Colorectal Cancer Test Use from the 2005 National Health Interview Survey

Screening has been shown to reduce colorectal cancer mortality (1-6). Based on this evidence, major national organizations, including the American Cancer Society (7), the U.S. Preventive Services Task Force (8), and a consortium of medical professional organizations (9), recommend regular colorectal cancer screening for persons ages ≥50 years. Recommended options for colorectal cancer screening in 2005 included an annual home fecal occult blood test (FOBT), colorectal endoscopy (either flexible sigmoidoscopy every 5 years or colonoscopy every 10 years), or double-contrast barium enema every 5 years. Despite these recommendations, prior surveys conducted from 2000 to 2004 indicated that many people were not getting tested for colorectal cancer (10-12).

Achieving high levels of colorectal cancer testing in all major subgroups of the population is important for reducing mortality from colorectal cancer. We therefore analyzed data from the 2005 National Health Interview Survey (NHIS) on use of home FOBT and colorectal endoscopy. We sought to estimate the proportion of the U.S. population that is having these tests within recommended time intervals and the extent to which this proportion varies by sociodemographic, health-care access, and other health-related factors. In addition, we examined reported reasons for not having a colorectal cancer test.

The NHIS is an in-person survey of the civilian, noninstitutionalized U.S. population that is conducted by the National Center for Health Statistics of the Centers for Disease Control and Prevention (13). A representative sample of households is selected using a multistage cluster sample design (13). U.S. Census Bureau interviewers visit each selected household to administer the survey. The Cancer Control Module and Sample Adult Core obtain additional information from one randomly selected adult in the family (14). The interviewed sample for the 2005 Cancer Control Module included 30,873 of 39,227 eligible adults for a conditional response rate of 79%. The conditional response rate is the rate only for those adults identified as eligible and does not take into account family nonresponse. The unconditional response rate for the Cancer Control Module was 68%, calculated by multiplying the conditional rate by the family response rate of 86% (14).

For the current analysis, only the 13,480 respondents ages ≥50 years were included, because it is recommended that colorectal cancer screening be initiated at age 50 years for average-risk persons who do not have a family history of the disease or certain other risk factors (7-9). Respondents with a personal history of colorectal cancer or missing information on history of colorectal cancer were excluded from all analyses (n = 211).

Respondents were asked if they had ever had a sigmoidoscopy, colonoscopy, or proctoscopy and when they had their most recent exam. They were also given a brief description of each type of endoscopy and asked whether their most recent exam was a sigmoidoscopy, colonoscopy, proctoscopy, or something else. The term “proctoscopy” was kept in the survey for consistency with earlier surveys. Proctoscopy is an older procedure that uses a rigid instrument shorter than the instrument used for flexible sigmoidoscopy. Respondents who had never had one of these exams or had not had an exam in the past 10 years were asked the most important reason they had not had the exam and whether in the past 12 months a doctor or other health professional had recommended that they have a sigmoidoscopy or colonoscopy. For our analysis, we classified any endoscopy (sigmoidoscopy, colonoscopy, or proctoscopy) within the past 10 years as having been done within the recommended time interval, because 10 years is the maximum interval recommended for any endoscopic procedure (7, 9). We were unable to analyze use of each type of endoscopic test separately for the primary analyses because the survey only asked about the type of endoscopy for the most recent procedure. In addition, respondents may not accurately report which type of endoscopy they had (15, 16).

For a small number of respondents, it was unclear if they had an endoscopy within the recommended time interval. These respondents reported that their most recent endoscopy was >5 years but <10 years ago but did not specifically report that this most recent endoscopy was a colonoscopy (reported that this most recent endoscopy was a sigmoidoscopy or proctoscopy or did not provide interpretable information on the type of endoscopy). Many of these respondents would have met testing recommendations because they had a FOBT within the past year, because their most recent endoscopy was in fact a colonoscopy, or because they had a colonoscopy before their most recent endoscopy but still within 10 years. For our primary analyses, we assumed that these respondents had been tested within the recommended time interval. However, we did a sensitivity analysis to examine the effect of this assumption. For this sensitivity analysis, we assumed that the 183 respondents who reported that their most recent endoscopy was a sigmoidoscopy or proctoscopy >5 years but <10 years ago (and did not report a FOBT within the past year) did not have colorectal cancer testing within the recommended time interval. The 46 respondents who did not provide interpretable information on the type of endoscopy (and did not report a FOBT within the past year) were excluded from this sensitivity analysis.

