Abstract
Background
Black Americans receive a diagnosis at later stage of lung cancer more often than White Americans. We undertook a population-based study to identify factors contributing to racial disparities in lung cancer stage of diagnosis among low-income adults.
Research Question
Which multilevel factors contribute to racial disparities in stage of lung cancer at diagnosis?
Study Design and Methods
Cases of incident lung cancer from the prospective observational Southern Community Cohort Study were identified by linkage with state cancer registries in 12 southeastern states. Logistic regression shrinkage techniques were implemented to identify individual-level and area-level factors associated with distant stage diagnosis. A subset of participants who responded to psychosocial questions (eg, racial discrimination experiences) were evaluated to determine if model predictive power improved.
Results
We identified 1,572 patients with incident lung cancer with available lung cancer stage (64% self-identified as Black and 36% self-identified as White). Overall, Black participants with lung cancer showed greater unadjusted odds of distant stage diagnosis compared with White participants (OR,1.29; 95% CI, 1.05-1.59). Greater neighborhood area deprivation was associated with distant stage diagnosis (OR, 1.58; 95% CI, 1.19-2.11). After controlling for individual- and area-level factors, no significant difference were found in distant stage disease for Black vs White participants. However, participants with COPD showed lower odds of distant stage diagnosis in the primary model (OR, 0.72; 95% CI, 0.53-0.98). Interesting and complex interactions were observed. The subset analysis model with additional variables for racial discrimination experiences showed slightly greater predictive power than the primary model.
Interpretation
Reducing racial disparities in lung cancer stage at presentation will require interventions on both structural and individual-level factors.
Key Words: lung cancer, multilevel factors, racial disparities, stage at diagnosis, United States
Graphical Abstract
Take-home Points.
Study Question: Which multilevel factors contribute to racial disparities in stage of lung cancer at diagnosis?
Results: A greater percentage of Black participants than White participants recruited primarily from community health centers received a diagnosis of distant-stage lung cancer, although a racial disparity was not apparent after adjusting for individual- and area-level factors. Including psychosocial variables (eg, racial discrimination experiences) slightly improved model predictive power in a follow-up model.
Interpretation: Reducing racial disparities in lung cancer stage at presentation will require interventions on both structural and individual-level factors.
Considerable racial disparities exist in mortality from lung cancer, the leading cause of cancer-related death in the United States. Black Americans have poorer survival rates for most cancer types, including lung cancer.1 The higher lung cancer mortality rate is particularly evident among Black men when compared with White men (54.0 per 100,000 population vs 47.0 per 100,000 population, respectively).1 Stage at diagnosis is an important factor influencing lung cancer mortality and survival.2,3 When lung cancer is diagnosed at a localized stage, curative surgical treatment options are available and the 5-year relative survival rate is 60%.1 However, a large percentage of patients with lung cancer receive a diagnosis of distant stage disease, when the 5-year relative survival rate is only 6%.1 Furthermore, a greater percentage of Black Americans than White Americans receive a diagnosis of stage IV lung cancer (42% vs 36%),4 which may contribute to the poorer lung cancer survival among this population.1,5
Although numerous studies have investigated racial disparities in lung cancer survival,2,6,7 few studies have identified factors associated with racial disparities in lung cancer stage at diagnosis, particularly among low-income and medically underserved populations. Existing research has focused on individual-level factors (eg, age, smoking status, comorbidities) associated with disparities in lung cancer outcomes; however, growing research suggests that area-level factors (eg, neighborhood deprivation and availability of health-care facilities) are important predictors of cancer outcomes.8, 9, 10, 11, 12 We aimed to address these knowledge gaps by evaluating racial disparities in lung cancer stage at diagnosis and identifying multilevel factors contributing to stage of lung cancer at presentation.
Study Design and Methods
Study Design and Participants
The Southern Community Cohort Study (SCCS) was established in 2002 to examine disparities in cancer and other chronic diseases. Between 2002 and 2009, the SCCS enrolled nearly 86,000 English-speaking adults between 40 and 79 years of age who were living in 12 southeastern states. About two-thirds of SCCS participants self-identified as Black. The SCCS recruited participants largely through community health centers (86%), organizations that provide health care to low-income and uninsured people. As described elsewhere,13,14 the remaining 14% of participants were recruited via an age-, sex-, and race-stratified random sample of the general population in the study states to ensure additional income and educational diversity in the cohort. Participants completed a survey at baseline and up to four additional surveys during the SCCS follow-up period. The overall SCCS was approved by institutional review boards at Vanderbilt University and Meharry Medical College, and all participants provided written informed consent. The Vanderbilt University Medical Center Institutional Review Board approved the procedures for this specific study, which uses SCCS data (#191867). Additional study details are provided elsewhere.13, 14, 15
Individual-Level Factors
Incident lung cancers were identified by linkage with state cancer registries through the end of follow-up in 2019. Stage at lung cancer diagnosis was derived using the American Joint Committee on Cancer TNM staging system guidelines (6th and 7th editions). For participants missing American Joint Committee on Cancer information on stage at diagnosis, we used the Surveillance, Epidemiology, and End Results summary stage variable. Here, local disease was equivalent to stage I, regional disease was equivalent to stage II/III, and distant disease was equivalent to stage IV.2 Our outcome, stage at lung cancer diagnosis, was coded as distant vs not distant (ie, localized and regional stage at diagnosis were collapsed into the not distant category). Social, behavioral, and medical history information were ascertained at study baseline via questionnaire. We assessed comorbidities using the previously validated Charlson comorbidity index.16,17 Participant pack-year smoking history at the time of lung cancer diagnosis was estimated using the most recent follow-up questionnaires (methodology described elsewhere).18
Area-Level Factors
Each SCCS participant’s residential address (latitude and longitude) at the time of the baseline survey was geocoded and used to determine the county, 2000 US Census tract, and 2000 US Census block group. Addresses were matched to the street level for 91% of residential addresses in the SCCS cohort. Zip code centroid was used for the 9% of addresses that did not match to a street address or where only a post office box or rural delivery route was provided. Addresses were used to calculate an area deprivation index based on the formula developed by Messer and colleagues.19 Area deprivation provides a percentile measure of the socioeconomic status of the US Census block group containing the participant residential address. For ease of interpretation, we categorized area deprivation into quartiles ranging from least deprived to most deprived. Residential addresses also were used to determine rural or urban status based on the US Department of Agriculture Rural-Urban Continuum Codes of the county of residence. We also geocoded hospital and National Cancer Institute Comprehensive Cancer Center locations and used participant residential addresses to determine geographic distance from the nearest hospital and Comprehensive Cancer Center. Table 1 presents the full list of individual- and area-level variables included in the primary analysis.
Table 1.
