Descriptive epidemiology of colorectal cancer in the United States, 1998–2001†‡
We appreciate the in-kind support from all the contributors to this monograph and also are grateful for the contributions of Jessica King for the preparation of analytic files and to Faruque Ahmed for his leadership of the colorectal cancer monograph project. The following registries contributed data to the production of this supplement: AL, AK, AZ, CA, CO, CT, DC, FL, Metro Atlanta (Georgia), HI, ID, IL, IN, IA, KS, KY, LA, ME, MA, MI, MN, MO, MT, NE, NJ, NM, NY, NC, OH, OK, OR, PA, RI, SC, TX, UT, VT, WA, WV, WI, and WY.
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.
This article presents an overview of the descriptive epidemiology of colorectal cancer (CRC) in the United States and provides detailed methodologic information on the cancer incidence data from population-based cancer registries that were used by most of the articles in this supplement. Information in this article and the other articles in the supplement will be important for guiding cancer control and prevention efforts related to reducing the burden of CRC.
Colorectal cancer (CRC) is the third most common cancer and the third leading cause of cancer deaths among men and women in the United States.1 It is the second most common cancer among black, Asian/Pacific Islander (A/PI), and Hispanic women and the second leading cause of cancer deaths among American Indian/Alaska Native (AI/AN) men.1 For 2005, it was estimated that about 145,290 people in the United States would be diagnosed with CRC and approximately 56,290 people would die of the disease.2 The vast majority of colorectal cancers and deaths can be prevented by applying existing knowledge about cancer prevention—largely related to modifying behavioral risk factors such as physical inactivity, being overweight or obese, smoking, alcohol consumption, a diet high in red meat, and inadequate intake of fruits and vegetables—and by using recommended screening tests.3, 4
Recognizing the need to better inform the public, the media, health professionals, and the comprehensive cancer control community about the burden of CRC, the Centers for Disease Control and Prevention's (CDC's) National Program of Cancer Registries (NPCR) led the effort to produce this first-ever supplement describing the epidemiology of CRC in the United States. Using data mainly from the CDC's NPCR and the National Cancer Institute's (NCI's) Surveillance, Epidemiology, and End Results (SEER) programs,5, 6 investigators from the CDC, NCI, population-based cancer registries, academic institutions, ORC Macro, and the American Cancer Society collaborated to produce a series of articles highlighting the magnitude of the burden of CRC in the United States.
Most of the articles presented in this supplement used high-quality cancer incidence data from central registries covering 39 states, the District of Columbia, and metropolitan Atlanta for the period 1998 through 2001. All central registries are members of the North American Association of Central Cancer Registries (NAACCR).7 Cancer incidence data are collected by hospitals and other facilities that diagnose or treat cancer and are reported to central cancer registries. Although the vast majority of cancer cases are still reported by hospitals, increasingly data are obtained from nonhospital sources such as pathology laboratories, radiation facilities, freestanding surgical centers, long-term care facilities, and physicians' offices. Central cancer registries use death certificate data to update vital status of cases already in the registry and as a case-finding tool in identifying cases that should have been reported to the registry but were not.
Registries included in this report cover approximately 88% of the U.S. population. This cancer incidence dataset contains 542,149 cases of invasive CRC (Table 1), making it the largest CRC incidence file to be analyzed to date. Information from this supplement will be important for guiding cancer control and prevention efforts related to reducing the burden of CRC.
