- Email: [email protected]
The Effect of Socioeconomic Status on Treatment and Mortality in Non-Small Cell Lung Cancer Patients
Journal Pre-proof The effect of socioeconomic status on treatment and mortality in non-small cell lung cancer patients Peggy J. Ebner, BA, Li Ding, MD MPH, Anthony W. Kim, MD MS, Scott M. Atay, MD, Mimi J. Yao, BA, Omar Toubat, BA, P. Michael McFadden, MD, Alex A. Balekian, MD MS, Elizabeth A. David, MD, MAS PII:
S0003-4975(19)31240-8
DOI:
https://doi.org/10.1016/j.athoracsur.2019.07.017
Reference:
ATS 32939
To appear in:
The Annals of Thoracic Surgery
Received Date: 13 February 2019 Revised Date:
10 June 2019
Accepted Date: 5 July 2019
Please cite this article as: Ebner PJ, Ding L, Kim AW, Atay SM, Yao MJ, Toubat O, McFadden PM, Balekian AA, David EA, The effect of socioeconomic status on treatment and mortality in nonsmall cell lung cancer patients, The Annals of Thoracic Surgery (2019), doi: https://doi.org/10.1016/ j.athoracsur.2019.07.017. 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 effect of socioeconomic status on treatment and mortality in non-small cell lung cancer patients Running Head: Examining health disparities in NSCLC
Peggy J. Ebner, BA3 Li Ding, MD MPH3 Anthony W. Kim, MD MS1 Scott M. Atay, MD1 Mimi J. Yao, BA3 Omar Toubat, BA3 P. Michael McFadden, MD1 Alex A. Balekian, MD MS2 Elizabeth A. David MD, MAS1*
1
Division of Thoracic Surgery
2
Division of Pulmonary and Critical Care Medicine
3
Keck School of Medicine, University of Southern California, Los Angeles, CA
Poster Presentation at the 55th Annual Meeting of the Society of Thoracic Surgeons, San Diego, CA USA
*Corresponding Author: Elizabeth A David M.D. 1510 San Pablo St., Suite 514 Los Angeles, CA 90033 E: [email protected]
Word Count: 4,491/4,500
1
Abstract (248/250 words) Background: Treatment decisions for patients with non-small cell lung cancer (NSCLC) are based upon patient and tumor characteristics, including socioeconomic status (SES) factors. The objective was to assess the contribution of SES factors to treatment and outcomes among patients with stage I NSCLC.
Methods: The National Cancer Database was queried for operable patients with stage I NSCLC. Patients were divided into 3 treatment groups: primary resection (SUR), nonstandard treatments (chemotherapy with or without radiation) (NST), and no therapy (NoT). The SES of patients who made up the treatment groups was assessed and the 5-year survival of all groups was analyzed.
Results: The cohort included 69,168 patients with stage I NSCLC. Each of these patients had between 0 and 5 SES risk factors. The factors associated with no surgery were: low income, nonwhite race, low high school graduation rate, Medicaid or no insurance, rural residence, and distance <12.5 miles from treatment facility. Patients with multiple SES risk factors have a linearly increasing odds of undergoing NST and a quadratically increasing odds of undergoing NoT (up to OR=4.7; 95% CI 3.44-6.30 for those with 5 factors). SUR was associated with significantly longer 5-year survival (71.8%) compared to NST (22.7%) and NoT (21.8%), (p<0.0001).
Conclusions: SES factors increase the risk of undergoing guideline discordant therapy for Stage I NSCLC. As the number of SES factors increases, the odds of NoT rises quadratically while the odds of NST rises constantly. The SUR group had significantly longer survival than NST and NoT groups.
2
The actual 5-year survival rate for stage I lung cancer can range from 56% to 92% in the United States.1,2 It has been proposed that the reasons for this variability in cancer survival may include social factors such as income, ethnicity, and rural environment. These factors can contribute not only to the inability to access prompt and effective care, but can create disparities in physicians’ and patients’ knowledge and awareness of the disease and options for treatment.3,4,5,6 In addition to patient and tumor characteristics, socioeconomic status (SES) factors may influence therapeutic decisions. For example, differences in the receipt of treatment between black and white patients have been demonstrated for early stage lung cancer.7,8 It has also been established that this disparity exists independently of factors such as current health insurance or lack of an established physician before cancer diagnosis, although these factors have been shown to contribute as well. 9,10,11 The interaction and possible additive nature of factors such as these is a much more complex question that has not yet been addressed in the current literature on health disparities and cancer outcomes. It is hypothesized that SES as measured by wealth, education, race, insurance, and living environment contribute to a failure to receive guideline concordant treatment for early stage lung cancer. The objective of this study was to assess the contribution of these SES factors to the therapies provided and their respective outcomes for patients with stage I NSCLC.
Patients and Methods Data Source The National Cancer Database (NCDB) is an oncology database assembled from hospital records that contains information about patient demographics, treatment modality, and outcomes.12 Over 1,500 Commission on Cancer (CoC) Accredited treatment facilities contribute to the database. The collective data from these hospitals represents approximately 70% of the patients diagnosed with cancer annually in the United States and contains as many as 34 million historical records. The NCDB is sponsored by the American College of Surgeons and the
3
American Cancer Society. This study was exempt from review by the University of Southern California Institutional Review Board.
Patient Selection The NCDB Participant User Data File (2004-2014) was queried to identify patients with clinical stage I NSCLC (T1a,T1b,T2aN0 disease) with a tumor size <4 cm and with a CharlsonDeyo score of 0 or 1 per American Joint Commission on Cancer 7th edition staging. These patients represent a cohort for whom primary resection would constitute guideline concordant therapy.13 This cohort was further divided into those who underwent primary resection (SUR), those who underwent non-standard therapy including chemotherapy with or without radiation therapy (NST), and those who underwent no therapy (NoT) (Figure 1). The SUR group included those who underwent lobectomy or bilobectomy only and did not include wedge resections, sublobar resections, or pneumonectomies. Patients treated with radiation alone (including stereotactic body radiotherapy) were excluded from this analysis.
Variables We focused our analysis on six SES factors from the NCDB: income, race, education, insurance type, residence in relation to an urban area, and great circle distance (distance to treatment facility). They were dichotomized to facilitate comparison. Income was separated into categories of <$38,000 per year vs ≥$38,000 per year based upon the lowest quintile represented in the NCDB. Race was categorized as white vs nonwhite, with the nonwhite category including all other races recorded in the NCDB. Education was dichotomized as living in a zip code where >21% of the population had no high school education vs <20.9% of the population with no high school education. Insurance groups included Medicaid or “Not Insured” vs all other forms listed in the NCDB (Private Insurance/Managed Care, Medicare, and Other Government).10,11 Patients were dichotomized into those living in metropolitan vs
4
nonmetropolitan environments based upon the classification schemes used by the US Department of Agriculture Economic Research Service.14 Demographic variables included age, sex, tumor size, Charlson-Deyo score, and American Joint Committee on Cancer clinical stage. All patients who were classified into the SUR and NST groups underwent their therapy within 60 days of their diagnosis.15,16 Patients in the NoT group did not undergo surgery, radiation therapy, or any form of chemotherapy at any time.
