Published ahead of print on February 25, 2005, doi:10.1164/rccm.200411-1475OC
© 2005 American Thoracic Society doi: 10.1164/rccm.200411-1475OC
Ethnic Disparities in the Treatment of Stage I Non–Small Cell Lung CancerDivisions of General Internal Medicine and Pulmonary, Critical Care Medicine, and Sleep Medicine, and Department of Health Policy, Mount Sinai School of Medicine; and Department of Radiology, New York-Presbyterian Hospital/Weill Cornell Medical Center, New York, New York Correspondence and requests for reprints should be addressed to Juan P. Wisnivesky, M.D., M.P.H., Department of Medicine, Mount Sinai School of Medicine, One Gustave L. Levy Place, Box 1087, New York, NY 10029. E-mail: juan.wisnivesky{at}mssm.edu
Rationale: Important variations exist in the treatment of non–small cell lung cancer. Because resection is the most effective treatment for patients with early disease, disparities in surgical rates can generate considerable differences in outcomes. Objective: We analyzed data from a national population-based registry to evaluate disparities in the treatment of Hispanic and white patients with stage I lung cancer and to assess the extent to which these inequalities explain survival differences. Methods: This study included 16,036 Hispanic and white patients with stage I lung cancer diagnosed between 1991 and 2000. Cases were identified from the Surveillance, Epidemiology, and End Results registry. Survival was compared among white and Hispanics using Kaplan-Meier curves. Stratified survival curves and Cox regression were used to evaluate whether inequalities in stage (IA vs. IB) and resection could explain survival differences. Results: Hispanics had worse overall and lung cancer–specific survival compared with whites (p = 0.04 and 0.008, respectively). Five-year lung cancer survival was 54% for Hispanics versus 62% for whites. Hispanics were more frequently diagnosed with stage IB disease (p = 0.0002) and less likely to undergo resection (p = 0.03). Among resected patients, survival was similar for the two groups, as it was among those who did not undergo unresection. After adjusting for surgery and stage, there was no difference in survival between groups. Conclusions: Hispanics with stage I lung cancer had worse survival as compared with whites. These disparities are largely explained by lower rates of resection and higher probability of diagnosis at stage IB. Future work must delineate why Hispanics are receiving less surgery.
Key Words: disparities • ethnicity • lung cancer • treatment Lung cancer is the leading cause of cancer death among Hispanic men and the second leading cause of death among Hispanic women (1). Meaningful long-term survival, unfortunately, is only possible for non–small cell lung cancer cases diagnosed at an early stage and treated by surgical resection (2). Patients with surgically resected stage I disease have at least a 65% probability of surviving 5 years or longer (3–5). Conversely, patients with early stages of lung cancer who do not undergo resection have a median survival of less than 2 years (6). Variations in lung cancer treatment and survival because of racial and social factors have been reported in several studies. Greenwald and coworkers (7) found that, among patients with stage I lung cancer, African Americans and those with lower socioeconomic status were less likely to undergo surgery and less likely to attain 5-year survival. Bach and colleagues (8) reported lower survival rate among African-American patients with stage I and II lung cancer and found that survival differences were explained by lower rates of surgical treatment among blacks. Although Hispanics have a lower incidence of lung cancer than whites, they are more likely to be diagnosed at an advanced stage, and Hispanic males tend to have worse survival compared with whites (1, 9). Samet and colleagues (10) found that older age and Hispanic ancestry were associated with lower rates of surgical treatment in a cohort in New Mexico. Although differences between African Americans and whites in resection rates and lung cancer survival have been widely reported, the extent of disparities among Hispanics, the largest ethnic minority in the United States, and the reason for these disparities remain incompletely understood. We used nationally representative, population-based cancer data to assess whether there are disparities in survival between Hispanics and white patients with stage I non–small cell lung cancer. We were particularly interested to examine whether apparent disparities in survival could be explained by differences in the rates of surgical resection and stage distribution at diagnosis.
Cases were selected for the Surveillance, Epidemiology, and End Results (SEER) program, a national database that collects data on all incident cancer cases in selected areas of the United States (11, 12). From the SEER registry 2003, we identified all cases of non–small cell lung cancer diagnosed during the last decade reported (1991–2000). Among these subjects, we identified 19,160 cases of stage I cancer according to the American Joint Committee on Cancer criteria (13). Information regarding the stage and racial and ethnic distribution of individuals in the SEER registry has been published (1, 11). Additional details on the method for selecting the study population and the statistical analysis are provided in an online supplement. From this group, we identified 16,036 subjects classified as white or Hispanic. Ethnic data in SEER are coded in accordance with the Office of Management and Budget directive (14), which defines ethnicity as "Hispanic origin" or "not of Hispanic origin." Individuals were classified as white if their race was coded as white and their ethnicity was not designated as Hispanic. We excluded non-Hispanic patients whose race was recorded as African American, American Indian, Asian/Pacific Islander, or unknown. Therefore, comparisons are made between Hispanics and non-Hispanic whites, henceforth called "whites." Cases were classified as resected if the SEER site-specific variable indicated that a surgical procedure (local resection, segmentectomy, wedge resection, lobectomy, and partial or total pneumonectomy) had been performed. Information about age at diagnosis, sex, race, marital status, and cause of death was obtained from SEER. Because SEER does not have information on income, we linked the registry with annual per capita income county-level data from the Area Resource File and approach used previously (15, 16).