Table 1. Incidence Counts of Invasive Colorectal Cancer by Registry, United States, 1998–2001*
|District of Columbia||1375|
|Metro Atlanta (Georgia)||3912|
Epidemiology of Colorectal Cancer
CRC is one of the most common cancers and one of the biggest killers worldwide, with economically developed countries having the highest incidence.8 CRC is cancer that develops in the colon or rectum (Table 2). Over two thirds of CRCs in the United States are found in the colon (Table 3).2, 9
Table 2. SEER Site Recode for Colorectal Cancer
| Cecum||C18.0||Excluding lymphomas (histology codes 9590-9989)|
| Ascending colon||C18.2|
| Hepatic flexure of colon||C18.3|
| Transverse colon||C18.4|
| Splenic flexure of colon||C18.5|
| Descending colon||C18.6|
| Sigmoid colon||C18.7|
| Large intestine, NOS||C18.8-C18.9, C26.0|
| Rectosigmoid junction||C19.9||Excluding lymphomas (histology codes 9590-9989)|
| Rectum, NOS||C20.9|
Table 3. Invasive Colorectal Cancer by Selected Features, United States, 1998–2001*,†
|Overall||66.1 (100)||47.8 (100)||55.6 (100)|
| White||65.7 (88.1)||47.0 (87.1)||55.1 (87.6)|
| Black||70.1 (8.3)||53.7 (9.5)||60.2 (8.9)|
| Asian/Pacific Islander||48.1 (2.2)||35.3 (2.1)||40.9 (2.2)|
| Other||– (0.6)||– (0.5)||– (0.5)|
| Unknown||– (0.8)||– (0.8)||– (0.8)|
| Non-Hispanic||66.9 (94.8)||48.5 (95.3)||56.4 (95.0)|
| Hispanic||53.9 (5.2)||36.5 (4.7)||43.8 (4.9)|
| Unknown||– (0.0)||– (0.0)||– (0.0)|
| Proximal colon||25.4 (37.5)||21.7 (46.0)||23.2 (41.8)|
| Distal colon||17.3 (26.5)||11.5 (23.7)||14.0 (25.1)|
| Colon, not otherwise specified||4.0 (5.8)||3.0 (6.4)||3.4 (6.1)|
| Rectum||19.5 (30.2)||11.6 (23.9)||15.1 (27.1)|
| Localized||23.6 (35.4)||16.4 (33.9)||19.5 (34.7)|
| Regional||25.4 (38.3)||19.1 (39.5)||21.8 (38.9)|
| Distant||10.9 (16.7)||7.9 (16.1)||9.2 (16.4)|
| Unstaged||6.7 (9.6)||4.8 (10.4)||5.6 (10.0)|
| Yes||63.8 (96.9)||46.0 (95.9)||53.6 (96.4)|
| No||2.3 (3.1)||1.8 (4.1)||2.0 (3.6)|
More than 95% of CRCs are adenocarcinomas.2, 10 Most sporadic CRCs develop from benign adenomas or polyps.11–13 A lengthy time, estimated at 5–10 years, is required for the development of malignancy from adenoma.11 The risk of developing a colorectal adenoma is nearly 19% in the general population and 2–5% of sporadic polyps will develop into an invasive cancer.14 Lifetime risk of being diagnosed with CRC is about 5.9% for men and 5.5% for women.15 About 75% of CRC cases occur in persons without a family history of CRC or a predisposing illness.16, 17 Approximately 15% of CRCs are familial; the most common inherited conditions are familial adenomatous polyposis and hereditary nonpolyposis CRC and occur at an earlier age.11, 14
CRC incidence is >50 times higher in persons aged 60–79 years than in those <40 years.2 However, CRC appears to be increasing among younger persons.18, 19 CRC is one of the 10 most commonly diagnosed cancers among men and women aged 20–49 years and ranks among the top four cancers in men aged 40–49 years regardless of race.20 Disparities in the incidence of CRC by race/ethnicity, sex, socioeconomic status, rural/urban residence, and geographic region are described in the articles by Matanoski et al., Wu et al., Fairley et al., Cress et al., Lai et al., and Coughlin et al.20–25
Male and female CRC incidence rates declined from the mid-1980s to the mid-1990s, followed by a short period of stabilization, and from 1998–2002 rates declined 2.5% per year for males and 1.5% per year for females.26 Before 1987, incidence rates for white males were higher than for black males and approximately equal for black and white females. Since that time, incidence rates have been higher for black males and females than for whites.2, 26, 27 Rates for A/PIs, AI/ANs and Hispanic males and females remain lower than for whites. The only group to show significant declines during 1992–2002 is whites.26
Between 1992 and 2002, CRC death rates decreased significantly for white, black, and A/PI males and females but not for AI/ANs or Hispanics.26 The decline has been smaller among blacks than whites, resulting in a growing disparity between blacks and whites.15, 28
Since the 1960s, survival rates for CRC have been increasing.9 The 5-year relative survival rate for CRC increased more than 10% for both males and females, from 52% to 63% in females and from 50% to 64% in males.2, 29 The survival rate varies by stage at diagnosis with a survival of 91% for localized stage and 9% for distant stage.29 Among blacks, the 5-year relative survival rate improved from 45% in 1974–1976 to 55% for the period 1995–2000. This improvement was smaller than among whites for the same period (from 50% to 64%).15 The disparity in survival can partially be attributed to later stage at diagnosis for blacks (35% localized) than for whites (39% localized).15, 30, 31 From 1992 to 2000, the percentage of CRCs diagnosed at a localized stage for Hispanics and AI/ANs (35%) was similar to blacks while A/PIs (39%) showed a similar pattern to whites with a slightly smaller percentage of CRCs diagnosed at a distant stage (17% vs. 19%).2