Outcomes of Interest The primary outcome was odds of undergoing NoT or NST compared to undergoing SUR. The secondary outcome was overall 5-year survival among the three groups studied. The influence of number of SES factors per patient was also analyzed to determine its role in treatment group selection.
Statistical Analysis Descriptive analyses were performed to report the frequency counts and percentages for categorical variables. Multinomial logistic regression was used to test each SES factor for contribution and then consolidate them into a final model adjusting for demographic variables. In this logistic regression, each SES factor risk was counted as 1 (range 0-6, no patient had all 6 SES factors) with surgery only (SUR) as the reference group. Kaplan Meier and log rank test was used for time to event analysis. The significance level of <0.05 for two-sided tests was used. The analysis was performed using SAS software 9.4 (Cary, NC, USA).
Results A total of 69,168 patients met inclusion criteria for this study. Of those patients, 74% (51,208) underwent SUR, 9% (6,369) underwent NST, and 17% (11,591) underwent NoT (Figure 1).
5
In a bivariate analysis of the dichotomized demographic factors comparing the higher risk group to its reference group the following factors were associated with patients undergoing surgery for their disease: female sex, younger age (≤69 years), higher income (>$38,000 per year), higher rates of high school education (zip code with >21% having a high school education), insurance including private insurance/managed care, Medicare, and Other Government, longer travel distance to a treatment facility (>12.5 miles) and residence in a metro area (urban areas with populations greater than 50,000 people). Low income, living in a community with low rates of high school graduation, Medicaid or no insurance, close proximity to a treatment center, and residence in a rural area contributed to patients undergoing no treatment or nonsurgical treatments for their disease (Table 2). After adjusting for other patient characteristics and clinical factors, a multivariate logistic regression was performed to predict treatment with NST or NoT with SUR as the reference group. In this analysis, all 6 SES factors were individually associated with an increased odds of NoT and NST (Table 2). For NoT patients, except for residence in relation to an urban area, all five other factors contributed significantly to an increased odds of not undergoing surgical treatment. For patients who underwent non-standard treatments, all six SES risk factors were associated with patients not undergoing surgery. To assess the odds of undergoing NoT compared to SUR, logistic regression was performed. This analysis showed that multiple SES risk factors create a combined effect greater than the sum of their individual effects (Figure 2A). This resulted in a quadratic rise in odds of receiving no treatment for patients with increasing numbers of SES risk factors. For patients who underwent NST rather than SUR, the relationship of the odds of nonstandard treatment to the number of risk factors is approximately linear with each additional SES factor associated with a more consistent rise in odds of undergoing nonsurgical treatment (Figure 2B). The longest observed 5-year survival was among the SUR group (71.8%) which was significantly longer than 22.7% and 21.8% for the NST and NoT groups, respectively (Figure 3).
6
Comment Characterizing socioeconomic status is a challenging task. Many methods have been proposed to quantify SES, but no perfect system exists and each is subject to its own inherent bias. The model in this study sought to assess the association between SES and NSCLC outcomes based upon the factors available in the NCDB. Rather than examine any one variable, the NCDB data was examined and stratified based upon multiple SES factors. This assessment is unique because it allows for the possibility that multiple SES factors may be compounding the barriers that patients face to receiving guideline concordant care. In this paradigm, patients had a disproportionately large increase in the odds of undergoing no treatment (NoT) when they had multiple SES risk factors. The odds of undergoing nonsurgical treatment (NST) also increased but did not rise as sharply. This implies that patients with the SES factors examined in this study are not simply at risk of receiving guideline discordant treatment, but that this risk plays a measurable role in health outcomes. A patient with 5 SES risk factors has inordinately higher odds of not receiving the same care as someone with higher income, private insurance and an urban residence, or even the same care as a person with 1 SES risk factor. As interest in examining the effect of SES on health care has grown, techniques for measuring SES have evolved. In many assessments of disparity, race is used as a proxy, not only for income but often for SES as a whole.17 While well-intended, this approach may be problematic. Although patients may face barriers because of a lack of resources, they may also experience other factors independent of their material wealth, such as bias by providers.18 Furthermore, it has been proposed that more complete models of SES should include not only material capital such as income, but also human capital, expressed as education or skill level and social capital such as membership in social networks.17 Both education and living environments that include supportive social communities have been established to play an important role in successful cancer treatment and recovery.19,20
7
Associations have already been drawn between a single factor, such as race, and a difference in the receipt of care. Black patients are less likely to receive surgery for early stage lung cancer than white patients and have decreased survival.7 It has also been shown that rural communities and communities where residents have low rates of high school graduation have decreased access to surgical services.21 An association has also been established between lower SES and reduced likelihood of receiving any treatment for lung cancer.22,23 Despite the fact that these factors are known to contribute to health disparities, the combined effects of these circumstances constitutes a literature gap in current knowledge. This analysis focused on the receipt of guideline concordant treatment (SUR) compared with receipt of NoT or NST. Radiation therapy was not included in the analysis, in order to provide a streamlined comparison between the currently accepted standard treatment for early stage NSCLC (SUR) and non-surgical treatments (NST or NoT). Comparisons exist between the effectiveness of surgery and Stereotactic Body Radiation Therapy (SBRT).24,25 Whether SBRT can be considered an equivalent treatment for inoperable patients is a subject of ongoing debate outside the scope of this study.22 While receipt of prompt treatment may not be the only factor contributing to poor outcomes, stage at diagnosis has previously been shown to have the largest effect on differences in cancer survival by race.26 It is often proposed that this reflects decreased access to healthcare and bias in screening.27 In this analysis, patients were selected by cancer stage at presentation, providing for a homogenous group among whom treatment modalities and subsequent survival could be compared. Patients who underwent surgical treatment had dramatically higher survival rates (71.8% vs 22.7% and 21.8% for NST and NoT, respectively) than patients with nonsurgical or no treatment despite the fact that all patients presented with similar disease caught early in its course. While social and environmental factors may confound long-term survival, the survival disparity among this patient cohort appears to be attributable mainly to treatment modality. This observation suggests that if efforts can be made to ensure
8
