Diabetes mellitus is associated with short-term postoperative outcomes in patients undergoing vats lobectomy for lung cancer: a 7-year retrospective analysis Authors: Ping Song, Hongbin Liu, Huafen Li, Yan Jiang, Chaoyang Zheng, Hong Ding, Zeyong Zhang, Zhaoxia Sun Published: 2026 ABSTRACT AbstractBackgroundDiabetes mellitus (DM) is a prevalent metabolic disorder and a well-recognized perioperative risk factor across non-cardiac surgical populations. However, its correlative association with short-term postoperative complications after video-assisted thoracoscopic surgery (VATS) lobectomy remains insufficiently characterized. This study explores the relationship between DM and postoperative outcomes using a large national inpatient database.MethodsA retrospective cohort study was conducted based on the 2016–2022 Nationwide Inpatient Sample. Adult patients undergoing VATS lobectomy for lung cancer were identified via International Classification of Diseases, 10th Revision coding, and participants were stratified by the presence or absence of DM. Multivariable logistic regression were used to evaluate the correlative links between DM and postoperative complications after adjusting for demographic, comorbid, and hospital-related covariates.ResultsA total of 34,398 eligible patients were included in the final cohort, among whom 7,515 (21.84%) had a diagnosis of DM. Diabetic patients were older and had a higher overall comorbidity burden compared with non-diabetic counterparts, accompanied by a higher crude incidence of in-hospital postoperative complications. After multivariable adjustment, DM was associated with elevated odds of acute respiratory insufficiency [odds ratio (OR) 1.177; 95% confidence interval (CI): 1.077 to 1.286; P<0.001], myocardial infarction (OR 1.714; 95% CI: 1.561 to 1.882; P<0.001), delirium (OR 1.509; 95% CI: 1.198 to 1.902; P<0.001), requirement for mechanical ventilation (OR 1.208; 95% CI: 1.033 to 1.412; P=0.02), and supraventricular arrhythmia (OR 1.143; 95% CI: 1.073 to 1.218; P<0.001). An inverse correlation between DM and acute pneumothorax was also observed (OR 0.869; 95% CI: 0.778 to 0.971; P=0.01).ConclusionsThis large nationwide observational cohort reveals correlative associations between DM and increased risks of cardiopulmonary and neurological complications following VATS lobectomy for lung cancer. These observational findings provide supportive insights for perioperative risk assessment and optimized metabolic management in diabetic patients undergoing minimally invasive thoracic surgery.Keywords: Video-assisted thoracoscopic surgery lobectomy (VATS lobectomy), diabetes mellitus (DM), postoperative complications, lung cancer Highlight box Key findings • In a nationwide cohort of 34,398 patients undergoing video-assisted thoracoscopic surgery (VATS) lobectomy for lung cancer, 21.84% of individuals had diabetes mellitus (DM). After multivariable adjustment, DM was associated with elevated risks of acute respiratory insufficiency [odds ratio (OR) 1.177; 95% confidence interval (CI): 1.077 to 1.286; P<0.001], myocardial infarction (OR 1.714; 95% CI: 1.561 to 1.882; P<0.001), delirium (OR 1.509; 95% CI: 1.198 to 1.902; P<0.001), requirement for mechanical ventilation (OR 1.208; 95% CI: 1.033 to 1.412; P=0.02), and supraventricular arrhythmia (OR 1.143; 95% CI: 1.073 to 1.218; P<0.001). The inverse correlation between DM and acute pneumothorax may plausibly stem from administrative coding discrepancies. DM patients also presented higher hospitalization costs and broader length-of-stay dispersion, while in-hospital mortality was comparable to non-DM patients. What is known and what is new? • DM is a well-established perioperative risk factor in non-cardiac surgery, while its prognostic relevance to short-term outcomes after VATS lobectomy for lung cancer remains inconsistent across existing studies. • This large national study explored and quantified the associations between DM and multiple short-term postoperative complications, providing real-world observational evidence with modest effect sizes and inherent database limitations. What is the implication, and what should change now? • This observational analysis offers supplementary references for perioperative