Preventive Medicine 29, 466–477 (1999)Article ID pmed.1999.0566, available online at http://www.idealibrary.com on

Use of Cancer Screening Practices by Hispanic Women:Analyses by Subgroup

Ruth E. Zambrana, Ph.D.,*,1 Nancy Breen, Ph.D.,† Sarah A. Fox, Ed.D.,‡ andMary Lou Gutierrez-Mohamed, Ph.D.§

*Social Work Program, George Mason University, Fairfax, Virginia 22030; †Applied Research Program, Division of Cancer Controland Population Sciences, National Cancer Institute, Na

,

data collection approaches. q 1999 American Health Foundation

a mammogram in the previous year. Despite increasedscreening among all women, low-income women,

‡Department of Medicine, University of California, Los Angelesand §Jackson Heart Study, Jackson State U

Objectives. This study compares the use of three can-cer screening practices (Pap smear, mammogram, andclinical breast examination) 3 years prior to interviewamong five subgroups of Hispanic women, and exam-ines whether sociodemographic; access; health behav-ior, perception, and knowledge; and acculturation fac-tors predict screening practices for any subgroup.

Methods. Descriptive and multiple logistic regres-sion analyses were conducted with data pooled fromthe 1990 and 1992 National Health Interview Surveyson women who reported that they were Hispanic. Thestudy sample includes 2,391 respondents: 668 Mexican-American, 537 Mexican, 332 Puerto Rican, 143 Cuban,and 711 other Hispanic women.

Results. Subgroup profiles reveal differences in edu-cation, health insurance, use of English language, andscreening use. Mexican women were the least likely tobe screened with any procedure. Logistic regressionresults for each screening practice show that havinga usual source of care was a positive predictor for ob-taining each of the three screening practices withinthe last 3 years. Being married, being more than 50years of age, and having knowledge of breast self-exam-ination were all predictors of having a Pap smear. Hav-ing health insurance and ever having had a clinicalbreast examination and Pap smear were predictors ofhaving a mammography, while age, knowledge ofbreast self-examination, ever having had a Pap smear

and mammogram, and being a nonsmoker all predictedhaving a clinical breast examination.

Conclusions. We conclude that access factors andprior screening are more strongly associated with cur-rent screening than are language and ethnic factors.

1 To whom correspondence and requests for reprints should be ad-dressed to the current address at University of Maryland CollegePark, Department of Women’s Studies, 2101 Woods Hall, CollegePark, Maryland 20742. E-mail: rz23@umail.umd.edu.

46

tional Institutes of Health, Bethesda, Maryland 20892;and Rand Corporation, Santa Monica, California 90407-2138;niversity, Jackson, Mississippi 39216-4506

Our data confirm that a disproportionate percentageof Hispanic women are low income and at risk of beingunderscreened. Our findings from a nationally repre-sentative sample of Hispanics have implications forprovider practices, ethnic-specific community inter-ventions, and future development of measures and

and Academic Press

Key Words: cancer screening; women’s health; Lat-ina; ethnicity.

INTRODUCTION

Early detection of breast and cervical cancer is consis-tent with national objectives to promote cancer screen-ing behaviors [1,2]. The Healthy People 2000 Objectivesfor Hispanic2 women are to increase to 60% thosewomen aged 50 and older who have received a clinicalbreast examination (CBE) and a mammogram withinthe past 1–2 years, and to increase to at least 95% theproportion of women aged 18 and older who have everreceived a Pap smear [3]. In 1987, 13% of Hispanicwomen reported having a mammogram in the past yearcompared with 18% of non-Hispanic white women [4].By 1992, rates for every group had doubled and His-panic women were as likely as other groups to have had

women over 65, and Hispanic women still remain atthe greatest risk of not being screened [5].

2 “Hispanic women” and “Latino women” are used to refer to womenof Hispanic origin from Mexico, Puerto Rico, Cuba, and Central andSouth America. The designation “Hispanic” is a federal designation,and is used in national and state reporting systems. “Latino” is aself-designated term by members of different groups. “Hispanic” isused in this paper without preference or prejudice.

6 0091-7435/99 $30.00Copyright q 1999 by American Health Foundation and Academic Press

All rights of reproduction in any form reserved.

E

CANCER SCREENING US

Hispanics, who constitute nearly 11% of the U.S. pop-ulation, represent regional and socioeconomically dis-tinct subgroups3 [6–10]. In 1990, the Hispanic popula-tion consisted of Mexican-origin (63%), Puerto Rican(11%), Cuban (5%), Central and South American (14%),and Other Hispanic (8%).4 Hispanics are geographicallyconcentrated in five states: California (34%), Texas(19%), Florida (7%), New York (10%), and Illinois (4%)[8]. Hispanic women currently represent more than 9%of the total female U.S. population, and are expectedto increase to more than 15% by the year 2020 [6].Hispanic women are younger, have lower median an-nual earnings, and are more likely to be unemployedor employed in low-wage, service-sector jobs than His-panic men or non-Hispanic women [8,11]. Data on 1993family income by Hispanic subgroup reveal that 37%of Puerto Ricans, 30% of Mexican-Americans, 18% ofCubans, and approximately 23% of Other Hispanic fam-ilies were below the poverty level compared with 7.6%of non-Hispanic white families [12]. The rationale forexamining Hispanics by subgroups is strongly arguedby several authors [7–10]. For example, Massey states,“Hispanics do not comprise a single coherent commu-nity. Rather, they are a disparate collection of nationalorigin groups with heterogeneous experiences of settle-ment, immigration, political participation, and eco-nomic incorporation into the United States” [9], p. 454.In this paper, we compare five subgroups of Hispanicwomen on their use of three cancer screening proce-dures (Pap smear, mammogram, and CBE) and test forsignificant predictors of screening practices.

Hispanic women tend to be younger when diagnosedwith breast or cervical cancer, present with these can-cers at later stages of diagnosis, and have higher mor-tality rates than non-Hispanic white women [13–16].Hispanic women are less likely to have visited a physi-cian in the last year, to have had a mammogram anda Pap smear, and to know cancer warning signs [17].The underlying factors that place Hispanic women atrisk are low-income status, lack of access to health care,and institutional barriers [18–30].

Screening is positively associated with younger age,higher income, greater educational level, being mar-ried, health insurance, having a usual source of care,and knowledge and use of prior cancer screening tests[18–25]. Among U.S. women, Pap smear screening

3 Subgroup is used throughout the paper to refer to individualswho self-report Hispanic ethnicity and identify with a particularnational origin such as Mexican-American or Puerto Rican.

