Initial Patient Characteristics of TSOG 102: A Multicenter Prospective Registry of Active Surveillance in Patients with Multiple Ground Glass Opacities.

OBJECTIVE: Presentation with multiple ground glass opacities (GGOs) is an increasingly common occurrence, and the optimal management of these lesions is unclear. Active surveillance has been increasingly adopted as a management strategy for other low-grade malignancies. We hypothesized that active surveillance could be a feasible and safe option for patients with multiple GGOs. METHODS: Patients with ≥2 GGOs (ground glass predominant, <50% solid, ≤3 cm) were enrolled in a multi-institutional registry and prospectively followed up on active surveillance with computed tomography scans every 6 to 12 months. Each GGO was catalogued and measured individually at each follow-up visit. RESULTS: Target accrual was met, with 337 patients from 23 institutions. The mean age was 70 years (interquartile range, 65-77 years), and 74% were women. Most were former (70%) or current (9%) smokers, with a mean exposure of 30 pack-years (interquartile range [IQR], 15-44 pack-years). Half of the patients (51%) had a previous lung cancer, and the majority (86%) were already under surveillance at the time of study entry. The median number of GGOs per patient was 3 (IQR, 2-5), with a total of 1467 GGOs under surveillance. The median GGO size was 0.9 cm (IQR, 0.7-1.3 cm). Most GGOs were 0.5 to 1 cm in size. CONCLUSIONS: Active surveillance, rather than immediate intervention, was an acceptable option to patients, and accrual to this registry trial was feasible. Safety endpoints and long-term outcomes will be assessed in the planned 5-year follow-up in accordance with the protocol.

Association Between Utilization of Services and Perioperative Outcomes for Lung Cancer Resection.

INTRODUCTION: Lung cancer resection has largely focused on perioperative outcomes (eg, mortality) to benchmark performance. While variations in perioperative outcomes and in utilization of services (eg, ambulatory procedures, hospitalization) have been independently demonstrated, there has been limited evaluation of associations between these outcomes. We evaluated the association between perioperative outcomes and utilization of services to evaluate provider performance across a broader context of care. PATIENTS AND METHODS: This was a retrospective cohort study of patients undergoing lung cancer resection in 2017 to 2019. We utilized hierarchical logistic regression models to determine risk- and reliability-adjusted mortality and risk-adjusted utilization of services, at the hospital-level. We then evaluated utilization of services across quartiles of perioperative mortality. RESULTS: A total of 15,168 patients across 297 hospitals underwent lung cancer resection. Mean risk- and reliability-adjusted 90-day mortality varied between 1.58% (95% CI, 1.54%-1.62%) and 2.74% (95% CI, 2.59%-2.90%) across quartiles. Risk-adjusted utilization of all ambulatory procedures was highest in the best performing (lowest mortality) quartile at 37.7% (95% CI, 33.6%-41.8%). Additionally, risk-adjusted inpatient utilization prior to and after surgery was lowest in the best performing quartile at 15% (95% CI, 13.7%-16.3%) and 19.3% (95% CI, 17.5%-21.0%), respectively. CONCLUSIONS: Hospitals with the lowest perioperative mortality demonstrated trends towards using more outpatient resources prior to surgery, but fewer inpatient services surrounding lung cancer resection. This correlation highlights the importance of incorporating utilization of services in addition to other metrics to profile the efficiency and effectiveness of centers performing lung cancer resection across a broader spectrum of care.

Data Valuation with Gradient Similarity.