risk stratification in diabetic patients scheduled for VATS lobectomy. It suggests the value of routine preoperative glycemic assessment, optimized perioperative metabolic regulation, and enhanced clinical surveillance for cardiopulmonary complications in this high-risk subgroup. Introduction Lung cancer remains one of the most frequently diagnosed malignancies worldwide and continues to be the leading cause of cancer-related mortality. In the United States, an estimated 229,410 new lung cancer cases and 124,990 cancer deaths are expected in 2026, contributing to growing detection of early-stage tumors eligible for surgical resection (1). Surgical resection is the primary curative approach for operable lung cancer, and operative techniques have evolved from conventional open thoracotomy to widely adopted minimally invasive video-assisted thoracoscopic surgery (VATS) (2-4). Compared with open thoracotomy, VATS confers substantial advantages, including reduced postoperative pain, faster functional recovery, and lower rates of postoperative complications (5-7). Given the notable differences in surgical trauma and postoperative recovery pathways between open and VATS procedures, risk factors identified from general non-cardiac or mixed thoracic cohorts may not be directly generalizable to patients undergoing VATS lobectomy alone. The global prevalence of diabetes mellitus (DM) is projected to reach 578 million by 2030 and 700 million by 2045. As a common chronic metabolic comorbidity, DM is closely linked to elevated perioperative susceptibility to infection, prolonged hospitalization, and cardiovascular events, posing substantial challenges to perioperative surgical management (8,9). Nevertheless, existing studies examining the relationship between DM and postoperative outcomes following lung cancer resection remain inconsistent (10-14). Some studies have reported an association between DM and higher risks of adverse perioperative events (11,13), whereas others found no significant correlation between DM and overall postoperative complications (12,14). Such discrepancies are largely attributable to heterogeneous surgical strategies, mixed open and minimally invasive cohorts, and small single-center sample sizes. To date, the specific prognostic correlation between DM and short-term outcomes after isolated VATS lobectomy remains poorly defined. With the expanding clinical application of VATS lobectomy and the growing number of diabetic patients requiring thoracic surgery, targeted evidence focusing on this specific population is warranted. We therefore hypothesized that DM is correlated with adverse short-term in-hospital outcomes among lung cancer patients undergoing VATS lobectomy. This study aimed to examine whether DM-related perioperative prognostic patterns are specific to the modern minimally invasive VATS setting, rather than merely reflecting general surgical risks in diabetic populations. We present this article in accordance with the STROBE reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-0925/rc). Methods This retrospective study utilized data from the 2016–2022 National Inpatient Sample (NIS), a large nationally representative inpatient database sponsored by the Agency for Healthcare Research and Quality as part of the Healthcare Cost and Utilization Project. The NIS covers approximately 20% of all U.S. hospitalizations annually and includes around 8 million discharge records, providing robust real-world inpatient healthcare data. All analyzed datasets were de-identified. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. We identified patients aged 18 years or older with a primary diagnosis of lung cancer who underwent VATS lobectomy, identified via ICD-10 diagnostic and procedural codes. The ICD-10 codes used for cohort identification and outcome definitions are presented in Table S1. Elixhauser comorbidities were ascertained using the ICD-10-CM v2022.1 standardized software. Patients with a metastatic cancer diagnoses were excluded, with the exception of those with diagnoses of secondary or unspecified malignant neoplasms. We also removed patients with missing demographic data. DM was identified using corresponding ICD-10 diagnostic codes. The study population