4 The Census Bureau definition of persons of Other Hispanic originare those whose origins are Spain, the Spanish-speaking countriesof Central and South America, or the Dominican Republic, or personsidentifying themselves generally as Spanish, Spanish-American, His-panic, Hispano, or Latino.

BY HISPANIC WOMEN 467

rates decline with increasing age. Higher levels of edu-cation are associated with more mammography screen-ing [23,24]. These patterns are similar for Hispanicwomen. Hispanic women who are older than 65 yearsof age, less educated, or poor are less likely to use mam-mography [25]. Stein et al. found that for Hispanicwomen, financial indicators, primarily lack of insur-ance, reduced use of mammography more than culturaldifferences [26]. A recent study on access to health carefound that health insurance coverage was higher forEnglish-speaking women of Mexican origin who re-ported a family income above the poverty level, hadhigher levels of education, were employed, and wereyounger, as compared with Spanish-speaking Mexican-origin women [27]. Puerto Rican and Mexican-originwomen, aged 50 to 64, had the lowest percentage ofhealth insurance coverage. While 95.8% of all womenaged 65 and older report having Medicare coverage,only 88.1% of Mexican-American women, 86.5% of Cu-ban women, and 77.4% of Puerto Rican women haveMedicare coverage5 [23].

Although the preponderance of evidence shows thatlow socioeconomic status is an important predictor ofcancer screening practices among Hispanic women [22–30], screening is most strongly associated with knowl-edge of screening, prior screening, and physician recom-mendation for a screening [31–40]. Hispanic women 40years of age and older are less likely to have knowledgeof mammography (31.6%) or CBE (13.4%) than non-Hispanic white women (12.2 and 7.9%, respectively)[1]. Analyses of 1992 National Health Interview Survey(NHIS) data found that women 50 years and older whohad received other preventive services from their pro-viders, such as CBE and mammogram, were more likelyto have had a Pap smear in the last 3 years [41]. Physi-cians not recommending a Pap smear or mammogramdue to cost of tests, women’s lack of knowledge aboutscreening tests, or lack of health insurance are strongdeterminants of cancer screening underutilizationby elderly women, low-income women, and Hispanicwomen [36–40].

For Hispanic women, acculturation, income, and edu-cation are three interrelated factors associated withhealth behaviors [10,17,42–47]. Although there is lim-

ited consensus on how to measure acculturation in pub-lic health, agreement exists on two points: English lan-guage use is a principal marker of acculturation, andit is associated with country of birth, completed yearsof education, and number of years in the United States

5 To receive Medicare and social security entitlement, individualsmust work a total of 5 years in a job where contributions are madeto the government. The lower rates of Medicare coverage amongHispanic women are a reflection of immigration at later ages andless likelihood of participation in the above-ground economy.

468 ZAMBRAN

[47]. Less acculturated Hispanics are often character-ized by limited education, low literacy levels, and useof Spanish language, compared with their more accul-turated counterparts. These attributes are, in turn,linked to decreased access to health services, lower lev-els of health knowledge, less communication with pro-viders [28,47–50], and less likelihood to use preventiveand primary care services, to have annual Pap smears,and, for those over 50 years of age, to report ever havinghad a mammogram [51–54].

Based on factors associated with cancer screeningpractices for Hispanic women reported in the literature,four questions were posed: (a) Are Hispanic women withmore education, a usual source of care, and who speakEnglish more likely to be screened than less educatedwomen, without a usual a source of care and who arenot English speaking? (b) Are Hispanic women withprior screening more likely to have had more recentscreening procedures than Hispanic women withoutprior screening? (c) Are Mexican women, who have theleast education and lowest income, least likely of allHispanic groups to have had recent screening proce-dures? (d) Are Cuban women, who have more educationand higher income, most likely to be screened of allsubgroups?

To analyze cancer screening practices among sub-groups of Hispanic women, we employ a social inequal-ity model. This paradigm proposes that low-incomeindividuals are more exposed to risk factors and experi-ence more barriers in the health care system thanhigher-income individuals [55,56]. For Hispanicwomen, the model posits that socioeconomic character-istics, including income, education, and literacy levels,reduce both knowledge of health practices and accessto health care services which, in turn, make cancerscreening less likely. Acculturation, measured by lan-guage use, place of birth, and length of time in theUnited States, mediates the relationship between socio-economic status and screening practices. In contrast tostandard models of health care utilization that focus onindividual behaviors exclusive of socioenvironmentalconstraints, the social inequality model takes into ac-count how social, cultural, and economic context shapeHispanic women’s use of cancer screening [57].

We pooled 2 years of nationally representative popu-lation samples from the NHIS. This afforded us aunique opportunity to examine the use of preventivecancer screening practices for total Hispanic womenand by subgroup. In this paper, we present the firstnational estimates of cancer screening rates and socio-

demographic, access, and other health behavior vari-ables for total Hispanic women and by subgroup. Webelieve that these new estimates will be useful in formu-lating more effective targeted objectives for Hispanicsubgroups for the Year 2010 Healthy People Objectives.

A ET AL.

METHOD

Data from the 1990 and 1992 NHISs were pooledto increase the statistical power to analyze the fivesubgroups. These data included comparable questionsand showed similar screening rates for all women. Be-cause the 1990 sample constituted 70% of the totalpooled sample and the 1992 sample constituted 30% ofthe total pooled sample (1,667 in 1990 and 724 in 1992),we adjusted the sample weights in the combined analy-sis by multiplying them by the factors 0.7 and 0.3 in1990 and 1992, respectively.

The NHIS is a continuing annual nationwide surveyof approximately 49,000 households of the civilian non-institutionalized population of the United States. Ques-tions on cancer screening practices were included inthe Health Promotion and Disease Prevention Supple-ment of the 1990 survey and in the Cancer ControlSupplement of the 1992 survey. The study sampleincludes 2,391 adult female Hispanic respondents,18 years of age and older. Hispanics were groupedaccording to self-reported ethnic identification intoprecoded categories: Mexican-American (668), Mexican(537), Puerto Rican (332), Cuban (143), and OtherHispanic (711).

Measures and Analyses

Dependent variables include three cancer screeningpractices: Pap smear, mammogram, and CBE. Women18 years of age and older were asked questions regard-ing Pap smears and women 35 years of age and olderwere asked questions regarding mammogram and CBE.We constructed dichotomous dependent variables forscreening: a woman was either screened 3 years priorto the interview or she was not. This yielded comparablevariables for the three screening practices.