Characteristics of Participants With a Diagnosis of Lung Cancer in the Southern Community Cohort Study, 2002 through 2019 (N = 1,572)
| Characteristic | White (n = 568) | Black/African American (n = 1,004) | Total (N = 1,572) |
|---|---|---|---|
| Stage at diagnosis | |||
| Distant | 247 (43.5) | 500 (49.8) | 747 (47.5) |
| Not distant (localized or regional) | 321 (56.5) | 504 (50.2) | 825 (52.5) |
| Age at diagnosis, y | |||
| Mean ± SD | 64.4 ± 8.75 | 61.7 ± 8.75 | 62.7 ± 8.85 |
| Median (minimum, maximum) | 65.0 (43.0, 87.0) | 61.0 (42.0, 91.0) | 62.0 (42.0, 91.0) |
| Sex | |||
| Female | 329 (57.9) | 469 (46.7) | 798 (50.8) |
| Male | 239 (42.1) | 535 (53.3) | 774 (49.2) |
| Education | |||
| Less than high school | 194 (34.2) | 431 (42.9) | 625 (39.8) |
| High school graduate | 190 (33.5) | 320 (31.9) | 510 (32.4) |
| More than high school | 179 (31.5) | 236 (23.5) | 415 (26.4) |
| Missing | 5 (0.9) | 17 (1.7) | 22 (1.4) |
| Household annual income, US$ | |||
| < $15,000 | 337 (59.3) | 668 (66.5) | 1005 (63.9) |
| ≥ $15,000 | 217 (38.2) | 308 (30.7) | 525 (33.4) |
| Missing | 14 (2.5) | 28 (2.8) | 42 (2.7) |
| Marital status | |||
| Married | 216 (38.0) | 275 (27.4) | 491 (31.2) |
| Separated or divorced | 199 (35.0) | 352 (35.1) | 551 (35.1) |
| Widowed | 82 (14.4) | 121 (12.1) | 203 (12.9) |
| Single | 53 (9.3) | 234 (23.3) | 287 (18.3) |
| Missing | 18 (3.2) | 22 (2.2) | 40 (2.5) |
| Has health insurance | |||
| No | 188 (33.1) | 391 (38.9) | 579 (36.8) |
| Yes | 372 (65.5) | 592 (59.0) | 964 (61.3) |
| Missing | 8 (1.4) | 21 (2.1) | 29 (1.8) |
| Pack-year of smoking | |||
| Mean ± SD | 49.3 ± 32.0 | 28.2 ± 23.0 | 35.8 ± 28.5 |
| Median (minimum, maximum) | 45.0 (0, 198) | 23.0 (0, 225) | 30.0 (0, 225) |
| Missing | 19 (3.3) | 36 (3.6) | 55 (3.5) |
| History of COPD | |||
| No | 413 (72.7) | 878 (87.5) | 1291 (82.1) |
| Yes | 149 (26.2) | 109 (10.9) | 258 (16.4) |
| Missing | 6 (1.1) | 17 (1.7) | 23 (1.5) |
| Prior cancer diagnosis | |||
| No | 435 (76.6) | 911 (90.7) | 1346 (85.6) |
| Yes | 120 (21.1) | 67 (6.7) | 187 (11.9) |
| Missing | 13 (2.3) | 26 (2.6) | 39 (2.5) |
| Charlson comorbidity index score | |||
| 1 | 49 (8.6) | 188 (18.7) | 237 (15.1) |
| 2 | 108 (19.0) | 238 (23.7) | 346 (22.0) |
| 3 | 118 (20.8) | 233 (23.2) | 351 (22.3) |
| 4 | 118 (20.8) | 164 (16.3) | 282 (17.9) |
| 5 | 76 (13.4) | 91 (9.1) | 167 (10.6) |
| ≥ 6 | 78 (13.7) | 51 (5.1) | 129 (8.2) |
| Missing | 21 (3.7) | 39 (3.9) | 60 (3.8) |
| Ever underwent colonoscopy | |||
| No | 349 (61.4) | 751 (74.8) | 1100 (70.0) |
| Yes | 208 (36.6) | 225 (22.4) | 433 (27.5) |
| Missing | 11 (1.9) | 28 (2.8) | 39 (2.5) |
| Area deprivation index quartile | |||
| 1 (least deprived) | 253 (44.5) | 122 (12.2) | 375 (23.9) |
| 2 | 188 (33.1) | 203 (20.2) | 391 (24.9) |
| 3 | 82 (14.4) | 304 (30.3) | 386 (24.6) |
| 4 (most deprived) | 34 (6.0) | 371 (37.0) | 405 (25.8) |
| Missing | 11 (1.9) | 4 (0.4) | 15 (1.0) |
| Distance from comprehensive cancer center, 100-mile increments | |||
| Mean ± SD | 1.01 ± 0.68 | 0.953 ± 0.89 | 0.973 ± 0.82 |
| Median (minimum, maximum) | 0.841 (0.0071, 3.86) | 0.831 (0.0045, 3.94) | 0.835 (0.0045, 3.94) |
| Missing | 6 (1.1) | 15 (1.5) | 21 (1.3) |
Data are presented as No. (%), unless otherwise indicated.
Primary Analysis
Summary statistics and unadjusted associations were computed as an exploratory step. Logistic regression was used to model factors associated with distant stage at diagnosis. For model selection we used a 10-fold cross-validated least absolute shrinkage and selection operator (LASSO) model to identify features predictive of distant stage at diagnosis.20 The cross-validated LASSO model selection process is a regularization technique. It uses shrinkage to remove features from a model that are not necessary for predictive performance. In this process, we considered 30 main effects and all two-way interaction effects with race. The 10-fold cross-validated area under the receiver operating characteristic curve was used to measure prediction accuracy at each λ value. e-Figure 1 displays the 10-fold cross-validated LASSO variable selection procedure and each model’s associated predictive accuracy. The chosen model then was relaxed fully for inference and interpretation purposes. Key confounders were added to the model and chosen interactions were removed to facilitate inference on the key features. Modifications were conducted only if the overall predictive accuracy remained stable. We also explored random forest models as a robustness check on the logistic and LASSO assumptions, but the random forest models did not perform better nor did they lead to new insights. Because of the computational complexity of simultaneous multiple imputation and cross-validation, we used the complete case data for the model selection process and imputation in the final regression results. Complete cases were defined as the individuals with no missing data for all predictors under consideration in the variable selection algorithm. After identifying the chosen variable selection set of predictors, we used multiple imputation techniques.21 We used the Multivariate Imputation by Chained Equations using the “mice” R package version 3.11.0 (R Foundation for Statistical Computing) to refit the model accounting for predictor missingness in the analysis set.21,22
Subset Analysis of Follow-up Data
We conducted a subset analysis with additional data available from the second SCCS follow-up questionnaire. The second follow-up questionnaire asked whether the participant had adverse experiences in adulthood, such as whether the participant’s spouse, family member, or close friend had ever physically hurt them, psychologically harmed them, or threatened them with a gun or weapon. The second follow-up questionnaire also asked questions about participant experiences of racial discrimination and childhood living environment. Because this additional information was not available from the baseline survey data used for our primary analyses (described previously herein), we conducted a subset analysis with this follow-up data to investigate whether we obtained improved model fit and predictive power when these new variables were included. The sample for this subset analysis included those who responded to and provided complete information on the variables of interest in the second follow-up questionnaire (Table 2). Similar to our methods for the primary analysis, we applied a 10-fold cross-validated LASSO approach with 30 main effects and all two-way interactions for the subset analysis model construction.20 For the subset analysis, the fully relaxed LASSO model was not modified for inference purposes.
