ORIGINAL PAPER

Perceived risk of breast cancer among Latinas attendingcommunity clinics: risk comprehension and relationshipwith mammography adherence

Kristi D. Graves Æ Elmer Huerta Æ Jennifer Cullen Æ Elizabeth Kaufman ÆVanessa Sheppard Æ George Luta Æ Claudine Isaacs Æ Marc D. Schwartz ÆJeanne Mandelblatt

Received: 23 October 2007 / Accepted: 8 July 2008 / Published online: 15 August 2008

� Springer Science+Business Media B.V. 2008

Abstract

Objective To describe breast cancer risk perceptions,

determine risk comprehension, and evaluate mammogra-

phy adherence among Latinas.

Methods Latina women age C35, primarily from Central

and South America, were recruited from community-based

clinics to complete in-person interviews (n = 450). Risk

comprehension was calculated as the difference between

numeric perceived risk and Gail risk score. Based on rec-

ommended guidelines from the year data were collected

(2002), mammography adherence was defined as having a

mammogram every one to two years for women C40 years

of age.

Results Breast cancer risk comprehension was low, as

81% of women overestimated their risk and only 6.9% of

women were high risk based on Gail risk scores. Greater

cancer worry and younger age were significantly associated

with greater perceived risk and risk overestimation. Of

women age eligible for mammography (n = 328), 29.0%

were non-adherent to screening guidelines. Adherence was

associated with older age, (OR = 2.99, 95% CI = 1.76–

5.09), having insurance (OR = 1.81, 95% CI = 1.03–

3.17), greater acculturation (OR = 1.18, 95% CI =

1.02–1.36), and higher breast cancer knowledge

(OR = 2.03, 95% CI = 1.21–3.40).

Conclusions While most Latinas over-estimated their

breast cancer risk, older age, having insurance, being more

acculturated, and having greater knowledge were associ-

ated with greater screening adherence in this Latino

population. Perceived risk, risk comprehension, and cancer

worry were not associated with adherence. In Latinas,

screening interventions should emphasize knowledge and

target education efforts at younger, uninsured, and less

acculturated mammography-eligible women.

Keywords Latina � Breast cancer � Risk perception �Risk comprehension � Mammography adherence �Acculturation

Introduction

The rapid growth and aging of the Latino population, along

with acculturation to US lifestyles that increase the risk of

breast cancer [1], will increase the absolute number of Latino

women (hereafter referred to as Latinas) at risk for breast

Supported by Grants U01CA86114, U01CA114593, K05CA96940

(JM), and K07CA131172 (KG) from the National Cancer Institute.

K. D. Graves (&) � E. Huerta � E. Kaufman � V. Sheppard �M. D. Schwartz � J. Mandelblatt

Department of Oncology, Cancer Control Program, Lombardi

Comprehensive Cancer Center, Georgetown University,

3300 Whitehaven Street, NW, Suite 4100, Washington,

DC 20007, USA

e-mail: kdg9@georgetown.edu

E. Huerta

Washington Hospital Center, Washington, DC 20010, USA

J. Cullen

American Legacy Foundation, Washington, DC 20036, USA

G. Luta

Department of Biostatistics, Bioinformatics, and

Biomathematics, Lombardi Comprehensive Cancer Center,

Georgetown University, Washington, DC 20007, USA

C. Isaacs

Clinical Breast Cancer Program, Lombardi Comprehensive

Cancer Center, Georgetown University, Washington, DC 20057,

USA

123

Cancer Causes Control (2008) 19:1373–1382

DOI 10.1007/s10552-008-9209-7

cancer over the coming decades. Although breast cancer

incidence rates remain lower among Latinas compared to

non-Hispanic Whites and African-American women [2],

Latinas are more likely to be diagnosed with advanced breast

cancer compared to non-Hispanic White women [3]. Despite

these trends, rates of mammography adherence remain

suboptimal in the Latina population [4–7].

There are several possible reasons for low screening

adherence in Latinas, including low breast cancer knowl-

edge, lack of insurance/access to care, and low education [5,

7–10]. Results have been equivocal for the impact of

acculturation (and English ability) on screening adherence

[8, 11–15]. These reasons for non-adherence map onto the

predisposing, enabling, and need factors described in the

Behavioral Model for Vulnerable Populations [16]. Spe-

cifically, the predisposing and enabling factors that likely

impact mammography behavior include sociodemograph-

ics, insurance, knowledge, access to care, and acculturation,

while need factors could include perceived breast cancer

risk, comprehension of risk, and breast cancer worry.

In other populations, perceptions of breast cancer risk

appear to impact screening behaviors [17, 18]. For instance,

research conducted with African American women indi-

cates a non-linear relationship between perceived risk and

screening adherence [19]. However, we know little about

the impact of perceived risk on screening in Latinas. We are

aware of only one study that has examined perceived risk

and mammography adherence in Latinas [10]. In that study,

perceived risk was measured as comparative risk (likeli-

hood of getting cancer compared to other people their age)

and was unrelated to mammography outcomes. Moreover,

the construct of breast cancer risk comprehension, repre-

senting how closely a woman’s perceived risk matches her

objective risk (based on family history, reproductive his-

tory, and prior biopsies) has only been explored with

predominately Caucasian samples [20–22]. To our knowl-

edge, other types of perceived risk (e.g., absolute risk,

numerical risk), risk comprehension, and cancer worry have

not been assessed in prior work with Latinas. Thus, con-

structs associated with perceived breast cancer risk remain

under explored in broad groups of Latinas, particularly as

these factors may relate to mammography adherence.

