Racial Disparities and Survival for Nonsmall-Cell Lung Cancer in a Large Cohort of Black and White Elderly Patients

Transcription

1 Racial Disparities and Survival for Nonsmall-Cell Lung Cancer in a Large Cohort of Black and White Elderly Patients Dale Hardy, PhD 1 ; Rui Xia, MS 1 ; Chih-Chin Liu, MS 1 ; Janice N. Cormier, MD, MPH 2 ; Zhannat Nurgalieva, MD, MS 1 ; and Xianglin L. Du, MD, PhD 1,3 BACKGROUND: This study aimed to examine disparities in survival and associated factors for patients with nonsmall-cell lung cancer (NSCLC) and to determine whether racial disparities varied over time ( , , and ). METHODS: The authors studied 70,901 patients aged 65 years with stage I-IV NSCLC identified from Surveillance, Epidemiology, and End Results/Medicare data. Multivariate time-to-event survival analyses were completed using Cox proportional regression modeling. RESULTS: The 5-year observed lung cancer-specific survival rates were 52.7% for whites and 47.5% for blacks with stage I-II disease, and 17.7% and 19.6% for whites and blacks, respectively at stages III-IV. After controlling for standard treatment, socioeconomic status (SES), and other factors, there were no significant differences in all-cause mortality, or lung cancer-specific mortality between black and white patients with stage I-II or III-IV lung cancer. However, blacks had an increased risk for overall all-cause mortality at stage I-IV (hazard ratio [HR], 1.24; 95% confidence interval, ), and during at stage III-IV for all-cause mortality (HR, 1.22; 95% CI, ) and lung cancer-specific mortality (HR, 1.24; 95% CI, ). Standard treatment was significantly associated with increased survival, whereas poor SES was associated with increased mortality. CONCLUSIONS: There were no significant differences in survival between blacks and whites with NSCLC within stage stratifications after adjusting for covariates, except for black patients at overall stage for all-cause mortality and at stage III-IV diagnosed in Receiving stage-specific evidence-based standard therapy was associated with significantly increased survival. Cancer 2009; 115: VC 2009 American Cancer Society. KEY WORDS: nonsmall-cell lung cancer, survival, survival rates, race/ethnicity, standard treatment, health disparity. Lung cancer is the second most commonly diagnosed cancer, and it is the leading cause of cancer-associated mortality in the United States, claiming >158,000 lives in Although overall survival rates Corresponding author: Dale Hardy, PhD, RD CDE, The University of Texas School of Public Health, Division of Epidemiology, 1200 Herman Pressler Drive, RAS-E645, Houston, TX 77030; dale.s.hardy@uth.tmc.edu 1 Division of Epidemiology and Disease Control, The University of Texas School of Public Health, Houston, Texas; 2 Department of Surgical Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, Texas; 3 Center for Health Services Research, The University of Texas School of Public Health, Houston, Texas We acknowledge the creation of this database from the National Cancer Institute, Centers for Medicare and Medicaid Services, and Surveillance, Epidemiology, and End Results tumor registries. The analyses, interpretation, and reporting are the sole responsibilities of the authors. Received: November 19, 2008; Revised: January 28, 2009; Accepted: March 3, 2009 Published online July 22, 2009 in Wiley InterScience ( DOI: /cncr.24521, Cancer October 15,

2 from lung cancer have been increasing steadily over the past decade, 2 disparities in survival outcomes for blacks compared with whites have been noted in the administration of stage-specific treatment modalities for surgery, chemotherapy, and radiation therapy. 3 Surgical resection is the potentially curative treatment for patients with stage I and II lung cancer. 4 Surgical resection followed by cisplatin-based adjuvant chemotherapy for patients with stages I, II, and III nonsmall-cell lung cancer (NSCLC) has been shown to improve survival compared with surgical resection alone. 5-7 Chemoradiation or platinum-based chemotherapy appears to be beneficial in patients with unresectable stage III or stage IV disease. 8 Differences in survival rates among races may be explained by several factors besides disparities in treatment received. 9 Advanced stage of disease at the time of diagnosis, which is a known adverse prognostic factor for survival, is more common in blacks. 