Abstract
Background: Despite overall incidence reduction in colorectal cancer (CRC) the past 32 years, unexplained incidence and mortality rates have increased significantly in younger adults ages 20-49. To improve understanding of sex-specific differences among this population, we aimed to determine the variance in early-onset CRC (EOCRC) survival among US men diagnosed with CRC before age 50, while considering individual- and county-level CRC outcome determinants. Methods: Hotspots (i.e., counties with high EOCRC mortality rates) were derived from Centers for Disease Control and Prevention data from 1999-2017, and linked to SEER data for men aged 15-49 years with CRC. Cox proportional hazards models were used to compare CRC-specific survival probability and hazard in hotspots versus non-significant counties. A generalized R2 was used to estimate the total variance in EOCRC survival explained by clinicodemographic and county-level determinants. Results: We identified 232 hotspot counties for EOCRC-214 (92%) of which were in the South. In hotspots, 1,009 men were diagnosed with EOCRC and 31,438 in non-significant counties. After adjusting for age, race, tumor stage and grade, surgery, chemotherapy, radiation therapy, and marital status, men residing in hotspot counties had higher hazard of CRC-specific death (HR 1.24, 95% CI, 1.12-1.36). Individual/county-level factors explained nearly 35% of the variation in survival, and adult smoking served as the strongest county-level determinant of EOCRC survival. Conclusion: Distinct geographic patterns of EOCRC were predominantly located in the southern US. Survival after EOCRC diagnosis was significantly worse among men residing in hotspot counties.
Keywords: Colorectal cancer, early-onset, racial disparities, SEER, smoking
Introduction
Colorectal cancer (CRC) affects 1 in 23 men and 1 in 25 women in their lifetime [1]. Among individuals aged ≥50, declining CRC incidence and mortality rates are attributable partly to increased CRC screening utilization and adjuvant therapy advancements [2,3]. However, these reductions in disease burden have coincided with increased CRC incidence among individuals aged <50 (early-onset CRC [EOCRC]), such that individuals born circa 1990 have double and quadruple the risk of colon and rectal cancers, respectively, compared with similarly aged adults born around 1950 [4]. While approximately 1 in 10 new CRC diagnoses affect young individuals, the causes underlying this increase in EOCRC incidence remain unexplained [2-13]. Moreover, it is predicted that by 2030, CRC incidence rates will increase by 90%-124% among Americans aged 20-34 and by 28%-46% among those aged 35-49 [11].
EOCRC harbors a distinct clinical and molecular phenotype compared with CRC seen among individuals aged ≥50 [14-18]. Studies have demonstrated that individuals diagnosed with EOCRC present with more advanced and aggressive stages of the disease [3,5,6,19], while others have reported that younger individuals have better outcomes compared with their older counterparts [5,6,19]. These complex findings may be attributed to differences in environmental, geographical, and lifestyle factors (e.g., diet, obesity, sedentary behaviors), as well as sex and race/ethnicity [20].
In the US, CRC incidence and survival differ by sex, race/ethnicity, and geography. Regarding sex-specific differences, CRC incidence rates among men are nearly one-third higher than that of women, with mortality rates also 40% higher in men than in women [2]. Racial/ethnic disparities have grown more pronounced [3,7,8,21], with survival after CRC diagnosis poorer among African Americans/Blacks compared with Whites, even among patients with early-stage CRC [9,22-24]. In particular, non-Hispanic (NH) Black men have the lowest five-year CRC survival and age-adjusted mortality rates across all racial/ethnic and sex subgroups [25]. Geographically, patterns of CRC incidence and mortality have shifted over time across the US. Once highest in the Northeast, CRC mortality rates are now highest in the South and Midwest-a shift largely explained by higher birthrates and poorer economic status among Southern and Midwestern Blacks [2]. This shift holds true for EOCRC, with previous studies [26-30] identifying similar differences in geographical regions. Moreover, previous research by Siegel et al. derived contemporary spatial clusters of US counties with high CRC rates based on county-level mortality data from 1970-2011 and identified three distinct hotspots [30], yet failed to elucidate factors contributing to the troubling and sharp rise in EOCRC. Accordingly, our study aimed to identify mortality hotspots specific to men with EOCRC-a missed opportunity to improve our understanding of EOCRC disparities while controlling for sex-specific differences. Further, we examined differences in the individual- and county-level characteristics between EOCRC hotspots and non-hotspots, as well as study factors that explain better survival among men residing within EOCRC hotspots.
Material and methods
EOCRC hotspots
To identify EOCRC hotspots, we obtained county-level estimates using our previously described geospatial methodology [31]. Using data from the Centers for Disease Control and Prevention’s (CDC’s) underlying causes of death file [32], EOCRC deaths were defined as deaths among US residents aged 15-54 from 1999-2017. (Those aged 50-54 were included to account for patients diagnosed at age 49 with standardized 5-year follow-ups). EOCRC county-level frequencies, crude rates, and age-adjusted rates were identified using International Classification of Diseases, Tenth Revision (ICD-10) codes for colon- and rectum-specific cancers (Supplementary Table 1). Geospatial analyses were performed using three geospatial autocorrelation measures: empirical Bayes (EB) smoothed EOCRC mortality rates, local indicators of spatial association (LISA), and the Getis-Ord Gi* statistic [33-35]. Contiguous US counties were categorized as hotspots if they had high rates of EOCRC mortality based on all three geospatial methodologies (i.e., in the fifth quintile of smoothed EB EOCRC mortality rates (Supplementary Figure 1), a high-high cluster using LISA (Supplementary Figure 2), and an EOCRC hotspot as defined by the Getis-Ord Gi* statistic (Supplementary Figure 3)) [13,35]. All other US counties were categorized as non-significant spots. These categories were subsequently linked to Surveillance, Epidemiology, and End Results (SEER) data [36], with patients residing in either an EOCRC hotspot or a non-significant county according to their county of residence.
Study population
Data were obtained from the National Cancer Institute (NCI) SEER program database (November 2018 submission) [36], which covers nearly one-third of the US population and includes detailed information from 18 population-based registries on demographics, clinical characteristics, and survival for each cancer diagnosis. Study participants were NH-White, NH-Black, and Hispanic adults or adolescents aged 15-49 at primary CRC diagnosis. A total of 32,447 men in the SEER database were diagnosed with EOCRC from 1999-2016, after excluding those diagnosed with a prior malignant cancer (n=2,211), those with an unknown surgical procedure (n=243), those with missing follow-up time (n=28), and those residing in Alaska and Hawaii (n=764). Residents of Alaska and Hawaii were excluded because residence categorization in a geospatial hotspot was dependent upon counties located in the 48 contiguous states.
Statistical analysis
Differences in patient-level characteristics and county-level determinants between men residing in EOCRC hotspots and men living in non-significant counties were examined using Chi-square tests for categorical variables, analysis of variance (ANOVA) for parametric continuous variables, and Wilcoxon rank-sum tests for non-parametric continuous variables. Survival time was calculated from date of diagnosis to the last date of follow-up or the date of death. Patient follow-up was within 22 months of the annual survey submission date (November 2017). A two-sided P-value <0.05 was considered statistically significant. All county population proportion estimates included the total adult population.
Kaplan-Meier curves were used to compare overall hotspot residence survival among NH-Black, NH-White, and Hispanic patients. To determine the association of hotspot residence with CRC survival, multilevel regression models accounting for clustering among SEER registry groups were employed. Adjusted and unadjusted Cox proportional hazards models estimated EOCRC survival. Hazard ratios (HRs) and corresponding 95% confidence intervals (CIs) were estimated. Age was included in adjusted models since the association of patient-level characteristics with residence in a hotspot was not defined a priori. Models were adjusted for patient-level factors found in bivariate analysis to reach statistical significance and in a separate analysis they were also adjusted for smoking at the community level which is known to be associated with cancer mortality. Other community level covariates were not considered in the multivariable modeling to avoid multicollinearity as they all were found to have significant correlation with smoking.