The purpose of this study is to describe the correlates of

perceived risk and comprehension of risk as well as to assess

the degrees of association between risk and a related risk

construct—cancer worry—on adherence to mammography

guidelines after considering covariates (e.g., predisposing

factors) in a sample of Central and South American Latinas

attending community health clinics. We hypothesized that

younger age and a more significant family history would be

associated with greater perceived risk [23]. We also hypoth-

esized that Latina women, similar to other women [20, 22],

would overestimate their actual risk of breast cancer. In

addition, we hypothesized that adherence to mammography

recommendations would be associated with higher perceived

risk for breast cancer, greater (more accurate) risk compre-

hension, and more cancer worry. Finally, we explored the

impact of level of acculturation on adherence to mammog-

raphy guidelines, hypothesizing that greater acculturation

would be associated with higher levels of adherence.

Materials and methods

Setting

In the Washington DC metropolitan area, Latinos make up

15% of the total population. Many Latinos in the area are

recently immigrated and typically come to the US from

Central and South American countries. The predominant

countries of origin for the DC Latino population are El Sal-

vador, Ecuador, Peru, and Columbia [24]. Data were

collected from women recruited from three community-

based clinics in the DC area that provide services at no cost to

primarily uninsured Latino individuals. The three commu-

nity clinics are non-federally funded and the majority of the

staff at the clinics is bilingual, English- and Spanish-speak-

ing. Non-physician staff at each clinic served as

interviewers. The clinics are part of the Latin American

Cancer Research Coalition, a National Cancer Institute

Community Network Program.

The present study was part of an IRB-approved larger

trial investigating an intervention to educate Latina patients

about a breast cancer prevention trial (Study of Tamoxifen

and Raloxifene; STAR trial) [24]. During the larger trial

women were screened for STAR eligibility and in a base-

line interview answered questions about willingness to

participate in clinical trials, use of screening, and their

perceptions of risk of breast cancer. The present study used

data from the baseline interviews of the larger trial.

Sample

The sample consisted of 450 Latinas who were 35 years of

age and older and were seen for a medical appointment at

one of the three clinics. More than half of patients served at

these clinics are women (Range: 55–83% of clinic popu-

lation). The average age of the women served at these

clinics is 57 years, and El Salvador is the country of origin

for the largest segment of clinic patients (Range: 42–51%

of clinic populations). In addition, most patients are

monolingual in Spanish and on average, have completed

eight years of education. The women served at the three

clinics are similar in terms of demographics, education,

country of origin, and language abilities to women at the

other clinics within the Latin American Cancer Research

Coalition [25].

1374 Cancer Causes Control (2008) 19:1373–1382

123

Of the 450 women who completed interviews, three

were missing data related to either actual and/or perceived

breast cancer risk. Thus, the final sample consisted of 447

women; for analyses related to mammogram adherence, we

included the 328 women who were 43 years or older, as

described below.

Recruitment and survey procedures

Using appointment logs and medical records, clinic staff

determined women’s eligibility prior to scheduled clinic

appointments. Women were eligible if they were 35 years of

age and older and attended an appointment at one of the three

clinics. Study staff approached all potentially eligible women

during their scheduled appointment for consent to participate.

A small number of women were missed during appointment

sessions due to administrative delays or other clinic flow

issues unrelated to the study questions. Of the eligible women

approached, 96% consented to participate. The IRB-approved

consent procedures and interviews were conducted in a pri-

vate area while women waited for their appointment. In-

person interviews were conducted in the language of the

patient’s choice using a computer-assisted format. Almost all

women elected to have their interview conducted in Spanish

(97.5%), and 95% agreed to the computer-assisted format.

The remaining 5% of responses were recorded on a hard copy

of the study questionnaire. The interview took *30 min to

complete. All interviewers were bilingual, completed exten-

sive training, and engaged in mock interviews.

Measures

We first piloted a brief version of the study interview in a

separate sample of 79 Latinas, assessing sociodemograph-

ics, acculturation, family history of cancer, medical history,

and perceived risk [26]. Women were able to understand

the questions and reported relatively high perceived risk

(e.g., 19% of pilot sample rated their risk for breast cancer

as ‘‘high’’).

Using our pilot results and following the factors outlined

in The Behavioral Model for Vulnerable Populations [16],

we assessed the following variables in the present study.

Predisposing and enabling factors

Sociodemographics

Information was collected on age, marital status, education,

insurance status, and employment. Variables were catego-

rized as follows: age (\50 years vs. C50 years), marital

status (married vs. not married), education (\High School

vs. CHigh School), insurance status (yes vs. no), and

employment (yes vs. no). We created a binary variable for

age (\50 years vs. C50 years) to account for the increase

in risk of breast cancer for women age 50 and over.

Acculturation

We assessed acculturation with three questions related to

language use [27]. Women were asked to identify the

language they (a) spoke at home, (b) thought in, and (c)

spoke outside the home with the following options: (1)

Only Spanish, (2) Spanish better than English, (3) Both

equally, (4) English better than Spanish, or (5) Only Eng-

lish. Level of acculturation was determined by summing

responses to these questions. Higher scores represent

higher levels of acculturation.

Cancer worry

We measured breast cancer worry with two questions used

in prior research [28, 29]. Specifically, we asked partici-

pants, ‘‘In the past month, how often have you thought

about your own chances of getting breast cancer?’’ and ‘‘In

the past month, how often have these thoughts affected

your mood?’’ Responses to both questions were on a Likert

scale (not at all or rarely, sometimes, often to a lot) and

were summed to create an overall cancer worry score.

Breast cancer knowledge

We used a 13-item scale to measure breast cancer knowl-

edge [24]. Response options were true, false, or unsure, and

questions assessed knowledge related to the etiology of

breast cancer, screening recommendations, population-

level rates of breast cancer, how age relates to risk for

breast cancer, and screening and treatment. Sample items

included ‘‘A hard blow to the breast may cause a woman to

get cancer later in life,’’ ‘‘Mammography can detect lumps

that can’t be felt,’’ and ‘‘If a woman gets regular mam-

mography, she does not need to do breast self exams or

have physical examinations.’’ Scores were calculated as

percent of correct responses, ranging from 0% to 100%.