selection flow diagram is depicted in Figure 1. Flow chart of study cohort selection. VATS, video-assisted thoracoscopic surgery. Four primary outcome categories were assessed: postoperative complications, in-hospital mortality, length of stay (LOS), and total hospitalization charges. Postoperative complications of interest included acute pneumothorax, acute respiratory insufficiency, perioperative blood transfusion, procedural bleeding, delirium, empyema (with and without fistula), mechanical ventilation requirement, myocardial infarction, pneumonia, pulmonary collapse, pulmonary edema, sepsis, shock, supraventricular arrhythmia. In-hospital mortality, LOS, and hospital charges were extracted from standard NIS administrative records. Statistical analysis All analyses were conducted using SPSS version 25.0. The Wilcoxon rank-sum test was applied to continuous variables and the χ2 test to categorical variables. Multivariate logistic regression was used to evaluate the association between DM and postoperative outcomes after adjusting for demographic characteristics, comorbidities, and hospital-related covariates. All adjusted variables are listed in Table 1. A two-sided P value ≤0.05 was defined as statistically significant. Table 1Variables used in binary logistic regression analysisVariables categoriesSpecific variablesPatient demographicsAge, sex (male and female), race (White, Black, Hispanic, Asian or Pacific Islander, Native American and Other)Hospital characteristicsType of admission (non-elective, elective), bed size of hospital (small, medium, large), teaching status of hospital (nonteaching, teaching), location of hospital (rural, urban), type of insurance (Medicare, Medicaid, private insurance, self-pay, no charge, other), location of the hospital (northeast, midwest, south, west)ComorbiditiesChronic pulmonary disease, hypertension, obesity, cigarette use, alcohol use, depressionThis table was adapted from an Open Access article (15) under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/). Variables used in binary logistic regression analysis This table was adapted from an Open Access article (15) under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/). Results Baseline characteristics A total of 34,398 patients who underwent VATS lobectomy between 2016 and 2022 were included in the final analysis. Compared with patients without DM, diabetic patients were significantly older (71 vs. 69 years, P<0.001), more likely to be males (51.8% vs. 41.9%, P<0.001) and had higher rates of chronic pulmonary disease (51.3% vs. 48.3%, P<0.001), hypertension (15.3% vs. 13.6%, P<0.001), obesity (25.3% vs. 11.3%, P<0.001) and cigarette use (51.7% vs. 49.7%, P<0.001). Although median LOS was identical between groups (4 days), the interquartile range was wider among patients with DM (3.00–7.00 vs. 3.00–6.00 days, P<0.001), indicating significant between-group heterogeneity in hospitalization duration. Additionally, the DM group incurred significantly higher median hospitalization charges ($89,119 vs. $84,380, P<0.001). Baseline demographics, comorbidities, and hospital characteristics are summarized in Tables 2,3. Table 2Patient characteristics after VATS lobectomy (2016–2022)CharacteristicsDiabetes (n=7,515)No diabetes (n=26,883)PAge (years)71.00 (65.00, 75.00)69.00 (62.00, 74.00)<0.001Gender<0.001 Male3,893 (51.8)11,272 (41.9) Female3,622 (48.2)15,611 (58.1)Race<0.001 White5,513 (73.4)21,763 (81.0) Black861 (11.5)1,872 (7.0) Hispanic418 (5.6)1,030 (3.8) Asian or Pacific Islander348 (4.6)909 (3.4) Native American14 (0.2)78 (0.3) Other361 (4.8)1,231 (4.6)Type of insurance<0.001 Medicare5,485 (73.0)17,360 (64.6) Medicaid366 (4.9)1,590 (5.9) Private insurance1,428 (19.0)7,098 (26.4) Self-pay53 (0.7)230 (0.9) No charge6 (0.1)33 (0.1) Other177 (2.4)572 (2.1)Admission type0.38 Elective7150 (95.1)25,643 (95.4) Non-elective365 (4.9)1,240 (4.6)Household income<0.001 Q12,057 (27.4)6,050 (22.5) Q21,948 (25.9)6,906 (25.7) Q31,860 (24.8)6,861 (25.5) Q41,650 (22.0)7,066 (26.3)Medical comorbidities Chronic pulmonary disease3,855 (51.3)12,981 (48.3)<0.001 Hypertension1,153 (15.3)3,655 (13.6)<0.001 Obesity1,899 (25.3)3,035 (11.3)<0.001 Cigarette use3,888 (51.7)13,354 (49.7)0.002 Alcohol use140 (1.9)800 (3.0)<0.001 