Fax: (404) 639-8117
Article first published online: 3 NOV 2004
Published 2004 by the American Cancer Society
Volume 101, Issue 12, pages 2851–2858, 15 December 2004
How to Cite
Armstrong, L. R., Thompson, T., Hall, H. I., Coughlin, S. S., Steele, B. and Rogers, J. D. (2004), Colorectal carcinoma mortality among Appalachian men and women, 1969–1999. Cancer, 101: 2851–2858. doi: 10.1002/cncr.20667
The views expressed herein do not necessarily reflect the opinions of the Centers for Disease Control or the U.S. Government.
This article is a U.S. Government work and, as such, is in the public domain in the United States of America.
- Issue published online: 1 DEC 2004
- Article first published online: 3 NOV 2004
- colorectal carcinoma;
- Appalachian region;
Colorectal carcinoma screening can reduce mortality, but residents of poor or medically underserved areas may face barriers to screening. The current study assessed colorectal carcinoma mortality in Appalachia, a historically underserved area, from 1969 to 1999.
All counties within the 13-state Appalachian region, which stretches from southern New York to northern Mississippi, were used to calculate annual death rates for the 31-year period. Joinpoint regression analysis was used to examine trends by age and race for the Appalachian region and the remainder of the United States. Five-year rates for 1995–1999 age-adjusted to the 2000 U.S. standard population were calculated by race and age group for the Appalachian region and elsewhere in the United States.
Trend analysis showed that colorectal carcinoma death rates among both racial and gender groups studied had declined in recent years. Despite this, the rates for white males and white females were still significantly higher in Appalachia than in the rest of the country at the end of the study period, 1999. Five-year colorectal carcinoma death rates among white males (ages < 50, 50–59, and 70–79 years) and white females (ages < 50, 50–59, 70–79, ≥ 80 years) were significantly higher in Appalachia than elsewhere in the United States, whereas rates among black females 60–69 and 70–79 years old were significantly lower in Appalachia.
The Appalachian region may benefit from targeted prevention efforts to eliminate disparities in the colorectal carcinoma death rates among subgroups. Further studies are needed to determine whether the higher death rates in specific Appalachian subgroups are related to a higher incidence of the disease, the cancer being at a later stage at diagnosis, poorer treatment, or other factors. Cancer 2004. Published 2004 by the American Cancer Society.
Recent evidence shows that screening for colorectal carcinoma is effective in reducing mortality rates.1, 2 However, the national colorectal carcinoma screening rate is only 41% for males and 37.5% for women age ≥ 50 years. Furthermore, rates are substantially lower than those within some demographic subgroups.3 There is some evidence that death rates from colon carcinoma in some areas of Appalachia appear to be higher than in the remainder of the United States.4 To determine the extent to which this may be true, we compared colorectal carcinoma death rates in various age/race/gender subgroups in Appalachia with rates in the same subgroups elsewhere in the United States.
Some areas of Appalachia have traditionally been medically underserved.5 For many residents, poverty, unemployment, rural residence, geographic isolation, lack of education, lack of child care services, and attitudinal and cultural factors may pose barriers to cancer screening.6–8 According to the Appalachian Regional Commission, 42% of the region's 22 million people live in rural areas, compared with 20% of the national population, and the area has a poverty level of 15.3% compared with 13.1% for the entire United States.9
MATERIALS AND METHODS
Appalachia is defined by the Appalachian Regional Commission as a 200,000-square mile region that follows the spine of the Appalachian Mountains from southern New York to northern Mississippi.10 As of 1992, the area contained 407 counties, including all of West Virginia and parts of 12 other states: Alabama, Georgia, Kentucky, Maryland, Mississippi, New York, North Carolina, Ohio, Pennsylvania, South Carolina, Tennessee, and Virginia.
