Association of rurality with survival and receipt of treatment in - PowerPoint PPT Presentation

Association of rurality with survival and receipt of treatment in early- stage non-small cell lung cancer Charles D. Nicoli, B.S. Brian L. Sprague, Ph.D. Nataniel H. Lester-Coll, M.D.

Disclosures • No conflicts of interest • University of Vermont Institutional Review Board Not Human Subjects Exemption: 18-0075

Background • Rural populations make up ~20% of the U.S. 1 • Rural areas of the U.S. are experiencing negative population growth. 2 • Recent decreasing trend of lung cancer mortality has been less pronounced in rural areas. 3-5 • Inconsistent findings of rurality’s impact on lung cancer outcomes. 3,4,7-10 • Stage I NSCLC is amenable to intervention with established treatment paradigm (lobectomy + mediastinal LN dissection or SBRT). 6

Methods: Data and Analysis • National Cancer Data Base (NCDB): joint venture of American College of Surgeons, American Cancer Society • Captures ~70% of invasive cancers in the U.S. • In 2005, captured estimated 82.1% of invasive lung & bronchus cancers in the U.S. 11

Survival Analysis • Unadjusted Survival • Log-Rank test, Kaplan-Meier method, Cox proportional-hazards model 12 • Multivariable survival modeling • Cox proportional-hazards model. 12 • Preceded by stepwise logistic regressions, included as covariates those associated with survival at p < 0.05

Methods: Guidelines-Concordant Treatment • Surgical: lobectomy with mediastinal lymph node dissection. 13,14 • Stereotactic Body Radiation Therapy (SBRT): • Defined according to Corso et al., 2017: 13,15 • Treatment modality: radiation therapy and not surgery • BED 10 between 40 and 300 Gy • < 10 fractions • No guidelines-concordant treatment: neither lobectomy w/LN dissection nor SBRT

Methods: Defining Rurality • NCDB includes the USDA’s Rural-Urban Continuum Codes (RUCC), 2013 version, in defining rurality of patient residence 1 2 3 4 5 6 7 8 9 Source: USDA, Economic Research Service using Data from the U.S. Census Bureau [16]

Figure 1: Creating a Rural Dichotomy

Methods: Defining Rurality 1 2 3 4 5 6 7 8 9 Source: USDA, Economic Research Service using Data from the U.S. Census Bureau [16]

Methods: Defining Rurality Rural Non-Rural Source: USDA, Economic Research Service using Data from the U.S. Census Bureau [16]

Figure 2. Exclusion Criteria

Sample Characteristics • 81.3% Non-Rural, 18.7% Rural Non-Rural Rural Overall n % n % n % AJCC Stage at Diagnosis Stage I 149,000 21.8 33,278 21.2 182,278 21.7 Stage II 45,972 6.7 11,842 7.5 57,814 6.9 Stage III 161,418 23.6 39,161 25.0 200,579 23.9 Stage IV 327,215 47.9 72,680 46.3 399,895 47.6 χ 2 p < 0.001

Sample Characteristics Rural patients of all stages ( χ 2 p < 0.001): • More male (57.3% vs. 52.5%) • More white (90.4% vs 79.0%) • Lived in areas of lower median annual income (< $38K; 38.2% vs. 17.6%) • Lived in areas of lower education level (>21% no H.S. diploma; 28.8% vs 17.6%) • More often received cancer care at community facilities (77.5% vs. 66.7%) • Less often had private insurance payor (24.0% vs. 30.1%)

Comorbidities • Measured by Charlson-Deyo Comorbidities Score (CDS) All Stages Stage I Non- Non- Rural Overall Rural Overall Rural Rural CDS % % 58.9 54.8 58.8 52.2 48.5 51.5 0 28.0 31.2 28.1 32.4 34.2 32.8 1 11.4 13.0 11.7 9.5 10.4 9.5 2 3+ 4.0 4.3 4.0 3.6 3.7 3.6 χ 2 p < 0.001 χ 2 p < 0.001