Colorectal Cancer Screening
Screening involves the detection of precancerous polyps and cancers in asymptomatic individuals.32–34 Current guidelines recommend screening beginning at age of 50 for average-risk persons and at an earlier age for persons at higher risk (e.g., family history of CRC).4 At least 60% of colorectal cancer deaths could be prevented through the use of screening.4, 35, 36
Although the percentage of adults aged 50 or older who have received a fecal occult blood test (FOBT) or endoscopic screening has increased, in 2000 only 43% of the U.S. population underwent such tests within the recommended time intervals.37 The low rate of participation in CRC screening, especially in comparison with breast and cervical cancer screening, may be due to a number of factors, including patient discomfort, cost, lack of awareness, and poor acceptability of current screening methods.38, 39
MATERIALS AND METHODS
We used data on newly diagnosed primary cancers from population-based cancer registries that participate in CDC's National Program of Cancer Registries (NPCR) and/or NCI's Surveillance, Epidemiology, and End Results (SEER) Program.5, 6 Both NPCR and SEER data are collected and reported using standard data items and uniform codes and procedures as documented by the NAACCR.40 Reportable cancers include in situ or invasive primary cancers of all sites except in situ cancer of the cervix uteri.41 Basal and squamous cell carcinomas of the skin are also excluded, with the exception of those on the skin of the genital organs. Tumor and demographic information for cancer cases are obtained from medical records, with a small percentage being obtained from death certificates only.
In addition to cancer incidence data, certain analyses in this supplement used cancer mortality and behavioral data. Mortality data from death certificates are reported to state vital statistics offices and are consolidated into a national database by CDC through the National Vital Statistics System, which covers 100% of the U.S. population deaths.5 Cause of death is coded using the International Classification of Diseases (ICD) in use at the time of death.42, 43 Effective with deaths occurring in 1999, the United States began using the Tenth Revision of this classification (ICD-10).5, 43 The CDC's Behavioral Risk Factor Surveillance System (BRFSS) is a representative, state-level telephone survey of the adult U.S. population. Conducted monthly throughout the year, it includes information on cancer risk behaviors and screening.44
The population denominator data for calculating cancer incidence and mortality rates were obtained from the 2000 U.S. Census and slightly modified by SEER for improving the accuracy of rates for the Hawaiian population.5 We linked the cancer incidence and mortality data to the U.S. Census 2000 data to obtain county-level socioeconomic information.45 We also linked to the U.S. Department of Agriculture's 2003 Rural-Urban Continuum Codes to obtain county-level rural/urban information.46 The linked socioeconomic and rural-urban information were not available for four states.
The study was approved by the CDC's Institutional Review Board (IRB). Non-CDC collaborators on research papers obtained local IRB approval as needed. To ensure confidentiality, a Data Use Agreement was signed by researchers.
NPCR data for this report were reported to CDC as of January 31, 2004. SEER data were reported to NCI as of November 1, 2003, and made available through the SEER Program public use data file released in April 2004.47 Data from states that are supported by both NPCR and SEER are presented as reported to CDC as of January 31, 2004, unless otherwise stated.5 Cancer incidence data were analyzed for cancer registries that met the following high-quality data criteria for all cancer sites combined for diagnosis years 1998 though 2001: case ascertainment was at least 90% complete as measured by methods developed by NAACCR; ≥97% of cases passed a standard set of computerized edits; ≤5% of cases were reported by death certificate only; ≤5% of cases were missing information on race; ≤3% of cases were missing information on sex; and ≤3% of cases were missing information on age.5
NPCR and SEER data from 39 states, metropolitan Atlanta, and the District of Columbia met the data quality criteria for inclusion in this report (Table 1). These registries cover approximately 88% of the U.S. population, 89% of whites, 81% of blacks, and 93% of A/PIs. For analyses on trends and multiple primary cancers, which require data over a longer period of time, SEER data were used. SEER data were available from nine registries (SEER 9) from 1975 onwards and for 12 registries (SEER 12) for 1992–200147; these SEER datasets cover 10% and 14% of the U.S. population, respectively.