We calculated colorectal carcinoma deaths from 1969 to 1999 for all Appalachian counties and the remainder of the United States from the Multiple Cause of Death Public-Use File death certificate data provided by the National Center for Health Statistics (Centers for Disease Control and Prevention).11 These data include the decedents' underlying cause of death, age, race, gender, and county of residence. Colorectal carcinoma was identified as the underlying cause of death in accordance with procedures specified by the World Health Organization (WHO).12 Deaths from colorectal carcinoma were defined as those deaths assigned morphology codes C18, C19, C20, or C26 as specified in the WHO International Classification of Diseases for Oncology.13
We calculated death rates per 100,000 population using Surveillance, Epidemiology, and End Results (SEER)*Stat statistical software Version 4.214 and age adjusted the results to the 2000 U.S. standard population.15 Denominator data were provided by the Population Estimates Program of the Bureau of the Census with support from the National Cancer Institute through an interagency agreement. We calculated annual death rates during 1969–1999 by gender, race (white or black), and age (< 50, 50–59, 60–69, 70–79, or ≥ 80 years) and used joinpoint regression models to describe changes in trends over time (Joinpoint Version 2.5).16 Joinpoint regression analysis has been described elsewhere.2, 16, 17 Briefly, joinpoint is a statistical software program that analyzes trends. It chooses the best fitting points (joinpoints) at the calendar year in which the amount of increase or decrease in a trend changes significantly. For the current analysis, we allowed up to three joinpoints for each model and preserved the type I error rate through a Bonferroni correction. We described trends for each line segment that was separated by the joinpoints in the final models by the estimated annual percent change (EAPC) and 95% confidence intervals (CI), and tested the rate of change for each trend to determine whether it was significantly different from zero.
For 1995–1999, we also calculated death rates, age adjusted within broad range categories, by age groups (< 50, 50–59, 60–69, 70–79, and ≥ 80 years), gender, and race in Appalachia and in the rest of the United States as a whole. To compare the death rates of different populations, we calculated the difference in rates, the standard error, and the z statistic. To calculate the 95% CI for death rates, we used the gamma method.18 The current study was exempt from institutional review board review because mortality data were analyzed exclusively and human subjects were not included.
From 1969 to 1999, 155,005 people in Appalachia and 1,519,912 elsewhere in the United States died of colorectal carcinoma. Whites accounted for 93.4% (n = 144,707) of all colorectal carcinoma deaths within Appalachia and 89.1% (n = 1,353,509) of colorectal carcinoma deaths elsewhere in the United States. Blacks accounted for 6.5% (n = 10,150) of all colorectal carcinoma deaths within Appalachia and 9.8% (n = 149,469) of deaths elsewhere in the United States. Persons of other races accounted for only 0.1% of all deaths within Appalachia (n = 148) and 1% of deaths elsewhere in the United States (n = 16,934). Because of the few deaths among people of other races, we limited our analysis to whites and blacks.
In 1969, at the beginning of the study period, white males in Appalachia had a significantly lower colorectal carcinoma death rate than white males elsewhere in the United States (P = 0.0001; Table 1). However, since the mid-1980s, the trends in death rates in the 2 regions have been similar (Fig. 1), although in 1999, the rate in Appalachia was slightly, but significantly, higher than elsewhere (P = 0.002). Among white males, colorectal carcinoma death rates in Appalachia increased by an average of 0.4% per year from 1969 to 1986 but decreased by an average of 1.7% per year from 1986 to 1999 (Fig. 1). Elsewhere in the United States, rates for white males decreased by 0.1% per year from 1969 to 1983, decreased by 1.5% per year from 1983 to 1990, and decreased by 2.3% per year from 1990 to 1999.