Independent variables are grouped into measures ofsocioeconomic status; access to health services; healthperceptions, behavior, and knowledge; and accultura-tion. Socioeconomic variables include age (18–34,35–49, and 501), education level (, 12 years, 12 yearscompleted, and $13 years), annual household income(,$20,000 and $$20,000), labor force status (employedvs other), and marital status (married or living togethervs not married). Access variables include health insur-ance status [private, public, government only (Medic-aid, Medicare, or both), and none] and usual source ofcare (yes/no). Health perceptions, behavior, and knowl-edge include six variables. Respondent-assessed health

status was measured using a 5-point Likert-type scale.Respondents were asked to rate their health from 1 to5 (1 5 excellent, 5 5 poor). These data were recodedinto three groups: 0 5 very good or excellent health,1 5 good, 2 5 fair (or poor). Five self-reported items,

E

CANCER SCREENING US

with yes/no response options, measured smoking sta-tus, knowledge of breast self-examination, and fre-quency of CBE, mammogram, and Pap smear. Accultur-ation proxy variables include language of interview,country of birth, and number of years living in theUnited States. Three variables were coded based onpublished evidence of differences in screening practicesamong Hispanic women by country of birth and lengthof residence [57,58]. Language of interview was usedas a trichotomous variable: English, Spanish, and both.The variables for country of birth and number of yearsin the United States were recoded into three categories:(a) women born in United States, (b) women not bornin the United States and residing in the United Statesfor 10 years or longer, (c) women not born in the UnitedStates and residing in the United States for less than10 years.

The SUDDAN program was used to compute popula-tion estimates for the pooled sample, including propor-tions and design-appropriate confidence intervals andtests for significance [59]. The pooled sample isweighted to the total population of the United Statesand, thus, is representative of each Hispanic subgroup(within the limitations of cluster sampling). SUDAANlogistic regression was used to test group differencesusing regression models for each screening practice.

Several limitations of the data warrant caution inthe interpretation of the findings. The sample size forthe Cuban group is small, and therefore its generaliz-ability may be limited. Other Hispanic group (n 5 711)contains a heterogeneous mix of Hispanics from Centraland South America and others who self-identified inthis category. Inferences cannot be drawn for this het-erogeneous group as they can for groups that represent

CBE, and mammography. For this study, we analyzed

a single nationality. Other possible limitations includewhether responses are accurate, given that some re-spondents may not know about or understand the proce-dures about which they were asked, and the quality ofon-site translation for the Spanish interviews [60,61].

RESULTS

Characteristics of the Sample

Table 1 displays data on sociodemographics; access toservices; health behaviors, perceptions, and knowledge;and acculturation variables for all Hispanic womenaged 18 or older in the pooled NHIS sample. Data areordered in Table 1 by their proportion in the Hispanicpopulation: Mexican-Americans (28%), Mexicans(22%), Puerto Ricans (14%), Cubans (6%), and all OtherHispanics (30%). Table 1 presents the data from whichFig. 1 is derived.

Figure 1 provides a visual comparison of the charac-teristics of each Hispanic subgroup and the total His-panic group. Demographic similarities between the fivesubgroups were employment status, perceived health,

BY HISPANIC WOMEN 469

and smoking behavior. About half (52%) of all Hispanicwomen are in the work force, with Mexican women leastlikely (46%) and Mexican-American (54%) and otherHispanic women (55%) equally likely to be employed.Cuban women are the least likely (49%) to perceivetheir health as excellent or very good, whereas 54% ofMexican and Puerto Rican women perceive this levelof health. Only 17% of the total sample are currentsmokers, and Mexican-American women are the leastlikely to smoke (11%).

Hispanic women as a group are younger than thetotal U.S. population [11,12]. Characteristics that varyby subgroups include age, education, and marital sta-tus. About half (51%) of Hispanic women are betweenages 18 and 34, and less than a fourth are older than50. Cuban women are older as a group, with half olderthan 50 and about one-third between ages 18 and 34.Mexican women are both the youngest (with 55% aged18–34) and the least educated (with more than three-fourths not completing high school). In terms of annualhousehold income (,$20,000), Mexican women (60%)are the poorest, and Cuban (42%) and Other Hispanicwomen (40%) are the least likely to be poor. Mexican(70%) and Cuban (62%) women are the most likely to bemarried, whereas Puerto Rican women are least likely(52%) to be married.

When examining health insurance coverage, PuertoRican women are the most likely to be insured (83%),and Mexican women the least likely to be insured (58%).Noteworthy, although only 58% of Mexican women re-port having health insurance, 65% report a usual sourceof care. Not unexpectedly, Cuban women are most likely(67%) to have private insurance and to have a usualsource of health care (83%).

Additional support for disaggregating Hispanics bysubgroup is gleaned by analyzing the considerable vari-ability on language of interview and birthplace (seeTable 1). As a group, 65% of Hispanic women took theinterview in English and 48% were born in the UnitedStates. A majority of Mexican-American women tookthe interview in English (86%). In contrast, 39% of Mex-ican and 44% of Cuban women took the interview inSpanish. The majority of Mexican-American women(92%) were born in the United States, whereas fewMexicans (13%) or Cubans (27%) were born in theUnited States.

Screening Practices by Hispanic Subgroup and Age

Overall and age-specific screening rates are pre-sented in Table 2 for each of the procedures: Pap smear,

women aged 18 and older for Pap screening, women 30and older for CBE, women 35 and older for mammogra-phy use, and Mexican women were least likely to bescreened for any of the three procedures. Cuban women

$10 years 17 (15–19) 1 (0–2)

In terms of mammography screening, Other Hispanic

,10 years 35 (32–38) 7 (5–9)

a Includes only the 50 states and the District of Columbia.b Mexican-Americans are the referent group.Source: Pooled NHIS data from 1990 and 1992.

younger than 50 years of age reported the highestscreening rates of Pap smear and CBE. Cuban women

50 years and older ranked fourth in screening rates forall three procedures. Mexican-American women hadthe highest rates of Pap smear screening (72%) amongthose 50 years and older, whereas Puerto Rican women

32 (27–38) 11 (6–15) 10 (6–14) 25 (20–29)55 (50–60) 49 (43–55) 63 (54–73) 34 (29–40)

of the same age had the highest rates of CBE (83%).