Receipt of Guideline-Concordant Treatment Non-Rural Rural Total n % n % n % Lobectomy 58,522 39.3 11,917 35.8 70,439 38.7 SBRT 16,196 10.9 3,682 11.1 19,878 10.9 No Guideline- Concordant 74,247 49.8 17,669 53.1 91,916 50.4 Treatment Total 148,965 100 33,268 100 182,233 100 χ 2 p < 0.001

Unadjusted Survival Unadjusted Cox PH Model Median OS Non Rural-Rural HR LR p (months) Diff. (months) (rural) 11.24 Non-Rural Stages All 10.18 1.06 1.079 <0.0001 Rural 11.04 Total Non-Rural 61.37 Stage I 50.3 11.07 1.184 <0.0001 Rural Total 59.17 Non-Rural 25.03 Stage II 23.2 1.83 1.069 <0.0001 Rural Total 24.57 12.81 Stage III Non-Rural 11.96 0.85 1.076 <0.0001 Rural Total 12.65 Stage IV 5.22 Non-Rural 4.83 0.39 1.097 <0.0001 Rural T otal 5.16

Figure 3. Unadjusted Survival: Stage I

Multivariable Model: Stage I HR 95% CI p HR 95% CI p Sex (ref: Male) Age 1.035 1.034, 1.036 <0.001 Female 0.715 0.704, 0.726 <0.001 Charlson-Deyo Score (ref: CDS = 0) Guideline-Concordant Treatment (ref: none) 1 1.097 1.079, 1.115 <0.001 Lobectomy 0.482 0.473, 0.490 <0.001 2 1.322 1.292, 1.352 <0.001 SBRT 0.950 0.927, 0.973 <0.001 3+ 1.677 1.620, 1.736 <0.001 Median Annual Income (ref: < $38,000) Insurance Payor (ref: Private) ≥ $38,000 0.935 0.915, 0.956 <0.001 Medicare 1.186 1.160, 1.212 <0.001 Education (ref: ≥ 21% lack HS diploma) Medicaid 1.566 1.504, 1.631 <0.001 < 21% Lack HS Diploma 0.919 0.898, 0.939 <0.001 Other Gov't Insur 1.237 1.159, 1.319 <0.001 Race (ref: White) Uninsured 1.387 1.302, 1.477 <0.001 Black 1.060 1.032, 1.090 <0.001 Location (ref: Northeast) Hispanic 0.794 0.751, 0.838 <0.001 Southeast 1.153 1.127, 1.180 <0.001 Asian 0.704 0.660, 0.751 <0.001 Midwest 1.164 1.137, 1.192 <0.001 Other 0.900 0.844, 0.960 0.001 West 1.117 1.085, 1.149 <0.001 Distance to Treatment Ctr 0.929 0.904, 0.955 <0.001 Pacific 1.058 1.026, 1.091 <0.001 Facility Type (ref: Community) Academic Medical Center 0.890 0.875, 0.906 <0.001 Rurality (ref: non-rural) Rural 1.085 1.062, 1.108 <0.001

Conclusions Rural patients with Stage I NSCLC: 1. Have a survival disparity of 11.07 months compared to non-rural patients 2. Have greater comorbidities at baseline 3. Less often received any form of accepted treatment (neither lobectomy nor SBRT) 4. Less often underwent first-line management with lobectomy 5. Rurality is an independent risk factor for decreased survival in multivariable modeling.

Limitations • Selection bias • NCDB does not include data on history of: • Tobacco use • Alcohol and drug use • Occupational exposures • Diet & exercise • Well-established difficulty in defining rurality • Did not consider adjuvant therapy or sublobar resections or other ablative procedures in stage I (substandard care)

Implications • Increased provision of treatment according to guidelines in rural patients with stage I NSCLC • Rural public health interventions aimed at decreasing burden of chronic health conditions ( ↓ comorbidities) • Next steps: • Large-database multivariable modeling incorporating smoking as a covariate • Temporal analysis of rural stage I survival disparity

Acknowledgements Mentors: • Nataniel H. Lester-Coll, M.D. • Brian L. Sprague, Ph.D.