Race, gender, and safety: exposing biases in minimally invasive lung cancer surgery research

Introduction

Video-assisted thoracic surgery (VATS) and robotic-assisted thoracic surgery (RATS) have been widely adopted as safe and effective alternatives to open thoracotomy for lung cancer (1-3). Minimally invasive techniques were commonly accepted as the new standard of care after comparable oncologic outcomes of VATS approaches were established by Flores and colleagues in 2009 (3). Since that time, many studies have been published examining the short- and long-term outcomes between MIS and open techniques with a focus on pain scores, hospital stay, and complications (1,2,4).

Unfortunately, the broader application of such research may be limited by results from individual surgeons in single hospital systems utilizing common practices. Structured in this fashion, insular studies may fail to capture larger patterns in surgical outcomes, especially as they pertain to health, gender and race-related factors on a national scale.

We sought to address these issues by evaluating the incidence of adverse events following minimally invasive and open lung resection in the Surveillance Epidemiology and End Results (SEER)-Medicare database. Our goal was to identify macroscopic trends in patient outcomes that supersede the practices of individual hospital groups and reflect widespread features in lung surgery. We present this article in accordance with the STROBE reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-1287/rc).

Methods

Data source and selection criteria

Data was queried from the linked Surveillance Epidemiology and End Results (SEER)-Medicare database. SEER compiles data on cancer incidence and mortality from population-based cancer registries across the United States. Registries included in SEER cover approximately 35% of the United States. Medicare eligibility begins at age 65 years, and the dataset includes Medicare beneficiaries who were residing in a SEER registry area at the time of a cancer diagnosis. Demographics and tumor characteristics were extracted from SEER, while information on surgery, readmission, and comorbidities was extracted from the Medicare linked claims.

There were 746,441 patients diagnosed with lung cancer between 1992–2015. Patients with date of birth and death agreement between SEER and Medicare, microscopically confirmed first or only cancer diagnosis of lung cancer, and with a reporting source other than autopsy or death certificate, were considered for inclusion (n=466,219). Based on the change from the International Classification of Diseases (ICD) 9 to ICD-10 on October 1, 2015, and in order to maintain consistent ICD coding versions across the cohort, those missing the month of diagnosis, or diagnosed after March 2015, were excluded (n=21,188). The remaining patients were queried for a lung cancer surgery within 6 months after diagnosis, based on ICD-9 procedure codes (n=60,507), and were limited to those with complete fee-for-service coverage, and no private insurance for at least 3 months after surgery (or until death) (n=49,140). The final cohort was limited to those with non-small cell histology, at least 65 years old at diagnosis, and discharged alive after receiving a lobectomy or limited resection in 2007 or later, based on the availability of detailed codes for minimally invasive surgery (MIS) (n=17,486). The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Outcomes and covariates

The primary outcome of interest was an adverse event within 30 or 90 days after discharge from the index surgery, defined as a readmission to the hospital or death. The secondary outcome of interest was length of stay (LOS) during the index surgery. The primary predictor of interest was the surgical approach, defined as open or MIS, based on ICD-9 procedure codes. Data on age at diagnosis, sex, race, marital status, stage, histology, and proximity to an urban area were extracted from SEER. Marital status was defined as married/domestic partner or unmarried, including single, divorced, and widowed. Stage was defined based on American Joint Committee on Cancer (AJCC) 6^th edition codes, while histology was classified as squamous cell carcinoma, adenocarcinoma, or other [including large-cell carcinoma, other carcinomas, and those with non-small cell lung cancer (NSCLC), not otherwise specified], based on codes from the International Agency for Research on Cancer (IARC) (5,6). Patients were classified by whether they lived in an urban commuting area at the time of diagnosis, based on 2010 Rural-Urban Commuting Area (RUCA) codes. Surgery information, readmission, LOS, and a readmission index score based on Elixhauser comorbidities were extracted from Medicare (7,8).