470 ZAMBRANA ET AL.

TABLE 1

Percent and Confidence Intervals on Selected Sociodemographics; Access to Services; Health Behaviors, Perceptions and Knowledge;and Acculturation of Hispanic Women Aged 18 and Older by Subgroup

Mexican-Americanb Mexican Puerto-Rican Cuban Other Hispanic

Totala n 5 668 n 5 537 n 5 332 n 5 143 n 5 711Characteristic N 5 2,391 (28%) (22%) (14%) (6%) (30%)

SociodemographicsAge

18–34 years 51 (49–53) 54 (49–58) 55 (52–59) 51 (44–57) 32 (23–40) 48 (44–51)35–49 years 27 (25–28) 23 (20–27) 28 (25–31) 31 (26–36) 18 (12–25) 28 (24–32)501 years 23 (21–25) 23 (19–27) 16 (13–20) 18 (14–23) 50 (44–56) 24 (21–28)

Educational attainment,12 years 58 (56–61) 58 (52–63) 77 (72–82) 55 (48–62) 48 (43–53) 48 (43–53)High school graduate only 22 (20–24) 28 (24–32) 13 (9–16) 24 (17–32) 24 (19–29) 22 (18–26)More than high school 20 (18–22) 15 (11–18) 10 (8–13) 21 (16–26) 28 (26–35) 30 (26–35)

Annual Household Income,$20,000 48 (45–51) 44 (39–50) 60 (56–64) 53 (43–62) 42 (33–50) 40 (35–45)$$20,000 52 (49–55) 56 (50–61) 40 (36–44) 47 (38–57) 59 (50–66) 60 (55–65)

Labor force statusEmployed 52 (49–54) 54 (49–58) 46 (40–52) 52 (46–58) 50 (42–58) 55 (51–58)Other 48 (46–51) 46 (42–51) 54 (48–60) 48 (43–54) 50 (42–58) 45 (42–49)

Marital statusMarried/live as married 61 (58–63) 58 (53–64) 70 (66–74) 52 (46–57) 62 (54–71) 58 (54–63)Not married 39 (37–42) 42 (36–47) 30 (26–34) 48 (43–54) 38 (29–46) 42 (37–46)

Access to ServicesHealth Insurance

Private 57 (53–60) 62 (55–70) 44 (39–49) 52 (44–60) 67 (61–73) 61 (56–66)Public 16 (14–18) 16 (12–21) 14 (10–18) 31 (25–36) 16 (12–20) 12 (9–16)None 27 (24–30) 21 (16–26) 43 (37–48) 17 (13–21) 17 (12–23) 27 (23–30)

Usual source of careHas usual source 74 (72–76) 77 (72–81) 65 (61–70) 77 (72–82) 83 (72–93) 76 (72–79)No usual source 26 (24–28) 23 (20–27) 35 (30–39) 23 (18–29) 17 (7–28) 24 (21–28)

Health Behaviors, Perceptions, and KnowledgeRespondent-assessed health status

Excellent, very good 54 (51–58) 51 (45–57) 54 (47–61) 54 (44–63) 49 (42–55) 59 (54–64)Good 30 (28–33) 35 (31–40) 29 (25–33) 27 (20–33) 32 (24–39) 28 (24–33)Fair, poor 15 (13–17) 13 (10–17) 17 (12–22) 19 (14–25) 20 (14–26) 13 (10–15)

Smoking statusCurrent 17 (15–20) 21 (17–25) 11 (7–15) 19 (15–24) 16 (11–22) 19 (15–22)Other 83 (80–85) 79 (75–83) 89 (85–93) 81 (76–85) 84 (78–89) 81 (78–85)

AcculturationLanguage of interview

English 65 (62–69) 86 (80–93) 39 (32–46) 69 (61–77) 44 (35–52) 70 (65–74)Spanish 21 (17–23) 5 (2–9) 40 (36–44) 15 (9–21) 46 (37–55) 19 (15–22)English and Spanish 13 (11–16) 8 (5–12) 21 (16–26) 16 (11–21) 10 (5–15) 12 (9–15)

Nativity and length of stayBorn in United States (1) 48 (44–53) 92 (90–95) 13 (10–15) 40 (34–46) 27 (19–35) 41 (34–48)

women, across age groups, were the most likely to bescreened overall, followed by Cuban women 35–49years old and Mexican-American women 50 yearsand older.

a

FIG. 1. Percent distribution of sociodemographic and access charsubgroups.

Logistic Regression Findings

Separate logistic regression models were run for eachscreening modality. To avoid problems associated withmissing data, the Pap regression included all women 18years of age or older. The CBE and the mammographyregression equations included Hispanic women 35

years and older, so that mammography and CBE couldbe used as independent variables in their respectiveequations. Whether a woman had a Pap smear, mam-mogram, or CBE in the 3 years prior to the interviewwas regressed on the variables shown in Table 1.

cteristics, by total Hispanic (with confidence interval) and Hispanic

Table 3 displays the results from the logistic regres-sion analyses including the odds ratios and their cor-responding confidence intervals. Although Mexicanwomen were less likely to report being screened as dis-played in Table 2, the multivariate analysis indicatesthat factors other than ethnicity may account for thisresult. Instead, being married, having a usual source

CANCER SCREENING USE BY HISPANIC WOMEN 471

of care, and knowledge of how to do a breast self-exami-nation (BSE) all predicted higher Pap smear screening,while older age predicted lower screening. Reported useof mammography in the last 3 years was associated

35–49 years 44 (39–49) 43 (34–52) 37 (29–45) 39 (31–47) 43 (29–57) 54 (46–62)501 years 44 (39–49) 51 (42–61) 31 (19–42) 49 (38–61) 40 (21–60) 55 (46–63)

a Includes only the 50 states and the District of Columbia.b Mexican-Americans are the referent group.Source: Pooled NHIS data for 1990 and 1992.

with having insurance (public or private), having ausual source or care, and ever having had a Pap smearand a CBE. Younger age, having a usual source of care,knowledge of BSE, ever having had a Pap smear or amammogram, and not smoking all predicted having aCBE. The most consistent predictors of use of all threecancer screening tests were use of other preventive ser-vices and having a usual source of care.

