Perceived and Objective Breast Cancer Risk Assessment in Chilean Women Living in an Underserved Area

The burden of breast cancer in Chile has grown significantly. Currently, it represents the most frequently diagnosed cancer among women in Chile (1). The risk of developing breast cancer for Chilean women, approximately 3% to 3.5% (2, 3), is less than the risk observed for all United States (US) women combined (12.3%) and US Latinas (9.7%; ref. 4). Yet, the mortality-to-incidence ratio (MIR), a measure that approximates case-fatality rates, is approximately 29.8% in Chile, far greater than the 18.8% MIR reported for the US (5). Although new health policies directed at combating breast cancer and improving access to mammography were incorporated in 2005, increasing mammography access for women who are 50 years of age and older (6), data on mammography rates in some pilot areas of Chile suggested rates following program implementation were still as low 12% (7). Consequently, while increasing mammography access was an important step for Chile, other factors may also be important for breast cancer prevention efforts.

Perceived risk of disease, theoretically, is an important motivator of health behaviors to prevent, detect, and control cancer (8), as it is assumed that aligning an individual's perceived and actual risk of developing breast cancer, leading to a more realistic perceived risk, will motivate such behavior (8, 9). Although limited, the evidence among Latinas suggests perceived breast cancer risk may not be associated with screening behavior (10, 11). Nevertheless, it is important to better understand risk perception among this population, as the motivation for protective behaviors may be in the underlying meaning of risk appraisal (8).

Given the gap in literature about breast cancer risk perception in South American women, the goal of our study was to explore breast cancer risk perceptions and accuracy of perceived breast cancer risk among underserved Chilean women, age 50 to 70 years, in Santiago, Chile. We were interested in assessing risk perception using absolute risk, comparative risk, and numeric risk approaches, as well as the factors previously shown to be associated with breast cancer risk perception (9, 10, 12, 13). The information provided in this study may contribute to better inform health policy makers on how to deliver preventive messages to improve breast cancer prevention.

This study was conducted in El Castillo Oriente, Santiago, Chile, where the majority of residents were registered to receive free health care services at the university primary care clinic, including free screening mammography for women 50 to 70 years old (14). Women 50 to 70 years of age registered at the university primary clinic, who had not received a mammogram within the last 2 years, were eligible. Women with a previous breast cancer diagnosis were excluded. Using electronic medical records, eligible women were randomly selected and invited to participate. Among the 540 women invited, 12 refused to participate and 28 had a change of address/no contact, for a final sample of 500 women. Women who agreed to participate were asked to provide informed consent; consenting participants were interviewed in-person in the clinic.

Trained interviewers administered a structured questionnaire, including questions on sociodemographic factors, health behaviors, reproductive health, health care practices, and breast cancer prevention beliefs, attitudes, and practices. The questionnaire was developed from previous research and written at a low literacy level (15).

Participant information was collected on race/ethnicity, education, marital status, family monthly income, and occupation. We measured breast cancer knowledge using 5 items, on a 5-point Likert scale, related to breast cancer etiology, risk factors, screening, and prognosis (15, 16). Knowledge score was calculated as the percentage of correct responses.

The National Cancer Institute's Breast Cancer Risk Assessment Tool (BCRAT), also known as the Gail Model (17), was used to calculate participants' 5-year and lifetime risk of developing invasive breast cancer. Among the BCRAT risk factors, only information on atypical hyperplasia was unavailable. Evidence suggests the BCRAT may underestimate absolute risk of invasive breast cancer in US Latinas (18).

Perceived risk of breast cancer was measured using 3 items: absolute risk—“how would you classify your probability of developing breast cancer?” (5-point Likert scale: 1-“very low” to 5-“very high”); comparative risk—“in general, how would you classify your risk of developing breast cancer compared with a woman who is about your age?” (7-point Likert scale: 1-“much lower,” to 7-“much higher”); and numeric risk—“what level of risk do you think you have for developing breast cancer during your lifetime?” [Scale from 0% (no possibility of developing breast cancer) to 100% (I will develop breast cancer either way)] (15). We categorized responses as follows: absolute risk (low: “very low/low; average: “not low or high,” and high: “high/very high”) and comparative risk (lower: “much lower/lower/a little lower;” equal: “about the same” and higher: “much higher/higher/a little higher”).