Statistical Analysis The Kaplan-Meier method was used to estimate survival rates (17). First, we compared the overall (all-cause mortality) survival for Hispanics and whites. Because the purpose of the study was to evaluate disparities in the treatment of lung cancer, we used lung cancer–specific mortality for all other comparisons because it allows for controlling for unrelated causes of death. To estimate cancer-specific survival, deaths attributed to causes other than lung cancer were censored at the date of death. To evaluate whether differences in the rate of surgical treatment between Hispanics and whites were attributable to unbalances in the distribution of coexisting illnesses, survival curves were constructed considering only deaths from causes other than lung cancer as an event (non–lung cancer survival). When stratifying by ethnicity, these survival curves represent the overall burden of comorbid conditions influencing survival in an ethnic group. Adjusted associations between ethnicity and mortality were evaluated using the Cox regression (18). The effect of Hispanic ethnicity on survival was estimated adjusting for age, sex, marital status, and estimated per capita income. The analysis was repeated by controlling for surgical treatment and stage at diagnosis to assess whether the effect of ethnicity on lung cancer survival persisted after adjusting for unbalances in these factors. Analyses were performed using the SAS 9.0 (SAS, Cary, NC) statistical package.
Patient Characteristics A total of 16,036 eligible patients with stage I primary non–small cell lung cancer were identified from SEER. Of these patients, 686 (4.3%) were Hispanics and 15,350 (95.7%) white. The clinicopathologic characteristics of these patients are reported in Table 1. There were no significant differences between Hispanic and white patients regarding sex, marital status, or the histologic distribution of tumors. Hispanic patients were younger (p = 0.03) and more likely to be diagnosed with stage IB tumors (p = 0.0002) than whites.
Unadjusted Survival and Association with Ethnicity All-cause mortality as well as lung cancer–specific survival was significantly worse in Hispanics compared with whites (p = 0.04 and 0.008, respectively). The lung cancer–specific 5-year survival rate for Hispanics was 54.2% (95% confidence interval, 49.4–59.1%) compared with 62.2% (95% confidence interval, 61.3–63.2%) for whites (Figure 1). Whites were more likely to undergo surgical resection compared with Hispanics (86 vs. 83%, p = 0.03; Table 1). Among patients who had surgical resection, whites had lung cancer–specific survival rates that were not statistically better than Hispanics (p = 0.12). Those who did not undergo surgery had similar lung cancer survival rates (p = 0.52), regardless of their ethnicity (Figure 2).
Figure 3 shows the Kaplan-Meier curve for causes of deaths other than lung cancer. Non–lung cancer survival for Hispanics and whites was not significantly different (p = 0.8). As expected, resected cases had better non–lung cancer survival rates than those who did not undergo surgery (p < 0.0001). White and Hispanic patients who underwent resection had similar survival from causes other than lung cancer (p = 0.5). Among patients who did not undergo surgical resection, however, Hispanics had a significantly better non–lung cancer survival than whites (p = 0.009), suggesting that unresected Hispanic patients had overall lower rates of serious coexistent diseases or other favorable characteristics compared with unresected whites.
Adjusted Survival Analysis We performed Cox regression analysis to evaluate if survival differences among whites and Hispanics persisted after adjusting for age, sex, marital status, and estimated per capita income, and to explore the degree to which the association between Hispanic ethnicity and cancer-specific survival can be explained by disparities in the rate of surgery or stage at diagnosis. As seen in Table 2, in the proportional hazards model adjusted for age, sex, marital status, and mean income in the geographic area (Model 1), Hispanics had significantly increased hazard of lung cancer death (hazard ratio, 1.23; 95% confidence interval, 1.08–1.41). After controlling for surgical treatment received and stage at diagnosis (IA vs. IB; Model 2), Hispanic ethnicity was no longer associated with increased risk of lung cancer death (hazard ratio, 1.09; 95% confidence interval, 0.95–1.24), suggesting that differences in resection rates and stage likely explain the unadjusted relationship. Older age, having a stage IB cancer, and lower mean per capita income were significantly associated with increased risk of lung cancer death. As expected, surgical resection significantly reduced the risk of lung cancer death (hazard ratio, 0.22; 95% confidence interval, 0.21–0.24).