Additional analyses were conducted to examine thepotential confounding effects of use of other preventiveservices and access variables with our acculturationproxy variables. Logistic regression models were runfor each screening procedure excluding use of other pre-ventive service variables, access to health care vari-ables, and both of these variables. The results of theseanalyses were not significantly different from our mainanalysis with two exceptions: (a) When access to healthcare was excluded in the model of predictors of havinghad a mammography, women over 50 who received amammogram were less likely to receive a Pap screening

472 ZAMBRANA ET AL.

TABLE 2

Percent and Confidence Intervals of Hispanic Women Screened with Pap Smear, Clinical Breast Examination, and Mammographywithin 3 Years Prior to Interview by Subgroup

Mexican-Totala Americanb Mexican Puerto Rican Cuban Other Hispanic

N 5 2,288 n 5 640 n 5 507 n 5 317 n 5 141 n 5 683

Pap ($18 years)Total 77 (74–79) 80 (75–84) 72 (69–75) 77 (71–83) 73 (65–82) 78 (74–82)18–34 years 78 (75–82) 81 (74–87) 71 (65–76) 81 (75–88) 78 (69–88) 81 (76–87)35–49 years 82 (77–87) 83 (72–94) 82 (76–88) 74 (61–88) 91 (78–103) 83 (77–89)501 years 67 (63–71) 72 (63–81) 59 (51–67) 70 (59–80) 63 (49–78) 67 (60–75)

Clinical breast examination ($30 years) N 5 1,560 n 5 408 n 5 338 n 5 205 n 5 121 n 5 488Total 79 (77–82) 80 (78–87) 74 (69–78) 83 (76–90) 77 (69–84) 80 (75–86)30–49 years 83 (80–86) 84 (76–91) 78 (73–83) 83 (76–91) 86 (82–91) 85 (81–89)501 years 73 (69–77) 74 (66–82) 62 (50–73) 83 (72–93) 70 (60–81) 77 (69–85)

Mammography ($35 years) N 5 1,218 n 5 315 n 5 245 n 5 161 n 5 100 n 5 397Total 46 (42–49) 54 (48–60) 35 (28–42) 43 (36–49) 41 (28–55) 47 (40–54)

and CBE, and (b) for the three screening procedures,if access to health care is available, language appearsto assume a secondary role and is not a barrier to cancerscreening. (These results are not shown, and can beobtained from authors on request.)

DISCUSSION

This study explored intergroup variation among His-panic women in their use of three cancer screeningpractices and tested for significant predictors of eachscreening practice. Consistent with other studies [35–

41], we found that usual source of care and prior screen-ing predicted current screening practices for all sub-groups of Hispanic women. The women who reporteda mammogram in the last 3 years were almost one andone-half times more likely to know how to do a BSEand more than four times more likely to have ever hada CBE. Also, the women who reported having had aCBE were twice as likely to know how to do a BSE andsix and one-half times more likely to ever had a Papsmear. Our data confirm that like all women, Hispanicwomen with access to health care are more likely toengage in more preventive health behaviors than arewomen without access to health care [29,40]. A newfinding in our analysis showed that many more womenover 50 received a mammogram than a Pap smear orCBE. Important appropriateness of care questions sur-face regarding whether these women are obtaining thefull range of primary care services in a physician’s officein addition to regular screening or whether they aresimply receiving a mammogram. The extent to whichHispanic women are receiving segmented care meritsfurther research.

Contrary to our expectations and evidence from otherstudies [17,37,47–53], increased education, Englishlanguage use, and Cuban subgroup did not predict useof screening. Screening rates for Mexican women, espe-cially those older than 50, were lower than for other

subgroups in the bivariate analyses. However, the mul-tivariate analysis demonstrated that they under-screened due to factors other than ethnicity and lan-guage. In addition we found that Cuban women, who

CANCER SCREENING USE BY HISPANIC WOMEN 473

TABLE 3

Multiple Logistic Regression Analysis of Predictors of Having Had a Cancer Screening Test in the Last 3 Years (N 5 2,391)

Had clinical breastHad a Pap smear Had a mammogram Examination

OR 95% CI OR 95% CI OR 95% CI

SociodemographicsHispanic subgroup

Mexican 1.13 0.58–2.19 0.75 0.41–1.37 0.88 0.39–1.98Puerto-Rican 0.96 0.56–1.64 0.89 0.49–1.61 1.34 0.49–3.70Cuban 0.81 0.40–1.63 0.64 0.28–1.45 0.59 0.24–1.44Other Hispanic 1.14 0.69–1.89 1.27 0.82–1.95 0.81 0.40–1.66Mexican-American 1.00 1.00 1.00

Age501 years 0.43* 0.29–0.65 1.30 0.96–1.77 0.61* 0.41–0.9118–34 years 1.19 0.77–1.8335–49 years 1.00 1.00 1.00

Educational attainment,12 years 1.00 0.70–1.45 1.43 1.01–2.03 0.55 0.31–1.99$13 years 1.04 0.64–1.69 1.19 0.73–1.94 1.04 0.44–2.42High school graduate only 1.00 1.00 1.00

Labor force statusOther 0.70 0.49–1.00 0.78 0.51–1.20 1.12 0.66–1.90Employed 1.00 1.00 1.00

Marital statusNot married 0.46* 0.32–0.65 1.02 0.75–1.38 0.98 0.59–1.63Married/living as married 1.00 1.00 1.00

Annual Household Income,$20,000 1.22 0.81–1.84 0.73 0.51–1.05 1.25 0.77–2.05$$20,000 1.00 1.00 1.00

Access to servicesHealth insurance

None 0.76 0.50–1.15 0.60* 0.38–0.92 1.17 0.66–2.08Public 0.77 0.47–1.27 0.83 0.46–1.48 0.69 0.34–1.41Private 1.00 1.00 1.00

Usual source of careNo usual source 0.29* 0.21–0.40 0.48* 0.34–0.68 0.36* 0.23–1.56Has a usual source 1.00 1.00 1.00

Health Behaviors, Perceptions, and Knowledge (Respondent-Assessed)Health status

Fair, poor 0.82 0.56–1.18 1.03 0.62–1.72 0.66 0.43–1.01Good 0.91 0.63–1.32 1.24 0.87–1.78 1.03 0.67–1.58Excellent, very good 1.00 1.00 1.00

Smoking statusCurrent 1.02 0.69–1.51 0.76 0.52–1.11 0.55 0.34–0.89*Other 1.00 1.00 1.00

Other preventive testsKnows how to do BSE 1.05 1.76–3.73 1.26 0.89–1.78 1.99* 1.14–3.47Ever had CBE 2.56* — 4.38* 1.76–10.94Ever had Pap smear — — 2.77* 1.15–6.65 6.58* 2.57–16.85Ever had mammogram — — — — 3.44* 2.08–5.70

AcculturationLanguage of interview

English and Spanish 0.77 0.45–1.32 0.66 0.38–1.14 2.07 1.16–3.70Spanish 0.73 0.45–1.18 1.03 0.65–1.64 1.26 0.78–2.03English 1.00 1.00 1.00

Nativity and length of stay,10 years 1.05 0.58–1.92 1.09 0.54–2.19 0.90 0.38–2.14

$10 years 1.24 0.73–2.11 1.22 0.75–1.97 1.07 0.58–1.95Born in United States 1.00 1.00 1.00

* Significance: P , 0.05Note. The 1.00 are the referent group.Source: Pooled NHIS data in 1990 and 1992.

