Obesity and Awareness of Obesity as Risk Factors for Breast Cancer in Six Ethnic Groups

Abstract

Objective: To document BMI and knowledge regarding obesity as a risk factor for breast cancer among subpopulations of African-, Caribbean-, and European-American women and to consider the variables predicting obesity in these diverse groups.

Research Methods and Procedures: A stratified cluster-sampling plan was used to recruit 1364 older women from Brooklyn, NY, during 2000–2002. Two groups were born in the United States (African Americans and European Americans), whereas others were from the English-speaking Caribbean, Haiti, the Dominican Republic, and Eastern Europe. Participants provided demographics, height and weight measures, and estimates of the risk obesity posed for breast cancer.

Results: Women from all groups were significantly overweight (BMI > 25 kg/m²), although European Americans were lowest, followed by Dominicans and Haitians; African-American and English-speaking Caribbean women fell into the obese range, even when background variables were controlled. Knowledge of obesity as a breast cancer risk factor was also poor across groups, but Dominicans and Haitians had the lowest scores on knowledge. Importantly, knowledge was not associated with BMI in the overall sample, even when controlling for demographics and ethnicity, although logistic regressions comparing normal weight women with overweight and obese groupings suggested some knowledge of breast cancer risk in the overweight, but not the obese, group.

Discussion: The findings remind health professionals of the need to consider more specific ethnic groupings than has hitherto been the case, as well as consider how ethnic and cultural variables may influence perceptions of obesity and its relation to cancer risk.

Introduction

The fact that a significant proportion of breast cancers could be prevented through avoiding certain behaviors is well accepted. Although there are a large number of risk factors that are not readily amenable to intervention (e.g., age, late onset of first pregnancy, early menarche, and late menopause), lifestyle patterns that produce excess body mass are avoidable. Excess body mass is one of the most readily preventable risk factors for breast cancer (1,2), and estimates suggest that excess body mass alone can directly explain at least 10% of female cancers (3), as well as being relevant to secondary detection and prevention (4,5,6,7).

Disagreement remains regarding which elements of excess mass are most directly associated with breast cancer risk and for whom. Some research has suggested that excess central obesity or increased upper body fat is the most useful operationalization of excess mass in predicting increased risk (8,9,10), although other epidemiological studies have failed to find this relation (11,12). Other researchers argue that only BMI consistently shows a relation with breast cancer risk, at least among postmenopausal women (3,13). The distinction between the risk posed by excess mass before and after menopause is an important one (14), with greater BMI producing increased breast cancer risk among postmenopausal women (3,13) and decreased risk in premenopausal women (15). This study draws from a large community sample of women between 50 and 70 years of age and employs BMI as its measure of excess mass.

Irrespective of the precise anthropometric mechanisms linking excess mass to breast cancer, recent national data have suggested that obesity rates are increasing (16,17) and that upward of 105 million Americans are now overweight or obese (18). According to the definition of the NIH (19), >60% of the adult population are overweight (BMI ≥ 25 kg/m²), and nearly 50 million adults are clinically obese (BMI ≥ 30 kg/m²). Indeed, the trend of increasing obesity is moving away from the Healthy People 2010 goal of an obesity prevalence of 15% (20). This trend is evident among women from all ethnic groups, including African Americans (21). In 1991, 19.3% of African Americans were obese. By 1988, nearly 27% met the BMI criterion for obesity (21). In terms of breast cancer, research has shown that body mass may account for nearly 30% of the later-stage breast cancer diagnosis risk faced by African-American women (22), although whether this is a primary prevention or a detection issue is unclear. Culturally, African-American women have more trouble losing weight or maintaining weight once lost (23), placing them at increased risk for obesity-related breast cancers.

Complicating this picture is the fact that, although BMI differences between women of European and African descent are well documented (24,25), there is virtually no descriptive information regarding the body mass characteristics of subpopulations of African- and Caribbean-American women in the United States. Few studies have included sufficient numbers of African-American women (26), much less considered the uniqueness of Caribbean subpopulations (27). Instead, minority women are subsumed within the “black” and “Hispanic” rubrics (28,29), the implicit assumption being that there is no variation in risk factors within these groups. However, ethnic categories such as “African American” are arbitrary, in this case encompassing U.S.-born African Americans, immigrants from Africa, and individuals of African descent from the West Indies (30,31,32,33,34).

