. The mean age was 58.90±10.36 years, and 48.8% were female, with a mean BMI of 23.31±2.99 kg/m². The baseline characteristics of the training and validation cohorts are presented in Table S6.

The model’s visualization is shown in Figure 3A,3B, and the ROC curves for the prediction model in the training and validation cohorts are shown in Figure 3C. The AUCs were 0.770 (95% CI: 0.755–0.786) in the training cohort and 0.758 (95% CI: 0.735–0.782) in the validation cohort. Calibration curves demonstrated good agreement between the predicted and observed probabilities of POC (Figure 4A,4B). DCA showed a positive net benefit within a threshold range of 0.0 to 0.8, indicating the model’s practical clinical value (Figure 4C,4D). CIC analysis revealed that beyond a threshold probability of 0.6, the model’s predicted POC incidence closely matched the actual incidence in the high-score population. Furthermore, as the threshold probability increased, the agreement between predicted and actual incidences improved, indicating the model’s high clinical predictive validity (Figure 4E,4F).

Figure 3 Visualization and receiver operating characteristic curve of the POC predictive model. (A) Correlation between risk scores and the probability of POC. (B) Risk nomogram for the POC predictive model. (C) Receiver operating characteristic curves demonstrating the performance of the POC predictive model in the training and validation cohorts. AUC, area under the curve; CI, confidence interval; POC, postoperative cough.

Figure 4 Calibration curves, decision curve analysis, and clinical impact curves for the POC predictive model in the training and validation cohorts. (A,B) Calibration curves assess the agreement between predicted probabilities and observed event probabilities, reflecting the calibration performance of the model. A closer alignment of the apparent curve with the ideal curve indicates better calibration. (C,D) Decision curve analysis evaluates the net clinical benefit of the model across different risk thresholds, aiding in the selection of optimal thresholds to maximize decision-making outcomes. (E,F) Clinical impact curves demonstrate the practical utility of the model in clinical applications by illustrating its potential to improve real-world outcomes under specific high risk thresholds. POC, postoperative cough.

Based on the frequency of POC, the decision tree algorithm stratified the study population into five risk groups (low, low-middle, middle, high-middle, and high) with cutoff values of 15, 26, 36.5, and 40.5. In the training cohort, the number and proportion of patients experiencing POC in the five groups were: 117 (13.8%), 227 (31.2%), 270 (31.7%), 434 (55.6%), and 427 (61.8%) (Figure S2). In the validation cohort, the number and proportion of patients experiencing POC in the five groups were: 53 (14.8%), 108 (31.2%), 117 (33.8%), 179 (56.8%), and 177 (58.0%) (Figure S2). Compared to the low-risk group, the high-middle risk group had significantly higher ORs for POC incidence (OR: 7.81 in the training cohort and 7.60 in the validation cohort). Similarly, the high-risk group had markedly higher ORs (OR: 10.11 in the training cohort and 7.98 in the validation cohort) (Figure S2).

Discussion

POC, though common after thoracic surgery, often receives insufficient attention from clinicians (2). While less critical than complications like hemorrhage or pulmonary infection, and pulmonary embolism, serious POC can cause chest pain, vomiting, fractures, and pneumothorax, significantly affecting patient recovery and quality of life (6-8). Previous studies on POC risk factors have shown varied results due to different sample sizes (2,3,5,9). Meta-analysis allows us to combine multiple studies, increasing statistical power and improving research reliability (37). Our meta-analysis of 23 cohort studies identified 12 key risk factors for POC: age ≥60 years, BMI <24 kg/m², preoperative cough, right lung surgery, lobectomy, subcarinal lymph node dissection, mediastinal lymph node dissection, closure of bronchial stump with stapler, peritracheal lymph node dissection, postoperative acid reflux, preoperative respiratory training, and tracheal intubation time ≥172 min. Following principles of scientific practicality and evidence-based medicine, we integrated meta-analysis findings and selected 8 out of 12 risk factors to develop a risk prediction model. The model underwent external training and validation using single-center data, demonstrating good discriminative ability with AUCs of 0.770 (95% CI: 0.755–0.786) in the training cohort and 0.758 (95% CI: 0.735–0.782) in the validation cohort. This model can help doctors identify high-risk patients early and implement appropriate preventive measures.