2 Overall, the 5-year survival rate for NSCLC is 15.2% within the 17 Surveillance, Epidemiology, and End Result (SEER) areas. However, for the time period between 1992 and 2003, 5-year survival rates stratified by disease stage and race showed 42% versus 50% survival for African Americans compared with whites with localized NSCLC (stages I and II), and 13% versus 16% survival for regional disease (stage III), which represented 38% of cases, and 2% survival for both racial groups with distant disease stage (stage IV), which included 40% of cases diagnosed. Despite advances in therapies, blacks continue to have lower survival rates compared with whites. 2 It is unclear whether disparities in survival for blacks have improved recently compared with whites for those who have received treatment according to stage-specific guidelines. 10 The purpose of this research is to examine survival outcomes for black and white patients with NSCLC to determine whether racial disparities persist after adjusting for established risk factors, and to investigate whether racial disparities in survival have changed over time according to the years of diagnosis: , , and We hypothesized that after adjusting for clinicopathologic factors and confounders, there would be no significant disparities in survival for blacks compared with whites. MATERIALS AND METHODS Data Source This study used the SEER-Medicare linked data for 93% complete linkage records in subjects 65 years diagnosed with NSCLC. The SEER program collects high-quality data on cancer-related variables pertaining to cancer incidence, cancer mortality, and cancer survival in 17 US geographical areas. Medicare is the official federal health insurance for people 65 years. 11 Patients who did not have full coverage for Medicare parts A and B, which pays for hospital expenses and physicians visits, or were members of a health maintenance organization in the year their diagnosis was made were excluded from the study. The University of Texas Health Science Center at Houston approved the study protocol. Study Population The study population included 89,834 Medicare patients 65 years who were diagnosed with American Joint Committee on Cancer stage I-IV NSCLC from January 1991 to December 2002 and resided in 1 of the SEER areas in the United States. Of the total study sample, 75,141 (83.6%) were white, and 7960 (8.9%) were black. Patients in socioeconomic status (SES) groups with missing cases (n ¼ 852) and patients with missing or unknown tumor stage information (n ¼ 11,510) were excluded from the analyses, leaving a total sample of 70,901 patients for the multivariate analyses. Study Variables The outcome variables included vital status and the time-to-event from the date of diagnosis until death, censoring, or last follow-up. Tumor stage was dichotomized by grouping patients with stage I and II disease and those with stage III and IV disease. The exposure variable, race/ethnicity, was designated as whites (non-hispanic whites) and blacks (non-hispanic African Americans and blacks from other countries). The standard therapy variable for the treatment of NSCLC patients was constructed according to the National Comprehensive Cancer Network guidelines 12 to define stage-specific therapy. In brief, surgical resection alone was defined as standard therapy for patients with 4808 Cancer October 15, 2009

3 Racial Disparities and Lung Cancer Survival/Hardy et al stage I disease. Standard therapy for stage II was defined as surgical resection with chemotherapy or chemoradiation. For patients with stage IIIA disease, standard therapy included surgical resection with chemotherapy and radiation therapy or surgical resection and chemoradiation. For patients with stage IIIB and IV disease, standard therapy included chemotherapy and radiation therapy. Surgical resection included SEER codes (10-70, 90) for local surgical excision, partial wedge/segmental resection, partial lobectomy, resection of the whole lung, or at least 1 lobe, but less than the whole lung; or if there were Medicare claims of resection through the International Classification of Disease (ICD)-9 procedure codes ( ), 13 and Current Procedural Terminology (CPT) codes ( , 32657, 32663). 14 Information on chemotherapy identified from Medicare claims are described elsewhere. 15,16 Patients who were treated with chemotherapy were identified using the ICD-9 procedure code 9925 for injection or infusion of cancer chemotherapeutic substance, CPT codes ( , J8510, J8520, J8521, J8530-J8999, J9000-J9999, Q0083-Q0085) for chemotherapy administration, revenue center codes 0331 (chemotherapy injected), 0332 (chemotherapy oral), 0335 (chemotherapy intravenous), and ICD-9 Clinical Modification V codes (V58.1, V66.2, or V67.2) for follow-up examination or care after chemotherapy. Radiation therapy included external beam radiation, radioactive implants, radioisotopes, or other radiation as coded in SEER, 17 or Medicare claims indicated by ICD-9 procedure codes ( ) for inpatient or outpatient radiation therapy facility claim, CPT codes ( or ) for physician or outpatient claim for delivery of radiation treatment or clinical bradytherapy, and revenue codes (0330 or 0333) for radiation therapy. 