In this national population-based study of 16,036 patients, Hispanics with potentially curable stage I lung cancer had worse all-cause mortality and lung cancer–specific survival compared with whites. Hispanics were less likely to undergo surgical resection, and they presented with stage IB disease more frequently than whites. The observed difference in resection rates does not appear to be explained by the presence or absence of serious coexisting illnesses. Among patients who underwent surgical treatment, survival was similar for the two ethnic groups, as it was among unresected patients. Both groups benefited equally from appropriate surgery, suggesting that the observed ethnic disparity in survival is not explained by differences in the biological aggressiveness of lung cancer or other unmeasured biological determinants of survival. Our multivariate analyses suggest that most of the unadjusted difference in survival can be explained by unequal rates of surgical resection and differences in the proportion of patients presenting with stage IB disease. Several recent studies have demonstrated disparities in the care of African Americans with non–small cell lung cancer (7, 8, 19). Much less is known about inequalities in the treatment of Hispanics diagnosed with lung cancer. One study focusing on variations of cancer therapy related to age found that older Hispanics in New Mexico were less likely to receive surgery (10). This study, however, was based on a regional tumor registry and did not evaluate differences in survival. The observed differences in the rates of surgery by ethnicity is concerning and may be from multiple factors. Treatment decisions may relate to patient-related factors (e.g., insurance status, health beliefs, risk aversion, and cultural issues, among others) or physician factors, or be modified by the interaction between patients and physicians (patient–physician communication) (20–22). As with other diseases (23–25), more in-depth, prospective studies will be needed to evaluate the various factors that may explain these disparities in the treatment and delays in diagnosis. Identifying these factors and determining their relative importance can guide future interventions to eliminate disparities among white and Hispanic patients with lung cancer. Several strengths and limitations should be noted. The generalizability of our findings should be strong. An advantage of using the SEER registry is that it contains population-based data and therefore is less affected by referral patterns and other sources of bias that might be associated with hospital-based case series. Levels of ascertainment within participating areas have been reported to be as high as 98%, showing that most eligible cases are captured in the registry. Geographic areas included in the SEER registry encompass various levels of urbanization of a large percentage of the U.S. population, thus providing an insight at the national level. Given the different regions included in SEER, the registry should be representative of most of the Hispanic subgroups in the United States. Long-term follow-up information available in SEER allowed us to demonstrate disparities in survival for up to 10 years, showing that these differences are not transient trends. In SEER registries, Hispanic ethnicity is coded according to medical records or through a match to a Spanish surname list. This method may be less accurate than using self-reported ethnicity and may result in undercounting Hispanics (26). Misclassification of Hispanics as whites, however, should bias our results toward the null. In addition, the approach to classifying ethnicity that we used has been applied in several published studies using SEER data to evaluate the outcomes of Hispanics and/or assess disparities in care according to ethnicity (1, 9, 27–29). Information regarding the cause of death for cases in the SEER registry is abstracted from death certificates. Although the death certificate is an important source of data on disease incidence, prevalence, and mortality, inaccuracies in the reported cause of death in this document have been described (30, 31). For lung cancer, however, the underlying cause of death has been found to be more than 90% accurate in a large registry (32). In addition, estimates of lung cancer survival rates using SEER are similar to those reported in other studies using hospital-based cases series (13). No individual-level data were available on income or insurance status, which may vary with ethnicity and could affect access to and quality of health care. We used an aggregate measure of income as a surrogate for socioeconomic status of each patient. Our measure, however, was collected at the county level and therefore is probably a suboptimal marker of the patient's socioeconomic status. However, all the patients in the study had the diagnosis of non–small cell lung cancer and the stage of disease established, which meant that they had extensive involvement with the health care system. Thus, these differences are not related to undiagnosed disease in patients without access to care. In addition, no data on comorbidity are available in the SEER database. However, the similar rates of non–lung cancer survival between Hispanics and whites up to 10 years after diagnosis suggest that they had similar rates of serious comorbid conditions. If anything, unresected Hispanics had better noncancer survival than unresected whites, suggesting that they may have had fewer serious comorbid conditions. This approach, however, should be validated. A potential advantage of this approach is that it measures a hard outcome, such as mortality resulting from serious comorbid conditions, rather than relying, for example, on a comorbidity index that predicts 1-year mortality. In addition, survival data were nearly 100% complete in SEER, whereas only 25% of the patients in the study published by Bach and colleagues (8) had enough data available to calculate a comorbidity index. Although recent data suggest a potential role for adjuvant chemotherapy for early-stage lung cancer (33, 34), surgical resection is the only current available intervention that considerably improves the prognosis of these patients. This study shows lower survival rates of Hispanic patients with stage I non–small cell lung cancer, as compared with whites. The observed survival difference did not persist after adjusting for differences in surgical treatment and stage at diagnosis. Future work should explore why Hispanics are less likely to undergo resection and more likely to present with more advanced stage disease. Understanding these differences is important, because improving the detection at an early stage and improving the rate of surgical resection among Hispanics and other minorities may be a valuable means of improving the outcomes of patients with lung cancer and reducing ethnic disparities.
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org Conflict of Interest Statement: J.P.W. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; T.M. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; C.H. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; P.H. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; M.C.I. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; E.A.H. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form November 4, 2004; accepted in final form February 17, 2005
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
2 test for proportions and t test for continuous variables. 