Research from the United Kingdom has suggested that Caribbean women are at high risk for weight-related health problems, with an average BMI of 28.0 kg/m² (35). Another study of 314 Barbadian and 487 Dominican women residing in the Caribbean reported BMIs of 26.8 kg/m² for Barbadian women and 28.5 kg/m² for Dominican women between 20 and 55 years of age (36). On the basis of waist-hip measurements, another study of 485 patients attending a diabetes referral clinic in Jamaica has suggested a 90% prevalence of obesity among Jamaican women (37). With few exceptions, however, researchers do not yet have the data needed to evaluate the obesity-related disease risk to which the large and growing groups of immigrant women from the Caribbean countries of Jamaica, Haiti, the Dominican Republic, and the islands of the English-speaking Caribbean may be exposed. In what seems to be the single available large-scale community survey that examines African subpopulations within the continental United States, Consedine et al. (31) found that African Americans had a higher BMI than either European Americans or English-speaking Caribbean women (28.54, 26.57, and 27.52 kg/m², respectively).

In addition to the need for descriptive data, research that considers why obesity is increasing at a time when adults should, in theory, be increasingly aware of the risk that obesity poses to health is also needed. One possibility has been that, because of ethnic differences in preferred body types (38), the meaning of body size (39), or differences in education, minority women do not realize the danger that obesity poses. African-American women are less aware of dietary practices that might reduce breast cancer, even when prompted (40), and there is a strong association between poorer education and less knowledge (40,41). How differing levels of knowledge relate to BMIs among well-defined minority groups is unclear at this time.

Other data in psychosocial oncology have, however, established a link between knowledge about a health behavior and the frequency of that behavior. In breast cancer screening research, poorer knowledge has been associated with poorer screening (4,5,6,7). To our knowledge, however, there are no studies documenting awareness of obesity as a risk factor for breast cancer and its relation to BMI in subpopulations of African- and Caribbean-American women, although such research would help identify the subgroups of women at particular risk for obesity-related breast cancers.

Given the underdeveloped state of the literature, this study was framed as an exploratory one in which we aimed to document BMI scores, knowledge regarding the role of obesity in breast cancer, and the associations or disjunctions between these two variables in women from six ethnic groups.

Research Methods and Procedures

Participants

The participants in this study were 1364 older adult women (age, 59.2 ± 6.5 years) drawn from a stratified cluster-sampling plan. Based on the 1990 Census files, data on census blocks were gathered from the Household Income and Race Summary Tape File 3A. Blocks were stratified by ethnic group and income (high, medium, and low). Random selection without replacement was used to choose samples of block groups from each stratum. Interviewers were sent to conduct interviews with respondents who lived within the selected blocks to participate in a study of “women's health.” They were paid $25.00 for their participation. Because of our particular interest in examining possible differences within diverse subpopulations of African and Caribbean Americans, we differentiated between U.S.-born African Americans and Caribbean immigrant groups who were from (1) the English-speaking islands (i.e., Barbados, Trinidad and Tobago, Jamaica), (2) the Spanish-speaking territory of the Dominican Republic, and (3) the Creole-speaking territory of Haiti. U.S.-born European Americans served as a contrast group, with immigrants from Russia, the Ukraine, and Belarus serving as a white immigrant control group. These groups, as Eastern Slavs, are ethnically similar (42,43) and, for this study, were combined to yield a sample size of sufficient magnitude with which to compare them to the other African- and European-American groups.

Procedures

All measures comprising the research protocol were piloted on samples of women from the different ethnic groups to ensure adequate levels of comprehension. The protocol was translated into Creole and Spanish for the Haitian and Dominican samples, respectively, and, following standard ethnographic procedures, back-translated into English to ensure protocol comparability across groups. The study and its measures were reviewed by and approved of by the Long Island University Institutional Review Board before the study commenced, and approval for subsequent years was obtained. Respondents were interviewed by interviewers from their same racial and language group.

Measures

Demographics Questionnaire

Respondents provided information on age, ethnicity, household income, level of education, and marital status.