18 Other covariates considered in adjusted analyses were age grouped in 5-year increments (65-69, 70-74, 75-79, 80-84, 85 years), sex, marital status (married or unmarried), diagnosis years ( , , ), 17 SEER areas, tumor size (<1, 1-2, 2.1-3, 3.1-4, >4 cm), tumor grade (well, moderate, poor, undifferentiated), positive number of nodes (1, 2-3, 4-5, 6-9, 10-51), and histology (nonspecific cancer, squamous cell, adenocarcinoma, bronchus, adenosquamous cell). Comorbidity score (0, 1, 2, 3, 4) was calculated using the Deyo adaptation of the Charlson comorbidity index, 19,20 with several procedure codes that reflected the Romano revision. 21 The percentage of individuals living below the poverty line at the census tract level was used to define SES quartile categories. 22 Statistical Analysis Analyses were conducted using STATA, version 10.1 (STATACORP, College Station, Tex). Descriptive statistics for patients with stage I-II and III-IV disease were performed using Pearson chi-square tests (Table 1). The observed survival rates were calculated for racial groups and for different disease risk factors by stages (Table 2). The P value approach was used in the modelbuilding process in choosing variables for each model at stage I-II and III-IV as described by Klein and Moeschberger. 23 One of the goals in modeling was to develop parsimonious models at each stage of stratification. Univariate and multivariate regression analyses were performed to assess the risk of dying for blacks compared with whites after receiving stage-specific treatment for NSCLC and adjusting for other factors (Table 3). The effects of sex, SES, and receipt of standard therapy were assessed over time. In addition, univariate and multivariate regression analyses were performed to assess the survival of blacks compared with whites within diagnosis years , , and (Table 4). Interactions were assessed for biological plausibility in all models using methods described by Rothman 24 and Andersson et al. 25 Inherent in this process, biologically plausible effect modifiers were identified with the race/ethnicity exposure variable and all-cause/ lung-cancer survival using the likelihood ratio test with a P value cutoff point of.05. Confidence intervals were constructed using the Hosmer and Lemeshow delta method based on Taylor expansion of variances and covariances. 26 These interaction terms were included along with single-order covariates in the model. Sensitivity analyses were performed to estimate how an added adjusted prevalence of cigarette smoking might affect disparities in mortality between blacks and whites (Table 5) using methods described by Lin and colleagues. 27 Cancer October 15,

4 Table 1. Distribution of Patient Characteristics for Nonsmall-Cell Lung Cancer by and Tumor Disease Stage Diagnosed Between 1991 and 2002 and Follow-Up Through 2005 Characteristic Stage I-II % Stage III-IV % Whites, n518,233 Blacks, n51618 White, n546,381 Age, y Sex Men Women SES Highest quartile nd quartile rd quartile Lowest quartile Unknown Married Married Unmarried Unknown Year of diagnosis SEER registry area Connecticut Detroit Hawaii Iowa New Mexico Seattle Utah Atlanta/rural Georgia Kentucky Louisiana New Jersey California Blacks, n55359 Comorbid score (Continued) 4810 Cancer October 15, 2009

5 Racial Disparities and Lung Cancer Survival/Hardy et al Table 1. (Continued) Characteristic Stage I-II % Stage III-IV % Whites, n518,233 Blacks, n51618 White, n546,381 Histology Cancer, nonspecific Large cell Squamous cell Adenocarcinoma Bronchus Adenosquamous cell Tumor grade Well Moderate Poor Undifferentiated Missing/unknown Tumor size, cm < Missing Surgery No Yes Chemotherapy No Yes Radiation therapy No Yes Standard therapy No Yes Blacks, n55359 SES indicates socioeconomic status; SEER, Surveillance, Epidemiology, and End Results. P values were calculated using Pearson chi-square tests of hypothesis for independence to assess statistical significance in relationships between ethnic groups. All variables had significant P values <.001, except sex at stage I-II and radiation therapy stage III-IV; P value <.05. Year of diagnosis and chemotherapy at stage I-II, and year of diagnosis stage III-IV, P value >.05. RESULTS Descriptive Patient Characteristics by Ethnicity The initial study cohort consisted of 83,101 NSCLC patients, of whom 75,141 (90.4%) were whites and 7960 (9.6%) were blacks. The mean age at diagnosis was 73.5 years for blacks and 74.8 years for whites. The median overall survival was 6.48 months for whites (95% confidence interval [CI], ) and 5.95 months for blacks (95% CI, ). The median lung cancer-free survival was the same for both races; months (95% CI, ) for whites and months (95% CI, ) for blacks. Table 1 shows the distribution of various patient characteristics stratified by black versus white and disease Cancer October 15,