Statistical analysis

All analyses were stratified by the type of surgery. Patients with open and MIS surgical approaches were compared on demographics and clinical characteristics using χ² tests for categorical variables and t-tests for continuous variables. Multivariable logistic regression was used to assess the factors independently associated with surgical approach, based on odds ratio (OR) and 95% confidence interval (CI). Multivariable logistic regression was then used to assess the independent association of surgical approach with an adverse event within 30 or 90 days, after adjusting for age, sex, race, marital status, stage, histology, urban commuting area, and Elixhauser comorbidity index. Adverse events within 90 days were also assessed using a time-to-event approach using Kaplan-Meier curves and Cox proportional hazards models. As a secondary outcome, LOS during the index surgery was natural log (ln) transformed, and multivariable linear regression was used to assess the independent association with surgical approach. Multivariable analyses were conducted on the subset of patients with complete data. All analyses were conducted using Statistical Analysis System (SAS) software, v 9.4 (SAS Institute, Cary, North Carolina).

Results

There were 5,185 patients who underwent limited resection, 2,287 with an open approach, and 2,898 with an MI approach. Compared to those undergoing open surgery, those undergoing MI surgery were statistically significantly (all P<0.001) more likely to be older (74.7 vs. 74.0 years), female (59.4% vs. 52.3%), to have adenocarcinoma (62.9% vs. 54.6%), to live in an urban commuting area (92.1% vs. 86.2%), and to have a lower comorbidity score (10.0 vs. 11.8) (Table 1). These associations remained after adjusting for all covariates (Table 2).

There were 12,301 patients in the lobectomy cohort, 8,182 with open surgery, and 4,119 with MIS surgery. Those undergoing MIS were statistically significantly (all P<0.001) older (73.8 vs. 73.1 years), and more likely to be female (57.3% vs. 51.1%), diagnosed at stage I/II (79.8% vs. 75.9%), have adenocarcinoma (63.7% vs. 54.8%), to live in an urban commuting area (92.2% vs. 85.2%), and to have a lower comorbidity score (8.8 vs. 10.4). There were also significant differences in race across surgical approach (P<0.001), with Black patients less likely to receive MIS (4.5% vs. 5.9%), compared to open surgery (Table 1). After adjustment for all covariates, results remained consistent (Table 2).

After adjusting for all covariates, there was no significant association between surgical approach and 30-/90-day adverse events among those who received a limited resection [OR_adj(30-day): 1.07, 95% CI: 0.91–1.26; OR_adj(90-day): 1.05, 95% CI: 0.92–1.20]. Older patients, those with later-stage cancers, and those with higher comorbidity scores were more likely to have an adverse event within 30 days. Later-stage patients, those with non-adenocarcinoma histologies, and those with higher comorbidity scores were more likely to have an adverse event within 90 days (Table 3).

Within the lobectomy cohort, those who received MIS were significantly less likely to have an adverse event within 30 days (OR_adj: 0.87, 95% CI: 0.77–0.98) or 90 days (OR_adj: 0.87, 95% CI: 0.79–0.96). Patients who are older, males, of other races, unmarried, with squamous cell histology, and with higher comorbidity scores were more likely to have an adverse event within both 30 and 90 days. Those with stage III/IV and those with other histologies were also more likely to have an adverse event within 90 days (Table 3).

Results from the time-to-event analysis for 90-day adverse events were similar, with no significant difference in hazard ratio (HR) for MIS compared to open surgery in the limited resection cohort (HR_adj: 1.06, 95% CI: 0.95–1.18), and significantly lower risk in the lobectomy cohort (HR_adj: 0.88, 95% CI: 0.81–0.96).

After adjusting for all covariates, the ln-LOS was significantly shorter for MIS, compared to open surgery for both the limited resection cohort (β_adj: −0.436, 95% CI: −0.475 to −0.396) and the lobectomy cohort (β_adj: −0.354, 95% CI: −0.376 to −0.333) (Table 4).

Discussion

The intention of this study was to evaluate the rates of adverse events, including readmission, following minimally invasive and open lung resection. Fortunately, our investigation has revealed several important and less commonly discussed features of lung surgery concealed in the data.

