SEARCH

SEARCH BY CITATION

Keywords:

  • nonsmall cell lung cancer;
  • disparities;
  • survival;
  • socioeconomic status

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. FUNDING SOURCES
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

BACKGROUND:

Studies examining the impact of lower socioeconomic status (SES) on the outcomes of patients with nonsmall cell lung cancer (NSCLC) are inconsistent. The objective of this study was to clearly elucidate the association between SES, education, and clinical outcomes among patients with NSCLC.

METHODS:

The study population was derived from a consecutive, retrospective cohort of patients with NSCLC who received treatment within the Duke Health System between 1995 and 2007. SES determinants were based on the individual's census tract and corresponding 2000 Census data. Determinants included the percentage of the population living below poverty, the median household income, and the percentages of residents with at least a high school diploma and at least a bachelor's degree. The SES and educational variables were divided into quartiles. Statistical comparisons were performed using the 25th and 75th percentiles.

RESULTS:

Individuals who resided in areas with a low median household income or in which a high percentage of residents were living below the poverty line had a shorter cancer-specific 6-year survival than individuals who resided in converse areas (P = .0167 and P = .0067, respectively). Those living in areas in which a higher percentage of residents achieved a high school diploma had improved disease outcomes compared with those living in areas in which a lower percentage attained a high school diploma (P = .0033). A survival advantage also was observed for inhabitants of areas in which a higher percentage of residents attained a bachelor's degree (P = .0455).

CONCLUSIONS:

Low SES was identified as an independent prognostic factor for poor survival in patients with both early and advanced stage NSCLC. Patients who lived in areas with high poverty levels, low median incomes, and low education levels had worse mortality. Cancer 2012. © 2012 American Cancer Society.

Lung cancer is the leading cancer killer in the United States and results in more than 160,000 deaths per year. Nearly 60% of patients with lung cancer will die within 1 year of diagnosis. Traditionally, personal factors, such as smoking status, age, sex, occupational exposure, etc, have been associated with lung cancer mortality. These individual determinants may not be the only factors that affect the survival of patients with NSCLC. Thus, having a clear understanding of the factors that influence this disease is imperative.

Social scientists have long investigated the power that income inequality and deprivation has had on our population. The impact of socioeconomic status (SES) on health outcomes has gained much attention in recent years. Several investigators have studied an individual's SES and the variables on which SES is based, ie, income, education, wealth, and occupation, and their effect on health outcomes. Several studies have demonstrated that individuals with lower SES have inferior disease outcomes across a wide range of disease states, including cardiovascular disease and cancer.1-4 It has been observed that economically disadvantaged individuals are more likely to smoke, less likely to exercise, and les likely to eat a well balanced diet. In addition, a lack of advanced, formal education tends to result in employment in occupations that have an increased rate of exposure to pollutants. Working, families with lower income also may exceed the financial minimums necessary to participate in publicly funded health insurance programs. This results in noninsurance or under insurance along with decreased access to adequate preventative health care, such as smoking prevention programs.

There have been some published reports examining individual and group SES and outcomes with inconsistent results.5-10 The objective of the current study was to more clearly elucidate any association between SES, education, and clinical outcomes among patients with nonsmall cell lung cancer (NSCLC). Uniquely, we created a consecutive database of all patients who were treated at a comprehensive cancer center in the southeastern United States. Duke University Medical Center is an internationally recognized, quaternary care center with 1 of the largest integrated, multidisciplinary thoracic oncology programs in North America. Therefore, these data should represent a patient population that could be considered “a best-case scenario,” because these patients were able and willing to seek care from a top-tier lung cancer treatment facility.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. FUNDING SOURCES
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

Study Population

The study population was derived from a consecutive, retrospective cohort of patients with pathologically confirmed NSCLC who received treatment within the Duke Health System (Duke University Medical Center, Durham Regional Hospital, and Duke Comprehensive Cancer Center) between 1995 and 2007. These data were obtained from the Duke Comprehensive Cancer Center Tumor Registry under an institutional review board-approved protocol. It is required to maintain >90% follow-up of all newly diagnosed lung cancer patients. The Tumor Registry data included TNM staging, histology, sex, race, age at diagnosis, address at the time of diagnosis, and follow-up and treatment data. All patients had a follow-up period of at least 2 years. Only patients with complete tumor staging and a home address that could be located in the 2000 Census were included in the final study cohort.