that patients from at-risk SES groups undergo guideline concordant therapy, their survival will improve. These findings are also supported by previous analyses that have shown that when outcomes for patients who undergo surgery are compared by race their survival is similar.7 The importance of attributing increased mortality to lack of surgical treatment and not to late presentation cannot be overstated, as it challenges the assumption that patients are to blame for not seeking medical care sooner, and places the onus of providing guideline concordant care on providers and the healthcare system. Several other factors have been suggested to play a role in these well-described health disparities. Although Medicaid expansions under the Affordable Care Act did not go into effect until January of 2014, they may have influenced some part of the 2010-2014 dataset. Despite the potential for a slight increase in the Medicaid population at that time, it still appears that the most accurate way to dichotomize health insurance status is to include Medicaid and uninsured patients together. Many papers have shown poorer outcomes for Medicaid patients than those with private insurance or Medicare,10,11,28 and a recent publication has even found that the Medicaid expansion may be worsening health disparity.29 Differences in interactions with providers among patients of different race may also play a role. Analyses of the differences in treatment among Native American groups compared to non-Hispanic White patients have identified the importance of shared decision making between surgeons and patients as an important factor in outcomes.30 Lack of education, low income, and rural residence could easily contribute to a failure in shared decision making, which is therefore a possible explanation for the connection between low SES and treatment modality. Studies regarding sexual and gender minorities have also identified fear of bias as an important factor influencing how patients interact with the healthcare system.31 This analysis also revealed an apparent paradox in that patients who traveled further to their healthcare center fared better in terms of treatment type. This interesting finding has been
9
explored in several publications. It is often attributed to redirection of patients who are traveling to higher volume academic centers over local community hospitals.32,33 The strength of this analysis lies in the large patient cohort, which allows for a unique deconstruction of the factors that contribute to a lack of surgical treatment for early stage lung cancer. The limitations include the retrospective nature of the data, an inability to define why patients did not undergo surgery, and the complexity of quantifying SES. As with any database of the size and scope of the NCDB, our analysis is limited by available data. It would strengthen our analysis to have data on pulmonary function and performance status available for study. This type of granular information is necessary to determine which patients are considered reasonable candidates for surgical treatment.13 Not having this type of data confounds the decisions to not pursue surgery. Not all disparities can be explained by SES, and any attempt to compare the experiences of societal groups must acknowledge the difficulty of measuring patient care with numerical data. In this analysis all 6 SES factors were weighted as 1 to allow for the evaluation of their additive effect on treatment and survival disparity and the factors represented in this study did not interact in a colinear fashion. The reality of their impact on an individual’s barriers to care may be much more nuanced and furthermore may vary from patient to patient and region to region. This challenge is compounded by the fact that health and social factors are not independent. For example, poor health can influence income when a patient’s ability to work is affected.34 Future study quantifying the actual impact of an individual SES factor relative to another SES factor could elucidate these relationships and provide healthcare providers with insight into how to most effectively address disparity in the future.
Conclusion The presence of SES factors increases the odds of not undergoing guideline concordant care for Stage I NSCLC. Patients with multiple SES risk factors have a quadratic increase in
10
odds of undergoing NoT and a linear increase in odds of undergoing nonsurgical treatments rather than SUR. Patients who undergo surgery have the longest 5-year survival, while patients who underwent NST or NoT have dramatically diminished 5-year survival. Surgeons and healthcare professionals need to be aware of these differences in treatment so that the medical community can work towards understanding and mitigating these disparities in the future.
11
References 1. American Cancer Society Cancer Facts and Figures 2018. https://www.cancer.org/content/dam/cancer-org/research/cancer-facts-and-statistics/annualcancer-facts-and-figures/2018/cancer-facts-and-figures-2018.pdf. Published 2018. Accessed April 14, 2018. 2. Investigators TIELCAP. Survival of Patients with Stage I Lung Cancer Detected on CT Screening. N Engl J Med. 2006;355(17):1763-1771. 3. Lisovicz N, Wynn T, Fouad M, Partridge EE. Cancer health disparities: what we have done. Am J Med Sci. 2008;335(4):254-259. 4. Marmot M, Friel S, Bell R, Houweling TA, Taylor S. Closing the gap in a generation: health equity through action on the social determinants of health. Lancet. 2008;372(9650):16611669. 5. Cykert S, Dilworth-Anderson P, Monroe MH, et al. Factors associated with decisions to undergo surgery among patients with newly diagnosed early-stage lung cancer. JAMA. 2010;303(23):2368-2376. 6. Margolis ML, Christie JD, Silvestri GA, Kaiser L, Santiago S, Hansen-Flaschen J. Racial differences pertaining to a belief about lung cancer surgery: results of a multicenter survey. Ann Intern Med. 2003;139(7):558-563. 7. Bach PB, Cramer LD, Warren JL, Begg CB. Racial Differences in the Treatment of Early-Stage Lung Cancer. N Engl J Med. 1999;341(16):1198-1205. 8. Balekian AA, Wisnivesky JP, Gould MK. Surgical Disparities Among Patients With Stage I Lung Cancer in the National Lung Screening Trial. Chest. 2019;155(1):44-52. 9. Gross CP, Smith BD, Wolf E, Andersen M. Racial disparities in cancer therapy: did the gap narrow between 1992 and 2002? Cancer. 2008;112(4):900-908. 10. Slatore CG, Au DH, Gould MK, American Thoracic Society Disparities in Healthcare Group. An official American Thoracic Society systematic review: insurance status and disparities in lung cancer practices and outcomes. Am J Respir Crit Care Med. 2010;182(9):1195-1205. 11. Niu X, Roche LM, Pawlish KS, Henry KA. Cancer survival disparities by health insurance status. Cancer Med. 2013;2(3):403-411. 12. Bilimoria KY, Stewart AK, Winchester DP, Ko CY. The National Cancer Data Base: A Powerful Initiative to Improve Cancer Care in the United States. Ann Surg Oncol. 2008;15(3):683-690. 13. Ettinger DS, Wood DE, Aisner DL, et al. Non-Small Cell Lung Cancer, Version 5.2017, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc Netw. 2017;15(4):504-535. 14. USDA ERS - Rural Classifications. https://www.ers.usda.gov/topics/rural-economypopulation/rural-classifications.aspx#.UdsKZ6UoI64. Accessed August 28, 2018. 15. Alsamarai S, Yao X, Cain HC, et al. The Effect of a Lung Cancer Care Coordination Program on Timeliness of Care. Clin Lung Cancer. 2013;14(5):527-534. 16. Samson P, Patel A, Garrett T, et al. Effects of Delayed Surgical Resection on ShortTerm and Long-Term Outcomes in Clinical Stage I Non-Small Cell Lung Cancer. Ann Thorac Surg. 2015;99(6):1906-1912; discussion 1913. 17. Gornick ME. Measuring the Effects of Socioeconomic Status on Health Care. Guidance for the National Healthcare Disparities Report. https://www.ncbi.nlm.nih.gov/books/NBK221050/. Published 2002. Accessed August 28, 2018. 18. Shavers VL, Brown ML. Racial and ethnic disparities in the receipt of cancer treatment. J Natl Cancer Inst. 2002;94(5):334-357. 19. Pokhrel A, Martikainen P, Pukkala E, Rautalahti M, Seppä K, Hakulinen T. Education, survival and avoidable deaths in cancer patients in Finland. Br J Cancer. 2010;103(7):11091114.
