|
Does Medicare Coverage Of Colonoscopy Reduce Racial/Ethnic Disparities In Cancer Screening Among the Elderly?
Racial and ethnic disparities in colorectal cancer screening have been documented extensively in the literature. In July 2001 Medicare began covering colonoscopy for average-risk beneficiaries. We examined the effect of Medicare reimbursement on the rate and disparity of colorectal cancer screening among the elderly in the United States. This policy alleviated the screening disparity between non-Hispanic whites and blacks, but the gap between Hispanics and non-Hispanic whites has widened. Overall, fewer than half of the elderly are screened, even though Medicare now covers colonoscopy.
COLORECTAL CANCER IS THE THIRD most commonly diagnosed malignancy and the second leading U.S. cause of cancer death.1 The association between colorectal cancer screening and reduced colorectal cancer mortality is well established; hence, screening of average-risk populations is recommended.2 Despite these recommendations, colorectal cancer screening in the United States is used much less than that for breast or cervical cancer. A study using data from the National Health Interview Survey (NHIS) reported 67 percent and 80 percent screening rates for mammography and Pap smears, respectively, but a rate of less than 40 percent for colorectal cancer screening.3 Racial and ethnic disparities in colorectal cancer screening have been documented extensively; most studies have found a lower rate of screening among racial/ethnic minorities.4 Lack of health insurance and variations in coverage are two of the most frequently cited reasons for the observed racial/ethnic disparities in use of or access to health care.5 Disparities in colorectal cancer screening have persisted among Medicare beneficiaries.6 One plausible explanation is that previous studies reporting racial/ethnic disparities in colorectal cancer screening were based on data collected in the 1980s or 1990s, when Medicare coverage of colorectal cancer screening was limited to "high-risk" beneficiaries. The lack of insurance coverage for colonoscopy for those of average risk might have created access barriers for the lower-income elderly, and it likely led to the previously observed racial/ethnic disparities in screening among Medicare beneficiaries. Medicare began paying for colonoscopy screening for average-risk beneficiaries on 1 July 2001.7 Medicare reimbursement for this screening procedure had been limited to those at high risk, whereas fecal occult blood test and flexible sigmoidoscopy were covered for all elderly beneficiaries.8 Medicare’s reimbursement for colonoscopy reduces financial barriers to screening and should increase its use among the elderly. Two studies have documented an increase in the rate of endoscopic colorectal cancer screening among Medicare beneficiaries and those ages 50–64.9 One reported a significant (p < .01) increase in use of colonoscopy in a tertiary institution within six months after the policy went into effect.10 The other observed an increase in the rate of screening from 4.6 percent before to 14.2 percent after the coverage change among average-risk patients in a large network of gastrointestinal physicians.11 To our knowledge, no studies have examined the effect of this Medicare policy on the rate of colorectal cancer screening in a nationally representative sample of the U.S. elderly population or its impact on screening disparities among the elderly in various racial/ethnic groups. Our study is intended to fill that gap.
Data source. For this study, we used the 2000 Cancer Control Module (CCM) and 2003 Cancer Screening Supplement (CSS) of the NHIS. The NHIS is a nationally representative probabilistic annual health survey.12 It has been the primary source of information on the health of civilian, noninstitutionalized U.S. households since 1957. A CCM was administered in the 2000 NHIS to examine cancer-screening behavior and cancer risk factors such as diet, nutrition, physical activity, and family history. The 2003 NHIS also fielded a CSS, but it covered only some of the questions asked in 2000. To examine whether Medicare coverage of colonoscopy changed the pattern of screening colonoscopy use among the elderly, we selected two years of NHIS data, 2000 and 2003, which contain information on the use of colonoscopy. The two years represent pre- and postcoverage periods. Our study was limited to respondents age sixty-five and older. Key variables. The key variables for the study were the use of endoscopic colorectal cancer screening (the outcome of interest), and Medicare beneficiaries’ racial and ethnic characteristics (the explanatory factors of interest). We identified the use of colorectal cancer screening from the NHIS survey question, "Have you EVER HAD a sigmoidoscopy, colonoscopy, or proctoscopy?" If a respondent answered "yes," we considered him or her to have been screened. Race/ethnicity was classified into four categories: Hispanic, non-Hispanic white, non-Hispanic black, and non-Hispanic other races; included in the "other races" were American Indians/Alaska Natives, Asians, Native Hawaiians and other Pacific Islanders.
Other explanatory variables.