Socioeconomic and Educational Determinants

SES determinants were created by cross-referencing the individual's primary residence at the time of diagnosis, including zip code and state, with the 2000 Census data to determine the individual's census tract. According to the United States Census Bureau, a census tract is a statistical subdivision of counties that include between 1500 and 8000 individuals. Census tracts are comprised of a relatively homogenous population in terms of characteristics and socioeconomic determinants, such as education, poverty level, and household income. From the census tract, income and education data representative of the area of domicile were abstracted for each patient. Two income determinants for each patient were defined: 1) the percentage of the population within the area of domicile living below the poverty line and 2) the median household income of the population within the area of domicile. The percentage below the poverty line is defined as the percentage of the population within the area of domicile that live below the poverty line divided by the total number of residents within the census tract. The median household income is defined as the median income of all of the households located within the census tract, including those of single individuals and unrelated groups of 2 or more. Two education determinants for each patient were defined: 1) the percentage of residents that attained at least a high school diploma and 2) the percentage of residents that attained at least a bachelor's degree. These determinants are defined as the percentage of the population within the tract that attained the specific education level divided by the total number of residents within the census tract.

Statistical Analysis

Statistical analysis was performed on the cohort as a whole and included patients with all stages of NSCLC. Categorical data are presented as percentages and are compared using the Pearson chi-square test. Continuous data are presented as medians and are compared using the Wilcoxon test. Survival was defined as the period between the initial diagnosis and the date of cancer-specific death or last follow-up. Survival estimates were determined using the Kaplan-Meier method, and any difference in survival between groups was assessed using the log-rank test. Significant univariate associations with survival were assessed for independence using Cox logistic regression modeling. The SES and educational determinant variables were divided into quartiles for the total study cohort. Statistical comparisons were performed using the first (25th percentile) and third (75th percentile) quartiles. Statistical significance was assumed for a P value < .05. For the purpose of data analysis, patients who were alive at the end of the 6-year study period were censored.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. FUNDING SOURCES
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

Patient Demographics

The study population was comprised of 4820 patients with NSCLC (Table 1). The mean age at diagnosis was 65 years, and the median follow-up was 40.18 months. The study population was comprised of residents of the states of Virginia, South Carolina, West Virginia, Georgia, and Florida; and >75% of the population resided in the state of North Carolina.

Table 1. Population Demographics
VariableNo. (%)
  1. Abbreviations: NOS, not otherwise specified; NSCLC, nonsmall cell lung cancer; SD, standard deviation.

Mean±SD age [range], y65 ± 11 [20-105]
Sex 
 Men2743 (56.9)
 Women2077 (43.1)
Race 
 White3909 (81.1)
 African American892 (18.5)
 Other19 (0.4)
Histology 
 Adenocarcinoma1939 (40.2)
 Squamous2022 (41.9)
 Large cell/undifferentiated104 (2.2)
 NSCLC NOS755 (15.7)
Stage 
 In situ12 (0.25)
  I1154 (23.9)
  II319 (6.6)
  III1168 (24.2)
  IV1449 (30.1)
 Localized, not otherwise stated132 (2.7)
 Direct extension, not otherwise stated55 (1.1)
 Involves regional lymph nodes,  not otherwise stated72 (1.5)
 Direct extension and regional lymph nodes,  not otherwise stated42 (0.9)
 Regional NOS, not otherwise stated3 (<0.1)
 Distant involvement, not otherwise stated305 (6.3)
 Unknown, not otherwise stated109 (2.3)

Socioeconomic and Educational Determinants

Kaplan-Meier survival analysis of the socioeconomic determinants across all disease stages indicated that, within an individual's area of domicile, the median household income and the percentage of the population living below the poverty line are indicators of disease outcome. Figure 1 reveals that those residing in an area with a low median household income and a high percentage of individuals living below the poverty line had a shorter cancer-specific 6-year survival than those living in areas with high median household incomes and a low number of individuals living below the poverty line (P = .0167 and P = .0067, respectively). Those patients within the 75th percentile for median household income within the area of domicile had a median survival of 16.4 months versus 13.87 months for those in the 25th percentile. With regard to the percentage of residents living below the poverty line within an area of residence, those patients in the 25th percentile (ie, those residing in the quartile with the smallest percentage below poverty level) had a median survival of 16.7 months compared with those in the 75th percentile, who had a median survival of 13.30 months.

thumbnail image

Figure 1. These charts illustrate the impact of socioeconomic status determinants on the outcome of patients with nonsmall cell lung cancer.