Responses of unsure were coded as incorrect. In order to

account for a bimodal distribution of knowledge scores, we

created a binary breast cancer knowledge score with a

median split to represent low and high knowledge.

Need factors

Perceived risk

Perceived risk was assessed with three distinct items. First,

an absolute estimate of perceived risk was obtained

through responses on a 3-point Likert scale (‘‘not at all’’ to

‘‘definitely’’) to the question, ‘‘How likely do you think it is

Cancer Causes Control (2008) 19:1373–1382 1375

123

that you will develop breast cancer?’’ To assess compar-

ative risk, participants were asked ‘‘How do you think your

risk of dying from breast cancer compares to an average

women your own age?’’ on a 3-point Likert scale (‘‘lower

than average’’ to ‘‘higher than average’’) [30]. For numeric

risk, participants were asked to rate their likelihood of

developing breast cancer on a scale from 0 (definitely will

not get breast cancer) to 20 (definitely will get breast

cancer). The items were not combined into scales.

Medical history/Gail model risk

We used the Gail risk model to estimate objective risk [31],

with scores determined by age at menarche (e.g., \12,

12–13, and C14), biopsy history (calculated separately for

women \50 and women C50 by number of biopsies),

pregnancy history (age at first live birth), history of

atypical hyperplasia, number of first degree relatives

(mother, daughter, sister) with breast or ovarian cancer, and

current age. High (C1.7%) and low (\1.7%) objective risk

were categorized as the 5-year risk of disease, with high

risk scores being equal to or greater than the 5-year risk of

disease among the average 60-year old woman [31, 32].

Risk comprehension

Based on prior research, we calculated risk comprehension

as the difference between subjective numeric risk estimates

and objective Gail score percentage risk [20]. We converted

numeric risk and Gail scores to a common metric. Specifi-

cally, numeric risk was multiplied by five, so that the original

scale (0–20) became a 0–100 scale (Range = 5–100) and

5-year Gail model risk scores were multiplied by 10 to pro-

vide percentages between 1.0 and 100 (Range = 1.66–38.3).

We then subtracted objective risk from perceived numeric

risk. We categorized risk comprehension as accurate if the

difference between subjective and objective risk estimates

was B10 points in either direction, an underestimate if the

difference was [10 points below objective risk, and an

overestimate if[10 points above objective risk. Only 2% of

the sample (n = 9) underestimated risk, so the underesti-

mation and accurate risk categories were collapsed, as done

in prior research [22].

Health behavior outcomes

Adherence to mammography recommendations

Mammography use was classified as never, ever, and recent

(B2 years). Specifically, women were asked if they had

ever had a mammogram (yes/no), when they had their most

recent mammogram (within 1–2 years, [2 years but

\3 years ago, or [3 years ago), and whether they had a

mammogram during the two years before the most recent

mammogram (yes/no). For these analyses, we only inclu-

ded women age 43 or older (n = 328) to match guidelines

outlined by the National Cancer Institute for mammogra-

phy in 2002, the year in which the data were collected [33].

Women age 43 and older were considered adherent if they

reported having a recent mammogram (B2 years). Women

age 45 and older were considered adherent if they reported

having a recent mammogram (B2 years) and having a

mammogram prior to the most recent mammogram.

Analysis

After generating descriptive statistics to characterize the

sample, we conducted bivariate analyses using v2 tests, t tests,

and Pearson correlation coefficients to determine associations

between study variables and perceived risk, risk comprehen-

sion, and adherence to mammography guidelines.

Depending on the scale of each outcome variable—

continuous, or binary—and the scale of the independent

variable—categorical or continuous—we used ANOVAs,

linear regression, and logistic regression to identify vari-

ables that were associated with each outcome. Specifically,

we entered significant (p \ .05) bivariate predictors of our

outcomes (perceived risk, risk comprehension, and adher-

ence) as well as predictors associated with the

predisposing, enabling, and need factors from our con-

ceptual model, into the multivariate analyses. We used

logistic regression models for our statistical analyses

investigating predictors of risk comprehension and adher-

ence to mammography. The final models were selected

using backward elimination procedures. This approach

allowed us to control for relevant sociodemographics (e.g.,

education, insurance status, clinic site) while assessing the

effect of the main predictors. In subsequent sensitivity

analyses we evaluated the effect of potential correlation of

the outcomes within clinic by using related logistic

regression models based on Generalized Estimating

Equations (GEE) with an exchangeable working correla-

tion structure. The estimated within-clinic correlations

were practically null; and the overall results were not

changed. Thus, we elected to report the results from the

previously described logistic regression models.

Results

Sample characteristics

In the final sample of 447 women, the mean age was 50.5

(SD = 10.5) and the majority of women were married

(64%), had less than a high school education (73%), and

1376 Cancer Causes Control (2008) 19:1373–1382

123

were uninsured (68%). Fewer women had experienced a

breast biopsy (12%) or had a family history of breast

cancer (5%). The largest number of participants (n = 172;

38.2%) were born in El Salvador, and the mean level of

acculturation was low (see Table 1).

Rates of perceived risk and risk comprehension

Based on absolute estimates of perceived risk, one-quarter

(26.2%) of the sample considered themselves to be at high

risk for breast cancer, meaning that they rated their risk as

3, or ‘‘definitely’’ will get breast cancer. For comparative

risk estimates, 24.4% reported their risk of breast cancer as

higher than the average woman’s risk. The average

numeric estimate of perceived risk was 9.0 (SD = 4.8). All

estimates of risk (absolute, comparative, numeric) were

statistically significantly correlated (r’s ranging from .22 to

.38, all p’s \ .0001).