6 Table 2. Observed Survival Rates for Blacks and Whites With Nonsmall-Cell Lung Cancer Diagnosed Between 1991 and 2002 With Follow-Up Through 2005 Observed Survival Rates 1 Year 3 Year 5 Year All Causes Lung Cancer All Cause Lung Cancer All Causes Lung Cancer Stage I-II White Black Sex Men White Black Women White Black White Black White Black White Black Std therapy White Black No Std therapy White Black Stage III-IV White Black Sex Male White Black Female White Black White Black White Black White Black Std therapy White Black No Std therapy White Black Std indicates standard Cancer October 15, 2009

7 Racial Disparities and Lung Cancer Survival/Hardy et al Table 3. Effect of Various Factors on in Patients With Nonsmall-Cell Lung Cancer Diagnosed Between 1991 and 2002 All-Cause, n519,519 Hazard Ratios and 95% Confidence Intervals Stage I-II Stage III-IV Lung Cancer-Specific, n519,393 All-Cause, n580,519 Lung Cancer-Specific, n547,521 Adjusted model White Black 1.22 ( ) 1.00 ( ) 1.07 ( ) 1.04 ( ) Sex Men Women 0.78 ( )* 0.89 ( )* 0.92 ( )* 0.95 ( )* SES Highest quartile nd quartile 1.08 ( ) 1.01 ( ) 1.03 ( )* 1.03 ( ) 3rd quartile 1.23 ( )* 1.01 ( ) 1.05 ( )* 1.05 ( )* Lowest quartile 1.38 ( )* 0.99 ( ) 1.09 ( )* 1.04 ( )* Std therapy No Yes 0.50 ( )* 0.54 ( )* 0.70 ( )* 0.93 ( )* Crude model White Black 1.20 ( )* 1.15 ( )* 1.04 ( )* 0.95 ( )* SES indicates socioeconomic status; Std, standard. Crude models represent unadjusted models. Lung cancer-specific mortality (LCM) adjusted model at stage I-II included race, sex, SES, standard therapy, tumor size, number of nodes, stratified diagnosis years ( , , ), comorbidity score; stage III-IV adjusted model included LCM stage I-II adjusted model plus race-number of nodes interaction. All-cause mortality model at stage I-II included LCM stage I-II adjusted model plus age, histology, marital status, Surveillance, Epidemiology, and End Results areas, and race-ses interaction; stage III-IV model included LCM stage I-II adjusted model plus race-number of nodes interaction. * Significant hazard ratios. stage (I-II and III-IV). Slightly higher percentage of whites with NSCLC were diagnosed with early cancers (24.3% vs 20.4%), whereas blacks were diagnosed with more advanced stage disease (67.4% vs 61.8%). A greater proportion of blacks were clustered in the lowest (poorest) SES quartile compared with whites. More blacks were unmarried at the time of diagnosis. More whites had a comorbidity score of <1 compared with blacks, who tended to have comorbidity scores of 4. The majority of blacks and whites were diagnosed with squamous cell cancer and adenocarcinoma, but more blacks than whites were diagnosed with unspecified stage III-IV disease. A greater percentage of whites compared with blacks (53.6% vs 41.8%) received standard therapy including surgical resection for stage I-II disease compared with blacks, who were more likely to receive chemotherapy and radiation therapy for similar stage I-II disease. Basic characteristics of persons with missing stage information were similar to results in Table 1, except for surgery and chemotherapy given, in which there were no significant differences between racial groups. In addition, we compared the patients included and those excluded because of their health maintenance organization or because they did not have full coverage for Medicare parts A and B (n ¼ 32,571) in terms of age, sex, race, tumor stage, tumor grade, year of diagnosis, and SEER areas, and found these 2 groups had similar distribution of these factors. For example, the percentages of several factors between the 2 groups included blacks (8.9% vs 8.8%), women (43.8% vs 42.4%), tumor stage IV (33.8% vs 34.6%), and poor tumor grade (29.5% vs 30.1%). Cancer October 15,