474 ZAMBRAN

like Mexican women, interviewed mostly in Spanish,reported higher rates of mammography screening thanMexican women. While higher education levels, longerresidence in the United States, and greater access toinsurance may promote increased adherence to screen-ing practices [10–13], our small sample of Cubanwomen may not provide adequate power to show theexpected association between more education or moreincome and higher screening rates.

Our findings offer additional evidence that Hispanicwomen confront increased socioeconomic barriers to ac-cess and health knowledge as posited in the social in-equality model [55,56]. We found a shared set of attri-butes among all subgroups, including high percentagesof women who did not complete high school (rangesfrom 48 to 77%), had incomes less than $20,000 (ranges40 to 60%), and did not have health insurance (ranges17–43%), suggesting that these socioeconomic factorsmay contribute to less knowledge about and access tohealth care which makes cancer screening less likely[1–5,28,47–50,67–70]. In a recent study, Vernes andAbrahmse (1996) found that “low income had a largernegative effect on Hispanics than on other racial/ethnicgroups with respect to high school graduation, college-going and college-continuity” [62,p. 52]. Accordingly,acculturation proxy variables that have been shown inprevious studies to be highly associated with educationand income for Hispanic groups were not significantmediators of screening practices among low-incomeHispanic women in our study [10,17,42–47]. In fact, ouranalyses showed that language is not a major barrier toscreening if access to care is available. Consistent withother studies [19,28,29], we conclude that access factorsand prior screening are more strongly associated withcurrent screening than are language and ethnic factors.Our data confirm that a disproportionate percentage ofHispanic women are low income and at risk of beingunderscreened [11,15,16]. Experiential accounts andcase studies suggest that additional barriers to screen-ing and lower screening rates exist in many low-incomeHispanic communities [36,53,69,70].

Since screening rates are higher than might be ex-pected among our Hispanic sample, we can only inferthat recent intervention strategies, including increasedpublic education and community outreach, have beeneffective in promoting cancer screening behaviors ofHispanic women [4,16,63–69]. Our findings have impli-cations for provider practices, targeted health promo-tion activities, and future research directions. For His-panic women, particularly Mexican women, not having

a usual source of care lessens the likelihood of obtaininga screening recommendation from a physician [67,71].In fact, physicians may not be aware of the lack of priorscreening among their low-income patients or, as other

A ET AL.

researchers have found, do not make screening recom-mendations due to cost concerns [35,36]. Although phy-sicians have been encouraged to take advantage of ev-ery clinical opportunity to promote regular screening,physician recommendation patterns vary, especiallywhen women are culturally different from providersand low income [65–67,71–73]. If all physicians were toinform their Hispanic patients of preventive screeningpractices, recommend they screen according to clinicalguidelines, and refer them to appropriate facilities, wewould expect a considerable increase in the use of pre-ventive screening services in Hispanic communities[69–71].

Community-based health promotion activities, par-ticularly patient education regarding cancer risks, havebeen shown to increase cancer screening practicesamong Hispanic women [63,64,68,70]. Successful inter-ventions in different regions of the country include in-volving Hispanic community members (communityhealth workers and health providers) in health educa-tion and outreach efforts; use of Spanish-language me-dia, especially public radio announcements on healthprevention; and use of existing community networks,such as churches, to promote cancer screening practices[64,76]. Pasick et al. suggest that tailoring interven-tions at the community level can move us beyond simplerace and ethnic categories to social factors that directlyinfluence behavior and health, such as living environ-ment, opportunities and barriers that affect health be-liefs, and ethnic-specific behaviors [68]. Programs tai-lored to communities in which Hispanic women areconcentrated should take into account not only level ofhealth knowledge, language preference, and life cir-cumstances, but also available community health re-sources [4,14,68–70]. In this study, despite using apooled sampled from two nationally representative sur-veys, we did not find any statistically significant effecton ethnicity using conventional measures. Thus we con-cur with Passick et al. who suggest that social factors,such as poverty, rather than ethnicity alone, shouldserve as the indicator of who needs health services, andthat ethnicity (defined as a shared dynamic culturalidentity) should be used to inform providers on whatservices to deliver and guide how services are delivered.

To continue to monitor Hispanic intergroup variationin cancer screening, we recommend that measures ofhealth-specific sociocultural behaviors that may influ-ence screening practices be included in future surveysconducted at the regional and local levels [64]. Patient–provider communication items, such as physician rec-ommendation and doctor–patient race and ethnic con-

cordance [77], should be included in future surveys,since communication with a provider has been shownto be strongly associated with increased screening forwomen of all ages and racial and ethnic groups [64–69].

E

CANCER SCREENING US

Sample sizes of Hispanic subgroups and geographic cov-erage in national surveys should be increased, as shouldadministration of surveys in language of preference,Spanish or English, by bilingual and ethnic-specific in-terviewers. We also recommend the development anduse of measures of literacy so that the association be-tween literacy, knowledge levels, and screening can betested [60].

In conclusion, community-based programs targeted

to ethnic-specific groups appear to have been effectivein increasing screening practices in various geographicregions, and should be expanded, particularly forwomen of Mexican origin, to help meet the objectivesfor Hispanic women for 2000 and beyond.

ACKNOWLEDGMENTS

We gratefully acknowledge Rich Snyder and James Cucinelli forprogramming assistance, and Barry Graubard, Ph.D., for statisti-cal consultation.

REFERENCES

1. Cancer facts and figures for minority Americans 1991. Atlanta,GA: American Cancer Society, 1991:1.

2. Cancer facts and figures 1997. Atlanta, GA: American CancerSociety, 1997.

3. Healthy People 2000 progress report for Hispanic Americans.Washington, DC: Office of Disease Prevention and Health Promo-tion, 1993.

4. Breen N, Kessler L. Changes in the use of screening mammogra-phy: evidence from the 1987 and 1990 National Health InterviewSurveys. Am J Public Health 1994;84:62–7.

5. Siegel PZ, Frazier EL, Mariolis P, Brackbill RM, Smith C. Behav-ioral risk factor surveillance, 1991: monitoring progress towardthe nation’s Year 2000 health objectives. Morb Mortal WklyRep 1993;42.

6. Montgomery PA. The Hispanic population in the United States.U.S. Bureau of the Census, Current Population Reports, SeriesP20–475. Washington, DC: U.S. Govt. Printing Office, 1994.

7. Schur CL, Bernstein AB, Berk ML. The importance of distin-guishing Hispanic subpopulations in the use of medical care.Med Care 1987;25:627–41.