Obesity

In accordance with NIH and WHO guidelines, as well as research among postmenopausal women (3,13), we used BMI as our measure of obesity. Measures of height and weight were taken and combined by dividing weight in kilograms by height in meters squared to produce a BMI. Within current classifications, being “overweight” is indicated by a BMI of 25 to <30 kg/m² and “obese” by a BMI ≥30 kg/m² (19).

Awareness of Obesity as a Breast Cancer Risk Factor

A single item asked women to indicate how likely they thought it was that “being overweight” was a risk factor for breast cancer. Respondents used a rating scale with values ranging from 1 (extremely unlikely) to 6 (extremely likely). The variable was normally distributed.

Results

Preliminary Analyses

We first present demographic characteristics of the sample, means and SD of BMI, and knowledge of obesity as a risk factor for breast cancer broken down by ethnic group and a table of intercorrelations among the study variables. These are followed by several models depicting patterns of relationship among the variables.

Ethnic Group Differences on Demographics, BMI, and Knowledge

Table 1 presents the means and SD of the demographics of the sample, BMI, and accuracy of knowledge that obesity represents a risk factor for breast cancer, broken down by ethnic group. As indicated, Haitians and Eastern Europeans were older than other groups, European Americans had the highest income, Haitians and Dominicans had the lowest education, and a higher proportion of African-American vs. other women were single. The data on BMI indicated that the means for all groups placed them in the “overweight” category by national guidelines (19), but that the mean score for African Americans fell in the obese range (BMI ≥ 30 kg/m²) and that of the English-Caribbean women was borderline obese. European-American women had the lowest scores (significantly lower than African Americans and English Caribbeans). In terms of accuracy of knowledge about obesity being a risk factor for breast cancer, scores from 1 to 3 indicated that women thought that being overweight was moderately unlikely (3) to extremely unlikely (1) to be a risk factor; as such, our data indicated that all ethnic groups seriously underestimated the risk that excess mass poses for breast cancer.

Table 1.  Background variables, BMI, and accuracy of knowledge that being overweight is a risk factor for breast cancer by ethnic group (N = 1364)

	Ethnic group
Variable	African American (N = 295)	English Caribbean (N = 299)	Haitian (N = 305)	Dominican (N = 160)	Eastern European (N = 151)	European American (N = 154)	F value	Post hoc comparisons
* Marital status dummy coded such that 0 = single and 1 = married. † Participants rated the extent to which they considered being overweight as a risk factor for breast cancer on a scale of 1 = extremely unlikely to 6 = extremely likely. Lower scores indicate poorer knowledge. ‡ p < .01; AA, African American; EC, English-speaking Caribbean; H, Haitian; D, Dominican; EE, Eastern European; EA, U.S.-born European American.
Age	58.9 (6.2)	58.4 (7.0)	60.4 (6.5)	58.2 (6.1)	60.8 (6.1)	59.4 (6.5)	5.7‡	H, EE > AA, EC, D
Income	30.0 (26.5)	34.9 (20.3)	22.5 (15.5)	25.3 (14.5)	30.1 (22.7)	46.3 (34.6)	26.5‡	EA > all; EC > H, D; EE > H
Education	13.5 (12.1)	13.3 (8.3)	7.4 (8.1)	7.2 (8.2)	16.2 (13.0)	14.3 (13.1)	194.4‡	EE > EA > AA, EC > H, D
Marital status*	0.21 (0.41)	0.31 (0.47)	0.37 (0.48)	0.52 (0.50)	0.58 (0.49)	0.42 (0.50)	17.6‡	EA, EE, D, H > AA
BMI	31.0 (6.7)	29.7 (4.5)	28.7 (5.1)	28.5 (5.0)	29.3 (6.3)	27.6 (5.8)	9.9‡	AA > H, D, EA; EC > EA
Knowledge of risk†	2.8 (1.4)	2.7 (1.3)	1.9 (1.3)	1.9 (1.4)	2.9 (1.4)	3.1 (1.3)	31.3‡	D, H < AA, EC, EE, EA; EC < EA

Intercorrelations among the Study Variables

Table 2 presents the intercorrelations among BMI, knowledge scores, ethnicity, and the other demographic variables. As indicated, BMI was positively associated with being African American and negatively associated with being Dominican and Haitian (vs. European American). Knowledge of obesity as a risk factor for breast cancer was negatively associated with being Dominican and Haitian and positively associated with being English Caribbean, African American, and Eastern European. Of note, however, knowledge of obesity as a risk factor was not associated with BMI in zero-order relations. We next looked at critical cut-points for overweight and obesity.

