Ananthakrishnan AN, Schellhase KG, Sparapani RA, Laud PW, Neuner JM. Disparities in Colon Cancer Screening in the Medicare Population. Arch Intern Med. 2007;167(3):258-264. doi:10.1001/archinte.167.3.258
Copyright 2007 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2007
Colorectal cancer is the third most common cancer in the United States, but the rate of screening remains low. Since 2001, Medicare has provided coverage of colonoscopy for colorectal cancer screening in individuals at average risk, but little is known about the effect of this coverage on screening or disparities in screening practices.
We examined the Medicare physician/supplier billing claims file for New York, Florida, and Illinois for the years 2002 and 2003. Using a previously employed algorithm, we identified the rates of colorectal screening tests in individuals at average risk. We performed multivariate logistic regression analysis to calculate the effects of sex, racial/ethnic, and socioeconomic characteristics on screening. We also looked for interactions between socioeconomic and demographic variables.
A total of 596 470 Medicare beneficiaries were included in the study. Approximately 18.3% of the population had undergone a screening colon test during the study period. Nonwhite persons were less likely to be screened for colorectal cancer than were white persons (relative risk [RR], 0.52; 95% confidence interval [CI], 0.50-0.53). The lowest RR of screening colonoscopy in women compared with men was in the oldest age group and the highest income tertile (RR for whites, 0.64; 95% CI, 0.59-0.70). Higher income level was associated with screening colonoscopy in white patients (men: RR, 1.19; 95% CI, 1.14-1.25; women: RR, 1.09; 95% CI, 1.05-1.15) but not in nonwhite patients (men: RR, 0.97; 95% CI, 0.78-1.22; women: RR, 0.94; 95% CI, 0.78-1.14).
Despite the expansion of Medicare coverage for colorectal cancer screening, there still remain significant disparities between sex and racial/ethnic groups in screening practices.
Colorectal cancer is the third most common cancer in men and women in the United States, with an estimated 145 290 new cases in 2005, and the second leading cause of cancer deaths.1 There is evidence that screening for colorectal cancer decreases incidence and mortality from the disease. The US Preventive Services Task Force recommends colon cancer screening for all persons at average risk who are older than 50 years with any of the following tests: fecal occult blood test (FOBT) annually, flexible sigmoidoscopy every 5 years, double-contrast barium enema (DCBE) every 5 years, flexible sigmoidoscopy every 5 years plus FOBT annually, or colonoscopy every 10 years.2 Despite similar guidelines from the American Gastroenterological Association3 and the American Cancer Society,4 the rates of screening in the population at average risk remain low.5- 10
Previous studies have identified racial and ethnic differences in colorectal cancer care and outcomes.6,11- 14 African Americans and individuals of Hispanic or other minority racial groups are more likely to have their conditions diagnosed when the cancer is at an advanced stage,14 are less likely to receive chemotherapy,13 and have a higher mortality rate due to colon cancer.14 Contributing to and compounding this problem is the lower rate of screening in these racial/ethnic groups.15 Gender and socioeconomic disparities have also been identified in screening practices.9,10,15,16
More than 95% of US residents 65 years or older receive inpatient and outpatient health services through Medicare. Beginning on January 1, 1998, Medicare began reimbursement for colon cancer screening for persons at average risk with an annual FOBT, flexible sigmoidoscopy, or DCBE. Beginning on July 1, 2001, Medicare coverage was expanded to cover screening colonoscopy every 10 years for individuals at average risk. Few population-based studies have examined the rates of various colorectal cancer screening methods and disparities in screening practices after this expanded coverage. We performed this study to identify sex, age, socioeconomic, and racial/ethnic disparities in colon cancer screening practices in Medicare beneficiaries in Illinois, New York, and Florida since institution of the expanded coverage.