To estimate total EOCRC variance explained by each explanatory factor (clinicodemographic and county-level determinants) and all combined, we calculated generalized R2 using methods adapted for Cox proportional hazards model developed by Allison [37] and based onthe Cox and Snell [38] method. Data were analyzed using SAS v9.4 (SAS Institute Inc., Cary, NC) and GeoDa v1.6.7.9 (https://geodacenter.github.io/), and mapped using ArcGIS v10.5 (Esri, Redlands, CA). Additional methods are detailed in Supplementary Methods. This study was exempt from approval by the Institutional Review Boards of the University of Utah and Augusta University because the datasets used are publicly available.
Results
Hotspot characteristics
Based on geospatial analyses using national mortality data from 1999-2017, a composite hotspot map was created for EOCRC mortality (Figure 1). A total of 232 (7.5%) of 3,108 contiguous US counties were EOCRC hotspots. Overall, 92% (214 of 232) of the hotspot counties were located in the South and 8% (18 of 232) were located in the Midwest (P<0.01; Supplementary Table 2). Hotspot counties and associated EOCRC mortality are outlined in Supplementary Table 3. Approximately 3.11% of men diagnosed with EOCRC in SEER (n=1,009) resided in hotspot counties (Figure 1; Table 1). The mean EOCRC diagnosis age was 42.73 and did not significantly differ by hotspot residence (Table 1). Compared to men living in non-significant counties, men residing in hotspot counties were more likely to be NH-Black (30.82% vs 13.06%) and less likely to be Hispanic (1.68% vs 16.65%; P<0.01). Men diagnosed with EOCRC living in hotspots were significantly less likely to be married or to have a domestic partner compared to counterparts residing in non-hotspot counties (52.1% vs 56.5%, respectively; P<0.01). Approximately one-third of men (28.75%) had right-sided tumors. However, tumor sidedness did not significantly differ by hotspot residential status. Men residing in hotspots were more likely to be diagnosed with metastatic disease (stage IV CRC) compared to those residing in non-significant spots (2.58% vs 1.94%; P<0.01) (Table 1).
Figure 1.
Early-onset colorectal cancer (CRC) mortality hotspots across the contiguous United States: US residents, 1999-2017.
Table 1.
Summary of Demographic and Clinical Characteristics by EOCRC Hotspot Classification Among US Men: SEER18, 1999-2016
| Number of Cases (N=32,447) | EOCRC Hot-spot Residence N (%) | |||
|---|---|---|---|---|
|
|
|
|||
| N (%) or Mean (SE)c | Hotspota | Non-Significant | P-valueb | |
|
| ||||
| N (%) or Mean (SE)c | N (%) or Mean (SE)c | |||
| Total | 32447 (100.0) | 1009 (3.11) | 31438 (96.89) | <.01 |
| Mean survival, monthsd | 128.65 (0.55) | 113.76 (3.01) | 129.04 (0.56) | <.01 |
| Age in years, mean | 42.73 (0.04) | 42.98 (0.21) | 42.73 (0.04) | .23 |
| Age at diagnosis (%) | ||||
| 15-29 years | 1738 (5.35) | 53 (5.26) | 1685 (5.36) | .17 |
| 30-39 years | 6225 (19.19) | 184 (18.24) | 6183 (19.21) | |
| 40-49 years | 24,484 (75.46) | 772 (76.51) | 23,712 (75.42) | |
| Race (%) | ||||
| NH-White | 19,657 (60.58) | 663 (65.71) | 18,994 (60.42) | <.01 |
| NH-Black | 4417 (13.61) | 311 (30.82) | 4106 (13.06) | |
| Hispanic | 5251 (16.18) | 17 (1.68) | 5234 (16.65) | |
| Other | 3122 (9.87) | 18 (1.78) | 3104 (9.63) | |
| Marital status (%) | ||||
| Single or never married | 9478 (29.21) | 291 (28.84) | 9187 (29.22) | <.01 |
| Married or domestic partner | 18,293 (56.38) | 526 (52.13) | 17,767 (56.51) | |
| Divorced, separated, or widowed | 2821 (8.69) | 116 (11.50) | 2705 (8.60) | |
| Unknown | 1855 (5.72) | 76 (7.53) | 1779 (5.66) | |
| AJCC stage (%) | ||||
| 0-I | 5019 (15.47) | 189 (18.73) | 4830 (15.36) | .05 |
| II | 5968 (18.39) | 170 (16.85) | 5798 (18.44) | |
| III | 8551 (26.38) | 260 (25.77) | 8291 (26.37) | |
| IV | 7115 (21.93) | 222 (22.00) | 6893 (28.93) | |
| Unknown | 5794 (17.86) | 168 (16.65) | 5626 (17.90) | |
| Tumor Sidedness (%) | ||||
| Righte | 9329 (28.75) | 288 (28.54) | 9041 (28.76) | .88 |
| Leftf | 23118 (71.25) | 721 (71.46) | 22397 (71.24) | |
| Grade (%) | ||||
| I (well differentiated) | 3060 (9.43) | 65 (6.44) | 2995 (9.53) | <.01 |
| II (moderately differentiated) | 17,822 (54.93) | 581 (57.58) | 17,241 (54.84) | |
| III (poorly differentiated) | 5385 (16.60) | 152 (15.06) | 5233 (16.65) | |
| IV (undifferentiated) | 636 (1.96) | 26 (2.58) | 610 (1.94) | |
| Unknown | 5544 (17.09) | 185 (18.33) | 5359 (17.05) | |
| Surgery (%) | 27,561 (84.94) | 858 (85.03) | 26,703 (84.94) | .93 |
| Chemotherapy (%) | 18,574 (57.24) | 18,005 (57.27) | 569 (56.39) | .58 |
| Radiation Therapy (%) | 7626 (23.50) | 253 (25.07) | 7373 (23.45) | .23 |
Patients residing in counties with high EOCRC mortality rates (fulfilling all three criteria for geographic clustering).
p-value calculations determined using Chi-square, t tests, log-rank test (survival), or Wilcoxon rank-sum test as appropriate. P value calculations do not include unknown values.
Presented as number (column percentage) or mean (standard error).
Calculated using Kaplan-Meier method (or product limit method).
Defined as ICD-O-3 codes 180, 181, 182, 183, and 184.
Defined as ICD-O-3 codes 185, 186, 187, 188, 189, 199, 209, and 260.
Men living in hotspot counties at EOCRC diagnosis were significantly more likely to reside in areas with greater proportions of NH-Whites (66.70% vs 55.05%; P<0.01) and NH-Blacks (27.89% vs 11.77%; P<0.01; Table 2) compared to men living in non-significant counties. Moreover, hotspots were more likely than non-significant counties to have higher poverty rates (26.57% vs 16.77%), greater prevalence of adult obesity (34.94% vs 25.89%), more physical inactivity (32.31% vs 21.63%), fewer exercise opportunities (52.67% vs 83.36%), more limited access to healthy foods (8.58% vs 4.66%), lower college completion rates (16.14% vs 30.68%), higher adult smoking rates (23.97% vs 15.44%), higher uninsured rates (20.06% vs 17.91%), fewer primary care physicians (58.28 vs 75.45 per 100,000 population), increased rurality (43.66% vs 12.70%), and more violent crimes (488.7 vs 406.75 crimes per 100,000 persons) (all P<0.01).
Table 2.