For objective risk, only 6.9% of the sample was high

risk as defined by the eligibility requirements for the STAR

Trial (i.e., 5-year Gail score of C1.7%). Based on differ-

ences between objective and subjective (i.e., perceived

numeric risk) estimates, more than three-quarters (81%,

n = 362) of the sample overestimated their breast cancer

risk.

Variables associated with perceived breast cancer risk

We examined bivariate associations among our three per-

ceived risk variables (absolute, comparative, and numeric)

and predisposing/enabling (age, education, race, insurance,

marital status, clinic site, acculturation, cancer worry, breast

cancer knowledge), and need (Gail score) factors. The three

perceived risk variables were all significantly associated

with cancer worry, with higher levels of cancer worry

associated with higher perceived risk. In addition, clinic site

Table 1 Characteristics

of Latinas by breast cancer

risk comprehension

a Difference between objective

risk (Gail score) and subjective

numeric perceived risk;b Includes nine participants

who underestimated their cancer

risk;

* p \ .05; ** p \ .01,

*** p \ .001

Characteristic Risk comprehensiona

Accurate riskb

(n = 85)

Overestimated risk

(n = 362)

Total sample

(n = 447)

Age, Mean (SD) ** 53.46 (10.5) 49.84 (10.3) 50.53 (10.5)

Range 35–81 years, Median = 49 years

\50 32 (7.2) 192 (42.9) 85 (19.0)

C50 53 (11.9) 170 (38.0) 362 (81.0)

Marital status

Married, n (%) 58 (68.2) 228 (63.0) 286 (64.0)

Unmarried, n (%) 27 (31.8) 134 (37.0) 161 (36.0)

Education

BHigh School, n (%) 60 (70.6) 266 (73.5) 326 (72.9)

[High School, n (%) 25 (29.4) 96 (26.5) 121 (27.1)

Insurance

Yes, n (%) 32 (37.7) 112 (30.9) 144 (32.2)

No, n (%) 53 (62.3) 250 (69.1) 303 (67.8)

Birthplace*

El Salvador, n (%) 24 (28.2) 148 (40.9) 172 (38.5)

Other, n (%) 61 (71.8) 214 (59.1) 275 (61.5)

Clinic

A, n (%) 31 (36.5) 118 (32.6) 149 (33.3)

B, n (%) 30 (35.3) 119 (32.9) 149 (33.3)

C, n (%) 24 (28.2) 125 (34.51) 149 (33.3)

Acculturation, Mean (SD) 5.57 (2.57) 5.25 (2.02) 5.31 (2.13)

Range 4–14, Median = 4.00

Cancer worry, Mean (SD)*** 2.68 (0.99) 3.12 (1.10) 3.08 (1.09)

Range 2–5, Median = 3.00

Breast Cancer Knowledge, Mean (SD) 67.81 (12.4) 66.27 (13.1) 66.56 (12.9)

Range: 0–100, Median = 64.30

High ([64.30) 41 (9.2) 44 (9.9) 85 (19.0)

Low (B64.30) 162 (36.2) 200 (44.7) 362 (81.0)

Cancer Causes Control (2008) 19:1373–1382 1377

123

was associated with absolute perceived risk (see Table 2).

No other variables were significantly associated with the

perceived risk measures. In adjusted ANOVA models for

our two categorical perceived risk outcomes (absolute and

comparative risk), clinic site (F = 6.3, p = .01; F = 4.7,

p = .03), and cancer worry (F = 11.8, p \ .001; F = 5.2,

p = .02) were independently associated with absolute risk

and comparative risk, respectively. In a multiple regression

model, cancer worry was independently associated with

numeric perceived risk (t = 5.3, p \ .001; b = 0.3), with

greater worry associated with greater perceived risk.

Variables associated with risk comprehension

Overall, 85 participants held accurate estimates of breast

cancer risk and 362 overestimated risk. Risk overestima-

tion was associated with being born in El Salvador,

younger age, and higher breast cancer worry (Table 2). No

other variables were associated with risk comprehension.

In the logistic regression model with backward elimination

of variables, older age (OR = 1.84, 95% CI = 1.13, 3.01),

and lower cancer worry (OR = 0.64, 95% CI = 0.50–

0.82) were independent predictors of accurate risk com-

prehension (Table 3). Thus, risk overestimation was more

likely to occur in younger women and in women with

greater cancer worry.

Adherence to mammography recommendations

Of women age 43 and older, 29% (n = 95) were classified

as not adherent to mammography guidelines. Women from

different countries had varying rates of non-adherence to

mammography guidelines, with women from Guatemala

reporting the highest rates of non-adherence (50.0%), fol-

lowed by Bolivia (41.2%), Peru (38.1%), El Salvador

(30.0%), Nicaragua (23.5%), Columbia (18.2%), Honduras

(18.2%), and Ecuador (15.4%).

Variables associated with adherence at the bivariate and

multivariate levels included age 50 years or older,

(OR = 2.99, 95% CI = 1.76–5.09), having insurance

(OR = 1.81, 95% CI = 1.03–3.17), greater acculturation

(OR = 1.18, 95% CI = 1.02–1.36), and higher breast

cancer knowledge (OR = 2.03, 95% CI = 1.21–3.40; see

Tables 4 and 5). Country of birth, level of education,

cancer worry, and risk comprehension were not associated

with adherence. Women at high-risk for breast cancer

based on Gail scores (n = 31) had the same rate of non-

adherence to mammography guidelines (29%) as women in

the entire sample.

Discussion

To our knowledge, this is the first study to evaluate the

relationships among perceived risk, risk comprehension,

cancer worry, and mammography adherence while consid-

ering demographics, acculturation, and breast cancer

knowledge in Latinas. Overall, Latinas significantly over-

estimated their cancer risk. Interestingly, risk perceptions

and cancer worry were not associated with adherence,

although age, insurance status, acculturation, and knowledge

were independently associated with screening behavior.