We examined the Medicare physician/supplier file, which is derived from Medicare Part B claims for physician services, and the denominator file, which contains information on beneficiary entitlement, enrollment in Medicare Parts A and B, Medicare health maintenance organizations, and residential ZIP code. These files are administrative databases maintained by the Center for Medicare and Medicaid services. A one-fifth sample of eligible subjects enrolled in both Parts A and B, exclusively receiving their care through the Medicare fee-for-service system in each of the 3 states of New York, Florida, and Illinois during 2002 and 2003, was identified. Individuals with a personal history of colon polyps (International Classification of Diseases, Ninth Revision [ICD-9] codes V12.72, 211.3, and 211.4), colon cancer (ICD-9 V10.05 and V10.06), or inflammatory bowel disease (ICD-9 555.x, 556.x, 558.2, and 558.9) were excluded from the analysis because they are at higher risk of colon cancer and the same screening guidelines are not used as for individuals who are at average risk. Subjects older than 90 years or with missing data (n = 724) on race/ethnicity or sex were excluded from the analysis. The study was approved by the institutional review board of the Medical College of Wisconsin.
Use of Medicare claims is a well-validated method to determine health care usage rates.17 A previously used algorithm was employed to identify specific colon cancer screening tests.7,8 Using Medicare claims data, screening procedures were identified using the following Health Care Financing Administration Common Procedure Coding System and Current Procedural Terminology codes: colonoscopy (44388, 44389, 44392, 44393, 44394, 45378, 45380, 45383, 45384, 45385, G0105, and G0121), sigmoidoscopy (45300, 45305, 45308, 45309, 45315, 45320, 45330, 45331, 45333, 45338, 45339, and G0104), DCBE (74270, 74280, G0106, G0120, and G0122), and FOBT (G0107 and G0328). The G0107 code for FOBT refers to the recommended 3-card test and not the in-office single-card test.
We considered the procedures to be screening tests if they were coded using the relevant Health Care Financing Administration Common Procedure Coding System codes or using the appropriate ICD-9 codes for screening (V76.51 and V76.51). They were also considered to be screening tests if they were not associated with any of the following symptom ICD-9 codes in the 3 months preceding the procedure: abdominal pain (787.3, 789.0x, and 789.6x), gastrointestinal tract bleeding (578.x), positive FOBT (792.1), weight loss (783.2), iron deficiency anemia (280.x), or anemia, unspecified (285.9). Colonoscopy, flexible sigmoidoscopy, and DCBE were defined as invasive tests. We included only the first test performed in each subject during the study period.
Data from the US census linked with the Medicare database were used to determine the socioeconomic characteristics of study subjects' ZIP codes of residence.18 Educational achievement (percent of residents in the ZIP code who graduated from high school), and ZIP code per capita income level were divided into tertiles for the combined study population from the 3 states. Educational achievement was defined as low if the proportion of persons who graduated from high school in the ZIP code was less than 0.79, medium if between 0.79 and 0.87, and high if greater than 0.87. We used the combined study population for division into tertiles because there was no significant difference among the 3 states insofar as the tertile categories. Age was divided into 5-year intervals from 65 to 79 years, plus a final category comprising individuals aged 80 to 90 years. Race/ethnicity was classified as white or nonwhite.
Data analysis was performed using Stata 8.0 for Windows (Stata Corp, College Station, Tex). We calculated the percentages of patients who had undergone any colorectal cancer screening procedure, and for each individual test we analyzed the screening practices by age group, sex, race/ethnicity, educational achievement, per capita income level, and state of residence using univariate analysis. All variables showing a significant association (P<.05) with the outcome were included in the final multivariate model to arrive at adjusted estimates. In addition, we analyzed interactions between the demographic variables (age, sex, and race/ethnicity) and the socioeconomic markers (educational achievement and per capita income level) and between the various demographic variables. Interactions that were statistically significant (P<.05) were included in the final model. For the outcome “screening colonoscopy,” the reference population comprised individuals who had not undergone any screening test or had undergone screening using the other 3 tests. The purpose of this analysis was to identify characteristics that were specifically associated with selecting colonoscopy as the screening method. Likelihood ratio tests were used to compare the various models, and our final model was the most parsimonious after including the significant interaction terms. We also checked for the presence of outliers and influential observations. The odds ratios obtained were converted to relative risk (RR) because of the common incidence of the outcomes of interest.19
A total of 694 451 Medicare beneficiaries from Florida (n = 269 418), Illinois (n = 182 291), and New York (n = 242 742) were considered eligible for the study. After exclusion of individuals with a personal history of colon cancer (n = 49 077) or other high-risk conditions (n = 48 904), we arrived at the final study population (n = 596 470). There were an approximately equal number of subjects in each age category, and more women than men in each of the 3 states (Table 1). Most of the population was white (89.5%), with African Americans (7.0%) and Hispanics (2.1%) representing the largest minority groups.