depression1,000 (13.3)3,466 (12.9)0.35Data are presented as n (%) or median (interquartile range). VATS, video-assisted thoracoscopic surgery. Patient characteristics after VATS lobectomy (2016–2022) Data are presented as n (%) or median (interquartile range). VATS, video-assisted thoracoscopic surgery. Table 3Hospital characteristics and outcome (2016–2022)CharacteristicsDiabetesNo diabetesPRegion of hospital<0.001 Northeast1,672 (22.2)6,665 (24.8) Midwest1,747 (23.2)6,129 (22.8) South3,023 (40.2)9,835 (36.6) West1,073 (14.3)4,254 (15.8)Bed size of hospital0.009 Small790 (10.5)2,915 (10.8) Medium1,949 (25.9)6,507 (24.2) Large4,776 (63.6)17,461 (65.0)Hospital urban status7,264 (96.7)26,148 (97.3)0.005Hospital teaching status6,427 (85.5)23,242 (86.5)0.04Length of stay ( days)4.00 (3.00, 7.00)4.00 (3.00, 6.00)<0.001Total charge ($)89,119 (61,532, 136,332)84,380 (58,750, 128,210)<0.001In-hospital mortality83 (1.1)269 (1.0)0.43Data are presented as n (%) or median (interquartile range). Hospital characteristics and outcome (2016–2022) Data are presented as n (%) or median (interquartile range). Postoperative complications In unadjusted analyses, DM patients exhibited higher crude rates of acute respiratory insufficiency (11.0% vs. 8.4%; P<0.001), pneumonia (4.5% vs. 3.8%; P=0.005), pulmonary collapse (10.6% vs. 9.5%; P=0.005), use of mechanical ventilation (3.2% vs. 2.4%; P<0.001), myocardial infarction (10.5% vs. 5.8%; P<0.001), delirium (1.5% vs. 0.9%; P<0.001), and supraventricular arrhythmia (24.8% vs. 20.4%; P<0.001) compared non-DM patients. By contrast, the crude incidence of acute pneumothorax was lower among diabetic patients (5.8% vs. 6.7%; P=0.005). After multivariable adjustment, DM was associated with increased risks of acute respiratory insufficiency [odds ratio (OR) 1.177; 95% confidence interval (CI): 1.077 to 1.286; P<0.001], myocardial infarction (OR 1.714; 95% CI: 1.561 to 1.882; P<0.001), delirium (OR 1.509; 95% CI: 1.198 to 1.902; P<0.001), requirement for mechanical ventilation (OR 1.208; 95% CI: 1.033 to 1.412; P=0.02), and supraventricular arrhythmia (OR 1.143; 95% CI: 1.073 to 1.218; P<0.001). A negative correlative trend between DM and acute pneumothorax persisted after adjustment (OR 0.869; 95% CI: 0.778 to 0.971; P=0.01). All multivariable regression outcomes are presented in Table 4. Table 4Relationship between diabetes mellitus and postoperative complicationsComplicationsUnivariate analysisMultivariate logistic regressionDiabetes, n (%)No diabetes, n (%)ARDPOR95% CIPPulmonary Acute respiratory insufficiency826 (11.0)2,255 (8.4)2.6<0.0011.1771.077–1.286<0.001 Pneumonia340 (4.5)1,026 (3.8)0.70.0051.1000.965–1.2530.16 Pulmonary collapse797 (10.6)2,557 (9.5)1.10.0051.0410.954–1.1350.37 Pulmonary edema31 (0.4)102 (0.4)00.681.0110.667–1.5320.96 Empyema with and without fistula59 (0.8)210 (0.8)00.970.9600.712–1.2940.79 Acute pneumothorax433 (5.8)1,790 (6.7)−0.90.0050.8690.778–0.9710.01Respiratory support Mechanical ventilation243 (3.2)643 (2.4)0.8<0.0011.2081.033–1.4120.02Others Myocardial infarction787 (10.5)1,560 (5.8)4.7<0.0011.7141.561–1.882<0.001 Delirium112 (1.5)253 (0.9)0.6<0.0011.5091.198–1.902<0.001 Supraventricular arrhythmia1,860 (24.8)5,487 (20.4)4.4<0.0011.1431.073–1.218<0.001 Sepsis97 (1.3)288 (1.1)0.20.111.0640.836–1.3550.61 Shock24 (0.3)69 (0.3)00.361.0630.657–1.7220.80 Blood transfusion241 (3.2)801 (3.0)0.20.311.0590.911–1.2320.45 Bleeding complicating the procedure17 (0.2)59 (0.2)00.911.0200.585–1.7790.94ARD, Absolute risk difference; CI, confidence interval; OR, odds ratio. Relationship between diabetes mellitus and postoperative complications ARD, Absolute risk difference; CI, confidence interval; OR, odds ratio. Discussion This large national cohort analysis shows that DM is associated with a higher overall postoperative complication burden and increased hospitalization costs among patients undergoing VATS lobectomy for lung cancer, which is consistent with existing perioperative evidence in surgical populations (9,16). Previous studies have reported inconsistent prognostic findings regarding DM in lung cancer surgery. Some investigations found no significant link between DM and postoperative complications or prolonged hospital stay (12,14), while