Respondents were also asked if they had ever had a blood stool test using a home test kit (home blood stool test or FOBT) and when they had their most recent home blood stool test. We only included information on use of home blood stool tests, not office tests, because national guidelines recommend use of home blood stool tests (7, 9). Respondents who had never had a home blood stool test or had not had one in the past year were asked the most important reason they had not had the exam and whether in the past 12 months a doctor or other health professional had recommended that they have a home blood stool test. For our analysis, we defined a FOBT within the past year as having been done within the recommended time interval as recommended by national guidelines (7, 9).

For respondents ages ≥50 years with no history of colorectal cancer, the percentage missing the information needed to calculate each colorectal cancer testing variable was 10% for FOBT within the past year and 9% for endoscopy within the past 10 years, resulting in a sample size of 11,918 for FOBT within the past year, 12,045 for endoscopy within the past 10 years, and 11,943 for either test within the recommended time interval. No questions about double-contrast barium enema were asked on the NHIS.

Colorectal cancer tests done for any indication were included in the analysis because the reported indication for a colorectal cancer test may not be accurate (15), and even if a test were conducted for nonscreening purposes, a person would have been considered effectively screened.

Variables examined in relation to colorectal cancer test use are listed in Table 1. All variables were self-reported. Respondents who reported they were of multiple races but also reported a primary race were included in the primary race category. The “other” race category includes respondents whose primary race was American Indian/Alaska Native, because the sample size was too small to report them separately. Respondents who reported multiple races but did not select a primary race were also included in the “other” category. Missing data for race and ethnicity were imputed using hot-deck imputation (13, 14). Missing data for family income were imputed using multiple imputation (13, 14).

Respondents were asked if there was a place that they usually go to when they are sick or need advice about their health. Respondents reporting the hospital emergency room as their only usual source of health care were included among those with no usual source of health care, whereas those respondents who reported other places were classified as having a usual source of health care.

Health-care coverage was categorized into eight groups: (a) private only; (b) Medicare only; (c) Medicare plus private coverage (including Medigap); (d) Medicare plus Medicaid; (e) Medicaid only; (f) military health care including Tricare, VA, or CHAMP-VA (with or without other health-care coverage); (g) other health-care coverage, and (h) no coverage or single service only. The “other” category for health-care coverage includes coverage provided by the Indian Health Service, a state-sponsored health plan, or some other government program. Respondents who had one of these types of health-care coverage in addition to Medicare, Medicaid, or private insurance were included in the “other” category as were the small number of respondents who had Medicaid in addition to private insurance.

The following formula was used to calculate metabolic equivalents from the number of minutes per week of moderate and vigorous exercise: 4.5 × number of minutes of moderate exercise per week + 7.0 × number of minutes of vigorous exercise per week (17). Physical activity was categorized as no activity (metabolic equivalents = 0), some activity (metabolic equivalents = 1 to <675), or meets/exceeds recommendations (metabolic equivalents ≥ 675) according to the Surgeon General’s recommendation of 30 min of moderate activity per day at least 5 days a week (18).

To provide national estimates of the prevalence of colorectal cancer test use, responses were weighted to reflect the probability of selection with adjustments for nonresponse and poststratification. Proportions with 95% confidence interval (95% CI) based on a logit transformation were calculated using SAS version 9.1.3 (19) and SUDAAN version 9.0 (20) to account for the complex survey design. Proportions in Table 1 were age standardized to the 2000 U.S. standard million population using the direct method (21).

Adjusted percentages (predictive margins) in Table 2 were computed from multiple logistic regression models controlling for all of the variables in Table 2 (22, 23). Predictive margins are a type of direct standardization that averages the predicted values from the logistic regression models over the covariate distribution in the study population (22, 23). The predictive margin for a specific group represents the average predicted response if everyone in the sample had been in that group. The P values in Table 2 are based on an overall Wald χ² test for association from multivariate logistic regression models. The family history of colorectal and noncolorectal cancer variables were excluded from the predictive margins model due to the large amount of missing data. The mammogram and Papanicolaou smear variables were also excluded because they pertain only to women. Number of physician visits was also not included in the model due to difficulties in interpretation of this variable because a physician visit is usually needed to obtain a FOBT kit or endoscopy.

In 2005, the age-standardized proportion of respondents who reported a colorectal cancer test within the recommended time interval (either FOBT within 1 year or endoscopy within 10 years) was 50.0% (95% CI, 48.8-51.2; Table 1). The proportion reporting an endoscopy within 10 years was 45.2% (95% CI, 44.0-46.4), whereas the proportion reporting a FOBT within 1 year was 12.0% (95% CI, 11.3-12.7; Table 1). Among respondents who reported an endoscopy within the past 10 years, most (83.8%) reported that their most recent exam was a colonoscopy (including colonoscopies that were done to follow-up other colorectal cancer screening tests), whereas 11.7% reported a sigmoidoscopy as their most recent test.