We included age and sex in addition to race/ethnicity. Age was categorized into three groups: 65–74, 75–84, and 85+, since studies have shown a nonlinear relationship between age and screening.13 We described geographic characteristics by census region and location in a metropolitan statistical area (MSA). Because of confidentiality concerns, the MSA variable is no longer available in the 2003 NHIS. Education attainment was categorized into four groups; we dichotomized marital status as married or not married and place of birth as U.S.-born or foreign-born. Income was classified in four income levels and unknown.14 Barriers to access included the lack of a usual source of care or lack of supplemental insurance (private insurance or Medicaid). Health status was measured as self-perceived health and self-reported cancer history (yes or no). All explanatory variables are listed in Exhibit 1
Statistical analysis. We first compared the differences in colorectal cancer screening rates between the 2000 and 2003 samples for the four defined racial/ethnic groups, using Wald chi-square statistics.15 We then used multivariate logistic regression to examine the differences in the likelihood of screening across racial/ethnic groups, while controlling for other factors that have been found to be associated with this screening in the literature, such as socioeconomic status or access barriers. We employed appropriate weighting procedures in Stata 8.0 and defined statistical significance as p < .05.
Descriptive statistics. Exhibit 1  compares the characteristics of the 2000 and 2003 NHIS populations. On average, a significantly higher proportion of the elderly in 2003 were in the oldest age group (eighty-five years and older), and a lower proportion were at the lowest education level (less than high school) or in the poor/near-poor income group. No differences were found in the distribution of other population characteristics.
Racial/ethnic groups.
In 2003, the rate of endoscopic colorectal cancer screening was significantly higher than in 2000 (p < .001) (Exhibit 1
Comparisons of the odds of screening in multivariate logistic models. After other confounders were controlled for, in 2000, non-Hispanic blacks were less likely to have had endoscopic colorectal cancer screening compared with non-Hispanic whites; the odds of screening increased in 2003, and the previously observed disparities between these two groups were no longer significant in 2003 (Exhibit 3  ). In contrast, the odds of screening declined for Hispanics between the two study years, and the differences between Hispanics and non-Hispanic whites became significant in 2003 (p = .048). No statistically significant differences were found between non-Hispanic whites and non-Hispanic others in either year.
Associations between the rate of endoscopic colorectal cancer screening and other covariates were similar to those reported in the literature. However, we found that some associations appeared to have grown stronger over time, while others had become weaker. Most noticeably, sex differences in screening were more apparent over time. Moreover, elderly in the low-income group in 2000 were not more likely to have been screened than those in the poor/near-poor income group, whereas the difference was significant in 2003 (p < .001). Similarly, having a history of cancer was not significantly associated with screening in 2000 but became significant in 2003 (p = .041). The positive effect of having a usual source of care also became more pronounced over time. The observed differences between U.S.- and foreign-born groups and across geographic areas were no longer significant in 2003 (p = .207). All other covariates showed a similar trend over time.
The expansion of Medicare reimbursement of colonoscopy to the average-risk population on 1 July 2001 was intended to reduce the economic access barrier to colorectal cancer screening, including that associated with race/ethnicity. We conclude that although the above goal has been achieved overall, all sectors of the population have not benefited equally; there is still room for improvement. We found that even after Medicare began covering colonoscopy, more than half of elderly respondents reported never having this screening procedure. The disparities in screening rates between non-Hispanic whites and non-Hispanic blacks have been reduced: They were no longer statistically significant after Medicare began covering colonoscopy. However, the disparities between the non-Hispanic whites and Hispanics became significant after Medicare coverage began. A number of factors might have contributed to a lower rate of endoscopic colorectal cancer screening among elderly Hispanics and a slower rate of catching up after Medicare coverage became available. First, a disproportionately higher percentage of Hispanics were in the poor/near-poor income group; Hispanics accounted for less than 6 percent of the U.S. elderly population, but approximately 10 percent of the elderly in this income group were Hispanic. Also, a large percentage of Hispanics were foreign-born. Additionally, Hispanic beneficiaries were less likely than others to have a usual source of care or private supplemental insurance. All of these factors might have contributed to the lower rate of screening observed in both years. After controlling for these and other factors, we found no significant association between Hispanic ethnicity and endoscopic colorectal cancer screening in 2000 but strong evidence of disparities in screening among Hispanics in 2003. This suggests that factors uniquely associated with a lower rate of screening among Hispanic Medicare beneficiaries make this population more vulnerable even after economic barriers to screening have been removed. These might include a lack of awareness of the availability of Medicare coverage or cultural factors. There are certainly cultural differences between Hispanics and blacks that could lead to the observed findings. These include barriers to seeking health care such as language and health literacy, a nihilistic view of a cancer diagnosis (as a death sentence), preference for not knowing about the disease before it becomes symptomatic, and many others.16 However, in most studies, the most consistent predictor of cancer