Table 2.  Intercorrelations among the study variables: BMI, demographics, knowledge of obesity as a risk factor for breast cancer, and ethnicity

Variables	BMI	Age	Income	Education	Marital status	Knowledge	D	H	EC	AA	EE
* p < 0.05; † p < 0.01. AA, African American; EC, English-speaking Caribbean; H, Haitian; D, Dominican; EE, Eastern European.
BMI		−0.02	−0.04	−0.08*	0.03	0.03	−0.06*	−0.06*	0.03	0.15†	−0.00
Age			−0.21†	−0.10†	−0.08*	−0.01	−0.06*	0.09†	−0.07†	−0.03	0.08*
Income				0.36†	0.25†	0.12†	−0.08†	−0.19†	0.09†	−0.02	−0.01
Education					0.11†	0.20†	−0.18†	−0.26†	−0.07†	−0.02	0.45†
Marital status						−0.03	0.11†	0.00	−0.06*	−0.17†	0.16†
Knowledge							−0.14†	−0.24†	0.06*	0.10†	0.10†
Dominican								−0.20†	−0.19†	−0.19†	−0.13†
Haitian									−0.28†	−0.28†	−0.19†
English-Caribbean										−0.28†	−0.19†
African American											−0.19†
Eastern European

Proportions of Each Ethnic Group in Normal, Overweight, and Obese Groups

To present the data on body mass in a more clinically relevant manner—by weight groupings—Table 3 presents the proportion of each ethnic group that was normal weight (BMI < 25 kg/m²), overweight (BMI ≥25 but <30 kg/m²), and clinically obese (BMI ≥ 30 kg/m²). As indicated, European Americans had the highest proportion of normal weight women, Dominicans had the highest proportion of overweight women, and African American and African Caribbeans had the highest proportions of clinically obese women.

Analytic Strategy for Models Depicting the Pattern of Interrelations among the Variables

To assess the relative contribution of risk factors for overweight and obesity, we first conducted a discriminant function analysis with the three weight groups (normal, overweight, and obese) as the grouping variable; this identified two functions we interpreted as obesity and overweight. Because there were clearly two differentiated and significant functions and because obesity is considered a risk factor for breast cancer, while being overweight is less clear of a risk, we subsequently conducted two logistic regressions to separately assess the contribution of demographic, knowledge, and ethnicity variables to the prediction of obesity and overweight to obtain odds ratios for the predictor variables. Finally, we regressed the same set of variables on BMI to take advantage of the full variability of the BMI data.

Discriminant Function Analysis

A discriminant function analysis was performed using demographic variables, knowledge of risk of breast cancer, and ethnicity as predictors of membership in the three weight groups (normal weight, overweight, and obese). The overall Wilks’ Λ was significant, Λ = 0.95; p < 0.01; χ² (20, N = 1364) = 74.02; p < 0.01, indicating that, collectively, the predictors differentiated among the three weight groups. After removal of the first function, there was still a strong association between groups and predictors, χ² (9, N = 1364) = 20.72 (p < 0.01). The two discriminant functions accounted for 72.3% and 27.7%, respectively, of the between-group variability.

Table 4 presents the standardized within-group correlations between the predictors and the two discriminant functions. The first discriminant function included being married, being African American and English Caribbean (vs. European American), and lower income. The second function included higher income, greater knowledge of obesity as a risk factor for breast cancer, younger age, and being Dominican or Haitian vs. European American and being European American vs. Eastern European. The means on the discriminant function helped interpret these two functions. The means for the first function were smallest for the normal weight group and greatest for the obese group (0.21 vs. −0.33). The means for the second function were greatest for the overweight group and lowest for the normal weight group (0.15 vs. −0.12, respectively). Thus, it seems that obesity vs. normal weight is predicted by ethnicity (African American, English Caribbean), low income, and being married, whereas being overweight vs. normal weight is predicted by higher income, greater knowledge of breast cancer risk, and being Dominican and Haitian.