8. Campell P. Population projections for states by age, race andsex: 1993–2020, U.S. Bureau of the Census, Current PopulationReports, P25-1111. Washington, DC: U.S. Govt. Printing Of-fice, 1994.

9. Massey D. Latinos, poverty, and the underclass: a new agendafor research. Hisp J Behav Sci 1993;15:449–75.

10. Vega, W.A., Amaro, H. Latino outlook: good health, uncertainprognosis. Annu Rev Public Health 1994;15:39–67.

11. Perez SM. Untapped potential: a look at Hispanic women in theUS. Washington, DC: National Council of La Raza, 1996.

12. U.S. Department of Commerce, Bureau of the Census. The na-tion’s Hispanic population—1994, Statistical Brief, SB/95-25.Washington, DC: U.S. Govt. Printing Office, 1995.

13. Ramirez-Gutierrez A, Valdez R, Carter-Pokras O. Cancer. In:Molina C, Aguirre-Molina M, editors. Latino health in the US:a growing challenge. Washington, DC: American Public HealthAssociation, 1994:211–46.

14. Ramirez AG, Villareal R, Suarez L, Flores ET. The emerging

BY HISPANIC WOMEN 475

hispanic population: a foundation for cancer prevention and con-trol. J Natl Cancer Inst 1995;18:1–9.

15. Trapido EJ, Valdez RB, Obeso JL, Strickman-Stein N, Rotger A,Perez-Stable EJ. Epidemiology of cancer among Hispanics in theUnited States. J Natl Cancer Inst 1995;18:17–28.

16. Valdez RB, Delgado DJ, Cervantes, RC, Bowler, S. Cancer in USLatino communities: an exploratory review. Santa Monica, CA:Rand, 1993.

17. Balcazar H, Castro FG, Krull JL. Cancer risk reduction in Mexi-can American women: the role of acculturation, education, andhealth risk factors. Health Educ Q 1995;22:61–84.

18. Miller AM, Champion, VL. Attitudes about breast cancer andmamography: racial, income and educational differences. WomenHealth 1997:26(1):41–63.

19. Lewin-Epstein N. Determinants of regular source of health carein black, Mexican, Puerto Rican, and non-Hispanic white popula-tions. Med Care 1991;29:543–57.

20. Wolfgang PE, Semeiks PA, Burnett WS. Cancer incidence in NewYork City Hispanics, 1982 to 1985. Ethn Dis 1991;3:263–72.

21. Makuc D, Freid VM, Parsons PE. Health insurance and cancerscreening among women. Advance Data from Vital and HealthStatistics, No. 254. Hyattsville, MD: National Center for HealthStatistics, 1994.

22. Fulton, JP, Rakowski W, Jones AC. Determinants of breast cancerscreening among inner-city Hispanic women in comparison withother inner-city women. Public Health Rep 1995;110:476–82.

23. National Center for Health Statistics. Health, United States,1995. Hyattsville, MD: Public Health Service, 1996.

24. Tortolero-Luna G, Glober GA, Villarreal R, Palos G, Linares A.Screening practices and knowledge, attitudes, and beliefs aboutcancer among Hispanic and non-Hispanic white women 35 yearsold or older in Nueces County, Texas. J Natl Cancer Inst 1995;18:49–56.

25. Calle EE, Flanders WD, Thun MJ, Martin LM. Demographicpredictors of mammography and pap smear screening in USwomen. Am J Public Health 1993;83:53–60.

26. Stein JA, Fox SA, Murata PJ. The influence of ethnicity, socioeco-nomic status, and psychological barriers on use of mammography.J Health Soc Behav 1991;32:101–13.

27. de la Torre A, Friis R, Hunter HR, Garcia L. The health insurancestatus of US Latino women: a profile from the 1982–1984HHANES. Am J Public Health 1995;86:533–7.

28. Schur CL, Leigh LA, Berk ML. Health care use by Hispanicadults: financial vs. non-financial determinants. Health CareFinan Rev 1995:17(2):71–88.

29. Richardson JL, Marks G, Solis JM, Collins LM, Birba L, HisserichJC. Frequency and adequacy of breast cancer screening amongelderly Hispanic women. Prev Med 1987:16:761–74.

30. Zapka J, Estabrook B. Breast and cervical cancers—early detec-tion and control: interpersonal strategies and channels. WellnessPerspect Res Theory Pract 1995;11(2):40–78.

31. Brown ER, Wyn R, Cumberland WG, Yu H, Abel E, GellbergL, Ng L. Women’s health-related behaviors and use of clinicalpreventive services. Los Angeles: UCLA Center for Health PolicyResearch, 1995.

32. Fox SA, Roethzheim R. Screening mammography and older His-panic women: current status and issues. Cancer 1994; 1;74(7 suppl):2028–33.

33. Suarez L, Weiss N, Rainbolt T, Pulley L. Effect of social networks

on cancer-screening behavior of older Mexican-American women.J Natl Cancer Inst 1994;86:775–9.

34. Morgan C, Park E, Cortes DE. Beliefs, knowledge, and behaviorabout cancer among urban hispanic women. J Natl CancerInst 1995;18:57–63.

476 ZAMBRAN

35. Fox, SA, Stein JA. The effect of physician-patient communicationon mammography utilization by different ethnic groups. MedCare 1991;29:1065–82.

36. Goldman DA, Simpson DM. Survey of El Paso physician’s breastand cervical cancer screening attitudes and practices. J Commu-nity Health 1994;19:75–85.

37. Stein JA, Fox SA. Language preference as an indicator of mam-mography use among Hispanic women. J Natl Cancer Inst 1990;82:1715–1716.

38. Vernon SW, Vogel VG, Halabi S, Jackson GL, Lundy RO, PetersGN. Breast cancer screening behaviors and attitudes in threeracial/ethnic groups. Cancer 1992;69:165–74.

39. Potvin L, Camirand J, Beland F. Patterns of health servicesutilization and mammography use among women aged 50–59years in the Quebec Medicare system. Med Care 1995;33:515–30.

40. Rakowski W, Rimer BK, Bryant SA. Integrating behavior andintention regarding mammography by respondents in the 1990National Health Interview Survey of health promotion and dis-ease prevention. Public Health Rep 1993;108:605–24.

41. Gutierrez-Mohamed ML. The effect of age on differences in useof Pap smear screening. Unpublished doctoral dissertation.Baltimore, MD: Johns Hopkins University, 1995.