Our study identified age as a risk factor for POC. Unlike the meta-analysis by Wu et al. (4), we further analyzed age as a dichotomous variable, taking into account different age cutoff values and variable types across included studies. We found that patients under 60 years old who underwent thoracoscopic surgery were more likely to develop POC. This may be attributed to younger patients having more sensitive receptors to physical stimulation (38,39).

Studies show conflicting results regarding the impact of gender on POC (2,9,14,40). Some suggest females have a higher incidence, potentially due to several factors (41,42): (I) estrogen may sensitize vagal C-afferent nerve fibers in the trachea, bronchi, and carina by activating transient receptor potential (TRP) V1/A1 channels. (II) Females often exhibit stronger immune responses, increasing the risk of severe postoperative infections or inflammation, which can trigger cough when chemical receptors are sensitized. (III) Females may be more prone to psychological issues such as anxiety and depression after surgery, which can worsen coughing symptoms (43).

However, other studies report no significant gender differences. Lai et al. (14) found a roughly equal gender distribution among Chinese patients with chronic cough, while Mu et al. (2) and Xie et al. (40) also found no statistical gender difference in POC incidence. Similarly, our study showed that female gender was not a risk factor in either the VATS and thoracotomy group (OR: 1.565, 95% CI: 0.797–3.071) or the VATS-only group (OR: 1.211, 95% CI: 0.400–3.663). Given the heterogeneity of findings, large-scale, multinational studies are needed to clarify whether gender differences in POC truly exist.

Studies (9,18) have identified surgery-related factors as risk factors for POC, including the surgical approach (lobectomy, segmentectomy, wedge resection), lung resection location, and the extent of lymph node dissection. Mechanisms linking surgical factors to POC include: (I) local inflammation of lung tissue and surrounding nerves, (II) postoperative airway distortion, (III) irritation from surgical scars or foreign materials like sutures and stapler cartridges, and (IV) local pleuritis and pleural effusion. A prospective study by Pan et al. (3). analyzing 135 lung surgery patients identified lobectomy as a risk factor for POC through multivariate regression analysis. Similarly, Lin et al.’s (44) longitudinal study found that patients undergoing thoracoscopic sublobar resection had better health-related quality of life scores and faster cough recovery compared to those undergoing thoracoscopic lobectomy. Consistent with these findings, our study showed that lobectomy posed a higher risk of POC than sublobar resection (OR: 2.718, 95% CI: 1.666–4.434). We suggest that the larger pulmonary parenchyma resection in lobectomy may cause bronchial abnormalities due to residual lung expansion and diaphragm elevation. These structural changes, such as bronchial distortion and obstruction, likely increase airway sensitivity, leading to POC (45).

The incidence of POC varies with the location of lung resection. Retrospective studies by Xie et al. (40) and Cui et al. (30). found that right lung and upper lobe resections were associated with a higher risk of POC, aligning with our meta-analysis. Seok et al. (46) and Bu et al. (47). attributed the increased incidence in right upper lobectomy to bronchial deformation and distortion caused by inferior pulmonary ligament release. Compared to lower lobectomy, upper lobectomy typically creates a larger residual cavity, triggers more severe pleural reactions, and increases pleural effusion, all contributing to a higher risk of cough. Additionally, the higher cough incidence after right lung surgery may be linked to differences in mediastinal lymph node dissection techniques and anatomical variations in bronchial bifurcation angles between the right and left lungs.

Lymph node dissection is a standard procedure in lung cancer surgery (48), but studies by Sawabata et al. (1) and Lin et al. (9). have identified mediastinal and peritracheal lymph node dissection as risk factors for increased POC. Similarly, our study found that mediastinal, subcarinal, and peritracheal lymph node dissection were associated with a higher risk of POC. Cough receptors, primarily located in the larynx, trachea, bronchi, and carina, may be affected during lymph node dissection, as energy devices can damage vagal nerve fibers and disrupt cough reflex pathways. Additionally, one study (10) found that lymph node sampling resulted in a lower incidence of POC compared to complete lymph node dissection. Huang et al. (49). reported that filling the residual cavity with fat tissue after lymph node dissection reduced POC, though the mechanisms remain unclear and require further investigation.