Summary of county-level characteristics by EOCRC hotspot classification among US men: SEER, 1999-2016 linked with 2014 American community survey and county health rankings county-level data
| County Characteristic | EOCRC Hotspot Residence Among All Men | |||
|---|---|---|---|---|
|
| ||||
| Hotspota n=1,009 (3.11%) | Non-significant n=31,438 (96.89%) | P-valueb | Spearman (ρ) Correlation Coefficient | |
|
| ||||
| Presented as Mean (SE)c | ||||
| Race | ||||
| % NH-White | 66.70 (0.58) | 55.05 (0.12) | <0.01 | 0.95 |
| % NH-Black | 27.89 (0.54) | 11.77 (0.07) | <0.01 | 0.14 |
| % Hispanic | 2.62 (0.04) | 21.86 (0.10) | <0.01 | -0.25 |
| % Household income <$20,000 | 26.57 (0.18) | 16.77 (0.03) | <0.01 | 0.24 |
| % Access to exercise opportunities | 52.67 (0.52) | 83.36 (0.11) | <0.01 | -0.24 |
| % Limited access to healthy foods | 8.58 (0.15) | 4.66 (0.02) | <0.01 | 0.14 |
| % Obesity | 34.94 (0.09) | 25.89 (0.03) | <0.01 | 0.26 |
| % Smoking | 23.97 (0.10) | 15.44 (0.03) | <0.01 | 0.24 |
| % Completed college | 16.14 (0.14) | 30.68 (0.06) | <0.01 | -0.24 |
| % Physical inactivity | 32.31 (0.10) | 21.63 (0.03) | <0.01 | 0.27 |
| % Unemployed | 8.11 (0.07) | 9.25 (0.01) | <0.01 | -0.09 |
| % Uninsured | 20.06 (0.05) | 17.91 (0.03) | <0.01 | 0.08 |
| PCPd per 100,000 persons | 58.28 (0.84) | 75.45 (0.15) | <0.01 | -0.10 |
| % Non-urban (rural) | 43.66 (0.83) | 12.70 (0.12) | <0.01 | 0.22 |
| Violent crimes per 100,000 persons | 488.71 (8.86) | 406.75 (1.27) | <0.01 | 0.07 |
County characteristic determined by patient FIPS code. Patients residing in counties with high EOCRC mortality (fulfilling all three criteria for geographic clustering).
Significance determined using Wilcoxon test.
Mean (standard error).
PCP = primary care physicians.
Hot-spots and CRC-specific survival
Men living in hotspots had poorer CRC survival compared with those in non-significant counties (113.76 vs 129.04 months, respectively; P<0.001; Table 1). By race/ethnicity, NH-White men residing in EOCRC hotspots experienced significantly worse CRC survival compared with NH-White men in non-significant counties (P<0.01; Figure 2A). In particular, NH-White men living in hotspots saw a 10-year survival rate of 49.31% (± SE 2.38%), while their non-significant counterparts saw 58.76% (± SE 0.42%) survival during the same time period (data not shown). However, racial disparities in survival by residential hotspot area did not persist among NH-Black or Hispanic men (P=0.59 and P=0.23, respectively; Figure 2B and 2C). Among men diagnosed with EOCRC, those residing in hotspots demonstrated a 24% higher hazard for CRC-specific death compared with those residing in non-significant counties (HR 1.24, 1.12-1.36; Table 3). However, after adjusting for county-level smoking, men in hotspots demonstrated a 12% higher hazard for CRC-specific death (HR 1.12, 1.01-1.24). Compared with NH-White men, NH-Black (HR 1.31, 1.25-1.38) and Hispanic (HR 1.12, 1.07-1.19) patients demonstrated a 31% and 12% increased risk for CRC-specific death, respectively, after adjusting for smoking. Men who underwent surgical resection for CRC had a 61% reduced risk for CRC-specific death compared to those who did not undergo surgery (HR 0.39, 0.37-0.41). In addition, for each unit increase in the county-level proportion of adult smokers, men diagnosed with EOCRC were nearly four times more likely to die from CRC (HR 3.71, 2.52-5.45). When limited to men residing in hotspot counties, later stage (stage II-IV) CRC diagnosis was associated with increased CRC-specific mortality risk, after adjusting for county-level smoking (Table 4). Specifically, compared to stage 0 or I diagnosis, men diagnosed with stage II (HR 2.45, 1.60-3.75), stage III (HR 4.18, 2.76-6.32), and stage IV (HR 10.83, 7.21-16.25) had 2.5 times, 4 times, and nearly 11 times greater risk of CRC-specific mortality, respectively. Furthermore, within hotspot counties, the severity of tumor grade was associated with increased CRC-specific mortality risk, with poorly differentiated tumors (HR 1.87, 1.03-3.40) and undifferentiated tumors (HR 2.60, 1.21-5.61) having nearly 2 times and 2.6 times greater mortality risk compared to well differentiated tumors, respectively. Within hotspot counties, single men (HR 1.43, 1.13-1.79) and those who were divorced, separated, or widowed (HR 1.43, 1.05-1.95) had 43% increased risk for CRC mortality compared to married or coupled men.
Figure 2.
Kaplan-Meier CRC-specific survival curves for (A) non-Hispanic White, (B) non-Hispanic Black, and (C) Hispanic men with EOCRC by hotspot region: SEER18 1999-2015.
Table 3.
Multivariable Cox Proportional Hazards Regression Models for CRC-Specific Death Among Men with EOCRC
| CRC Survival | |||
|---|---|---|---|
|
| |||
| No. of Deaths (%)a | Adjusted HR (95% CI)b,c,d excluding smoking | Adjusted HR (95% CI)b,c,d including smoking | |
| EOCRC hotspot | |||
| Non-significant | 11,183 (34.47) | Ref | Ref |
| Hotspot | 420 (1.29) | 1.24 (1.12-1.36) | 1.12 (1.01-1.24) |
| Age (in years) | |||
| 15-24 | 193 (0.59) | 0.83 (0.72-0.96) | 0.83 (0.71-0.96) |
| 25-29 | 336 (1.04) | 0.87 (0.78-0.97) | 0.87 (0.77-0.97) |
| 30-34 | 675 (2.08) | 0.85 (0.78-0.92) | 0.84 (0.78-0.92) |
| 35-39 | 1442 (4.44) | 0.95 (0.90-1.01) | 0.95 (0.90-1.01) |
| 40-44 | 2714 (8.36) | 0.91 (0.87-0.95) | 0.91 (0.87-0.95) |
| 45-49 | 6243 (24) | Ref | Ref |
| Race/ethnicity | |||
| Non-Hispanic White | 6867 (21.16) | Ref | Ref |
| Non-Hispanic Black | 1987 (6.12) | 1.32 (1.26-1.39) | 1.31 (1.25-1.38) |
| Hispanic | 1814 (5.59) | 1.09 (1.03-1.14) | 1.12 (1.07-1.19) |
| Asian/Pacific Islander | 827 (2.55) | 0.95 (0.88-1.02) | 1.00 (0.93-1.07) |
| Other/Unknown | 108 (0.33) | 0.67 (0.55-0.81) | 0.66 (0.55-0.80) |
| Marital status | |||
| Single or never married | 3934 (12.12) | 1.38 (1.32-1.44) | 1.39 (1.33-1.45) |
| Married or domestic partner | 5886 (18.14) | Ref | Ref |
| Divorced, separated, or widowed | 1286 (3.96) | 1.37 (1.29-1.46) | 1.36 (1.28-1.45) |
| Unknown | 497 (1.53) | 1.02 (0.93-1.12) | 1.01 (0.92-1.11) |
| AJCC stage | |||
| 0 or I | 653 (2.01) | Ref | Ref |
| II | 1258 (3.88) | 1.76 (1.60-1.94) | 1.76 (1.59-1.94) |
| III | 2926 (9.02) | 3.34 (3.05-3.66) | 3.34 (3.05-3.66) |
| IV | 5825 (17.95) | 14.06 (12.85-15.39) | 14.12 (12.90-15.47) |
| Unknown | 941 (2.90) | 1.70 (1.54-1.89) | 1.70 (1.53-1.88) |
| Tumor grade | |||
| I (well differentiated) | 606 (1.87) | Ref | Ref |
| II (moderately differentiated) | 5825 (17.95) | 1.24 (1.14-1.35) | 1.23 (1.13-1.34) |
| III (poorly differentiated) | 2785 (8.58) | 1.97 (1.80-2.16) | 1.97 (1.80-2.15) |
| IV (undifferentiated) | 314 (0.97) | 2.39 (2.08-2.74) | 2.33 (2.03-2.68) |
| Unknown | 2073 (6.39) | 1.32 (1.20-1.45) | 1.32 (1.20-1.45) |
| Tumor surgery | |||
| Yes | 8332 (25.68) | 0.39 (0.37-0.40) | 0.39 (0.37-0.41) |
| No | 3271 (10.08) | Ref | Ref |
| Chemotherapy | |||
| Yes | 8150 (25.12) | 0.83 (0.79-0.87) | 0.83 (0.79-0.87) |
| No or unknown | 3453 (10.64) | Ref | Ref |
| Radiation therapy | |||
| Yes | 2949 (9.09) | 1.06 (1.01-1.11) | 1.06 (1.02-1.11) |
| No or unknown | 8654 (26.67) | Ref | Ref |
| Smokinge | N/A | - | 3.71 (2.52-5.45) |
| Generalized R2 | 35% | ||
Number of events/deaths and row (strata) proportion.