12
20. Kroenke CH, Kubzansky LD, Schernhammer ES, Holmes MD, Kawachi I. Social networks, social support, and survival after breast cancer diagnosis. J Clin Oncol. 2006;24(7):1105-1111. 21. Khubchandani JA, Ingraham AM, Daniel VT, Ayturk D, Kiefe CI, Santry HP. Geographic Diffusion and Implementation of Acute Care Surgery. JAMA Surg. 2018;153(2):150. 22. Forrest LF, Adams J, Wareham H, Rubin G, White M. Socioeconomic Inequalities in Lung Cancer Treatment: Systematic Review and Meta-Analysis. Mathers CD, ed. PLoS Med. 2013;10(2):e1001376. 23. David EA, Daly ME, Li C-S, et al. Increasing Rates of No Treatment in Advanced-Stage Non–Small Cell Lung Cancer Patients: A Propensity-Matched Analysis. J Thorac Oncol. 2017;12(3):437-445. 24. Puri V, Crabtree TD, Kymes S, et al. A comparison of surgical intervention and stereotactic body radiation therapy for stage I lung cancer in high-risk patients: A decision analysis. J Thorac Cardiovasc Surg. 2012;143(2):428-436. 25. Rosen JE, Salazar MC, Wang Z, et al. Lobectomy versus stereotactic body radiotherapy in healthy patients with stage I lung cancer. J Thorac Cardiovasc Surg. 2016;152(1):44-54.e9. 26. Ellis L, Canchola AJ, Spiegel D, Ladabaum U, Haile R, Gomez SL. Racial and Ethnic Disparities in Cancer Survival: The Contribution of Tumor, Sociodemographic, Institutional, and Neighborhood Characteristics. J Clin Oncol. 2018;36(1):25-33. 27. Japuntich SJ, Krieger NH, Salvas AL, Carey MP. Racial Disparities in Lung Cancer Screening: An Exploratory Investigation. J Natl Med Assoc. 2018;110(5):424-427. 28. Baicker K, Taubman SL, Allen HL, et al. The Oregon Experiment — Effects of Medicaid on Clinical Outcomes. N Engl J Med. 2013;368(18):1713-1722. 29. Andrews C. Unintended consequences: Medicaid expansion and racial inequality in access to health insurance. Health Soc Work. 2014;39(3):131-133. 30. Morris AM, Doorenbos AZ, Haozous E, Meins A, Javid S, Flum DR. Perceptions of cancer treatment decision making among American Indians/Alaska Natives and their physicians. Psychooncology. 2016;25(9):1050-1056. 31. Gibson AW, Radix AE, Maingi S, Patel S. Cancer care in lesbian, gay, bisexual, transgender and queer populations. Futur Oncol. 2017;13(15):1333-1344. 32. Vetterlein MW, Löppenberg B, Karabon P, et al. Impact of travel distance to the treatment facility on overall mortality in US patients with prostate cancer. Cancer. 2017;123(17):3241-3252. 33. Ryan S, Serrell EC, Karabon P, et al. The Association between Mortality and Distance to Treatment Facility in Patients with Muscle Invasive Bladder Cancer. J Urol. 2018;199(2):424429. 34. Crimmins E, Preston S, Cohen B. Explaining Divergent Levels of Longevity in HighIncome Countries. National Research Council (US) Panel on Understanding Divergent Trends in Longevity in High-Income Countries. http://doi.wiley.com/10.1177/0884533612444541. Published 2011. Accessed August 28, 2018.
13
Figure Legends Figure 1. Consort diagram of the NCDB cohort.
Figure 2. Risk of undergoing NoT (A) and NST (B) based on number of SES factors. Odds ratios of undergoing NST or NoT were plotted against increasing number of SES factors. SUR was used as the comparator group. The NoT line rises quadratically while the NST relationship is more linear. A relatively small sample size of patients with 5 SES factors is likely responsible for the lower OR of undergoing NST for this group. Abbreviations: NST, non-standard treatment; NoT, no treatment; SUR, surgery only; SES, socioeconomic status.
Figure 3. Kaplan-Meier analysis of overall survival by treatment group. Five-year survival by treatment type. Abbreviations: NST, non-standard treatment; NoT, no treatment; SUR, surgery only; SES, socioeconomic status.
Figure 4. Kaplan-Meier analysis of overall survival by number of SES factors. Relatively higher survival for patients with 5 SES factors is likely due to small sample size of this subgroup.
14
Table 1. Demographic and clinical characteristics of patients from the NCDB Characteristic
NST No.
%
NoT No.
Gender Female 3258 51.15 6238 Male 3111 48.85 5353 Age ≤ 69 2739 43.01 3779 > 69 3630 56.99 7812 Race/ethnicity White 5568 87.42 9854 Other 767 12.04 1653 AJCC Clinical T Stage 1 3226 50.65 5229 1A 1087 17.07 2970 1B 956 15.01 1994 2A 1100 17.27 1398 Charlson-Deyo Score 0 4169 65.46 7736 1 2200 34.54 3855 Income < 1409 22.12 2407 $38,000 ≥ 4862 76.34 8985 $38,000 Education <20.9% No HS Diploma 5093 79.97 9201 >21% No HS Diploma 1180 18.53 2193 Insurance Not Insured or Medicaid 365 5.73 792 Other 5906 92.73 10617 Urban vs Rural Residence Metro 5226 82.05 9755 Non942 14.79 1366 Metro Great Circle Distance ≤ 12.5 3812 59.85 7298 miles >12.5 2463 38.67 4097 miles Abbreviations: NST, non-standard treatment; NoT, no treatment; SUR, surgery only.
%
SUR No.
%
p-Value
53.82 46.18
29722 21486
58.04 41.96
<.0001
32.60 67.40
28904 22304
56.44 43.56
<.0001
85.01 14.26
45970 4887
89.77 9.54
<.0001
45.11 25.62 17.20 12.06
23091 14747 8718 4652
45.09 28.80 17.02 9.08
<.0001
66.74 33.26
30654 20554
59.86 40.14
20.77
8011
15.64
77.52
42705
83.40
79.38 18.92
43557 7186
85.06 14.03
<.0001
6.83 91.60
2555 48075
4.99 93.88
<.0001
84.16 11.79
43717 5859
85.37 11.44
<.0001
62.96
28036
54.75
<.0001
35.35
22711
44.35
<.0001
<.0001
15
Table 2. Significant factors predictive of treatment group for the NCDB patient cohort NST vs SUR OR
NoT vs SUR
95% CI
p-Value
OR
<0.0001
Ref 1.50
95% CI
p-Value
Race/ethnicity White Other
Ref 1.19
≥ $38,000
Ref
< $38,000
1.34
1.24
1.45
<0.0001
<20.9% No HS Diploma >21% No HS Diploma
Ref 1.16
1.07
1.26
0.0004
Private Insurance/ Medicare/ Managed Care/ Other
Ref
Not Insured or Medicaid
1.36
1.09
1.30
1.41
1.60
<0.0001
1.22
1.15
1.30
<0.0001
Ref 1.28
1.20
1.36
<0.0001
1.84
2.21
<0.0001
1.16
1.34
0.875
1.38
1.52
<0.0001
Income Ref
Educatio n
Insurance Ref
1.21
1.53
<0.0001
2.02
Urban vs Rural Residence Metro
Ref
Non-Metro
1.47
Ref 1.35
1.61
<0.0001
1.25
Great Circle Distance >12.5 miles
Ref
≤ 12.5 1.39 1.31 1.48 <0.0001 miles Abbreviations: NST, non-standard treatment; NoT, no treatment; SUR, surgery only.