8 Table 4. Effect of Various Factors on in Patients With Nonsmall-Cell Lung Cancer Stratified by Diagnosis Years, All-Cause Hazard Ratios and 95% Confidence Intervals Lung Cancer-Specific All-Cause Lung Cancer-Specific All-Cause Lung Cancer-Specific Stage I-II Adjusted model Whites Blacks 1.11 ( ) 0.97 ( ) 1.28 ( ) 1.10 ( ) 1.34 ( ) 0.97 ( ) Sex Men Women 0.77 ( )* 0.90 ( )* 0.78 ( )* 0.91 ( )* 0.78 ( )* 0.86 ( )* SES Highest quartile nd quartile 1.05 ( ) 1.02 ( ) 1.03 ( ) 0.88 ( )* 1.16 ( )* 1.13 ( )* 3rd quartile 1.24 ( ) 1.06 ( ) 1.17 ( ) 0.91 ( ) 1.30 ( )* 1.05 ( ) Lowest quartile 1.27 ( ) 0.98 ( ) 1.41 ( ) 0.91 ( ) 1.53 ( )* 1.08 ( ) Std therapy No Yes 0.50 ( )* 0.58 ( )* 0.52 ( )* 0.58 ( )* 0.48 ( )* 0.46 ( )* Crude model White Black 1.17 ( )* 1.07 ( ) 1.24 ( )* 1.18 ( )* 1.19 ( )* 1.19 ( )* Stage III-IV Adjusted model Whites Blacks 0.96 ( ) 0.87 ( ) 0.97 ( ) 0.97 ( ) 1.22 ( )* 1.24 ( )* Sex Men Women 0.91 ( )* 0.99 ( ) 0.91 ( )* 0.94 ( )* 0.93 ( )* 0.94 ( )* SES Highest quartile nd quartile 1.07 ( )* 1.06 ( )* 1.04 ( ) 1.04 ( ) 1.00 ( ) 0.99 ( ) 3rd quartile 1.07 ( )* 1.07 ( )* 1.04 ( ) 1.06 ( )* 1.03 ( ) 1.03 ( ) Lowest quartile 1.10 ( )* 1.05 ( ) 1.08 ( )* 1.09 ( )* 1.06 ( )* 1.00 ( ) Std therapy No Yes 0.71 ( )* 0.93 ( )* 0.69 ( )* 0.94 ( )* 0.69 ( )* 0.91 ( )* Crude model White Black 0.98 ( )* 0.93 ( )* 1.00 ( ) 0.92 ( )* 1.07 ( )* 1.00 ( )* SES indicates socioeconomic status; Std, standard. Refer to Table 3 for variables that were adjusted for in analyses, except for stratified years of diagnosis. * Significant hazard ratios (P <.05). Observed 1-Year, 3-Year, and 5-Year Survival Rates The observed overall and lung cancer-free unadjusted survival rates are presented in Table 2. Whites appear to have longer survival compared with blacks, even when stratified by year of diagnosis. Both whites and blacks were noted to have improved lung cancer-free survival over time (54.5% to 60.6%) compared with blacks (50.4% to 54.4%), but observed survival rates were higher in white women, who tended to survive longer than men. When standard 4814 Cancer October 15, 2009

9 Racial Disparities and Lung Cancer Survival/Hardy et al Table 5. Sensitivity Analysis of the Potential Impact of Smoking Status on Racial Disparities in Survival Smoking Prevalence, % Hazard Ratio for Due to Smoking Hazard Ratios and 95% Confidence Intervals Stage I-II Stage III-IV Blacks Whites All-Cause Lung Cancer-Specific All-Cause Lung Cancer ( ) 1.00 ( ) 1.07 ( ) 1.04 ( ) ( ) 0.94 ( ) 1.01 ( ) 0.98 ( ) ( ) 0.89 ( ) 0.96 ( ) 0.93 ( ) ( ) 0.84 ( ) 0.91 ( ) 0.88 ( ) ( ) 0.80 ( ) 0.86 ( ) 0.84 ( ) ( ) 1.00 ( ) 1.07 ( ) 1.04 ( ) ( ) 1.02 ( ) 1.10 ( ) 1.06 ( ) ( )* 1.04 ( ) 1.12 ( ) 1.09 ( ) ( )* 1.07 ( ) 1.15 ( )* 1.11 ( ) ( )* 1.09 ( )* 1.17 ( )* 1.14 ( ) ( ) 1.00 ( ) 1.07 ( ) 1.04 ( ) ( )* 1.04 ( ) 1.12 ( ) 1.08 ( ) ( )* 1.08 ( ) 1.17 ( )* 1.13 ( ) ( )* 1.13 ( )* 1.21 ( )* 1.18 ( )* ( )* 1.17 ( )* 1.26 ( )* 1.22 ( )* ( ) 1.00 ( ) 1.07 ( ) 1.04 ( ) ( )* 1.06 ( ) 1.14 ( )* 1.10 ( ) ( )* 1.12 ( )* 1.20 ( )* 1.17 ( )* ( )* 1.18 ( )* 1.27 ( )* 1.23 ( )* ( )* 1.24 ( )* 1.33 ( )* 1.29 ( )* * Significant hazard ratios. therapy was administered, both blacks and whites were noted to have comparable stage-specific survival. Unadjusted and Adjusted Hazard Ratios In crude models (Table 3) for both all-cause mortality and lung cancer-specific mortality, blacks appeared to have an increased risk of dying from lung cancer compared with whites, except for lung cancer-specific mortality in those with stage III-IV disease, in whom there was a slightly lower risk (hazard ratio [HR], 0.95; 95% CI, ). However, after adjustment for confounders and standard treatment, there were no differences in survival between the 2 races. Further assessment of other variables revealed that there was an increased survival for women compared with men across stage I-IV disease for both all-cause mortality and lung cancer-specific mortality. Patients at the poorest SES quartile had increased risks of all-cause mortality and lung cancer-specific mortality compared with patients in the highest SES quartile. Patients who received standard therapy had significantly lower risk for all-cause and lung cancer-specific mortality. In separate models, there were no significant differences in survival for overall mortality for stage I-IV in lung cancer (1.09, ); however, blacks had an increased risk (1.24, ) of dying from all-cause mortality after adjusting for confounders. Stratified by Years of Diagnosis Table 4 presents the risk of mortality for blacks compared with whites, stratified by year of diagnosis. There was no significant difference in all-cause or lung cancer-specific mortality between blacks and whites across the 3 time periods of diagnosis, except during the period between 2000 and 2002, during which blacks with stage III-IV disease had an increased risk of all-cause mortality (1.22, ) and lung cancer-specific mortality (1.24, ). Women appeared to have lower risk of mortality than men across these time periods. Overall in these 3 time periods, patients living in the poorest quartile of SES had higher risks of all-cause mortality when compared Cancer October 15,