|Race/gender/year||Appalachia||Elsewhere in the United States|
|Rate||95% CIb||Rate||95% CIb||P valuec|
|White males, 1969||31.0||29.5–32.6||34.1||33.6–34.7||0.0001|
|White males, 1999||26.6||25.5–27.6||24.8||24.5–25.1||0.002|
|Black males, 1969||24.5||20.0–30.1||27.3||25.8–28.9||0.27|
|Black males, 1999||35.2||30.2–40.8||33.9||32.7–35.2||0.65|
|White females, 1969||25.9||24.8–27.1||26.9||26.5–27.3||0.11|
|White females, 1999||18.6||17.9–19.3||17.3||17.1–17.6||0.001|
|Black females, 1969||21.5||17.6–26.1||25.7||24.4–27.0||0.06|
|Black females, 1999||22.7||19.6–26.2||25.5||24.6–26.4||0.10|
Colorectal carcinoma death rates among black males both in Appalachia and elsewhere in the United States were more variable from year to year than rates for whites (Fig. 2). In 1969, the colorectal carcinoma death rate among black males was lower than the rate among black males elsewhere in the United States (Table 1) but the difference was not statistically significant (P = 0.27). Thirty years later, the death rates among black males in Appalachia and elsewhere in the United States had increased, but again, the difference was not statistically significant (P = 0.65). In Appalachia, the death rates among black males increased from 1969 to 1991 at an average rate of 1.7% per year (Fig. 2). Thereafter, they declined at approximately 1.2% per year, but this trend was not statistically significant. Outside Appalachia, the rates among black males decreased until 1972, increased by an average rate of 4.6% per year from 1972 to 1975, increased by 1.2% per year from 1975 to 1990, and decreased by 0.9% per year from 1990 to 1999.
The rates among white females steadily declined both in Appalachia and elsewhere in the United States (Fig. 3). The rates decreased at 1.2% per year in Appalachia from 1969 to 1999 and at a slightly faster rate of 1.9% per year elsewhere in the United States from 1982 to 1999. In 1969, colorectal carcinoma death rates among white females in Appalachia (Table 1) and elsewhere in the United States were similar (P = 0.11). By 1999, the death rates among white females had decreased both in Appalachia and elsewhere in the United States, although the rate was slightly but significantly higher in Appalachia (P = 0.001). In 1969, the colorectal carcinoma death rate among black females in Appalachia was slightly, but not significantly, lower than the rate for black females elsewhere in the United States (P = 0.06), and the death rates for both groups had changed little by 1999 (P = 0.10; Table 1). The annual death rates among black females in Appalachia and elsewhere in the United States both increased from 1969 to 1985. However, from 1985 to 1999, the death rates among black females decreased by 1.6% per year in Appalachia and by 0.3% per year elsewhere in the United States (Fig. 4).
The results of our analysis of rate trends from 1969 to 1999 by gender and age among Appalachian residents and residents elsewhere in the United States, in which we used joinpoint regression to determine the EAPC, is summarized in Table 2. The death rates decreased significantly in recent years for all groups except females < age 50 years in Appalachia whose rates increased from 1992 to 1999 but this increase was not statistically significant.
|Region Gender/age (yrs)||Years||Trend 1||Trend 2||Trend 3|
|EAPC||95% CI||P valuec||Years||EAPC||95% CI||P valuec||Years||EAPC||95% CI||P valuec|
|Males, < 50||1969–1999||−0.7||(−1.0–−0.4)||< 0.001|
|Males, 60–69||1969–1985||0.3||(−0.2–0.9)||0.19||1985–1999||−1.3||(−1.9–−0.7)||< 0.001|
|Males, 70–79||1969–1987||0.2||(−0.2–0.6)||0.39||1987–1999||−1.9||(−2.6–−1.2)||< 0.001|