Risk comprehension was measured using participants' perceived numeric risk and BCRAT estimates (10, 19, 20). Risk comprehension was calculated as the difference (d) between participants' perceived numeric risk and BCRAT lifetime risk (d = perceived numeric risk − BCRAT risk). Risk comprehension categories were defined as accurate: d≤ ± 10, underestimate: −10 < d, and overestimate: d > 10 (10, 19, 20).

Absolute risk, comparative risk, and BCRAT 5-year risk estimates were used to measure accuracy of perceived risk. Perceived absolute risk categories were defined as: accurate, if women with BCRAT 5-year risk 1.67 or more responded as “high” or women with BCRAT 5-year risk less than 1.67 responded as “low” or “average”; underestimate, women with BCRAT 5-year risk 1.67 or more selected “low” or “average”; and overestimate, if women with BCRAT 5-year risk less than 1.67 selected “high.” Categories for perceived comparative risk were defined similarly: accurate, if women with BCRAT 5-year risk 1.67 or more responded as “higher” or women with BCRAT 5-year risk less than 1.67 responded to as “lower” or “equal”; underestimate, if women with BCRAT 5-year risk 1.67 or more selected “lower” or “equal”; and overestimate, if women with BCRAT 5-year risk less than 1.67 selected “higher.”

Descriptive statistics were used to assess participants' sociodemographic, breast cancer knowledge, and BCRAT risk factors. Pearson's χ² and Fisher's exact tests were used to assess for associations between BCRAT risk category and perceived risk estimates. Multivariate logistic regression analyses were conducted to examine factors associated with risk comprehension and accuracy of perceived risk, including ethnicity/indigenity, education, age, and breast cancer knowledge. A significance level of P < 0.05 was used for all statistical analyses.

Participants' sociodemographic, breast cancer knowledge and BCRAT characteristics are shown in Table 1. The majority of participants were White or Mestizo with low education and low socioeconomic status. The mean age of participants was approximately 57 years old, with the majority giving first live birth at age less than 20 years, and having no family history of breast cancer or personal history of breast biopsy. Participants' mean BCRAT 5-year and lifetime risk estimates were approximately 0.8% and 5.0%, respectively.

Table 2 shows the distribution of perceived absolute and comparative risk, by BCRAT risk category. Women at high risk of breast cancer were significantly more likely than average risk women to consider themselves to be at high risk, on the basis of perceived absolute risk (P < 0.001). Similar results were observed for perceived comparative risk, although a greater proportion of women at average risk considered themselves to be at equal or higher risk of breast cancer relative to a woman of the same age. Among all participants, the mean perceived numeric risk of developing breast cancer in one's lifetime was 33.2% (SE = 1.3).

A greater proportion of average risk women accurately estimated their breast cancer risk on the basis of perceived absolute risk (86.3%) and comparative risk (77.5%) compared with high risk women (52.6% and 57.9%, respectively; Table 2). Two-thirds of participants (67%) overestimated their risk of developing breast cancer, on the basis of risk comprehension.

Table 3 shows results of multivariate regression analyses, which indicate that younger age and higher BCRAT 5-year risk were significantly associated with overestimation of perceived absolute risk [OR, 0.90; 95% confidence interval (CI), 0.84–0.96; P < 0.01 and OR, 6.95; 95% CI, 3.20–15.11; P < 0.01, respectively]. Higher breast cancer knowledge was significantly associated with lower odds of overestimating comparative risk [OR, 0.72; 95% CI, 0.57–0.91; (P < 0.01)]. None of the variables included in the multivariate regression were found to be were significantly associated with the risk comprehension.