10 with those in the top quartile of SES. Patients who received standard therapy were at a significantly lower risk of all-cause or lung cancer-specific mortality than those who did not receive standard therapy regardless of the year of diagnosis or tumor stage. Significant interaction terms used in model adjustment for stage I-II were race with number of nodes for stage III-IV for all-cause mortality and lung cancer-specific mortality and race with SES level for stage I-II for allcause mortality. There was no interaction term used in the model for stage I-II for lung cancer-specific mortality. Sensitivity Analysis for Smoking Smoking has been identified as a strong risk factor not only for lung cancer occurrence but also for treatment benefits and lung cancer survival. 28 Because of a lack of smoking data in our study, we present Table 5 as a sensitivity analysis to estimate the potential impact of smoking rates on disparities in survival between blacks and whites. Smoking prevalence was estimated as 21.9% in white adults and 23.0% in black adults according to reports by the Centers for Disease Control and Prevention, 29 and a HR of 1.55 was estimated to be associated with mortality because of smoking. 30 Sensitivity analyses were performed by varying estimates for smoking between the 2 groups to assess how sensitive survival disparities between blacks and whites are to changes in smoking prevalence. Overall, the study results for both all-cause and lung cancer-specific mortality were not found to be sensitive to the changes in smoking prevalence between blacks and whites. The risk of mortality between blacks and whites became significant only when the smoking prevalence was in most scenarios 10% higher in whites than in blacks and the HR for smoking was at least DISCUSSION This study examined disparities in survival between blacks and whites with stage I-IV NSCLC. We found that there were no differences in survival for blacks compared with whites when stages were stratified into stage I-II and III- IV, after adjusting for standard treatment and other confounders. However, when looking at overall survival for the whole group for stages I-IV, blacks had an increased risk for all-cause mortality. Blacks with stage III-IV disease were also noted to have an increased risk of mortality during the time period Receipt of standard treatment was associated with increased all-cause and lung cancer-specific survival across various stages. Women had a lower risk of mortality compared with men. Patients in the lowest quartile of SES had an increased risk of allcause mortality at stages I-IV and lung cancer-specific mortality at stage III-IV disease. These study results are consistent with findings from previously published studies. 31,32 Lathan et al 31 found no significant difference in survival between black and white patients who underwent surgery for stage I-III NSCLC. Greenwald et al 32 found decreased survival for blacks with early stage NSCLC in age- and sex-adjusted models; however, when model adjustment included income, there was a slightly lower risk of mortality for blacks (risk ratio, 0.96; P <.0001). Bryant and Cerfolio 33 also noted no significant differences in survival rates for black and white NSCLC patients after matching on several factors and adjusting for lifestyle and treatment factors. Temporal trends in survival were also analyzed by Owonikoko et al 34 for the time periods and These authors found better survival with treatment for the latter stratification years for all age groups in the SEER elderly population of blacks, whites, and patients of other races, with younger ages having the best outcome. There was a worsening outcome in patients who did not receive surgery or radiation during Our study found a disparity in survival at stage III-IV for , and as a whole for stages I-IV for all-cause mortality for blacks compared with whites. The latter suggests that blacks are more likely to die from other causes, indicating disparities in general health. Those patients who received standard therapy had increased survival. There are several strengths in the current study. The study includes a large population-based cohort of patients with up to 16 years of follow-up, selected from SEER areas that account for about 26% of the US population. SEER data not only provide reliable information on tumor stage, grade, and demographics, but also include information on long-term follow-up of vital status. Medicare claims allowed us to calculate a patient comorbidity score, which is known to be a potential strong confounder of mortality. In addition, the receipt of chemotherapy can be identified from Medicare claims data, 15 so that standard therapy for stage I-II disease (chemotherapy as 4816 Cancer October 15, 2009