High-quality data is crucial for accurate machine learning and actionable analytics, however, mislabeled or noisy data is a common problem in many domains. Distinguishing low- from high-quality data can be challenging, often requiring expert knowledge and considerable manual intervention. Data Valuation algorithms are a class of methods that seek to quantify the value of each sample in a dataset based on its contribution or importance to a given predictive task. These data values have shown an impressive ability to identify mislabeled observations, and filtering low-value data can boost machine learning performance. In this work, we present a simple alternative to existing methods, termed Data Valuation with Gradient Similarity (DVGS). This approach can be easily applied to any gradient descent learning algorithm, scales well to large datasets, and performs comparably or better than baseline valuation methods for tasks such as corrupted label discovery and noise quantification. We evaluate the DVGS method on tabular, image and RNA expression datasets to show the effectiveness of the method across domains. Our approach has the ability to rapidly and accurately identify low-quality data, which can reduce the need for expert knowledge and manual intervention in data cleaning tasks.

Factors Associated With Margin Positivity After Lung Resection Surgery.

INTRODUCTION: The purpose of this study is to utilize a representative national sample to investigate the factors associated with margin positivity after attempted surgical resection. Given the changes in surgical approaches to lung cancer for the last 10 years, margin positivity and outcomes between robotic, video assisted thoracoscopic surgery (VATS) and open surgical resections may vary. METHODS: This retrospective cohort study utilized the National Cancer Database. Patients with non-small-cell lung cancer, 18 or older and who had a surgical lung resection between 2010 and 2019 were included. Demographic data, along with patient-level clinical variables were extracted. Patient-level outcome variables including 30-day, 90-day mortality and readmission rates were analyzed. Univariable and multivariable logistic regression was utilized to assess factors associated with margin positivity. RESULTS: A total of 226,884 patients were identified. Of the total cohort, 9229 had positive margins (4.2%). Patients with positive margins had statistically significant increased 30-day, 90-day mortality, as well as increased readmission rate. Older age, male sex, patients undergoing an open resection, patients who underwent a wedge resection, higher clinical stage, larger tumor size, squamous and adenosquamous histologies, and higher Charlson-Deyo Comorbidity Index were all associated with having a positive margin after resection. CONCLUSION: In conclusion, there was no difference in margin positivity when comparing robotic and VATS resection, however, open resection had increased rates of margin positivity. Increasing tumor size, clinical stage, squamous and adenosquamous histologies, male sex, and patients undergoing a wedge resection were all associated with increased rates of margin positivity.

Graph Structured Neural Networks for Perturbation Biology.

Computational modeling of perturbation biology identifies relationships between molecular elements and cellular response, and an accurate understanding of these systems will support the full realization of precision medicine. Traditional deep learning, while often accurate in predicting response, is unlikely to capture the true sequence of involved molecular interactions. Our work is motivated by two assumptions: 1) Methods that encourage mechanistic prediction logic are likely to be more trustworthy, and 2) problem-specific algorithms are likely to outperform generic algorithms. We present an alternative to Graph Neural Networks (GNNs) termed Graph Structured Neural Networks (GSNN), which uses cell signaling knowledge, encoded as a graph data structure, to add inductive biases to deep learning. We apply our method to perturbation biology using the LINCS L1000 dataset and literature-curated molecular interactions. We demonstrate that GSNNs outperform baseline algorithms in several prediction tasks, including 1) perturbed expression, 2) cell viability of drug combinations, and 3) disease-specific drug prioritization. We also present a method called GSNNExplainer to explain GSNN predictions in a biologically interpretable form. This work has broad application in basic biological research and pre-clincal drug repurposing. Further refinement of these methods may produce trustworthy models of drug response suitable for use as clinical decision aids. Availability and implementation: Our implementation of the GSNN method is available at https://github.com/nathanieljevans/GSNN. All data used in this work is publicly available.

Wedge Resection Outcomes: A Comparison of Video-Assisted and Robot-Assisted Wedge Resections.

BACKGROUND: Quality of oncologic resection for early-stage non-small cell lung cancer (NSCLC) may differ by surgical approach. Minimally invasive surgery has become the standard for surgical treatment of NSCLC. Our study compares quality of wedge resection by video-assisted thoracoscopic surgery (VATS) vs robotic video-assisted thoracoscopic surgery (RVATS). We hypothesized that RVATS would result in higher quality resections and improved patient outcomes. METHODS: A retrospective cohort analysis was completed using the National Cancer Database for patients with clinical stage 1 NSCLC with tumor size ≤2 cm who underwent a minimally invasive surgery wedge resection from 2010 to 2019. Wedge resections approached with RVATS were compared with VATS. A 1:1 propensity score matched analysis was performed. RESULTS: The cohort included 16,559 patients; 80.4% (13,406) received VATS and 18.9% (3153) received RVATS. Compared with RVATS, a VATS approach was associated with a lower likelihood of lymph nodes being examined (59.0% vs 75.2%; P < .001), fewer nodes dissected (median, 4 vs 5; P < .001), and less adjuvant systemic therapy administered (1.3% vs 2.2%; P < .001). Propensity score matching resulted in 2590 balanced pairs. Statistical significance was maintained for likelihood of lymph nodes examined, number of nodes dissected, and adjuvant systemic therapy administered. There was no significant difference in nodal upstaging after propensity score matching (3.7% vs 4.3%; P = .37). CONCLUSIONS: Compared with the VATS approach, wedge resections by RVATS for early-stage NSCLC were more likely to be associated with increased lymph nodes resected. These data may support increased use of RVATS for wedge resections.

Association between surgical quality and long-term survival in lung cancer.

OBJECTIVES: There are significant variations in both perioperative and long-term outcomes after lung cancer resection. While perioperative outcomes are often used as comparative measures of quality, they are unreliable, and their association with long-term outcomes remain unclear. In this context, we evaluated whether historical perioperative mortality after lung cancer resection is associated with 5-year survival. PATIENTS AND METHODS: The National Cancer Database (NCDB) was queried to identify patients diagnosed with non-small cell lung cancer (NSCLC) in 2010-2016 who underwent surgical resection (n = 234200). Hospital-level reliability-adjusted 90-day mortality rate quartiles for 2010-2013 was used as the independent variable to analyze 5-year survival for patients diagnosed in 2014-2016 (n = 85396). RESULTS: There were 85,396 patients in the 2014-2016 cohort across 1,086 hospitals. Overall observed 90-day mortality rate was 3.2% (SD 17.6%) with 2.6% (SD 16.0%) for the historically best performing quartile vs. 3.9% (SD 19.4%) for the worst performing quartile (p < 0.0001). Patients who underwent resection at hospitals with the best historical mortality rate had significantly better 5-year survival across all stages compared to those treated at hospitals in the worst performing quartile in multivariate Cox regression analysis (all stages - HR 1.21 [95% CI 1.15-1.26]; stage I - HR 1.19 [95% CI 1.12-1.25]; stage II - HR 1.20 [95% CI 1.09-1.32]; stage III - HR 1.36 [95% CI 1.20-1.54]) and Kaplan-Meier survival estimates (all stages - p < 0.0001, stage I - p < 0.0001; stage II - p = 0.0004; stage III - p < 0.0001). CONCLUSION: With expanded lung cancer screening criteria and likely increase in early-stage detection, profiling performance is paramount to ensuring mortality benefits. We found that episodes surrounding surgical resection may be used to profile long-term outcomes that likely reflect quality across a broader context of care. Evaluating lung cancer care quality using perioperative outcomes may be useful in profiling provider performance and guiding value-based payment policies.

Leveraging influencers to increase HPV vaccination intention: The impact of message framing and health regulatory fit using repeated measures.

Through a series of two online experiments that incorporate a repeated measures design, we examine how message framing (loss versus gain) within Instagram influencers' posts interact with consumers' health regulatory orientation (promotion versus prevention) to impact HPV vaccination intention between two data collection points. Findings indicate that among those who are more prevention oriented, exposure to a loss-framed influencer advertisement was effective at increasing intention to receive an HPV vaccine relative to those that were exposed to gain-framed influencer advertisement. Based on these findings we offer theoretical and managerial implications.