We conducted a sensitivity analysis (see Materials and Methods for details) where we assumed that the respondents who reported that their most recent endoscopy was a sigmoidoscopy or proctoscopy >5 years but <10 years ago (and did not report a FOBT within the past year) did not have colorectal cancer testing within the recommended time interval. Under these assumptions, the overall percentage tested was 48.4% (95% CI, 47.2-49.6), very similar to the 50.0% presented in Table 1.

For respondents who reported an endoscopy within the past 10 years, 72.8% (95% CI, 71.3-74.3) said that their most recent endoscopy was part of a routine exam, whereas 24.3% (95% CI, 22.9-25.8) said that they had the endoscopy because of a problem. Among respondents who reported a FOBT within the past year, 91.0% (95% CI, 89.2-92.6) said that their most recent FOBT was part of a routine exam, whereas 7.8% (95% CI, 6.3-9.5) said that they had the FOBT because of a problem.

We examined the age-standardized proportions who reported a colorectal cancer test within the recommended time interval among subgroups defined by sociodemographic, health-care access, and other health-related characteristics (Table 1). The proportion who reported colorectal cancer testing in the various subgroups ranged from a high of 68% among respondents with military health-care coverage or a family history of colorectal cancer to <25% among respondents who did not have health-care coverage, did not have a usual source of health care, or did not visit a physician in the past year.

The largest differences in the age-standardized proportions reporting colorectal cancer testing were between subgroups defined by socioeconomic status and health-care-related factors (Table 1). The proportion who had a colorectal cancer test was 37.0% for respondents with less than a high school education compared with 60.7% for college graduates. A similar difference was observed between respondents in low-income and high-income categories. For respondents who had not visited a physician in the last year, the proportion who reported colorectal cancer testing was only 19.5%. Women who reported not having a recent mammogram or Papanicolaou smear were also considerably less likely to report a colorectal cancer test compared with women who reported having had these tests recently. Respondents with a usual source of health care or with health-care coverage were much more likely to report a colorectal cancer test than those without health-care coverage or a usual source of health care.

Moderate differences in the age-standardized proportions reporting colorectal cancer test use were observed between subgroups defined by other sociodemographic and health-related characteristics (Table 1). The age-standardized proportion reporting colorectal cancer test use was lower among Blacks and Asians than among Whites, and was lower among Hispanics than among non-Hispanics. There were also moderate differences in the age-standardized proportions reporting colorectal cancer test use by age, marital status, physical activity, smoking status, alcohol use, and personal or family history of cancer. We did not observe meaningful differences in colorectal cancer test use by gender, general health status, or body mass index.

The factors associated with FOBT and with endoscopy were generally similar (Table 2). When the variables included in Table 2 were adjusted for each other, the associations with colorectal cancer test use were generally attenuated. The associations with education, income, and Hispanic ethnicity were weakened after adjustment for other factors. In addition, the difference in colorectal cancer test use between Blacks and Whites was eliminated after adjustment for other factors. However, even after adjustment for other factors, health-care coverage and having a usual source of health care remained strongly associated with colorectal cancer test use.

Among those respondents who had not had a FOBT or endoscopy within the recommended time interval, the most commonly reported reason for not having a colorectal cancer test was “never thought about it” (Table 3). Reasons for not having a FOBT and reasons for not having an endoscopy were similar, although the following reasons were slightly more likely to be reported for endoscopy: “have not had any problems,” “put it off,” “too expensive/no insurance,” and “too painful/unpleasant/embarrassing.” The following reasons were slightly more likely to be reported for FOBT: “never thought about it,” “doctor did not order it,” and “did not need it.” The distribution of reasons was similar for those who never had either test compared with those who had a test but did not have it recently (Table 3).

In the NHIS, the proportion of people ages ≥50 years who had colorectal cancer testing within recommended time intervals has increased modestly since 2000 (50.0% in 2005 versus 42.5% in 2000; ref. 10). This increase appears to have been due entirely to an increase in endoscopy (45.2% in 2005 versus 33.9% in 2000), because FOBT use has declined (12.0% in 2005 versus 17.1% in 2000; ref. 10). The increase in endoscopy from 2000 to 2005 occurred both in people ages 50 to 64 years and in those ages ≥65 years (10). Several factors may have contributed to the increase in endoscopy use, including the addition of Medicare coverage in 2001 for screening colonoscopy in people not at high risk, media coverage of colonoscopy, and changes in physician and patient preferences (24-26). Despite the modest increase between 2000 and 2005, the proportion with recommended colorectal cancer testing in 2005 was still lower than for mammography (27).

