RSS Feeds
Exploring and Crossing the Disparity Divide in Cancer Mortality
Disparities in access to health care and in health outcomes are a significant challenge in the United States. A disproportionate burden of deaths due to cancer is borne by black populations for breast, colorectal (1), and cervical (2, 3) cancer and by persons with lower educational levels for most types of cancer (1, 4, 5). Hispanic women have higher rates of late-stage cervical cancer diagnoses than do non-Hispanic white women (1). The excess burden of death from breast cancer among African-American women occurs despite a lower incidence of breast cancer than that for white women (6).
Current evidence-based methods of screening for colon, cervical, and breast cancer (7-9) enable early detection of lesions, but access to and use of these screening tests vary by race and socioeconomic status. The general trend is that minority persons and those with lower incomes have lower rates of screening (1, 10). Encouraging data from the recent National Health Interview Survey (11) and Behavioral Risk Factor Surveillance Survey (12) that report similar use of mammography across racial groups rely on self-report, an approach that lends itself to inaccuracies (13).
This issue includes 2 articles relevant to cancer screening in special populations. One article from the Breast Cancer Surveillance Consortium (BCSC) reports on 2 potential sources of the disproportionate burden of death due to breast cancer borne by African-American women (14). The second article describes the results of a randomized, controlled trial aimed at countering 1 of those sources, namely less frequent mammography screening. This group tested “prevention coaches” as a means to increase use of cancer screening services by largely Hispanic and African-American patients at community and migrant health centers in New York City (15).
Smith-Bindman and colleagues (14) used mammography registry data pooled in the BCSC to assess relations between use of mammography and disparities in outcomes among more than 1 million women from diverse racial and ethnic backgrounds who have cancer. Between 1996 and 2002, 1 010 515 women in the consortium had 2 588 479 screening mammographies; 17 558 of these women received diagnoses of breast cancer. Cancer outcomes were provided by links to data from the Surveillance, Epidemiology, and End Results program. Women self-reported race at the time of mammography; consortium data and self-reporting were used to assess time since the last mammogram.
The researchers found that the rates of breast cancer were similar between African-American and white women (relative risk, 1.04 [95% CI, 0.95 to 1.1]), whereas Hispanic (relative risk, 0.76), Asian (relative risk, 0.76), and Native American (relative risk, 0.53) women had significantly lower rates of breast cancer. African-American and Hispanic women were more likely than white women to be diagnosed with advanced-stage tumors, and African-American, Hispanic, and Native American women were more likely than white women to be diagnosed with high-grade tumors. Women of different races had varied frequency of mammography screening, and white women were more likely than all other groups to be “adequately screened.” In all racial groups, the rates of larger, higher-grade, and more advanced tumors increased with the time since the last mammogram. In models controlling for recency of mammography, these differences were no longer present for rates of advanced-stage, large, or lymph node–positive breast cancer between white and African-American women. However, African-American women had less favorable tumor grades than did white women, even after accounting for screening frequency.
The authors suggest that more frequent screening might improve mortality rates, especially among women who have been never screened or those who have been infrequently screened. This supposition is buoyed by the strength of the association between frequency of screening and rates of advanced cancer at diagnosis and the robustness of this association across groups of women with different racial and ethnic backgrounds. However, their finding is only an association and cannot establish causation; it is possible that other unmeasured and unknown factors are responsible for risk for advanced cancer and lack of regular screening. The finding of a persistently significant difference in the proportion of high-grade breast cancer among African-American and white women suggests that some portion of the excess mortality burden is not amenable to more frequent screening; in this regard, the authors refer to “indirect evidence of biological differences.” However, biological differences are not the only way to consider this finding; other possible explanations would necessarily include environmental as well as genetic and other factors (16).
We do well to consider the double-edged sword of database research. The authors used the BCSC to great advantage in looking at the association of recent mammography with breast cancer outcomes in minority ethnic groups rarely studied with this power. However, they are limited to conclusions only about the sample who used the consortium's services, and to what happened, but not why. For example, this study does not shed light on why frequency of mammography is lower among minority women than among white women. Although the authors do not provide data on insurance or socioeconomic status, their population is probably nearly universally insured. Selection bias may seem like an unlikely criticism to raise against a study with more than 1 million women, but before making conclusions about an association of tumor grade with race, researchers should remember that the population of African-American women in this study may or may not resemble the population of African-American women in the United States.
The appropriate role of the construct of race in the study of health in a U.S. setting is a topic of debate (17, 18). One hopes that African-American women's increased risk for high–histologic grade tumors will ultimately help in identifying which, if any, biological factors contribute to differential mortality and how to target those factors with therapies.
The data presented by Smith-Bindman and colleagues (14) provide an incentive for efforts to increase regular mammography among all U.S. women. One such effort is described in another article in this issue. Dietrich and colleagues (15) report a targeted multilingual intervention aimed at increasing screening rates for breast, cervical, and colorectal cancer among low-income urban women. Participants were drawn from 11 community and migrant health centers in New York City and were either recruited during clinic visits or referred by a clinician. The study included women who were 50 to 69 years of age who were overdue for at least 1 cancer screening test. The women who were enrolled were primarily Spanish-speaking (>60%) and were covered by Medicaid (>75%). Most had been enrolled at the community and migrant health center for more than 3 years (>67%). Of those reporting race, 38% were self-identified as black. The intervention, a series of partially scripted, motivational telephone interviews with trained prevention care managers, was administered in a block randomized fashion by health center, duration of enrollment, and number of screenings that were not up-to-date (1 vs. 2 or more). The control group received a pamphlet and 1 telephone call.
The intervention lasted for 18 months. Prevention care managers did not order screening tests themselves, although they were authorized at 2 centers to mail cards for fecal occult blood testing to women who expressed willingness to complete the test. Women who told the care managers that they had difficulty communicating with their providers were given brightly colored cards listing the tests for which they were overdue to share with their clinician at the next visit. The prevention managers reached more than 90% of women by telephone and contacted each woman a mean of 4 times (range, 1 to 20 times).
Cancer screening in the intervention group increased during the intervention period, with increases of 7 percentage points for Papanicolaou smears (from 71% to 78%), 10 percentage points for mammography (from 58% to 68%), and 24 percentage points for colorectal cancer screening (from 39% to 63%). All increases were significantly greater than those of the control group.
The investigators are to be lauded for testing an intervention aimed at reducing cancer disparities in low-income populations. The setting, multiple languages, and cultural appropriateness of the intervention clearly enhanced the results. Such individualized, patient-centered, and patient-activating interventions are clearly part of the next wave of interventions (19). Although baseline rates of “up-to-date” Papanicolaou smears were quite high in the study sample to begin with, and cynics would contend that Papanicolaou smear and mammography screening were only incrementally increased by the intervention, every reader should be impressed with the increase in colorectal cancer screening by approximately 25 percentage points. This increase alone might be worth the cost of the intervention. However, a few hard questions follow. For example, what was the cost of the intervention? And how rapidly did the intervention's effect decay?
The study sample consisted of established users of the community health center system. It remains to be seen whether this approach could work among a more transient population. The authors do not tell us anything about the 768 persons who declined to participate; such information could be useful in understanding which segments of the community and migrant health center populations would be left behind if this system were widely adopted.
Disparities in health and health outcomes are well-documented; the time is ripe for closing those gaps. The lessons of successful change efforts that aim at the clinician–patient interface may improve patient and clinician behavior in other encounters and settings. Somehow, America must meet the challenge of providing effective health care to all our citizens.
Article and Author Information
-
Note: The author of this commentary is responsible for its content. No statement in the commentary should be construed as an official position of the Agency for Healthcare Research and Quality or the U.S. Department of Health and Human Services.
-
Grant Support: None.
-
Potential Financial Conflicts of Interest: None disclosed.
-
Requests for Single Reprints: Mary B. Barton, MD, MPP, Center for Primary Care, Prevention and Clinical Partnerships, Agency for Healthcare Research and Quality, 540 Gaither Road, Rockville, MD 20850; e-mail, mbarton{at}ahrq.gov.