Ref 1.45
16
S0003-4975(19)31240-8
DOI:
https://doi.org/10.1016/j.athoracsur.2019.07.017
Reference:
ATS 32939
To appear in:
The Annals of Thoracic Surgery
Received Date: 13 February 2019 Revised Date:
10 June 2019
Accepted Date: 5 July 2019
Please cite this article as: Ebner PJ, Ding L, Kim AW, Atay SM, Yao MJ, Toubat O, McFadden PM, Balekian AA, David EA, The effect of socioeconomic status on treatment and mortality in nonsmall cell lung cancer patients, The Annals of Thoracic Surgery (2019), doi: https://doi.org/10.1016/ j.athoracsur.2019.07.017. 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 effect of socioeconomic status on treatment and mortality in non-small cell lung cancer patients Running Head: Examining health disparities in NSCLC
Peggy J. Ebner, BA3 Li Ding, MD MPH3 Anthony W. Kim, MD MS1 Scott M. Atay, MD1 Mimi J. Yao, BA3 Omar Toubat, BA3 P. Michael McFadden, MD1 Alex A. Balekian, MD MS2 Elizabeth A. David MD, MAS1*
1
Division of Thoracic Surgery
2
Division of Pulmonary and Critical Care Medicine
3
Keck School of Medicine, University of Southern California, Los Angeles, CA
Poster Presentation at the 55th Annual Meeting of the Society of Thoracic Surgeons, San Diego, CA USA
*Corresponding Author: Elizabeth A David M.D. 1510 San Pablo St., Suite 514 Los Angeles, CA 90033 E: [email protected]
Word Count: 4,491/4,500
1
Abstract (248/250 words) Background: Treatment decisions for patients with non-small cell lung cancer (NSCLC) are based upon patient and tumor characteristics, including socioeconomic status (SES) factors. The objective was to assess the contribution of SES factors to treatment and outcomes among patients with stage I NSCLC.
Methods: The National Cancer Database was queried for operable patients with stage I NSCLC. Patients were divided into 3 treatment groups: primary resection (SUR), nonstandard treatments (chemotherapy with or without radiation) (NST), and no therapy (NoT). The SES of patients who made up the treatment groups was assessed and the 5-year survival of all groups was analyzed.
Results: The cohort included 69,168 patients with stage I NSCLC. Each of these patients had between 0 and 5 SES risk factors. The factors associated with no surgery were: low income, nonwhite race, low high school graduation rate, Medicaid or no insurance, rural residence, and distance <12.5 miles from treatment facility. Patients with multiple SES risk factors have a linearly increasing odds of undergoing NST and a quadratically increasing odds of undergoing NoT (up to OR=4.7; 95% CI 3.44-6.30 for those with 5 factors). SUR was associated with significantly longer 5-year survival (71.8%) compared to NST (22.7%) and NoT (21.8%), (p<0.0001).
Conclusions: SES factors increase the risk of undergoing guideline discordant therapy for Stage I NSCLC. As the number of SES factors increases, the odds of NoT rises quadratically while the odds of NST rises constantly. The SUR group had significantly longer survival than NST and NoT groups.
2
The actual 5-year survival rate for stage I lung cancer can range from 56% to 92% in the United States.1,2 It has been proposed that the reasons for this variability in cancer survival may include social factors such as income, ethnicity, and rural environment. These factors can contribute not only to the inability to access prompt and effective care, but can create disparities in physicians’ and patients’ knowledge and awareness of the disease and options for treatment.3,4,5,6 In addition to patient and tumor characteristics, socioeconomic status (SES) factors may influence therapeutic decisions. For example, differences in the receipt of treatment between black and white patients have been demonstrated for early stage lung cancer.7,8 It has also been established that this disparity exists independently of factors such as current health insurance or lack of an established physician before cancer diagnosis, although these factors have been shown to contribute as well. 9,10,11 The interaction and possible additive nature of factors such as these is a much more complex question that has not yet been addressed in the current literature on health disparities and cancer outcomes. It is hypothesized that SES as measured by wealth, education, race, insurance, and living environment contribute to a failure to receive guideline concordant treatment for early stage lung cancer. The objective of this study was to assess the contribution of these SES factors to the therapies provided and their respective outcomes for patients with stage I NSCLC.
Patients and Methods Data Source The National Cancer Database (NCDB) is an oncology database assembled from hospital records that contains information about patient demographics, treatment modality, and outcomes.12 Over 1,500 Commission on Cancer (CoC) Accredited treatment facilities contribute to the database. The collective data from these hospitals represents approximately 70% of the patients diagnosed with cancer annually in the United States and contains as many as 34 million historical records. The NCDB is sponsored by the American College of Surgeons and the
3
American Cancer Society. This study was exempt from review by the University of Southern California Institutional Review Board.
Patient Selection The NCDB Participant User Data File (2004-2014) was queried to identify patients with clinical stage I NSCLC (T1a,T1b,T2aN0 disease) with a tumor size <4 cm and with a CharlsonDeyo score of 0 or 1 per American Joint Commission on Cancer 7th edition staging. These patients represent a cohort for whom primary resection would constitute guideline concordant therapy.13 This cohort was further divided into those who underwent primary resection (SUR), those who underwent non-standard therapy including chemotherapy with or without radiation therapy (NST), and those who underwent no therapy (NoT) (Figure 1). The SUR group included those who underwent lobectomy or bilobectomy only and did not include wedge resections, sublobar resections, or pneumonectomies. Patients treated with radiation alone (including stereotactic body radiotherapy) were excluded from this analysis.
Variables We focused our analysis on six SES factors from the NCDB: income, race, education, insurance type, residence in relation to an urban area, and great circle distance (distance to treatment facility). They were dichotomized to facilitate comparison. Income was separated into categories of <$38,000 per year vs ≥$38,000 per year based upon the lowest quintile represented in the NCDB. Race was categorized as white vs nonwhite, with the nonwhite category including all other races recorded in the NCDB. Education was dichotomized as living in a zip code where >21% of the population had no high school education vs <20.9% of the population with no high school education. Insurance groups included Medicaid or “Not Insured” vs all other forms listed in the NCDB (Private Insurance/Managed Care, Medicare, and Other Government).10,11 Patients were dichotomized into those living in metropolitan vs
4
nonmetropolitan environments based upon the classification schemes used by the US Department of Agriculture Economic Research Service.14 Demographic variables included age, sex, tumor size, Charlson-Deyo score, and American Joint Committee on Cancer clinical stage. All patients who were classified into the SUR and NST groups underwent their therapy within 60 days of their diagnosis.15,16 Patients in the NoT group did not undergo surgery, radiation therapy, or any form of chemotherapy at any time.
