Population-based study of the geographic variation in colon cancer incidence in Alabama: relationship to socioeconomic status indicators and physician density.

Objectives: The objective of this population-based study was to examine the relationship between race, socioeconomic characteristics (socioeconomic status, SES), physician density, and colon cancer incidence in Alabama.

Methods: Data for 5,788 colon cancer cases from 1996 to 1999 provided by the Alabama Statewide Cancer Registry are linked to county-level measures of SES, including median household income, percentage of high school graduates, percentage of families below poverty level, and occupational and health care factors. Poisson regression is used to model the predictors adjusting for age, gender, and race.

Results: Blacks had higher incidences of colon cancer compared with whites and presented with later stages (20.4% versus 14.8% for distant disease (P = 0.0089). After controlling for race, gender, and age at diagnosis, significant associations were detected between colon cancer incidence and higher education (RR = 1.10; 95% CI, 1.03-1.17), and increased number of physicians per 1,000 (RR = 1.14; 95% CI, 1.06-1.22). The county percentage of families below poverty is associated inversely with localized disease and positively with distant stage.

Conclusions: Colon cancer incidence varies geographically across Alabama and is positively related to aggregate SES factors, including education and physician density. A higher incidence of distant disease is related to black race and increased poverty. Health disparities in colon cancer across Alabama warrant further investigation.

Key Words: colon cancer, race, socioeconomic status, incidence, disparities

*********

Despite reported declines in mortality rates due to colorectal cancer (CRC) in the United States, (1,2) this neoplasm remains the second leading cause of cancer death for the nation and an important health concern for all adults and for minority groups in particular. In addition, striking differences between blacks and whites have been found for CRC incidence (62.6/100,000 for blacks and 53.9/100,000 for whites) and for CRC-related deaths (28.5/100,000 for blacks and 20.7/100,000 for whites). (1,3) Recently, 5-year relative survival rates for all stages of colon cancer were 63% for whites compared with 53% for blacks in the United States. (4) Similar patterns of CRC incidence and mortality rates have been observed for the state of Alabama (1996 to 2000), where age-adjusted incidence rates per 100,000 are reported at 57.9 and 45.2 for black men and women, respectively, compared with 57.0 and 39.2 for white men and women, respectively. Reported death rates for the same period were 33.0% and 21.5% for black men and women, respectively, compared with 22.7% and 15.2% for white men and women, respectively. (5)

Measures of socioeconomic status (SES) have been found to be prominently associated with these differences in CRC incidence among population groups in several epidemiologic investigations. In many countries, a positive socioeconomic gradient has been observed with an increased risk for colon cancer incidence, associated with higher education levels, (6) white-collar occupations, (7) and higher income. (8) A lower risk of colon cancer was also found to be associated with lower social strata in both genders. (9) However, such a positive social gradient with regard to CRC incidence has not been consistently supported by other data, particularly in the United States, where many investigations have reported higher CRC incidence and mortality rates among groups of lower social status and lower income. (10-14) The general degree of poverty of an area, independent of other environmental and individual risk factors, has been shown (15-17) to affect the risk for certain diseases and influence health outcomes and quality of life of individuals directly or indirectly. Marcella et al, (18) investigating socioeconomic differences in mortality rates from CRC in the United States, found that cases from communities with a low percentage of high school graduates and high unemployment tended to have higher colorectal cancer death rates. The findings for CRC and SES are more consistent for death than incidence.

Motivated by the geographic variation in colorectal cancer incidence by gender and race across Alabama, we investigated the association of selected area-based SES indicators on cancer incidence.