Latinas’ overestimation of their breast cancer risk is

consistent with findings among Caucasian women [20, 22].

Similarly, the significant relationship between cancer

Table 2 Correlates of risk perception and risk comprehension

Characteristic Perceived risk variables

Absolute

risk

Comparative

risk

Numeric

risk

Risk

comprehension

Age -0.07 -0.002 -0.02 -0.14**

Marital status -0.05 -0.03 -0.05 0.04

Education -0.02 -0.05 -0.07 -0.03

Insurance

status

0.01 -0.07 -0.04 -0.06

Birth place 0.06 0.02 0.08 0.10*

Clinic site 0.11* -0.08 0.001 0.04

Acculturation 0.01 -0.05 -0.03 -0.05

5-year Gail

risk

-0.07 -0.01 -0.07 –

Cancer worry 0.20*** 0.13** 0.28*** 0.17***

Breast cancer

knowledge

0.02 0.03 -0.03 -0.05

* p \ .05; ** p \ .01; *** p \ .001

Note: Values in the table represent the Pearson-Product Correlation

Coefficients for continuous variables (age, education, acculturation,

5-year Gail risk, Cancer worry, Breast cancer knowledge) and Point-

biserial correlation coefficients for dichotomous variables [insurance

status (yes/no), marital status (yes/no), birth place (El Salvador/

other)]. Dashes indicate that the correlation between the 5-year Gail

risk score and the Risk comprehension score was not evaluated

because the Gail risk score was used to create the Risk comprehension

score

Table 3 Variables associated with risk comprehension

Variable Accurate risk comprehension

OR (95%CI)

Age (C50 years vs. \50 years) 1.84 (1.13, 3.01)*

Cancer worrya 0.64 (0.50, 0.82)**

a Cancer worry measured as a continuous variable; higher scores

represent greater cancer worry

* p \ .05; ** p \ .001

Note: Odds Ratios calculated from parameter estimates of the logistic

regression models with backward elimination of the following vari-

ables: education (v2 = 0.03, p [ .05), acculturation (v2 = 0.25,

p [ .05), insurance status (v2 = 1.08, p [ .05), clinic (v2 = 0.80,

p [ .05), knowledge (v2 = 1.86, p [ .05), and country of origin

(v2 = 3.51, p [ .05)

1378 Cancer Causes Control (2008) 19:1373–1382

123

worry and perceived risk is also consistent with prior

research indicating a strong and significant relationship

between these two variables [30, 34]. In these prior studies,

cancer worry has been treated as both an independent

predictor of perceived risk and as an outcome of perceived

risk. Thus, the direction and nature of this relationship

among Latina women warrants further evaluation in studies

with prospective research designs.

Slightly less than one-third (29%) of Latinas were non-

adherent to mammography guidelines. Consistent with

prior research with Latinas who identified as Cuban,

Mexican, Puerto Rican, and Central American [10], breast

cancer knowledge was significantly associated with

screening adherence in our population of both Central and

South American Latinas, with lower knowledge associated

with lower rates of adherence. Women from the various

Central and South American countries reported differing

rates of adherence. Due to low sample sizes of women from

certain countries, we were not able to evaluate these dif-

ferences statistically; however, future research with Latinas

could explore whether individual cultural differences

account for these discrepancies. Similar to prior research,

sociodemographic characteristics including younger age

and not having insurance were associated with non-

adherence to mammography guidelines [5, 7–10]. Our

findings of the positive association between acculturation

and adherence may clarify equivocal results in prior stud-

ies. Women who were less acculturated reported lower

rates of adherence. Other culturally relevant factors that

may be associated with adherence include cancer fatalism

and medical mistrust, as these constructs may influence risk

perception and mammography adherence in Latinas [35].

Examination of these pathways will be important to

explore in future research.

Our finding that Latinas between ages 43 and 50 were

less likely to adhere to mammography screening guidelines

than women age 50 and older is consistent with cancer

screening research conducted with Latinas [5] and other

populations [36]. These results indicate that women eligi-

ble for mammography may not be initiating this screening

behavior at the appropriate time. Although the Latina

population in the present study had access to community

health clinics, actual access to free or reduced-cost mam-

mography services was not assessed.

Unlike predictors of mammography adherence in pre-

dominately non-Hispanic White samples and contrary to

expectation in the present study, perceived risk, risk

Table 4 Unadjusted associations between variables of interest and

adherence to screening guidelines (n = 328)

Variable Adherent

(n = 233)

Non-adherent

(n = 95)

Age**

Mean (SD) 55.3 (8.3) 53.7 (9.9)

\50, n (%) 61 (18.6) 44 (13.4)

C50, n (%) 172 (52.4) 51 (15.6)

Marital status

Married, n (%) 163 (49.7) 63 (19.2)

Unmarried, n(%) 70 (21.3) 32 (9.8)

Education**

BHigh School, n (%) 171 (52.1) 76 (23.2)

[High School, n (%) 62 (18.9) 19 (5.8)

Insurance **

Yes, n (%) 85 (25.9) 23 (7.0)

No, n (%) 148 (45.1) 72 (22.0)

Birthplace

El Salvador, n (%) 84 (25.6) 36 (11.0)

Other, n (%) 149 (45.4) 59 (18.0)

Clinic

A, n (%) 67 (20.4) 35 (10.7)

B, n (%) 85 (25.9) 29 (8.8)

C, n (%) 81 (24.7) 31 (9.5)

Acculturation, Mean (SD)** 5.3 (2.3) 4.8 (1.5)

Cancer worry, Mean (SD) 3.02 (1.07) 3.08 (1.10)