Download figure to PowerPoint

Survival analysis of the educational determinants across all disease stages indicated that the educational level of the population within an individual's area of domicile impacts cancer-specific 6-year survival. Figure 2 reveals that those residing within areas that had a higher percentage of residents who have not achieved at least a high school diploma realized a decreased disease outcome compared with those who resided in an area with a higher percentage of the population having attained at least a high school diploma (P = .0033). Those patients in the 75th percentile had a median survival of 16.2 months compared with 14.03 months for those in the 25th percentile. It is noteworthy that, when considering the impact of a more advanced formal education in the form of the attainment of at least a bachelor's degree, a survival advantage was observed (P = .0455), but the survival advantage was not as great as the advantage observed with regard to the attainment of at least a high school diploma. Those patients in the 75th percentile had a median survival of 15.5 months versus 14.27 months for those in the 25th percentile. The percentage living in poverty, the median income, and the percentages with a high school diploma and a bachelor's degree were highly correlated, as expected. In addition, Pearson correlation analysis indicated a moderately strong association between the percentage with a high school diploma and median income and between the percentage with a bachelor's degree and median income. Analysis of the percentage with a high school diploma versus median income resulted in an R value of 0.7699, thereby accounting for nearly 60% of covariance. In addition, analysis of the percentage with a bachelor's degree versus median income resulted in an R value of 0.7371, thereby accounting for 54% of covariance. Therefore, a multivariate model that included these factors did not demonstrate independence of the 4 variables. A larger population would be necessary to delineate the independence of education and income.

thumbnail image

Figure 2. These charts illustrate the impact of educational determinants on the outcome of patients with nonsmall cell lung cancer.

Download figure to PowerPoint

Analysis of SES and educational determinants and their association with the stage of disease at presentation revealed that neither the percentage of residents who lived below poverty within an individual's area of residence (Fig. 3) nor the percentage of residents with at least a high school education had any impact on the stage of disease at presentation. Analyses were also performed using median income and the percentage with at least a bachelor's degree. These determinants also did not reveal a statistically significant impact on stage of presentation. These data are not shown.

thumbnail image

Figure 3. These charts illustrate the effect of socioeconomic status and education on disease stage at presentation in patients with nonsmall cell lung cancer. H.S. indicates high school; N.S., nonsignificant.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. FUNDING SOURCES
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

Lung cancer continues to be a significant health issue in the United States. This is especially true within the southeastern region of the country (ie, Virginia, West Virginia, North Carolina, South Carolina, Georgia, and Florida). This area consistently realizes a lung cancer incidence rate above the national average; and, although, strides have been made nationally to reduce the incidence of lung cancer through tobacco control measures, the southeast still is burdened by a smoking incidence rate (22.24 of 100,000 population) that is well above the national average (19.7 of 100,000 population). In addition, this region is 1 of the most economically challenged areas of the country, and it is estimated that poverty affects approximately 15% of the population (with the exception of Virginia and Maryland).11

Multiple studies have demonstrated that factors indicating lower SES have a strong association with smoking status and persistence of smoking.12-19 However, with respect to lung cancer outcomes, evidence supporting the impact of SES on disease survival is contradictory. Some investigators have reported that there is no association between NSCLC mortality and SES,5, 6, 10 whereas others have reported an association only in patients with stage I disease.7, 20