others indicated correlations with adverse perioperative events and greater healthcare resource utilization (13,17,18). Such discrepancies arise largely from heterogeneous surgical approaches, mixed open and minimally invasive cohorts, and single-center sample limitations, highlighting the need for targeted analyses focused specifically on VATS lobectomy populations. The present study utilized standardized ICD-10 coding to further clarify the prognostic correlates of DM within this large national cohort. Our analysis shows that DM correlates with higher rates of acute respiratory insufficiency, mechanical ventilation requirement, myocardial infarction, and supraventricular arrhythmia. Accumulated literature suggests impaired host immune defense serves as a plausible underlying mechanism (19). Persistent hyperglycemia disrupts neutrophil and macrophage function, induces chronic low-grade inflammation and endothelial dysfunction, and impairs respiratory mucosal barrier integrity, which may render diabetic individuals more susceptible to respiratory complications (20-24). These biological pathways may partially explain the higher complication burden observed in the DM group. Notably, although several associations reached statistical significance, the corresponding effect sizes were generally modest. In large administrative database analyses, small between-group differences can achieve statistical significance driven by enormous sample size and may not always translate into tangible clinical implications. Accordingly, our findings should be interpreted as hypothesis-generating observational evidence rather than definitive mechanistic conclusions. The observed association between DM and postoperative delirium requires cautious interpretation. Delirium is a multifactorial condition related to advanced age, baseline frailty, preoperative cognitive impairment, intraoperative hypoxia, sedative exposure, and delayed postoperative mobilization (25-29). Since these granular clinical variables are unavailable in the NIS database, the detected correlation likely reflects unmeasured baseline physiological vulnerability among diabetic patients, rather than a direct independent effect of hyperglycemia alone. The inverse correlation between DM and acute pneumothorax should be regarded as a methodological artifact inherent to administrative inpatient database research. From a thoracic surgical perspective, no plausible biological mechanism supports a protective effect of DM against alveolar air leakage or pleural effusion development. This counterintuitive trend may be attributable to inconsistent coding differentiation between minor asymptomatic and clinically significant pneumothorax, variability in postoperative imaging and chest drain management, and statistical inflation from multiple endpoint comparisons. Thus, this negative association represents only a numerical statistical trend without clear physiological or clinical relevance. Although median LOS was comparable between the two groups, diabetic patients exhibited broader LOS dispersion and higher hospitalization costs. On the one hand, these disparities are partly driven by unfavorable baseline sociodemographic and clinical characteristics. Diabetic individuals in our cohort were older with a greater comorbidity burden, less likely to have private insurance and showed distinct distributions in household income, racial composition, and hospital geographic region. On the other hand, heterogeneous postoperative recovery trajectories further account for the widened LOS distribution: diabetic patients with well-controlled preoperative glycemia and stable baseline status demonstrated recovery timelines similar to non-diabetic counterparts and achieved timely hospital discharge (30,31). In contrast, those with long-standing suboptimal glycemic control presented compromised immune function and poorer surgical and pulmonary tissue repair capacity, accompanied by elevated postoperative complication risks and prolonged in-hospital observation (32,33). Such polarized recovery patterns collectively contributed to greater LOS variability in the DM group. Despite higher overall complication rates in diabetic patients, no significant intergroup difference in