11 Racial Disparities and Lung Cancer Survival/Hardy et al adjuvant therapy) and for stage III-IV disease (chemotherapy as primary or palliative treatment) can reasonably be defined. In addition, we were able to compare survival of blacks with whites for stage-specific mortality over time from 1991 to Several limitations should be noted in this study. First, there was a lack of information on smoking, which is known to be an important risk factor for lung cancer survival. 28,30,33 It is also unknown if the prevalence of smoking differed between black and white patients. However, the sensitivity analysis to evaluate smoking as an unmeasured confounder showed that the study results were not sensitive to changes of smoking prevalence among the 2 populations. For example, racial disparities in survival would be affected only when there is a 10% difference in smoking prevalence between blacks and whites, with a higher prevalence in whites, and the HR of mortality because of smoking is >1.50. Our study groups are population-based cohorts, and because blacks were reported to have a higher prevalence of smoking than whites (23.0% vs 21.9% as reported by the Centers for Disease Control and Prevention), 29 it is unlikely that the magnitude and direction would change substantially if smoking was included in the analysis adjustment. Second, other variables that have been previously identified to be associated with lung cancer survival were not included in the current analysis, including hospital volume characteristics 35 and the presence of lung cancer symptoms, which is said to be an important predictor of survival in African Americans. 36 In addition, there were a large number of cases with missing values, and those who belonged to a health maintenance organization and/or were without Medicare parts A and B were excluded from the study. Although it is not known how these patients fared in their disease outcome, the baseline characteristics were mostly similar to patients included in the study, reducing the possibility of selection bias in affecting study findings. Furthermore, our population was limited to persons 65years old, so the current findings cannot be generalized to persons <65 years of age. In conclusion, there were no significant differences in survival between blacks and whites with NSCLC within stage stratifications after adjusting for covariates, except for black patients at overall stage for all-cause mortality and at stage III-IV disease diagnosed in These findings were not sensitive to changes in unbalanced smoking prevalence for blacks and whites. Poor SES was independently associated with higher mortality even in this population of patients covered by the Medicare healthcare program. These findings support the importance of receiving the stage-specific evidence-based standard therapy, which played an important role in reducing survival disparities. Conflict of Interest Disclosures This study was supported by a grant from the Agency for Healthcare Research and Quality (RO1-HS016743). References 1. US Cancer Statistics Working group. United States Cancer Statistics: 2004 Incidence and. Atlanta, GA): Department of Health and Human Services, Centers for Disease Control and Prevention, and National Cancer Institute; Available at: statistics/ Accessed May 12, American Cancer Society. Cancer Facts and Figures for African Americans Atlanta, GA: American Cancer Society; Available at: downloads/stt/caff2007aaacspdf2007.pdf Accessed May 12, National Cancer Institute. SEER Survival Rates for Lung Cancer. Available at: lungb.html?statfacts_page¼lungb.html&x¼16&y¼13 Accessed May 16, American Cancer Society. Cancer Statistics Atlanta, GA: American Cancer Society; Arriagada R, Bergman B, Dunant A, Le Chevalier T, Pignon JP, Vansteenkiste J; International Adjuvant Lung Cancer Trial Collaborative Group. Cisplatin-based adjuvant chemotherapy in patients with completely resected nonsmall-cell lung cancer. N Engl J Med. 2004;350: Douillard JY, Rosell R, De Lena M, et al. Adjuvant vinorelbine plus cisplatin versus observation in patients with completely resected stage IB-IIIA nonsmall-cell lung cancer (Adjuvant Navelbine International Trialist Association [ANITA]): a randomised controlled trial. Lancet Oncol. 2006;7: Winton T, Livingston R, Johnson D, et al; National Cancer Institute of Canada Clinical Trials Group; National Cancer Institute of the United States Intergroup JBR.10 Trial Investigators. Vinorelbine plus cisplatin vs. observation in resected nonsmall-cell lung cancer. N Engl J Med. 2005;352: Souquet PJ, Chauvin F, Boissel JP, Bernard JP. Meta-analysis of randomised trials of systemic chemotherapy versus supportive treatment in non-resectable nonsmall cell lung cancer. Lung Cancer. 1995; 12 suppl 1:S147-S154. Cancer October 15,