screening use is a recommendation from the health care provider.17 The second most consistent predictor is the patient’s perception of risk. Both of these factors are strongly influenced by the prevalence of the disease in the population of interest. Therefore, it is important to note that both the incidence (39.0 per 100,000) and the mortality (14.2 per 100,000) associated with colorectal cancer among U.S. Hispanics are much lower than those observed among non- Hispanic whites (53.6 and 20.1 per 100,000, respectively) and blacks (62.4 and 27.9 per 100,000, respectively).18 Given the high incidence and mortality of colorectal cancer among blacks, it is perhaps not surprising that their physicians would be more likely to recommend screening and that these patients would more likely comply, compared with Hispanics. This is a possible explanation, although not a justification, for the low rates of colorectal cancer screening among Hispanics. However, this observation points to the importance of tailoring interventions to specific populations. Specifically, among Hispanics and their physicians, awareness of risk could be an important target for intervention. Although the coverage of colonoscopy was intended to reduce the economic burden of colorectal cancer screening for Medicare beneficiaries, such benefits are subject to the Part B deductible and a 20 percent copayment. The estimated cost of a colonoscopy, based on Medicare-allowable payments in 2000, was $695.95 without a biopsy and $1,003.76 with a biopsy.19 Therefore, beneficiaries paid a $139.20 copayment or more, depending on whether a biopsy was performed, plus an annual deductible of $110 (in 2005) for Part B services. This out-of-pocket amount could be prohibitive for poor and near-poor beneficiaries. Some dually eligible beneficiaries avoid the out-of-pocket expenses through Medicaid coverage; however, as of 2004, screening colonoscopy was not covered by Medicaid in thirty-two states.20 The lack of Medicaid coverage in many states probably explains the lack of association between Medicaid supplemental insurance and screening while private supplemental insurance was positively associated with screening. The increasing odds of screening found among beneficiaries in the low-income group, compared with those in the poor/near-poor group, suggest that the coverage of colonoscopy might have been effective in increasing the rate of screening among people in the former group but has remained ineffective for those in the latter. Study limitations. We examined the impact of the expansion of Medicare coverage of colonoscopy on the rate of screening by comparing the rate before and after the coverage policy took effect. Because other time-varying factors such as an increasing awareness of the importance of colorectal cancer screening or a change in diet and physical activity might have also contributed to a higher rate of screening in 2003, our findings are best interpreted as variations in the odds of screening when the economic barrier of screening was reduced. It is difficult to quantify the extent to which these variations were attributable to the Medicare policy. Our analysis assumed that nearly all elderly are covered by Medicare. Although this assumption was supported in the published statistics, studies have found that Hispanics are overrepresented among the uninsured elderly.21 Therefore, some of the observed disparity in colorectal cancer screening between the Hispanic and non-Hispanic white elderly could have been attributable to a lack of Medicare insurance among Hispanics. We could not address this issue because the low uninsurance rate (less than 5 percent) in the elderly population left us with an extremely small number of observations for the uninsured subgroup in each racial/ethnic group. Future studies should consider combining several years of NHIS data to explore this issue. We defined colorectal cancer screening as ever having had an endoscopic examination. The NHIS also asked those who had received an examination the main reason for their most recent examination, in an attempt to differentiate examinations performed for screening from those for diagnosis or confirmation. Our analysis did not make such distinctions for two reasons. First, recent receipt of a confirmatory or diagnostic test does not necessarily imply that no earlier screening examination was done. Second, we believe that the elderly who have never had any examination pose the greatest public health concern. However, if we redefined our outcome of interest as endoscopic examinations for screening purposes only, we would have to combine two very heterogeneous groups, those never screened and those with an examination for nonscreening purposes, as the "no screening" group. Future studies are recommended to further explore this topic using more advanced multivariate methods such as ordered logit models. Policy implications. On 15 September 2004 the Centers for Medicare and Medicaid Services elicited participation from health care providers and organizations in the Cancer Prevention and Treatment Demonstration for Ethnic and Racial Minorities. African American, Hispanic, Asian American, Pacific Islander, and American Indian Medicare beneficiaries are eligible for recruitment. One of the purposes of this project is to "eliminate disparities in the rate of preventive cancer screening measures."22 Our study makes a timely observation to inform this demonstration: It identifies Hispanic beneficiaries as those most urgently in need of increased colorectal cancer screening. Further, we have shown that demonstrations focusing on Hispanics should explore whether the rate of screening can be improved by waiving the Medicare copayment and deductible for colonoscopy.
Ya-Chen Tina Shih (yashih{at}mdanderson.org) is an associate professor in the Section of Health Services Research, Department of Biostatistics and Applied Mathematics, M.D. Anderson Cancer Center at the University of Texas in Houston. Lirong Zhao is a statistical analyst there, and Linda Elting is a professor. The authors gratefully acknowledge editors of Health Affairs and two anonymous reviewers for their valuable comments on an earlier draft.
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Study Data And Methods
Study Results