Methods

Study sample and variables

The Alabama Statewide Cancer Registry (ASCR) provided data on colon cancer cases for the years 1996 to 1999. Only cases coded as colon cancer (C18.0-C18.9) with complete data on race and gender over 14 years of age were included in this investigation. Cases coded as rectal cancer (C19.9-C20.9) were excluded because the clinical and epidemiologic features of rectal cancer have been considered different from those for colon cancer. For descriptive maps, age-standardized rates were computed for each gender and race group by county using the 2000 US population and age-adjusted weights (http://www.cdc.gov/nchs/data/statnt/statnt20.pdf). Geographic maps were created in EpiMap (http://www.cdc.gov), with shape files from the geological survey of Alabama (http://www.gsa.state.al.us).

Data from the 1990 Census for the State of Alabama (19) were used to derive area-based SES characteristics, specific to each county. The main measures of area-based SES selected for this analysis were education (percentage of 25-year-old individuals and older who had completed 12 years of schooling), income (median household income), and percentage of families below the poverty level. Secondary measures included occupation as assessed by the percentage of patients 16 years and older employed in the civilian labor force (finance, insurance and real estate, public administration, wholesale and retail trade, manufacturing, agriculture/forestry/fishing, and the percentage employed in health services), and the unemployed civilian population, aged 16 years and older. In addition, access to care was assessed by both the number of practicing physicians and the number of hospital beds per 1,000 people. Population estimates for the 5-year age strata (15 to 19, 20 to 24,... 80 to 84, > 85) for Alabama counties for 1996 to 1999 were taken from the US 1990 Census and used as denominators in computing cancer incidence rates.

Statistical analyses

Descriptive statistics were computed as mean [+ or -] SD or proportions, as indicated. Linear correlations were computed as Pearson correlations. The outcome variable for modeling is the colon cancer incidence rate computed for each gender, race, 5-year age strata, and county during the time period of 1996 to 1999. Independent predictors included age as a continuous variable (midpoint of 5-year intervals), gender (0 = M, 1 = F), and race (0 = white, 1 = black). Independent county level predictors included median household income (in $1,000.00 units), percentage of high school graduates, percentage of families below poverty level, and occupational factors. Significant univariate factors were considered in a multivariable model, and models were compared by the deviance criterion for assessing goodness of fit. In this paper, GLIMMIX, an extension of PROC MIXED available from SAS[R] (Version 9.0 SAS Institute Inc, SAS[R], Cary, NC) was used to develop the estimates for the coefficients. A sample code for the model is given in the Appendix. For all analyses, a P value of less than 0.05 was deemed statistically significant.

Results

A total of 5,788 colon cancer cases were included in the analysis. The proportion of male cases was 48.1%. The age distribution of these cases was 15 to 29 (0.48%), 30 to 49 (7.98%), 50 to 69 (37.41%), and 70 and over (54.13%). The median age at diagnosis was 71.0 years. The geographic variation in colon cancer rates obtained from the ASCR for the time period of 1996 to 1999 for white and black males are mapped in quartiles in Figure 1, a and b. Although there is similarity in the rates for males across races, there is some variation geographically. For females, the geographic variation by race is more marked, with black females displaying higher rates than white females (Fig. 2, A and B).

The State of Alabama displays heterogeneity in resources with greater rates of practicing physicians concentrating in the urban areas (such as Jefferson, Mobile, Montgomery, and Tuscaloosa counties) compared with rural areas (such as Coosa, Henry, Lowndes, and Washington counties). There is greater educational attainment in the urban areas, and highest poverty in the counties in the so-called "black belt" region. Overall, the state has a mean high school educational attainment at 59.8%, which ranks 46th in the nation for high school completion. The mean poverty level is 17.5%, with the lowest median incomes in Greene ($11,990) and Wilcox counties ($12,437) and highest in Shelby County ($36,852). The majority of the individuals who live below the poverty threshold have no health insurance.

[FIGURE 1 OMITTED]

Pearson correlation coefficients were calculated for all county level indicators (data not shown). As expected, a significant percentage of high school graduates and those with median household income are positively correlated, whereas the percentage of families below the poverty level are inversely correlated with both a percentage of high school graduates and median household income. The percentage of persons employed in manufacturing or unskilled labor is negatively correlated with income and education. The number of physicians per 1,000 is significantly related to education, income, and employment in health services.

Univariate

The results of Poisson models of single factors and their relationship to colon cancer for all stages as well as stratified by stage are shown in Table 1. For all cases, the number of physicians was related to increased incidence (RR = 1.14; 95% CI, 1.06-1.22). For each 10% increase in the percentage of persons employed in health services, the colon cancer incidence rate also increased by 50% (RR = 1.52; 95% CI, 1.20-1.52). Other factors significantly associated with an increase in colon cancer incidence were percent employed in wholesale/retail trades and in public administration. By contrast, factors such as the percentage of persons employed in manufacturing and in agriculture/forestry/fishing were negatively associated with colon cancer incidence. As the percentage of persons employed in manufacturing increased, colon cancer incidence rates decreased (RR = 0.92; 95% CI, 0.87-0.97), and for each 10% increase in persons employed in agriculture/forestry/fishing, the colon cancer rate decreased by about 30% (RR = 0.69; 95% CI, 0.51-0.93).

[FIGURE 2 OMITTED]

For localized disease alone, race was not a significant factor. Controlling for race, age, and gender, there was a significant but inverse relationship between incidence of localized colon cancer and employment in agriculture/forestry/fishing (RR = 0.52; 95% CI, 0.34, 0.80). For every 10% increase in percentage employed in these occupations, there is a corresponding decrease in risk of localized disease (RR = 0.52; 95% CI, 0.34, 0.80). The effect of physicians per 1,000 in the same model is only of borderline significance (P = 0.06). Education was positively associated to incidence of localized disease, but this association is not statistically significant (P = 0.07). An increase in percentage of families below poverty is associated with a slight decrease in localized disease (RR = 0.87; 95% CI, 0.77-0.98), whereas income is of borderline significance (P = 0.05).

For regional stage, race is not significant. SES factors significantly related to regional colon cancer included percentage of high school graduates (RR = 1.10 for every 10% increase), physicians/100,000 (RR = 1.16 for every 100 increase), employment in health services (RR = 1.72 for every 10% increase), employment in wholesale and retail trades (RR = 1.34 for every 10% increase), and employment in manufacturing (RR = 0.90 for every 10% increase).

For the small proportion of patients with distant disease, race (RR = 1.46; 95% CI, 1.29, 1.66) and education level (RR = 1.13; 95% CI, 1.11, 1.16) are main factors significantly associated with an increase in incidence of advanced stage of disease in a model including age and gender. Other factors related to increased distant disease incidence included employment in health services and wholesale/retail trade. Employment in manufacturing is inversely related to distant disease incidence (RR = 0.86; 95% CI, 0.79, 0.83).

Multivariate 

In multivariable models of all stages of disease, including both education and number of physicians, the positive effect of the number of physicians (RR = 1.08) persisted, whereas that of education became nonsignificant (P = 0.09). Likewise, when education and occupation in manufacturing were considered in the same model, the effect of education was no longer significant (P = 0.46). More of the variability in the total colon cancer incidence rates between counties in this total model was explained by the number of physicians per 1,000 (reflecting access to care) than by either education or occupation. The model including age, race, gender, and number of physicians had a smaller deviance compared with the model including age, race, gender, and education.

For individuals diagnosed with incident localized disease, in a multivariable model including gender and age, the significant factor was county level employment in agriculture, forestry, and fishing with individuals residing in these areas having a reduced risk (RR = 0.94; 95% CI, 0.90, 0.98). The primary predictor of incident regional disease in a model including age and gender was number of physicians per 1,000 (RR = 1.01, P = 0.002). The best fit model for distant disease included age, gender, black race (RR = 1.45), and education (RR = 1.13).

Discussion

Our results confirm earlier findings of higher incidences of colon cancer in counties with higher SES indicators and among blacks. (2,12,13,20,21) The finding of an increased risk for colon cancer among blacks, even after adjusting for age at diagnosis, gender, and county-based SES, suggests that given the same SES level, blacks have higher overall colon cancer rates than whites.

There is a positive effect of the number of county physicians and level of education on the total colon cancer incidence, which is not mediated by age, gender, income, or race. Furthermore, the number of physicians in a county appears to explain more of the intercounty variability in colon cancer rates for all stages than did education, occupation, income or poverty. Lack of routine visits to primary care physicians is indeed recently identified as one of the strongest predictors for inadequate CRC screening and advanced stage at presentation. (22,23)

Characteristics and factors associated with patients and providers, (24-26) as well as system-specific barriers, (23,27,28) have been connected to the underutilization of CRC screening despite its established effectiveness. The Behavioral Risk Factor Surveillance System (BRFSS) in 1999 reported that only 33.6% of responders 50 years or older in the United States (19,535 of 63,000) had sigmoidoscopy/colonoscopy within the preceding 5 years. (8) Between 1995 and 2001, only 35 to 39% of eligible persons in Alabama (39% whites and 33% blacks), reported having had a sigmoidoscopy/colonoscopy as recommended. (5)

The positive relationship of SES with digestive cancers has been hypothesized to reflect a lifestyle in which individuals have a dietary intake with higher levels of nutrients such as fats and lower levels of fruit and vegetable consumption, which causes a positive energy balance over an extended period of time, known to be associated with higher risk for colon cancer. (29,30) Recent dietary guidelines aimed at colorectal cancer prevention were unlikely to have been adopted in a time frame relevant for individuals included in this analysis. Other lifestyle factors such as physical activity have been shown to have a significant and independent impact on colon cancer incidence and stage of diagnosis. (30-32) As noted by Slattery, (30) 13% of colon cancers can be attributed to physical inactivity. Epidemiologic studies found that both occupational and recreational activities were associated with lower risk of colorectal cancer. (33,34) A meta-analysis of the association of physical activity and colorectal cancer has shown a reduced risk in physically active males for occupational and recreational activities and a reduced risk in women for increased recreational activities. (35)

Although data about individual income, education, and behavior was not available for our analysis, we could hypothesize that both recreational and occupational activities play a protective role in colon cancer. Our data show a decreased risk of localized disease in counties with a greater proportion of occupations that include agriculture, forestry, and fishing. In support of this theory, findings from a study of mortality rates among Alabama workers and farmers by Austin et al suggested a "lower mortality from all malignant neoplasms, specifically for cancers of the respiratory and digestive systems" among Alabama farmers. (36) It is likely that the effect of occupation on incidence may be mediated by physical activity.

The striking finding in this study is that among patients diagnosed with distant disease, there is an independent effect of race in models that include education, type of employment, or physician density. Even residing in counties with higher levels of education and white-collar employment, blacks were more likely to have higher distant disease rates than whites. Although occupational status and education show differences in the strength of association with distant cancer incidence, the patterns of association were the same. Therefore, our findings reinforce the conclusion reached by others (12,37,38) that the patterns of incidence of distant colon cancer among blacks are not completely explained by SES factors alone and that other cultural or genetic factors may be at play.

An unexpected finding is that our investigation did not reveal a direct significant association between colon cancer incidence rates and income alone, as reflected by median household income. One possible explanation for this finding may be that median household income is not a good measure of income. As noted by Gordon, (39) there is only a 55 to 75% overlap between low income and poor households, reflecting the fact that low income may not consistently correlate with actual poverty.

In Alabama, almost half of the population lives in rural areas that are potentially underserved, and according to recent Census estimates, (40) the state ranked 6th in the percentage of people living below the poverty level. In addition, although they constitute a quarter of the state population, 31% of blacks had income determined to be below the federal poverty level. (41) The current disparity in health care delivery in Alabama is related to colon cancer incidence and possibly to diagnosis. In general, when compared with white non-Hispanics, blacks have higher rates of poverty, lower educational status, and less access to health care coverage or to a regular source of primary care. All these characteristics have been strongly linked to lower utilization of screening and delayed diagnosis for CRC, which directly affects mortality rates in these population groups, since survival rate for CRC is 90% when diagnosed at a localized stage but declines to 10% when the cancer has reached distant organs. (42)

These analyses have several limitations. Several potential threats to the validity of the results were considered. In this investigation, county of diagnosis may not measure an individual's lifetime exposure history, and county-level variables may not measure an individual's SES. County measures were selected as convenient geographical units, and it is not clear whether reducing the unit to a smaller scale, such as a census tract, would lead to a better measure. Smaller units will be more socioeconomically homogeneous, but it may be the characteristics of the larger surrounding area that are important in assessing the cancer incidence of the smaller units. The small number of cancer cases in each county (median, 0 cases) by each race/gender/age group precluded analysis at units smaller than the county level.

The completeness of the Alabama Statewide Cancer Registry was considered in summarizing the validity of our study's conclusions. The case ascertainment rate for colorectal cancer averaged over 95% in white males and about 91% in white females from 1996 to 1999. The ascertainment rate averaged 88% in black males and 91% in black females. The expected rates were derived using the Surveillance Epidemiology and End Results (SEER) rates in the calculations, and previous reports have shown that cancer incidence rates in predominantly black, rural-poor populations in the southern states are about 15% lower for colorectal cancer than for the SEER sites. (43) In general, SEER rates tend to provide higher estimates of the number of cases of Alabama counties than actual numbers collected from surveillance data. Therefore, the cancer experience probably is different for the Alabama black population, and the use of the ASCR cancer data is valid for this investigation.

Conclusions

The findings in this study that relate to SES can be viewed as proxies for unmeasured mechanisms or actual exposures. The causal mechanisms that link the SES effects to health outcomes may operate on pathways that have yet to be identified. Colon cancer incidence varies geographically across Alabama and is positively related to aggregate SES factors, including education and physician density. Incidence of distant disease is related to black race and increased poverty. Health disparities in colon cancer across Alabama warrant further investigation.

Appendix

Sample SAS code for Models (note: requires macro GLIMMIX, can be downloaded from SAS Institute):

% glimmixs (data = ses, stmts = %str(class county; model sumcount/sumagpop = gender age racer perpov/solution; random county;), error = poisson, link = log) run

References

1. Weir HK, Thun MJ, Hankey BF, et al. Annual report to the nation on thestatus of cancer, 1975-2000, featuring the uses of surveillance datafor cancer prevention and control. J NatlCancer Inst 2003;95:1276-1299.

2. Ries LA, Wingo PA, Miller DS, et al. The Annual Report to the Nation on the Status of Cancer, 1973-1997, with a special section on colorectal cancer. Cancer 2000;88:2398-2424.

3. Ries LAG, Eisner MP, Kosary CL, et al (eds) SEER Cancer Statistics Review, 1973-1998. Bethesda, MD: National Cancer Institute, 2001.

4. Ries LAG, Eisner MP, Kosary CL, et al (eds). SEER Cancer Statistics Review, 1973-1999 National Cancer Institute, Bethesda MD. Table VI-1. Available from URL: http://seer.cancer.gov/csr/1973_1999/, 2002 [Accessed March, 2004].

5. Alabama Statewide Cancer Registry (ASCR). Alabama Cancer Facts & Figures, 2003. Alabama Department of Public Health. Available at: http://www.adph.org/cancer_registry. Accessed May, 2004.

6. Tavani A, Fioretti F, Franceschi S, et al. Education, socioeconomic status and risk of cancer of the colon and rectum. Int J Epidemiol 1999;28:380-385.

7. Hsing AW, McLaughlin JK, Chow WH, et al. Risk factors for colorectal cancer in a prospective study among US white men. Int J Cancer 1998;77:549-553.

8. Pukkala E. Cancer risk by social class and occupation. A survey of 109,000 cancer cases among Finns of working age. Cited in: Kogevinas M, Pearce N, Susser M and Boffetta P. (eds.) Social Inequalities and Cancer, IARC Scientific Publications No. 138. International Agency for Research on Cancer, Lyon, 1997.

9. Faggiano F, Partanen T, Kogevinas M, Boffetta P. Socioeconomic Differences in Cancer Incidence and Mortality. In: Kogevinas M, Pearce N, Susser M, and Boffetta P, eds. Social Inequalities and Cancer. IARC Scientific Publications No 138, International Agency for Research Against Cancer, World Health Organization, 1997.

10. Singh GK, Miller BA, Hankey BF, Edwards BK. Area Socioeconomic Variations in U.S. Cancer Incidence, Mortality, Stage, Treatment, and Survival, 1975-1999. NCI Cancer Surveillance Monograph Series, Number 4. Bethesda, MD: National Cancer Institute, 2003. NIH Publication No. 03-5417.

11. Freeman HP, Alshafie, TA. Colorectal Carcinoma in poor blacks. Cancer 2002;94:2327-2332.

12. Mandelblatt J, Andrews H, Kao R, et al. The late diagnosis of colorectal cancer: demographic and socioeconomic factors. Am J Public Health 1996;86:1794-1797.

13. Mayberry RM, Coates RJ, Hill HA, et al. Determinants of black/white differences in colon cancer survival. J Natl Cancer Inst 1995;87:1686-1693.

14. Dayal H, Polissar L, Yang CY, et al. Race, socioeconomic status and other prognostic factors for survival from colorectal cancer. J Chron Dis 1987;40:857-864.

15. Waitzman NJ, Smith KR. Phantom of the area: poverty-area residence and mortality in the United States. Am J Public Health 1998;88:973-976.

16. Troutt, DD. The thin red line: How the poor still pay more. West Coast Regional Office of Consumers Union, San Francisco, 1994.

17. Haan MM, Kaplan G, Camacho, T. Poverty and health: prospective evidence from the Alameda County Study. Am J Epidemiol 1987;125:989-998.

18. Marcella S, Miller JE. Racial differences in colorectal cancer mortality: the importance of stage and socioeconomic status. J Clin Epidemiol 2001;54:359-366.

19. US Bureau of the Census. Census of Population and Housing, 1990: Summary tape of file 3, Technical documentation. Washington, DC: US Bureau of the Census, 1991.

20. Cooper GS, Yuan Z, Rimm AA. Racial disparity in the incidence and case-fatality of colorectal cancer: analysis of 329 United States counties. Cancer Epidemiol Biomarkers Prev 1997;6:283-285.

21. Gorey KM, Vena JE. The association of near poverty status with cancer incidence among black and white adults. J Community Health 1995;20:359-366.

22. Ioannou N, Chapko MK, Dominitz JA. Predictors of colorectal cancer screening participation in the United States. Am J Gastroenterol 2003;98:2082-2091.

23. Zapka JG, Puleo E, Vickers-Lahti M, et al. Healthcare system factors and colorectal cancer screening. Am J Prev Med 2002;23:28-35.

24. Menon U, Champion VL, Larkin GN, et al. Beliefs associated with fecal occult blood test and colonoscopy use at a worksite colon cancer screening program. J Occup Environ Med 2003;45:891-898.

25. Shapiro JA, Seeff LC, Nadel MR. Colorectal cancer-screening tests and associated health behaviors. Am J Prev Med 2001;21:132-137.

26. Cooper GS, Fortinsky RH, Hapke R, et al. Primary care physician recommendations for colorectal cancer screening: patient and practitioner factors. Arch Intern Med 1997;157:1946-1950.

27. Cokkinides VE, Chao A, Smith RA, et al. Correlates of underutilization of colorectal cancer screening among US adults, age 50 years and older. Prev Med 2003;36:85-91.

28. US Preventive Task Force. Screening for colorectal cancer: recommendations and rationale. Ann Intern Med 2002;137:129-131.

29. Slattery ML, Potter, JD. Physical activity and colon cancer: confounding or interaction? Med Sci Sports Exerc 2002;34:913-919.

30. Slattery ML. Diet, lifestyle and colon cancer. Semin Gastrointest Dis 2000;11:142-146.

31. Slattery ML, Edwards S, Curtin K, et al. Physical activity and colorectal cancer. Am J Epidemiol 2003;158:214-224.

32. Thune I, Lund E. Physical activity and risk of colorectal cancer in men and women. Br J Cancer 1996;73:1134-1140.

33. Tomeo CA, Colditz GA, Willett WC, et al. Harvard Report on Cancer Prevention, Volume 3: prevention of colon cancer in the United States. Cancer Causes Control 1999;10:167-180.

34. Colditz GA, Cannuscio CC, Frazier AL. Physical activity and reduced risk of colon cancer: implications for prevention. Cancer Causes Control 1997;8:649-667.

35. Samad AK, Taylor RS, Marshall T, Chapman MA. A meta-analysis of the association of physical activity with reduced risk of colorectal cancer. Colorectal Dis 2005;7:204-213.

36. Austin SG, Huang N, Woernle CW. PMR study of mortality among Alabama workers and farmers. Am J Int Med 1995;27:29-36.

37. Williams DR. Race/ethnicity and socioeconomic status: measurement and methodological issues. Int J Health Serv 1996;26:483-505.

38. Baquet CR, Horm JW, Gibbs T, et al. Socioeconomic factors and cancer incidence among blacks and whites. J Natl Cancer Inst 1991;83:551-557.

39. Gordon D. Census based deprivation indices: their weighting and validation. J Epidemiol Community Health 1995;49(Suppl 2):S39-S44.

40. US Census: American Community Survey Ranking Tables, 2002-Percent of People Below Poverty Level (State Level). Available at: http://www.census.gov/acs/www/Products/Ranking/2002/R01T040.htm.

41. US Census Bureau, Census 2000. Detailed Tables, American Fact Finder. P159B. Poverty status in 1999 by Age in Black or African American Alone. Available at: http://factfinder.census.gov.

42. American Cancer Society. Colorectal Cancer Facts and Figures Special Edition 2005. Atlanta: American Cancer Society, 2005.

43. Weiss HL, Soong SJ, Partridge EE, et al. Cancer incidence among predominantly black, rural-poor populations in southern states. South Med J 1997;90:986-992.

It is true that free inquiry is integral to our humanity, but so too is
morality. So the quest for knowledge must be tempered with moral
concern.
--Bernard Rollin

Michele P. L. Shipp, MD, DRPH, Renee Desmond, PHD, Neil Accortt, PHD, Reda J. Wilson, MPH, Mona Fouad, MD, MPH, and Mohamad A. Eloubeidi, MD, MHS 

From the School of Public Health, Ohio State University, Columbus, OH; Comprehensive Cancer Center, University of Alabama at Birmingham; Cancer Surveillance Branch, Division of Cancer Prevention and Control, Centers for Disease Control and Prevention, Atlanta, GA; the Department of Medicine, Division of Preventive Medicine, University of Alabama at Birmingham; and the Department of Medicine, Division of Gastroenterology and Hepatology, University of Alabama at Birmingham.

Reprint requests to Dr. Mohamad A. Eloubeidi, 408 Lyons Harrison Research Building, 701 19th Street South, Birmingham, AL 35294-0007. Email: [e-mail]

Accepted July 29, 2005.

This study was supported by a grant from the National Cancer Institute (5 KO7 CA92142).

