Fax: (504) 568-2493
Associations of subsite-specific colorectal cancer incidence rates and stage of disease at diagnosis with county-level poverty, by race and sex†
Article first published online: 26 JUN 2006
Published 2006 by the American Cancer Society
Supplement: Descriptive Epidemiology of Colorectal Cancer in the United States, 1998–2001, Utilizing Data from the NPCR and SEER Programs
Volume 107, Issue Supplement 5, pages 1121–1127, 1 September 2006
How to Cite
Wu, X., Cokkinides, V., Chen, V. W., Nadel, M., Ren, Y., Martin, J. and Ellison, G. L. (2006), Associations of subsite-specific colorectal cancer incidence rates and stage of disease at diagnosis with county-level poverty, by race and sex. Cancer, 107: 1121–1127. doi: 10.1002/cncr.22009
The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention
- Issue published online: 21 AUG 2006
- Article first published online: 26 JUN 2006
- Manuscript Accepted: 21 MAR 2006
- Manuscript Received: 2 MAR 2006
- Centers for Disease Control
- Prevention's National Program of Cancer Registries
- colorectal cancer;
- proximal colon cancer;
- distal colon cancer;
- rectal cancer;
- county poverty;
- stage at diagnosis;
- cancer registry;
This study examined associations of subsite-specific colorectal cancer incidence rates and stage of the disease with county-level poverty.
The 1998–2001 colorectal cancer incidence data, covering 75% of the United States population, were from 38 states and metropolitan areas. The county-level poverty data were categorized into 3 groups according to the percentage of the population below the poverty level in 1999: <10% (low-poverty), 10%–19% (middle-poverty), and ≥20% (high-poverty). Age-adjusted subsite-specific incidence rates (for all ages) and stage-specific incidence rates (for ages ≥50) were examined by race (whites and blacks), sex, and the county's poverty level. The differences in the incidence rates were examined using the 2-tailed z-statistic.
The incidence rates of proximal colon cancer were higher among white males (11% higher) and white females (15% higher) in the low-poverty than in the high-poverty counties. No differences across county poverty levels were observed among whites for distal colon and rectal cancers or among blacks for all the subsites. The late-to-early stage incidence rate ratios were higher in the high-poverty than in the low-poverty counties among white and black males for distal colon and rectal cancers, among white females for distal colon cancer, and among black females for rectal cancer. For proximal colon cancer, however, the late-to-early stage rate ratios were similar across all county poverty levels.
Higher incidence rates of proximal cancer were observed among white males and females in the low-poverty counties relative to the high-poverty counties. The higher late-to-early stage rate ratios in high-poverty than in low-poverty counties is observed for distal colon and rectal cancers, but not for proximal colon cancer. Cancer 2006. © 2006 American Cancer Society.
Previous reports have documented variations in colorectal cancer subsite-specific incidence rates and stage distributions by race and sex.1–5 For instance, incidence rates of proximal colon cancer are higher among blacks than among whites for both males and females, whereas incidence rates of distal colorectal cancer are higher among whites than among blacks for males but not females.1, 2 It has also been noted that cancers in the proximal colon are less likely than those in the distal colon and rectum to be localized disease at diagnosis regardless of race or sex.1, 2 It is not clear whether socioeconomic status (SES) plays a part in the variations of the subsite-specific incidence by race and sex. None of the previous studies on SES and colorectal cancer incidence has explored subsite-specific colorectal cancer incidence.6–10 Although SES may not be a direct determinant of subsite-specific colorectal cancer risk and stage of the disease, it is associated with risk factors and use of screening tests for colorectal cancer.11–13 The objective of this study was to examine the associations of subsite-specific colorectal cancer incidence rates and risk of late-stage disease at diagnosis with county poverty level by race and sex.