Outcomes of Interest The primary outcome was odds of undergoing NoT or NST compared to undergoing SUR. The secondary outcome was overall 5-year survival among the three groups studied. The influence of number of SES factors per patient was also analyzed to determine its role in treatment group selection.
Statistical Analysis Descriptive analyses were performed to report the frequency counts and percentages for categorical variables. Multinomial logistic regression was used to test each SES factor for contribution and then consolidate them into a final model adjusting for demographic variables. In this logistic regression, each SES factor risk was counted as 1 (range 0-6, no patient had all 6 SES factors) with surgery only (SUR) as the reference group. Kaplan Meier and log rank test was used for time to event analysis. The significance level of <0.05 for two-sided tests was used. The analysis was performed using SAS software 9.4 (Cary, NC, USA).
Results A total of 69,168 patients met inclusion criteria for this study. Of those patients, 74% (51,208) underwent SUR, 9% (6,369) underwent NST, and 17% (11,591) underwent NoT (Figure 1).
5
In a bivariate analysis of the dichotomized demographic factors comparing the higher risk group to its reference group the following factors were associated with patients undergoing surgery for their disease: female sex, younger age (≤69 years), higher income (>$38,000 per year), higher rates of high school education (zip code with >21% having a high school education), insurance including private insurance/managed care, Medicare, and Other Government, longer travel distance to a treatment facility (>12.5 miles) and residence in a metro area (urban areas with populations greater than 50,000 people). Low income, living in a community with low rates of high school graduation, Medicaid or no insurance, close proximity to a treatment center, and residence in a rural area contributed to patients undergoing no treatment or nonsurgical treatments for their disease (Table 2). After adjusting for other patient characteristics and clinical factors, a multivariate logistic regression was performed to predict treatment with NST or NoT with SUR as the reference group. In this analysis, all 6 SES factors were individually associated with an increased odds of NoT and NST (Table 2). For NoT patients, except for residence in relation to an urban area, all five other factors contributed significantly to an increased odds of not undergoing surgical treatment. For patients who underwent non-standard treatments, all six SES risk factors were associated with patients not undergoing surgery. To assess the odds of undergoing NoT compared to SUR, logistic regression was performed. This analysis showed that multiple SES risk factors create a combined effect greater than the sum of their individual effects (Figure 2A). This resulted in a quadratic rise in odds of receiving no treatment for patients with increasing numbers of SES risk factors. For patients who underwent NST rather than SUR, the relationship of the odds of nonstandard treatment to the number of risk factors is approximately linear with each additional SES factor associated with a more consistent rise in odds of undergoing nonsurgical treatment (Figure 2B). The longest observed 5-year survival was among the SUR group (71.8%) which was significantly longer than 22.7% and 21.8% for the NST and NoT groups, respectively (Figure 3).
6
Comment Characterizing socioeconomic status is a challenging task. Many methods have been proposed to quantify SES, but no perfect system exists and each is subject to its own inherent bias. The model in this study sought to assess the association between SES and NSCLC outcomes based upon the factors available in the NCDB. Rather than examine any one variable, the NCDB data was examined and stratified based upon multiple SES factors. This assessment is unique because it allows for the possibility that multiple SES factors may be compounding the barriers that patients face to receiving guideline concordant care. In this paradigm, patients had a disproportionately large increase in the odds of undergoing no treatment (NoT) when they had multiple SES risk factors. The odds of undergoing nonsurgical treatment (NST) also increased but did not rise as sharply. This implies that patients with the SES factors examined in this study are not simply at risk of receiving guideline discordant treatment, but that this risk plays a measurable role in health outcomes. A patient with 5 SES risk factors has inordinately higher odds of not receiving the same care as someone with higher income, private insurance and an urban residence, or even the same care as a person with 1 SES risk factor. As interest in examining the effect of SES on health care has grown, techniques for measuring SES have evolved. In many assessments of disparity, race is used as a proxy, not only for income but often for SES as a whole.17 While well-intended, this approach may be problematic. Although patients may face barriers because of a lack of resources, they may also experience other factors independent of their material wealth, such as bias by providers.18 Furthermore, it has been proposed that more complete models of SES should include not only material capital such as income, but also human capital, expressed as education or skill level and social capital such as membership in social networks.17 Both education and living environments that include supportive social communities have been established to play an important role in successful cancer treatment and recovery.19,20
7
Associations have already been drawn between a single factor, such as race, and a difference in the receipt of care. Black patients are less likely to receive surgery for early stage lung cancer than white patients and have decreased survival.7 It has also been shown that rural communities and communities where residents have low rates of high school graduation have decreased access to surgical services.21 An association has also been established between lower SES and reduced likelihood of receiving any treatment for lung cancer.22,23 Despite the fact that these factors are known to contribute to health disparities, the combined effects of these circumstances constitutes a literature gap in current knowledge. This analysis focused on the receipt of guideline concordant treatment (SUR) compared with receipt of NoT or NST. Radiation therapy was not included in the analysis, in order to provide a streamlined comparison between the currently accepted standard treatment for early stage NSCLC (SUR) and non-surgical treatments (NST or NoT). Comparisons exist between the effectiveness of surgery and Stereotactic Body Radiation Therapy (SBRT).24,25 Whether SBRT can be considered an equivalent treatment for inoperable patients is a subject of ongoing debate outside the scope of this study.22 While receipt of prompt treatment may not be the only factor contributing to poor outcomes, stage at diagnosis has previously been shown to have the largest effect on differences in cancer survival by race.26 It is often proposed that this reflects decreased access to healthcare and bias in screening.27 In this analysis, patients were selected by cancer stage at presentation, providing for a homogenous group among whom treatment modalities and subsequent survival could be compared. Patients who underwent surgical treatment had dramatically higher survival rates (71.8% vs 22.7% and 21.8% for NST and NoT, respectively) than patients with nonsurgical or no treatment despite the fact that all patients presented with similar disease caught early in its course. While social and environmental factors may confound long-term survival, the survival disparity among this patient cohort appears to be attributable mainly to treatment modality. This observation suggests that if efforts can be made to ensure
8
that patients from at-risk SES groups undergo guideline concordant therapy, their survival will improve. These findings are also supported by previous analyses that have shown that when outcomes for patients who undergo surgery are compared by race their survival is similar.7 The importance of attributing increased mortality to lack of surgical treatment and not to late presentation cannot be overstated, as it challenges the assumption that patients are to blame for not seeking medical care sooner, and places the onus of providing guideline concordant care on providers and the healthcare system. Several other factors have been suggested to play a role in these well-described health disparities. Although Medicaid expansions under the Affordable Care Act did not go into effect until January of 2014, they may have influenced some part of the 2010-2014 dataset. Despite the potential for a slight increase in the Medicaid population at that time, it still appears that the most accurate way to dichotomize health insurance status is to include Medicaid and uninsured patients together. Many papers have shown poorer outcomes for Medicaid patients than those with private insurance or Medicare,10,11,28 and a recent publication has even found that the Medicaid expansion may be worsening health disparity.29 Differences in interactions with providers among patients of different race may also play a role. Analyses of the differences in treatment among Native American groups compared to non-Hispanic White patients have identified the importance of shared decision making between surgeons and patients as an important factor in outcomes.30 Lack of education, low income, and rural residence could easily contribute to a failure in shared decision making, which is therefore a possible explanation for the connection between low SES and treatment modality. Studies regarding sexual and gender minorities have also identified fear of bias as an important factor influencing how patients interact with the healthcare system.31 This analysis also revealed an apparent paradox in that patients who traveled further to their healthcare center fared better in terms of treatment type. This interesting finding has been
9
explored in several publications. It is often attributed to redirection of patients who are traveling to higher volume academic centers over local community hospitals.32,33 The strength of this analysis lies in the large patient cohort, which allows for a unique deconstruction of the factors that contribute to a lack of surgical treatment for early stage lung cancer. The limitations include the retrospective nature of the data, an inability to define why patients did not undergo surgery, and the complexity of quantifying SES. As with any database of the size and scope of the NCDB, our analysis is limited by available data. It would strengthen our analysis to have data on pulmonary function and performance status available for study. This type of granular information is necessary to determine which patients are considered reasonable candidates for surgical treatment.13 Not having this type of data confounds the decisions to not pursue surgery. Not all disparities can be explained by SES, and any attempt to compare the experiences of societal groups must acknowledge the difficulty of measuring patient care with numerical data. In this analysis all 6 SES factors were weighted as 1 to allow for the evaluation of their additive effect on treatment and survival disparity and the factors represented in this study did not interact in a colinear fashion. The reality of their impact on an individual’s barriers to care may be much more nuanced and furthermore may vary from patient to patient and region to region. This challenge is compounded by the fact that health and social factors are not independent. For example, poor health can influence income when a patient’s ability to work is affected.34 Future study quantifying the actual impact of an individual SES factor relative to another SES factor could elucidate these relationships and provide healthcare providers with insight into how to most effectively address disparity in the future.