POC after pulmonary resection is influenced by not only surgical factors but also anesthetic factors. A randomized controlled trial by Xu et al. (50). found that endotracheal intubation significantly increased the incidence of POC compared to laryngeal mask anesthesia. Similarly, studies by Chen et al. (12). and Hung et al. (51). showed that maintaining spontaneous breathing during anesthesia reduced POC and accelerated recovery. Our study also found that endotracheal intubation lasting longer than 172 minutes significantly increased the risk of POC, likely due to compression, irritation, and potential injury to the bronchi caused by the intubation.

Anesthesia duration has been linked to POC, as longer durations can result in prolonged endotracheal intubation, increased airway irritation, and greater exposure to residual anesthetic gases. Studies by Lin et al. (9). and Pan et al. (3). found a significant increase in POC incidence when anesthesia duration exceeded 164 and 153 minutes, respectively. However, our findings showed no association between anesthesia duration (per 1-minute increment) and POC (OR: 1.034, 95% CI: 0.996–1.074). We hypothesize that this discrepancy may result from the differing cutoff values for anesthesia time in previous studies, complicating meta-analysis. Additionally, after excluding five studies that treated anesthesia duration as a dichotomous variable, our meta-analysis of studies using continuous variables showed evidence of publication bias (Egger’s test P=0.017). We speculate that the relationship between anesthesia duration and POC may be non-linear, requiring confirmation through future large-scale studies.

Wu et al.’s (4) meta-analysis found that patients with preoperative cough had a higher likelihood of developing POC, likely due to their pre-existing airway hyperreactivity. This finding aligns with our meta-analysis results. Therefore, administering medications that reduce airway hyperreactivity before surgery might help lower the incidence of POC. Additionally, some researchers have explored preoperative respiratory training as a preventive measure against POC. Dong et al. (23). divided study participants into two groups based on whether they received preoperative respiratory training and found that such training served as a protective factor against POC, which was consistent with both Yu et al.’s findings (19) and our study results. Preoperative respiratory training can enhance respiratory muscle strength, alleviate muscle tension, promote recovery of breathing capacity, facilitate timely clearance of airway secretions, and ultimately reduce the occurrence of cough.

In our study, postoperative acid reflux was identified as a risk factor for POC, consistent with several previous studies (1,3,19). This association can be explained by two mechanisms: First, gastroesophageal reflux stimulates the vagus nerve from the esophagus to the bronchi, triggering the release of neuropeptides A, B, and substance P, which leads to neurogenic inflammation and bronchospasm, ultimately promoting the cough reflex (3,4,52). Second, acidic reflux contents can directly enter the bronchi, causing irritative cough. Therefore, proper positioning during bed rest and dietary care may be helpful for preventing postoperative reflux (52). For patients presenting with preoperative malignancy, retching, heartburn, or chest pain, the use of a reflux diagnostic questionnaire is recommended to assess the severity of reflux (3). Furthermore, a multidisciplinary approach should be considered to evaluate whether prophylactic use of proton pump inhibitors and prokinetic agents is necessary before surgery to prevent POC in patients with gastroesophageal reflux.

Among the aforementioned risk factors, most can be controlled and managed during the preoperative and intraoperative periods, except for age. This controllability is crucial for preventing POC. Based on our meta-analysis, we developed a predictive model that assigns numerical values to different risk factors. This model allows quick assessment of a patient’s risk for POC, considering preoperative surgical and anesthetic plans. After external validation, the model showed strong performance across various measures, including the ROC curve, calibration curve, DCA, and CIC, with good alignment between predicted and actual values, indicating strong predictive accuracy.

In this study, we visualized the model using nomograms and curves, and simplified the formulas into scoring tables for easier interpretation. Clinical staff can use this model to calculate both current risk and potential risk of POC after eliminating high-risk factors, highlighting the benefits of risk factor management. This supports more effective preoperative health education. Surgeons and anesthesiologists can also use the model to adjust surgical and anesthetic plans, enhancing its clinical value.

