Surgeon Quality and Patient Survival After Resection for Non–Small-Cell Lung Cancer

PURPOSE

The quality and outcomes of curative-intent lung cancer surgery vary in populations. Surgeons are key drivers of surgical quality. We examined the association between surgeon-level intermediate outcomes differences, patient survival differences, and potential mitigation by processes of care.

PATIENTS AND METHODS

Using a baseline population-based surgical resection cohort, we derived surgeon-level cut points for rates of positive margins, nonexamination of lymph nodes, nonexamination of mediastinal lymph nodes, and wedge resections. Applying the baseline cut points to a subsequent cohort from the same population-based data set, we assign surgeons into three performance categories in reference to each metric: 1 (<25th percentile), 2 (25th-75th percentile), and 3 (>75th percentile). The sum of performance scores created three surgeon quality tiers: 1 (4-6, low), 2 (7-9, intermediate), and 3 (10-12, high). We used chi-squared, Wilcoxon-Mann-Whitney, and Kruskal-Wallis tests to compare patient characteristics between the baseline and subsequent cohorts and across surgeon tiers. We applied Cox proportional hazards models to examine the association between patient survival and surgeon performance tier, sequentially adjusting for clinical stage, patient characteristics, and four specific processes.

RESULTS

From 2009 to 2021, 39 surgeons performed 4,082 resections across the baseline and subsequent cohorts. Among 31 subsequent cohort surgeons, five were tier 1, five were tier 2, and 21 were tier 3. Tier 1 and 2 surgeons had significantly worse outcomes than tier 3 surgeons (hazard ratio [HR], 1.37; 95% CI, 1.10 to 1.72 and 1.19; 95% CI, 1.00 to 1.43, respectively). Adjustment for specific processes mitigated the surgeon-tiered survival differences, with adjusted HRs of 1.02 (95% CI, 0.8 to 1.3) and 0.93 (95% CI, 0.7 to 1.25), respectively.

CONCLUSION

Readily accessible intermediate outcomes metrics can be used to stratify surgeon performance for targeted process improvement, potentially reducing patient survival disparities.

4 metrics of lung cancer surgeons in performance tiers that predict long-term survival disparities.

INTRODUCTION

Although improving in recent years, the aggregate 5-year survival of all patients diagnosed with lung cancer in the United States is still only approximately 23%.¹ Most 5-year survivors have had curative-intent surgery for non–small-cell lung cancer (NSCLC). However, fewer than 50% of recipients of surgical resection survive 5 years.^2,3 In CALGB/Alliance 140503, patients with stage IA NSCLC had a 5-year disease-free survival of 64% and two thirds of deaths were attributed to lung cancer.⁴ Long-term survival disparities after curative-intent resection of NSCLC are well-described at the patient and institution levels.^5-8 Older patients, men, persons who actively smoke, those with more comorbidities, and the socioeconomically disadvantaged are at higher risk for poor long-term survival,^5-7 as are recipients of surgery at low-volume, nonteaching, and community-level institutions.^6,8 These factors are not readily corrected.

CONTEXT

• Key Objective

• Can readily retrievable intermediate surgical outcome benchmarks be used to categorize lung cancer surgeons into performance stata associated with long-term patient survival differences, potentially permitting identification of targets for corrective intervention?

• Knowledge Generated

• Four intermediate outcomes—rates of resections with positive margins, nonexamination of lymph nodes, nonexamination of mediastinal lymph nodes, and wedge resections—could be used to create aggregate performance scores, with which surgeons can be categorized into performance tiers associated with patient survival disparities. Adjustment for specific processes involved in curative-intent lung cancer surgery—preoperative positron emission tomography–computed tomography scanning, invasive mediastinal nodal staging, technique of surgical resection, and surgery with a lymph node collection kit—eliminated the survival disparities between tiers, suggesting the possibility of corrective intervention.

• Relevance (T.E. Stinchcombe)

• Determining the appropriate metrics to assess surgical quality and implementing process improvements are inherently challenging. This study provides preliminary data for future studies to improve surgical quality and patient outcomes.*

  *Relevance section written by JCO Associate Editor Thomas E. Stinchcombe, MD.