Approximately 18.3% of the study population had undergone a screening test during 2002-2003. A larger percentage of the population had undergone a screening colon test in Florida (21.5%) compared with Illinois (16.6%) and New York (16.2%). Almost 22% of individuals aged 65 to 69 years had undergone a screening test, compared with only 11.7% in the population older than 80 years.
Overall screening for colon cancer varied by race/ethnicity, income level, and educational achievement but not by sex. An equal proportion (18.3%) of men and women had undergone a screening colon test. Blacks (9.7%) and Hispanics (8.1%) had lower rates of colon cancer screening compared with whites (19.3%). Individuals living in ZIP codes with a higher per capita income were more likely to undergo a colon screening test than were those living in ZIP codes with a lower per capita income (21.0% and 14.6% in the highest and lowest tertiles, respectively).
Fecal occult blood testing and colonoscopy were the most common tests, accounting for 53.7% and 42.1%, respectively, of all colorectal screening tests. Colonoscopy was used more commonly than the other screening tests among men, younger age groups, and nonwhite patients. No such association was seen for income or educational achievement levels.
We explored the interactions of sex with income level, educational achievement, race/ethnicity, and age group. There was a significant interaction between age group and sex (P <.001) and between income level and sex (P <.05) for all screening tests and in an analysis restricted to screening colonoscopy alone. There was also a significant interaction between race/ethnicity and both sex and income level for screening colonoscopy (P <.05). There was no interaction between race/ethnicity and the other demographic and socioeconomic variables when analyzed for the outcome “all screening tests.” After stratifying by age group and income level, we did not find a significant interaction between sex and educational achievement.
Analysis of sex disparities in all screening tests and screening colonoscopy revealed interesting patterns (Table 2). Younger women were as likely as or more likely than men to undergo any screening test. However, the RR of undergoing any screening test for women compared with men was lower in the highest income tertile compared with the lowest income tertile. We analyzed the sex disparity in screening colonoscopy after stratifying by race/ethnicity. Among whites, the disparity was highest in the older age groups. It was also most prominent in the highest income tertile. In the age group 65 to 69 years, the RR for women to undergo a screening colonoscopy in the lowest income tertile was 0.86 (95% confidence interval [CI], 0.81-0.92) compared with 0.80 in the highest tertile (95% CI, 0.76-0.83). This pattern was consistent across all age groups. Among nonwhites, the sex disparity was significant only in the oldest age groups, and the difference was again largest in the highest income tertile.
Individuals residing in a higher per capita income ZIP code were more likely to undergo any screening test (Table 3). Among whites, higher per capita income was also associated with greater rates of screening colonoscopy. This association was stronger in men (RR for the highest income tertile, 1.19; 95% CI, 1.14-1.25) than in women (RR, 1.09; 95% CI, 1.05-1.15). Income was not associated with higher rates of screening colonoscopy in nonwhites.
Beneficiaries older than 80 years were less likely to undergo any screening test compared with those aged 65 to 69 years (Table 4). This disparity was greater in women (RR, 0.44; 95% CI, 0.43-0.45) than in men (RR, 0.61; 95% CI, 0.59-0.63). Older persons were also less likely to undergo a screening colonoscopy or FOBT.