Our study has several limitations. First, we did not account for the time-to-event effect, as follow-up data for POC were unavailable in both the training and validation cohorts. As a result, the model predicts whether POC will occur, but not when. Future studies should prospectively collect data on the timing and status of POC to enhance risk analysis. Second, the model was trained and validated using data from a single center, which may limit its generalizability. Future research should include prospective, large-scale, multi-center data for improved model robustness. Third, while 76% of our cohort underwent VATS, this homogeneity prevented meaningful analysis of surgical approach as a predictor. Future multicenter studies should compare cough incidence between stapled vs. hand-sewn bronchial closures. Fourth, some risk factors, such as age ≥60 years (I²=82.7%), peritracheal lymph node dissection (I²=81.8%), and lobectomy (I²=75.9%), showed high heterogeneity across the original studies. Although, we addressed this with sensitivity analysis, subgroup analysis, and trim-and-fill analysis, this heterogeneity could still affect the model’s predictive accuracy. Lastly, the meta-analysis primarily involved Asian populations (from China and Japan), which may limit the model’s applicability to other global populations.

Conclusions

This study provides evidence-based insights into the risk factors for POC and presents a validated predictive model for identifying high-risk patients. The model’s strong predictive accuracy and clinical utility enable targeted preoperative and postoperative interventions, such as respiratory training and surgical modifications, to mitigate POC risk. These findings offer a practical tool for improving postoperative management, with potential to enhance recovery and quality of life in patients undergoing pulmonary resection. Further validation in diverse populations is recommended to extend its applicability.

Acknowledgments

We greatly appreciate the assistance of the staff of the Department of Thoracic Surgery, West-China Hospital, Sichuan University, and thank them for their efforts.

Footnote

Reporting Checklist: The authors have completed the PRISMA and TRIPOD reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-572/rc

Data Sharing Statement: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-572/dss

Peer Review File: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-572/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-572/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work, ensuring that questions related to the accuracy or integrity of any part of the study are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Board of West China Hospital, Sichuan University (No. 2024-1177), 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