Cancer incidence variables used for this report include: age at diagnosis, sex, race, ethnicity, site recode, behavior, histology, grade, SEER summary stage, microscopic confirmation, sequence number, type of reporting source, year of diagnosis, residence (state and county), and county poverty level, education level, and rural–urban continuum code. The SEER site recode for CRC cases comprises International Classification of Diseases for Oncology, Third Edition (ICD-O-3)48 site codes C18.0-C18.9, C19.9, C20.9, C26.0; lymphomas originating in the lymphatic tissue of the colon and rectum are excluded (Table 2).49 Histology, behavior (in situ/invasive), and grade are coded using the ICD-O-3 system.48 Cases diagnosed before 2001 were coded according to the ICD-O, Second Edition (ICD-O-2)50 and then converted to ICD-O-3 codes using an algorithm.51 CRC cases diagnosed in 1998–2000 were staged using the 1977 SEER summary staging system52 whereas cases diagnosed in 2001 were staged using the 2000 SEER summary staging system.53 The differences between the 1977 and 2000 SEER staging systems are negligible for colorectal cancer.25 The summary stage categories for invasive colorectal cancer are localized (confined to colon or rectum), regional (direct extension of the cancer to adjacent organs/tissues or to regional lymph nodes), distant (metastasis to other areas of the body), and unstaged.
For identifying colorectal cancer deaths, the cause of death recode developed by the SEER Program for maximizing consistency with the cancer incidence data was used.54 Mortality variables include age at death, sex, race, ethnicity, cause of death recode, year of death, residence (state and county), and county poverty level, education level, and county rural-urban continuum code.
Race groups include white, black, A/PI, other, and unknown. Ethnicity was categorized as Hispanic, non-Hispanic, or unknown. Race and ethnic groups were kept separate unless otherwise stated. The Hispanic ethnicity data for NPCR registries are classified according to the NAACCR Hispanic Identification Algorithm (NHIA) that augments the NAACCR Spanish/Hispanic Origin data element by using birthplace, race, and/or surname associations with Hispanic ethnicity. The SEER registries identify Hispanic ethnicity according to the NAACCR Spanish/Hispanic Origin data element only.5 For mortality data, Hispanic ethnicity is reported as recorded on the death certificate.
Incidence and death rates are per 100,000 persons; incidence rates are based on invasive cases only, unless otherwise specified. The average annual incidence rate denotes the sum of the number of cases reported between 1998 and 2001 divided by the sum of the annual denominators. Rates were age-adjusted by 19 age groups (<1 year, 1–4 years, 5–9 years, 10–14 years, 15–19 years, … ≥85 years) to the 2000 U.S. standard population by the direct method.5 Ninety-five percent confidence intervals for the age-standardized rates were computed using the gamma method.55 Age-specific rates were not age-adjusted. Cancer cases with unknown sex or age were excluded from all analyses. SEER*Stat software was used for the analysis unless otherwise stated.56 Cell suppression was applied for <16 cases.
Descriptive Information on Colorectal Cancer Cases
The 1998–2001 study data file contains 542,149 cases of invasive primary colorectal cancer, of which 49.6% were among females (Table 3). Of the females, 87.1% were white, 9.5% were black, and 2.1% were A/PI. Of the males, 88.1% were white, 8.3% were black, and 2.2% were A/PI. Race information was available for 99.2% of the cases. Hispanic or non-Hispanic identification was known for >99.9% of cases (Table 3). Subsite information was available for 93.9% of cases. Information on stage at diagnosis was included for 94.4% of cases, excluding the data for California outside of Los Angeles, San Francisco-Oakland, and San Jose-Monterey for which stage was not available in the study data file. Diagnoses for 96.4% of the cases were microscopically confirmed.