MATERIALS AND METHODS
The 1998–2001 cancer incidence data from 41 population-based registries, which are the participants in the Centers for Disease Control and Prevention's National Program of Cancer Registries (NPCR) and the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) Program, were used for this colorectal cancer monograph (see the introductory article of this monograph).14 For this particular study, 4 states (Alaska, Hawaii, Illinois, and Minnesota) were excluded because county-level poverty data were unavailable. Likewise, because the only summary stage data for California were from metropolitan Los Angeles and the Greater Bay Area, this study was limited to those California data. The 1998–2001 data from these registries met the U.S. Cancer Statistics publication standard15 and covered 75% of the U.S. population, 75% of whites, and 72% of blacks.
This study included incident primary in situ and invasive colorectal cancers, which were coded according to the International Classification of Diseases for Oncology, Third Edition (ICD-O-3) topography (site) codes C18.0-C18.9, C19.9, C20.9, and C26.0.16 To conform to published population-based cancer incidence statistics, we used only invasive cases to calculate the incidence rates. Both in situ and invasive cancers were included for stage analysis. Excluded were death-certificate-only and autopsy-only cancers, which accounted for 1.5% of all cancers, and malignant lymphomas in the colon and rectum. Whites and blacks were the only race-specific categories included. We also excluded 208 cases with invalid county poverty-level data.
County Poverty-level Grouping
County-level poverty data for all races combined were categorized into 3 poverty groups according to the percentage of the population below the poverty level in 1999 as estimated by the U.S. Bureau of Census. The following cutpoints for the county poverty groups were used: <10% (low-poverty), 10%–19% (middle-poverty), and ≥20% (high-poverty). Areas with a poverty level of ≥20% are considered severely disadvantaged areas.17
Anatomic Subsite Grouping
Colorectal cancers were grouped into 3 major anatomic subsites: proximal colon, distal colon, and rectum. Proximal colon included the cecum (C18.0), ascending colon (C18.2), hepatic flexure (C18.3), transverse colon (C18.4), and splenic flexure (C18.5). Distal colon included the descending colon (C18.6) and the sigmoid colon (C18.7). Rectum included the rectosigmoid junction (C19.9) and rectum not otherwise specified (C20.9). Cancers of the appendix (C18.1), colon not otherwise specified (C18.9), overlapping subsite (C18.8), and intestinal tract not otherwise specified (C26.0), which accounted for 6% of all colorectal cancers in the study data file, were included in the statistics for the all-colorectal-cancers-combined but not in the specific-anatomic-subsite analyses.
Colorectal cancers diagnosed in the years from 1998 through 2000 were staged using the 1977 SEER summary staging system,18 and cancers diagnosed in 2001 were staged using the 2000 SEER summary staging system.19 Superior mesenteric lymph nodes for proximal colon cancer and internal iliac lymph nodes for rectal cancer are considered as distant lymph nodes in the 2000 system but as regional lymph nodes in the 1977 system. Because we combined in situ and localized categories as early-stage disease and combined regional and distant stages as late-stage disease, the differences between the 2 staging systems would not affect our results. Stage data from the 2 systems, therefore, were combined in this study. Only stage data for those aged 50 and more are presented because colorectal cancer screening in the average-risk population is recommended beginning at age 50. Since North Carolina, Texas, and Wisconsin did not submit their data for in situ cancers, including them in the stage data analyses may have artificially decreased the percentage of early-stage disease. Given that the impact was less than 1%, we decided to include these 3 states in the stage data analyses. Stage-specific incidence rates and rate ratios were based on cancers with stage information.
Subsite-specific average annual age-adjusted incidence rates (all ages) and stage-specific age-adjusted incidence rates (ages ≥50) per 100,000 were computed by race (whites, blacks) and sex (males, females) for each of the county poverty groups (<10%, 10%–19%, ≥20%). The 2000 U.S. standard population was used for age-adjustment of the rates. The rate ratios of the incidence rates in the low- and middle-poverty groups versus the incidence rates in the high-poverty group as well as the late-to-early stage rate ratios by county poverty level were calculated and tested with the 2-tailed z-statistic.20 Tests of significance for all comparisons were at a level of P < 0.05.
This study included a total of 476,871 eligible colorectal cancers diagnosed in the years from 1998 to 2001, of which 6% were in situ and 94% were invasive. Cancers of the proximal colon accounted for 42% of the cancers, the distal colon for 25%, the rectum for 27%, and others for 6%. Among those aged 50 and more, the percentages of colorectal cancer by stage at diagnosis were 6% in situ, 34% localized, 37% regional, 15% distant, and 8% unknown stage.
Associations of Subsite-specific Incidence Rates With the County Poverty Level
The incidence rates of all colorectal cancers combined were slightly higher (P < 0.05) among white males (3% higher) and females (6% higher) in the low-poverty than in the high-poverty counties (Table 1). Slightly larger differences were observed for proximal colon cancer (11% higher among white males and 15% higher among white females) but not for distal colon and rectal cancers. The differences between the middle- and high-poverty counties were smaller and were statistically only for proximal colon cancer (but not for all colorectal cancers combined and distal colon and rectal cancers). Among black males and females, no significant differences were observed except for rectal cancer among black males (8%) between the middle- and high-poverty counties (Table 1).
|Race/Sex/Anatomicsubsite||Subsite distribution||Age-adjusted rate by poverty level||Rate ratios (95% Confidence interval)|
|Lowc (%)||Middled (%)||Highe (%)||Lowc||Middled||Highe||Lowc/Highe||Middlec/Highe|
|White males (n = 205,925)|
|Colon and rectumf||100.0||100.0||100.0||66.8||64.5||64.8||1.03 (1.01–1.05)*||1.00 (0.98–1.01)|
|Proximal colong||37.5||37.1||35.1||25.7||24.4||23.2||1.11 (1.08–1.14)*||1.05 (1.02–1.08)*|
|Distal colonh||26.7||26.7||27.9||17.6||17.1||17.8||0.99 (0.95–1.02)||0.96 (0.93–0.99)|
|Rectumi||30.5||30.5||30.7||19.9||19.4||19.5||1.02 (0.99–1.05)||0.99 (0.96–1.02)|
|White females (n = 200,072)|
|Colon and rectumf||100.0||100.0||100.0||48.1||46.2||45.2||1.06 (1.04–1.08)*||1.02 (1.00–1.04)|
|Proximal colong||46.4||45.9||43.0||21.9||20.7||19.0||1.15 (1.12–1.18)*||1.09 (1.06–1.12)*|
|Distal colonh||23.6||23.6||25.1||11.6||11.1||11.6||1.00 (0.96–1.03)||0.96 (0.93–1.00)|
|Rectumi||24.1||24.3||24.8||11.9||11.6||11.5||1.03 (0.99–1.07)||1.01 (0.97–1.04)|
|Black males (n = 19,980)|
|Colon and rectumf||100.0||100.0||100.0||68.3||68.9||66.7||1.02 (0.97–1.09)||1.03 (1.00–1.07)|
|Proximal colong||42.3||40.6||41.2||29.6||28.9||28.1||1.05 (0.97–1.14)||1.03 (0.97–1.09)|
|Distal colonh||24.8||26.3||26.4||17.1||18.1||17.4||0.98 (0.89–1.07)||1.04 (0.97–1.11)|
|Rectumi||26.3||25.6||23.9||16.8||16.6||15.3||1.10 (0.99–1.20)||1.08 (1.01–1.16)*|
|Black females (n = 22,640)|
|Colon and rectumf||100.0||100.0||100.0||51.6||52.7||52.1||0.99 (0.95–1.04)||1.01 (0.98–1.04)|
|Proximal colong||46.2||46.3||45.7||24.4||24.6||23.8||1.02 (0.96–1.09)||1.03 (0.98–1.08)|
|Distal colonh||26.4||25.3||25.4||13.4||13.3||13.2||1.01 (0.92–1.09)||1.00 (0.94–1.07)|
|Rectumi||20.6||21.0||20.7||10.3||10.9||10.8||0.96 (0.86–1.03)||1.01 (0.95–1.09)|
Associations of Subsite-specific Stage-specific Incidence Rates With the County Poverty Level
For proximal colon cancer, the age-adjusted incidence rates of late-stage disease (regional and distant) were statistically higher (52%–82%) than the rates of early-stage disease (in situ and localized) regardless of race, sex, or county poverty level, and the late-to-early stage incidence rate ratios were similar across county poverty levels within the same race and sex groups (Table 2). In contrast, for distal colon and rectal cancers, the differences in the rates between late-stage and early-stage disease were much less and the late-to-early stage rate ratios varied strikingly by county poverty level. The late-to-early stage rate ratios were much higher in the high-poverty counties than in the low- and middle-poverty counties among white males and black males for distal colon and rectal cancers, among white females for distal colon cancer, and among black females for rectal cancer.
|Race/Sex/Anatomic subsite||Low povertyc||Middle povertyd||High povertye|
|Early||Late||Rate ratio (95% CI)||Early||Late||Rate ratio (95% CI)||Early||Late||Rate ratio (95% CI)|
|White males (n = 187,554)|
|Colon and rectumf||101.5||123.9||1.22 (1.20–1.23)*||93.5||117.6||1.25 (1.23–1.27)*||88.0||117.9||1.34 (1.30–1.39)*|
|Proximal colong||35.4||53.5||1.52 (1.47–1.56)*||32.4||50.1||1.54 (1.52–1.59)*||30.8||46.6||1.52 (1.43–1.61)*|
|Distal colonh||30.9||31.8||1.03 (1.00–1.05)*||28.7||30.4||1.05 (1.03–1.09)*||26.5||33.2||1.25 (1.18–1.33)*|
|Rectumi||32.1||32.6||1.02 (0.99–1.01)||29.3||31.3||1.06 (1.04–1.09)*||27.4||31.5||1.15 (1.08–1.22)*|
|White females (n = 180,354)|
|Colon and rectumf||67.7||90.7||1.33 (1.32–1.35)*||62.3||85.2||1.37 (1.35–1.39)*||59.6||82.0||1.37 (1.33–1.43)*|
|Proximal colong||28.2||46.8||1.67 (1.61–1.69)*||25.8||43.5||1.69 (1.67–1.72)*||23.6||39.4||1.67 (1.59–1.75)*|
|Distal colonh||18.6||20.5||1.09 (1.06–1.14)*||16.9||19.7||1.16 (1.14–1.19)*||17.2||20.3||1.18 (1.10–1.27)*|
|Rectumi||18.7||18.8||1.00 (0.97–1.04)||17.6||17.6||1.00 (0.97–1.02)||16.9||17.4||1.03 (0.96–1.11)|
|Black males (n = 17,026)|
|Colon and rectumf||90.9||134.6||1.47 (1.35–1.61)*||86.6||133.6||1.54 (1.49–1.61)*||84.0||129.8||1.54 (1.45–1.67)*|
|Proximal colong||37.8||64.6||1.72 (1.49–1.96)*||34.8||62.4||1.79 (1.69–1.89)*||35.7||59.0||1.67 (1.49–1.82)*|
|Distal colonh||26.7||33.3||1.25 (1.05–1.47)*||26.7||35.1||1.32 (1.22–1.41)*||25.4||34.8||1.36 (1.20–1.56)*|
|Rectumi||23.2||29.4||1.27 (1.06–1.52)*||21.9||26.5||1.20 (1.11–1.32)*||18.7||27.0||1.45 (1.25–1.67)*|
|Black females (n = 19,193)|
|Colon and rectumf||64.1||98.6||1.54 (1.41–1.67)*||65.3||99.8||1.54 (1.47–1.59)*||63.9||100.2||1.56 (1.47–1.67)*|
|Proximal colong||29.1||51.3||1.76 (1.56–2.00)*||29.6||51.5||1.72 (1.64–1.82)*||28.5||51.6||1.82 (1.67–1.96)*|
|Distal colonh||18.1||25.7||1.41 (1.22–1.67)*||18.2||25.1||1.37 (1.28–1.47)*||19.0||23.9||1.25 (1.12–1.41)*|
|Rectumi||14.3||15.6||1.10 (0.91–1.32)||14.7||16.8||1.15 (1.05–1.23)*||13.5||18.2||1.35 (1.18–1.54)*|
This is the first study to examine the association of the subsite-specific incidence of colorectal cancer with county-level poverty in the United States. This study found that incidence rates for all colorectal cancers combined were slightly higher among white males and females in low-poverty counties than in high-poverty counties. These differences were observed for proximal colon cancer only, not for distal colon and rectal cancers. Although subsite-specific risk factors for colorectal cancer have not been definitely determined, the current literature suggests that people with low SES are more likely than those with high SES to be exposed to certain factors that increase risk for colorectal cancer.21–26 According to well-documented historical risk factor patterns by SES, it is possible that our results partly relate to different profiles of risk factor distributions by SES, which were apparent years before. It has been suggested that, decades ago, people with high SES may have consumed more meat and fat than those with low SES.6 The substantial positive relationship between area SES and colorectal cancer mortality in the United States during the 1950s and 1960s and between 1975 and the early 1990s supports this speculation.7, 27 Although the dietary habits in high SES populations have changed over time, the effect of those earlier habits on colorectal cancer incidence could have persisted because of the long latency between exposure to risk factors and diagnosis of colorectal cancer.
In this study it is not clear why the differences in the incidence rates between low- and high-poverty counties in white males and females were observed for proximal colon cancer but not for distal colon and rectal cancers. Colorectal cancer screening may have played a part. Because colorectal cancer screening varies by area poverty and its effect also differs by colorectal cancer subsite,7, 12, 13, 28, 29 screening may be partly responsible for the observed subsite-specific incidence pattern. We are uncertain whether other factors contribute to this pattern.
Our study found that the association between county-level poverty and proximal colorectal cancer incidence rates occurred among whites but not among blacks. Area-level socioeconomic measures (such as income) have been shown to correlate better with individual-level socioeconomic measures for whites than for blacks.30 Thus, it seems legitimate that an association between county-level poverty and colorectal cancer incidence would be more discernible among whites than among blacks.
Our study is the first to report that age-adjusted late-to-early stage rate ratios are higher in high-poverty than in low-poverty counties for distal colon and rectal cancers, but not for proximal colon cancer. The finding indicates that those residing in counties with high levels of poverty are more likely to receive a diagnosis of late-stage disease than those residing in counties with low levels of poverty. Previous studies have suggested that living in a low-SES area is the most important determinant of late stage of colorectal cancer for all age, race, sex, and source-of-care groups.31, 32 An ecological analysis of the Behavioral Risk Factor Surveillance System (BRFSS) has shown that states with a high rate of poverty tend to have a lower prevalence of colorectal cancer screening utilization.7 The data from the 2000 National Health Interview Survey (NHIS) also showed that people in low-SES groups, indicated by low income and low educational attainment, were less likely to have colorectal cancer screening than were their counterparts in high-SES groups.33 The variation of late-to-early stage incidence rate ratios by county poverty reflects the actual effect of cancer screening on stage-specific incidence rates.
The similar late-to-early stage rate ratios across county poverty levels for proximal cancer regardless of race and sex may suggest that the impact of colorectal cancer screening on reducing late-stage of disease does not differ for proximal colon cancer by county poverty level. Stage shift from late-to-early disease due to screening for proximal colon cancer may not be apparent because the use of colonoscopy for screening purposes generally is still low. Although colorectal cancer screening has been shown to be effective in reducing the incidence and mortality rates of this disease,34–38 its effect on the incidence rates of proximal colon cancer has not been reported. To better understand the impact of screening on proximal colon cancer incidence rates, more research is needed.
Three limitations of this study must be noted. First, poverty level may vary greatly within some counties. County-level poverty may not reflect the true area characteristics of poverty level.17 The possible misclassification of poverty level may diminish or increase the differences in subsite-specific and stage-specific colorectal cancer incidence rates and rate ratios across county poverty groups. Measures for smaller areas such as census tract would be preferable when they are available. Second, migration between different county poverty areas may have an effect on the observed incidence and stage patterns. However, since the information about migration of county populations over the years was not available for this study, such an effect cannot be assessed. Third, the stage distributions were based on cases with known stage, which may bias the stage statistics. There were 3 state cancer registries that did not report in situ cases for this study. The late-to-early stage rate ratios could be slightly higher than what they should be. Since the staging data from these 3 states accounted for only a small proportion of the cases, our results are unlikely to be affected by the absence of those in situ cases.
Many questions remain unanswered regarding the variations of subsite-specific colorectal cancer incidence by race and sex. This study is the largest thus far to examine these relationships. Although we cannot provide definitive answers for observed subsite-specific colorectal cancer incidence patterns by county poverty level, our findings may stimulate ideas and research strategies to further enhance cancer control and prevention for colorectal cancer. The variations in late-to-early stage incidence rate ratios by county poverty level suggest an effect of colorectal cancer screening. This finding supports the need to target socioeconomically disadvantaged areas such as high poverty counties.
The authors appreciate the in-kind support from all the contributors to this monograph and also are grateful to Jessica King for contributing to the preparation of analytic files and to Faruque Ahmed for his leadership of the colorectal cancer monograph project
- 7Area socioeconomic variations in U.S. cancer incidence, mortality, stage, treatment, and survival, 1975–1999. NCI cancer surveillance monograph series, No. 4. Bethesda, MD: National Cancer Institute; 2003. NIH publication no. 03-5417., , , .
- 14Descriptive epidemiology of colorectal cancer in the United States, 1998–2001. Cancer. 2006; 107: 5(Suppl)., , , , .
- 15U.S. Cancer Statistics Working Group. United States cancer statistics: 2000 incidence. Atlanta: Department of Health and Human Services, Centers for Disease Control and Prevention and National Cancer Institute, 2003.
- 16International Classification of Diseases for Oncology, 2nd ed. Geneva, Switzerland: World Health Organization; 1990., , .
- 18NIH. Summary staging guide for the cancer Surveillance, Epidemiology, and End Results Reporting (SEER) program. Bethesda, MD: National Cancer Institute, National Institutes of Health; 1977. Surveillance, Epidemiology, and End Results Program. NIH publication no. 86–2313. (Reprinted July 1986.)
- 19NIH. Summary Staging Manual 2000. Bethesda, MD: National Cancer Institute, National Institutes of Health; 2001. Surveillance, Epidemiology, and End Results Program.
- 20Statistical Methods in Cancer Res, Vol. IV: Descriptive Epidemiology. Lyon, France: International Agency for Research on Cancer; 1994., , .
- 26National Center for Health Statistics. Healthy people 2000 Final Review. Hyattsville, MD: Public Health Service; 2001.
- 27Cancers of the large intestine. In: SchottenfeldD, FraumeniJF, eds. Cancer Epidemiology and Prevention, 2nd ed. New York: Oxford University Press; 1996: 813–840., .
- 29Incidence of colorectal cancer in West Virginia from 1993–1999: An update by gender, age, subsite and stage. W V Med J. 2003; 99: 182–186., , , .