12 9. Jemal A, Clegg LX, Ward E, et al. Annual report to the nation on the status of cancer, , with special feature regarding survival. Cancer. 2004;101: Bach PB, Schrag D, Brawley OW, Galaznik A, Yakren S, Begg CB. Survival of blacks and whites after a cancer diagnosis. JAMA. 2002;287: National Cancer Institute. Surveillance, Epidemiology, and End Results Program: Overview of the SEER Program. Available at: Accessed May 20, National Comprehensive Cancer Network. Nonsmall cell lung cancer. Available at: Accessed May 19, US Health Care Financing Administration, Practice Management Information Corporation. HCPCS 1994 National Level II Medicare Codes: HCFA Common Procedure Coding System. Los Angeles, CA: Practice Management Information Corp.; World Health Organization, WHO Collaborating Centers for Classification of Diseases. International Statistical Classification of Diseases and Related Health Problems. 10th rev ed. Geneva, Switzerland: World Health Organization; Du X, Goodwin JS. Patterns of use of chemotherapy for breast cancer in older women: findings from Medicare claims data. J Clin Oncol. 2001;19: Warren JL, Harlan LC, Fahey A, et al. Utility of the SEER-Medicare data to identify chemotherapy use. Med Care. 2002;40(8 suppl):iv-55-iv Cunningham J,National Cancer Institute. The SEER Program Code Manual. Rev ed. Bethesda, MD: Cancer Statistic Branch, National Cancer Institute, National Institute of Health; Du X, Freeman JL, Goodwin JS. Information on radiation treatment in patients with breast cancer: the advantages of the linked Medicare and SEER data. Surveillance, Epidemiology and End Results. J Clin Epidemiol. 1999;52: Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis. 1987;40: Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45: Romano PS, Roos LL, Jollis JG. Adapting a clinical comorbidity index for use with ICD-9-CM administrative data: differing perspectives. J Clin Epidemiol. 1993;46: Du XL, Fang S, Vernon SW, et al. Racial disparities and socioeconomic status in association with survival in a large population-based cohort of elderly patients with colon cancer. Cancer. 2007;110: Klein JP, Moeschberger ML. Survival Analysis: Techniques for Censored and Truncated Data. 2nd ed. New York, NY: Springer; Rothman KJ. Epidemiology: An Introduction. New York, NY: Oxford University Press; Andersson T, Alfredsson L, Kallberg H, Zdravkovic S, Ahlbom A. Calculating measures of biological interaction. Eur J Epidemiol. 2005;20: Hosmer DW, Lemeshow S. Confidence interval estimation of interaction. Epidemiology. 1992;3: Lin DY, Psaty BM, Kronmal RA. Assessing the sensitivity of regression results to unmeasured confounders in observational studies. Biometrics. 1998;54: Tammemagi CM, Neslund-Dudas C, Simoff M, Kvale P. Smoking and lung cancer survival: the role of comorbidity and treatment. Chest. 2004;125: Adult Cigarette Smoking in the United States. Centers for Disease Control, Atlanta, GA. May 29, Available at: cig_smoking/index.htm Accessed on June 30, Tsao AS, Liu D, Lee JJ, Spitz M, Hong WK. Smoking affects treatment outcome in patients with advanced nonsmall cell lung cancer. Cancer. 2006;106: Lathan CS, Neville BA, Earle CC. The effect of race on invasive staging and surgery in nonsmall-cell lung cancer. J Clin Oncol. 2006;24: Greenwald HP, Polissar NL, Borgatta EF, McCorkle R, Goodman G. Social factors, treatment, and survival in earlystage nonsmall cell lung cancer. Am J Public Health. 1998; 88: Bryant AS, Cerfolio RJ. Impact of race on outcomes of patients with nonsmall cell lung cancer. J Thorac Oncol. 2008;3: Owonikoko TK, Ragin CC, Belani CP, et al. Lung cancer in elderly patients: an analysis of the Surveillance, Epidemiology, and End Results database. J Clin Oncol. 2007; 25: Neighbors CJ, Rogers ML, Shenassa ED, Sciamanna CN, Clark MA, Novak SP. Ethnic/racial disparities in hospital procedure volume for lung resection for lung cancer. Med Care. 2007;45: Tammemagi CM, Neslund-Dudas C, Simoff M, Kvale P. Lung carcinoma symptoms an independent predictor of survival and an important mediator of African American disparity in survival. Cancer. 2004;101: Cancer October 15, 2009