Access to Guideline Concordant Care for Node-Positive Non-Small Cell Lung Cancer in the United States.

BACKGROUND: This study sought to determine whether seeking care at multiple Commission on Cancer (CoC) hospitals is associated with different rates of receiving guideline-concordant care (GCC) among patients with non-small cell lung cancer (NSCLC). METHODS: The National Cancer Database was queried for the years 2004 to 2018 for patients with margin-negative pT1 to pT3 N1 to N2 M0 noncarcinoid NSCLC without neoadjuvant therapy. GCC was defined as chemotherapy for pN1 disease and as chemotherapy with or without radiation for pN2 disease. Patients who received care at >1 facility were examined separately. Factors previously associated with barriers to care were compared between groups. Kaplan-Meier analysis with log-rank tests analyzed 5-year overall survival (OS). Propensity score matching was performed to compare the effect sizes of race, insurance status, and income. RESULTS: In total 44,531 patients met inclusion criteria, 11,980 (26.9%) of whom sought care at >1 CoC institution. Among patients with pN1 disease, 5565 (76.7%) received GCC if they visited >1 facility vs 13,995 (68.5%) patients who sought care at 1 facility (P < .001). For patients with pN2 disease, 3991 (84.4%) received GCC if they visited >1 facility vs9369 (77.4%) patients receiving care at 1 facility (P < .001). Visiting >1 facility was associated with higher OS at 5 years (4784 [54.35%] vs 10,215 [45.62%]; P < .001). CONCLUSIONS: Visiting >1 CoC institution is associated with higher rates of GCC for individuals with pN1 to pN2 lung cancer. Patients who received care at >1 facility had higher OS at 5 years. Further study is warranted to identify factors associated with the ability of patients to seek care at multiple facilities.

Robotic First Rib Resection in Thoracic Outlet Syndrome: A Systematic Review of Current Literature.

Thoracic outlet syndrome (TOS) involves the compression of neurovascular structures in the thoracic outlet. TOS subtypes, including neurogenic (nTOS), venous (vTOS), and arterial (aTOS) are characterized by distinct clinical presentations and diagnostic considerations. This review explores the incidence, diagnostic challenges, and management of TOS with a focus on the innovative approach of Robotic First Rib Resection (R-FRR). Traditional management of TOS includes conservative measures and surgical interventions, with various open surgical approaches carrying risks of complications. R-FRR, a minimally invasive technique, offers advantages such as improved exposure, reduced injury risk to neurovascular structures, and shorter hospital stays. A comprehensive literature review was conducted to assess the outcomes of R-FRR for TOS. Data from 12 selected studies involving 397 patients with nTOS, vTOS, and aTOS were reviewed. The results indicate that R-FRR is associated with favorable intraoperative outcomes including minimal blood loss and low conversion rates to traditional approaches. Postoperatively, patients experienced decreased pain, improved function, and low complication rates. These findings support R-FRR as a safe and effective option for medically refractory TOS.

Is wedge a dirty word? Demographic and facility-level variables associated with high-quality wedge resection.

Objectives: Although sublobar resections have gained traction, wedge resections vary widely in quality. We seek to characterize the demographic and facility-level variables associated with high-quality wedge resections. Methods: The National Cancer Database was queried from 2010 to 2018. Patients with T1/T2 N0 M0 non-small cell lung cancer 2 cm or less who underwent wedge resection without neoadjuvant therapy were included. A wedge resection with no nodes sampled or with positive margins was categorized as a low-quality wedge. A wedge resection with 4 or more nodes sampled and negative margins was categorized as a high-quality wedge. Facility-specific variables were investigated via quartile analysis based on the overall volume and proportion of high-quality wedge or low-quality wedge resections performed. Results: A total of 21,742 patients met inclusion criteria, 6390 (29.4%) of whom received a high-quality wedge resection. Factors associated with high-quality wedge resection included treatment at an academic center (3005 [47.0%] vs low-quality wedge 6279 [40.9%]; P < .001). The 30- and 90-day survivals were similar, but patients who received a high-quality wedge resection had improved 5-year survival (4902 [76.7%] vs 10,548 [68.7%]; P < .001). Facilities in the top quartile by volume of high-quality wedge resections performed 69% (4409) of all high-quality wedge resections, and facilities in the top quartile for low-quality wedge resections performed 67.6% (10,378) of all low-quality wedge resections. A total of 113 facilities were in the top quartile by volume for both high-quality wedge and low-quality wedge resections. Conclusions: High-quality wedge resections are associated with improved 5-year survival when compared with low-quality wedge resections. By volume, high-quality wedge and low-quality wedge resections cluster to a minority of facilities, many of which overlap. There is discordance between best practice guidelines and current practice patterns that warrants additional study.

Individual- and neighborhood-level characteristics of lung cancer screening participants undergoing telemedicine shared decision making.

BACKGROUND: Although lung cancer screening (LCS) for high-risk individuals reduces lung cancer mortality in clinical trial settings, many questions remain about how to implement high-quality LCS in real-world programs. With the increasing use of telemedicine in healthcare, studies examining this approach in the context of LCS are urgently needed. We aimed to identify sociodemographic and other factors associated with screening completion among individuals undergoing telemedicine Shared Decision Making (SDM) for LCS. METHODS: This retrospective study examined patients who completed Shared Decision Making (SDM) via telemedicine between May 4, 2020 - March 18, 2021 in a centralized LCS program. Individuals were categorized into Complete Screening vs. Incomplete Screening subgroups based on the status of subsequent LDCT completion. A multi-level, multivariate model was constructed to identify factors associated with incomplete screening. RESULTS: Among individuals undergoing telemedicine SDM during the study period, 20.6% did not complete a LDCT scan. Bivariate analysis demonstrated that Black/African-American race, Medicaid insurance status, and new patient type were associated with greater odds of incomplete screening. On multi-level, multivariate analysis, individuals who were new patients undergoing baseline LDCT or resided in a census tract with a high level of socioeconomic deprivation had significantly higher odds of incomplete screening. Individuals with a greater level of education experienced lower odds of incomplete screening. CONCLUSIONS: Among high-risk individuals undergoing telemedicine SDM for LCS, predictors of incomplete screening included low education, high neighborhood-level deprivation, and new patient type. Future research should focus on testing implementation strategies to improve LDCT completion rates while leveraging telemedicine for high-quality LCS.

Intraoperative Versus Preoperative Diagnosis of Lung Cancer: Differences in Treatments and Patient Outcomes.

OBJECTIVES: Non-small cell lung cancer (NSCLC) is frequently diagnosed during surgical resection. It remains unclear if lack of preoperative tissue diagnosis influences likelihood of receipt of guideline-concordant care or postoperative outcomes. METHODS: A retrospective cohort analysis was completed utilizing the National Cancer Database for patients undergoing lung resection with clinical stage 1 NSCLC from 2004 to 2018. Diagnosis during resection was defined as zero days between diagnosis and definitive lung resection. Patients receiving neoadjuvant therapy were excluded. Subgroup analyses were completed by resection type, including wedge resection. RESULTS: The cohort included 91,328 patients, 33,517 diagnosed during definitive resection and 57,811 diagnosed preoperatively. For patients diagnosed preoperatively, median time from diagnosis to surgery was 42 days (interquartile range 28-63 days). Patients diagnosed intraoperatively had smaller median tumor size (1.7 cm vs. 2.5 cm, P < .01) and were more likely to undergo wedge resection (10,668 [31.8%] vs. 7,617 [13.2%], P < .01). Intraoperative diagnosis resulted in lower likelihood of nodal sampling (27,356 [81.9%] vs. 53,183 [92.4%], P < .01) and nodal upstaging (2,482 [9.7%] vs. 7701 [15.5%], P < .01). Amongst patients with intraoperative diagnoses, those treated via wedge resection were less likely to undergo lymph node sampling (5,515 [52.0%] vs. 5,606 [61.1%], P < .01). Amongst patients with positive lymph nodes, patients diagnosed intraoperatively were less likely to receive adjuvant therapy (1,677 [5.0%] vs. 5,669 [9.8%], P < .01). CONCLUSIONS: Preoperative tissue diagnosis of NSCLC is associated with more frequent lymph node harvest, increased rates of upstaging and receipt of adjuvant therapy. Preoperative workup may contribute to increased rates of guideline-concordant lung cancer care.

Disparities in lung cancer.

Lung cancer is the second most common cancer and the leading cause of cancer death among men and women in the United States. Despite a substantial decline in lung cancer incidence and mortality across all races in the last few decades, medically underserved racial and ethnic minority populations continue to carry the greatest burden of disease throughout the lung cancer continuum. Black individuals experience a higher incidence of lung cancer due to lower rates of low-dose computed tomography screening, which translate into advanced disease stage at diagnosis and poorer survival outcomes compared with White individuals. With respect to treatment, Black patients are less likely to receive gold standard surgery, have access to biomarker testing or high-quality treatment compared with White patients. The reasons for those disparities are multifactorial and include socioeconomic (eg, poverty, lack of health insurance, and inadequate education), and geographic inequalities. The objective of this article is to review the sources of racial and ethnic disparities in lung cancer, and to propose recommendations to help address them.

Defining risk factors for mortality after emergent hiatal hernia repair in the era of minimally invasive surgery.

BACKGROUND: Risk factors for mortality following emergent hiatal hernia (HH) repair in the era of minimally invasive surgery remain poorly defined. METHODS: Data was obtained from the National Inpatient Sample (NIS), National Readmissions Database, and National Emergency Department Sample for patients undergoing HH repair between 2010 and 2018. Univariate and multivariate logistic regression analyses reported with odds ratio (OR) and 95% confidence intervals (CI) were performed to identify factors associated mortality. RESULTS: Via the NIS, mortality rate was 2.2% (147 patients). Via the NEDS, the mortality rate was 3.6% (303 patients). On multivariate analysis, predictors of mortality included age (OR 1.05, CI: 1.04,1.07), male sex (OR 1.49, CI: 1.06,2.11), frailty (OR 2.49, CI: 1.65,3.75), open repair (OR 3.59, CI: 2.50,5.17), and congestive heart failure (OR 2.71, CI: 1.81,4.06). CONCLUSIONS: There are multiple risk factors for mortality after hiatal hernia repair. There is merit to a laparoscopic approach even in emergent settings.

Allopatric mosaics in the Indo-West Pacific crab subfamily Chlorodiellinae reveal correlated patterns of sympatry, genetic divergence, and genitalic disparity.

Molecular studies have revealed that many species once thought to be wide-ranging in the Indo-West Pacific contain allopatric mosaics of endemic lineages. These lineages provide compelling evidence that substantial time is needed to evolve isolating mechanisms sufficient to permit successful secondary sympatry, and that divergence is initiated in allopatry. In this context, questions arise regarding the nature, timing, and origin of isolating mechanisms that permit secondary sympatry. We present a phylogeny of the crab subfamily Chlorodiellinae which displays allopatric mosaics within species. These allopatric lineages typically do not have divergent male genitalia, while older sympatric lineages do. We tested the relationship between genetic distance (proxy for time), sympatry, and the divergence of male genitalic morphology. Our results suggest that male genitalic divergence is not involved in the initiation of speciation in chlorodielline crabs, having likely occurred only after isolation began in allopatry. However, morphological evolution of genitalia seemingly does play an important role in completing the process of speciation in these crabs.