Safety is the selection bias: patients undergoing MIS are healthier with earlier-stage disease

For both limited resection and lobectomy, patients undergoing MIS are healthier with lower co-morbid scores (10.0 vs. 11.8, and 8.8 vs. 10.4, respectively) and are more likely to have early-stage disease (stage I or II). It is surprising that there was no difference when comparing MIS and open surgery for patients undergoing limited resection for early- and late-stage cancers (P=0.56). But this may reflect the use of wedge resection as a treatment in early lung cancer, and diagnosis in late.

For limited resection, there was no association between surgical approach and 30- or 90-day adverse events. Adverse events were independently associated with later stage cancers, higher co-morbidity scores, older patients and non-adenomatous histology. Among lobectomy patients, MIS was significantly less likely to have an adverse event within both 30- (OR_adj: 0.87, 95% CI: 0.77–0.98) and 90-days (OR_adj: 0.87, 95% CI: 0.79–0.96). Adverse events were associated with a similar combination of features, including later-stage cancers, high co-morbidity scores, older patients, squamous cell histology, and male patients. Of note, squamous cell carcinomas are more likely to be centrally located, which helps to explain this result (9).

The common conclusion drawn by these types of studies is that MIS is independently superior and de facto safer than open surgery (10). Instead, the finding likely reflects the good judgement of operating surgeons. There are multiple forms of bias in large-scale studies comparing minimally invasive and open surgery (11). Chief among them are selection biases that tend to favor minimally invasive outcomes in a population of healthier patients with earlier-stage disease, as we observe in this study (11). These patients should be expected to have fewer adverse events, but because of health and anatomic conditions that permit a minimally invasive technique, not because of the surgery itself. Of equal importance, procedural codes used in national databases are often generated by how the surgery ended instead of how it began. As a result, procedures that were planned as VATS or RATS but were converted due to operative difficulty, technical complication, or emergency will be counted as open procedures.

The impact of surgical decision-making is largely neglected in the literature. Unfortunately, without clinical context, retrospective studies create a false sense of security that minimally invasive techniques are somehow better, regardless of clinical context. Instead, it is more likely that surgeons are making safe choices of who to operate on and how to operate.

LOS is shorter following MIS, but what does that actually mean?

For limited resection, the LOS was 7.9 and 5.3 days for open and MIS, respectively (P<0.001). For lobectomy, the relationship is similar, with a LOS of 8.4 and 6.0 days (P<0.001). Despite the statistical significance, it is difficult to interpret the clinical importance of these findings.

As seen above, patients undergoing open surgery are less healthy with more advanced disease. This will have a substantial clinical impact that increases the odds of a longer post-operative recovery or adverse events overall.

Some studies argue that patients have greater pain levels following thoracotomy, which requires longer admission (10,12). However, even this observation is not universally accepted. The VIOLET trial, a blinded, randomized control study published by Dr. Lim and colleagues, examined outcomes following VATS and open lobectomy (2). While their focus was to examine post-surgical recovery of physical function, part of their protocol included blinding patients from their procedure (2). They accomplished this by placing a bandage on the chest large enough to conceal all operative incisions. Before discharge, the patients were then asked which group they had been assigned to: VATS surgery or open thoracotomy (2). Only 51.2% of VATS patients, and 45.74% of thoracotomy patients were able to correctly identify their surgical approach, where a score of 50% would suggest random chance. This means that a majority of patients in both groups were unable to correctly guess their procedure (13). Worse, more than half of thoracotomy patients, who should have meaningfully worse pain scores, thought they had small incisions (2). To note, the authors also publish that more thoracotomy patients require pain medications for >5 weeks following surgery, with a similar trend in pain scores. However, there were also no significant differences in performance metrics or quality of life (2). Post-operative pain is only a meaningful factor in surgical decision-making if the experience affects the patients. If patients are unable to tell which surgery they had, it strongly suggests that pain may be equivocal between the two procedures and should not, by itself, affect surgeon’s judgment.