Surgeon-level factors have also been associated with long-term survival differences. Surgery performed by general surgeons or cardiovascular surgeons who do not focus on general thoracic surgery and surgeons with low case volumes has been associated with worse outcomes.^9-11 Provider-level disparities in surgical outcomes may be related to differences in processes of care. Processes of care are more readily susceptible to corrective interventions than structural drivers of outcomes disparities.^12,13 Ideally, there should be a strong link between process and outcome to justify the disruptiveness of practice change. Nevertheless, intervention at the surgeon- or surgery team–level might be a viable strategy for improving overall lung cancer survivorship in populations. For example, improving the selection of patients for surgical resection through guideline-concordant preoperative invasive nodal staging and using a lymph node specimen collection kit to improve the quality of pathologic nodal staging are associated with better long-term survival.^14,15

We examined readily retrievable, survival-impactful intermediate surgical outcomes. Hypothesizing that high-quality processes are transferable, we categorized surgeons into quality attainment tiers using intermediate outcomes metrics, compared the survival of patients who had curative-intent NSCLC resection by surgeons in the different tiers, and explored the potentially mitigating effects of specific processes.

PATIENTS AND METHODS

The Mid-South Quality of Surgical Resection Consortium

The Mid-South Quality of Surgical Resection (MS-QSR) consortium consists of 12 hospitals in five contiguous Dartmouth Hospital Referral Regions in Eastern Arkansas, North Mississippi, and Western Tennessee. With Institutional Review Board permission at all participating institutions, we constructed databases of all lung cancer resections at each hospital from 2009 to 2021. Eligible hospitals had five or more curative-intent lung cancer resections per year.¹⁵ Hospital location within the prespecified, contiguous Hospital Referral Regions was important to minimize leakage to nonparticipating institutions and improve the capture of all cases performed by surgeons who might operate at more than one hospital.¹⁶ We updated patient death information every 6 months from each institution's tumor registry, most recently on August 31, 2021, and censored all patients still alive on that date. The institutional review boards at each participating institution approved this research with a waiver of informed consent.

Surgeon Cohorts

From 2009 to 2021, we conducted a multiple baseline, staggered implementation study of a lymph node specimen collection kit designed to improve the quality of pathologic nodal staging for resected NSCLC.¹⁵ For this analysis, we used the defined cohort of surgeons captured within these five Hospital Referral Regions as a convenient sample, encapsulating all their curative-intent resections while present in the cohort. We aggregated cases by surgeon, irrespective of institution.

A preimplementation phase included all resections from 2009 until each institution's kit use activation date.¹⁵ We used the surgical resections in this phase to develop surgeon-level cut points of quality benchmarks (Appendix Figs A1 and A2, online only). The postimplementation phase included all resections performed when the kit became available for use at each institution. We used the cut points developed from the preimplementation phase to categorize surgeons into tiers for survival comparisons.

Intermediate Outcome Quality Metrics

We evaluated surgeon performance using four specific quality measures—rates of resection with positive margins,¹⁷ wedge resection,¹⁸ resection without lymph node examination (pNX),^19,20 and resection without mediastinal lymph node examination.²¹ These metrics have each been linked with poorer long-term survival, are primarily within the control of surgery teams, and are readily accessible in the pathology report.

To objectively describe the MS-QSR cohort quality characteristics (but not to rank surgeon performance), we categorized the completeness of resection by the International Association for the Study of Lung Cancer (IASLC) R-factor and the National Comprehensive Cancer Network (NCCN) recommendations for resection quality.^22,23 The IASLC R-factor defines resections as complete only with the combination of negative margins, systematic or lobe-specific nodal dissection, no reported extracapsular extension of nodal metastasis, and no metastasis in the highest resected mediastinal lymph node; incomplete if positive margins, extracapsular lymph node extension, nonresection of known nodal disease, or positive pleural or pericardial fluid cytology are reported; and uncertain, when margins are negative, but nodal dissection was inadequate, the highest resected mediastinal lymph node is involved, carcinoma in situ is present at the bronchial resection margin, or pleural lavage cytology is positive.^22,24,25 The NCCN quality measure is defined as the combination of anatomic resection, negative margins, examination of hilar/intrapulmonary lymph nodes, and ≥3 mediastinal lymph node stations.^23,26

Performance Cut Points and Surgeon Performance Tiers

To segregate surgeons into performance tiers, we first excluded resections after neoadjuvant therapy (which can confound several of the metrics) and then excluded surgeons with 15 or fewer resections in the entire cohort (Appendix Fig A1). From the preimplementation resections, we derived cut points for each of the four outcomes metrics, defined using the 25th and 75th percentiles. We then applied these cut points to all eligible postimplementation phase resections by surgeons active in that phase.