Table 4.  Discriminant function analysis: standardized within-group correlations between the predictors of weight groups (normal, overweight, obese) and the discriminant functions (N = 1364)

	Correlation coefficients with discriminant function
Variable	Function 1	Function 2
* Largest absolute correlation between each variable and each of the two discriminant functions.
African American	0.43*	−0.30
English Caribbean	0.42*	0.33
Education	−0.36*	0.16
Marital status	0.17*	−0.10
Income	−0.20	0.44*
Knowledge of risk	0.12	0.35*
Dominican	−0.17	0.20*
Age	0.03	−0.19*
Haitian	−0.17	−0.18*
Eastern European	−0.15	−0.14*

Logistic Regressions

We next conducted logistic regressions to assess the predictive value of the demographic, knowledge, and ethnicity variables separately in the case of the two overweight groups vis-à-vis the normal weight group. This yielded odds ratios for the important predictor variables. Tables 5 and 6 display the findings. As indicated in Table 5, all five ethnic groups were significantly more overweight than European Americans, with odds ratios ranging from 1.93 in the case of Eastern Europeans to 3.47 in the case of English Caribbeans; curiously, knowledge of overweight as a risk factor for breast cancer was positively associated with being overweight. Table 6, comparing normal weight women with obese women, indicated that all ethnic groups except the Dominican group were significantly more likely to be obese vs. normal weight than European-American women, with odds ratios ranging from 1.87 in the case of Haitians to 3.82 in the case of English Caribbeans. In this regression, knowledge was not associated with being obese.

Table 5.  Logistic regression analysis of BMI as a function of demographic variables, awareness of obesity as a risk factor, and ethnicity: normal weight (34.3%) vs. overweight (65.7%)

Variables	Odds ratio	95% Confidence interval	p
N = 845. Marital status is coded so that married = 1 and single = 0 (never married, widowed, divorced, or separated). Ethnicity is dummy-coded with European Americans as the reference group (coded 0) and African Americans, English-speaking Caribbeans, Haitians, Dominicans, and Eastern Europeans coded as 1. Knowledge: extent to which respondent rated the likelihood that being overweight was a risk factor for breast cancer on a scale of 1 (extremely unlikely) to 6 (extremely likely).
Age	1.00	0.98 to 1.03	0.82
Education	0.88	0.60 to 1.30	0.52
Income	1.00	1.00 to 1.00	0.23
Marital status	1.12	0.81 to 1.55	0.50
Knowledge of risk	1.15	1.03 to 1.29	0.02
Ethnicity
African American	2.47	1.47 to 4.14	0.001
English Caribbean	3.47	2.04 to 5.88	0.0001
Haitian	2.41	1.42 to 4.09	0.001
Dominican	2.38	1.31 to 4.33	0.005
Eastern European	1.93	1.06 to 3.52	0.03

Table 6.  Logistic regression analysis of BMI as a function of demographic/structural variables, awareness of obesity as a risk factor, and ethnicity: normal weight (35.8%) vs. obese (64.2%)

Variables	Odds ratio	95% Confidence interval	p
N = 809. Marital status is coded so that married = 1 and single = 0 (never married, widowed, divorced, or separated). Ethnicity is dummy coded with European Americans as the reference group (coded 0) and African Americans, English-speaking Caribbeans, Haitians, Dominicans, and Eastern Europeans coded as 1. Knowledge: extent to which respondent rated the likelihood of being overweight as a risk factor for breast cancer on a scale of 1 (extremely unlikely) to 6 (extremely likely).
Age	0.99	0.97 to 1.02	0.88
Education	0.68	0.45 to 1.03	0.07
Income	1.00	1.00 to 1.00	0.56
Marital status	1.47	1.05 to 2.06	0.02
Knowledge of risk	1.07	0.96 to 1.20	0.23
Ethnicity
African American	3.65	2.12 to 6.26	0.0001
English Caribbean	3.82	2.20 to 6.63	0.0001
Haitian	1.87	1.07 to 3.29	0.03
Dominican	1.66	0.88 to 3.11	0.12
Eastern European	2.41	1.29 to 4.52	0.01

Multiple Regression with BMI as the Outcome Variable

Table 7 displays the β weights, SE, and p values for the regression of demographic variables, knowledge, and ethnicity on BMI. As indicated, lower education, being married vs. single, and being African American, English Caribbean, and Eastern European were associated with increasing levels of BMI. As in the zero-order relations, awareness of obesity as a risk factor for breast cancer did not make a significant independent contribution to BMI after controlling for demographics and ethnicity.