Minority racial/ethnic groups were only half as likely as whites to undergo any screening test (RR, 0.52; 95% CI, 0.50-0.53) or a screening colonoscopy (Table 5). There was also a trend for greater disparity in screening colonoscopy in the highest income tertile (Table 6). Among the racial/ethnic subgroups, Hispanics had the lowest RR of undergoing a screening test compared with whites (RR, 0.47; 95% CI, 0.45-0.50; data not shown). Residence in a ZIP code with a greater proportion of high school graduates was strongly associated with undergoing colon cancer screening (RR, 1.52; 95% CI, 1.48-1.55 for the highest tertile).
We found a low rate of colorectal cancer screening in a study population at average risk consisting of Medicare patients in 3 geographically diverse states studied after the expansion in Medicare coverage to include screening colonoscopy in individuals at average risk. An equal percentage of men and women underwent colon cancer screening tests, but women were less likely to undergo invasive screening tests. Nonwhites and persons residing in ZIP codes with low per capita income were less likely to undergo colon cancer screening tests. Persons residing in ZIP codes with low per capita income were also less likely to undergo any colon cancer screening. To our knowledge, ours is the first study to analyze sex, racial/ethnic, and socioeconomic disparities since institution of the expanded Medicare coverage.
Just over 18% of the population underwent any colorectal cancer screening test in our 2-year study, yielding an annual screening rate of 9%. Our finding of 3.8% for annual colonoscopic screening is higher than the 1.44% reported by Ko et al.8 While this difference may be due to geographic variation, it is also consistent with an increased use of colonoscopy for colon cancer screening since expansion of Medicare coverage, as in 2 previous studies.20,21
We found that women were more likely than men to undergo a screening test but were less likely to undergo an invasive screening test or screening colonoscopy and more likely to undergo FOBT. This sex disparity in screening colonoscopy in whites, who formed a major portion of our study population, was greatest in the highest per capita income tertile. To our knowledge, this pattern has not been reported previously in reference to colon cancer screening practices. A theory of diminishing returns has been used to explain similar findings for other health outcomes in minority racial/ethnic groups.22 Future studies should further examine a possible sex-income interaction, particularly since colon cancer rates are higher in affluent populations.23,24 The higher use of FOBT and lower rates of invasive tests in women have been identified in other studies10,11,16 and have been associated with physician sex.25 Female patients also may perceive colorectal cancer to be predominantly a disease of men26 or prefer FOBT to invasive screening tests.15 A preference for FOBT may be harmful; while all 4 screening tests are recommended equally by the professional societies, one recent study found that in women only 33% of advanced neoplasias detected at colonoscopy would have been detected at flexible sigmoidoscopy.27 Inasmuch as overall colorectal cancer screening rates are much lower than for breast and cervical cancer, other preventive care measures must be used to educate women about colon cancer.10
Elderly persons in our study, particularly those 80 years and older, were less likely to undergo screening tests or screening colonoscopy. While current guidelines do not specify an upper age limit for colorectal cancer screening, recent studies have questioned the life expectancy benefit obtained from performing screening colonoscopy in very elderly persons.28 The benefits of screening an elderly person must be weighed against the risk of the procedure as well as competing comorbid conditions.
Nonwhites in our study had lower rates of screening, and even with adjustment for socioeconomic status were much less likely to undergo screening tests. The lower rates of screening in this population is worrisome because both colon cancer incidence12 and mortality14 are higher than in whites. Lack of screening has also been associated with a higher incidence of cancers diagnosed at an advanced stage.29 There was a greater racial/ethnic disparity in screening colonoscopy in the higher income tertiles, and, after stratifying by race/ethnicity, income level was not associated with screening in nonwhites. These findings suggest that, while higher income, in general, is associated with increased screening, the benefits of a higher socioeconomic class is not evident in minority racial/ethnic groups,22 thus emphasizing the need to target this population as a whole rather than limiting interventions to only individuals with lower income levels.
