Stereotactic Body Radiation Therapy vs Surgery for Stage IA Non–Small Cell Lung Cancer

Importance: Stereotactic body radiation therapy (SBRT) is a well-established alternative to surgery for nonoperable patients with early-stage non-small cell (NSCLC) lung cancer. However, the role of SBRT in operable patients is unclear as 3 randomized clinical trials comparing SBRT with surgery were closed early because of poor accrual. Existing observational comparisons are limited by confounding because patients who receive SBRT are generally frailer and more comorbid. Objective: To compare cancer-specific mortality between surgery and SBRT using proximal causal inference to adjust for unmeasured confounding due to frailty and comorbidity using proxy variables. Design, Setting, and Participants: This cohort study was a national database analysis of the Surveillance, Epidemiology, and End Results (SEER)-Medicare dataset. Three treatment effect estimates were obtained with increasing control of confounding: unadjusted, standard-regression adjusted for measured confounders as in existing studies, and proximally adjusted for unmeasured confounders. Participants were patients with stage IA NSCLC who underwent sublobar resection or SBRT from 2010 to 2019, followed up through December 21, 2021. Data analysis was performed from April 2023 to August 2026. Exposures: Sublobar resection vs SBRT. Negative control exposures were constructed from billing codes. Main Outcome and Measures: Lung cancer-specific mortality. Negative control outcomes were constructed from other-cause mortality. Results are reported as hazard differences (HD) and 95% CI. Results: A total of 4251 patients met inclusion and exclusion criteria; 2581 (60.7%) were female and 1670 (39.3%) were male. Their mean (SD) age was 75.6 (6.3) years. A total of 2558 patients underwent SBRT and 1693 underwent surgery. In the unadjusted and covariate-adjusted analyses, SBRT was associated with higher lung cancer cause-specific mortality, HD of 0.047 person-years (95% CI, 0.038-0.055) and 0.024 person-years (95% CI, 0.014-0.034), respectively. However, there was also compelling evidence of confounding due to a strong estimated association with other-cause mortality. In the proximal analysis, other-cause mortality together with negative control exposures were used to indirectly control for hidden confounders, yielding an HD of 0.012 person-years (95% CI, 0.0013-0.027). Conclusions and Relevance: This study empirically showed that existing observational comparisons of SBRT and surgery are biased because of unmeasured confounding. When proximal causal inference was used to better control for confounding due to frailty and comorbidity, there continued to be a smaller but significant improvement in outcomes associated with surgery, though the effects of residual bias due to differences in pathologic and clinical staging could not be ruled out. This study demonstrates that proximal causal inference can be used to control for unmeasured confounders in a variety of settings.

Authors

Institutions

Publication Details

Journal
JAMA Surgery
Published
2026-11-18
DOI
https://doi.org/10.1001/jamasurg.2026.5510
Citations
1
Primary Topic
Lung Cancer Diagnosis and Treatment
Type
article
Field-Weighted Citation Impact
3.10
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Stereotactic Body Radiation Therapy vs Surgery for Stage IA Non–Small Cell Lung Cancer

Chi‐Fu Jeffrey Yang, Eric J. Tchetgen Tchetgen, Kendrick Qijun Li, Zhi Ven Fong et al.
1 citations
JAMA Surgery
Lung Cancer Diagnosis and Treatment
3.10
article

Stereotactic Body Radiation Therapy vs Surgery for Stage IA Non–Small Cell Lung Cancer

Chi‐Fu Jeffrey Yang, Eric J. Tchetgen Tchetgen, Kendrick Qijun Li, Zhi Ven Fong, David Kozono, George C. Linderman, Alexandra L. Potter, Timothy Hwang
article en
1 citations

Abstract

Importance: Stereotactic body radiation therapy (SBRT) is a well-established alternative to surgery for nonoperable patients with early-stage non-small cell (NSCLC) lung cancer. However, the role of SBRT in operable patients is unclear as 3 randomized clinical trials comparing SBRT with surgery were closed early because of poor accrual. Existing observational comparisons are limited by confounding because patients who receive SBRT are generally frailer and more comorbid. Objective: To compare cancer-specific mortality between surgery and SBRT using proximal causal inference to adjust for unmeasured confounding due to frailty and comorbidity using proxy variables. Design, Setting, and Participants: This cohort study was a national database analysis of the Surveillance, Epidemiology, and End Results (SEER)-Medicare dataset. Three treatment effect estimates were obtained with increasing control of confounding: unadjusted, standard-regression adjusted for measured confounders as in existing studies, and proximally adjusted for unmeasured confounders. Participants were patients with stage IA NSCLC who underwent sublobar resection or SBRT from 2010 to 2019, followed up through December 21, 2021. Data analysis was performed from April 2023 to August 2026. Exposures: Sublobar resection vs SBRT. Negative control exposures were constructed from billing codes. Main Outcome and Measures: Lung cancer-specific mortality. Negative control outcomes were constructed from other-cause mortality. Results are reported as hazard differences (HD) and 95% CI. Results: A total of 4251 patients met inclusion and exclusion criteria; 2581 (60.7%) were female and 1670 (39.3%) were male. Their mean (SD) age was 75.6 (6.3) years. A total of 2558 patients underwent SBRT and 1693 underwent surgery. In the unadjusted and covariate-adjusted analyses, SBRT was associated with higher lung cancer cause-specific mortality, HD of 0.047 person-years (95% CI, 0.038-0.055) and 0.024 person-years (95% CI, 0.014-0.034), respectively. However, there was also compelling evidence of confounding due to a strong estimated association with other-cause mortality. In the proximal analysis, other-cause mortality together with negative control exposures were used to indirectly control for hidden confounders, yielding an HD of 0.012 person-years (95% CI, 0.0013-0.027). Conclusions and Relevance: This study empirically showed that existing observational comparisons of SBRT and surgery are biased because of unmeasured confounding. When proximal causal inference was used to better control for confounding due to frailty and comorbidity, there continued to be a smaller but significant improvement in outcomes associated with surgery, though the effects of residual bias due to differences in pathologic and clinical staging could not be ruled out. This study demonstrates that proximal causal inference can be used to control for unmeasured confounders in a variety of settings.

JAMA Surgery
St. Jude Children's Research Hospital (US), Harvard University (US), Massachusetts General Hospital (US), Dana-Farber Cancer Institute (US), Mayo Clinic in Arizona (US), Mayo Clinic Hospital (US), Dana-Farber/Harvard Cancer Center (US), University of Pennsylvania (US)
Good health and well-being
Openalex Percentile: Top 16%
Lung Cancer Diagnosis and Treatment
3.10
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon