NIH Public Access
Author Manuscript
Soc Sci Med. Author manuscript; available in PMC 2009 March 1.
NIH-PA Author Manuscript
Published in final edited form as:
Soc Sci Med. 2008 March ; 66(5): 1204–1216.
Provider and Clinic Cultural Competence in a Primary Care Setting
Kathryn A Paez, PhD(c), MBA, RN,
Johns Hopkins University School of Nursing, Baltimore, Maryland UNITED STATES
Jerilyn K Allen, RN, ScD, FAAN,
Johns Hopkins University School of Nursing, Baltimore, Maryland, jallen@son.jhmi.edu
Kathryn A Carson, ScM, and
Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore,
Maryland, kcarson@jhmi.edu
NIH-PA Author Manuscript
Lisa A Cooper, MD, MPH
Departments of Epidemiology, and Health, Behavior and Society, Johns Hopkins Bloomberg School
of Public Health; Division of General Internal Medicine, Department of Medicine, Johns Hopkins
University School of Medicine; Welch Center for Prevention, Epide, lisa.cooper@jhmi.edu
Abstract
A multilevel approach that enhances the cultural competence of clinicians and healthcare systems is
suggested as one solution to reducing racial/ethnic disparities in healthcare. The primary objective
of this cross-sectional study was to determine if there is a relationship between the cultural
competence of primary care providers and the clinics where they work. Forty-nine providers from
23 clinics in Baltimore, Maryland and Wilmington, Delaware, USA. completed an on-line survey
which included items assessing provider and clinic cultural competence. Using simple linear
regression, it was found that providers with attitudes reflecting greater cultural motivation to learn
were more likely to work in clinics with a higher percent of nonwhite staff, and those offering cultural
diversity training and culturally adapted patient education materials. More culturally appropriate
provider behavior was associated with a higher percent of nonwhite staff in the clinic, and culturally
adapted patient education materials. Enhancing provider and clinic cultural competence may be
synergistic strategies for reducing healthcare disparities.
Keywords
NIH-PA Author Manuscript
USA; cultural competence; disparities; primary care providers; health organizations; healthcare
Introduction
In recent years, racial and ethnic disparities in health status and the delivery of healthcare have
come to the forefront of healthcare research and policy. These inequities have been documented
and summarized in numerous publications, most notably the Institute of Medicine (Smedley,
Stith, & Nelson, 2003). As the evidence of poorer minority health and treatment has
accumulated, the emphasis of public policy and research initiatives has shifted from further
Corresponding Author: Ms. Kathryn A Paez, Johns Hopkins University, Baltimore, Maryland, UNITED STATES,
kpaez1@son.jhmi.edu.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting
proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could
affect the content, and all legal disclaimers that apply to the journal pertain.
Paez et al.
Page 2
cataloging the problems to identifying and fostering the implementation of effective strategies
to remedy disparities.
NIH-PA Author Manuscript
Both the IOM report and the National Standards for Culturally and Linguistically Appropriate
Services (CLAS), promulgated by the Office of Minority Health (2000), recommend, as one
among a number of strategies, a multilevel approach that enhances the cultural competence of
clinicians and healthcare systems to improve racial and ethnic minority health (Smedley et al.,
2003). Since publication of the CLAS Standards, there have been significant efforts by
government and private organizations to provide payers and providers with user-friendly action
plans and tools to improve cultural competence in the delivery of healthcare. Elevating the
cultural competence of clinicians and health care systems, however, is not a straightforward
task. Cultural competence is a challenging term to define, making it difficult for the average
healthcare organization to independently develop and execute an effective action plan for
improvement. What constitutes cultural competence can vary by healthcare organization,
provider-type, organizational and community resources, and patient populations (Office of
Minority Health, 2000)
NIH-PA Author Manuscript
Cultural competence has been broadly defined as “a set of congruent behaviors, attitudes, and
policies that come together in a system, agency or among professionals that enable effective
work in cross-cultural situations” (Cross, Bazron, Dennis, & Isaacs, 1989). Culture refers to
the integrated patterns of human behavior that include language, thoughts, communications,
actions, customs, beliefs, values and institutions of racial, ethnic, religious or social groups
(Office of Minority Health, 2000), while competence signifies possession of functionally
adequate knowledge, judgment, practical and thinking skills to perform in a desired way
(Merriam-Webster, 2002).