Many publications report a different finding. But to the authors’ knowledge, this is the only study that has successfully limited the influence that knowing the approach may have on subjective pain reporting.

Lung cancer is a women’s disease

The data convincingly shows that women are more likely to undergo minimally invasive approaches for lung resection, but it also shows that women are more likely to receive lung surgery overall. From 1992 to 2015, 9,451 women underwent surgery for lung cancer compared to only 8,035 men. In 2023, there were an estimated 120,790 women diagnosed with lung cancer compared to 117,550 men (14). There were a total of 59,910 deaths from lung cancer in women, which is only 9,560 less than breast, ovarian, and uterine cancer combined (n=69,470) (15). By comparison, there were 67,160 deaths in men.

This has many potential explanations. Different patterns of diagnosis, compliance with medical care, screening programs, and risk factor mitigation all contribute to lung cancer outcomes, even before considering the possibility of gender-related incidence of resectable disease, or tumors with favorable responses to therapy. Whatever the cause, more women are being diagnosed with lung cancer and treated with lung surgery. While this should emphasize the importance of lung screening programs to male patients, it should also codify the importance of lung cancer as a women’s health issue.

Black Americans undergo minimally invasive lobectomy less frequently than White Americans

Though there are many potential explanations for this finding, the most troubling is the possibility of widespread disparity in access to healthcare. Unfortunately, racial inequity in both lung cancer screening and treatment are well documented in the literature (16-18).

But of graver concern is the observation that patients who undergo open lobectomy tend to be less healthy, with more advanced disease. Instead of restricted access to modern surgical resources, it is possible that Black Americans are less likely to receive MIS because of a wider healthcare crisis. Namely, that at the time of diagnosis, Black Americans present with more severe concomitant illness and later-stage cancers. This suggests that Black Americans may be receiving worse medical care overall or may not have access to lung cancer screening (14). The treatment of lung cancer begins with education, prevention, and surveillance programs before diagnosis and therapy. If any of these elements are unavailable, there would be significant downstream consequences for treatment options at the time of diagnosis.

There are several important limitations in this study. First, large-scale national databases reflect broad trends but may have narrow application to regional or individual practices. Second, the SEER Medicare database will have reporting bias, since the data included only reflects Medicare patients. Finally, the study involves data from 1992 to 2015. While this may favor more complete reporting, it may not reflect current trends.

Conclusions

In this analysis, we have identified several significant differences between minimally invasive and open lung surgery. However, it is not clear if any of these statistical outcomes translate to actual clinical benefits. Patients undergoing MIS are healthier with earlier-stage disease, and would therefore be expected to have lower rates of adverse events. Patients undergoing thoracotomy have more comorbid conditions, more advanced disease, and include operative catastrophes that were converted from video- or robotic-assisted procedures. Such biases in patient selection and surgical judgment are inherent to retrospective studies and will obscure outcomes-based research.

However, the data does reinforce the imperative that we consider lung cancer a women’s health issue. It also raises the distinct possibility that Black Americans do not have equal access to lung cancer screening and treatment. Or worse, that diagnoses are occurring later in Black patients with limited access to health care resources overall.

We must also remember that patient selection and operative decision-making are silent biases inherent to these retrospective studies. While strong statistical results should support the continued use of VATS and robotic surgery, they should not substitute good surgical judgment. Poorer outcomes, blood transfusions, longer hospital stays and complications may reflect the conditions that require thoracotomy instead of its consequences. Open surgery should be viewed as a tool, not as a failure, especially when surgeons are confronted by critical conditions, complex anatomy, or advanced disease. The technical ability of a surgeon to complete a minimally invasive procedure has no virtue if it does not confer an advantage to the patient.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-1287/rc

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-1287/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-1287/coif). R.F. serves as an unpaid editorial board member of Translational Cancer Research from May 2024 to December 2026. The other 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 and its subsequent amendments.