Table 2.
Characteristics of Participants With a Diagnosis of Lung Cancer in the Southern Community Cohort Study, Follow-up Data, 2002 through 2019 (n = 472)
| Characteristic | White (n = 189) | Black/African American (n = 283) | Total (N = 472) |
|---|---|---|---|
| Stage at diagnosis | |||
| Distant | 56 (29.6) | 107 (37.8) | 163 (34.5) |
| Not distant (localized or regional) | 133 (70.4) | 176 (62.2) | 309 (65.5) |
| Age at diagnosis, y | |||
| Mean ± SD | 59.3 ± 8.08 | 54.0 ± 7.85 | 56.1 ± 8.36 |
| Median (minimum, maximum) | 60.0 (42.0, 77.0) | 53.0 (40.0, 79.0) | 56.0 (40.0, 79.0) |
| Sex | |||
| Female | 115 (60.8) | 173 (61.1) | 288 (61.0) |
| Male | 74 (39.2) | 110 (38.9) | 184 (39.0) |
| Education | |||
| Less than high school | 48 (25.4) | 100 (35.3) | 148 (31.4) |
| High school graduate | 68 (36.0) | 93 (32.9) | 161 (34.1) |
| More than high school | 72 (38.1) | 87 (30.7) | 159 (33.7) |
| Missing | 1 (0.5) | 3 (1.1) | 4 (0.8) |
| Household income, US$ | |||
| < 15,000 | 99 (52.4) | 168 (59.4) | 267 (56.6) |
| ≥ 15,000 | 87 (46.0) | 107 (37.8) | 194 (41.1) |
| Missing | 3 (1.6) | 8 (2.8) | 11 (2.3) |
| Marital status | |||
| Married | 71 (37.6) | 85 (30.0) | 156 (33.1) |
| Separated or divorced | 68 (36.0) | 98 (34.6) | 166 (35.2) |
| Widowed | 27 (14.3) | 28 (9.9) | 55 (11.7) |
| Single | 15 (7.9) | 67 (23.7) | 82 (17.4) |
| Missing | 8 (4.2) | 5 (1.8) | 13 (2.8) |
| Has health insurance | |||
| No | 56 (29.6) | 115 (40.6) | 171 (36.2) |
| Yes | 131 (69.3) | 164 (58.0) | 295 (62.5) |
| Missing | 2 (1.1) | 4 (1.4) | 6 (1.3) |
| Charlson comorbidity index score | |||
| 1 | 17 (9.0) | 51 (18.0) | 68 (14.4) |
| 2 | 37 (19.6) | 73 (25.8) | 110 (23.3) |
| 3 | 38 (20.1) | 72 (25.4) | 110 (23.3) |
| 4 | 33 (17.5) | 42 (14.8) | 75 (15.9) |
| 5 | 30 (15.9) | 22 (7.8) | 52 (11.0) |
| ≥ 6 | 27 (14.3) | 14 (4.9) | 41 (8.7) |
| Missing | 7 (3.7) | 9 (3.2) | 16 (3.4) |
| Ever experienced discrimination because of race or ethnicity | |||
| No | 159 (84.1) | 177 (62.5) | 336 (71.2) |
| Yes | 30 (15.9) | 106 (37.5) | 136 (28.8) |
| Had adult adverse experience | |||
| No | 116 (61.4) | 184 (65.0) | 300 (63.6) |
| Yes | 73 (38.6) | 99 (35.0) | 172 (36.4) |
| Lived with mother and father during childhood | |||
| No | 45 (23.8) | 112 (39.6) | 157 (33.3) |
| Yes | 140 (74.1) | 161 (56.9) | 301 (63.8) |
| Missing | 4 (2.1) | 10 (3.5) | 14 (3.0) |
| Lived with father only during childhood | |||
| No | 182 (96.3) | 266 (94.0) | 448 (94.9) |
| Yes | 3 (1.6) | 7 (2.5) | 10 (2.1) |
| Missing | 4 (2.1) | 10 (3.5) | 14 (3.0) |
| Lived with grandparent, aunt, uncle, or other relative during childhood | |||
| No | 171 (90.5) | 240 (84.8) | 411 (87.1) |
| Yes | 14 (7.4) | 33 (11.7) | 47 (10.0) |
| Missing | 4 (2.1) | 10 (3.5) | 14 (3.0) |
| Distance from cancer center, 100-mile increments | |||
| Mean ± SD | 1.01 ± 0.67 | 1.00 ± 0.95 | 1.01 ± 0.84 |
| Median (minimum, maximum) | 0.855 (0.00914, 3.86) | 0.823 (0.00551, 3.86) | 0.837 (0.00551, 3.86) |
| Missing | 4 (2.1) | 6 (2.1) | 10 (2.1) |
| Distance from hospital, 100-mile increments | |||
| Mean ± SD | 0.277 ± 0.250 | 0.260 ± 0.352 | 0.267 ± 0.315 |
| Median(minimum, maximum) | 0.194 (0.0044, 0.928) | 0.0680 (0.0040, 2.59) | 0.123 (0.0040, 2.59) |
| Missing | 4 (2.1) | 6 (2.1) | 10 (2.1) |
| Rural-urban continuum code | |||
| Metro | 148 (78.3) | 222 (78.4) | 370 (78.4) |
| Nonmetro | 41 (21.7) | 61 (21.6) | 102 (21.6) |
Data are presented as No. (%), unless otherwise indicated.
Results
Descriptive Characteristics of Study Participants
Of 2,114 SCCS participants with a diagnosis of incident lung cancer, 1,572 of them had available lung cancer stage information. Of those 1,572 participants, approximately 64% self-identified as Black and 36% self-identified as White. The mean ± SD age of participants was 62.7 ± 8.85 years. Approximately half of participants were women (50.8%). Most participants had a high school education or less (72.2%), had an annual household income of < $15,000 (63.9%), and had health insurance (61.3%). The mean ± SD pack-years smoked at lung cancer diagnosis was 35.8 ± 28.5 years. On average, participants lived 97.3 ± 82.2 miles from a comprehensive cancer center. Overall, 47.5% of participants had received a diagnosis of distant stage disease; however, a greater percentage of Black than White participants had received a diagnosis of distant stage lung cancer (49.8% vs 43.5%, respectively) (Table 1). A more detailed presentation of the percent of participants with diagnoses of localized, regional, and distant stages by participant race is presented in Figure 1. Characteristics of SCCS participants included in the primary analysis are provided in Table 1.
Figure 1.
Bar graph showing the percentage of cases of lung cancer diagnosed at localized, regional, and distant stage by patient race among patients with incident lung cancer in the Southern Community Cohort Study, 2002 through 2019 (N = 1,572).