The median age for invasive colorectal cancer cases was 74.2 years for females and 70.7 years for males (Table 4). The vast majority of invasive CRC cases were staged, with unstaged cases ranging from a low of 6.7% for A/PI males to a high of 11.6% for black females (Table 5). About 97% of the invasive colorectal cancers were reported from hospital sources, and 1.5% were identified solely from autopsy reports or death certificates (data not shown). Because cancer registry records represent tumors as opposed to persons, the 542,149 cases may include multiple primary cancers of the colon and rectum that were reported during the study period. The 542,149 cases of CRC occurred among 523,450 persons.
Table 4. Invasive Colorectal Cancer by Age and Race/Ethnicity, United States, 1998–2001*,†
|0–19||0.1 (0.0)||– (0.1)||– (0.1)||– (0.1)||0.1 (0.0)|
|20–24||0.8 (0.1)||0.6 (0.1)||0.9 (0.3)||0.6 (0.3)||0.8 (0.1)|
|25–29||1.9 (0.2)||2.3 (0.4)||2.0 (0.6)||1.4 (0.7)||2.0 (0.3)|
|30–34||3.9 (0.5)||4.4 (0.8)||3.4 (1.1)||3.2 (1.4)||4.0 (0.5)|
|35–39||7.6 (1.1)||9.8 (1.9)||7.0 (2.1)||6.4 (2.5)||8.0 (1.2)|
|40–44||14.5 (2.0)||19.2 (3.6)||13.0 (3.5)||12.2 (3.8)||15.2 (2.2)|
|45–49||29.5 (3.6)||40.6 (6.2)||24.9 (5.8)||23.6 (5.6)||30.7 (3.9)|
|50–54||56.9 (6.0)||73.4 (8.8)||44.1 (8.4)||49.2 (8.7)||58.7 (6.4)|
|55–59||98.1 (8.0)||128.5 (11.0)||74.4 (10.0)||80.4 (9.8)||100.9 (8.3)|
|60–64||162.4 (10.5)||194.5 (13.2)||107.9 (11.4)||129.0 (12.0)||164.9 (10.8)|
|65–69||246.9 (14.0)||254.2 (14.1)||168.5 (13.4)||207.5 (15.0)||247.0 (14.0)|
|70–74||328.7 (16.9)||334.1 (14.1)||236.5 (14.7)||268.0 (14.8)||328.4 (16.5)|
|75–79||404.5 (16.4)||410.6 (12.4)||300.0 (13.5)||340.0 (12.4)||404.7 (15.9)|
|80–84||475.1 (11.9)||464.7 (7.7)||340.4 (8.4)||388.1 (7.2)||474.0 (11.4)|
|≥85||538.2 (9.0)||465.1 (5.6)||400.2 (6.7)||422.3 (5.7)||532.9 (8.6)|
|Median age (y)||71.2||66.4||67.5||66.7||70.7|
|0–19||0.1 (0.0)||− (0.0)||−(0.1)||0.1 (0.2)||0.1 (0.0)|
|20–24||0.7 (0.1)||0.8 (0.1)||−(0.2)||0.7 (0.3)||0.7 (0.1)|
|25–29||1.7 (0.2)||2.0 (0.4)||1.2 (0.5)||1.3 (0.6)||1.8 (0.2)|
|30–34||3.6 (0.4)||4.2 (0.8)||3.2 (1.2)||2.7 (1.2)||3.7 (0.5)|
|35–39||6.7 (0.9)||9.1 (1.8)||5.9 (2.0)||6.3 (2.6)||7.1 (1.1)|
|40–44||13.3 (1.9)||18.0 (3.3)||11.2 (3.6)||12.5 (4.3)||14.0 (2.1)|
|45–49||24.7 (3.1)||34.9 (5.5)||22.4 (6.4)||21.6 (5.9)||25.9 (3.4)|
|50–54||42.5 (4.7)||63.9 (8.0)||40.4 (9.5)||36.8 (7.8)||45.1 (5.2)|
|55–59||66.6 (5.9)||92.7 (8.7)||54.2 (8.8)||58.2 (9.0)||69.5 (6.2)|
|60–64||107.2 (7.8)||138.1 (10.6)||73.3 (9.5)||80.3 (9.9)||109.9 (8.1)|
|65–69||163.2 (10.9)||183.3 (12.2)||118.9 (13.1)||126.0 (12.9)||164.8 (11.1)|
|70–74||220.0 (14.5)||243.3 (13.7)||160.7 (14.5)||167.5 (14.0)||221.9 (14.4)|
|75–79||287.2 (16.9)||304.6 (13.5)||201.7 (12.8)||209.9 (12.4)||288.5 (16.5)|
|80–84||360.3 (15.5)||362.9 (10.6)||261.6 (9.7)||266.9 (9.4)||360.7 (14.8)|
|≥85||411.7 (17.1)||384.5 (10.9)||297.4 (8.2)||304.1 (9.6)||410.2 (16.3)|
|Median age (y)||74.8||69.5||68.2||68.2||74.2|
Table 5. Invasive Colorectal Cancer by SEER Summary Stage* and Race/Ethnicity, United States, 1998–2001†,‡
| Localized||23.7 (35.9)||21.2 (30.0)||17.7 (35.7)||17.5 (31.3)|
