Despite growing interest in factors associated with cancer-related disparities and interventions to reduce disparities, information is still limited regarding differences in screening practices by geographic location and effective tailored interventions for specific racial/ethnic groups. African American and Latina women have disproportionately higher death rates due to breast cancer yet nationally lower incidence rates than those of their non-Hispanic white counterparts (1). Although screening rates for breast cancer have increased for African American and Latina women, breast cancer cases among these populations are often diagnosed at later stages, which limits treatment options (1,2). Although national data illuminate prevailing disparities, data are not available for women who are categorized as non-Hispanic white but who consider themselves to be of a different racial/ethnic group and are low-income or medically underserved or both. In this study we included Arab women traditionally categorized as non-Hispanic white but representing 10 northern African countries and 12 Persian Gulf countries (3) who are not ethnically identifiable in national databases such as the Behavioral Risk Factor Surveillance System (BRFSS) (4).

As women's screening rates for breast cancer increase nationally, barriers to screening remain for African American, Arab, and Latina women from poor backgrounds. Multiple factors are associated with breast cancer screening rates among these groups of women, including socioeconomic status, health insurance coverage, usual source of care, perceptions and fears about breast cancer, race, ethnicity, age, and knowledge of breast cancer screening (5-8).

A special edition of the Michigan Cancer Behavioral Risk Factor Survey (BRFS) conducted in 2006 (9) disaggregated Arab women from non-Hispanic white women. The survey showed that screening rates for annual mammography and clinical breast examination were 53% (95% confidence interval [CI], ±6%) and 46% (95% CI, ±20%) for African American and Latina women aged 40 years or older, respectively, which were lower than rates for the general Michigan population (54% [95% CI, ±4%]) and for Arab women (68% [95% CI, ±24%]). Nationally, the 2005 BRFSS indicated that recent mammography screening rates for African American women aged 40 years or older were approximately 67% and for Latinas aged 40 years or older were approximately 65%, comparable to rates for non-Hispanic white women (63%). Mammogram screening data serve as a proxy for adherence to guidelines that recommend yearly mammograms for women aged 40 or older (10). However, the combination of mammography screening and clinical breast examination would give a more accurate picture of current breast cancer screening behaviors overall; guidelines also recommend clinical breast examinations every 3 years for women in their 20s and 30s and every year for women aged 40 years or older (10).

We studied differences in knowledge of and socioeconomic factors associated with breast cancer screening and assessed baseline and postintervention test scores on the knowledge of breast cancer screening practices among African American, Arab, and Latina women who participated in the Kin Keeper Cancer Prevention Intervention, a family-focused educational intervention for women (11). We hypothesized that, whereas all 3 population groups should be targeted for the educational intervention, Arab and Latina women have lower socioeconomic status and lower levels of knowledge about screening for breast cancer than do African American women. These racial/ethnic differences are important for the design of tailored interventions because they are likely to affect the screening behaviors and, ultimately, the breast health outcomes among these groups of women.

Sociodemographic characteristics differed significantly by race/ethnicity (Table 1). African American women had higher levels of education, employment, access to health insurance, and income. Approximately 53% of Arab women and 41% of Latina women had not completed high school education or obtained a general equivalency diploma, compared with approximately 3% of African American women (P < .001). Overall, 61% of the women were aged 40 years or older, the age category of women recommended for yearly mammography screening (10). Scores on baseline knowledge test (Figure 1) and postintervention tests (Figure 2) were compared by race/ethnicity. At baseline, African American women had higher knowledge scores; more than 30% obtained a perfect pretest score compared with 9% for Arab women and 5% for Latina women. Mean pretest scores were 3.91 of 5 (SD, 0.92) for African Americans, 3.02 (SD, 1.12) for Latinas, and 2.92 (SD, 1.18) for Arabs. Posttest scores for Latina and Arab women increased significantly (P < .001), as depicted by distribution, which is relatively skewed compared with the pretest distribution in Figure 1. Posttest median scores were 5 for African American women and 4 for Latina and Arab women. The percentage of women in each racial/ethnic group that answered each question correctly varied (Table 2). Question 4, which asked the women to differentiate between types of screening test (self-examination, clinical examination, and mammography) proved to be the most difficult (Table 2) for all racial/ethnic groups.

In the pretest, Arab women had lower odds of correctly answering the first question (who does a breast self-examination?) than did African American women (odds ratio [OR], 0.27; 95% CI, 0.08-0.89), adjusted for other sociodemographic variables (Table 3). Similarly, for the second question (who does a clinical breast examination?), and the fourth question, which tested the ability to distinguish screening types, Arab women were less likely to answer correctly in the pretest. However, for the third question (who does a mammogram?) and the last question, which distinguished who performs each screening type, the odds of answering the question correctly in the pretest were lower for Arab women but not significantly.

For Latinas, pretest results were somewhat similar to those of Arabs. The GEE-based analysis showed that Latinas had lower odds of correctly answering each question in the pretest than African Americans, except for questions 3, 4, and 5, in which the GEE-based odds were not significant. Latinas had significantly lower odds of knowing the differences between breast self-examination, clinical breast examination, and mammogram than did African Americans.

Posttest GEE estimates show that, compared with African American women, Arab women had lower odds of correctly answering questions 1 and 4, whereas Latina women had significantly lower odds of correctly answering question 4.

To our knowledge this is the first study to empirically compare knowledge of breast cancer screening types among African American, Arab, and Latina women. It yielded 3 major findings: 1) knowledge of breast cancer screening practices was highly associated with race/ethnicity; 2) Arabs and Latinas had similar patterns of knowledge and lower levels of education and insurance coverage compared with African American women; and 3) knowledge of breast cancer screening significantly increased for Latinas and Arabs after community-based intervention, as evidenced by posttest scores.

Overall, the results show that, controlling for sociodemographic characteristics, Arab and Latina women had lower pretest and posttest scores for breast cancer screening knowledge compared with African American women, despite significantly improving their scores in the posttest. Respondents did not fully understand the differences in breast cancer screening types.

African American women's socioeconomic characteristics were highly associated with their higher knowledge levels. Overall, women with higher education levels had higher baseline knowledge of breast cancer screening and reported higher levels of breast cancer screening regardless of race/ethnicity. This finding is consistent with the results of previous studies that show education and access to health care are major predictors of cancer screening knowledge and practices (20,21).

The higher scores on knowledge of breast cancer screening among African American women may also be a result of benefits they received from earlier cancer disparities research that focused on African Americans and breast cancer screening awareness campaigns conducted in English. For non–English-speaking Latinas and Arabs, breast cancer awareness campaigns using Spanish and Arabic messages and materials (12,22) have been made available more recently. Also, popular breast cancer websites do not offer educational materials in Arabic. This disparity might explain the lower baseline scores among Arab and Latina women. More than half the Arab (54%) and Latina (54%) women reported not having health insurance. Other factors — low educational levels and limited English language proficiency, combined with low health literacy, limited availability of linguistically and culturally appropriate materials on breast cancer, and limited experience with the health care system — are strongly associated with knowledge barriers regarding breast cancer screening (23).

Irrespective of a woman's education level, posttest scores increased among the 3 groups; Latina and Arab women achieved the largest gains in knowledge of breast cancer screening. These findings demonstrate that information on breast cancer screening can be effectively provided when interventions address barriers. The Kin Keeper intervention mitigated 3 barriers: low education levels, literacy, and limited English language proficiency (24). If the intervention had not been administered orally and without language preference, 76 (68%) of 112 Arab women would have been missed. Oral administration of the questionnaire permitted the researchers to measure what respondents knew regardless of the respondents' ability to read Arabic or English. Translated materials that are linguistically and culturally appropriate and at an appropriate literacy level (14) increase the effectiveness of cancer screening interventions for women of specific ethnicities.

For Arab women, community-based interventions are necessary to reduce disparities, given that they are more likely than non-Hispanic white women to have irregular screenings and cancer that is detected at later stages (24). Language, access to health care, and geographic residence have also been found to adversely affect Latinas' screening practices (25,26).

Although CHWs completed the same training (27) and had the same basic information regarding specifics of breast cancer education, they had flexibility to tailor their home education sessions. The Latina and Arab CHWs found this especially helpful when talking with older participants, who did not speak or understood very little English, or women who did not socialize much and had various cultural or educationally shaped perceptions of cancer (28). The CHWs' knowledge of their respondent population, and the CHWs' sensitivity to women's perceptions, cultural and language nuances, and questions informed the educational intervention to ensure that information was transmitted in the context of women's lives. The posttest scores suggest that the intervention was effective. Data were not collected on whether racial/ethnic concordance of CHW with respondent increased participation or learning; it is an area for future inquiry.

Methodologic cautions are warranted in terms of instrument and identification of populations. As observed, future research needs to include definitions of medical screening procedures for women to ensure their understanding of survey questions being administered, particularly for low-income women with limited proficiency in English. A community-based study, by design, represents the unique needs of a particular geographic area and designated population groups. Therefore, the results are not meant to be applicable to the general US population. Other limitations include lack of longitudinal data to report knowledge retention and length of residency status of Latinas and Arabs who selected a language other than English.

The inclusion of Arab women presented several challenges. Although the Arab world comprises various countries, including 10 in Africa (3), Arabs have traditionally been classified as white and combined with other non-Hispanic whites and nonmultiracial groups in survey data. Therefore, their unique health status is masked. Furthermore, cancer is not readily discussed among Arab groups because of its severity and the perception that all cancer is hereditary (29). The similar socioeconomic characteristics of Arab and Latina women were predictive of lower baseline knowledge levels, suggesting that low education levels and lack of health insurance coverage are the most powerful predictors of knowledge of breast cancer screening regardless of ethnicity.

Because of the complexity of cancer disparities, education interventions must be developed that are appropriate to the linguistic, health literacy, and cultural needs of participants. Population-specific materials need to be administered in conjunction with community-based participants such as CHWs. Increasing women's knowledge about breast cancer screening is an important first step, but moving women in the direction of adherence to breast cancer screening guidelines and assuring their access to health care services would reduce disparities in breast cancer death and illness.