Of 1,572 participants with lung cancer stage data, 472 participants (30.0%) responded to the second follow-up questionnaire administered between 2012 and 2015. Similar to the full dataset, the subset analysis found a greater percentage of Black participants than White participants in the sample had received a diagnosis of distant stage lung cancer (37.8% vs 29.6%, respectively) (Table 2). Overall, the demographics of participants in the subset analysis (Table 2) were similar to the demographics of participants in the primary analysis (Table 1). Of note for the additional follow-up questions, 28.8% of subset participants had experienced discrimination based on their race or ethnicity, and more than twice the percentage of Black participants had experienced racial discrimination compared with White participants (37.5% vs 15.9%, respectively). About 36% of participants included in the subset analysis had an adverse experience during their adult life, such as being threatened with a weapon or gun. Most participants lived with their mother and father during childhood (63.8%) and resided in a metropolitan area (78.4%) (Table 2).
Associations Between Individual- and Area-Level Factors and Distant Stage Lung Cancer Diagnosis
e-Table 1 presents unadjusted results from the primary analysis. Overall, Black participants showed greater odds of distant stage diagnosis compared with White participants (OR, 1.29; 95% CI, 1.05-1.59). Participants who identified as male or who were single also had greater odds of distant stage at diagnosis compared with participants who were female or married, respectively (male participants: OR, 1.31 [95% CI, 1.07-1.59]; single participants: OR, 1.56 [95% CI, 1.16-2.09]). Greater area deprivation also was associated with distant stage diagnosis because participants living in neighborhoods with the greatest deprivation (ie, quartile 4) were 58% more likely to receive a diagnosis of distant stage lung cancer than those living in neighborhoods with the least deprivation (ie, quartile 1; OR, 1.58; 95% CI, 1.19-2.11). Conversely, participants who had health insurance, had a history of COPD, or had undergone a colonoscopy were less likely to have received a diagnosis of distant stage lung cancer (e-Table 1). Similarly, participants with the highest Charlson comorbidity index score (≥ 6) showed lower odds of distant stage of disease at diagnosis than those with the lowest score (score of 1) (e-Table 1).
Although the primary analysis model was not highly predictive (cross-validated area under the receiver operating characteristic curve, 0.561), the following variables remained in the model after LASSO shrinkage: race, age at lung cancer diagnosis, sex, educational attainment, household income, marital status, health insurance status, pack-year smoking history, history of COPD, prior cancer diagnosis, Charlson comorbidity index score, history of colonoscopy screening, area deprivation, and distance from a comprehensive cancer center. After controlling for these individual- and area-level factors, no significant difference was found in the odds of distant stage of disease for Black participants compared with White participants (OR, 0.49; 95% CI, 0.08-3.04). Participants with a diagnosis of COPD were less likely to receive a diagnosis of distant-stage lung cancer (OR, 0.72; 95% CI, 0.53-0.98). Although not statistically significant, participants with health insurance also showed lower odds of distant stage at diagnosis than those without insurance (OR, 0.81; 95% CI, 0.64-1.02). Additionally, results suggest that for an individual residing 100 miles from a comprehensive cancer center, the odds of distant stage at diagnosis were 1.08 times higher when compared with an individual residing 0 miles from a comprehensive cancer center, but this result also was not statistically significant. Furthermore, the model indicated nonlinear relationships between race and comorbidity status and between race and area deprivation. Table 3 presents results from the adjusted primary analysis model. Figure 2 displays ORs of interest (and their CIs) from the primary model for visual inspection. This is one way to visualize and compare the estimated effects that are driving model predictions. For example, visual inspection of Figure 2 shows that the estimated OR for insurance coverage is < 1, suggesting that individuals with insurance coverage had lower odds of distant stage at diagnosis than those without insurance coverage.
Table 3.
Adjusted Associations Between Individual- and Area-Level Characteristics and Distant-Stage Lung Cancer Diagnosis Among Patients With Incident Lung Cancer in the Southern Community Cohort Study, Primary Analysis of Imputed Baseline Data, 2002 through 2019 (N = 1,572)
| Characteristic | ORa | 95% CI | P Value |
|---|---|---|---|
| Main effects | |||
| Black/African American race (reference: White) | 0.49 | 0.08-3.04 | .441 |
| Age at diagnosis | 0.98 | 0.96-1.00 | .060 |
| Male sex (reference: female sex) | 1.32 | 0.91-1.93 | .145 |
| Education (reference: less than high school) | |||
| High school | 1.23 | 0.97-1.58 | .090 |
| More than high school | 1.28 | 0.97-1.68 | .084 |
| Household income ≥ $15,000 (reference: < $15,000) | 0.92 | 0.72-1.17 | .493 |
| Marital status (reference: married) | |||
| Separated or divorced | 0.94 | 0.62-1.42 | .780 |
| Widowed | 1.53 | 0.88-2.65 | .132 |
| Single | 1.08 | 0.57-2.06 | .811 |
| Health insurance coverage (reference: no coverage) | 0.81 | 0.64-1.02 | .068 |
| Pack-years smoked | 1.00 | 1.00-1.00 | .801 |
| History of COPD (reference: no history) | 0.72 | 0.53-0.98 | .040 |
| Prior cancer diagnosis (reference: no diagnosis) | 0.98 | 0.69-1.40 | .930 |
| Charlson comorbidity index score (reference: 1) | |||
| 2 | 1.46 | 0.73-2.97 | .290 |
| 3 | 1.70 | 0.84-3.49 | .145 |
| 4 | 1.91 | 0.94-3.94 | .076 |
| 5 | 2.11 | 0.98-4.61 | .059 |
| ≥ 6 | 1.31 | 0.58-2.98 | .521 |
| Underwent colonoscopy (reference: never underwent) | 0.93 | 0.73-1.19 | .576 |
| Area deprivation (reference: quartile 1, least deprived) | |||
| Quartile 2 | 1.29 | 0.87-1.91 | .207 |
| Quartile 3 | 0.82 | 0.48-1.38 | .464 |
| Quartile 4 (most deprived) | 2.14 | 1.00-4.68 | .051 |
| Distance from comprehensive cancer center, 100-mile increments | 1.08 | 0.95-1.23 | .224 |
| Interaction effectsb | |||
| Black/African American × age at diagnosis | 1.02 | 0.99-1.05 | .193 |
| Black/African American × male sex | 0.98 | 0.62-1.54 | .925 |
| Black/African American × separated or divorced | 1.23 | 0.73-2.07 | .430 |
| Black/African American × widowed | 0.86 | 0.42-1.76 | .682 |
| Black/African American × single | 1.39 | 0.67-2.91 | .377 |
| Black/African American × area deprivation quartile 2 | 1.00 | 0.55-1.85 | .989 |
| Black/African American × area deprivation quartile 3 | 2.37 | 1.21-4.71 | .013 |
| Black/African American × area deprivation quartile 4 | 0.72 | 0.30-1.71 | .456 |
| Black/African American × Charlson comorbidity index score 2 | 0.54 | 0.24-1.18 | .125 |
| Black/African American × Charlson comorbidity index score 3 | 0.44 | 0.19-0.98 | .046 |
| Black/African American × Charlson comorbidity index score 4 | 0.36 | 0.15-0.81 | .015 |
| Black/African American × Charlson comorbidity index score 5 | 0.43 | 0.17-1.06 | .067 |
| Black/African American × Charlson comorbidity index score ≥ 6 | 0.70 | 0.26-1.91 | .492 |
Model fit statistics: cross-validated area under the receiver operating characteristic curve, 0.561; pseudo R2 = 0.045.