| Regional||25.4 (38.5)||25.9 (37.3)||19.7 (40.8)||21.5 (40.0)|
| Distant||10.6 (16.3)||15.0 (21.9)||8.1 (16.9)||9.4 (17.8)|
| Unstaged||6.5 (9.4)||8.3 (10.9)||3.7 (6.7)||6.5 (10.9)|
| Localized||16.3 (34.4)||16.2 (30.2)||12.0 (34.0)||12.0 (32.0)|
| Regional||18.9 (39.7)||20.5 (38.2)||15.1 (42.1)||14.4 (39.1)|
| Distant||7.6 (15.7)||10.7 (20.1)||5.8 (16.7)||6.4 (18.0)|
| Unstaged||4.6 (10.2)||6.3 (11.6)||2.9 (7.2)||4.3 (11.0)|
Limitations of Data
The data have some potential limitations. First, data for 2001 used ICD-O-3 coding, whereas 1998-2000 data used ICD-O-2 coding. A comparison of ICD-O-3 and ICD-O-2 showed that the malignant-to-borderline change would result in the exclusion of only 0.006% of CRC cases. For terms that changed morphology codes, 0.02% of CRC cases would be affected. Therefore, the change in ICD-O had a negligible effect. Second, in situ CRC cases are not always captured or reported. However, most analyses are based on invasive cases only. Third, although the NAACCR Race and Ethnicity Identifier Assessment affirmed the importance of publishing cancer rates by race and ethnicity,57 inconsistencies in the collection and coding of data on race and Hispanic origin and their impact on incidence and mortality statistics have been reported.57, 58 The rates for populations other than white and black may be affected by problems in ascertaining race/ethnicity information from basic records (e.g., medical records, death certificates, and census reports). Finally, the 1998–2001 dataset pertains to 88% of the U.S. population covered by the eligible cancer registries. The excluded population may have different CRC rates.
Overview of Chapters
This supplement includes 10 articles that focus on major issues in colorectal cancer. Six articles are based on the 1998–2001 dataset that includes both NPCR and SEER data. Matanoski et al. assess the characteristics of colon and rectal cancers.24 Wu et al. examine association of subsite-specific CRC incidence rate and stage at diagnosis with county-poverty level.25 Stewart et al., who included 522,630 microscopically confirmed cases of CRC, present the largest U.S. population-based study to date to characterize CRC histology.10 Fairley et al. describe the burden of CRC and tumor characteristics in younger adults.20 Lai et al. assess variation in CRC incidence in 9 geographic divisions of the United States based on race, sex, and stage at diagnosis.23 Coughlin et al. examine CRC incidence among adults by residence in rural, suburban, or metropolitan areas.21
Cress et al. use 1992–2001 SEER data to examine temporal changes in CRC incidence,22 whereas Ahmed et al. use 1975–2001 SEER data to examine the excess risk of noncolonic primary cancers among CRC survivors.59 Parikh-Patel et al. utilize data from the California Cancer Registry to examine stage at diagnosis with respect to urban/rural residence.60 Finally, Coughlin et al. describe eight intervention projects funded by CDC's Division of Cancer Prevention and Control intended to promote CRC screening among diverse populations.61