Conclusion The presence of SES factors increases the odds of not undergoing guideline concordant care for Stage I NSCLC. Patients with multiple SES risk factors have a quadratic increase in
10
odds of undergoing NoT and a linear increase in odds of undergoing nonsurgical treatments rather than SUR. Patients who undergo surgery have the longest 5-year survival, while patients who underwent NST or NoT have dramatically diminished 5-year survival. Surgeons and healthcare professionals need to be aware of these differences in treatment so that the medical community can work towards understanding and mitigating these disparities in the future.
11
References 1. American Cancer Society Cancer Facts and Figures 2018. https://www.cancer.org/content/dam/cancer-org/research/cancer-facts-and-statistics/annualcancer-facts-and-figures/2018/cancer-facts-and-figures-2018.pdf. Published 2018. Accessed April 14, 2018. 2. Investigators TIELCAP. Survival of Patients with Stage I Lung Cancer Detected on CT Screening. N Engl J Med. 2006;355(17):1763-1771. 3. Lisovicz N, Wynn T, Fouad M, Partridge EE. Cancer health disparities: what we have done. Am J Med Sci. 2008;335(4):254-259. 4. Marmot M, Friel S, Bell R, Houweling TA, Taylor S. Closing the gap in a generation: health equity through action on the social determinants of health. Lancet. 2008;372(9650):16611669. 5. Cykert S, Dilworth-Anderson P, Monroe MH, et al. Factors associated with decisions to undergo surgery among patients with newly diagnosed early-stage lung cancer. JAMA. 2010;303(23):2368-2376. 6. Margolis ML, Christie JD, Silvestri GA, Kaiser L, Santiago S, Hansen-Flaschen J. Racial differences pertaining to a belief about lung cancer surgery: results of a multicenter survey. Ann Intern Med. 2003;139(7):558-563. 7. Bach PB, Cramer LD, Warren JL, Begg CB. Racial Differences in the Treatment of Early-Stage Lung Cancer. N Engl J Med. 1999;341(16):1198-1205. 8. Balekian AA, Wisnivesky JP, Gould MK. Surgical Disparities Among Patients With Stage I Lung Cancer in the National Lung Screening Trial. Chest. 2019;155(1):44-52. 9. Gross CP, Smith BD, Wolf E, Andersen M. Racial disparities in cancer therapy: did the gap narrow between 1992 and 2002? Cancer. 2008;112(4):900-908. 10. Slatore CG, Au DH, Gould MK, American Thoracic Society Disparities in Healthcare Group. An official American Thoracic Society systematic review: insurance status and disparities in lung cancer practices and outcomes. Am J Respir Crit Care Med. 2010;182(9):1195-1205. 11. Niu X, Roche LM, Pawlish KS, Henry KA. Cancer survival disparities by health insurance status. Cancer Med. 2013;2(3):403-411. 12. Bilimoria KY, Stewart AK, Winchester DP, Ko CY. The National Cancer Data Base: A Powerful Initiative to Improve Cancer Care in the United States. Ann Surg Oncol. 2008;15(3):683-690. 13. Ettinger DS, Wood DE, Aisner DL, et al. Non-Small Cell Lung Cancer, Version 5.2017, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc Netw. 2017;15(4):504-535. 14. USDA ERS - Rural Classifications. https://www.ers.usda.gov/topics/rural-economypopulation/rural-classifications.aspx#.UdsKZ6UoI64. Accessed August 28, 2018. 15. Alsamarai S, Yao X, Cain HC, et al. The Effect of a Lung Cancer Care Coordination Program on Timeliness of Care. Clin Lung Cancer. 2013;14(5):527-534. 16. Samson P, Patel A, Garrett T, et al. Effects of Delayed Surgical Resection on ShortTerm and Long-Term Outcomes in Clinical Stage I Non-Small Cell Lung Cancer. Ann Thorac Surg. 2015;99(6):1906-1912; discussion 1913. 17. Gornick ME. Measuring the Effects of Socioeconomic Status on Health Care. Guidance for the National Healthcare Disparities Report. https://www.ncbi.nlm.nih.gov/books/NBK221050/. Published 2002. Accessed August 28, 2018. 18. Shavers VL, Brown ML. Racial and ethnic disparities in the receipt of cancer treatment. J Natl Cancer Inst. 2002;94(5):334-357. 19. Pokhrel A, Martikainen P, Pukkala E, Rautalahti M, Seppä K, Hakulinen T. Education, survival and avoidable deaths in cancer patients in Finland. Br J Cancer. 2010;103(7):11091114.