Table 7.  Multiple regression of demographic variables, knowledge, and ethnicity on BMI (N = 1364)

Variables	ß	SE	p
Marital status is coded so that married = 1 and single = 0 (never married, widowed, divorced, or separated). Ethnicity is dummy coded with European Americans as the reference group (coded 0) and African Americans, English-speaking Caribbeans, Haitians, Dominicans, and Eastern Europeans coded as 1. Knowledge: extent to which respondent rated the likelihood of being overweight as a risk factor for breast cancer, on a scale of 1 (extremely unlikely) to 6 (extremely likely). R = 0.22, R^2 = 0.05, F (10, 1353) = 7.18, p < 0.0001.
Age	−0.02	0.02	0.38
Education	−1.28	0.41	0.002
Income	0.00	0.00	0.26
Marital status	0.99	0.33	0.003
Knowledge of risk	0.01	0.11	0.38
Ethnicity
African American	3.10	0.57	0.0001
English Caribbean	1.69	0.57	0.003
Haitian	0.50	0.60	0.41
Dominican	0.03	0.67	0.96
Eastern European	1.89	0.67	0.005

Discussion

In view of the growing problem posed by obesity among U.S. women (16,17), this study sought to determine differences in the degree to which women from six well-defined ethnic groups were overweight or obese, as well as to explore the association between women's awareness of obesity as a breast cancer risk factor and their actual BMI. Consistent with national trends (National Center for Health Statistics 2002), our study of 1364 community-dwelling women between the ages of 50 and 70 years showed that, on average, all women, irrespective of ethnic group, had BMIs that have been associated with a significant increase in breast cancer risk in women past menopause (3,13,44). At 27.6 kg/m², the lowest mean BMI in our sample—that of European-American women—was nonetheless in the clinically overweight range (19).

In addition, while researchers have historically ignored immigrant subpopulations of African- and Caribbean-American women in the United States (27,28,29,30,31), our data suggest that this process must not be allowed to continue. When operationalized precisely, there were significant differences across six ethnic groups. U.S.-born African-American women had a significantly greater average BMI (31.0 kg/m²) than women from all other groups, except women of English-Caribbean heritage. Women from the English-speaking Caribbean (29.7 kg/m²), Haiti (28.7 kg/m²), the Dominican Republic (28.5 kg/m²), and Eastern Europe (29.3 kg/m²) did not differ from one another, although women from the English-speaking Caribbean had significantly higher BMIs than European-American women.

These data, some of the first to describe American samples of Caribbean women, suggest that they may have greater BMIs than has been implied by previous studies. The BMI of our sample of English-speaking Caribbean women (29.7 kg/m²) was slightly greater than the BMI of 28 kg/m² reported for Caribbean women in the United Kingdom by Cruikshank et al. (35) and that of 26.8 kg/m² for the 314 women from Barbados described by Tull et al. (36), as well as that reported for 435 U.S.-dwelling, English-speaking Caribbean women by Consedine et al. (31) (27.5 kg/m²). The BMI of our sample of 160 Dominican women (28.5 kg/m²) was identical to the value reported in a study of 487 Dominican women still residing in the Caribbean (36) and, at 32.5% obese (see Table 2), similar to the 40% obesity rate reported for Dominican women in another study (45).

To our knowledge there have been no previously reported studies of BMI among U.S.-dwelling Eastern-European or Haitian women. However, these women, along with two other minority groups, were more likely than European-American women to be obese than within the normal weight range, even in analyses controlling for background demographic variables including age, education, income, and marital status. Logistic regressions controlling for these variables showed minority groups to be between 1.93 (Eastern European) and 3.47 (English Caribbean) times as likely to be overweight vs. normal weight than European-American women, and 1.87 (Haitian) and 3.82 (English Caribbean) times as likely to be obese vs. normal weight than European-American women.

In general, the BMIs in our samples of women were slightly higher than those reported for the comparable ethnic groups in previous research, particularly where the comparison sample was of non-U.S. origin. This may be because our sample was older or more impoverished than those previously reported or may reflect the influence of the American diet and issues associated with immigration, acculturation, and adjustment to the lack of traditional food sources. Obesity rates tend to be lower in developing countries (46), and there may be difficulties in providing health information in a manner that is sufficiently sensitive to groups new to the majority culture (47).