Breast cancer knowledge,

Mean (SD)*

68.8 (10.4) 62.7 (17.1)

High, n (%) 122 (37.2) 35 (10.7)

Low, n (%) 111 (33.8) 60 (18.3)

Risk comprehension

Accurate, n (%) 55 (16.7) 14 (4.3)

Overestimate, n (%) 178 (54.3) 81 (24.7)

* p \ .05; ** p \ .001

Note: Percents calculated as the percent of the total 328 participants

included in these analyses

Table 5 Variables associated with adherence to mammography

guidelines

Variable Adherence to mammography

guidelinesa

OR (95% CI)

Age (C50 years vs. \50 years) 2.99 (1.76, 5.09)***

Insurance(Yes vs. No) 1.81 (1.03, 3.17)*

Acculturationb 1.18 (1.02, 1.36)*

Breast cancer knowledge (High

vs. Low)

2.03 (1.21, 3.40)**

a n = 328b Acculturation measured as a continuous variable; higher scores

represent greater acculturation

* p \ .05; ** p \ .01; *** p \ .001

Note: Odds Ratios calculated from parameter estimates of the logistic

regression models with backward elimination of the following vari-

ables: cancer worry (v2 = 0.06, p [ .05), country of origin

(v2 = 1.02, p [ .05), education (v2 = 1.51, p [ .05), clinic

(v2 = 1.33, p [ .05), and risk comprehension (v2 = 1.48, p [ .05)

Cancer Causes Control (2008) 19:1373–1382 1379

123

comprehension and cancer worry were not associated with

adherence in Latinas, whereas breast cancer knowledge

was associated with adherence. Thus, for Latinas, educa-

tion about breast cancer and mammography may be more

valuable intervention targets than efforts to improve risk

comprehension, particularly among newly immigrated

and/or mono-lingual Latinas. Moreover, these interventions

may be particularly useful in younger Latinas who are

appropriate candidates for mammography, but have not yet

initiated screening.

Study strengths and limitations

Strengths of this study include the focus on an understudied

and primarily uninsured Latina population recruited from

community health clinics. Latinas in our sample were from

predominately Central American countries, whereas prior

research has often focused on women from Mexico and

Puerto Rico. Our inclusion of multiple risk perception

measures allowed us to evaluate the relationship of each

measure with adherence. Although the measurement

approaches (absolute, comparative, and numeric perceived

risk) were associated with one another, none were associ-

ated with mammography adherence. Finally, we had a very

high participation rate, likely due to our use of bilingual

study interviewers and conduct of the survey within clinics

that were trusted by our study population [37, 38].

Limitations to the present study should be noted, includ-

ing use of a clinic-based convenience sample and assessment

of mammography adherence through self-report. Our clinic-

based sample may differ from non-clinic based samples

regarding both perceived risk and mammography adherence,

and thus results may not generalize to Latina women who do

not seek care through community health clinics. For exam-

ple, women seen through community health clinics may be

more likely to engage in breast cancer screening behaviors

compared to women who do not regularly seek health care. In

contrast, patients at community clinics may engage in less

cancer prevention behavior than women who are seen reg-

ularly by a primary care physician. In terms of self-report of

mammography adherence, some evidence indicates women

over-report adherence [39], whereas other evidence suggests

women are able to report mammography adherence with

acceptable sensitivity and specificity [40, 41]. If women in

the present study over-reported adherence, then an even

greater proportion of this population may have been non-

adherent to screening.

Adherence to mammography screening guidelines in the

present study followed 2002 recommendations of the

National Cancer Institute, which at the time advised

women 40 years of age and older to have a mammogram

every 1–2 years. Other agencies (e.g., American Cancer

Society) had different recommendations in 2002, such that

women over age 50 should have a mammogram every year

[42]. As such, our data reflect a conservative estimate of

the number of women who were non-adherent to mam-

mography screening guidelines. In addition, the current

sample consisted of a small number of women at high risk

of breast cancer based on objective criteria, thus limiting

our power to detect significant relationships within this

sub-sample. Use of the Gail model in a population of La-

tinas could also be considered a limitation as few studies

have used this model in racial and ethnic minority popu-

lations [43, 44]. Moreover, even less is known about how

the Gail model applies to particular subgroups of Latinas

(e.g., Salvadoran women).

This study has important implications for primary care

providers and clinics treating Latinas. Our high participa-

tion rate indicates that Latinas are willing to engage in

dialog about breast cancer screening [24]. The key finding

from this research stresses the importance of education to

improve breast cancer knowledge in this population—

teaching Latinas more about breast cancer, including

symptoms (and lack thereof), screening/early detection

recommendations, and risk factors—may be an effective

approach to improve mammography adherence. These

education efforts may not benefit from any focus on

improving the risk comprehension in Latinas, as our data

indicate that risk comprehension is not associated with

screening behavior. Future research would be strengthened

by use of prospective designs, ascertainment of actual

mammography behaviors, and inclusion of larger numbers

of high-risk Latinas. Finally, future work should build on

and expand existing efforts to intervene with Latinas so

that cultural influences are appropriately considered when

educating these women about breast cancer and breast

screening behaviors [45].

Conclusion

While most Latinas over-estimated their breast cancer risk,

older age, having insurance, being more acculturated, and

having greater knowledge were associated with greater

screening adherence in this Latino population. Perceived

risk, risk comprehension, and cancer worry were not

associated with adherence. In Latinas, screening interven-

tions should emphasize knowledge and target education

efforts at younger, uninsured, and less acculturated mam-

mography-eligible women.

Acknowledgments We would like to thank all the women who

participated in this study, as well as Susan Marx for assistance with

manuscript preparation. Our sincere appreciation also goes to Janet

Canar, Jyl Pomeroy, and Yosselyn Rodriguez for data collection at

each of the clinics; and to Michael Sanchez, Maria Lopez-Class, and

Barbara Kreiling for information about the LACRC clinics.