The proportion of respondents ages ≥50 years who reported colorectal cancer testing within recommended time intervals was slightly higher in the 2004 Behavioral Risk Factor Surveillance System survey than in the 2005 NHIS (11). The differences in proportions who had colorectal cancer testing between the two surveys may have been due to differences in mode of administration and response rates, because the Behavioral Risk Factor Surveillance System is administered by telephone and has a lower response rate. Behavioral Risk Factor Surveillance System rates were also higher than NHIS rates in earlier years both for colorectal cancer testing and for mammography (10, 28, 29).

Rates of colorectal cancer testing in the 2005 NHIS were not high in any population subgroup examined. Even among college graduates, only 61% had colorectal cancer testing within recommended time intervals. As in 2000 (10), the proportion who had a colorectal cancer test was very low among people without health-care coverage and among those without a usual source of health care.

The factors that we found to be associated with colorectal cancer testing were similar to the factors associated with colorectal cancer testing in earlier NHIS and Behavioral Risk Factor Surveillance System surveys (7, 10, 12, 30-35). Sociodemographic factors such as education, age, and income and factors related to healthcare access such as health-care coverage and having a usual source of health care were both important correlates of colorectal cancer test use. After adjustment for multiple factors, health-care coverage and having a usual source of health care were still strong correlates of colorectal cancer test use. However, the association between colorectal cancer test use and certain sociodemographic factors, including education, income, and race, was considerably attenuated by adjustment. In particular, the difference in colorectal cancer test use between Blacks and Whites was eliminated by adjustment for health-care coverage and other factors. These results suggest that much of the difference in colorectal cancer test use among sociodemographic groups may be attributable to differences in health-care access factors.

Factors that involve contact with a physician, including the number of physician visits in the last year and mammography or Papanicolaou smear use, were also associated with colorectal cancer test use. Reasons for these associations may include the fact that a physician visit may be an opportunity for the physician to discuss or offer colorectal cancer screening but could also be because a physician visit is usually needed to obtain a FOBT test kit or an endoscopy.

The most commonly reported reason in the 2005 NHIS for not having a colorectal cancer test was “never thought about it,” which was reported by about half of the respondents who did not have a colorectal cancer test within the recommended time interval. The second most common reason was “doctor did not order it,” which was reported by about one-fifth of respondents. These results are very similar to the results of the 2000 NHIS (10) and support the continued critical need for making the general public more aware of the importance of colorectal cancer screening as well as the need for physicians to advise their patients to get screened.

There are several limitations to this analysis. Some respondents may have incorrectly reported their use of colorectal cancer tests. However, studies comparing self-report of colorectal cancer test use to information from medical records have generally found moderate to good agreement between the two data sources, with better agreement when sigmoidoscopy and colonoscopy are grouped together (15, 16, 36-39). Because the NHIS only asked about the type of endoscopy for the most recent endoscopic procedure and because respondents may not accurately report the type of endoscopy (15, 16), we were unable to analyze use of sigmoidoscopy separately from colonoscopy for the primary analyses. In addition, we did not have information on double-contrast barium enema use because the NHIS did not include questions about this procedure. However, data from the Medicare program indicate that double-contrast barium enema is infrequently used. A study of 153,469 Medicare beneficiaries found that only 0.1% had a barium enema between 1998 and 2004 (40). The volume and percentage of barium enemas provided to Tricare and Medicare beneficiaries decreased from 1999 to 2001, with an increase in colonoscopies (41). In a 1999 to 2000 survey, primary care physicians infrequently reported that they recommended double-contrast barium enema to average-risk patients ages ≥50 years for colorectal cancer screening (42).

In conclusion, colorectal cancer testing results from the 2005 NHIS show that although some progress has been made in the last few years, about half of Americans ≥50 years are still not having appropriate colorectal cancer testing. Results from the 2005 NHIS also show important differences in colorectal cancer testing between subgroups of the population, with particularly low colorectal cancer testing rates among people with no health-care coverage or usual source of health care. These results indicate that increased efforts to expand health-care coverage and/or provide colorectal cancer testing to people without health-care coverage are needed to increase colorectal cancer screening and thereby reduce colorectal cancer incidence and mortality.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.