Adjusted for age, race, marital status, stage, grade, surgery, chemotherapy, radiation therapy, and smoking (depending on column), while accounting for clustering by SEER registry.
HR = hazard ratios, estimated using multilevel Cox proportional hazards regression.
Bold indicates significance with P-value ≤ 0.05.
County-level proportion of adult smoking.
Table 4.
Multivariable cox proportional hazards models for CRC-specific death among men with EOCRC stratified by hotspot Counties
| Among Men Living in EOCRC Hotspots | Among Men Living in Non-Significant Counties | |||
|---|---|---|---|---|
|
|
|
|||
| No. of Deaths (%)a | HR (95% CI)b,c,d | No. of Deaths (%)a | HR (95% CI)b,c,d | |
| Age (in years) | ||||
| 15-24 | 5 (1.19) | 0.48 (0.19-1.20) | 188 (1.68) | 0.83 (0.72-0.97) |
| 25-29 | 10 (2.38) | 0.77 (0.40-1.48) | 326 (2.92) | 0.87 (0.77-0.97) |
| 30-34 | 25 (5.95) | 0.73 (0.48-1.11) | 650 (5.81) | 0.85 (0.79-0.93) |
| 35-39 | 46 (10.95) | 0.85 (0.62-1.18) | 1396 (12.48) | 0.96 (0.90-1.02) |
| 40-44 | 92 (21.90) | 0.88 (0.69-1.13) | 2622 (23.45) | 0.91 (0.87-0.95) |
| 45-49 | 242 (57.62) | Ref | 6001 (53.66) | Ref |
| Race/ethnicity | ||||
| NH-White | 269 (64.05) | Ref | 6598 (59.00) | Ref |
| NH-Black | 139 (33.10) | 1.14 (0.92-1.42) | 1848 (16.53) | 1.33 (1.26-1.40) |
| Hispanic | 7 (1.67) | 1.33 (0.62-2.84) | 1807 (16.16) | 1.12 (1.07-1.19) |
| Asian/PI | 2 (0.48) | 0.74 (0.18-3.04) | 825 (7.38) | 1.00 (0.93-1.07) |
| Other/Unknown | 3 (0.71) | 1.44 (0.45-4.63) | 105 (0.94) | 0.65 (0.54-0.80) |
| Marital status | ||||
| Single or never married | 133 (31.67) | 1.43 (1.13-1.79) | 3801 (33.99) | 1.38 (1.32-1.44) |
| Married or domestic partner | 209 (49.76) | Ref | 5677 (50.76) | Ref |
| Divorced, separated, or widowed | 53 (12.62) | 1.43 (1.05-1.95) | 1233 (11.03) | 1.36 (1.28-1.45) |
| Unknown | 25 (5.95) | 1.09 (0.71-1.69) | 472 (4.22) | 1.00 (0.91-1.10) |
| AJCC stage | ||||
| 0 or I | 38 (9.05) | Ref | 615 (5.50) | Ref |
| II | 60 (14.29) | 2.45 (1.60-3.75) | 1198 (10.71) | 1.74 (1.57-1.92) |
| III | 118 (28.10) | 4.18 (2.76-6.32) | 2808 (25.11) | 3.33 (3.03-3.66) |
| IV | 174 (41.43) | 10.83 (7.21-16.25) | 5651 (50.53) | 14.32 (13.04-15.72) |
| Unknown | 30 (7.14) | 1.44 (0.88-2.38) | 911 (8.15) | 1.72 (1.54-1.91) |
| Grade | ||||
| I (well differentiated) | 13 (3.10) | Ref | 593 (5.30) | Ref |
| II (moderately differentiated) | 212 (50.48) | 1.14 (0.65-2.02) | 5613 (50.19) | 1.23 (1.13-1.35) |
| III (poorly differentiated) | 87 (20.71) | 1.87 (1.03-3.40) | 2698 (24.13) | 1.97 (1.80-2.16) |
| IV (undifferentiated) | 15 (3.57) | 2.60 (1.21-5.61) | 299 (2.67) | 2.31 (2.01-2.66) |
| Unknown | 93 (22.14) | 1.89 (1.04-3.43) | 1980 (17.71) | 1.30 (1.18-1.43) |
| Tumor surgery | ||||
| Yes | 317 (31.42) | 0.39 (0.30-0.51) | 8015 (25.49) | 0.39 (0.37-0.41) |
| No | 103 (10.21) | Ref | 3168 (10.08) | Ref |
| Chemotherapy | ||||
| Yes | 276 (65.71) | 0.78 (0.61-1.01) | 7874 (70.41) | 0.83 (0.79-0.87) |
| No or Unknown | 144 (34.29) | Ref | 3309 (29.59) | Ref |
| Radiation therapy | ||||
| Yes | 112 (26.67) | 1.04 (0.82-1.32) | 8346 (74.63) | 1.07 (1.02-1.12) |
| No or Unknown | 308 (73.33) | Ref | 2837 (25.37) | Ref |
| Smokinge | N/A | 8.43 (0.84-84.57) | N/A | 3.54 (2.39-5.23) |
| Generalized R2 | 32% | 35% | ||
Number of events/deaths and row (strata) proportion.
Adjusted for age, race, marital status, AJCC stage, grade, surgery, chemotherapy, radiation therapy, and smoking, while accounting for clustering by SEER registry.
HR = hazard ratios, estimated using multilevel Cox proportional hazards regression.
Bold indicates significance with P-value ≤ 0.05.
County-level proportion of adult smoking.
Within non-significant counties, younger men aged 15-24 (HR 0.83, 0.72-0.97), 25-29 (HR 0.87, 0.77-0.97), and 30-34 (HR 0.85, 0.79-0.93) saw a 13% to 17% decrease in CRC-specific mortality risk compared to older men aged 45-49 (Table 4). Within non-significant counties, NH-Blacks (HR 1.33, 1.26-1.40) and Hispanics (HR 1.12, 1.07-1.19) had a 33% and 12% greater risk of CRC-specific death than NH-Whites, respectively. Furthermore, patients who had received chemotherapy had a 17% decreased risk for CRC mortality (HR 0.83, 0.79-0.87), while those who had received radiation therapy had a 7% increased risk for CRC mortality (HR 1.07, 1.02-1.12). Additionally, for each unit increase in the county-level proportion of adult smokers, men in non-significant counties were 3.5 times more likely to die from CRC-related mortality (HR 3.54, 2.39-5.23). Like the trends witnessed within hotspots, marital status and tumor stage, grade, and surgery were all associated with risk of CRC-specific mortality among men in non-significant spots.
We estimated the proportion of variance explained in CRC survival by clinicodemographic determinants and county-level smoking among men with EOCRC (Tables 3, 4). Among all men, the full model determinants-including smoking-accounted for 35% of the variation in survival; the presence of smoking in the model had minimal influence on the proportion of variance explained (Table 3). Among all determinants, AJCC stage explained the largest proportion (15.4%) of variance in survival. Among men living in CRC hotspots, all determinants in the model accounted for nearly 32% of the survival variation (Table 4), with AJCC stage contributing to 13.8% of the variance. Among those living in non-significant counties, full model variables explained roughly 35% of the CRC survival variation, with AJCC stage accounting for 15.5% of the variance.
Discussion
Our analysis of CDC mortality data identified hotspot counties with high EOCRC mortality rates, mostly concentrated in the Southern US. Our analysis of outcomes for 32,447 men aged 15-49 diagnosed with CRC between 1999-2016 identified significantly worse survival for men residing in hotspot counties, who had 24% greater hazard of CRC-specific mortality compared to their non-significant counterparts. These findings are novel, given our study is one of the first to identify EOCRC mortality hotspots in the US and the first population-based study to identify county-level CRC outcome determinants specific to men diagnosed with CRC before age 50.