42. Solis JM, Marks G, Garcia M, Sheldon A. Acculturation, accessto care, and use of preventive services: findings from HHANE’S1982–84. Am J Public Health 1990(suppl);80:11–19.

43. Molina C, Zambrana RE, Aquirre-Molina M. The influence ofculture, class and environment on health care. In: Molina C,Aquirre-Molina, M, editors. Latino health in the US: a growingchallenge. Washington DC, American Public Health Associa-tion, 1994.

44. Rogler LH, Cortes DE, Malgady RG. Acculturation and mentalhealth status among hispanics. Am Psychol 1991;46:585–97.

45. Cobas JA, Balcazar H, Benin MB, Keith VM, Chong Y. Accultura-tion and low birthweight infants among Latino women: a reanaly-sis of HHANES data with structural equation models. Am JPublic Health 1996;86:394–6.

46. Scribner R. Paradox as paradigm: the health outcomes of MexicanAmericans. Am J Public Health 1996:86:303–4.

47. Suarez L. Pap smear and mammogram screening in MexicanAmerican women: The effects of acculturation. Am J PublicHealth 1994;84:742–6.

48. Zambrana RE, Ellis, B. Contemporary research issues in His-panic/Latino women’s health. In: Adams DL, editor. Health issuesfor women of color: a cultural diversity health perspective. Thou-sand Oaks, CA: Sage, 1995.

49. Giachello A. Issues of access and use. In: Molina C, Aquirre-Molina M, editors. Latino health in the US: A growing challenge.Washington, DC: American Public Health Association, 1994.

50. Elder JP, Castro FP, DeMoor CC, Mayer J, Candelaria JI, Camp-bell N, et al. Differences in cancer-risk-related behaviors in Lat-ino and Anglo adults. Prev Med 1991;20:751–63.

51. Coe K, Harmon MP, Castro FG, Campell N, Mayer JA, Elder JP.Breast self-examination: knowledge and practices of Hispanicwomen in two southwestern metropolitan areas. J CommunityHealth 1994;19:433–48.

52. Perez-Stable EJ, Sabogal F, Otero-Sabogal R. Use of cancer-screening test in the San Francisco bay area: comparison of Lat-inos and Anglos. J Natl Cancer Inst 1995;18:147–53.

53. Gonzalez MO. Cancer of the cervix in the Rio Grande Valley of

south Texas. In: Jones LA, editor. Minorities and cancer. NewYork: Springer-Verlag, 1989:193–8.

54. Harlan LC, Bernstein AB, Kessler LG. Cervical cancer screening:who is not screened and why? Am J Public Health 1991;81:885–90.

A ET AL.

55. Williams DR. Socioeconomic differentials in health: a review andredirection. Soc Psyc Q 1990;53(2):81–99.

56. Adler NE, Boyce T, Chesney MA, Folkman S, Syme L. Socioeco-nomic inequalities in health. JAMA 1993;269:3140–5.

57. Leclere FB, Jenseen L, Biddlecom AE. Health care utilization,family context, and adaption among immigrants to the UnitedStates. J Health Soc Behav 1994;35:370–84.

58. Stephen EH, Foote K, Hendershot GE, Schoenborn CA. Healthof the foreign-born population: United States, 1989–90. Advancedata from Vital and Health Statistics, No. 241. Hyattsville, MD:National Center for Health Statistics, 1994.

59. SUDAAN: Survey Data Analysis Software, Release 6.34. Re-search Triangle Park, NC: Research Triangle Institute, Septem-ber 1993.

60. D’Onofrio CN, Pasick, RJ. Improving health surveys for multieth-nic populations: final report of Grant Award Number U83/CCU908697-01. Union City, CA: Northern California CancerCenter, January 14, 1997.

61. Amaro H. Using national health data systems to inform Hispanicwomen’s health. Paper presented at the National Center forHealth Statistics, Public Health Conference on Records and Sta-tistics, Washington, DC, July 19–21, 1993.

62. Vernez G, Abrahamse A. How immigrants fare in US education.Santa Monica, CA: Rand, 1996.

63. Cancer Research in Hispanic Populations in the United States.J Natl Cancer Inst 1995;18:1–169.

64. Perez Stables EJ, Otero-Sabogal R, Sabogal F, Napoles-Springer.Pathway to early cancer detection for Latinas: en accion contrael cancer. Health Educ Q 1996; 23(suppl):541–59.

65. Fox SA, Murata PJ, Stein JA. The impact of physician complianceon screening mammography for older women. Arch InternMed 1991;151:50–6.

66. Fox SA, Siu AL, Stein JA. The importance of physician communi-cation on breast cancer screening of older women. Arch InternMed 1994;154:2058–68.

67. National Cancer Institute Breast Cancer Screening Consortium.Screening mammography: a missed clinical opportunity?JAMA 1990;264:54–8.

68. Pasick RJ, D’Onofrio C, Otero-Sabogal R. Similarities and differ-ences across cultures: questions to inform a third generation forhealth promotion research. Health Educ Q 1996;23(suppl):S142–61.

69. Brown CL. Screening patterns for cervical cancer: how best toreach the unscreened population. J Natl Cancer Inst 1996;21:7–11.

70. National Coalition of Hispanic Health and Human Services Orga-nizations. Meeting the health promotion needs of Hispanic com-munities. Am J Health Promotion 1995;9(4):300–11.

71. Flack JM, Amaro H, Jenkins W, Kunitz S, Levy J, Mixson M,et al. Epidemiology of minority health. Health Psychol 1995;14:592–600.

72. Auerbach JD, Riger AE. Women’s health research: public policyand sociology. J Health Soc Behav 1995; issue:115–31.

73. Lavizzo-Mourey R, Mackenzie ER. Cultural competence: essen-tial measurements of quality for managed care organizations.Ann Intern Med 1996;10:919–20.

74. Komaromy M, Grumbach K, Drake M, Vranizan K, Lurie N,

CANCER SCREENING USE

Keane D, et al. The role of black and Hispanic physicians inproviding health care for underserved populations. N Engl JMed 1996:1305–10.

75. Cantor JC, Miles EL, Baker LC, Barker DC. Physician serviceto the underserved: implications for affirmative action in medicaleducation. Inquiry 1996,33:167–80.

BY HISPANIC WOMEN 477

76. Fox SA, Stern JA, Gonzalez RE, Farrenkopf M, Dellinger A. Atrial to increase mammography utilization among Los AngelesHispanics. J Health Care Poor Underserved 1998;9(3).

77. Gray B, Stoddard JJ. Patient–physician pairing: does racial andethnic congruity influence selection of a regular physician. JCommunity Health 1997:22(4):247–59.