1380 Cancer Causes Control (2008) 19:1373–1382

123

References

1. John EM, Phipps AI, Davis A, Koo J (2005) Migration history,

acculturation, and breast cancer risk in Hispanic women. Cancer

Epidemiol Biomarkers Prev 14:2905–2913. doi:10.1158/1055-

9965.EPI-05-0483

2. Carozza SE, Howe HL (2006) Patterns of cancer incidence

among US Hispanics/Latinos, 1995–2000. Cancer Causes Control

17:1067–1075. doi:10.1007/s10552-006-0045-3

3. Howe HL, Wu X, Ries LA, Cokkinides V, Ahmed F, Jemal A

et al (2006) Annual report to the nation on the status of cancer,

1975–2003, featuring cancer among U.S. Hispanic/Latino popu-

lations. Cancer 107:1711–1742. doi:10.1002/cncr.22193

4. Abraido-Lanza AF, Chao MT, Gammon MD (2004) Breast and

cervical cancer screening among Latinas and non-Latina whites.

Am J Public Health 94:1393–1398

5. Gorin SS, Heck JE (2005) Cancer screening among Latino sub-

groups in the United States. Prev Med 40:515–526. doi:10.1016/

j.ypmed.2004.09.031

6. Mandelblatt JS, Gold K, O’Malley AS, Taylor K, Cagney K,

Hopkins JS et al (1999) Breast and cervix cancer screening

among multiethnic women: role of age, health, and source of

care. Prev Med 28:418–425. doi:10.1006/pmed.1998.0446

7. Thompson B, Coronado GD, Solomon CC, McClerran DF,

Neuhouser ML, Feng Z (2002) Cancer prevention behaviors and

socioeconomic status among Hispanics and non-Hispanic whites

in a rural population in the United States. Cancer Causes Control

13:719–728. doi:10.1023/A:1020280427712

8. Abraido-Lanza AF, Chao MT, Gates CY (2005) Acculturation

and cancer screening among Latinas: results from the National

Health Interview Survey. Ann Behav Med 29:22–28.

doi:10.1207/s15324796abm2901_4

9. Brown WM, Consedine NS, Magai C (2006) Time spent in the

United States and breast cancer screening behaviors among

ethnically diverse immigrant women: evidence for accultura-

tion? J Immigr Minor Health 8:347–358. doi:10.1007/s10903-

006-9005-y

10. Ramirez AG, Suarez L, Laufman L, Barroso C, Chalela P (2000)

Hispanic women’s breast and cervical cancer knowledge, atti-

tudes, and screening behaviors. Am J Health Promot 14:292–300

11. Echeverria SE, Carrasquillo O (2006) The roles of citizenship

status, acculturation, and health insurance in breast and cervical

cancer screening among immigrant women. Med Care 44:788–

792. doi:10.1097/01.mlr.0000215863.24214.41

12. O’Malley AS, Kerner J, Johnson AE, Mandelblatt J (1999)

Acculturation and breast cancer screening among Hispanic

women in New York City. Am J Public Health 89:219–227

13. Peragallo NP, Fox PG, Alba ML (2000) Acculturation and breast

self-examination among immigrant Latina women in the USA.

Int Nurs Rev 47:38–45. doi:10.1046/j.1466-7657.2000.00005.x

14. Ruiz MS, Marks G, Richardson JL (1992) Language accultura-

tion and screening practices of elderly Hispanic women. The role

of exposure to health-related information from the media. J Aging

Health 4:268–281. doi:10.1177/089826439200400206

15. Zambrana RE, Breen N, Fox SA, Gutierrez-Mohamed ML (1999)

Use of cancer screening practices by Hispanic women: analyses by

subgroup. Prev Med 29:466–477. doi:10.1006/pmed.1999.0566

16. Gelberg L, Andersen RM, Leake BD (2000) The Behavioral

Model for Vulnerable Populations: application to medical care

use and outcomes for homeless people. Health Serv Res 34:

1273–1302

17. Katapodi MC, Lee KA, Facione NC, Dodd MJ (2004) Predictors

of perceived breast cancer risk and the relation between perceived

risk and breast cancer screening: a meta-analytic review. Prev

Med 38:388–402. doi:10.1016/j.ypmed.2003.11.012

18. Moser RP, McCaul K, Peters E, Nelson W, Marcus SE (2007)

Associations of perceived risk and worry with cancer health-

protective actions: data from the Health Information National

Trends Survey (HINTS). J Health Psychol 12:53–65.

doi:10.1177/1359105307071735

19. Calvocoressi L, Stolar M, Kasl SV, Claus EB, Jones BA (2005)

Applying recursive partitioning to a prospective study of factors

associated with adherence to mammography screening guide-

lines. Am J Epidemiol 162:1215–1224. doi:10.1093/aje/kwi337

20. Daly MB, Lerman CL, Ross E, Schwartz MD, Sands CB, Masny

A (1996) Gail model breast cancer risk components are poor

predictors of risk perception and screening behavior. Breast

Cancer Res Treat 41:59–70. doi:10.1007/BF01807037

21. Kelly K, Leventhal H, Marvin M, Toppmeyer D, Much J, Der-

mody J et al (2004) Subjective and objective risk of breast cancer

in Ashkenazi Jewish individuals at risk for BRCA1/2 mutations.

