Socio Economic position (SEP) explains more of the disparities for African-American versus white women in the United States compared with other ethnic comparisons. The role of SEP appears to be smaller in more recently published papers. We also found that the differences in breast cancer survival between ethnic groups may in part be explained by BMI, but there is little evidence to implicate smoking or alcohol consumption as explanatory factors for this inequality. Furthermore, given social patterning of BMI and other lifestyle habits, it is possible that our results for SEP and BMI are measuring the same effect.

A suggestion frequently made regarding ethnic inequalities in health is that there may be an underlying genetic basis for these inequalities. Although there is a lack of major systematic genetic differences between ethnic groups, there are extensive differences in lifestyle, suggesting that health disparities are most likely driven by environmental factors (1, 2). In relation to breast cancer survival, although differences in certain allele frequencies have been related to prognosis, we are not aware of any studies demonstrating that these differences could explain ethnic inequalities in survival. On the contrary, numerous studies point to equally plausible, and more coherent, alternative explanations for the observed inequalities. The majority of this evidence relates to Asian women. Comparisons within one ethnic group cannot be explained by differences in genetic makeup. Chuang et al. (69) found that Chinese women born in the United States had better survival than Chinese women born in East Asia (HR = 1.22, 95% CI: 1.06, 1.40). Similarly, changes in survival across generations among immigrant women are an indicator of the importance of environmental and cultural factors. Pineda et al. (70) demonstrated such changes for Chinese and Japanese, although not for Filipino, women. These variations were almost fully explained by demographic and stage- and treatment-related factors.

Some environmental exposures that are culturally related are likely to persist across generations. Furthermore, living as a first-generation immigrant in a country poses its own challenges, and linguistic and cultural barriers in access to care are likely to be important (69). The evidence above is supplemented by observations of changes over time in survival inequalities between ethnic groups. Jatoi et al. (71) documented widening inequality in all-cause mortality following breast cancer between black women and white women in the United States. This effect was driven by the most recently diagnosed cohort of women (1995–1999). Such changes are more plausibly explained by improvements in the health system being better tailored to a dominant ethnic group. Finally, the importance of SEP and BMI in explaining inequalities, as shown in this review, adds to the evidence that genetic differences between ethnic groups are unlikely to be important determinants of inequalities.

Evidence is accumulating that women of different ethnicities experience disproportionate risks of various breast cancer subtypes. For example, we have shown that, in New Zealand, Māori and Pacific women are more likely than non-Māori/non-Pacific women to have human epidermal growth factor receptor-2–positive breast cancer (72). Māori women are less likely and Pacific women more likely than non-Māori/non-Pacific women to have a negative estrogen receptor and progesterone receptor status. In the United States, Hispanic women are more likely to have estrogen receptor and progesterone receptor negative breast cancer compared with non-Hispanic white women (73), and black women are more likely than white women to have triple-negative breast cancer (74). When Chinese women were compared with white American women, no differences in estrogen receptor or progesterone receptor status were found (69). Because receptor-negative breast cancer is not amenable to hormonal therapy, these ethnic differences could explain some of the inequalities in survival. However, even among women with triple-negative breast cancer, 5-year relative survival was lowest in the non-Hispanic black group (74). The presence of differential subtypes of disease could be due to differential risk factors, and breast cancer epidemiologists should refine their outcomes to account for these differences.

Our final analysis showed that adjusting the breast cancer survival inequalities for measures of SEP had a greater impact in studies of African-American versus white women compared with other ethnic comparisons. This finding is partly due to the higher crude inequality in survival between African-American and white women (Figure 3A), so there was more inequality to “explain” through adjustment. However, it is probably also partly due to the higher proportion of African Americans (compared with other ethnic minorities) living in poverty (25% vs. 8%) (75) and the probable higher level of resulting socioeconomic homogeneity in the census groups used to assign SEP in the studies included in this review. The effect of adjusting ethnic inequalities in health for SEP is strongly affected by the choice of SEP measure, and health inequalities between different ethnic groups respond in different ways to this adjustment (76). Therefore, the aggregation of different ethnic groups, as seen in several of the included studies (e.g., the grouping “Asian/Pacific Islander” in the United States and “South Asian” in the United Kingdom) could be masking the real effect that various individual-level measures of SEP would have on ethnic inequalities in survival.

One final study, which followed up women diagnosed with breast cancer in the Women’s Health Initiative, found that after adjusting for age, stage of disease, BMI, and whether the women was in the clinical or observational arm of the study, a significant increased risk of death still remained for African-American compared with white women (50).

Five studies (9 comparisons) contributed to the overall crude analyses. The pooled hazard ratio showed that women in the minority ethnic groups were more likely to die during follow-up (HR = 1.38, 95% CI: 1.16, 1.64) (Figure 3A). However, there was significant heterogeneity between studies (P < 0.001).

Correspondence to Fiona McKenzie, Centre for Public Health Research, Massey University, Private Bag 756, Wellington, New Zealand (e-mail: f.j.mckenzie@massey.ac.nz).

Source:

http://epirev.oxfordjournals.org/content/31/1/52.full?sid=2aab17bb-27b8-4a3e-b7ae-2895120a53da#sec-7