Surgical Outcomes in the National Lung Screening Trial Compared With Contemporary Practice.

BACKGROUND: The National Lung Screening Trial (NLST) established a role for lung cancer screening. Mortality benefits with screening are predicated on successful treatment with low surgical mortality. Given variations observed in perioperative outcomes after lung cancer resection, it remains unknown whether benefits observed in the NLST are generalizable to a broader population. We sought to determine whether NLST perioperative outcomes are reflective of contemporary practice in a national cohort. METHODS: We identified patients diagnosed with non-small cell lung cancer who underwent lung resection in the 2014 to 2015 National Cancer Database (NCDB) and the NLST. We compared demographic and cancer characteristics in both datasets. We used hierarchical logistic regression to compare 30-day and 90-day postoperative mortality across facilities in both datasets. RESULTS: In all, 65054 patients in NCDB and 1003 patients in the NLST treated across 1119 NCDB hospitals and 33 NLST hospitals were included. After risk and reliability adjustment, mean 30-day and 90-day mortality were significantly higher among NCDB hospitals (mean 30-day, 2.2 [95% confidence interval (CI), 2.2 to 2.2] vs 1.8 [95% CI, 1.8 to 1.8], P < .001; mean 90-day, 4.2 [95% CI, 4.2 to 4.3] vs 2.9 [95% CI, 2.9 to 2.9], P < .001). Variation in risk- and reliability-adjusted 30-day mortality (95% CI, 1.1% to 4.9%) and 90-day mortality (95% CI, 2.6% to 9.7%) was observed among NCDB hospitals. Adjusted mortality was similar among NLST facilities (30 days, 1.8% to 1.8%; 90 days, 2.9% to 2.9%). CONCLUSIONS: Risk- and reliability-adjusted postoperative mortality varies widely in a national cohort compared with outcomes observed in the NLST. Efforts to minimize this variation are needed to ensure that benefits of lung cancer screening are fully realized in the United States.

Hospital Utilization of Stereotactic Body Radiation Therapy and Rates of Surgical Refusal.

BACKGROUND: Increasingly, stereotactic body radiation therapy (SBRT) is used for patients unfit for or unwilling to undergo operation for early-stage non-small cell lung cancer. It remains unclear how SBRT utilization has influenced patient refusal of surgical resection. METHODS: A retrospective cohort analysis was completed using the National Cancer Database for patients with T1/T2 N0 M0 lesions from 2008 to 2017. Facilities were categorized into tertiles by SBRT/surgery ratio for each year of analysis. Propensity score matching was used to compare rates of surgical refusal and rates of postrefusal receipt of SBRT. Multivariable regression analysis was performed to evaluate effect size. RESULTS: The study included 129 901 patients; 63 048 were treated at low-tertile SBRT/surgery facilities, 41 674 at middle-tertile SBRT/surgery facilities, and 25 179 at high-tertile SBRT/surgery facilities. Patients refusing surgery at high SBRT/surgery facilities had fewer comorbid conditions and smaller tumors. Rates of SBRT after surgical refusal differed (low SBRT/surgery facilities, 17.2%; high SBRT/surgery facilities, 55.9%; P < .001). In a matched cohort of 76 636, surgical refusal differed (low SBRT/surgery facilities, 4.2%; high SBRT/surgery facilities, 6.0%; P < .001). On multivariable regression, treatment at a top-tertile SBRT/surgery facility was the largest risk factor for surgical refusal (odds ratio, 3.82 [3.53-4.13]; P < .001) and was most strongly associated with postrefusal receipt of SBRT (odds ratio, 6.11 [5.09-7.34]; P < .001). CONCLUSIONS: Patients treated at high SBRT-using facilities are more likely to refuse surgical resection and more likely to receive radiation therapy after surgical refusal. Further analysis is needed to better understand patient refusal of surgery in the setting of early-stage non-small cell lung cancer.