Interaction effect between two variables.
Figure 2.
Coefficient plot of features that add to model predictive power among patients with incident lung cancer in the Southern Community Cohort Study, primary analysis, 2002 through 2019 (N = 1,572). The reference categories for these variables are: Black/African American reference of White, education reference of less than high school, deprivation index quartile reference of quartile 1, and comorbidity index reference of score = 1. The ":" symbol represents an interaction effect between two variables.
The subset analysis model was almost identically predictive when compared with the primary analysis model (cross-validated area under the receiver operating characteristic curve, 0.590 vs 0.561, respectively). In the subset analysis, the following variables remained in the model after LASSO shrinkage: race, sex, education, income, marital status, Charlson comorbidity index score, racial discrimination experience, adverse adult experience, childhood living situation (eg, whether the participant lived with their father only during childhood), distance from a comprehensive cancer center, distance from a hospital, and residence in a metro or nonmetro area. Interaction terms that were identified as important contributors to the model’s predictive validity included: male sex × lived with grandparent, aunt, uncle, or other relative during childhood; Charlson comorbidity index score of ≥ 6 × Black or African American race; and widowed marital status × had adult adverse experience. Table 4 presents the results from the subset analysis of the follow-up data, and Figure 3 displays ORs of interest from this subset model. Although not statistically significant, the model fit predicts that an individual 100 miles from a hospital is 2.15 times more likely to receive a diagnosis of distant-stage lung cancer than an individual 0 miles from a hospital, assuming all other variables held constant.
Table 4.
Adjusted Associations Between Individual- and Area-Level Characteristics and Distant-Stage Lung Cancer Diagnosis Among Patients With Incident Lung Cancer in the Southern Community Cohort Study, Subset Analysis of Imputed Follow-up Data, 2002 through 2019 (n = 472)
| Characteristic | ORa | 95% CI | P Value |
|---|---|---|---|
| Main effects | |||
| Black/African American race (reference: White) | 1.47 | 0.40-5.95 | .570 |
| Male sex (reference: female sex) | 1.70 | 0.79-3.73 | .177 |
| Education (reference: less than high school) | |||
| High school | 1.39 | 0.68-2.89 | .365 |
| More than high school | 1.49 | 0.70-3.18 | .300 |
| Household income ≥ $15,000 (reference: < $15,000) | 1.04 | 0.64-1.70 | .863 |
| Marital status (reference: married) | |||
| Separated or divorced | 1.08 | 0.46-2.59 | .859 |
| Widowed | 0.53 | 0.15-1.73 | .315 |
| Single | 1.86 | 0.73-4.81 | .197 |
| Charlson comorbidity index score (reference: 1) | |||
| 2 | 1.76 | 0.49-7.00 | .399 |
| 3 | 1.1 | 0.30-4.42 | .889 |
| 4 | 0.51 | 0.12-2.22 | .360 |
| 5 | 0.81 | 0.20-3.52 | .775 |
| ≥ 6 | 0.42 | 0.09-2.02 | .279 |
| Ever experienced racial discrimination | 0.56 | 0.18-1.55 | .287 |
| Had adult adverse experience | 0.97 | 0.40-2.32 | .949 |
| Lived with mother and father during childhood | 0.83 | 0.50-1.37 | .456 |
| Lived with father only during childhood | 0.81 | 0.04-6.95 | .866 |
| Lived with grandparent, aunt, uncle, or other relative during childhood | 1.81 | 0.71-4.58 | .210 |
| Distance from comprehensive cancer center, 100-mile increments | 1.06 | 0.79-1.41 | .690 |
| Distance from hospital, 100-mile increments | 2.15 | 0.30-15.13 | .441 |
| Nonmetro rural-urban continuum code (reference: metro) | 0.65 | 0.17-2.38 | .517 |
| Interaction effectsb | |||
| Male × separated or divorced | 0.96 | 0.32-2.85 | .939 |
| Male × widowed | 1.10 | 0.11-8.85 | .932 |
| Male × single | 1.21 | 0.36-4.15 | .757 |
| Male × lived with grandparent, aunt, uncle, or other relative during childhood | 0.07 | 0.01-0.36 | .004 |
| Separated or divorced × had adult adverse experience | 0.73 | 0.23-2.35 | .599 |
| Widowed × had adult adverse experience | 8.06 | 1.67-42.64 | .011 |
| Single × had adult adverse experience | 0.61 | 0.16-2.33 | .468 |
| Lived with father only during childhood × household income ≥ $15,000 | 9.43 | 0.44-496.30 | .190 |
| Charlson comorbidity index score 2 × Black/African American race | 0.50 | 0.10-2.21 | .367 |
| Charlson comorbidity index score 3 × Black/African American race | 0.35 | 0.07-1.62 | .185 |
| Charlson comorbidity index score 4 × Black/African American race | 1.40 | 0.25-7.68 | .698 |
| Charlson comorbidity index score 5 × Black/African American race | 1.01 | 0.16-6.08 | .995 |
| Charlson comorbidity index score 6 × Black/African American race | 5.40 | 0.73-41.51 | .099 |
| Black/African American race × distance from hospital | 1.30 | 0.26-6.93 | .757 |
| Ever experienced racial discrimination × Black/African American race | 1.86 | 0.58-6.56 | .308 |
| High school education × nonmetro rural-urban continuum code | 2.35 | 0.54-10.60 | .257 |
| More than high school education × nonmetro rural-urban continuum code | 2.33 | 0.58-9.67 | .237 |
| High school education × distance from hospital | 0.40 | 0.05-3.07 | .388 |
| More than high school education × distance from hospital | 0.57 | 0.08-3.81 | .559 |
| Distance from comprehensive cancer center × nonmetro rural-urban continuum code | 0.97 | 0.48-1.93 | .924 |
Model fit statistics: cross-validated area under the receiver operating characteristic curve, 0.590; pseudo R2 = 0.138.
Interaction effect between two variables.
Figure 3.
Coefficient plot of features that add to model predictive power among patients with incident lung cancer in the Southern Community Cohort Study, subset analysis of follow-up data, 2002 through 2019 (n = 472). The reference categories for these variables are: Black/African American reference of White, male reference of female, widowed reference of married, and comorbidity index reference of score = 1. The ":" symbol represents an interaction effect between two variables.