1. Sawabata N, Maeda H, Takeda S, et al. Persistent cough following pulmonary resection: observational and empiric study of possible causes. Ann Thorac Surg 2005;79:289-93. [Crossref] [PubMed]
2. Mu T, Li J, Huang Q, et al. Characteristics and Risk Factors for Persistent Cough After Pulmonary Resection. Ann Thorac Surg 2023;115:1337-43. [Crossref] [PubMed]
3. Pan LY, Peng LP, Xu C, et al. Predictive factors of cough after uniportal video-assisted thoracoscopic pulmonary resection. J Thorac Dis 2020;12:5958-69. [Crossref] [PubMed]
4. Wu Y, Song W, Zhu D, et al. Risk factors for cough after pulmonary resection. World J Surg Oncol 2023;21:348. [Crossref] [PubMed]
5. Li X, Li X, Zhang W, et al. Factors and potential treatments of cough after pulmonary resection: A systematic review. Asian J Surg 2021;44:1029-36. [Crossref] [PubMed]
6. Yang P, Cheville AL, Wampfler JA, et al. Quality of life and symptom burden among long-term lung cancer survivors. J Thorac Oncol 2012;7:64-70. [Crossref] [PubMed]
7. Young EC, Smith JA. Quality of life in patients with chronic cough. Ther Adv Respir Dis 2010;4:49-55. [Crossref] [PubMed]
8. Irwin RS, Glomb WB, Chang AB. Habit cough, tic cough, and psychogenic cough in adult and pediatric populations: ACCP evidence-based clinical practice guidelines. Chest 2006;129:174S-9S. [Crossref] [PubMed]
9. Lin R, Che G. Risk factors of cough in non-small cell lung cancer patients after video-assisted thoracoscopic surgery. J Thorac Dis 2018;10:5368-75. [Crossref] [PubMed]
10. Wu X, Xing H, Chen P, et al. Lymph Node Dissection Is a Risk Factor for Short-Term Cough after Pulmonary Resection. Curr Oncol 2022;29:294-307. [Crossref] [PubMed]
11. Gu S, Wang W, Wang X, et al. Effects of Preserving the Pulmonary Vagus Nerve Branches on Cough After Pneumonectomy During Video-Assisted Thoracic Surgery. Front Oncol 2022;12:837413. [Crossref] [PubMed]
12. Chen Z, Dong Q, Liang L. Effect of different thoracic anesthesia on postoperative cough. J Thorac Dis 2018;10:3539-47. [Crossref] [PubMed]
13. Liu W, Wu Q, Mao B, et al. Gender difference in the association between cough severity and quality of life among patients with postinfectious cough. Health Qual Life Outcomes 2021;19:34. [Crossref] [PubMed]
14. Lai K, Long L, Yi F, et al. Age and Sex Distribution of Chinese Chronic Cough Patients and Their Relationship With Capsaicin Cough Sensitivity. Allergy Asthma Immunol Res 2019;11:871-84. [Crossref] [PubMed]
15. Collins GS, Reitsma JB, Altman DG, et al. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement. BMJ 2015;350:g7594. [Crossref] [PubMed]
16. Lin R, Che G. Validation of the Mandarin Chinese version of the Leicester Cough Questionnaire in non-small cell lung cancer patients after surgery. Thorac Cancer 2018;9:486-90. [Crossref] [PubMed]
17. Sullivan LM, Massaro JM, D'Agostino RB Sr. Presentation of multivariate data for clinical use: The Framingham Study risk score functions. Stat Med 2004;23:1631-60. [Crossref] [PubMed]
18. Jiang W, Wang J, Shen X, et al. Establishment and Validation of a Risk Prediction Model for Early Diabetic Kidney Disease Based on a Systematic Review and Meta-Analysis of 20 Cohorts. Diabetes Care 2020;43:925-33. [Crossref] [PubMed]
19. Yu L, Zhang Z, Wang Z, et al. Analysis of Factors Associated with Cough Persistence After Thoracoscopic Lung Cancer Resection in Elderly Lung Cancer Patients and Discussion of Prevention Strategies. Altern Ther Health Med 2025;31:257-61. [PubMed]
20. He J, Cui C, Jia Y, et al. Influencing factors of persistent cough in patients with non-small cell lung cancer after thoracoscopic surgery. Chin J Intern Med 2022;28:2241-5.
21. Li L, Wang G, Wang K, et al. Analysis of risk factors for intolerable cough after pneumonectomy. Chin J Tuberc Respir Dis 2022;27:491-5.
22. Dong Y, Meng X, Liu X, et al. Analysis of risk factors for persistent cough in patients after thoracoscopic pulmonary resection. J Aerosp Med 2022;33:1475-7.
23. Dong X, Xu Q, Zhong J. Analysis of risk factors for postoperative chronic cough in lung cancer patients. J Clin Pulm Med 2022;37:783-6.
24. Mu T, Jiang G, Li X, et al. Multivariate analysis of cough after pulmonary resection in video-assisted thoracoscopic surgery. Chin J Minim Invasive Surg 2017;17:577-80.
25. Xin W, Chen X, Tang J, et al. Factors influencing persistent cough after secondary thoracoscopic resection in lung cancer patients. J Thorac Dis 2021;13:31-7.
26. Wu Z, Xu W, Xu S, et al. Risk factors and prevention strategies for chronic cough after robotic and video-assisted thoracoscopic surgery in non-small cell lung cancer patients. Chin J Thorac Cardiovasc Surg 2020;27:1274-80.
27. Ma H. Clinical study on chronic cough after pulmonary tumor surgery under thoracoscopy. [Dissertation]. Soochow University; 2019.