We categorized surgeons into three sequential quality groups for each outcomes metric: 1 (lowest percentile), 2 (intermediate, 25th-75th percentile), and 3 (highest percentile). After ranking for each quality metric, we summed up each surgeon's quality tier for the four quality metrics to create an aggregate score. Surgeons were placed into performance tiers on the basis of their aggregate score: tier 1 (lowest quality) surgeons had an aggregate score of ≤6; tier 2 (intermediate quality), 7-9; and tier 3 (best quality), ≥10 of a maximum of 12.

Statistical Methods

We summarized the demographic, clinical, care delivery, and quality characteristics of the analytic MS-QSR cohort with appropriate statistics such as frequency, percentages, medians, and IQRs. The primary analysis cohort excluded resections by surgeons with 15 or fewer resections. We compared patient characteristics across the pre- and postimplementation resections using chi-squared tests (Fisher's exact test when sample sizes were small) and Wilcoxon-Mann-Whitney tests. We also compared characteristics across aggregate surgeon tiers using chi-squared (Fisher's exact if sample sizes were small) and Kruskal-Wallis tests.

To examine the influence of surgeon performance on patient survival, we used several Cox proportional hazards models. The first model examined the unadjusted association between surgeon tier and the hazard of death (crude hazard ratio [crude HR]); the second (HR¹) adjusted for clinical stage; the third (HR²) also adjusted for important patient characteristics (age, number of comorbidities, sex, race, insurance, histology); and the fourth model (HR³) included provider-determined care delivery processes (preoperative positron emission tomography–computed tomography [PET-CT] scan use, preoperative invasive staging, surgical technique [open, video-assisted, or robotically assisted thoracoscopic surgery], and kit use). Given the intracluster dependence within surgeons, all models used a robust sandwich covariance estimator within surgeons.

In sensitivity analyses, we included all surgeons regardless of resection volume (Appendix Fig A2), adjusted for surgeon case volume, and developed the aggregate surgeon tiers excluding both wedge resections and wedge resection as an intermediate outcome metric. We used SAS 9.4 (2013, SAS Institute Inc, Cary, NC) for all statistical analyses and set all significance levels at 0.05.

Data Availability

We will adhere to the international committee of medical journal editors (ICMJE) and the ethical obligation of responsible sharing of data. However, the data generated in this study are not publicly available due to compliance restrictions for protection of patient privacy, but are available upon reasonable request from the corresponding author.

RESULTS

Surgeon Characteristics

From 2009 to 2021, the MS-QSR cohort included 4,249 resections, performed by 60 surgeons, including five general surgeons, 45 board-certified cardiothoracic surgeons, and 10 dedicated general thoracic surgeons. We excluded 167 resections performed after neoadjuvant therapy, for a full primary resection cohort of 4,082. We then eliminated 21 surgeons who had 15 or fewer resections from the primary analysis cohort, reducing the overall number of resections to 3,959 (Fig 1). Excluded were three general surgeons, 15 cardiothoracic surgeons, and three general thoracic surgeons. Of the remaining 39 surgeons, 24 performed resections in the preimplementation phase and 31 performed resections during the postimplementation phase. The distribution of surgeon specialty between the pre- and postimplementation phases was roughly similar, with 3%-8% of general surgeons, 70%-77% of cardiothoracic surgeons, and 19%-21% of general thoracic surgeons.

FIG 1.

Flow diagram. ^aResections were aggregated according to surgeon and used to derive cut points for quality metrics (16 surgeons were in both phases). ^bResections were aggregated according to surgeon and compared to preimplementation cut points for each quality metric. ^cTier resections and surgeons are exclusive.

Patient-Level Characteristics

Among the full primary resection MS-QSR cohort, the median patient age was 68 years (IQR, 62-74), 54% were male, 78% were White, 21% were Black, 48% had Medicare, and 34% had commercial insurance. Patients mostly had early-stage NSCLC (84% clinical stage I or II; 84% pathologic stage I or II); 14% had no mediastinal lymph nodes examined, 6% were pNX, and 4% had positive resection margins; 46% of resections used the kit (Table 1). Some cohort characteristics differed between the pre- and postimplementation phases: surgery was robotically assisted in 74% versus 48% and video-assisted in 7% versus 40%, respectively; pNX rates were 12% versus 4%; rates of nonexamination of mediastinal lymph nodes were 23% versus 10%; 15% versus 6% were wedge resections; 5% versus 4% had positive margins. The quality of resection, adjudged by both IASLC and NCCN benchmarks, was better in the postimplementation cohort (Appendix Table A1, online only).

TABLE 1.

Patient-Level Demographic and Clinical Characteristics

Quality Cut Points

The 24 surgeons in the preimplementation phase performed a total of 1,130 resections. We aggregated quality metrics for each surgeon, using the 25th and 75th percentiles to define the cut points. The median (IQR) for rates of positive margins among surgeons during the preimplementation phase was 5% (2%-8%); for wedge resections, it was 14% (8-18); for pNX, it was 8% (5-14); and for nonexamination of mediastinal lymph nodes, it was 17% (14-32).

Surgeon Performance Tiers

Using the quality cut points developed from the preimplementation data set, we categorized surgeon performance within the postimplementation data set across the four metrics. For resection with positive margins, four surgeons (13% of the 31 postimplementation surgeons) were in quality group 1, 12 (39%) in group 2, and 15 (48%) in group 3. For resections without mediastinal lymph nodes, seven surgeons (23%) were in group 1, two (7%) in group 2, and 22 (71%) in group 3; for pNX, five surgeons (16%) were in group 1, five (16%) in group 2, and 21 (68%) in group 3. For wedge resections, five (16%) were in group 1, eight (26%) in group 2, and 18 (58%) in group 3. Aggregating the four metrics, tiers 1 (worst) and 2 (intermediate) each had five surgeons (16%) and tier 3 (best) had 21 (68%). Four of the five surgeons in tier 1 were cardiothoracic surgeons, and the other was a dedicated general thoracic surgeon; all five surgeons in tier 2 were cardiothoracic surgeons; and tier 3 consisted of one general surgeon, 15 cardiothoracic surgeons, and five dedicated general thoracic surgeons.

Patient-Level Characteristics Across Surgeon Performance Tiers

Demographic and clinical characteristics varied across the three aggregate tiers. In particular, tier 1 had higher rates of Black patients (38% v 21% v 21%, tier 1 v 2 v 3; P < .001) and Medicare-insured patients (58% v 50% v 47%; P < .001). Intermediate outcomes that differed between tiers included rates of the following: pNX, 21% versus 8% versus 2% (P < .001); nonexamination of mediastinal lymph nodes, 47% versus 34% versus 5% (P < .001); margin positivity, 9% versus 3% versus 4% (P = .0364); and wedge resections, 26% versus 8% versus 5% (Table 2).

TABLE 2.

Patient-Level Demographic and Clinical Characteristics and Surgeon Characteristics According to Surgeon Tier

The distribution of desirable processes also differed significantly between tiers: preoperative PET-CT scans were used in 80%, 91%, and 84% (P < .001); invasive staging was used in 7% versus 10% versus 18% (P < .001); the procedure was robotically assisted in 0% versus 29% versus 44% and video-assisted in 17% versus 6 versus 12%, with the rest being open thoracotomies (P < .001); and the lymph node kit was used in 27% versus 32% versus 71% (P < .001; Table 2). Furthermore, the quality of surgery as evaluated by aggregate benchmarks also sequentially improved with surgeon tier: resections were IASLC R0 in 6% versus 14% versus 52% (P < .001) and 16% versus 27% versus 71% (P < .001) met the NCCN criteria for surgical quality (Table 3).

TABLE 3.

Stage Details and Quality of Surgery According to Surgeon Tier

Characteristics of Wedge Resections Across Surgeon Tiers

The size distribution of tumors that underwent wedge resection was similar across surgeon tiers, with 68% versus 74% versus 73%, respectively, of tier 1 versus 2 versus 3 wedge-resected tumors ≤2 cm (P = .9019), there were no significant differences in clinical N-category, with 88% versus 92% versus 92% being clinical N0 (P = .5054), respectively. However, 68% versus 72% versus 20% of wedge resections had no lymph nodes examined (P < .001); 73% versus 84% versus 24% had no mediastinal lymph nodes, and only 7% versus 4% versus 40% had two or more mediastinal lymph node stations sampled (P < .001; Appendix Table A2, online only).