Among individuals of all ages diagnosed with CRC, cancer death rates vary sharply by US geographic region [39]. Previous work by Siegel et al. derived contemporary spatial clusters of US counties with high CRC rates based on county-level mortality data from 1970-2011 and identified three distinct hotspots: the lower Mississippi Delta, west-central Appalachia, and eastern Virginia/North Carolina [40]. More recently, examination of statewide variation in EOCRC incidence revealed highest incidence rates in the South [30]. Aligned with these findings, our use of three geospatial techniques defined hotspots for CRC mortality specifically among individuals diagnosed with EOCRC-which includes counties clustered in southern to central Appalachia, the southern Mississippi River and eastern Texas, and the coastal southeast and eastern Virginia/North Carolina. Moreover, we observed that men diagnosed with EOCRC residing in hotspot counties experienced a significantly higher hazard of death compared with men residing in other US regions. Potential explanations for poorer EOCRC outcomes among men residing in these hotspots include an enduring history of unique challenges (e.g., inadequate access to care, poor health literacy, and low educational attainment) [41-44]. We also noted that hotspot counties had higher rates of poverty and uninsurance, as well as fewer primary care physicians. Among all determinants, AJCC clinical stage explained the largest proportion of the variance in EOCRC survival among men in hotspots and non-significant counties combined. As stage is a prognostic CRC survival determinant, these findings emphasize the importance of developing targeted prevention and treatment guidelines for EOCRC, particularly among individuals residing in regions with poorer EOCRC outcomes.
Importantly, beyond geographic disparities in survival among patients diagnosed with EOCRC, other differences persist by race/ethnicity and sex. In a previous study using the South Carolina Central Cancer Registry, Wallace et al. found that African-American patients aged <50 with advanced-stage CRC diagnosis had a significantly higher death risk than their European-American counterparts [45]. Consistent with these earlier findings [3,7-9,21-24], we reported higher EOCRC burden among NH-Black men in hotspots compared to non-significant counties, as well as worse survival among NH-Black men compared to NH-White men in non-significant counties. The disproportionate burden of EOCRC among NH-Black men may result from distinctive stressors coupled with cultural and social expectations that impact screening and care behaviors [46-49]. A US population-based study by Holowatyj et al. that identified racial/ethnic disparities among patients with EOCRC also reported significantly worse cancer-specific and overall survival among men compared with women [9]. Given the exacerbated burden of EOCRC among men, our study focused solely on men to identify determinants of geographic CRC survival variation while minimizing the potential impact that sex-specific differences may have on EOCRC-related outcomes. Moreover, our complimentary investigation into community health behaviors and variation in EOCRC survival among US women-explored in Holowatyj et al. [50]-reported findings in contrast to men, in which AJCC clinical stage and race/ethnicity accounted for a higher proportion of EOCRC survival variance among women in hotspot counties compared to other counties. Given these distinct variations in EOCRC survival, our results warrant further investigation into the health behaviors and molecular differences of EOCRC by race/ethnicity and sex to better understand the rising burden of this disease in young patients.
Our examination of geographic variance in EOCRC incidence and mortality among men explained by individual- and county-level characteristics is the first population-based study to shed light on modifiable determinants of EOCRC outcomes. Much of the variance accounted for in EOCRC survival in both hotspot and non-significant counties was explained by disease stage [51]. While estimated that approximately 14% of all US adults are current smokers, we observed 24% of the adult population residing in hotspot counties reported currently smoking and having smoked at least 100 cigarettes in their lifetime [52]. Moreover, we discovered that the effect of hotspots on CRC survival was reduced by 12% after adjusting for county-level smoking, suggesting that smoking may be a major contributor to the increased mortality within hotspots. Although the implementation of tobacco control programs has led to sharp declines in adult cigarette smoking over the last decade, cancers linked to tobacco use, including CRC [53-55], continue to account for approximately 4 out of every 10 cancer diagnoses and approximately 3 out of every 10 cancer deaths in the US [51,56,57]. As current cigarette smoking rates differ by US census region (with the highest rates among adults living in the Midwest and South), as well as by sex, race/ethnicity, education, income, and marital status [58], our findings increase awareness regarding the impact of patient characteristics on differences in EOCRC survival in hotspots versus other regions. A recent retrospective study by Wolbert et al. reported that the rate of early-onset rectal cancer in rural Appalachia was 1.5 times higher than national rates, and that smoking was strongly associated with early-onset rectal cancer [59]. Several studies have also reported that smoking is more strongly associated with rectosigmoid junction/rectal cancer than with colon cancer [60-62]. Given higher rates of smoking in men, combined with gene-tobacco interactions [63], future studies should examine how the carcinogenic effects of tobacco smoke may be uniquely contributing to EOCRC burden among men.
Current smoking is also associated with several CRC risk factors, including lower consumption of healthy foods (e.g., fruits, vegetables, fiber) [64-67] and greater physical inactivity [68]. Along with finding a higher prevalence of adult smokers in counties with high EOCRC mortality rates, we also identified significantly higher proportions of adults having no leisure-time physical activity and more limited access to healthy foods when compared with other areas. Over time, chronic excessive caloric intake and physical inactivity lead to energy imbalances, resulting in individuals becoming overweight or developing obesity, which increases CRC risk [69,70]. A recent study by Nguyen et al. also demonstrated that prolonged sedentary behaviors-a surrogate for a more-inactive lifestyle, are associated with increased EOCRC risk [71]. The United Health Foundation’s annual report recently documented that the states in which our hotspots were located are some of the least healthy states in the country [45]. With obesity rates having nearly tripled worldwide since 1975 and with more than 70% of US adults currently overweight or obese [72], further studies of the association between lifestyle factors-including smoking, obesity, and dietary patterns-and EOCRC incidence and mortality are needed.
Our study had several limitations. First, we were unable to estimate EOCRC mortality rates in areas with limited deaths, as CDC data are suppressed at the county level when the number of deaths is fewer than 10. However, our geospatial analysis was strengthened by the use of three spatial autocorrelation methods, including the EB smoothed-rate method that accounted for counties with few cases, and the examination of mortality data across the contiguous US over the study period. Additionally, the use of FIPS codes derived from CDC death certificates and SEER patient files is a robust method of identifying EOCRC hotspots among men. However, we were unable to account for length or change of residence, which could contribute to changes in socioeconomic status, access to and quality of care, and CRC outcomes. Moreover, SEER lacks data on individual-level characteristics and lifestyle factors known to be associated with CRC risk, including individual smoking status and environmental exposures (e.g., secondhand smoke), and family CRC history and hereditary cancer syndromes. Another limitation is our aggregation of SEER data, a 17-year period from 1999 to 2016, which allowed us to work with a large, cohesive sample size, but restricted our ability to account for temporal changes relevant to CRC during this timeframe, such as advancements in cancer prevention, treatment, and epidemiology, as well as evolving clinicodemographics. In addition, we linked 1999-2016 SEER data with 2014 county-level daasets, which applied a single fixed timestamp of health and community characteristics to years of older data, which likely experienced changing characteristics over time.
Conclusions
As one of the first studies to define geographic hotspots for EOCRC, we observed significantly worse survival after EOCRC diagnosis among men-particularly NH-Black men-residing in hotspot counties. Further studies of CRC-related health behaviors among NH-Black men diagnosed with EOCRC are needed and could have significant implications for cancer screening, early detection, and care. Given that CRC incidence rates among individuals aged <50 are continuing to rise with causes unknown [3,5,6,9,19,22-24], examination of individuallevel health behaviors and clinical characteristics among men diagnosed with EOCRC is needed to explore gene-environment interactions associated with geographic survival variation and to tailor clinical algorithms for early CRC detection.
Acknowledgements
The authors acknowledge Eleanor Mayfield, ELS, Phung Matthews, PharmD, and Tracy Rees, MFA and who provided editorial assistance. We also extend appreciation to Autumn Henson who assisted with preliminary data analysis support. This research was supported by the National Cancer Institute of the National Institutes of Health (NIH) under Award Numbers K01 CA234319 (CRR), T32 CA190194 (JXM), T32 HG008962 (ANH), P30 CA042014, the Huntsman Cancer Foundation, and the Health Studies Fund of the Department of Family & Preventative Medicine at the University of Utah. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH, Huntsman Cancer Foundation, or University of Utah.
Disclosure of conflict of interest
None.
Supporting Information
References
- 1.Siegel RL, Miller KD, Jemal A. Cancer statistics, 2019. Cancer. 2019;69:7–34. doi: 10.3322/caac.21551. [DOI] [PubMed] [Google Scholar]
- 2.American Cancer Society. Colorectal cancer facts & figures 2017-2019. Atlanta, GA: American Cancer Society; 2018. https://www.cancer.org/content/dam/cancer-org/research/cancer-facts-and-statistics/colorectal-cancer-facts-and-figures/colorectal-cancer-facts-and-figures-2017-2019.pdf. [Google Scholar]
- 3.Wang W, Chen W, Lin J, Shen Q, Zhou X, Lin C. Incidence and characteristics of young-onset colorectal cancer in the United States: an analysis of SEER data collected from 1988 to 2013. Clin Res Hepatol Gastroenterol. 2019;43:208–215. doi: 10.1016/j.clinre.2018.09.003. [DOI] [PubMed] [Google Scholar]
- 4.Siegel RL, Fedewa SA, Anderson WF, Miller KD, Ma J, Rosenberg PS, Jemal A. Colorectal cancer incidence patterns in the United States, 1974-2013. J Natl Cancer Inst. 2017;109 doi: 10.1093/jnci/djw322. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Abdelsattar ZM, Wong SL, Regenbogen SE, Jomaa DM, Hardiman KM, Hendren S. Colorectal cancer outcomes and treatment patterns in patients too young for average-risk screening. Cancer. 2016;122:929–934. doi: 10.1002/cncr.29716. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Kasi PM, Shahjehan F, Cochuyt JJ, Li Z, Colibaseanu DT, Merchea A. Rising proportion of young individuals with rectal and colon cancer. Clin Colorectal Cancer. 2019;18:e87–e95. doi: 10.1016/j.clcc.2018.10.002. [DOI] [PubMed] [Google Scholar]
- 7.Ellis L, Abrahão R, McKinley M, Yang J, Somsouk M, Marchand LL, Cheng I, Gomez SL, Shariff-Marco S. Colorectal cancer incidence trends by age, stage, and racial/ethnic group in California, 1990-2014. Cancer Epidemiol Biomarkers Prev. 2018;27:1011–1018. doi: 10.1158/1055-9965.EPI-18-0030. [DOI] [PubMed] [Google Scholar]
- 8.Murphy CC, Singal AG, Baron JA, Sandler RS. Decrease in incidence of young-onset colorectal cancer before recent increase. Gastroenterology. 2018;155:1716–1719. e4. doi: 10.1053/j.gastro.2018.07.045. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Holowatyj AN, Ruterbusch JJ, Rozek LS, Cote ML, Stoffel EM. Racial/ethnic disparities in survival among patients with young-onset colorectal cancer. J. Clin. Oncol. 2016;34:2148–2156. doi: 10.1200/JCO.2015.65.0994. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Siegel R, Jemal A, Ward EM. Increase in incidence of colorectal cancer among young men and women in the United States. Cancer Epidemiol Biomarkers Prev. 2009;18:1695–1698. doi: 10.1158/1055-9965.EPI-09-0186. [DOI] [PubMed] [Google Scholar]
- 11.Bailey C, Hu CY, You YN, Bednarski BK, Rodriguez-Bigas MA, Skibber JM, Cantor SB, Chang GJ. Increasing disparities in the agerelated incidences of colon and rectal cancers in the United States, 1975-2010. JAMA Surg. 2015;150:17–22. doi: 10.1001/jamasurg.2014.1756. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Sung H, Siegel RL, Rosenberg PS, Jemal A. Emerging cancer trends among young adults in the USA: analysis of a population-based cancer registry. Lancet Public Health. 2019;4:E137–E147. doi: 10.1016/S2468-2667(18)30267-6. [DOI] [PubMed] [Google Scholar]
- 13.Araghi M, Seorjomataram I, Bardot A, Ferlay J, Cabasag CJ, Morrison DS, De P, Tervonen H, Walsh PM, Bucher O, Engholm G, Jackson C, McClure C, Woods RR, Saint-Jacques N, Morgan E, Ransom D, Thursfield V, Moller B, Leonfellner S, Guren MG, Bray F, Arnold M. Changes in colorectal cancer incidence in seven high-income countries: a population-based study. Lancet Gastroenterol Hepatol. 2019;4:511–518. doi: 10.1016/S2468-1253(19)30147-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Stigliano V, Sanchez-Mete L, Martayan A, Anti M. Early-onset colorectal cancer: a sporadic or inherited disease? World J Gastroenterol. 2014;20:12420–30. doi: 10.3748/wjg.v20.i35.12420. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Mauri G, Sartore-Bianchi A, Russo AG, Marsoni S, Bardelli A, Siena S. Early-onset colorectal cancer in young individuals. Mol Oncol. 2019;13:109–131. doi: 10.1002/1878-0261.12417. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.O’Connell JB, Maggard MA, Livingston EH, Yo CK. Colorectal cancer in the young. Am J Surg. 2004;187:343–348. doi: 10.1016/j.amjsurg.2003.12.020. [DOI] [PubMed] [Google Scholar]
- 17.Holowatyj AN, Gigic B, Herpel E, Scalbert A, Schneider M, Ulrich CM MetaboCCC Consortium; ColoCare Study. Distinct molecular phenotype of sporadic colorectal cancer among young patients based on multiomics analysis. Gastroenterology. 2020;158:1155–1158. e2. doi: 10.1053/j.gastro.2019.11.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Archambault AN, Su YR, Jeon J, Thomas M, Lin Y, Conti DV, Win AK, Sokada LC, Landsdorp-Vogelaar I, Peterse EFP, Zauber AG, Duggan D, Holowatyj AN, Huyghe JR, Brenner H, Cotterchio M, Bezieau S, Schmit SL, Edlund CK, Southey MC, MacInnis RJ, Campbell PT, Chang-Claude J, Slattery ML, Chan AT, Joshi AD, Song M, Cao Y, Woods MO, White E, Weinstein SJ, Ulrich CM, Hoffmeister M, Bein SA, Harrison TA, Hampe J, Li CI, Schafmayer C, Offit K, Pharoah PD, Moreno V, Lindblom A, Wolk A, Wu AH, Li L, Gunter MJ, Gsur A, Keku TO, Pearlman R, Bishop DT, Castellvi-Bel S, Moreira L, Vodicka P, Kampman E, Giles GG, Albanes D, Baron JA, Berndt SI, Brezina S, Buch S, Buchanan DD, Trichopoulou A, Severi G, Chiraque MD, Sanchez MJ, Palli D, Kuhn T, Murphy N, Cross AJ, Burnett-Hartman AN, Chanock SJ, De la Chapelle A, Easton DF, Elliott F, English DR, Feskens DJM, Fitzgerald LM, Goodman PJ, Hopper JL, Hudson TJ, Hunter DJ, Jacobs EJ, Josh CE, Jury S, Markowitz SD, Milne RL, Platz EZ, Rennert G, Renner HS, Schumaker FR, Sandler Rs, Seminara D, Tangen CM, Thibodeau SN, Toland AE, van Duijnhoven FJB, Visvanathan K, Vodickova L, Potter JD, Mannisto S, Weigl K, Figueirodo J, Martin V, Larsson SC, Parfrey PS, Huang WY, Lenz HJ, Caselao JE, Gago-Dominguez M, Munoz-Garzon V, Manco C, Haiman CA, Wilkens LR, Seigel E, Barry E, Younghusband B, Van Guellpen B, Harlid S, Zeleniuch-Jacquotte A, Liang PS, Du M, Casey G, Lindor NM, Le Marchand L, Gallingher SJ, Jenkins MA, Newcomb PA, Gruber SB, Schoen RE, Hampel H, Corley DA, Hsu L, Peters U, Haves RB. Cumulative burden of colorectal cancer-associated genetic variants is more strongly associated with early-onset versus late-onset cancer. Gastroenterology. 2020;158:1274–1286. e12. doi: 10.1053/j.gastro.2019.12.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Holowatyj AN, Lewis MA, Pannier ST, Kirchhoff AC, Hardikar S, Figueiredo JC, Huang LC, Shibata D, Schmit SL, Ulrich CM. Clinicopathologic and racial/ethnic differences of colorectal cancer among adolescents and young adults. Clin Transl Gastroenterol. 2019;10:e00059. doi: 10.14309/ctg.0000000000000059. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Murphy CC, Wallace K, Sandler RS, Baron JA. Racial disparities in incidence of young-onset colorectal cancer and patient survival. Gastroenterology. 2019;156:958–965. doi: 10.1053/j.gastro.2018.11.060. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Ashktorab H, Vilmenay K, Brim H, Laiyemo AO, Kibreab A, Nouraie M. Colorectal cancer in young African Americans: is it time to revisit guidelines and prevention? Dig Dis Sci. 2016;61:3026–3030. doi: 10.1007/s10620-016-4207-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Rahman R, Schmaltz C, Jackson CS, Simoes EJ, Jackson-Thompson J, Ibdah JA. Increased risk for colorectal cancer under age 50 in racial and ethnic minorities living in the United States. Cancer Med. 2015;4:1863–1870. doi: 10.1002/cam4.560. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Griffin PM, Liff JM, Greenberg RS, Clark WS. Adenocarcinomas of the colon and rectum in persons under 40 years old. Gastroenterology. 1991;100:1033–1040. doi: 10.1016/0016-5085(91)90279-t. [DOI] [PubMed] [Google Scholar]
- 24.Williams R, White P, Nieto J, Vieira D, Francois F, Hamilton F. Colorectal cancer in African Americans: an update. Clin Transl Gastroenterol. 2016;7:e185. doi: 10.1038/ctg.2016.36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.DeSantis CE, Miller KD, Goding Sauer A, Jemal A, Siegel RL. Cancer statistics for African Americans, 2019. CA Cancer J Clin. 2019;69:211–233. doi: 10.3322/caac.21555. [DOI] [PubMed] [Google Scholar]
- 26.Mokdad AH, Dwyer-Lindgren L, Fitzmaurice C, Stubbs RW, Bertozzi-Villa A, Morozoff C, Charara R, Allen C, Naghavi M, Murray CJ. Trends and patterns of disparities in cancer mortality among US counties, 1980-2014. JAMA. 2017;317:388–406. doi: 10.1001/jama.2016.20324. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Naishadham D, Lansdorp-Vogelaar I, Siegel R, Cokkinides V, Jemal A. State disparities in colorectal cancer mortality patterns in the United States. Cancer Epidemiol Biomarkers Prev. 2011;20:1296–302. doi: 10.1158/1055-9965.EPI-11-0250. [DOI] [PubMed] [Google Scholar]
- 28.Lian M, Schootman M, Doubeni CA, Park Y, Major JM, Stone RA, Laiyemo AO, Hollenbeck AR, Graubard BI, Schatzkin A. Geographic variation in colorectal cancer survival and the role of small-area socioeconomic deprivation: a multilevel survival analysis of the NIH-AARP Diet and Health Study Cohort. Am J Epidemiol. 2011;174:828–38. doi: 10.1093/aje/kwr162. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Henry KA, Niu X, Boscoe FP. Geographic disparities in colorectal cancer survival. Int J Health Geogr. 2009;8:48. doi: 10.1186/1476-072X-8-48. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Siegel RL, Medhanie GA, Fedewa SA, Jemal A. State variation in early-onset colorectal cancer in the United States, 1995-2015. J Natl Cancer Inst. 2019;111:1104–1106. doi: 10.1093/jnci/djz098. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Moore JX, Royston KJ, Langston ME, Griffin R, Hidalgo B, Wang HE, Colditz G, Akinyemiju T. Mapping hot spots of breast cancer mortality in the United States: place matters for Blacks and Hispanics. Cancer Causes Control. 2018;29:737–750. doi: 10.1007/s10552-018-1051-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Centers for Disease Control and Prevention. About underlying cause of death. 1999-2017. https://wonder.cdc.gov/ucd-icd10.html.
- 33.Nassel AF, Root ED, Haukoos JS, McVaney K, Colwell C, Robinson J, Eigel B, Magid DJ, Sasson C. Multiple cluster analysis for the identification of high-risk census tracts for out-of-hospital cardiac arrest (OHCA) in Denver, Colorado. Resuscitation. 2014;85:1667–1673. doi: 10.1016/j.resuscitation.2014.08.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Anselin L. Local indicators of spatial association-LISA. Geogr Anal. 1995;27:93–115. [Google Scholar]
- 35.Getis A, Ord K. The analysis of spatial association by use of distance statistics. Geogr Anal. 1992;24:189–206. [Google Scholar]
- 36.National Cancer Institute, Division of Cancer Control and Population Sciences, Surveillance Research Program. Surveillance, epidemiology, and end results (SEER) program populations (1969-2015), released December 2016. https://www.seer.cancer.gov/popdata.
- 37.Allison PD. Survival analysis using the SAS system: a practical guide. Cary, NC: SAS Institute Inc.; 1995. [Google Scholar]
- 38.Cox DR, Snell EJ. Analysis of binary data. 2nd edition. London; New York: Chapman and Hall; 1989. [Google Scholar]
- 39.Siegel RL, Sahar L, Portier KM, Ward EM, Jemal A. Cancer death rates in US congressional districts. CA Cancer J Clin. 2015;65:339–44. doi: 10.3322/caac.21292. [DOI] [PubMed] [Google Scholar]
- 40.Siegel RL, Sahar L, Robbins A, Jemal A. Where can colorectal cancer screening interventions have the most impact? Cancer Epidemiol Biomarkers Prev. 2015;24:1151–1156. doi: 10.1158/1055-9965.EPI-15-0082. [DOI] [PubMed] [Google Scholar]
- 41.Cosby AG, Bowser DM. The health of the delta region: a story of increasing disparities. J Health Hum Serv Adm. 2008;31:58–71. [PubMed] [Google Scholar]
- 42.Friedell GH, Rubio A, Maretzki A, Garland B, Brown P, Crane M, Hickman P. Community cancer control in a rural, underserved population: the Appalachian Leadership Initiative on Cancer Project. J Health Care Poor Underserved. 2001;12:5–19. doi: 10.1353/hpu.2010.0523. [DOI] [PubMed] [Google Scholar]
- 43.Green JJ, Nylander AB. A community-based framework for understanding problems and exploring alternatives: connecting underemployment, poverty and access to health care in the Mississippi Delta. Columbia, MO: Rural Poverty Research Center; 2006. Rural Poverty Research Center working paper 06-02. [Google Scholar]
- 44.United Health Foundation. America’s health rankings: annual report 2018. Minnetonka, MN: United Health Foundation; 2018. [Google Scholar]
- 45.Wallace K, Hill EG, Lewin DN, Williamson G, Oppenheimer S, Ford ME, Wargovich MJ, Berger FG, Bolick SW, Thomas MB, Alberg AJ. Racial disparities in advanced stage colorectal cancer survival. Cancer Causes Control. 2013;24:463–471. doi: 10.1007/s10552-012-0133-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Williams DR. The health of men: structured inequalities and opportunities. Am J Public Health. 2003;93:724–731. doi: 10.2105/ajph.93.5.724. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Griffith DM, Ellis KR, Allen JO. An intersectional approach to social determinants of stress for African American men: men’s and women’s perspectives. Am J Mens Health. 2013;7:19S–30S. doi: 10.1177/1557988313480227. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Umberson D, Williams K, Thomas PA, Liu H, Thomeer MB. Race, gender and chains of disadvantage: childhood adversity, social relationships, and health. J Health Soc Behav. 2014;55:20–38. doi: 10.1177/0022146514521426. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Rogers CR, Mitchell JA, Franta GJ, Foster MJ, Shires D. Masculinity, racism, social support, and colorectal cancer screening uptake among African American men: a systematic review. Am J Mens Health. 2017;11:1486–1500. doi: 10.1177/1557988315611227. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Holowatyj AN, Langston M, Han Y, Viskochil R, Cao Y, Rogers CR, Lieu C, Moore JX. Community health behaviors and geographic variation in early-onset CRC survival among women. 2020 doi: 10.14309/ctg.0000000000000266. Revision Under Review. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. The health consequences of smoking-50 years of progress: a report of the surgeon general. Atlanta, GA: Centers for Disease Control and Prevention; 2014. [Google Scholar]
- 52.County Health Rankings & Roadmaps Program. “County health rankings 2014 data dictionary”. Robert wood Johnson foundation and the university of Wisconsin population health institute. 2016. http://www.countyhealthrankings.org/sites/default/files/DataDictionary_2014.pdf.
- 53.Giovannucci E, Martínez ME. Tobacco, colorectal cancer, and adenomas: a review of the evidence. J Natl Cancer Inst. 1996;88:1717–1730. doi: 10.1093/jnci/88.23.1717. [DOI] [PubMed] [Google Scholar]
- 54.Giovannucci E. An updated review of the epidemiological evidence that cigarette smoking increases risk of colorectal cancer. Cancer Epidemiol Biomarkers Prev. 2001;10:725–731. [PubMed] [Google Scholar]
- 55.Hannan LM, Jacobs EJ, Thun MJ. The association between cigarette smoking and risk of colorectal cancer in a large prospective cohort from the United States. Cancer Epidemiol Biomarkers Prev. 2009;18:3362–3367. doi: 10.1158/1055-9965.EPI-09-0661. [DOI] [PubMed] [Google Scholar]
- 56.Inoue-Choi M, Liao LM, Reyes-Guzman C, Hartge P, Caporaso N, Freedman ND. Association of long-term, low-intensity smoking with all-cause and cause-specific mortality in the National Institutes of Health-AARP Diet and Health Study. JAMA Intern Med. 2017;177:87–95. doi: 10.1001/jamainternmed.2016.7511. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 57.Nash SH, Liao LM, Harris TB, Freedman ND. Cigarette smoking and mortality in adults aged 70 years and older: results from the NIH-AARP cohort. Am J Prev Med. 2017;52:276–283. doi: 10.1016/j.amepre.2016.09.036. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 58.Wang TW, Asman K, Gentzke AS, Cullen KA, Holder-Hayes E, Reyes-Guzman C, Jamal A, Neff L, King BA. Tobacco product use among adults-United States, 2017. MMWR Morb Mortal Wkly Rep. 2018;67:1225–1232. doi: 10.15585/mmwr.mm6744a2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 59.Wolbert T, Leigh EC, Barry R, Thompson EC, Gress T, Ajmera A, Arrington AK. Later stage disease and earlier onset of rectal cancer: epidemiology and outcomes comparison of rectal cancer in a rural Appalachian area to state and national rates. Am Surg. 2018;84:1229–1235. [PubMed] [Google Scholar]
- 60.Hansen RD, Albieri V, Tjønneland A, Overvad K, Andersen KK, Raaschou-Nielsen O. Effects of smoking and antioxidant micronutrients on risk of colorectal cancer. Clin Gastroenterol Hepatol. 2013;11:406–415. doi: 10.1016/j.cgh.2012.10.039. [DOI] [PubMed] [Google Scholar]
- 61.Tsoi KK, Pau CY, Wu WK, Chan FK, Griffiths S, Sung JJ. Cigarette smoking and the risk of colorectal cancer: a meta-analysis of prospective cohort studies. Clin Gastroenterol Hepatol. 2009;7:682–688. doi: 10.1016/j.cgh.2009.02.016. [DOI] [PubMed] [Google Scholar]
- 62.Nisa H, Kono S, Yin G, Toyomura K, Nagano J, Mibu R, Tanako M, Kakeji Y, Maehara Y, Okamura T, Ikejiri K, Futami K, Maekawa T, Yasunami Y, Takenaka K, Ichimiya H, Terasaka R. Cigarette smoking, genetic polymorphisms and colorectal cancer risk: the Fukuoka Colorectal Cancer Study. BMC Cancer. 2010;10:274. doi: 10.1186/1471-2407-10-274. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 63.Taioli E. Gene-environment interaction in tobacco-related cancers. Carcinogenesis. 2008;29:1467–1474. doi: 10.1093/carcin/bgn062. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 64.Morabia A, Wynder EL. Dietary habits of smokers, people who never smoked, and exsmokers. Am J Clin Nutr. 1990;52:933–937. doi: 10.1093/ajcn/52.5.933. [DOI] [PubMed] [Google Scholar]
- 65.Dallongeville J, Marécaux N, Fruchart JC, Amouyel P. Cigarette smoking is associated with unhealthy patterns of nutrient intake: a meta-analysis. J Nutr. 1998;128:1450–1457. doi: 10.1093/jn/128.9.1450. [DOI] [PubMed] [Google Scholar]
- 66.Strine TW, Okoro CA, Chapman DP, Balluz LS, Ford ES, Ajani UA, Mokdad AH. Health-related quality of life and health risk behaviors among smokers. Am J Prev Med. 2005;28:182–187. doi: 10.1016/j.amepre.2004.10.002. [DOI] [PubMed] [Google Scholar]
- 67.Subar AF, Harlan LC, Mattson ME. Food and nutrient intake differences between smokers and non-smokers in the US. Am J Public Health. 1990;80:1323–1329. doi: 10.2105/ajph.80.11.1323. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 68.Choi J, Lee M, Lee JK, Kang D, Choi JY. Correlates associated with participation in physical activity among adults: a systematic review of reviews and update. BMC Public Health. 2017;17:356. doi: 10.1186/s12889-017-4255-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 69.Lauby-Secretan B, Scoccianti C, Loomis D, Grosse Y, Bianchini F, Straif K International Agency for Research on Cancer Handbook Working Group. Body fatness and cancer-viewpoint of the IARC Working Group. N Engl J Med. 2016;375:794–798. doi: 10.1056/NEJMsr1606602. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 70.Ulrich CM, Himbert C, Holowatyj AN, Hursting SD. Energy balance and gastrointestinal cancer: risk, interventions, outcomes and mechanisms. Nat Rev Gastroenterol Hepatol. 2018;15:683–698. doi: 10.1038/s41575-018-0053-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 71.Nguyen LH, Liu PH, Zheng X, Keum N, Zong X, Li X, Wu K, Fuchs CS, Ogino S, Ng K, Willett WC, Chan AT, Giovannucci EL, Cao Y. Sedentary behaviors, TV viewing time, and risk of young-onset colorectal cancer. JNCI Cancer Spectr. 2018;2:pky073. doi: 10.1093/jncics/pky073. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 72.National Center for Health Statistics. Health, United States, 2016: with chartbook on long-term trends in health. Hyattsville, MD: National Center for Health Statistics; 2017. [PubMed] [Google Scholar]
- 73.Minnesota Population Center. National historical geographic information system: version 2.0. Minneapolis, MN: University of Minnesota; 2011. [Google Scholar]
- 74.US Department of Agriculture, Economic Research Service. Rural-urban commuting area codes. https://www.ers.usda.gov/data-products/rural-urban-commuting-area-codes.aspx.
- 75.Gruca TS, Pyo TH, Nelson GC. Improving rural access to orthopaedic care through visiting consultant clinics. J Bone Joint Surg Am. 2016;98:768–774. doi: 10.2106/JBJS.15.00946. [DOI] [PubMed] [Google Scholar]
- 76.US Census Bureau. Geographic terms and concepts-census divisions and census regions. 2016. https://www2.census.gov/geo/pdfs/maps-data/maps/reference/us_regdiv.pdf.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.