To our knowledge, our study is among the first to evaluate perceived and objective breast cancer risk, as well as factors associated with overestimating breast cancer risk among women in South America. The study included women from different ethnicity groups and used a strong methodology applying 3 different risk perception measures and BCRAT lifetime risk rather than 5-year risk.

Our results indicate that 67% of Chilean women overestimated their lifetime risk of breast cancer, on the basis of risk comprehension estimates. These findings support the results of Graves and colleagues (10), who reported that the majority of Latinas (81%) overestimated their breast cancer risk. Consequently, as seen in other studies (20, 21), Latinas may have inaccurate perceptions of breast cancer risk, tending to overestimate, when asked to put a numerical value on their lifetime risk of developing breast cancer.

However, previous studies have also shown that individuals' perceived risk varies depending on the method used and corresponding response format (8, 22), suggesting participants may overestimate breast cancer risk when using numeric risk, yet are more accurate when using comparative or absolute risk (10, 22). Our study supports these findings, showing that while participants overestimated risk using a numerical approach, fewer women, overall, considered themselves to be at high or higher risk of developing breast cancer on the basis of absolute and comparative risk. Accordingly, using a numeric, open-ended scale to elicit perceived risk may present a difficult challenge to participants; furthermore, comparative and absolute perceived risk measures, often on the basis of categorical or ordinal responses, may allow individuals to place risk in a context of familiar and more intuitive events (22, 23). Therefore, comparative and absolute risk methods may better reflect women's interpretation of risk of developing breast cancer (24) and capture the accuracy of women's risk perception.

Prior studies assessing breast cancer risk perception suggest age, race/ethnicity, family history, knowledge, cancer worry, and education may be related to perceived breast cancer risk (9, 12). Among studies with Latinas, younger age (10, 13), higher breast cancer knowledge (12), and higher BCRAT 5-year risk scores or certain BCRAT risk factors (i.e., family history; refs. 12, 13) were significantly associated with overestimation or higher perceived risk. Our findings support and add to this literature indicating that younger age, increased breast cancer knowledge, and higher BCRAT risk scores were significantly associated with overestimation of breast cancer risk among Chilean women.

This study also has some limitations. First, our study was composed of underserved, Chilean women aged 50 to 70 years, who were registered at a primary care clinic; hence, our results are most generalizable to a similar population. Second, given that our sample included women who did not adhere with mammogram screening recommendations, their risk perception may differ from women who receive mammograms regularly. However, the limited evidence available in Chile shows that this represents a minority group. Third, we did not measure mammogram adherence and, therefore, were unable to assess how participants' risk perceptions were related to mammogram utilization.

In conclusion, this study shows that Chilean women living in an underserved community in Santiago may, as with many US Latinas, overestimate their risk of developing breast cancer. Importantly, the method used to measure perceived risk appears to affect women's responses and, therefore, accuracy of risk perception. While evidence on the impact of perceived risk on cancer prevention efforts among Latinas is limited, these findings improve our understanding of how Latina's appraise risk and factors associated with overestimating breast cancer risk. Further studies of how Latina's risk perception is associated with breast cancer prevention and control behaviors may lead to effective interventions that reduce the burden of breast cancer in this population.

No potential conflicts of interest were disclosed.

Conception and design: M.P. Banegas, K. Püschel, B. Thompson

Development of methodology: M.P. Banegas, K. Püschel, B. Thompson

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Püschel

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.P. Banegas, B. Thompson

Writing, review, and/or revision of the manuscript: M.P. Banegas, K. Püschel, J. Martinez, J.C. Anderson, B. Thompson

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.P. Banegas, J. Martinez, B. Thompson

Study supervision: M.P. Banegas, K. Püschel

This research was supported, in part, by the NIH, National Cancer Institute, Center for Population Health and Health Disparities (Grant number: 5 P50 CA148143) and the Fogarty International Center (Grant number: RO3 TW007900).