|Males, ≥ 80||1969–1986||1.4||(0.9–1.9)||< 0.001||1986–1999||−1.8||(−2.5–−1.1)||< 0.001|
|Females, < 50||1969–1992||−2.3||(−2.8–−1.8)||< 0.001||1992–1999||2.4||(−0.7–5.5)||0.13|
|Females, 50–59||1969–1999||−1.6||(−1.9–−1.3)||< 0.001|
|Females, 60–69||1969–1993||−1.2||(−1.4–−0.9)||< 0.001||1993–1999||−3.0||(−4.8–−1.1)||0.003|
|Females, 70–79||1969–1999||−1.4||(−1.6–−1.2)||< 0.001|
|Females, ≥ 80||1969–1981||0.6||(−0.04–1.3)||0.07||1981–1999||−1.0||(−1.4–−0.6)||< 0.001|
|Elsewhere in U.S.|
|Males, < 50||1969–1974||0.1||(−1.7–2.0)||0.90||1974–1999||−1.4||(−1.6–−1.3)||< 0.001|
|Males, 50–59||1969–1985||−0.4||(−0.7–−0.2)||0.002||1985–1999||−1.7||(−2.0–−1.4)||< 0.001|
|Males, 60–69||1969–1985||−0.3||(−0.5–−0.1)||0.001||1985–1999||−1.7||(−1.9–−1.5)||< 0.001|
|Males, 70–79||1969–1985||−0.2||(−0.3–−0.02)||0.03||1985–1999||−2.3||(−2.5–−2.2)||< 0.001|
|Males, ≥ 80||1969–1980||1.1||(0.8–1.3)||< 0.001||1980–1991||−0.8||(−1.1–−0.5)||< 0.001||1991–1999||−2.4||(−2.8, −2.0)||< 0.001|
|Females, < 50||1969–1989||−2.6||(−2.8–−2.3)||< 0.001||1989–1999||−0.7||(−1.3–−0.1)||0.04|
|Females, 50–59||1969–1987||−1.8||(−1.9–−1.6)||< 0.001||1987–1999||−2.4||(−2.7–−2.2)||< 0.001|
|Females, 60–69||1969–1983||−1.2||(−1.5–−0.9)||< 0.001||1983–1999||−2.1||(−2.4–−1.9)||< 0.001|
|Females, 70–79||1969–1983||−0.9||(−1.1–−0.7)||< 0.001||1983–1999||−2.1||(−2.3–−2.0)||< 0.001|
|Females, ≥ 80||1969–1984||−0.2||(−0.4–0.01)||0.07||1984–1999||−1.3||(−1.5–−1.1)||< 0.001|
Five-year, age-specific colorectal carcinoma death rates for Appalachia and for the rest of the United States combined (1995–1999), age adjusted within broad range categories, are shown in Table 3. Ninety-four percent of male decedents and 95% of female decedents were age ≥ 50 years when they died. In all demographic categories, colorectal carcinoma death rates increased with age. Among people in the oldest age group (age > 80 years), the death rates ranged from 204.6 per 100,000 for white females elsewhere in the United States to 318.7 per 100,000 for black males in Appalachia. In most age groups examined, the colorectal carcinoma death rates among whites in Appalachia were significantly higher than among whites elsewhere in the United States. The greatest differences between death rates in Appalachia and those elsewhere in the United States were among white males ages 70–79 years (rate difference, 4.9 per 100,000), white females ages 70–79 years (rate difference, 3.9 per 100,000), and white females age > 80 years (rate difference, 6.2 per 100,000).
|Gender/age (yrs)||Appalachia||Elsewhere in the United States|
|Count||Rate||95% CIb||Count||Rate||95% CIb||Differencec|
|Males, all ages||12,163||26.53||26.05–27.01||110,629||25.75||25.59–25.90||0.78c|
|Females, all ages||12,812||18.92||18.59–19.25||112,034||17.87||17.77–17.98||1.04c|
|Males, all ages||884||34.50||32.23–36.90||14,547||34.39||33.80–34.98||0.12|
|Females, all ages||946||22.62||21.20–24.13||16,321||25.52||25.13–25.92||−2.90c|
The death rate for colorectal carcinoma was significantly lower among black females in Appalachia than among black females elsewhere in the United States for all age groups combined (rate difference, 2.90 per 100,000), for those ages 60–69 years (rate difference, 17.9 per 100,000), and for those ages 70–79 years (rate difference, 15.0 per 100,000) (Table 3). The death rates for Appalachian black women aged ≥ 80 years was 13.3 per 100,000 lower than for black females elsewhere in the United States, but the difference was not statistically significant. The death rates among black males in Appalachia and elsewhere in the United States did not differ significantly for any age group.