Tammemagi et al evaluated the impact of several sociodemographic variables on lung cancer treatment and survival in 1155 patients in various disease stages.6 They observed that comorbidity, which is impacted by SES, had deleterious effects on lung cancer outcomes, but they did not observe that SES independently impacted adverse comorbidity. Bouchardy et al reported that age, histology, and disease stage were independent factors that influenced prognosis in a population of 428 patients who underwent surgical resection. Those authors suggest that SES was among the variables that did not appear to be associated with long-term survival.5 Herndon et al investigated the effect of SES on survival, as measured by education, in 1577 patients with stage III and IV lung cancer who were treated on 11 studies conducted by the Cancer and Leukemia Group B. Those investigators reported that education, as a surrogate of SES, was not predictive of survival, suggesting that, once enrolled onto a clinical trial, education has no impact on patients with small cell lung cancer or stage III or IV NSCLC.9 However, it could be argued that their study was influenced by selection bias, because patients who are more educated or have a better understanding of their disease are more likely to self-select into a clinical trial. Stavraky et al reported that no significant association was observed between the survival of their patients 224 lung cancer and the SES of their patients as measured by education.10

In the current study, we investigated neighborhood-level socioeconomic determinants and their effect on disease outcome by using information obtained from Duke University Medical Center in Durham, North Carolina. It is known that the environment and community where an individual resides can affect health outcomes in several disease processes. Based on this, we hypothesized that an individual's neighborhood-level determinants may be associated with NSCLC outcomes. These factors may influence an individual patient's willingness to seek care and may be associated with access to adequate care and available resources. Duke University Medical Center is centrally located within the southeast region of the country and serves a large number of patients with lung cancer from the surrounding southeastern states as well as a diverse population within the city of Durham, North Carolina. 2009 Census tract data from the Federal Financial Institutions Examination Council indicates that the 20 census tracts within a 10-mile radius of the hospital have various levels of the population living below the poverty level, and the average proportion is approximately 23.4%. The lowest area has 2.1% living below the poverty level, and the highest has 44.9%. Considering the close proximity of this major medical center with a comprehensive cancer center and the ability and willingness of those from surrounding states to seek care at such an institution, these data should represent a “best-case scenario” with regard to the effect of socioeconomic factors on disease outcome. However, the data suggest that, even with such a scenario, the negative effects of SES still are realized. The results from this study indicate that a low median income within an individual's area of domicile and a high percentage of the population living below the poverty level negatively impact cancer-specific disease outcomes regardless of disease stage within the study population. In addition, the education level of the population within an individual's area of domicile also appears to have a negative impact on disease outcomes. However, none of the socioeconomic determinants that were evaluated in our study appeared to be associated with the disease stage at presentation.

We realize that, like previous investigations into the impact of SES on lung cancer outcomes, the current study also has its limitations. It has been demonstrated that the incidence of smoking increases with lower SES and that smoking impacts adverse comorbidity. This study was limited by its inability to capture smoking history from tumor registry data as well as comorbid disease and functional status. However, we have started to gather these data for future analyses. Another limitation that is inherent in a study of this kind is that we were not able to access individual-level SES data. Therefore, although we used neighborhood socioeconomic characteristics in the current analysis as a surrogate, its use is controversial.21 However, we did include a large enough number of observations for appropriate statistical power, which should dramatically decrease the degree of error that results from estimating the SES based on a patient's area of domicile.

In conclusion, the current results indicate that low SES is an independent prognostic factor for poor survival in patients with both early and advanced stage NSCLC. Patients living in areas with high poverty levels, low median income, and low education levels have worse mortality despite having the same theoretical access to our tertiary, urban, academic center. These results present a dismal outlook when considering the effects of socioeconomic factors on those who are not able or willing to travel to a major medical center for treatment. Based on these findings, the first charge would be to determine why. Many suggest that disadvantaged patients present with more severe disease because of less resources and/or access. However, our study did not reveal a direct association between SES and the extent of disease at presentation. Overall, our study suggests that we cannot assume that more advanced disease is the main cause of the survival disparity. These provocative results need to be validated in a larger cohort of patients within the southeastern United States. This has led us to broaden our investigation of these factors to include the entire state of North Carolina as well as considering other health determinants and barriers associated with poor disease outcomes in patients with lung cancer. If validated, these results may impact the state's public health policy with respect to tobacco control and health disparities in oncology outcomes.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. FUNDING SOURCES
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES