- Email: [email protected]
The Society of Thoracic Surgeons Composite Score Rating for Pulmonary Resection for Lung Cancer
Journal Pre-proof The Society of Thoracic Surgeons Composite Score Rating for Pulmonary Resection for Lung Cancer Stephen R. Broderick, MD, MPHS, Maria Grau-Sepulveda, MD, MPH, Andrzej S. Kosinski, PhD, Paul A. Kurlansky, MD, David M. Shahian, MD, Jeffrey P. Jacobs, MD, Susan Becker, MBA, RN-BC, Malcolm M. DeCamp, MD, Christopher W. Seder, MD, Eric L. Grogan, MD, MPH, Lisa M. Brown, MD, MAS, William Burfiend, MD, Mitchell Magee, MD, Daniel P. Raymond, MD, Varun Puri, MD, MSCI, Andrew C. Chang, MD, Benjamin D. Kozower, MD, MPH PII:
S0003-4975(19)31604-2
DOI:
https://doi.org/10.1016/j.athoracsur.2019.08.114
Reference:
ATS 33176
To appear in:
The Annals of Thoracic Surgery
Received Date: 19 August 2019 Accepted Date: 19 August 2019
Please cite this article as: Broderick SR, Grau-Sepulveda M, Kosinski AS, Kurlansky PA, Shahian DM, Jacobs JP, Becker S, DeCamp MM, Seder CW, Grogan EL, Brown LM, Burfiend W, Magee M, Raymond DP, Puri V, Chang AC, Kozower BD, The Society of Thoracic Surgeons Composite Score Rating for Pulmonary Resection for Lung Cancer, The Annals of Thoracic Surgery (2019), doi: https:// doi.org/10.1016/j.athoracsur.2019.08.114. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 by The Society of Thoracic Surgeons
The Society of Thoracic Surgeons Composite Score Rating for Pulmonary Resection for Lung Cancer Running Head: STS Pulmonary Resection Composite Score
Stephen R Broderick MD, MPHS1, Maria Grau-Sepulveda MD, MPH2, Andrzej S Kosinski PhD2, Paul A Kurlansky MD3, David M Shahian MD4, Jeffrey P Jacobs MD5, Susan Becker MBA, RNBC6, Malcolm M DeCamp MD7, Christopher W Seder MD8, Eric L Grogan MD, MPH9, Lisa M Brown MD, MAS10, William Burfiend, MD11, Mitchell Magee MD12, Daniel P Raymond MD13, Varun Puri MD, MSCI14, Andrew C Chang, MD15, Benjamin D Kozower MD, MPH14 1
Division of Thoracic Surgery, Johns Hopkins Medical Institutions, Baltimore, Maryland
2
Duke Clinical Research Institute, Durham, North Carolina
3
Department of Surgery, Columbia University, New York, New York
4
Department of Surgery, Harvard Medical School, Boston, Massachusetts
5
Division of Cardiac Surgery, Johns Hopkins Medical Institutions, Baltimore, Maryland
6
Society of Thoracic Surgeons National Database, Chicago, Illinois
7
Division of Cardiothoracic Surgery, University of Wisconsin School of Medicine and Public Health
8
Department of Cardiovascular and Thoracic Surgery, Rush University Medical Center, Chicago, Illinois 9
Department of Surgery, Vanderbilt University, Nashville, Tennessee
10
Section of General Thoracic Surgery, University of California, Davis Health, Sacramento, California 11
St Luke’s Health Network, Bethlehem, Pennsylvania
12
Medical City Dallas Hospital, Dallas, Texas
13
Cleveland Clinic, Cleveland, Ohio
14
Department of Surgery, Washington University School of Medicine, St. Louis, Missouri
15
Section of Thoracic Surgery, Michigan Medicine, Ann Arbor, Michigan
Meeting Presentation: Presented at the 55th annual meeting of the Society of Thoracic Surgeons, San Diego, CA; Jan 26-29, 2019. Word Count: 4965
1
Corresponding Author: Stephen R Broderick MD, MPHS 600 N Wolfe Street Blalock 240 Baltimore, MD 21287 Email: [email protected]
The STS Executive Committee approved this document.
2
Abstract Background: The Society of Thoracic Surgeons (STS) General Thoracic Surgery Database (GTSD) has developed composite quality measures for lobectomy and esophagectomy. We sought to develop a composite measure including all resections for lung cancer. Methods: The STS lung cancer composite score is based on two outcomes: risk-adjusted mortality and morbidity. General Thoracic Surgery Database (GTSD) data were included from 1/2015 – 12/2017. “Star ratings” were created for centers with ≥ 30 cases using 95% Bayesian credible intervals. The Bayesian model was performed with and without inclusion of minimally invasive approach to assess the impact of approach on the composite measure. Results: The study population included 38,461 patients from 256 centers. Overall operative mortality was 1.3% (495/38,461); Major complication rate was 7.9% (3,045/38,461). Median nodes examined was 10 (IQR 5-16); Median nodal stations sampled was 4 (IQR 3-5). Positive resection margins were identified in 3.7% (1,420/38,461). 214 centers with ≥ 30 cases were assigned star ratings. There were 7 one-star, 194 two-star and 13 three-star programs. 70.6% of resections were performed through a minimally invasive approach. Inclusion of minimally invasive approach, which was adjusted for in previous models, altered the star ratings for 3% (6/214) of programs. Conclusions: Participants in the STS GTSD perform lung cancer resection with low morbidity and mortality. Lymph node data suggest participants are meeting contemporary staging standards. There is wide variability among participants in application of minimally invasive approaches. Risk adjustment for approach alters ratings in 3% of participants.
3
The Society of Thoracic Surgeons (STS) Quality Measurement Task Force (QMTF) has developed composite performance measures for common cardiac operations including coronary artery bypass grafting (CABG), valve replacement, and combined CABG and valve replacement [1-4]. The QMTF has also developed composite measures for two general thoracic operations: lobectomy [5] and esophagectomy [6]. These composite scores and corresponding star ratings are included in participant feedback reports for the STS General Thoracic Surgery Database (GTSD) and are available for public reporting. The GTSD task force previously developed risk-adjustment models for morbidity and mortality after lung cancer resection [7,8]. This report describes the development of a quality composite measure with star rating for that model encompassing a more recent three-year period (20152017) and includes all types of resections for lung cancer (wedge, segmentectomy, lobectomy, sleeve lobectomy, bilobectomy, pneumonectomy). Several additional outcomes that may be indicative of quality lung cancer resections are also assessed and reported. We also sought to assess the impact of adjustment for surgical approach (thoracotomy or minimally invasive) on the quality composite.
Patients and Methods Study Cohort The STS GTSD was queried for all patients who underwent pulmonary resection for primary non-small cell lung cancer (NSCLC) between January 1, 2015 and December 31, 2017. All data were captured using data collection form version 2.3. This time frame was selected to provide an adequate sample size that was reflective of current practice among GTSD participants and is consistent with use of a three year time frame in the lobectomy and esophageal resection composite models [5,6]. We excluded patients with non-elective status,
4
American Society of Anesthesiologists class V or VI, Zubrod score 4 or 5, occult or stage 0 tumors, or missing data for age, gender, FEV1%, discharge status, or pathologic stage. The final cohort consisted of 38,461 resections from 256 centers. Outcome Definitions Postoperative events were classified using STS GTSD definitions [9]. Operative mortality is defined as death within 30 days of surgery or at any time during the index hospitalization. Major morbidity was defined as the occurrence of one or more of the following: return to the operating room, pneumonia, initial ventilator support > 48hrs, respiratory failure, tracheostomy, ARDS, pulmonary embolus, bronchopleural fistula or myocardial infarction. Covariate Selection Covariates selected for risk adjustment were consistent with those utilized in the lobectomy composite model [8]. These include the following variables: age, gender, body mass index, hypertension, steroid use, congestive heart failure, coronary artery disease, peripheral vascular disease, reoperation, cerebrovascular disease, diabetes mellitus, receipt of neoadjuvant therapy, renal dysfunction, former smoker status, current smoker status, FEV1, Zubrod score, ASA class, pathologic stage (AJCC 7th edition) and operative procedure. The model was performed with and without inclusion of minimally invasive approach to assess its impact on the composite quality measure. Estimation of Composite Scores and Star Ratings The composite score for pulmonary resection for lung cancer is a combination of two riskadjusted outcome measures: operative mortality or presence of major complications. Participant risk-adjusted rates of these two endpoints were estimated in a Bayesian hierarchical model, as previously described. [5,6].
5
The composite score was calculated as a weighted sum of (1 minus the risk-adjusted mortality rate) and (1 minus the risk-adjusted major complication rate). Mortality and major complication scores were weighted inversely by their respective standard deviations (SDs) across participants and were normalized by the sum of inverses of these SDs. This is the same methodology used for other STS composite measures [1-6]. The composite scores were estimated for each STS GTSD participant and reported with 95% Bayesian credible interval (CrI). To assign GTSD participants into performance categories, individual risk-adjusted composite scores were compared to the mean composite score of all participants. Participants were classified into three groups: (1) lower-than-expected (1 star) if the 95% CrI for the composite score fell below the mean GTSD composite score; (2) as-expected, not statistically distinguishable from the mean (2-star) if the 95%CrI overlapped with the mean score; and (3) higher-than-expected (3-star) if the 95% CrI fell above the mean score. As minimally invasive approach to pulmonary resection may contribute to improved outcomes [10], “star ratings” were calculated for each participant with and without the inclusion of minimally invasive approach as a covariate to assess its impact on star ratings. Composite scores and ratings are reported for the analysis without inclusion of approach. Reliability of the Score Estimation The reliability estimate with its 95% CrI was obtained within the Bayesian model framework. The reliability estimate quantifies how well a model can distinguish accurately the performance of participants and is driven by each participant’s operative case volume, the degree of variation in true performance among participants, and measurement error [11]. Additional Outcome and Process Measures
6
In addition to the composite score we sought to evaluate other measures which may be indicative of quality pulmonary resection for NSCLC including adequacy of lymph node evaluation, proportion of resections with positive pathologic margins and proportion of resections performed using minimally invasive techniques. These additional measures are not included in the composite score but will be reported in participant feedback reports.
Results The study sample resulted in 38,461 lung cancer resections from 256 centers. The characteristics of the cohort and operations are shown in table 1. Composite Score and Ratings Rates of mortality and individual major morbidities are shown in table 2. The overall operative mortality rate for lung cancer resection was 1.3% (n=495); on a participant basis median operative mortality was 1.2% (IQR 0 to 2.1%). The overall rate of major morbidity was 7.9% (n=3,055); on a participant basis median major morbidity rate was 7.9% (IQR 5.1 to 11.2%). The relative weights for the mortality and morbidity components of the composite score were based on the reciprocal of the standard deviation such that mortality is weighted approximately 5 times that of a major complication (operative mortality weight = 0.84; morbidity weight = 0.16). Table 3 illustrates the reliability for the composite ratings based on volume thresholds. When all participants are included in the composite model, the reliability is 0.46. Using the threshold of 30 pulmonary resections over a three-year period, the reliability of the composite measure increases to 0.53 (95% CrI 0.42.4-0.61), similar to the reliability of the lobectomy composite model [5]. Therefore, only programs performing more than 30 pulmonary resections for lung cancer during the study period were eligible for star ratings.
7
The number of programs classified as high- (3-star), average- (2-star) and low-performing (1star) is shown in table 4. Of 214 programs with over 30 pulmonary resections for lung cancer, 3.3% were classified as 1-star (7/214), 90.7% were classified as 2-star (194/214) and 6.1% were classified as 3-star (13/214) programs withdecreasing rates of morbidity and mortality according to star rating. Figure 1 illustrates participants by increasing composite score. Participants toward the right of the curve, whose credible intervals do not cross the mean score are high-performing (3-star) programs. Participants to the left of the curve, whose credible intervals do not cross the mean score are low-performing (1-star) programs. There was significant variation in the utilization of minimally invasive techniques among all programs ranging from 0% to 100% (figure 2). Among participants performing ≥ 30 cases, the median percentage of cases performed using minimally invasive techniques was 75.4% (IQR 59.0-84.7). Inclusion of minimally invasive approach as a covariate altered the star ratings for 3% (6/214) programs. Three (1.5%) programs had an improvement and 3 (1.5%) programs had a decline in star rating with adjustment for approach. Additional Outcome Measures Adequacy of lymph node evaluation is shown in figure 3. The median number of nodal stations evaluated was 4.0 (IQR 3-5); the median number of lymph nodes evaluated was 10 (IQR 5-16). 5.8% of resections had no nodal stations evaluated; 78% of these were wedge resections. Positive pathologic margins were seen in 3.7% of resections. Table 5 demonstrates extent of resection according to pathologic stage. Most resections were lobectomies (70.7%); 22.3% were sublobar (wedge or segmentectomy) resections. Most sublobar resections (87%) were performed for pathologic stage I tumors. Most resections (70.6%) were performed using minimally invasive techniques; 26.2% of minimally invasive resections were performed using robotic assistance. Among resections performed for pathologic stage I tumors, 78% were performed using minimally invasive techniques.
8
Comment A composite quality measure for lung cancer resection in the STS GTSD was constructed to include all common pulmonary resections for lung cancer. This measure provides participants a means of comparing program performance and quality across the spectrum of lung cancer resections. In addition, this analysis provides up to date benchmarks for the proportion of lung cancer resections performed using minimally invasive techniques as well as nodal evaluation and positive margin status. The STS pulmonary resection composite identifies 10% of programs as statistically above or below average GTSD participant performance (3.7% below average, 6.5% above average). The percentage of above or below average performers could be altered by using an alternative CrI for the composite score. The 95% CrI was selected as it provides a balance between sensitivity and specificity. This is also consistent with other GTSD composite models [5,6]. The STS will use this composite model to provide participants with risk-adjusted outcomes for pulmonary resection for lung cancer with respect to operative mortality, major morbidity and the composite morbidity/mortality outcome. The additional measures including lymph node evaluation, proportion of resections with positive margins and utilization of minimally invasive techniques will also be reported but are not included in the composite score. The lobectomy and pulmonary resection composite measures are composed of two short-term outcomes: perioperative mortality and major morbidity. Long-term outcomes such as 90-day, 1-year or 5-year mortality are important measures in the evaluation of lung cancer resection. There is substantial evidence to suggest that in-hospital or 30-day measures of operative mortality likely do not capture all surgery-related deaths [12-15]. While long-term survival has been included in the STS GTSD data collection form for several years, the amount
9
of missing data for this variable precludes its inclusion in this composite measure. Future composite measures may incorporate longer term outcomes. Considerable evidence has shown that the application of minimally invasive techniques for pulmonary resection reduces perioperative morbidity and mortality [10,16-18]. As shown in figure 2, there is wide variation in the proportion of cases performed using minimally invasive techniques among centers. Assuming that use of minimally invasive approaches improves perioperative outcomes, programs with a high proportion of thoracoscopic or robotic resections may potentially receive a lower composite score as a result of adjustment for approach in the model. To assess for this we performed the composite score Bayesian models both with and without inclusion of approach as a covariate. Inclusion of approach resulted in a change in the star rating for 3% (6/214) of programs. The purpose of STS GTSD composite measures is to assess the quality of care provided by programs based on risk-adjusted outcomes (operative mortality and major morbidity). While surgical approach may impact outcomes [10,16-18], this variable is modifiable by the surgeon whereas other variables such as age, tumor stage, FEV1 and comorbid conditions are not. Thus, the composite measure and star ratings reported in this manuscript are those calculated without adjustment for surgical approach. Similarly, the reports provided to participants will include the composite without adjustment for approach. Consideration will be given by the GTSD and QMTF task forces to remove approach from risk adjustment models in future iterations of composite measures. Participants in the STS GTSD continue to perform pulmonary resection with low rates of mortality and morbidity. A 2-star program in this quality composite has a median operative mortality of 1.4% and a major complication rate of 8.2%. GTSD outcomes have previously demonstrated superior unadjusted discharge mortality rates, median length of stay and pulmonary complication rates for lobectomy when compared to the National Inpatient Sample
10
(NIS) [19]. Operative mortality for lung cancer resection in the GTSD has decreased from 2.2% in the years 2002-2008 [7] to 1.3% in the current analysis. Similarly, major morbidity has decreased from 8.6% to 7.9%. Potential explanations for these improved outcomes include the increased application of minimally invasive techniques and sublobar resections. Fernandez et al demonstrated 61.6% minimally invasive and 19.8% sublobar resections in a GTSD cohort from 2012-2014 [8] compared to 70.6% minimally invasive and 22.3% sublobar resections in the current analysis. In addition, the improved results may result from better patient selection as alternative non-surgical ablative therapies such as stereotactic radiotherapy for high-risk patients may contribute to the reduced operative mortality and morbidity. An analysis of the US Surveillance, Epidemiology, and End Results (SEER) Database from 1998-2009 reported that no lymph nodes were sampled in 13% of all lung cancer resections and 51% of sublobar resections [20]. Analysis of the National Cancer Database (NCDB) from 2003-2011 identified no lymph node evaluation in 28.8% of sublobar resections for clinical stage IA NSCLC [21]. Median lymph node counts in the NCDB are higher for patients who underwent lobectomy compared to sublobar resection [22]. Among all resections in the current GTSD analysis, 5.4% of the overall population and 21.4% of sublobar resections had no lymph nodes evaluated. A more recent analysis of the NCDB from 2004-2013 reported improvement in the number of nodes evaluated in resections of early stage NSCLC over the study period, reaching a median of 8 nodes in patients undergoing segmentectomy or lobectomy 2013 [23]. The median number of nodal stations evaluated for all resections in our cohort was 4; with a median of 10 nodes evaluated. These results suggest that GTSD participants are meeting contemporary pathological lymph nodal staging standards [24,25] and suggest that GTSD participants perform more thorough nodal evaluation than that observed in the SEER or NCDB samples.
11
Positive pathologic margins have been shown to negatively influence overall lung cancer survival after resection regardless of stage and occur in 4.4 - 4.7% in analyses of the NCDB [26,27]. The rate of positive margins in this GTSD cohort was 3.7% again suggesting that participants in the GTSD again outperform national registries. The appropriate extent of resection for early stage lung cancer remains a matter of debate in thoracic oncology. The current analysis of the GTSD demonstrates that sublobar resection was performed in 22.3% (15.4% wedge, 6.9% segmentectomy) of lung cancer patients, the majority (86.6%) of which had pathologic stage I disease. This represents an increase from previous analyses of the GTSD [8].
Among patients with pathologic stage I
disease, the majority (68.6%) underwent lobectomy and 76.5% of lobectomies were performed via minimally invasive approach. Most (86.1%) sublobar resections were performed with minimally invasive approaches. There are several limitations to this and other lung cancer surgery risk adjustment models. First, participation in the GTSD is voluntary and represents less than 50% of lung cancer resections in the United States [19]. Results from GTSD participants may not be generalizable to surgeons not participating in the database. Second, due to a number of programs performing relatively few procedures, the reliability of the model considering all participating centers is relatively low, thus the star rating was limited to programs performing at least 30 pulmonary resections over the 3-year time frame. This results in a more reliable model (95% CrI 0.42-0.61), but results in some participating centers not receiving star ratings due to an inadequate number of cases. This reliability estimate, however, is similar to that of the lobectomy, esophagectomy and adult cardiac surgery models [1-6]. Third, due to missing data in a large percentage of patients, diffusion capacity for carbon monoxide (DLCO) was excluded as a covariate in the risk adjustment model.
12
In conclusion, the STS has developed a two-domain composite performance measure for pulmonary resection for lung cancer. This is an extension of the previous lobectomy composite measure to include all common lung cancer resections. The composite measure identifies 10% of participants as statistically high or low performing programs. The provision of this data will facilitate programmatic benchmarking and quality improvement initiatives.
13
References 1. Shahian DM, O’Brien SM, Filardo G, et al. The Society of Thoracic Surgeons 2008 cardiac surgery risk models: part 1—coronary artery bypass grafting surgery. Ann Thorac surg 2009;88(suppl):2-22. 2. Shahian DM, He X, Jacobs JP, et al. The Society of Thoracic Surgeons isolated aortic valve replacement (AVR) composite score: a report of the STS Quality Measurement Task Force. Ann Thorac Surg 2012;94:2166-71. 3. Shahian DM, He X, Jacobs JP, et al. The STS AVR plus CABG composite score: a report of the STS Quality Measurement Task Force. Ann Thorac Surg 2014;97:1604-9. 4. Badhwar V, Rankin JS, He X, et al. The Society of Thoracic Surgeons mitral repair/replacement composite score: A report of the Society of Thoracic Surgeons Quality Measurement Task Force. Ann Thorac Surg 2016;101:2265-71. 5. Kozower BD, O’Brien SM, Kosinski AS, et al. The Society of Thoracic Surgeons Composite score for rating program performance for lobectomy for lung cancer. Ann Thorac Surg 2016;101:1379-86. 6. Society of Thoracic Surgeons General Thoracic Surgery Database Task Force. The Society of Thoracic Surgeons composite score for evaluating esophagectomy for esophageal cancer. Ann Thorac Surg 2017;103:1661-1667. 7. Kozower BD, ShengS, O’Brien SM, et al. STS database risk models: predictors of mortality and major morbidity for lung cancer resection. Ann Thorac Surg 2010;90:87583. 8. Fernandez FG, Kosinski AS, Burfeind W, et al. The Society of Thoracic Surgeons lung cancer resection risk model: Higher quality data and superior outcomes. Ann Thorac Surg 2016;102:370-7.
14
9.
The Society of Thoracic Surgeons national database. Available at https://www.sts.org/registries-research-center/sts-national-database/sts-generalthoracic-surgery-database.
10. Paul S, Altorki NK, Sheng S, et al. Thoracoscopic lobectomy is associated with lower morbidity than open lobectomy: a propensity-matched analysis from the STS database. J Thorac Cardiovasc Surg 2010;139:366-78. 11. Adams J. The reliability of provider profiling: a tutorial. 2009. Available at http://www.rand.org/pubs/technical_reports/tr653.html. Accessed September 21, 2015. 12. Fernando HC, Landreneau RJ, Mandrekar SJ, et al. Thirty- and ninety-day outcomes after sublobar resection with and without brachytherapy for non-small cell lung cancer: results from a multicenter phase III study. J Thorac Cardiovasc Surg 2011;142:1143-51. 13. Brunelli A, Dinesh P, Woodcock-Shaw J, et al. Ninety-day mortality after video-assisted thoracoscopic lobectomy: incidence and risk factors. Ann Thorac Surg 2017;104:10201026. 14. Schneider L, Farrokhyar F, Schieman C, et al. The burden of death following discharge after lobectomy. Eur J Cardiothorac Surg 2015;48:65-70. 15. Bryant AS, Rudemiller K, Cerfolio RJ. The 30- vs 90-day operative mortality after pulmonary resection. Ann Thorac Surg 2010;89:1717-23. 16. Stephens N, Rice D, Correa A, et al. Thoracoscopic lobectomy is associated with improved short-term and equivalent oncological outcomes compared with open lobectomy for clinical stage I non-small-cell lung cancer: a propensity matched analysis of 963 cases. Eur J Cardiothorac Surg 2014;46:607-13. 17. Donahoe LL, de Valence M, Atenafu EG, et al. High risk for thoracotomy but not thoracoscopic lobectomy. Ann Thorac Surg 2017;103:1730-1735.
15
18. Swanson SJ, Meyers BF, Gunnarsson CL, et al. Video-assisted thoracoscopic lobectomy is less costly and morbid than open lobectomy: a retrospective multiinstitutional database analysis. Ann Thorac Surg 2012;93:1027-32. 19. Lapar DJ, Bhamidipati CM, Lau CL, et al. The Society of Thoracic Surgeons General Thoracic Surgery Database: establishing generalizability to national lung cancer resection outcomes. Ann Thorac Surg 2012;94:216-21. 20. Osarogiagbon R, Yu Xinhua. Nonexamination of lymph nodes and survival after resection of non-small cell lung cancer. Ann Thorac Surg 2013;96:1178-1189. 21. Speicher PJ, Gu L, Gulack BC, et al. Sublobar resection for clinical stage IA Non-smallcell lung cancer in the United States. Clin Lung Cancer 2016;17:47-55. 22. Subramanian M, McMurry T, Meyers B, et al. Long-term results for clinical stage IA lung cancer: comparing lobectomy and sublobar resection. Ann Thorac Surg 2018;106:375381. 23. Krantz SB, Lutfi W, Kuchta K, et al. Improved lymph node staging in early-stage lung cancer in the National Cancer Database. Ann Thorac Surg 2017;104:1805-1814. 24. National Comprehensive Cancer Network clinical practice guidelines in oncology. Nonsmall cell lung cancer. Available at http://www.nccn.org/professionals/physician. Accessed 1/2/2019. 25. American College of Surgeons CoC Quality of Care Measures. Lung measure specifications. Available at https://www.facs.org/~/media/files/quality%20programs/cancer/mcdb/measure%20specs %20nscl.ashx. Accessed 1/2/2019. 26. Osarogiagbon RU, Lin CC, Smeltzer MP, et al. Prevalence, prognostic implications and survival modulators of incompletely resected non-small cell lung cancer in the U.S. National Cancer Data Base. J Thorac Oncol 2016;11:e5-16.
16
27. Lin CC, Smeltzer MP, Jemal A, et al. Risk-adjusted margin positivity rate as a surgical quality metric for non-small cell lung cancer. Ann Thorac Surg 2017;104:1161-1170.
17
Table 1. Characteristics of Cohort
Characteristics
Values (n = 38,461)
Age
67 +/- 9.5
Male
21,420 (44.3%)
Caucasian
33,066 (86%)
BMI
28 +/- 6
Hypertension
24,156 (62.8%)
Steroid use
1,108 (2.9%)
Congestive heart failure
1,009 (2.6%)
Coronary artery disease
8,005 (20.8%)
Peripheral vascular disease
3,627 (9.4%)
Reoperation
2,191 (5.7%)
Cerebrovascular disease
2,946 (7.7%)
Diabetes
7,647 (19.9%)
FEV1%
83 +/- 20
Neoadjuvant chemotherapy/radiation
2,500 (6.5%)
Renal Failure
627 (1.6%)
Former smoker
23,531 (61.2%)
Current smoker
9,100 (23.7%)
Zubrod score 0
17,478 (45.4%)
1
19,331 (50.3%)
2
1,411 (3.67%)
3
241 (0.6%)
ASA class I
73 (0.2%)
II
5,595 (14.6%)
18
III
29,676 (77.2%)
IV
3,117 (8.1%)
Extent of Resection 5,917 (15.4%) Wedge 2,659 (6.9%) Segmentectomy 27,201 (70.7%) Lobectomy 422 (1.1%) Sleeve Resection 1,152 (3.0%) Bilobectomy 1,110 (2.9%) Pneumonectomy Pathologic stage I
26,128 (67.9%)
II
7,441 (19.4%)
III
4,266 (11.1%)
IV
626 (1.6%)
19
Table 2: Overall Frequency and Participant Variation in Mortality/Morbidity
Main Outcomes Operative Mortality* Major Morbidity Composite**
Number
Overall Percent
495
1.29%
Participant Median (IQR) 1.2% (0-2.1)
3,045
7.92%
7.9% (5.1-11.2)
1,123
2.92%
2.7% (1.1-4.7)
1,379
3.59%
3.1% (1.3-5.3)
124
0.32%
1,025
2.67%
244
0.63%
0%
(0-1.0)
208
0.54%
0%
(0-0.8)
198
0.51%
0%
(0-0.8)
139
0.36%
0%
(0-0.5)
127
0.33%
0%
(0-0.5)
Components of Major Morbidity Return to OR Pneumonia Initial Vent Support > 48hrs Respiratory Failure Tracheostomy ARDS Pulmonary Embolus Bronchopleural Fistula Myocardial Infarction
0%
(0-0.4)
2.3% (0.6-3.9)
*Operative Mortality: in-hospital deaths and deaths within 30 days of surgery **Major Morbidity Composite: at least one component morbidity present
20
Table 3. Reliability for Pulmonary Resection Composite Ratings Based on Volume Thresholds
No. of participants Reliability 95% CrI
No Minimum
≥30 cases
≥50 cases
≥100 cases
≥150 cases
256
214
192
136
104
46.1% (36.6%54.9%)
52.5% (42.4%61.3%)
54.9% (44.7%63.6%)
61.1% (51.0%70.1%)
63.9% (53.0%73.5%)
21
Table 4. Participant Classification by Composite Rating and Construct Validity (Morbidity and Mortality Vary Across Star Ratings)
One star
Two stars
Three stars
Number of Programs
7
194
13
Operative mortality
1.5% (1.1%, 2.0%)
1.4% (1.2%, 1.5%)
0.3% (0.14%, 0.50%)
1.3% (1.2%, 1.4%)
10.9% (9.7%,12.2%)
8.2% (7.9%, 8.5%)
3.6% (3.1%, 4.3%)
7.9% (7.6%, 8.2%)
(95% CI) Major complication (95% CI)
All participants 214
22
Table 5. Extent of Resection and Surgical Approach by Pathologic Stage
Stage I Wedge MIS Thoracotomy Segmentectomy MIS Thoracotomy Lobectomy MIS Thoracotomy Sleeve Resection MIS Thoracotomy Bilobectomy MIS Thoracotomy Pneumonectomy MIS Thoracotomy Total MIS
Stage II
Stage III
Stage IV
342 (4.6%)
315 (7.4%)
136 (21.7%)
4,582 (89.4%)
272 (79.5%)
250 (79.4%)
123 (90.4%)
542 (10.6%)
70 (20.5%)
65 (20.6%)
13 (9.6%)
2,320 (8.9%)
207 (2.8%)
102 (2.4%)
30 (4.8%)
1,916 (82.6%)
146 (70.5%)
74 (72.5%)
22 (73.3%)
404 (17.4%)
61 (29.5%)
28 (27.5%)
8 (26.7%)
17,927 (68.8%)
5,843 (78.5%)
3,028 (70.9%)
403 (64.4%)
13,714 (76.5%)
3,584 (61.3%)
1,686 (55.7%)
258 (64.0%)
4,213 (23.5%)
2,259 (38.7%)
1,342 (44.3%)
145 (36.0%)
173 (0.7%)
160 (2.2%)
87 (20.4%)
2 (0.3%)
0
0
0
0
173 (100%)
160 (100%)
87 (100%)
2 (100%)
413 (1.6%)
430 (5.8%)
286 (6.7%)
23 (3.7%)
175 (42.4%)
140 (32.6%)
74 (25.9%)
7 (30.4%)
238 (57.6%)
290 (67.4%)
212 (74.1%)
16 (69.6%)
171 ( 0.6%)
459 (6.2%)
448 (10.5%)
32 (5.1%)
29 (17.0%)
51 (11.1%)
43 (9.6%)
2 (6.3%)
142 (83.0%)
408 (88.9%)
405 (90.4)
30 (93.8%)
26,128
7,441
4,266
626
20,416 (78.2%)
4,193 (56.3%)
2,127 (49.9%)
412 (65.8%)
5,712 ( 21.8%)
3,248 (43.6%)
2,139 (50.1%)
214 (34.2%)
5,124 (19.6%)
Thoracotomy
23
Figure Legends Figure 1. Participants Sorted According to Ascending Composite Score. Figure 2. Variation in Proportion of Minimally Invasive Procedures Among Participants with >15 Resections. Figure 3. Variation in Participant Median and Mean Number of Lymph Nodes Evaluated.
24
S0003-4975(19)31604-2
DOI:
https://doi.org/10.1016/j.athoracsur.2019.08.114
Reference:
ATS 33176
To appear in:
The Annals of Thoracic Surgery
Received Date: 19 August 2019 Accepted Date: 19 August 2019
Please cite this article as: Broderick SR, Grau-Sepulveda M, Kosinski AS, Kurlansky PA, Shahian DM, Jacobs JP, Becker S, DeCamp MM, Seder CW, Grogan EL, Brown LM, Burfiend W, Magee M, Raymond DP, Puri V, Chang AC, Kozower BD, The Society of Thoracic Surgeons Composite Score Rating for Pulmonary Resection for Lung Cancer, The Annals of Thoracic Surgery (2019), doi: https:// doi.org/10.1016/j.athoracsur.2019.08.114. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 by The Society of Thoracic Surgeons
The Society of Thoracic Surgeons Composite Score Rating for Pulmonary Resection for Lung Cancer Running Head: STS Pulmonary Resection Composite Score
Stephen R Broderick MD, MPHS1, Maria Grau-Sepulveda MD, MPH2, Andrzej S Kosinski PhD2, Paul A Kurlansky MD3, David M Shahian MD4, Jeffrey P Jacobs MD5, Susan Becker MBA, RNBC6, Malcolm M DeCamp MD7, Christopher W Seder MD8, Eric L Grogan MD, MPH9, Lisa M Brown MD, MAS10, William Burfiend, MD11, Mitchell Magee MD12, Daniel P Raymond MD13, Varun Puri MD, MSCI14, Andrew C Chang, MD15, Benjamin D Kozower MD, MPH14 1
Division of Thoracic Surgery, Johns Hopkins Medical Institutions, Baltimore, Maryland
2
Duke Clinical Research Institute, Durham, North Carolina
3
Department of Surgery, Columbia University, New York, New York
4
Department of Surgery, Harvard Medical School, Boston, Massachusetts
5
Division of Cardiac Surgery, Johns Hopkins Medical Institutions, Baltimore, Maryland
6
Society of Thoracic Surgeons National Database, Chicago, Illinois
7
Division of Cardiothoracic Surgery, University of Wisconsin School of Medicine and Public Health
8
Department of Cardiovascular and Thoracic Surgery, Rush University Medical Center, Chicago, Illinois 9
Department of Surgery, Vanderbilt University, Nashville, Tennessee
10
Section of General Thoracic Surgery, University of California, Davis Health, Sacramento, California 11
St Luke’s Health Network, Bethlehem, Pennsylvania
12
Medical City Dallas Hospital, Dallas, Texas
13
Cleveland Clinic, Cleveland, Ohio
14
Department of Surgery, Washington University School of Medicine, St. Louis, Missouri
15
Section of Thoracic Surgery, Michigan Medicine, Ann Arbor, Michigan
Meeting Presentation: Presented at the 55th annual meeting of the Society of Thoracic Surgeons, San Diego, CA; Jan 26-29, 2019. Word Count: 4965
1
Corresponding Author: Stephen R Broderick MD, MPHS 600 N Wolfe Street Blalock 240 Baltimore, MD 21287 Email: [email protected]
The STS Executive Committee approved this document.
2
Abstract Background: The Society of Thoracic Surgeons (STS) General Thoracic Surgery Database (GTSD) has developed composite quality measures for lobectomy and esophagectomy. We sought to develop a composite measure including all resections for lung cancer. Methods: The STS lung cancer composite score is based on two outcomes: risk-adjusted mortality and morbidity. General Thoracic Surgery Database (GTSD) data were included from 1/2015 – 12/2017. “Star ratings” were created for centers with ≥ 30 cases using 95% Bayesian credible intervals. The Bayesian model was performed with and without inclusion of minimally invasive approach to assess the impact of approach on the composite measure. Results: The study population included 38,461 patients from 256 centers. Overall operative mortality was 1.3% (495/38,461); Major complication rate was 7.9% (3,045/38,461). Median nodes examined was 10 (IQR 5-16); Median nodal stations sampled was 4 (IQR 3-5). Positive resection margins were identified in 3.7% (1,420/38,461). 214 centers with ≥ 30 cases were assigned star ratings. There were 7 one-star, 194 two-star and 13 three-star programs. 70.6% of resections were performed through a minimally invasive approach. Inclusion of minimally invasive approach, which was adjusted for in previous models, altered the star ratings for 3% (6/214) of programs. Conclusions: Participants in the STS GTSD perform lung cancer resection with low morbidity and mortality. Lymph node data suggest participants are meeting contemporary staging standards. There is wide variability among participants in application of minimally invasive approaches. Risk adjustment for approach alters ratings in 3% of participants.
3
The Society of Thoracic Surgeons (STS) Quality Measurement Task Force (QMTF) has developed composite performance measures for common cardiac operations including coronary artery bypass grafting (CABG), valve replacement, and combined CABG and valve replacement [1-4]. The QMTF has also developed composite measures for two general thoracic operations: lobectomy [5] and esophagectomy [6]. These composite scores and corresponding star ratings are included in participant feedback reports for the STS General Thoracic Surgery Database (GTSD) and are available for public reporting. The GTSD task force previously developed risk-adjustment models for morbidity and mortality after lung cancer resection [7,8]. This report describes the development of a quality composite measure with star rating for that model encompassing a more recent three-year period (20152017) and includes all types of resections for lung cancer (wedge, segmentectomy, lobectomy, sleeve lobectomy, bilobectomy, pneumonectomy). Several additional outcomes that may be indicative of quality lung cancer resections are also assessed and reported. We also sought to assess the impact of adjustment for surgical approach (thoracotomy or minimally invasive) on the quality composite.
Patients and Methods Study Cohort The STS GTSD was queried for all patients who underwent pulmonary resection for primary non-small cell lung cancer (NSCLC) between January 1, 2015 and December 31, 2017. All data were captured using data collection form version 2.3. This time frame was selected to provide an adequate sample size that was reflective of current practice among GTSD participants and is consistent with use of a three year time frame in the lobectomy and esophageal resection composite models [5,6]. We excluded patients with non-elective status,
4
American Society of Anesthesiologists class V or VI, Zubrod score 4 or 5, occult or stage 0 tumors, or missing data for age, gender, FEV1%, discharge status, or pathologic stage. The final cohort consisted of 38,461 resections from 256 centers. Outcome Definitions Postoperative events were classified using STS GTSD definitions [9]. Operative mortality is defined as death within 30 days of surgery or at any time during the index hospitalization. Major morbidity was defined as the occurrence of one or more of the following: return to the operating room, pneumonia, initial ventilator support > 48hrs, respiratory failure, tracheostomy, ARDS, pulmonary embolus, bronchopleural fistula or myocardial infarction. Covariate Selection Covariates selected for risk adjustment were consistent with those utilized in the lobectomy composite model [8]. These include the following variables: age, gender, body mass index, hypertension, steroid use, congestive heart failure, coronary artery disease, peripheral vascular disease, reoperation, cerebrovascular disease, diabetes mellitus, receipt of neoadjuvant therapy, renal dysfunction, former smoker status, current smoker status, FEV1, Zubrod score, ASA class, pathologic stage (AJCC 7th edition) and operative procedure. The model was performed with and without inclusion of minimally invasive approach to assess its impact on the composite quality measure. Estimation of Composite Scores and Star Ratings The composite score for pulmonary resection for lung cancer is a combination of two riskadjusted outcome measures: operative mortality or presence of major complications. Participant risk-adjusted rates of these two endpoints were estimated in a Bayesian hierarchical model, as previously described. [5,6].
5
The composite score was calculated as a weighted sum of (1 minus the risk-adjusted mortality rate) and (1 minus the risk-adjusted major complication rate). Mortality and major complication scores were weighted inversely by their respective standard deviations (SDs) across participants and were normalized by the sum of inverses of these SDs. This is the same methodology used for other STS composite measures [1-6]. The composite scores were estimated for each STS GTSD participant and reported with 95% Bayesian credible interval (CrI). To assign GTSD participants into performance categories, individual risk-adjusted composite scores were compared to the mean composite score of all participants. Participants were classified into three groups: (1) lower-than-expected (1 star) if the 95% CrI for the composite score fell below the mean GTSD composite score; (2) as-expected, not statistically distinguishable from the mean (2-star) if the 95%CrI overlapped with the mean score; and (3) higher-than-expected (3-star) if the 95% CrI fell above the mean score. As minimally invasive approach to pulmonary resection may contribute to improved outcomes [10], “star ratings” were calculated for each participant with and without the inclusion of minimally invasive approach as a covariate to assess its impact on star ratings. Composite scores and ratings are reported for the analysis without inclusion of approach. Reliability of the Score Estimation The reliability estimate with its 95% CrI was obtained within the Bayesian model framework. The reliability estimate quantifies how well a model can distinguish accurately the performance of participants and is driven by each participant’s operative case volume, the degree of variation in true performance among participants, and measurement error [11]. Additional Outcome and Process Measures
6
In addition to the composite score we sought to evaluate other measures which may be indicative of quality pulmonary resection for NSCLC including adequacy of lymph node evaluation, proportion of resections with positive pathologic margins and proportion of resections performed using minimally invasive techniques. These additional measures are not included in the composite score but will be reported in participant feedback reports.
Results The study sample resulted in 38,461 lung cancer resections from 256 centers. The characteristics of the cohort and operations are shown in table 1. Composite Score and Ratings Rates of mortality and individual major morbidities are shown in table 2. The overall operative mortality rate for lung cancer resection was 1.3% (n=495); on a participant basis median operative mortality was 1.2% (IQR 0 to 2.1%). The overall rate of major morbidity was 7.9% (n=3,055); on a participant basis median major morbidity rate was 7.9% (IQR 5.1 to 11.2%). The relative weights for the mortality and morbidity components of the composite score were based on the reciprocal of the standard deviation such that mortality is weighted approximately 5 times that of a major complication (operative mortality weight = 0.84; morbidity weight = 0.16). Table 3 illustrates the reliability for the composite ratings based on volume thresholds. When all participants are included in the composite model, the reliability is 0.46. Using the threshold of 30 pulmonary resections over a three-year period, the reliability of the composite measure increases to 0.53 (95% CrI 0.42.4-0.61), similar to the reliability of the lobectomy composite model [5]. Therefore, only programs performing more than 30 pulmonary resections for lung cancer during the study period were eligible for star ratings.
7
The number of programs classified as high- (3-star), average- (2-star) and low-performing (1star) is shown in table 4. Of 214 programs with over 30 pulmonary resections for lung cancer, 3.3% were classified as 1-star (7/214), 90.7% were classified as 2-star (194/214) and 6.1% were classified as 3-star (13/214) programs withdecreasing rates of morbidity and mortality according to star rating. Figure 1 illustrates participants by increasing composite score. Participants toward the right of the curve, whose credible intervals do not cross the mean score are high-performing (3-star) programs. Participants to the left of the curve, whose credible intervals do not cross the mean score are low-performing (1-star) programs. There was significant variation in the utilization of minimally invasive techniques among all programs ranging from 0% to 100% (figure 2). Among participants performing ≥ 30 cases, the median percentage of cases performed using minimally invasive techniques was 75.4% (IQR 59.0-84.7). Inclusion of minimally invasive approach as a covariate altered the star ratings for 3% (6/214) programs. Three (1.5%) programs had an improvement and 3 (1.5%) programs had a decline in star rating with adjustment for approach. Additional Outcome Measures Adequacy of lymph node evaluation is shown in figure 3. The median number of nodal stations evaluated was 4.0 (IQR 3-5); the median number of lymph nodes evaluated was 10 (IQR 5-16). 5.8% of resections had no nodal stations evaluated; 78% of these were wedge resections. Positive pathologic margins were seen in 3.7% of resections. Table 5 demonstrates extent of resection according to pathologic stage. Most resections were lobectomies (70.7%); 22.3% were sublobar (wedge or segmentectomy) resections. Most sublobar resections (87%) were performed for pathologic stage I tumors. Most resections (70.6%) were performed using minimally invasive techniques; 26.2% of minimally invasive resections were performed using robotic assistance. Among resections performed for pathologic stage I tumors, 78% were performed using minimally invasive techniques.
8
Comment A composite quality measure for lung cancer resection in the STS GTSD was constructed to include all common pulmonary resections for lung cancer. This measure provides participants a means of comparing program performance and quality across the spectrum of lung cancer resections. In addition, this analysis provides up to date benchmarks for the proportion of lung cancer resections performed using minimally invasive techniques as well as nodal evaluation and positive margin status. The STS pulmonary resection composite identifies 10% of programs as statistically above or below average GTSD participant performance (3.7% below average, 6.5% above average). The percentage of above or below average performers could be altered by using an alternative CrI for the composite score. The 95% CrI was selected as it provides a balance between sensitivity and specificity. This is also consistent with other GTSD composite models [5,6]. The STS will use this composite model to provide participants with risk-adjusted outcomes for pulmonary resection for lung cancer with respect to operative mortality, major morbidity and the composite morbidity/mortality outcome. The additional measures including lymph node evaluation, proportion of resections with positive margins and utilization of minimally invasive techniques will also be reported but are not included in the composite score. The lobectomy and pulmonary resection composite measures are composed of two short-term outcomes: perioperative mortality and major morbidity. Long-term outcomes such as 90-day, 1-year or 5-year mortality are important measures in the evaluation of lung cancer resection. There is substantial evidence to suggest that in-hospital or 30-day measures of operative mortality likely do not capture all surgery-related deaths [12-15]. While long-term survival has been included in the STS GTSD data collection form for several years, the amount
9
of missing data for this variable precludes its inclusion in this composite measure. Future composite measures may incorporate longer term outcomes. Considerable evidence has shown that the application of minimally invasive techniques for pulmonary resection reduces perioperative morbidity and mortality [10,16-18]. As shown in figure 2, there is wide variation in the proportion of cases performed using minimally invasive techniques among centers. Assuming that use of minimally invasive approaches improves perioperative outcomes, programs with a high proportion of thoracoscopic or robotic resections may potentially receive a lower composite score as a result of adjustment for approach in the model. To assess for this we performed the composite score Bayesian models both with and without inclusion of approach as a covariate. Inclusion of approach resulted in a change in the star rating for 3% (6/214) of programs. The purpose of STS GTSD composite measures is to assess the quality of care provided by programs based on risk-adjusted outcomes (operative mortality and major morbidity). While surgical approach may impact outcomes [10,16-18], this variable is modifiable by the surgeon whereas other variables such as age, tumor stage, FEV1 and comorbid conditions are not. Thus, the composite measure and star ratings reported in this manuscript are those calculated without adjustment for surgical approach. Similarly, the reports provided to participants will include the composite without adjustment for approach. Consideration will be given by the GTSD and QMTF task forces to remove approach from risk adjustment models in future iterations of composite measures. Participants in the STS GTSD continue to perform pulmonary resection with low rates of mortality and morbidity. A 2-star program in this quality composite has a median operative mortality of 1.4% and a major complication rate of 8.2%. GTSD outcomes have previously demonstrated superior unadjusted discharge mortality rates, median length of stay and pulmonary complication rates for lobectomy when compared to the National Inpatient Sample
10
(NIS) [19]. Operative mortality for lung cancer resection in the GTSD has decreased from 2.2% in the years 2002-2008 [7] to 1.3% in the current analysis. Similarly, major morbidity has decreased from 8.6% to 7.9%. Potential explanations for these improved outcomes include the increased application of minimally invasive techniques and sublobar resections. Fernandez et al demonstrated 61.6% minimally invasive and 19.8% sublobar resections in a GTSD cohort from 2012-2014 [8] compared to 70.6% minimally invasive and 22.3% sublobar resections in the current analysis. In addition, the improved results may result from better patient selection as alternative non-surgical ablative therapies such as stereotactic radiotherapy for high-risk patients may contribute to the reduced operative mortality and morbidity. An analysis of the US Surveillance, Epidemiology, and End Results (SEER) Database from 1998-2009 reported that no lymph nodes were sampled in 13% of all lung cancer resections and 51% of sublobar resections [20]. Analysis of the National Cancer Database (NCDB) from 2003-2011 identified no lymph node evaluation in 28.8% of sublobar resections for clinical stage IA NSCLC [21]. Median lymph node counts in the NCDB are higher for patients who underwent lobectomy compared to sublobar resection [22]. Among all resections in the current GTSD analysis, 5.4% of the overall population and 21.4% of sublobar resections had no lymph nodes evaluated. A more recent analysis of the NCDB from 2004-2013 reported improvement in the number of nodes evaluated in resections of early stage NSCLC over the study period, reaching a median of 8 nodes in patients undergoing segmentectomy or lobectomy 2013 [23]. The median number of nodal stations evaluated for all resections in our cohort was 4; with a median of 10 nodes evaluated. These results suggest that GTSD participants are meeting contemporary pathological lymph nodal staging standards [24,25] and suggest that GTSD participants perform more thorough nodal evaluation than that observed in the SEER or NCDB samples.
11
Positive pathologic margins have been shown to negatively influence overall lung cancer survival after resection regardless of stage and occur in 4.4 - 4.7% in analyses of the NCDB [26,27]. The rate of positive margins in this GTSD cohort was 3.7% again suggesting that participants in the GTSD again outperform national registries. The appropriate extent of resection for early stage lung cancer remains a matter of debate in thoracic oncology. The current analysis of the GTSD demonstrates that sublobar resection was performed in 22.3% (15.4% wedge, 6.9% segmentectomy) of lung cancer patients, the majority (86.6%) of which had pathologic stage I disease. This represents an increase from previous analyses of the GTSD [8].
Among patients with pathologic stage I
disease, the majority (68.6%) underwent lobectomy and 76.5% of lobectomies were performed via minimally invasive approach. Most (86.1%) sublobar resections were performed with minimally invasive approaches. There are several limitations to this and other lung cancer surgery risk adjustment models. First, participation in the GTSD is voluntary and represents less than 50% of lung cancer resections in the United States [19]. Results from GTSD participants may not be generalizable to surgeons not participating in the database. Second, due to a number of programs performing relatively few procedures, the reliability of the model considering all participating centers is relatively low, thus the star rating was limited to programs performing at least 30 pulmonary resections over the 3-year time frame. This results in a more reliable model (95% CrI 0.42-0.61), but results in some participating centers not receiving star ratings due to an inadequate number of cases. This reliability estimate, however, is similar to that of the lobectomy, esophagectomy and adult cardiac surgery models [1-6]. Third, due to missing data in a large percentage of patients, diffusion capacity for carbon monoxide (DLCO) was excluded as a covariate in the risk adjustment model.
12
In conclusion, the STS has developed a two-domain composite performance measure for pulmonary resection for lung cancer. This is an extension of the previous lobectomy composite measure to include all common lung cancer resections. The composite measure identifies 10% of participants as statistically high or low performing programs. The provision of this data will facilitate programmatic benchmarking and quality improvement initiatives.
13
References 1. Shahian DM, O’Brien SM, Filardo G, et al. The Society of Thoracic Surgeons 2008 cardiac surgery risk models: part 1—coronary artery bypass grafting surgery. Ann Thorac surg 2009;88(suppl):2-22. 2. Shahian DM, He X, Jacobs JP, et al. The Society of Thoracic Surgeons isolated aortic valve replacement (AVR) composite score: a report of the STS Quality Measurement Task Force. Ann Thorac Surg 2012;94:2166-71. 3. Shahian DM, He X, Jacobs JP, et al. The STS AVR plus CABG composite score: a report of the STS Quality Measurement Task Force. Ann Thorac Surg 2014;97:1604-9. 4. Badhwar V, Rankin JS, He X, et al. The Society of Thoracic Surgeons mitral repair/replacement composite score: A report of the Society of Thoracic Surgeons Quality Measurement Task Force. Ann Thorac Surg 2016;101:2265-71. 5. Kozower BD, O’Brien SM, Kosinski AS, et al. The Society of Thoracic Surgeons Composite score for rating program performance for lobectomy for lung cancer. Ann Thorac Surg 2016;101:1379-86. 6. Society of Thoracic Surgeons General Thoracic Surgery Database Task Force. The Society of Thoracic Surgeons composite score for evaluating esophagectomy for esophageal cancer. Ann Thorac Surg 2017;103:1661-1667. 7. Kozower BD, ShengS, O’Brien SM, et al. STS database risk models: predictors of mortality and major morbidity for lung cancer resection. Ann Thorac Surg 2010;90:87583. 8. Fernandez FG, Kosinski AS, Burfeind W, et al. The Society of Thoracic Surgeons lung cancer resection risk model: Higher quality data and superior outcomes. Ann Thorac Surg 2016;102:370-7.
14
9.
The Society of Thoracic Surgeons national database. Available at https://www.sts.org/registries-research-center/sts-national-database/sts-generalthoracic-surgery-database.
10. Paul S, Altorki NK, Sheng S, et al. Thoracoscopic lobectomy is associated with lower morbidity than open lobectomy: a propensity-matched analysis from the STS database. J Thorac Cardiovasc Surg 2010;139:366-78. 11. Adams J. The reliability of provider profiling: a tutorial. 2009. Available at http://www.rand.org/pubs/technical_reports/tr653.html. Accessed September 21, 2015. 12. Fernando HC, Landreneau RJ, Mandrekar SJ, et al. Thirty- and ninety-day outcomes after sublobar resection with and without brachytherapy for non-small cell lung cancer: results from a multicenter phase III study. J Thorac Cardiovasc Surg 2011;142:1143-51. 13. Brunelli A, Dinesh P, Woodcock-Shaw J, et al. Ninety-day mortality after video-assisted thoracoscopic lobectomy: incidence and risk factors. Ann Thorac Surg 2017;104:10201026. 14. Schneider L, Farrokhyar F, Schieman C, et al. The burden of death following discharge after lobectomy. Eur J Cardiothorac Surg 2015;48:65-70. 15. Bryant AS, Rudemiller K, Cerfolio RJ. The 30- vs 90-day operative mortality after pulmonary resection. Ann Thorac Surg 2010;89:1717-23. 16. Stephens N, Rice D, Correa A, et al. Thoracoscopic lobectomy is associated with improved short-term and equivalent oncological outcomes compared with open lobectomy for clinical stage I non-small-cell lung cancer: a propensity matched analysis of 963 cases. Eur J Cardiothorac Surg 2014;46:607-13. 17. Donahoe LL, de Valence M, Atenafu EG, et al. High risk for thoracotomy but not thoracoscopic lobectomy. Ann Thorac Surg 2017;103:1730-1735.