For all groups studied, the death rates from colorectal carcinoma in Appalachia have declined in recent years. These analyses mirror a recent nationwide study of colorectal carcinoma mortality rates that reported a decline in death rates from 1990 to 2000 for white males and females and black males and females in the United States as a whole.19 However, in our study, the recent rate of decline in death rates in some Appalachian demographic groups has been slower than the rate of decline in corresponding groups elsewhere in the United States.
The higher death rates among Appalachian whites could be caused by several factors, including higher colorectal carcinoma incidence, less thorough screening, and poorer cancer treatment. The percentage of Appalachian men and women who reported having a fecal occult blood test in the past year or a flexible sigmoidoscopy in the previous 5 years has not been determined, although colorectal carcinoma screening rates are low nationally.3 Results of a nationwide study showed that people living in rural areas and below the poverty line, conditions prominent in Appalachia, were less likely to receive colorectal carcinoma screening by fecal occult blood test.20 The cost of screening and a lack of knowledge about the importance of routine screening have also been cited as barriers to the use of preventive services by Appalachian populations.21 Regular and more frequent encounters with health care professionals may increase the opportunities for colorectal carcinoma screening, and a physician's recommendation that a patient be screened has been shown to increase the chances that patient will do so.22
People in Appalachia may also face other barriers to receiving proper diagnosis and treatment of colorectal carcinoma. In a study of prostate, breast, cervical, and colon carcinomas among people in the Appalachian region of South Carolina, researchers found that colon carcinoma was more likely to be diagnosed beyond the local stage.23 Access to prevention services24 and physician knowledge about the proper treatment of colorectal carcinoma25 have been shown to be lower in rural areas, and Appalachia as noted previously is disproportionately rural. A lack of recommended treatment may contribute to higher death rates in some groups. In one study, researchers found that poorer, uninsured patients and those insured by Medicaid were less likely than those insured by private health insurance to undergo surgery for colorectal carcinoma.26 Another study showed that patients who were older, insured by Medicaid or Medicare, or living in a zip code with a low per capita income were less likely to receive the recommended adjuvant therapy after surgery for Stage III colon carcinoma.27
As others have found for the U.S. population as a whole, we found the colorectal carcinoma death rates to be significantly higher among blacks than among whites in Appalachia. An analysis of national mortality data from 1993 to 1997 reported that black males and black females had significantly higher 5-year, age-adjusted colorectal carcinoma death rates than their white counterparts.2 In our study, the same pattern is reported for 5-year, age-adjusted colorectal carcinoma death rates for Appalachia from 1995 to 1999. One explanation for this may be found in recent reports of nationwide incidence rates showing that blacks were less likely than whites to have their colon carcinoma diagnosed while still localized and more likely to have it diagnosed after having spread to all anatomic subsites of the colon.28 Blacks are thus more likely than whites to be more severely ill when hospitalized for colorectal carcinoma and to have comorbid conditions that would complicate treatment and reduce their chance of surviving.29 In a study of Medicare beneficiaries, researchers also found that black patients were less likely to undergo surgical resection for colorectal carcinoma,30 another factor that may contribute to the higher death rate among blacks.