In addition to documenting BMI among understudied and at-risk groups of minority women, this study provided some preliminary data that document awareness of the link between obesity and breast cancer among women from these same groups. Women rated the extent to which they considered being overweight or obese as a risk factor for breast cancer on a scale where 1 = extremely unlikely and 6 = extremely likely. As indicated in Table 1, our data were clear in suggesting that women from all ethnic groups continue to seriously underestimate the risk that obesity or excess weight poses in terms of breast cancer risk.

Other analyses suggested no association between the two variables. In both the zero-order correlations and in a multiple regression controlling for demographics and ethnicity, with BMI as the dependent variable, there was no significant association between women's weight and their knowledge of obesity as a risk for breast cancer. However, discriminant function analyses and separate logistic regressions comparing obese and overweight women with normal weight women revealed a more differentiated pattern. These analyses suggested that some knowledge of breast cancer risk was present in the overweight groups but absent in the obese groups. Although the dietary habits of American women and the population at large put them at risk for becoming overweight, especially in later adulthood, and although we have become more overweight as a nation over the past two decades, some awareness of obesity as a risk factor for breast cancer and other cancers may be helping to keep overweight persons from going over the line toward obesity; the data from this study suggest that this may be true at least for certain groups of ethnic women living in urban environments.

Ethnic differences in education have been one major explanation for ethnic differences in knowledge about both primary preventive and risk behaviors (40,41), as well as ethnic differences in knowledge regarding secondary preventive behaviors, such as screening (48,49,50). There are, however, at least two good reasons to consider factors other than education. First, prior research has shown that ethnic differences in behavior may persist even when background variables are controlled (51,52). Second, our data simply do not support such a conclusion. As seen in Table 1, Eastern-European women were better educated than the European-American majority, who, in turn, reported greater education than African-American women. There were, however, no differences in the awareness of obesity as a risk factor across these three groups of women. Equally, African-American and English-speaking Caribbean women did not differ in terms of education, but their awareness of obesity as a risk factor for breast cancer did. Finally, we noted that ethnic differences in BMI remained in both multiple and logistic regressions with education, as well as other demographic factors, controlled.

The above findings immediately suggest two things. First, it is clear that awareness of obesity as a risk factor is not enough to motivate the lifestyle changes needed to alter it. African-American women, for example, remain clinically obese (on average) despite better-than-average awareness of how dangerous obesity is in terms of breast cancer. Second, such a finding hints at a clear need to expand the context in which we consider dietary behaviors and breast cancer knowledge, awareness, and risk perceptions. A growing literature documents dietary and eating pattern differences among ethnic groups, including Caribbean subpopulations (53,54,55,56). Other literature has suggested that cancer fatalism or the belief that nothing can be done to prevent or cure cancer may be particularly prevalent among African Americans (57,58), although these data have yet to consider ethnic subpopulations. One can readily imagine a situation in which cancer fatalism leads to continued engagement in risk behaviors simply because the individual does not believe that changing the behavior, in this case, weight, will make a difference.

Conclusion

This study documented BMI scores and the awareness of obesity as a risk factor for breast cancer among six ethnic groups of women. We found that the BMIs of the women from all six groups put the “average woman” in the clinically overweight range. Given that excess body mass is thought to be the most readily preventable risk factor for breast cancer (1,2), as well as other conditions such as diabetes and heart disease (59), and may explain a significant amount of breast cancers (3), these findings continue to describe the extent of the obesity epidemic in the United States. Also documented were ethnic differences in the extent to which women from distinct ethnic groups were overweight. Of particular note are the data showing differences among subpopulations of African- and Caribbean-American women. Ethnic differences in BMI were not reliably associated with ethnic differences in the awareness of obesity as a risk factor, suggesting that researchers must begin to develop a means of conceptualizing ethnic background in a manner that identifies the underlying elements of ethnic group membership that place women at risk. In the future, it will be these variables—whether structural, demographic, or psychological—that will be critical to the development of effective intervention strategies.

Acknowledgment

This research was supported by grants from the National Institute on Aging (KO7 AG00921), the National Institutes of General Medical Science (2SO6 GM54650), and the National Cancer Institute (1P20 CA 91372).