Genet Test 8:139–147

22. Lerman C, Lustbader E, Rimer B, Daly M, Miller S, Sands C et al

(1995) Effects of individualized breast cancer risk counseling: a

randomized trial. J Natl Cancer Inst 87:286–292. doi:10.1093/

jnci/87.4.286

23. Li R, Gilliland FD, Baumgartner KB, Samet J (2001) Family

history and risk of breast cancer in Hispanic and non-Hispanic

women: the New Mexico Women’s Health Study. Cancer Causes

Control 12:747–753. doi:10.1023/A:1011285205870

24. Mandelblatt J, Kaufman E, Sheppard VB, Pomeroy J, Kavanaugh

J, Canar J et al (2005) Breast cancer prevention in community

clinics: will low-income Latina patients participate in clinical

trials? Prev Med 40:611–618. doi:10.1016/j.ypmed.2004.09.004

25. Kreling BA, Canar J, Catipon E, Goodman M, Pallesen N,

Pomeroy J et al (2006) Latin American Cancer Research Coali-

tion. Community primary care/academic partnership model for

cancer control. Cancer 107:2015–2022. doi:10.1002/cncr.22145

26. Sheppard VB, Kaufman E, Pomeroy J, Kavanaugh J, O’Malley

A, Pallandre L et al (2004) Latina’s willingness to participate in

chemoprevention trials. Int J Cancer Prev 1:391–396

27. Marin G, Marin BV (1991) Research with Hispanic populations

(Applied Social Research Methods Series, vol. 23). Sage Publi-cations, Newbury Park, CA

28. Armstrong K, Micco E, Carney A, Stopfer J, Putt M (2005)

Racial differences in the use of BRCA1/2 testing among women

with a family history of breast or ovarian cancer. JAMA

293:1729–1736. doi:10.1001/jama.293.14.1729

29. Lerman C, Trock B, Rimer BK, Boyce A, Jepson C, Engstrom PF

(1991) Psychological and behavioral implications of abnormal

mammograms. Ann Intern Med 114:657–661

30. Lipkus IM, Kuchibhatla M, McBride CM, Bosworth HB, Pollak

KI, Siegler IC et al (2000) Relationships among breast cancer

perceived absolute risk, comparative risk, and worries. Cancer

Epidemiol Biomarkers Prev 9:973–975

31. Gail MH, Costantino JP, Bryant J, Croyle R, Freedman L, Hel-

zlsouer K et al (1999) Weighing the risks and benefits of

tamoxifen treatment for preventing breast cancer. J Natl Cancer

Inst 91:1829–1846. doi:10.1093/jnci/91.21.1829

32. McTiernan A, Kuniyuki A, Yasui Y, Bowen D, Burke W, Culver

JB et al (2001) Comparisons of two breast cancer risk estimates

in women with a family history of breast cancer. Cancer Epi-

demiol Biomarkers Prev 10:333–338

33. National Cancer Institute (2002) NCI statement on mammography

screening. http://www.cancer.gov/newscenter/mammstatement31

jan02. Accessed 29 March 2007]

34. McGregor BA, Bowen DJ, Ankerst DP, Andersen MR, Yasui Y,

McTiernan A (2004) Optimism, perceived risk of breast cancer,

and cancer worry among a community-based sample of women.

Health Psychol 23:339–344. doi:10.1037/0278-6133.23.4.339

Cancer Causes Control (2008) 19:1373–1382 1381

123

35. Luquis RR, Villanueva-Cruz IJ (2006) Knowledge, attitudes, and

perceptions about breast cancer and breast cancer screening

among Hispanic women residing in South Central Pennsylvania. J

Community Health 31:25–42. doi:10.1007/s10900-005-8187-x

36. O’Malley AS, Forrest CB, Mandelblatt J (2002) Adherence of

low-income women to cancer screening recommendations. J Gen

Intern Med 17:144–154. doi:10.1046/j.1525-1497.2002.10431.x

37. Sheppard VB, Zambrana RE, O’Malley AS (2004) Providing

health care to low-income women: a matter of trust. Fam Pract

21:484–491. doi:10.1093/fampra/cmh503

38. Sheppard VB, Cox LS, Kanamori MJ, Canar J, Rodriguez Y,

Goodman M et al (2005) Brief report: if you build it, they will

come: methods for recruiting Latinos into cancer research. J Gen

Intern Med 20:444–447. doi:10.1111/j.1525-1497.2005.0083.x

39. Armstrong K, Long JA, Shea JA (2004) Measuring adherence to

mammography screening recommendations among low-income

women. Prev Med 38:754–760. doi:10.1016/j.ypmed.2003.12.023

40. McPhee SJ, Nguyen TT, Shema SJ, Nguyen B, Somkin C, Vo P

et al (2002) Validation of recall of breast and cervical cancer

screening by women in an ethnically diverse population. Prev

Med 35:463–473. doi:10.1006/pmed.2002.1096

41. Tumiel-Berhalter LM, Finney MF, Jaen CR (2004) Self-report

and primary care medical record documentation of mammogra-

phy and Pap smear utilization among low-income women. J Natl

Med Assoc 96:1632–1639

42. Smith RA, Saslow D, Sawyer KA, Burke W, Costanza ME,

Evans WPIII et al (2003) American Cancer Society guidelines

for breast cancer screening: update 2003. CA Cancer J Clin 53:

141–169

43. Abu-Rustum NR, Herbolsheimer H (2001) Breast cancer risk

assessment in indigent women at a public hospital. Gynecol

Oncol 81:287–290. doi:10.1006/gyno.2001.6160

44. Grann VR, Jacobson JS, Troxel AB, Hershman D, Karp J, Myers

C et al (2005) Barriers to minority participation in breast carci-

noma prevention trials. Cancer 104:374–379. doi:10.1002/

cncr.21164

45. Thompson B, Coronado G, Chen L, Islas I (2006) Celebremos la

salud! a community randomized trial of cancer prevention

(United States). Cancer Causes Control 17:733–746. doi:10.1007/

s10552-006-0006-x

1382 Cancer Causes Control (2008) 19:1373–1382

123