Our study found that recent 5-year, age-adjusted death rates for black women in Appalachia are lower than those for black women living elsewhere in the United States for all ages combined, and for the 60–79 year age group. The reasons for this may be complex and are not easily answered by the available literature. According to a recent study, Appalachian women (all races combined) have slightly higher age-adjusted death rates from 1994 to 1998 than those in the entire United States.31 Further analysis of women residing in rural Appalachian counties shows colorectal carcinoma death rates to be equal to those in the entire United States and to be lower than death rates for the entire Appalachian region. A similar pattern was seen for males. The colorectal carcinoma death rates were higher in Appalachia than in the entire United States, but the rate was lower in the rural Appalachian counties than rates in either the United States or the entire Appalachian region. However, the authors did not analyze death rates by race group. Black females in our study may reside in rural areas of Appalachia and may not be exposed to the same factors or risk factors that contribute to colorectal carcinoma deaths as black females in nonrural Appalachian areas. Factors for black females in rural areas of Appalachia such as healthier eating habits, eating fruits, vegetables, and diets lower in fat may contribute to a reduced risk of dying of colorectal carcinoma. Patterns for colorectal carcinoma screening and polypectomy or differences in access to health care may also contribute to declining cancer deaths in some subgroups.32 However, a recent study of rural residents in the United States revealed that they were less likely to have routine colorectal carcinoma screening than those living in urban or suburban areas. These authors did not analyze data for the Appalachian region separately.33
There are several potential limitations to the current study. First, the mortality data were derived from death certificates, and the information on death certificates may be recorded inaccurately.34 However, studies of the accuracy of death certificates in the United States have shown them to be a valid method for determining colorectal carcinoma as the underlying cause of death,35–37 and another study of rural Kentucky death certificates validated their use in determining overall cancer mortality rates.6 Second, death certificate data do not include information about the decedents' stage of colorectal carcinoma at diagnosis, the histology, or the anatomic subsites to which the disease has spread. Only incidence studies can provide such information and may explain why slower rates of decline in death rates were seen in some Appalachian subgroups. Finally, we compared death rates in Appalachia with average death rates for the rest of the country, which is heterogeneous and includes areas with higher as well as lower rates. We demonstrate that there are race/gender/age subgroups within Appalachia that have higher death rates than the corresponding subgroups elsewhere in the United States. Further studies are needed to determine the cause for these higher rates, including studies of colorectal carcinoma screening rates and the stage at which the colorectal carcinoma is diagnosed and treated.
The authors thank Kevin Moran for his editorial suggestions regarding an earlier draft of the current article.
- 1U.S. Preventative Task Force on Colorectal Screening. Screening for colorectal cancer [monograph online]. Available from URL: http://www.ahcpr.gov/clinic/uspstf/uspscolo.htm [accessed June 6, 2003].
- 4Atlas of cancer mortality in the United States, 1950-1994. Bethesda: National Cancer Institute, 1999., , , et al.
- 5Health care services in Appalachia. In: CoutoRA, SimpsonNK, HarrisG, editors. Sowing seeds in the mountains: community-based coalitions for cancer prevention and control. Bethesda: Appalachia Leadership Initiative on Cancer, National Cancer Institute, 1994: 62–80..
- 9Appalachian Regional Commission. Poverty rates in Appalachia, 1990 [monograph online]. Available from URL: http://www.arc.gov/research/poverty/povmain.htm [accessed May 22, 2001].
- 10Appalachian Regional Commission. Counties in the Appalachian region [monograph online]. Available from URL: http://www.arc.gov/aboutarc/region/counties.htm [accessed May 22, 2001].
- 11National Center for Health Statistics. Vital statistics mortality data, multiple cause of death, 1969-1999 [machine-readable data tapes]. Hyattsville, MD: U.S. Department of Health and Human Services, 2002.
- 12World Health Organization. Manual of the international statistical classification of diseases, injuries, and causes of death, based on the recommendations of the ninth revision conference, 1975. Geneva: World Health Organization, 1977.
- 13PercyC, Van HoltenV, MuirC, editors. World Health Organization international classification of diseases for oncology. 2nd ed. Geneva: World Health Organization, 1990.
- 14National Cancer Institute Surveillance, Epidemiology, and End Results Program. SEER*Stat [computer software online]. Available from URL: http://seer.cancer.gov/seerstat [accessed January 2001].
- 15Age-adjusted death rates: trend data based on the year 2000 standard population. Natl Vital Stat Rep. 2001; 49: 1–6., .
- 16National Cancer Institute Statistical Research and Applications Branch. Joinpoint [computer software online]. Available from URL: http://srab.cancer.gov/joinpoint [accessed January 17, 2002].
- 19Cancer mortality surveillance—United States, 1990-2000. MMWR Surveill Summ. 2004; 53: 1–108., , , et al.
- 31Centers for Disease Control and Prevention. Cancer death rates—Appalachia, 1994-1998. MMWR Morb Mortal Wkly Rep. 2002; 51: 527–529.