12
20. Kroenke CH, Kubzansky LD, Schernhammer ES, Holmes MD, Kawachi I. Social networks, social support, and survival after breast cancer diagnosis. J Clin Oncol. 2006;24(7):1105-1111. 21. Khubchandani JA, Ingraham AM, Daniel VT, Ayturk D, Kiefe CI, Santry HP. Geographic Diffusion and Implementation of Acute Care Surgery. JAMA Surg. 2018;153(2):150. 22. Forrest LF, Adams J, Wareham H, Rubin G, White M. Socioeconomic Inequalities in Lung Cancer Treatment: Systematic Review and Meta-Analysis. Mathers CD, ed. PLoS Med. 2013;10(2):e1001376. 23. David EA, Daly ME, Li C-S, et al. Increasing Rates of No Treatment in Advanced-Stage Non–Small Cell Lung Cancer Patients: A Propensity-Matched Analysis. J Thorac Oncol. 2017;12(3):437-445. 24. Puri V, Crabtree TD, Kymes S, et al. A comparison of surgical intervention and stereotactic body radiation therapy for stage I lung cancer in high-risk patients: A decision analysis. J Thorac Cardiovasc Surg. 2012;143(2):428-436. 25. Rosen JE, Salazar MC, Wang Z, et al. Lobectomy versus stereotactic body radiotherapy in healthy patients with stage I lung cancer. J Thorac Cardiovasc Surg. 2016;152(1):44-54.e9. 26. Ellis L, Canchola AJ, Spiegel D, Ladabaum U, Haile R, Gomez SL. Racial and Ethnic Disparities in Cancer Survival: The Contribution of Tumor, Sociodemographic, Institutional, and Neighborhood Characteristics. J Clin Oncol. 2018;36(1):25-33. 27. Japuntich SJ, Krieger NH, Salvas AL, Carey MP. Racial Disparities in Lung Cancer Screening: An Exploratory Investigation. J Natl Med Assoc. 2018;110(5):424-427. 28. Baicker K, Taubman SL, Allen HL, et al. The Oregon Experiment — Effects of Medicaid on Clinical Outcomes. N Engl J Med. 2013;368(18):1713-1722. 29. Andrews C. Unintended consequences: Medicaid expansion and racial inequality in access to health insurance. Health Soc Work. 2014;39(3):131-133. 30. Morris AM, Doorenbos AZ, Haozous E, Meins A, Javid S, Flum DR. Perceptions of cancer treatment decision making among American Indians/Alaska Natives and their physicians. Psychooncology. 2016;25(9):1050-1056. 31. Gibson AW, Radix AE, Maingi S, Patel S. Cancer care in lesbian, gay, bisexual, transgender and queer populations. Futur Oncol. 2017;13(15):1333-1344. 32. Vetterlein MW, Löppenberg B, Karabon P, et al. Impact of travel distance to the treatment facility on overall mortality in US patients with prostate cancer. Cancer. 2017;123(17):3241-3252. 33. Ryan S, Serrell EC, Karabon P, et al. The Association between Mortality and Distance to Treatment Facility in Patients with Muscle Invasive Bladder Cancer. J Urol. 2018;199(2):424429. 34. Crimmins E, Preston S, Cohen B. Explaining Divergent Levels of Longevity in HighIncome Countries. National Research Council (US) Panel on Understanding Divergent Trends in Longevity in High-Income Countries. http://doi.wiley.com/10.1177/0884533612444541. Published 2011. Accessed August 28, 2018.
13
Figure Legends Figure 1. Consort diagram of the NCDB cohort.
Figure 2. Risk of undergoing NoT (A) and NST (B) based on number of SES factors. Odds ratios of undergoing NST or NoT were plotted against increasing number of SES factors. SUR was used as the comparator group. The NoT line rises quadratically while the NST relationship is more linear. A relatively small sample size of patients with 5 SES factors is likely responsible for the lower OR of undergoing NST for this group. Abbreviations: NST, non-standard treatment; NoT, no treatment; SUR, surgery only; SES, socioeconomic status.
Figure 3. Kaplan-Meier analysis of overall survival by treatment group. Five-year survival by treatment type. Abbreviations: NST, non-standard treatment; NoT, no treatment; SUR, surgery only; SES, socioeconomic status.
Figure 4. Kaplan-Meier analysis of overall survival by number of SES factors. Relatively higher survival for patients with 5 SES factors is likely due to small sample size of this subgroup.
14
Table 1. Demographic and clinical characteristics of patients from the NCDB Characteristic
NST No.
%
NoT No.
Gender Female 3258 51.15 6238 Male 3111 48.85 5353 Age ≤ 69 2739 43.01 3779 > 69 3630 56.99 7812 Race/ethnicity White 5568 87.42 9854 Other 767 12.04 1653 AJCC Clinical T Stage 1 3226 50.65 5229 1A 1087 17.07 2970 1B 956 15.01 1994 2A 1100 17.27 1398 Charlson-Deyo Score 0 4169 65.46 7736 1 2200 34.54 3855 Income < 1409 22.12 2407 $38,000 ≥ 4862 76.34 8985 $38,000 Education <20.9% No HS Diploma 5093 79.97 9201 >21% No HS Diploma 1180 18.53 2193 Insurance Not Insured or Medicaid 365 5.73 792 Other 5906 92.73 10617 Urban vs Rural Residence Metro 5226 82.05 9755 Non942 14.79 1366 Metro Great Circle Distance ≤ 12.5 3812 59.85 7298 miles >12.5 2463 38.67 4097 miles Abbreviations: NST, non-standard treatment; NoT, no treatment; SUR, surgery only.
%
SUR No.
%
p-Value
53.82 46.18
29722 21486
58.04 41.96
<.0001
32.60 67.40
28904 22304
56.44 43.56
<.0001
85.01 14.26
45970 4887
89.77 9.54
<.0001
45.11 25.62 17.20 12.06
23091 14747 8718 4652
45.09 28.80 17.02 9.08
<.0001
66.74 33.26
30654 20554
59.86 40.14
20.77
8011
15.64
77.52
42705
83.40
79.38 18.92
43557 7186
85.06 14.03
<.0001
6.83 91.60
2555 48075
4.99 93.88
<.0001
84.16 11.79
43717 5859
85.37 11.44
<.0001
62.96
28036
54.75
<.0001
35.35
22711
44.35
<.0001
<.0001
15
Table 2. Significant factors predictive of treatment group for the NCDB patient cohort NST vs SUR OR
NoT vs SUR
95% CI
p-Value
OR
<0.0001
Ref 1.50
95% CI
p-Value
Race/ethnicity White Other
Ref 1.19
≥ $38,000
Ref
< $38,000
1.34
1.24
1.45
<0.0001
<20.9% No HS Diploma >21% No HS Diploma
Ref 1.16
1.07
1.26
0.0004
Private Insurance/ Medicare/ Managed Care/ Other
Ref
Not Insured or Medicaid
1.36
1.09
1.30
1.41
1.60
<0.0001
1.22
1.15
1.30
<0.0001
Ref 1.28
1.20
1.36
<0.0001
1.84
2.21
<0.0001
1.16
1.34
0.875
1.38
1.52
<0.0001
Income Ref
Educatio n
Insurance Ref
1.21
1.53
<0.0001
2.02
Urban vs Rural Residence Metro
Ref
Non-Metro
1.47
Ref 1.35
1.61
<0.0001
1.25
Great Circle Distance >12.5 miles
Ref
≤ 12.5 1.39 1.31 1.48 <0.0001 miles Abbreviations: NST, non-standard treatment; NoT, no treatment; SUR, surgery only.
Ref 1.45
16