Surgeon Tiers and Aggregate Patient Survival

There were significant differences in patient survival across the surgeon tiers, with surgery by tier 3 surgeons associated with the best survival (log-rank P = .0066; Fig 2). With tier 3 as the reference, the crude HR for death among patients operated on by tier 1 and 2 surgeons was 1.37 (95% CI, 1.1 to 1.72) and 1.19 (95% CI, 1.00 to 1.43), respectively (Table 4). After adjusting for clinical stage (HR¹), it was 1.36 (95% CI, 1.08 to 1.71) and 1.22 (95% CI, 1.02 to 1.46), respectively. With further adjustment for patient demographics (HR²), surgery by tier 1 surgeons remained significantly more hazardous (HR², 1.31; 95% CI, 1.09 to 1.57), but there was no significant difference between tiers 2 and 3 (HR³, 1.10; 0.95 to 1.28). After adjustment for the preoperative PET-CT, invasive staging, technique of resection, and use of the kit (HR³), there were no significant differences between tiers 1 or 2 and tier 3 surgeons—1.18 (95% CI, 0.98 to 1.42) and 1.02 (95% CI, 0.81 to 1.27), respectively.

FIG 2.

Patient survival according to their surgeon's performance tier. Tier 1—low performers, tier 2—intermediate performers, and tier 3—high performers. Primary analysis excluding low-volume surgeons.

TABLE 4.

Comparative Death Hazards of Patients Operated Receiving Curative-Intent Lung Resection by Surgeons in Different Tiers

Sensitivity Analyses

In additional analyses including all 60 surgeons regardless of resection volume, surgery by tier 1 surgeons was persistently more hazardous than tier 3 and the difference between tier 2 and 3 surgeons, paradoxically, was primarily significant in all fully adjusted models (HR³) except for the quality metric no mediastinal examination (Table 5). Adjustment for surgeon resection volume (Appendix Table A3, online only) did not materially change the primary analysis findings. Excluding wedge resections from the intermediate outcomes used to create surgeon tiers weakened the model (Appendix Table A4, online only).

TABLE 5.

Comparative Death Hazards of Patients Operated Receiving Curative-Intent Lung Resection by Surgeons in Different Tiers

DISCUSSION

We used four specific, easily measured intermediate outcomes to categorize a population of surgeons into low-, intermediate-, and high-quality performance tiers and demonstrated survival differences in patients segregated by surgeon tier. The differences persisted after adjustment for clinical stage, and the difference between tier 1 and tier 3 surgeons persisted after additional adjustment for important patient-level characteristics, thereby associating surgeon-level quality with disparities in long-term patient survival. Surgeons with cumulative quality deviations had significantly worse overall patient survival.

Patients who undergo curative-intent surgery for lung cancer have survival challenges from the operation (avoiding postoperative mortality) and the cancer (avoiding long-term cancer-related mortality).^27,28 Postoperative mortality—at 30-90 days—has been closely examined,^29,30 and the multilevel factors driving disparities in this short-term outcome have been extensively characterized, including the concept of failure to rescue.^31-36 However, postoperative mortality rates after curative-intent lung cancer surgery are low, typically ranging from 0% to 5%,^37,38 whereas 5-year survival rates (profoundly influenced by the quality of oncologic resection) range from 30% to 60% between patients and institutions of different characteristics.^8,39-41 Although more deaths can be avoided by focusing quality initiatives on factors associated with long-term survival, long-term survival is harder to track, and disparities therein are relatively easier to ignore.³⁹

The oncologic quality of lung cancer resection can be distilled down to a single essence—the likelihood of complete resection. Traditionally measured by the absence of cancer at the surgical resection margin, the IASLC expanded the definition to include predictors of recurrence risk irrespective of margin status, introducing the concept of R-uncertain resections, which, despite negative margins, have residually high recurrence risk.^22,24,25,42 Suboptimal lymph node evaluation is the most common cause of R uncertainty.^24,42 Although the Lung Cancer Study Group established that lobectomy was associated with significantly lower local recurrence risk than sublobar resection,⁴³ two recent randomized controlled trials established that segmentectomy (JCOG 0802)⁴⁴ or wedge resection (CALBG/Alliance 140503)⁴ provides similar survival to lobectomy in patients with small (≤2 cm), peripheral stage IA NSCLC. Prerandomization eligibility required intraoperative confirmation of N0 disease after examination of three (JCOG0802) or two (CALGB/Alliance 140503) mediastinal lymph node stations.^4,44

We justify the inclusion of outlier behavior in the use of wedge resections as a quality metric because the traditional use of wedge resection in the United States does not meet the stringent eligibility criteria used in CALGB/Alliance 140503.^45-48 Indeed, in our cohort, 32%, 26%, and 26% of tier 1, 2, and 3 patients who had wedge resection had tumors >2 cm and node evaluation was poor, such that only 7%, 4%, and 40% in the respective tiers would have met criteria for inclusion in CALGB/Alliance 140503. Regarding the other metrics, resection with positive margin is clearly associated with significantly worse survival^17,49; although the optimal extent of lymphadenectomy may be debated, there is no controversy that nonexamination of mediastinal lymph nodes or, worse, nonexamination of lymph nodes is strongly associated with poor long-term survival.^19-21,42,50 Therefore, the four components of our surgeon-tiering system are robustly linked to survival.

Our analysis bears all the limitations of a retrospective review. The database has the advantages of granularity and inclusion of a population-based cohort, including almost all cases within a defined regional population. The relatively small number of surgeons (60 in the whole cohort and 39 in the primary analytic cohort) and the regional data set may raise questions about generalizability and restrict our ability to explore meaningful differences in surgeon characteristics. However, findings from this data set have been generalizable to the US and global experience.^42,51 Our results may be confounded by low surgeon volumes, and further studies with a larger number of surgeons and with a greater volume would be ideal. For this reason, we limited our primary analysis cohort to higher-volume surgeons and performed a sensitivity analysis including all surgeons in the cohort and adjusting for surgeon volume in sensitivity analysis, with similar results. Moreover, questions have been raised about the volume-outcome relationship for lung cancer surgery.^28,52

In summary, we used four readily retrievable, readily improved outcomes metrics to segregate surgeons into performance tiers associated with long-term survival disparity. Adjustment for potentially linked processes of care mostly eliminated the tier-associated disparities in patient hazard. Future work should further evaluate the use of such simple, measurable, and survival-impactful intermediate outcomes and more robustly test the survival impact of improved processes for lung cancer surgery.

APPENDIX 1. THE MID-SOUTH QUALITY OF SURGICAL RESECTION (MS-QSR) CONSORTIUM

Horace L. Wiggins, MD (St Bernard's Regional Medical Center, Jonesboro, AR), David Talton, MD (North Mississippi Medical Center, Tupelo, MS), Daniel R. Stevenson, MD (NEA Baptist Memorial Hospital, Jonesboro, AR), Albert. Koury, MD (Baptist Memorial Hospital—Mississippi Baptist Medical Center, Jackson, MS).

FIG A1.

Venn diagram illustrating the distribution of surgeons in the primary analysis cohort excluding low-volume surgeons who performed ≤15 resections in the cohort.

FIG A2.

Venn diagram illustrating the breakdown of all surgeons in the cohort.

TABLE A1.

Additional Patient Demographics and Characteristics Among the Full Cohort, Cohort Operated by Surgeons With More Than 15 Total Resections in the Database, and Patients in the Pre– and Post–Kit Implementation Cohorts

TABLE A2.

Characteristics of Wedge Resections Across the Three Tiers of Surgeons

TABLE A3.

Comparative Death Hazards of Patients Who Received Curative-Intent Lung Resection by Surgeons in Different Tiers (adjusting for surgeon case volume)

TABLE A4.

Comparative Death Hazards of Patients Who Received Curative-Intent Lung Resection by Surgeons in Different Tiers (excluding wedge resection in the model)

DATA SHARING STATEMENT

We will adhere to the international committee of medical journal editors (ICMJE) and the ethical obligation of responsible sharing of data. However, the data generated in this study are not publicly available due to compliance restrictions for protection of patient privacy, but are available upon reasonable request from the corresponding author.

AUTHOR CONTRIBUTIONS

Conception and design: Meredith A. Ray, Matthew P. Smeltzer, Nicholas R. Faris, Raymond U. Osarogiagbon

Financial support: Raymond U. Osarogiagbon

Administrative support: Carrie Fehnel, Nicholas R. Faris, Raymond U. Osarogiagbon

Provision of study materials or patients: Kourtney Dortch, Ganpat Valaulikar, Todd Robbins, Raymond U. Osarogiagbon

Collection and assembly of data: Olawale Akinbobola, Carrie Fehnel, Kourtney Dortch, Bradley Wolf, Ganpat Valaulikar, Hetal D. Patel, Todd Robbins, Nicholas R. Faris, Raymond U. Osarogiagbon

Data analysis and interpretation: Meredith A. Ray, Andrea Saulsberry, Bradley Wolf, Thomas Ng, Nicholas R. Faris, Raymond U. Osarogiagbon

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Surgeon Quality and Patient Survival After Resection for Non–Small-Cell Lung Cancer

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).