Discussion
This study assessed factors contributing to racial disparities in lung cancer stage at diagnosis among a largely medically underserved and low-income population. Overall, Black participants in the SCCS showed 29% greater odds of distant stage of lung cancer at diagnosis compared with White participants. However, a racial disparity in stage at diagnosis was not evident after adjusting for individual- and area-level factors. Of note, participants with diagnosis of COPD showed lower odds of distant stage of disease at diagnosis. Because COPD is a risk factor for lung cancer,23 it is possible that individuals with COPD and their physicians carefully monitored lung cancer symptoms, possibly leading to an earlier diagnosis.
Although the primary analysis and subset analysis of follow-up data resulted in models that were not highly predictive, we found that the subset analysis model containing psychosocial variables showed a small increase in predictive power. This finding suggests that psychosocial variables (eg, racial discrimination, adult adverse experiences, and childhood living experiences) may be important to focus on in research studies assessing lung cancer stage at diagnosis. Indeed, assessing psychosocial variables may improve model prediction (eg, by establishing new associations or reducing variability in other relationships in the model). Future studies should consider measuring these variables and should test specific hypotheses regarding how these factors may impact lung cancer outcomes.
Fundamental cause theory offers a framework for explaining and understanding why we did not observe racial disparities in stage of lung cancer at diagnosis in models adjusting for individual- and area-level factors. Link and Phelan24 proposed fundamental cause theory to explain persistent socioeconomic health inequities. The theory posits that those with higher socioeconomic status are better able to avoid morbidity and mortality risks using their more readily accessible flexible resources (eg, knowledge, money, power, prestige, and social connections). As such, when new technological innovations arise that help people avoid health risks (eg, new cancer screening and treatment methods), people with higher socioeconomic status will have better access to the innovation given their available resources. Recently, the authors extended the original theory to add that racism is also a fundamental cause of health inequities because racism causes racial differences in socioeconomic status.25 Accordingly, fundamental cause theory has been used to explain persistent racial and socioeconomic inequities in mortality for cancer types where screening tools exist to identify cancer at earlier stages (eg, colorectal).26, 27, 28 Importantly, although many studies have found that socioeconomic status explains some of the observed racial differences in various health outcomes, evidence also exists that racism causes racial differences in health outcomes and mortality rates independent of socioeconomic status.25,29, 30, 31, 32
For cancer types for which no established screening tool exists, racial disparities in cancer stage at diagnosis may be less evident because no screening tool exists that more privileged people can access differentially to reduce their risk of distant stage at diagnosis. Kim and colleagues33 tested this hypothesis by assessing racial differences in stage at diagnosis and mortality for ovarian cancer, a cancer type that lacks an established screening tool. Kim and colleagues found that race was not associated with stage at diagnosis, nor were most other variables in their model. It is possible that, similar to Kim and colleagues, we did not find racial disparities in stage at lung cancer diagnosis in our adjusted models because participants lacked access to a screening method for lung cancer. Although the landmark National Lung Screening Trial demonstrated that low-dose CT scan screening reduces lung cancer-specific mortality,34 the Centers for Medicare and Medicaid Services did not issue a coverage determination for lung cancer screening until 2015.35 Additionally, uptake of lung cancer screening in the United States is extremely low, with recent studies estimating uptake to be as low as 5% to 6%.36 As such, the general inaccessibility of screening during the study period (follow-up ended in 2019) may explain why we did not observe racial disparities in stage at diagnosis in adjusted models. Future research is needed to confirm these findings.
Our study has important limitations. The SCCS largely recruited low-income participants from community health centers in the southeastern United States. Although this sampling approach identified participants disproportionately experiencing health inequities in the United States, the design may have controlled for population variation and may have reduced our ability to detect racial disparities in stage at diagnosis, and thus may have limited the generalizability of our findings. Additionally, using participant residential addresses from a single time point to assess area-level factors, such as neighborhood deprivation, limits a more comprehensive assessment of potential neighborhood-level exposures. However, prior research found that most SCCS participants had identical neighborhood deprivation index values at baseline and follow-up.37 Furthermore, the baseline survey did not assess potentially important psychosocial variables (eg, racial discrimination experiences), which may have limited the predictive power of our primary analysis model. Despite these limitations, this study is the first, to our knowledge, to investigate racial disparities in lung cancer stage at diagnosis within a cohort of low-income adults living in the southeastern United States, where lung cancer survival rates are the poorest in the country.38
Conclusions
In a study of medically underserved adults with low income living in the southeastern United States, we found that a greater percentage of Black participants than White participants received a diagnosis of distant-stage lung cancer, although a racial disparity was not apparent after adjusting for individual- and area-level factors. These findings have important implications for clinicians and public health practitioners aiming to reduce lung cancer mortality and to facilitate diagnosis at an earlier stages of the disease. Efforts to promote lung cancer screening may benefit privileged populations disproportionately (eg, those with health insurance and reliable transportation) who can access screening readily.39 Without equitable lung cancer screening uptake, racial differences in stage at diagnosis may widen and exacerbate already existing racial disparities in lung cancer mortality and survival. For example, the 5-year relative survival rate for lung cancer is 22% among White Americans, but only 20% among Black Americans; this difference likely will widen if Black Americans do not have equitable access to early detection of lung cancer with screening.1 Recent evidence suggests Black people and people with low socioeconomic status already are less likely to receive lung cancer screening when eligible, showcasing the urgency of efforts to promote equitable screening uptake for early detection of disease.40 Research about novel ways to increase lung cancer screening uptake among diverse populations is growing (eg, hiring patient navigators and community health workers to support patients experiencing barriers to care),41, 42, 43, 44, 45 but more research is needed to promote equity in receipt of lung cancer screening. As lung cancer screening uptake improves and it becomes increasingly possible for people with more privilege and resources to reduce their risk of distant stage at diagnosis, researchers should explore whether racial disparities in lung cancer stage at diagnosis widen and are influenced by fundamental causes and psychosocial factors (eg, neighborhood deprivation, residential racial segregation, and racism in the health-care system). Indeed, racism and implicit biases among clinicians may reduce the likelihood that Black and low-income patients receive lung cancer screening referrals if harmful assumptions are made about patients’ willingness, ability, or eligibility to undergo screening. Furthermore, patients who experience racism—especially in health-care settings—may mistrust medical providers and be less willing to undergo lung cancer screening.46,47 Overall, the impact of psychosocial factors and structural racism on racial disparities in lung cancer stage at diagnosis deserves further attention.
Interpretation
A greater percentage of Black participants than White participants in this study received a diagnosis of distant-stage lung cancer, although a disparity was not apparent after adjusting for individual- and area-level factors. Future research is needed to investigate how psychosocial factors and structural racism are associated with racial disparities in lung cancer stage at diagnosis.