RELATED ARTICLE: Key Points

* Colon cancer incidence varies geographically across Alabama and is positively related to aggregate socioeconomic status factors, including education and physician density.

* Incidence of distant disease is related to black race and increased poverty.

* Health disparities across Alabama warrant further investigation.

Table. SES factors and their relationship to colon cancer incidence in
Alabama, 1996 to 1999 (a)

                                          RR (95% CI)
                                 All cases          Localized disease
Factor                           (n = 5788)            (n = 1952)

Patient factors
  Gender (female)            0.72 (0.68, 0.75)     0.67 (0.63, 0.70)
  Age                        1.078 (1.076, 1.079)  1.081 (1.078, 1,084)
  Race (black)               1.22 (1.06, 1.19)     1.03 (0.94, 1.13)
SES factors
  High school graduates (%)  1.10 (1.03, 1.17)     1.09 (0.99, 1.19)
  Median income              1.01 (0.99, 1.02)     1.02 (1.00, 1.03)
  Poverty (%)                0.98 (0.92, 1.08)     0.87 (0.77, 0.98)
  Physicans/1,000            1.14 (1.06, 1.22)     1.10 (1.00, 1.22)
  Hospital beds/1,000        1.02 (0.99, 1.04)     1.01 (0.97, 1.05)
Occupational factors
  Agriculture, forestry,     0.69 (0.51, 0.93)     0.52 (0.34, 0.80)
    fishing (%)
  Manufacturing (%)          0.92 (0.87, 0.97)     0.96 (0.89, 1.04)
  Wholesale, retail          1.28 (1.08, 1.52)     1.24 (0.96, 1.59)
    trade (%)
  Finance, insurance, real   1.32 (0.82, 1.91)     1.02 (0.62, 1.71)
    estate (%)
  Health services (%)        1.52 (1.20, 1.92)     1.26 (0.73, 1.81)
  Public administration (%)  1.25 (1.03, 1.52)     1.11 (0.84, 1.47)
  Unemployed (%)             0.97 (0.76, 1.24)     0.75 (0.68, 1.39)

                                            RR (95% CI)
                               Regional disease      Distant disease
Factor                           (n = 2592)             (n = 928)

Patient factors
  Gender (female)            0.73 (0.68, 0.78)     0.77 (0.69, 0.86)
  Age                        1.077 (1.075, 1.079)  1.070 (1.066, 1.074)
  Race (black)               1.05 (0.96, 1.14)     1.46 (1.29, 1.66)
SES factors
  High school graduates (%)  1.10 (1.01, 1.21)     1.13 (1.11, 1.16)
  Median income              1.01 (0.99, 1.02)     1.01 (0.99, 1.03)
  Poverty (%)                1.02 (0.91, 1.15)     1.05 (0.91, 1.20)
  Physicans/1,000            1.16 (1.05, 1.12)     1.18 (1.08, 1.29)
  Hospital beds/1,000        1.02 (0.98, 1.06)     1.04 (0.99, 1.56)
Occupational factors
  Agriculture, forestry,     0.83 (0.64, 1.26)     0.64 (0.58, 1.06)
    fishing (%)
  Manufacturing (%)          0.90 (0.84, 0.97)     0.86 (0.79, 0.93)
  Wholesale, retail          1.34 (1.06, 1.70)     1.39 (1.05, 1.84)
    trade (%)
  Finance, insurance, real   1.47 (0.90, 2.41)     1.56 (0.92, 2.65)
    estate (%)
  Health services (%)        1.72 (1.25, 2.38)     1.93 (1.41, 2.64)
  Public administration (%)  1.29 (0.99, 1.70)     1.29 (0.95, 1.77)
  Unemployed (%)             1.10 (0.73, 1.53)     1.09 (0.73, 1.63)

(a) SES, socioeconomic status; RR, relative risk.
All models include gender, age, race, and either single SES factor or
single occupational factor.