in-hospital mortality was identified. This non-significant finding is mainly attributed to the low overall in-hospital mortality rate and limited statistical power from rare mortality events. Additionally, socioeconomic factors related to financial constraints and early discharge practices may potentially mask genuine mortality disparities between groups. This study bears several inherent limitations typical of administrative database research. First, key granular thoracic surgical parameters, including preoperative pulmonary function, frailty status, tumor stage, operative duration, intraoperative blood loss, resected lobe classification, lymph node dissection extent, conversion to open surgery, and structured perioperative rehabilitation protocols are not captured in the NIS dataset. The absence of these variables limits in-depth mechanistic interpretation of cardiopulmonary outcomes and cannot fully exclude residual confounding. Second, DM was defined solely as a binary comorbidity based on ICD-10 coding, lacking detailed metabolic parameters such as HbA1c levels, disease duration, and insulin dependency. This simplified classification overlooks substantial metabolic heterogeneity among diabetic patients. Third, potential coding inaccuracies and underreporting of mild postoperative events may influence result reliability, and the database only captures in-hospital outcomes without long-term follow-up data. Even after multivariable adjustment, unmeasured baseline vulnerability and institutional practice variations may still introduce residual confounding. Clinically, our findings support the value of routine preoperative glycemic screening and prediabetes identification for patients scheduled for VATS lobectomy (34). Perioperative care should incorporate standardized preoperative glucose evaluation, targeted intra- and postoperative glycemic monitoring with clear therapeutic targets, rational nutritional adjustment, and standardized glycemic management protocols (35-39). Individualized perioperative risk stratification and metabolic surveillance are recommended for diabetic and prediabetic patients to mitigate cardiopulmonary, neurological, and economic postoperative burdens. In summary, this large national cohort analysis indicates that DM correlates with multiple short-term cardiopulmonary and neurological postoperative events after VATS lobectomy. Diabetic patients also show greater hospitalization variability and increased medical costs linked to demographic and clinical disparities. The inverse pneumothorax association represents a statistical artifact. Given modest effect sizes, inherent database constraints, and potential residual confounding, these results should be interpreted cautiously as hypothesis-generating observational evidence. Further prospective studies incorporating detailed surgical, metabolic, functional, and long-term outcome data are warranted to refine perioperative risk stratification and optimize individualized metabolic management for diabetic patients undergoing minimally invasive lung cancer resection. Conclusions In this nationwide analysis of 34,398 patients undergoing VATS lobectomy for lung cancer, DM correlates with increased risks of several short-term postoperative cardiopulmonary and neurological complications. These observational findings underscore the clinical value of tailored perioperative risk assessment and optimized metabolic monitoring for diabetic patients undergoing VATS lobectomy. Supplementary The article’s supplementary files as Acknowledgments None. 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 and its subsequent amendments. Footnotes Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-0925/rc Funding: This work was supported by Research project on health management of Southern Medical University (Grant No. YJY200507). Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-0925/coif). All authors report funding support from the Research project on health management of Southern Medical University (grant No. YJY200507) for the submitted work. The authors have no other conflicts of interest to declare. References Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-0925/prf Supplementary Materials The article’s supplementary files as