28. Liu M, Mao C, Wang Y, et al. Analysis of curative effects of left upper lobectomy with Da Vinci robot in lung cancer and influencing factors of postoperative chronic cough. Chin J Clin Oncol 2023;22:1507-11.
29. Gao Y, Zhang Z, Fu W, et al. Analysis of risk factors for persistent cough after non-small cell lung cancer segmentectomy under uniportal thoracoscopy. Shandong Med J 2022;62:77-9.
30. Cui H, Wang S, Gu Y. Risk factors of persistent cough after secondary thoracoscopic lung cancer resection in elderly patients. J Clin Pulm Med 2023;29:63-7.
31. Zhang Y, Xie G, Li B. Influencing factors for treating chronic cough after video-assisted thoracoscopic surgery in non-small cell lung cancer patients. Chin J Lung Cancer 2022;29:299-302.
32. Ma Z, Zhang A, Huang F, et al. Influencing factors of persistent dry cough in non-small cell lung cancer patients after pulmonary lobectomy. Chin J Oncol 2022;48:887-90.
33. Lin R, Che G, Xu Z, et al. Influencing factors of postoperative cough in lung resection patients with lung cancer: A single-center prospective study. Chin J Thorac Cardiovasc Surg 2017;24:748-52.
34. Yin Y, Rong G, Zhang F. Influencing factors of chronic cough in non-small cell lung cancer patients after minimally invasive video-assisted thoracoscopic surgery. Chin J Clin Thorac Cardiovasc Surg 2021;28:80-2.
35. Yang C, Du M. Influencing factors of persistent cough in early lung cancer patients after uniportal thoracoscopic surgery. World Latest Med Inf 2024;24:145-9.
36. Li L, Jiang P, Cheng N. Analysis of factors contributing to postoperative persistent cough in early-stage lung cancer patients. Chin J Geriatr 2023;31:940-2.
37. Caldwell DM. An overview of conducting systematic reviews with network meta-analysis. Syst Rev 2014;3:109. [Crossref] [PubMed]
38. Dicpinigaitis PV, Rauf K. The influence of gender on cough reflex sensitivity. Chest 1998;113:1319-21. [Crossref] [PubMed]
39. Kelsall A, Decalmer S, McGuinness K, et al. Sex differences and predictors of objective cough frequency in chronic cough. Thorax 2009;64:393-8. [Crossref] [PubMed]
40. Xie MR, Zhu YF, Zhou MQ, et al. Analysis of factors related to chronic cough after lung cancer surgery. Thorac Cancer 2019;10:898-903. [Crossref] [PubMed]
41. Zhu YF, Wu SB, Zhou MQ, et al. Increased expression of TRPV1 in patients with acute or chronic cough after lung cancer surgery. Thorac Cancer 2019;10:988-91. [Crossref] [PubMed]
42. Patberg KW. The female preponderance to cough hypersensitivity syndrome: another clue pointing to the role of TRPV1 in cough. Lung 2011;189:257-8. [Crossref] [PubMed]
43. French CT, Fletcher KE, Irwin RS. A comparison of gender differences in health-related quality of life in acute and chronic coughers. Chest 2005;127:1991-8. [Crossref] [PubMed]
44. Lin R, Chen W, Zhu L, et al. Comparison of postoperative cough-related quality of life and recovery between sublobectomy and lobectomy for early-stage non-small cell lung cancer patients: a longitudinal study. BMC Pulm Med 2022;22:154. [Crossref] [PubMed]
45. Lu XF, Min XP, Lu B, et al. Bronchial morphological changes are associated with postoperative intractable cough after right upper lobectomy in lung cancer patients. Quant Imaging Med Surg 2022;12:196-206. [Crossref] [PubMed]
46. Seok Y, Cho S, Lee JY, et al. The effect of postoperative change in bronchial angle on postoperative pulmonary function after upper lobectomy in lung cancer patients. Interact Cardiovasc Thorac Surg 2014;18:183-8. [Crossref] [PubMed]
47. Bu L, Yang AR, Peng H, et al. Dividing inferior pulmonary ligament may change the bronchial angle. J Surg Res 2016;201:208-12. [Crossref] [PubMed]
48. Ginsberg RJ, Rubinstein LV. Randomized trial of lobectomy versus limited resection for T1 N0 non-small cell lung cancer. Lung Cancer Study Group. Ann Thorac Surg 1995;60:615-22; discussion 622-3. [Crossref] [PubMed]
49. Huang J, Luo Q, Tan Q, et al. Evaluation of the surgical fat-filling procedure in the treatment of refractory cough after systematic mediastinal lymphadenectomy in patients with right lung cancer. J Surg Res 2014;187:490-5. [Crossref] [PubMed]
50. Xu K, Zhang Y, Cui Y, et al. Patient-reported outcomes of laryngeal mask anesthesia in thoracoscopic pulmonary wedge resection: A randomized controlled study. Thorac Cancer 2022;13:3192-9. [Crossref] [PubMed]
51. Hung MH, Chan KC, Liu YJ, et al. Nonintubated thoracoscopic lobectomy for lung cancer using epidural anesthesia and intercostal blockade: a retrospective cohort study of 238 cases. Medicine (Baltimore) 2015;94:e727. [Crossref] [PubMed]
52. Díaz Vico T, Elli EF. Clinical outcomes of gastroesophageal reflux disease-related chronic cough following antireflux fundoplication. Esophagus 2020;17:92-8. [Crossref] [PubMed]