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. Shagabayeva L, Fu B, Panda N, et al. Open, Video- and Robot-Assisted Thoracoscopic Lobectomy for Stage II-IIIA Non-Small Cell Lung Cancer. Ann Thorac Surg 2023;115:184-90. [Crossref] [PubMed]
2. Lim E, Harris RA, McKeon HE, et al. Impact of video-assisted thoracoscopic lobectomy versus open lobectomy for lung cancer on recovery assessed using self-reported physical function: VIOLET RCT. Health Technol Assess 2022;26:1-162. [Crossref] [PubMed]
3. Flores RM, Park BJ, Dycoco J, et al. Lobectomy by video-assisted thoracic surgery (VATS) versus thoracotomy for lung cancer. J Thorac Cardiovasc Surg 2009;138:11-8. [Crossref] [PubMed]
4. Chai T, Lin Y, Kang M, et al. Thoracotomy versus video-assisted thoracoscopic resection of lung cancer: A protocol for a systematic review and meta-analysis. Medicine (Baltimore) 2019;98:e14646. [Crossref] [PubMed]
5. Egevad L, Heanue M, Berney D, et al. Cancer Incidence in Five Continents. In: Curado M, Edwards B, Shin H, editors. Chapter 4: Histological groups. Lyon, France: International Agency for Research on Cancer; 2007:61-66.
6. Erdoğu V, Çıtak N, Sezen CB, et al. Comparison of 6th, 7th, and 8th editions of the TNM staging in non-small cell lung cancer patients: Validation of the 8th edition of TNM staging. Turk Gogus Kalp Damar Cerrahisi Derg 2022;30:395-403.
7. Rural-Urban Commuting Area Codes. Economic Research Service, United States Department of Agriculture. Available online: https://www.ers.usda.gov/data-products/rural-urban-commuting-area-codes
8. Moore BJ, White S, Washington R, et al. Identifying Increased Risk of Readmission and In-hospital Mortality Using Hospital Administrative Data: The AHRQ Elixhauser Comorbidity Index. Med Care 2017;55:698-705. [Crossref] [PubMed]
9. Sabbula BR, Gasalberti DP, Mukkamalla SKR, et al. Squamous Cell Lung Cancer. 2025; [PubMed]
10. Kim D, Woo W, Shin JI, et al. The Uncomfortable Truth: Open Thoracotomy versus Minimally Invasive Surgery in Lung Cancer: A Systematic Review and Meta-Analysis. Cancers (Basel) 2023;15:2630. [Crossref] [PubMed]
11. Housman B, Flores RM. Minimally Invasive vs Open Lobectomy for Lung Cancer: Safety Is the Selection Bias. Ann Thorac Surg 2023;115:191. [Crossref] [PubMed]
12. Maconachie R, Mercer T, Navani N, et al. Lung cancer: diagnosis and management: summary of updated NICE guidance. BMJ 2019;364:l1049. [Crossref] [PubMed]
13. Lim E, Darlison L, Edwards J, et al. Mesothelioma and Radical Surgery 2 (MARS 2): protocol for a multicentre randomised trial comparing (extended) pleurectomy decortication versus no (extended) pleurectomy decortication for patients with malignant pleural mesothelioma. BMJ Open 2020;10:e038892. [Crossref] [PubMed]
14. Cancer Facts & Figures 2023. Available online: https://www.cancer.org/research/cancer-facts-statistics/all-cancer-facts-figures/2023-cancer-facts-figures.html
15. Cancer Stat Facts: Common Cancer Sites. Available online: https://seer.cancer.gov/statfacts/html/common.html
16. Kathuria H, Wiener RS. Toward Racial Equity in Lung Cancer Screening Eligibility. J Clin Oncol 2024;42:2001-4. [Crossref] [PubMed]
17. Evans N 3rd, Grenda T, Alvarez NH, et al. Narrative review of socioeconomic and racial disparities in the treatment of early stage lung cancer. J Thorac Dis 2021;13:3758-63. [Crossref] [PubMed]
18. Harrison S, Judd J, Chin S, et al. Disparities in Lung Cancer Treatment. Curr Oncol Rep 2022;24:241-8. [Crossref] [PubMed]