This racial/ethnic disparity in colon cancer screening has been reported in most, but not all, other studies.9,15,16,30,31 In a recent study using the Medicare current beneficiary survey, O’Malley et al32 found that racial/ethnic differences were explained entirely by difference in socioeconomic status. We did not find that this completely accounted for disparities in our adjusted model. This difference between the 2 studies could be due to the limited ability to account for individual socioeconomic status in our study, but most studies of other conditions also continue to find independent associations with both race/ethnicity and socioeconomic status.33,34 Differences in screening rates between various racial/ethnic groups might be related in part to problems with health care access, attitudes of the community toward screening, cultural beliefs, and physician practices.35,36
We identified significant disparities in screening practices among the different income level and educational achievement tertiles. While lack of health insurance is a significant factor determining affordability of health care and health disparities, this problem is minimized to some extent with health insurance programs such as Medicare that provide similar benefits and coverage to most participating individuals. Among Medicare patients, lower socioeconomic status might reflect lower awareness and health literacy, both of which have been shown to affect screening practices.37- 39
For analysis of screening colonoscopy, both individuals who had not undergone any screening test and those who had undergone screening with the other methods constituted the reference population. We expect that including individuals who had undergone screening with other methods in the reference group would have resulted in our estimates being closer to null and that the true estimates of the disparities would be stronger if compared only with subjects who underwent no screening test.
There are a few limitations to our study. Because our study was cross-sectional, it is possible that individuals might have undergone an endoscopic screening test just before our study began and might have been current with screening during our study. However, because our study began soon after Medicare introduced coverage for colonoscopy, we expect the proportion of individuals who had undergone a screening colonoscopy before our study to be fairly small based on previously published data. Also, before colonoscopy, the most commonly used screening test was the FOBT, and individuals who had undergone the FOBT just before our study would still be expected to undergo a repeat evaluation during our study. Our outcome variables and exclusion criteria were based on the Health Care Financing Administration Common Procedure Coding System, and Current Procedural Terminology coding and might underestimate total tests performed or incorrectly allow a small number of patients at high risk for colon cancer into our study. While our discrimination between screening and diagnostic tests is also dependent on coding accuracy, we found similar sex, racial/ethnic, and socioeconomic differences in our analysis of all colon tests as for screening tests. Educational achievement and per capita income levels were recorded on the basis of ZIP code and, while this method might give fairly accurate results, it should not be taken to represent individual levels of per capita income and educational achievement. The relatively small number of nonwhites compared with whites in our population may have limited our power in examining multiple interactions between the variables in this subgroup, resulting in wider CIs. While inclusion of a random sample of the Medicare population from 3 states makes this study more representative and diverse than a single-state or non-Medicare sample, the composition of the population in these states differs from that of the general US population and the results must be extrapolated to other states or younger patients with caution.
Despite expanded Medicare coverage, there are still significant disparities in colorectal cancer screening practices. Women and nonwhites are less likely to undergo colonoscopy to screen for colorectal cancer. Levels of educational achievement and per capita income also strongly influence the odds of undergoing colon cancer screening. There are complex interactions between the demographic and socioeconomic variables that need to be considered in future studies. Further research is needed to determine the basis for the observed ongoing disparities to develop interventions to reduce and eliminate these differences. Policy initiatives are necessary to increase the awareness of colorectal cancer screening, especially in women and in racial/ethnic minorities, and to increase physician awareness about screening.
Correspondence: Joan M. Neuner, MD, MPH, Center for Patient Care and Outcomes Research, Medical College of Wisconsin, 8701 Watertown Plank Rd, Suite H2755, Milwaukee, WI 53226 (firstname.lastname@example.org).
Accepted for Publication: October 17, 2006.
Author Contributions:Study concept and design: Ananthakrishnan, Schellhase, Laud, and Neuner. Acquisition of data: Sparapani and Neuner. Analysis and interpretation of data: Ananthakrishnan, Schellhase, Laud, and Neuner. Drafting of the manuscript: Ananthakrishnan and Sparapani. Critical revision of the manuscript for important intellectual content: Ananthakrishnan, Schellhase, Laud, and Neuner. Statistical analysis: Ananthakrishnan, Sparapani, and Laud. Obtained funding: Neuner. Study supervision: Schellhase and Neuner.
Financial Disclosure: None reported.
Funding/Support: This study was supported in part by grant K08AG-021631 from the National Institutes of Health (Dr Neuner).