Funding/Support
Research reported in this publication was supported a grant from the Lung Cancer Research Foundation. This research was also supported by the National Cancer Institute, National Institutes of Health [Grant U01CA202979]. SCCS data collection was performed by the Survey and Biospecimen Shared Resource, which is supported in part by the Vanderbilt-Ingram Cancer Center [National Institutes of Health Grant P30 CA68485]. J. R. is supported by the Agency for Healthcare Research and Quality [Grant T32HS026122] and a Loan Repayment Award from the National Cancer Institute [Grant L60CA264691]. M. C. A. is funded by the National Cancer Institute [Grants 1U01CA253560 and 5R01CA251758].
Financial/Nonfinancial Disclosures
The authors have reported to CHEST the following: M. C. A. serves as an advisor for Guardant Health. None declared (J. R., M. H. M., C. M. M., J. D. B.).
Acknowledgments
Author contributions: M. C. A. takes responsibility for the content of the manuscript. J. R. contributed to the manuscript in the following areas: methodology, writing the original draft, reviewing and editing the draft, and visualization. M. H. M. contributed to the manuscript in the following areas: methodology, formal analysis, data collection, reviewing and editing of the manuscript, and visualization. C. M. M. contributed to the manuscript in the following areas: methodology and reviewing and editing the manuscript. J. D. B. contributed to the manuscript in the following areas: conceptualization, methodology, resource gathering, reviewing and editing the manuscript, supervision, and funding acquisition. M. C. A. contributed to the manuscript in the following areas: conceptualization, methodology, resource gathering, reviewing and editing the manuscript, supervision, project administration, and funding acquisition. All authors reviewed and approved the version submitted.
Role of sponsors: The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of this manuscript. The content is solely the responsibility of the authors and does not necessarily represent the official views of the funding organizations.
Additional information: The e-Figure and e-Table are available online under “Supplementary Data.”
Supplementary Data
References
- 1.Siegel R.L., Miller K.D., Fuchs H.E., Jemal A. Cancer statistics, 2022. CA Cancer J Clin. 2022;72(1):7–33. doi: 10.3322/caac.21708. [DOI] [PubMed] [Google Scholar]
- 2.Jones C.C., Mercaldo S.F., Blume J.D., et al. Racial disparities in lung cancer survival: the contribution of stage, treatment, and ancestry. J Thorac Oncol. 2018;13(10):1464–1473. doi: 10.1016/j.jtho.2018.05.032. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Walters S., Maringe C., Coleman M.P., et al. Lung cancer survival and stage at diagnosis in Australia, Canada, Denmark, Norway, Sweden and the UK: a population-based study, 2004-2007. Thorax. 2013;68(6):551–564. doi: 10.1136/thoraxjnl-2012-202297. [DOI] [PubMed] [Google Scholar]
- 4.Miller K.D., Nogueira L., Devasia T., et al. Cancer treatment and survivorship statistics, 2022. CA Cancer J Clin. 2022;72(5):409–436. doi: 10.3322/caac.21731. [DOI] [PubMed] [Google Scholar]
- 5.Giaquinto AN, Miller KD, Tossas KY, Winn RA, Jemal A, Siegel RL. Cancer statistics for African American/Black people 2022 [published online ahead of print February 10, 2022]. CA Cancer J Clin. https://doi.org/10.3322/caac.21718. [DOI] [PubMed]
- 6.Soneji S., Tanner N.T., Silvestri G.A., Lathan C.S., Black W. Racial and ethnic disparities in early-stage lung cancer survival. Chest. 2017;152(3):587–597. doi: 10.1016/j.chest.2017.03.059. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Hardy D., Xia R., Liu C.C., Cormier J.N., Nurgalieva Z., Du X.L. Racial disparities and survival for nonsmall-cell lung cancer in a large cohort of black and white elderly patients. Cancer. 2009;115(20):4807–4818. doi: 10.1002/cncr.24521. [DOI] [PubMed] [Google Scholar]
- 8.Landrine H., Corral I., Lee J.G.L., Efird J.T., Hall M.B., Bess J.J. Residential segregation and racial cancer disparities: a systematic review. J Racial Ethn Health Disparities. 2017;4(6):1195–1205. doi: 10.1007/s40615-016-0326-9. [DOI] [PubMed] [Google Scholar]
- 9.Adie Y., Kats D.J., Tlimat A., et al. Neighborhood disadvantage and lung cancer incidence in ever-smokers at a safety net health-care system: a retrospective study. Chest. 2020;157(4):1021–1029. doi: 10.1016/j.chest.2019.11.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Johnson A.M., Johnson A., Hines R.B., Mohammadi R. Neighborhood context and non-small cell lung cancer outcomes in Florida non-elderly patients by race/ethnicity. Lung Cancer. 2020;142:20–27. doi: 10.1016/j.lungcan.2020.01.012. [DOI] [PubMed] [Google Scholar]
- 11.Lathan C.S. Lung cancer care: the impact of facilities and area measures. Transl Lung Cancer Res. 2015;4(4):385–391. doi: 10.3978/j.issn.2218-6751.2015.07.23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Shah B.D., Tyan C.C., Rana M., et al. Rural vs urban inequalities in stage at diagnosis for lung cancer. Cancer Treat Res Commun. 2021;29 doi: 10.1016/j.ctarc.2021.100495. [DOI] [PubMed] [Google Scholar]
- 13.Signorello L.B., Hargreaves M.K., Blot W.J. The Southern Community Cohort Study: investigating health disparities. J Health Care Poor Underserved. 2010;21(1 suppl):26–37. doi: 10.1353/hpu.0.0245. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Signorello L.B., Hargreaves M.K., Steinwandel M.D., et al. Southern Community Cohort Study: establishing a cohort to investigate health disparities. J Natl Med Assoc. 2005;97(7):972–979. [PMC free article] [PubMed] [Google Scholar]
- 15.Southern Community Cohort Study Codebooks and documentation. Southern Community Cohort Study website. https://www.southerncommunitystudy.org/codebooks-and-documentation.html
- 16.Charlson M., Szatrowski T.P., Peterson J., Gold J. Validation of a combined comorbidity index. J Clin Epidemiol. 1994;47(11):1245–1251. doi: 10.1016/0895-4356(94)90129-5. [DOI] [PubMed] [Google Scholar]
- 17.Austin S.R., Wong Y.-N., Uzzo R.G., Beck J.R., Egleston B.L. Why summary comorbidity measures such as the Charlson comorbidity index and Elixhauser score work. Med Care. 2015;53(9):e65–e72. doi: 10.1097/MLR.0b013e318297429c. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Aldrich M.C., Mercaldo S.F., Sandler K.L., Blot W.J., Grogan E.L., Blume J.D. Evaluation of USPSTF lung cancer screening guidelines among African American adult smokers. JAMA Oncol. 2019;5(9):1318–1324. doi: 10.1001/jamaoncol.2019.1402. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Messer L.C., Laraia B.A., Kaufman J.S., et al. The development of a standardized neighborhood deprivation index. J Urban Health. 2006;83(6):1041–1062. doi: 10.1007/s11524-006-9094-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Tibshirani R. Regression shrinkage and selection via the Lasso. J R Stat Soc Series B Stat Methodol. 1996;58(1):267–288. [Google Scholar]
- 21.Harrell F.E. Regression Modeling Strategies: With Applications to Linear Models, Logistic and Ordinal Regression, and Survival Analysis. Second ed. Springer International Publishing; 2015. Missing data; pp. 45–61. [Google Scholar]
- 22.van Buuren S., Groothuis-Oudshoorn K. mice: Multivariate imputation by chained equations in R. J Stat Softw. 2011;45(3):1–67. [Google Scholar]
- 23.Sekine Y., Katsura H., Koh E., Hiroshima K., Fujisawa T. Early detection of COPD is important for lung cancer surveillance. Eur Respir J. 2012;39(5):1230–1240. doi: 10.1183/09031936.00126011. [DOI] [PubMed] [Google Scholar]
- 24.Link B.G., Phelan J. Social conditions as fundamental causes of disease. J Health Soc Behav. 1995;spec. no.:80–94. [PubMed] [Google Scholar]
- 25.Phelan J.C., Link B.G. Is racism a fundamental cause of inequalities in health? Annu Rev of Sociol. 2015;41(1):311–330. [Google Scholar]
- 26.Wang A., Clouston S.A., Rubin M.S., Colen C.G., Link B.G. Fundamental causes of colorectal cancer mortality: the implications of informational diffusion. Milbank Q. 2012;90(3):592–618. doi: 10.1111/j.1468-0009.2012.00675.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Clouston S.A.P., Link B.G. A retrospective on fundamental cause theory: state of the literature and goals for the future. Annu Rev Sociol. 2021;47(1):131–156. doi: 10.1146/annurev-soc-090320-094912. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Zapata-Moya Á.R., Willems B., Bracke P. The (re)production of health inequalities through the process of disseminating preventive innovations: the dynamic influence of socioeconomic status. Health Sociology Review. 2019;28(2):177–193. [Google Scholar]
- 29.Kehm R.D., Spector L.G., Poynter J.N., Vock D.M., Altekruse S.F., Osypuk T.L. Does socioeconomic status account for racial and ethnic disparities in childhood cancer survival? Cancer. 2018;124(20):4090–4097. doi: 10.1002/cncr.31560. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Williams D.R., Priest N., Anderson N.B. Understanding associations among race, socioeconomic status, and health: patterns and prospects. Health Psychol. 2016;35(4):407–411. doi: 10.1037/hea0000242. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Franks P., Muennig P., Lubetkin E., Jia H. The burden of disease associated with being African-American in the United States and the contribution of socio-economic status. Soc Sci Med. 2006;62(10):2469–2478. doi: 10.1016/j.socscimed.2005.10.035. [DOI] [PubMed] [Google Scholar]
- 32.Gupta A, Omeogu CH, Islam JY, Joshi AR, Akinyemiju TF. Association of area-level socioeconomic status and non-small cell lung cancer stage by race/ethnicity and health care-level factors: analysis of the National Cancer Database [published online ahead of print June 20, 2022]. Cancer. https://doi.org/10.1002/cncr.34327. [DOI] [PMC free article] [PubMed]
- 33.Kim S., Dolecek T.A., Davis F.G. Racial differences in stage at diagnosis and survival from epithelial ovarian cancer: a fundamental cause of disease approach. Soc Sci Med. 2010;71(2):274–281. doi: 10.1016/j.socscimed.2010.03.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.National Lung Screening Trial Research Team Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med. 2011;365(5):395–409. doi: 10.1056/NEJMoa1102873. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Centers for Medicare and Medicaid Services Decision memo for screening for lung cancer with low dose computed tomography (LDCT) (CAG-00439N). Centers for Medicare and Medicaid Services website. https://www.cms.gov/medicare-coverage-database/details/nca-decision-memo.aspx?NCAId=274
- 36.Fedewa S.A., Kazerooni E.A., Studts J.L., et al. State variation in low-dose CT scanning for lung cancer screening in the United States. J Natl Cancer Inst. 2020;113(8):1044–1052. doi: 10.1093/jnci/djaa170. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Sanderson M., Aldrich M.C., Levine R.S., Kilbourne B., Cai Q., Blot W.J. Neighbourhood deprivation and lung cancer risk: a nested case-control study in the USA. BMJ Open. 2018;8(9) doi: 10.1136/bmjopen-2017-021059. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Lee Y.-C., Calderon-Candelario R.A., Holt G.E., Campos M.A., Mirsaeidi M. State-level disparity in lung cancer survival in the United States. Original research. Front Oncol. 2020;10 doi: 10.3389/fonc.2020.01449. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Rivera M.P., Katki H.A., Tanner N.T., et al. Addressing disparities in lung cancer screening eligibility and healthcare access. An official American Thoracic Society statement. Am J Respir Crit Care Med. 2020;202(7):e95–e112. doi: 10.1164/rccm.202008-3053ST. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Sosa E, D’Souza G, Akhtar A, et al. Racial and socioeconomic disparities in lung cancer screening in the United States: a systematic review [published online ahead of print May 20, 2021]. CA Cancer J Clin. https://doi.org/10.3322/caac.21671. [DOI] [PMC free article] [PubMed]
- 41.Wang G.X., Baggett T.P., Pandharipande P.V., et al. Barriers to lung cancer screening engagement from the patient and provider perspective. Radiology. 2019;290(2):278–287. doi: 10.1148/radiol.2018180212. [DOI] [PubMed] [Google Scholar]
- 42.Triplette M., Thayer J.H., Pipavath S.N., Crothers K. Poor uptake of lung cancer screening: opportunities for improvement. J Am Coll Radiol. 2019;16(4 Pt A):446–450. doi: 10.1016/j.jacr.2018.12.018. [DOI] [PubMed] [Google Scholar]
- 43.Watson K.S., Siegel L.D., Henderson V.A., et al. The SHARED Project: a novel approach to engaging African American men to address lung cancer disparities. Am J Mens Health. 2020;14(5) doi: 10.1177/1557988320958934. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Williams L.B., McCall A., Joshua T.V., Looney S.W., Tingen M.S. Design of a community-based lung cancer education, prevention, and screening program. West J Nurs Res. 2019;41(8):1152–1169. doi: 10.1177/0193945919827261. [DOI] [PubMed] [Google Scholar]
- 45.Brown L.R., Sullivan F., Treweek S., et al. Increasing uptake to a lung cancer screening programme: building with communities through co-design. BMC Public Health. 2022;22(1):815. doi: 10.1186/s12889-022-12998-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Carter-Harris L., Slaven J.E., Monahan P.O., Draucker C.B., Vode E., Rawl S.M. Understanding lung cancer screening behaviour using path analysis. J Med Screen. 2019;27(2):105–112. doi: 10.1177/0969141319876961. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Griffith D.M., Bergner E.M., Fair A.S., Wilkins C.H. Using mistrust, distrust, and low trust precisely in medical care and medical research advances health equity. Am J Prev Med. 2021;60(3):442–445. doi: 10.1016/j.amepre.2020.08.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.