15
18. Swanson SJ, Meyers BF, Gunnarsson CL, et al. Video-assisted thoracoscopic lobectomy is less costly and morbid than open lobectomy: a retrospective multiinstitutional database analysis. Ann Thorac Surg 2012;93:1027-32. 19. Lapar DJ, Bhamidipati CM, Lau CL, et al. The Society of Thoracic Surgeons General Thoracic Surgery Database: establishing generalizability to national lung cancer resection outcomes. Ann Thorac Surg 2012;94:216-21. 20. Osarogiagbon R, Yu Xinhua. Nonexamination of lymph nodes and survival after resection of non-small cell lung cancer. Ann Thorac Surg 2013;96:1178-1189. 21. Speicher PJ, Gu L, Gulack BC, et al. Sublobar resection for clinical stage IA Non-smallcell lung cancer in the United States. Clin Lung Cancer 2016;17:47-55. 22. Subramanian M, McMurry T, Meyers B, et al. Long-term results for clinical stage IA lung cancer: comparing lobectomy and sublobar resection. Ann Thorac Surg 2018;106:375381. 23. Krantz SB, Lutfi W, Kuchta K, et al. Improved lymph node staging in early-stage lung cancer in the National Cancer Database. Ann Thorac Surg 2017;104:1805-1814. 24. National Comprehensive Cancer Network clinical practice guidelines in oncology. Nonsmall cell lung cancer. Available at http://www.nccn.org/professionals/physician. Accessed 1/2/2019. 25. American College of Surgeons CoC Quality of Care Measures. Lung measure specifications. Available at https://www.facs.org/~/media/files/quality%20programs/cancer/mcdb/measure%20specs %20nscl.ashx. Accessed 1/2/2019. 26. Osarogiagbon RU, Lin CC, Smeltzer MP, et al. Prevalence, prognostic implications and survival modulators of incompletely resected non-small cell lung cancer in the U.S. National Cancer Data Base. J Thorac Oncol 2016;11:e5-16.
16
27. Lin CC, Smeltzer MP, Jemal A, et al. Risk-adjusted margin positivity rate as a surgical quality metric for non-small cell lung cancer. Ann Thorac Surg 2017;104:1161-1170.
17
Table 1. Characteristics of Cohort
Characteristics
Values (n = 38,461)
Age
67 +/- 9.5
Male
21,420 (44.3%)
Caucasian
33,066 (86%)
BMI
28 +/- 6
Hypertension
24,156 (62.8%)
Steroid use
1,108 (2.9%)
Congestive heart failure
1,009 (2.6%)
Coronary artery disease
8,005 (20.8%)
Peripheral vascular disease
3,627 (9.4%)
Reoperation
2,191 (5.7%)
Cerebrovascular disease
2,946 (7.7%)
Diabetes
7,647 (19.9%)
FEV1%
83 +/- 20
Neoadjuvant chemotherapy/radiation
2,500 (6.5%)
Renal Failure
627 (1.6%)
Former smoker
23,531 (61.2%)
Current smoker
9,100 (23.7%)
Zubrod score 0
17,478 (45.4%)
1
19,331 (50.3%)
2
1,411 (3.67%)
3
241 (0.6%)
ASA class I
73 (0.2%)
II
5,595 (14.6%)
18
III
29,676 (77.2%)
IV
3,117 (8.1%)
Extent of Resection 5,917 (15.4%) Wedge 2,659 (6.9%) Segmentectomy 27,201 (70.7%) Lobectomy 422 (1.1%) Sleeve Resection 1,152 (3.0%) Bilobectomy 1,110 (2.9%) Pneumonectomy Pathologic stage I
26,128 (67.9%)
II
7,441 (19.4%)
III
4,266 (11.1%)
IV
626 (1.6%)
19
Table 2: Overall Frequency and Participant Variation in Mortality/Morbidity
Main Outcomes Operative Mortality* Major Morbidity Composite**
Number
Overall Percent
495
1.29%
Participant Median (IQR) 1.2% (0-2.1)
3,045
7.92%
7.9% (5.1-11.2)
1,123
2.92%
2.7% (1.1-4.7)
1,379
3.59%
3.1% (1.3-5.3)
124
0.32%
1,025
2.67%
244
0.63%
0%
(0-1.0)
208
0.54%
0%
(0-0.8)
198
0.51%
0%
(0-0.8)
139
0.36%
0%
(0-0.5)
127
0.33%
0%
(0-0.5)
Components of Major Morbidity Return to OR Pneumonia Initial Vent Support > 48hrs Respiratory Failure Tracheostomy ARDS Pulmonary Embolus Bronchopleural Fistula Myocardial Infarction
0%
(0-0.4)
2.3% (0.6-3.9)
*Operative Mortality: in-hospital deaths and deaths within 30 days of surgery **Major Morbidity Composite: at least one component morbidity present
20
Table 3. Reliability for Pulmonary Resection Composite Ratings Based on Volume Thresholds
No. of participants Reliability 95% CrI
No Minimum
≥30 cases
≥50 cases
≥100 cases
≥150 cases
256
214
192
136
104
46.1% (36.6%54.9%)
52.5% (42.4%61.3%)
54.9% (44.7%63.6%)
61.1% (51.0%70.1%)
63.9% (53.0%73.5%)
21
Table 4. Participant Classification by Composite Rating and Construct Validity (Morbidity and Mortality Vary Across Star Ratings)
One star
Two stars
Three stars
Number of Programs
7
194
13
Operative mortality
1.5% (1.1%, 2.0%)
1.4% (1.2%, 1.5%)
0.3% (0.14%, 0.50%)
1.3% (1.2%, 1.4%)
10.9% (9.7%,12.2%)
8.2% (7.9%, 8.5%)
3.6% (3.1%, 4.3%)
7.9% (7.6%, 8.2%)
(95% CI) Major complication (95% CI)
All participants 214
22
Table 5. Extent of Resection and Surgical Approach by Pathologic Stage
Stage I Wedge MIS Thoracotomy Segmentectomy MIS Thoracotomy Lobectomy MIS Thoracotomy Sleeve Resection MIS Thoracotomy Bilobectomy MIS Thoracotomy Pneumonectomy MIS Thoracotomy Total MIS
Stage II
Stage III
Stage IV
342 (4.6%)
315 (7.4%)
136 (21.7%)
4,582 (89.4%)
272 (79.5%)
250 (79.4%)
123 (90.4%)
542 (10.6%)
70 (20.5%)
65 (20.6%)
13 (9.6%)
2,320 (8.9%)
207 (2.8%)
102 (2.4%)
30 (4.8%)
1,916 (82.6%)
146 (70.5%)
74 (72.5%)
22 (73.3%)
404 (17.4%)
61 (29.5%)
28 (27.5%)
8 (26.7%)
17,927 (68.8%)
5,843 (78.5%)
3,028 (70.9%)
403 (64.4%)
13,714 (76.5%)
3,584 (61.3%)
1,686 (55.7%)
258 (64.0%)
4,213 (23.5%)
2,259 (38.7%)
1,342 (44.3%)
145 (36.0%)
173 (0.7%)
160 (2.2%)
87 (20.4%)
2 (0.3%)
0
0
0
0
173 (100%)
160 (100%)
87 (100%)
2 (100%)
413 (1.6%)
430 (5.8%)
286 (6.7%)
23 (3.7%)
175 (42.4%)
140 (32.6%)
74 (25.9%)
7 (30.4%)
238 (57.6%)
290 (67.4%)
212 (74.1%)
16 (69.6%)
171 ( 0.6%)
459 (6.2%)
448 (10.5%)
32 (5.1%)
29 (17.0%)
51 (11.1%)
43 (9.6%)
2 (6.3%)
142 (83.0%)
408 (88.9%)
405 (90.4)
30 (93.8%)
26,128
7,441
4,266
626
20,416 (78.2%)
4,193 (56.3%)
2,127 (49.9%)
412 (65.8%)
5,712 ( 21.8%)
3,248 (43.6%)
2,139 (50.1%)
214 (34.2%)
5,124 (19.6%)
Thoracotomy
23
Figure Legends Figure 1. Participants Sorted According to Ascending Composite Score. Figure 2. Variation in Proportion of Minimally Invasive Procedures Among Participants with >15 Resections. Figure 3. Variation in Participant Median and Mean Number of Lymph Nodes Evaluated.
24