NIH-PA Author Manuscript
Culturally competent providers and organizations possess the knowledge, attitude and skills
to overcome their own inherent barriers to quality minority care such as biases (e.g. racial/
ethnic prejudices, perceived lack of time, and yielding to seemingly overwhelming patient
social problems ), and service inaccessibility (e.g. inconvenient location, limited appointment
availability and lack of care coordination). In addition, culturally competent providers and
organizations develop approaches to compensate for patient characteristics that hinder the
patient’s ability to benefit from healthcare services. Patient obstacles to care include limited
English proficiency (LEP), low health literacy, fears (e.g. mistreatment, avoidance of
stigmatizing or grave diagnoses, and deportation), beliefs that preempt treatment (e.g. mistrust,
and aversion to medications, invasive treatment and preventive care), lack of knowledge
(understanding of health, management of acute and chronic illness, and Western healthcare
delivery norms and practices), lack of resources (e.g. insurance, funds for out-of-pocket
expenses, and transportation) and inability to leave their place of employment to attend medical
appointments (Martinez & Carter-Pokras, 2006). Cultural competence begins with
understanding the strengths and weaknesses of the healthcare organization and providers, and
the unique needs of the population being served. It is a process of increasing proficiency gained
from informal and formal cross-cultural experiences rather than an endpoint that is achieved
(Cross et al., 1989)
Our understanding of cultural competence in healthcare is in the formative stages. Limited
empirical research has methodically investigated the interdependence between clinician and
organizational cultural competence. The objectives of this study were to measure the cultural
competence of primary care providers and their perceptions of the cultural competence of the
clinics where they work; determine if there is a relationship between provider and clinic cultural
competence; and assess whether provider demographic characteristics were associated with
the cultural competence of providers. We hypothesized that there would be a positive
relationship between provider and clinic cultural competence and that compared to their
Soc Sci Med. Author manuscript; available in PMC 2009 March 1.
Paez et al.
Page 3
counterparts, providers who were women, liberally oriented and family physicians would have
knowledge, attitudes and behaviors that were more indicative of cultural competence.
NIH-PA Author Manuscript
Methods
Study design
We conducted a cross-sectional survey of clinicians participating in two clinical trials of
interventions to enhance patient-provider communication. The studies were not meant to
directly improve cultural competence. The study protocol was approved by the Institutional
Review Board, and all participants provided informed consent.
Study setting and population
NIH-PA Author Manuscript
The study took place in 23 community based, primary care clinics located in Baltimore,
Maryland (22 clinics) and Wilmington, Delaware (1 clinic) serving predominantly black (60%
to 100%) and white populations from a range of socioeconomic backgrounds. The clinics
represent 11 organizations including a federally qualified health center, community health
centers, university hospital-affiliated outpatient centers, and independent multi-specialty
practices. The study population consisted of 64 family and internal medicine physicians and 5
adult/family nurse practitioners who volunteered to participate in one of two randomized
clinical trials. Providers were eligible to participate in the trials if they practiced at least 20
hours per week and planned to stay at the clinical site for at least 12 months after enrollment.
Enrollment and data collection
During the post-intervention periods of the clinical trials, providers were invited by letter to
complete an Internet survey that was divided into three sections: questions about their clinic,
practice and experiences; questions about cultural issues in healthcare and knowledge, attitudes
and opinions about race and medical care; and two cognitive tests. Providers were told that the
goal of the study was to learn whether it is possible to administer a computerized test to
providers in order to measure their implicit attitudes (unconscious biases) toward race/
ethnicity. Provider characteristics, including race, were collected by self-report on enrollment
in the clinical trials or on the post-intervention Internet-survey.
Provider cultural competence measure
NIH-PA Author Manuscript
The operational framework for the provider cultural competence measure was drawn from the
CLAS Standards (Office of Minority Health, 2000) and the “Process of Cultural Competence
in the Delivery of Healthcare Services”, an operational model of cultural competence
(Campinha-Bacote, 2002). Provider cultural competence was divided into three conceptual
components: attitudes, self-reported behavior and knowledge. No one instrument could be
found in the literature that had established validity and reliability, and was appropriate for use
with practicing primary care physicians caring for inner city black patients (Gozu, Beach, Price,
Gary, Robinson, Palacio et al., 2007). Cultural competence instruments were reviewed to
identify subsets of items that best reflected the cultural needs of black and white patients in a
primary care setting and this study’s model for operationalizing cultural competence. (See
Gozu, 2007, for a list of the major instruments that were reviewed.) All items were evaluated
and revised in a multi-stage review process whereby experts in health disparities research made
recommendations to establish face validity of the item set. Items were pre-tested with eight
clinicians. Predictive validity of the measures was evaluated by examining the relationship of
the measures with provider characteristics that have been associated with the cultural
competence concepts by research.
Soc Sci Med. Author manuscript; available in PMC 2009 March 1.
Paez et al.
Page 4
NIH-PA Author Manuscript
Six items representing the attitude and awareness domains were selected from Dogra’s Cultural
Awareness Questionnaire and Godkin’s Modified-Cultural Competence Self-Assessment
Questionnaire (Dogra, 2001; Dogra & Stretch, 2001; Godkin & Savageau, 2001). Preference
was given to items that best captured the motivation to understand, accept and respect
differences; appreciation of other cultures; and the awareness of societal impact on
opportunities related to race (see Table 1). Because immigrants represent a very small
percentage of the patient population for these studies, items that concerned immigrant issues
were not included. Items reflecting controversial political perspectives were avoided.
Revisions were made to the wording of the items so that questions were personalized rather
than describing what a physician in general would think, and the items were placed within the
context of race rather than society in general or culture. A five-point Likert scale measuring
level of agreement (strongly disagree to strongly agree) was retained as the response set.
NIH-PA Author Manuscript
The Cultural Competence Assessment Instrument (CCA) was selected as the source for items
reflecting the behavior domain (Schim, Doorenbos, & Borse, 2005). Two strengths of the CCA
are that it assesses self-report of actual behavior rather than self-efficacy for performing a
behavior and it has been demonstrated to be reliable and valid (Doorenbos, Schim, Benkert,
& Borse, 2005; Schim et al., 2005; Schim, Doorenbos, Miller, & Benkert, 2003). Items were
selected from the CCA that had eigenvalue scores greater than .600 and that measured the
adaptation of care to meet the needs and expectations of diverse patients. If there were two
items that appeared to be similar, the item that was most relevant to the primary care setting
was selected. Five items were retained for measuring behavior (see Table 1). Since the CCA
was designed for a multidisciplinary setting (specifically hospice care), items were reworded
to be meaningful for primary care providers. The response set was modified from a five-point
Likert-type scale (always, often, at times, never, not sure) to a frequency format thus
standardizing the meaning to all respondents and making the response choices more exact.
Respondents were asked to select the percent of time (0–25, 26–50, 51–75, 76–100) that they
perform behaviors indicative of seeking culture-related information about patients; obtaining
feedback regarding their cross-cultural interaction skills; and adapting care to patients’ culture
and social situation (see Table 1).
NIH-PA Author Manuscript
Exploratory factor analysis was performed with cultural competence items grouped by attitude
or behavior using varimax rotation. Factors were retained that had an eigenvalue of ≥ 1.0. An
item was discarded if it was the only item that loaded on a factor or if there was limited variance
in response among the study participants (i.e. ≥ 90% of responses were in the same response
category). All but one of the 11 cultural competence items exhibited adequate variability of
response -- 90% of providers strongly agreed with a statement concerning equal access to
healthcare regardless of race or social class. This item was eliminated from the cultural attitude
item set. Bartlett’s Test of Sphericity (attitude, p=0.008; behavior, p≤0.001) indicated adequate
level of correlation to have an identity matrix, supporting the use of factor analysis.
The Kaiser-Meyer-Olkin (KMO) test evaluated adequacy of sample size. The KMO scores for
attitude (.618) and behavior (.669) met the minimum standard of .5 for continuing with factor
analysis (Hutcheson & Sofroniou, 1999). The sampling adequacy of individual variables was
tested using the anti-image correlation matrix. All variable values were above .5 indicating that
it was appropriate to continue with factor analysis using all of the items (Field A., 2005).
Communalities for each of the 2 categories ranged from .22 to .72 with 7 of 10 items scoring
above .40. The attitude items grouped into two factors explaining 59% of the variance while
behavior grouped into one factor explaining 42% of the variance. Items loaded from .52 to .
84 (attitude) and .47 to .83 (behavior) on the factors. The two attitude factors were interpreted
as the ‘motivation’ to learn about cultures within the provider’s practice and society, and
‘power/assimilation’ attitudes signifying awareness of white advantage and acceptance of a
Soc Sci Med. Author manuscript; available in PMC 2009 March 1.
Paez et al.
Page 5
NIH-PA Author Manuscript
racial group’s choice to retain distinct customs and values (see Table 1). Behavior was retained
as one factor and continued to represent self-reported frequency of seeking culture-related
information about patients; obtaining feedback regarding one’s cross-cultural interaction skills;
and adapting care to patients’ culture and social situation.
The Cronbach’s coefficients ranged from .50 to .64 for the attitude and behavior measures (see
Table 1). According to Nunnally (1967), these statistics are sufficient for preliminary research.
“In the early stages of research on predictor tests or hypothesized measures of a construct, one
saves time and energy by working with instruments that have only modest reliability, for which
purpose reliabilities of .60 or .50 will suffice” (Nunnally, 1967, p.226). Composite scores for
the measures were developed by averaging the item scores. A higher score indicates greater
cultural competence.
Knowledge was assessed using ten fact-based, multiple-choice items that were developed
drawing on the expertise of the investigators and content presented in the “Provider’s Guide
to Quality & Culture” (2004), an Internet based learning module funded by the U.S. Department
of Health and Human Services. The multiple choice items assessed knowledge of variations
in patterns of disease, risk factors and treatment, underlying factors creating disparities, and
professional and legal responsibilities related to care of minority groups. The knowledge score
consisted of the percent of items answered correctly.
NIH-PA Author Manuscript
Clinic cultural competence measure
The CLAS Standards were used as a guide for developing items to measure provider perception
of their clinic’s cultural competence (Office of Minority Health, 2000). Providers were asked
to respond to six questions concerning clinic provider and staff racial/ethnic diversity,
sponsorship of cultural diversity training for providers and staff, and provision of patient
education materials tailored to race/ethnicity or language.
Statistical analyses
Descriptive statistics characterized provider cultural competence and personal attributes, and
clinic cultural competence measures. Prior to analysis, negative items on the attitude measure
were reverse coded so a higher score indicated a more favorable attitude. Hispanic and East
Indian providers comprised a small sample so they were combined into an “other” category
while black, white, and Asian providers were classified in distinct race categories.
NIH-PA Author Manuscript
To evaluate the reliability of the clinic cultural competence measure, provider’s responses to
items evaluating clinic cultural competence were grouped by clinic. For dichotomous measures
(clinic sponsored cultural diversity training and patient education materials), the percent of
provider responses consistent with the clinic majority was calculated. There was 65%
consistency in provider response within clinics to the diversity training measure. Providers
were consistent 91% of the time in reporting the availability of adapted patient education
materials. Responses identifying the percent of white/nonwhite providers and staff were
deemed consistent if they were within 20% of the mean response for the clinic. Eighty three
percent of providers reported the percentage of nonwhite staff and providers within this
established range
Simple linear regression was used to determine if provider cultural competence varied by
provider characteristics and the provider’s report of their clinic’s cultural competence
characteristics. Multivariate linear regression was conducted including only those clinic and
provider characteristics that were significantly related to provider cultural competence as the
independent variables. Resistant linear regression models were fit if needed to reduce the
effects of extreme y-variable data values. A two-sided P value < 0.05 was considered
Soc Sci Med. Author manuscript; available in PMC 2009 March 1.
Paez et al.
Page 6
significant. Statistics were performed using STATA 8.2 for all descriptive and regression
statistics (STATA, 2003) and SPSS 14.0 for all psychometric analysis (SPSS, 2006).
NIH-PA Author Manuscript
Results
Provider characteristics
Forty-nine (71%) of the 69 providers enrolled in the clinical trials completed the cultural
competence portion of the Internet survey. Participants and non-participants were similar by
age, gender, race, specialty and years post-residency training, but participants were more likely
to be board certified than non-participants (98% vs. 80%, p<0.01). Participating providers
included 36 internists, 9 family physicians, and 4 nurse practitioners who tended to be
experienced, female and racially diverse (see Table 2). Approximately one-third of providers
reported ever attending cultural competence or diversity training, and of those attending
training, almost two-thirds rated the training as good to excellent. Providers reported seeing
more patients that were racially or ethnically discordant from themselves than concordant.
Over half of providers were very confident in caring for ethnic minorities and in providing care
to disadvantaged patients.
Provider cultural competence
NIH-PA Author Manuscript
Results from the attitude category showed that providers tended to strongly agree with
statements concerning the motivation to learn about cultures within their practice and society
while they responded more neutrally when asked their level of agreement with power/
assimilation statements concerning white advantage and conformity to white customs and
values (see Table 1). Providers practiced behaviors indicative of cultural competence (seeking
culture-related information about patients; obtaining feedback regarding one’s cross-cultural
interaction skills; and adapting care to patients’ culture and social situation) between 26%–
50% and 51%–75% of the time. The mean percent of knowledge items answered correctly was
approximately 80% (±13.74).
Clinic cultural competence
According to providers, their clinics’ providers and staff were racially diverse with the largest
nonwhite group being black for both groups (see Table 3). The majority of respondents reported
that their clinics offered diversity training to providers and staff, and patient education materials
tailored to race/ethnicity or language.
Association of provider cultural competence with clinic cultural competence
NIH-PA Author Manuscript
Attitude—Providers reporting a higher percent of nonwhite staff in their clinics, clinic
sponsorship of cultural diversity training for staff and physicians and the availability of adapted
patient education materials were more likely to have attitudes reflecting the motivation to learn
about cultures within their practice and society (see Table 4). No statistically significant
relationships were found between clinic cultural competence measures and power/assimilation.
Behavior—Providers who reported a higher percent of nonwhite staff in their clinics and the
availability of culturally and linguistically tailored patient education materials had increased
frequency of cultural competence behaviors than their counterparts (see Table 4).
Knowledge—Clinic cultural competence measures were not associated with clinical and
professional knowledge relevant to minority patient care (see Table 4).
Soc Sci Med. Author manuscript; available in PMC 2009 March 1.
Paez et al.
Page 7
Association of provider cultural competence with provider characteristics
NIH-PA Author Manuscript
Attitude—As hypothesized, women, liberals, and family physicians were more likely to agree
with power/assimilation statements reflecting cultural competence (see Table 5). Providers
who were very confident in caring for the disadvantaged were more likely to express culturally
competent power/assimilation attitudes than providers who were less than somewhat confident
in caring for the disadvantaged. In contrast, provider characteristics were not associated with
cultural motivation.
Behavior—Provider characteristics were generally not associated with behavior, although
black providers reported a greater frequency of culturally competent behavior than whites (see
Table 5).
Knowledge—Younger providers tended to score higher on the knowledge items than older
providers (see Table 5). Providers expressing less confidence in caring for minorities and
international medical graduates (IMGs) tended to have lower knowledge scores than their
counterparts (see Table 5).
Multivariate models
NIH-PA Author Manuscript
Multivariate models were developed for each of the four cultural competence measures (see
Table 6). The models included provider and clinic characteristics that were significantly
(p<0.05) associated with a provider cultural competence measure in the univariate analyses.
All four models as a whole were significant (p<0.01). Nonwhite staff (cultural motivation and
behavior models), liberal political orientation (power/assimilation model), and age and
confidence in caring for the disadvantaged (knowledge model) remained significant in the
multivariate analysis. All other provider and clinic characteristics were more weakly associated
with their respective provider cultural competence measure and had p-values ≥0.05 when
considered in the extended model.
Discussion
NIH-PA Author Manuscript
Our study is one of the first to link provider cultural competence with the cultural competence
of the clinics where they work. Our findings indicate that primary care providers who reported
that their clinics had adopted recommendations made in the CLAS standards were more likely
to have attitudes and behaviors that were culturally competent. However, this relationship
between providers and the clinics did not hold true for knowledge relevant to caring for a racial
minority population. This suggests that enhancing provider cultural competence (attitudes and
behaviors) and clinic cultural competence may be synergistic strategies for reducing healthcare
disparities.
There are at least three potential interpretations of the results from this cross-sectional study
that should be considered. First, clinics that have adopted more culturally competent practices
may influence providers to develop more culturally competent attitudes and behavior. Second,
more culturally competent providers may be attracted to work in clinics with a higher level of
cultural competence. Finally, providers with culturally competent attitudes and behaviors could
influence the cultural competence of the clinics where they practice by advocating for diversity
training, workforce diversity and use of culturally-appropriate patient education materials. Any
combination of these three conditions could exist within a given clinical setting.
The effect size linking a more diverse support staff with provider cultural competence was
small. Social learning theory and the attraction-selection-attrition cycle, however, supports the
existence of this relationship in the varying directions proposed. Social learning theory asserts
learning is affected by observing the behavior, attitudes and emotional reactions of others and
Soc Sci Med. Author manuscript; available in PMC 2009 March 1.
Paez et al.
Page 8
NIH-PA Author Manuscript
by personal experience. This effect is optimized when there is institutional support (Pettigrew
& Tropp, 2006). Support staff of color may increase provider awareness of cultural issues and
informally set the standards for behavior in interacting with racially discordant patients,
creating an atmosphere that is welcoming or not (Chrisman, 2007). They may also assist with
interpretation of patient interactions and advise providers on ways to more effectively interact
with their racially discordant patients (Kairys & Like, 2006; Pacquiao, 2007). Providers’
frequent contact with support staff in the provision of routine patient care and more limited
time to interact with colleagues during busy clinic hours may explain why a relationship was
seen with staff diversity and not provider diversity. Surrounding healthcare providers with a
diverse support staff may be one way to compensate for the short supply of minority healthcare
providers and to improve the delivery of culturally competent healthcare.
An alternative explanation for the association between staff diversity and provider cultural
competence is that providers who are more culturally competent may self-select into more
culturally competent organizations. Schneider’s attraction-selection-attrition cycle expands
upon this interpretation by suggesting not only that people are attracted to organizations based
on compatibility but organizations select people with common attributes (Schneider, Goldstein,
& Smith, 1995). Organizational homogeneity is perpetuated by people resigning or being fired
when they do not fit the work environment.
NIH-PA Author Manuscript
The availability of clinic sponsored cultural diversity programs was associated with provider
cultural motivation. Clinic sponsorship of such programs may serve as a surrogate for less
visible clinic characteristics that are related to providers’ motivation to learn about cultures
within their practice and society. In contrast, the availability of cultural diversity programs was
not associated with provider power/assimilation attitudes, culturally competent behavior, or
clinical and professional knowledge relevant to caring for minority groups.
The failure to see a relationship between attendance at a cultural competence training programs
(whether clinic sponsored or not) and provider cultural competence may reflect a lack of
standardization in training. The literature suggests that such training programs vary
considerably in curricular focus, teaching methods, and depth of experience (Beach, Price,
Gary, Robinson, Gozu, Palacio et al., 2005). It is possible, for example, that providers in this
study attended programs that only raised awareness of the importance of culture in the care of
patients but did not include experiential learning or strategies that build skills. Didactic
continuing education programs alone tend not to be effective in improving professional practice
(Cauffman, Forsyth, Clark, Foster, Martin, Lapsys et al., 2002; Davis, Thomson, Oxman, &
Haynes, 1995).
NIH-PA Author Manuscript
The provision of patient education materials adapted to the language or culture of the patients
served in a clinic was associated with provider cultural motivation and culturally competent
behaviors in the univariate analyses. Patient education materials tailored to culture and
language may serve as cues to providers to take a complete history that includes cultural factors
and to incorporate cultural issues important to the patient into the plan of care. In a synthesis
of the literature, Davis et al. (1995) concluded that such patient-mediated interventions are
quite effective at inducing behavior change in providers.
Knowing a provider group’s characteristics may provide some guidance to better tailoring
cultural competence program content to the needs of the target group. In this study, provider
characteristics were associated with some but not all of the cultural competence concepts.
Specifically, providers who were younger, female, politically liberal, a family physician, a U.S.
medical graduate were more likely to have attitudes and/or knowledge consistent with cultural
competence.
Soc Sci Med. Author manuscript; available in PMC 2009 March 1.
Paez et al.
Page 9
NIH-PA Author Manuscript
The relationship found between provider characteristics and power/assimilation attitudes is
consistent with the literature and supported our hypotheses. In this study, female, family
practice and liberal providers had power/assimilation attitudes that were more empathetic
toward the challenges faced by minority racial groups living in a Euro-Caucasian dominated
society. The association between female gender and a more positive attitude toward treating
minorities has also been found in studies with medical students (Crandall, George, Marion, &
Davis, 2003; Crandall, Reboussin, Michielutte, Anthony, & Naughton, 2007; Gurung & Mehta,
2001). Female physicians are more likely to attend to the psychosocial needs of patients and
to encourage patient participation.(Roter & Hall, 2004). More broadly, men, irrespective of
race or ethnicity, score higher than women on social dominance orientation, a measure of the
degree to which individuals desire and support group-based hierarchy and the dominance of
‘superior’ groups over ‘inferior’ groups (Pratto, Sidanius, Stallworth, & Malle, 1994; Sidanius,
Pratto, & Bobo, 1994).
NIH-PA Author Manuscript
The finding that family physicians tended to have culturally competent power/assimilation
attitudes is consistent with the founding principles of the family medicine movement, which
emphasize providing healthcare to the underserved and providing humanistic care (Stephens,
1998). In one study, medical students choosing family medicine as a career were more likely
to demonstrate a societal commitment and less likely to be concerned with personal prestige
(Wright, Scott, Woloschuk, Brenneis, & Bradley, 2004). According to some scholars, liberal
political orientation which indicates support for egalitarianism and the rights of minorities, and
the acceptance of culturally competent power/assimilation attitudes are synonymous
(Kerlinger, 1984). The belief that disadvantage such as poverty is largely mediated by societal
forces rather than factors within the control of the individual underlies liberal but not
conservative ideology (Gaertner, 1973).
Providers that were more confident in caring for disadvantaged patients tended to exhibit
attitudes reflecting understanding of the implications of white privilege and acceptance of other
cultural groups’ customs and values. It may be that providers who are comfortable providing
healthcare to patients who are less fortunate than themselves are those with the capacity to be
open-minded towards persons from diverse social backgrounds. This characteristic underlies
cultural humility, which is the ability to check the power imbalance in the provider-patient
relationship, practice ongoing self-reflection and self-critique, and seek to understand the
patient’s unique perspective (Tervalon & Murray-Garcia, 1998).
NIH-PA Author Manuscript
The incongruity between confidence in caring for minorities and knowledge may be explained
by students’ and less experienced clinicians’ blind spots about their lack of skills and abilities.
In one study, nursing students scored higher than experienced hospital and public health nurses
on a scale measuring confidence in performing transcultural nursing skills (Bernal & Froman,
1987). In another study, senior nursing students who received some cultural course content felt
less confident in providing culturally sensitive care than similar students not exposed to the
training (Alpers & Zoucha, 1996).
The inverse relationship of provider age with knowledge is consistent with curriculum trends
in U.S. medical and nursing schools and the literature. In response to the increasing cultural
diversity of the population, U.S. medical schools have increased their efforts to include
multicultural issues in their curricula. Comparison of studies conducted in 1978 and then in
2000 found a 45% increase (from 60% to 95%) in the integration of socio-cultural issues into
the curricula (Flores, Gee, & Kastner, 2000; Wyatt, Bass, & Powell, 1978). Similar studies of
U.S. nursing school curricula could not be found, but there has been an increasing number of
publications in the nursing literature exploring the content, teaching methods, evaluation and
faculty qualifications needed to ensure that nursing programs adequately prepares nurses to
meet the needs of culturally diverse patient populations.
Soc Sci Med. Author manuscript; available in PMC 2009 March 1.
Paez et al.
Page 10
NIH-PA Author Manuscript
Limitations of this study should be noted. First, a standardized measure of provider cultural
competence was not used. Although there are a number of measures of provider cultural
competence in the literature, few measures have been psychometrically tested (Gozu et al.,
2007) and none fit our provider and patient population, and our operational definition of cultural
competence. While our provider cultural competence measure has not been extensively tested,
it advances earlier studies of cultural competence in practice settings which have relied on
more limited measures such as providers’ ability to speak a second language, attendance at
cultural competence programs or global self-ratings cultural competence (Fernandez,
Schillinger, Grumbach, Rosenthal, Stewart, Wang et al., 2004; Mazor, Hampers, Chande, &
Krug, 2002; Wade & Bernstein, 1991)
Second, objective and comprehensive measures of clinic cultural competence would have
strengthened the study. Although the researchers were unable to collect objective data,
providers within clinics were reasonably consistent in their response to the clinic cultural
competence items. The clinic characteristics studied were limited to those that were visible to
providers in their daily work and did not include underlying organizational infrastructure (e.g.
governing board diversity, community and organizational assessment and the collection of race
and ethnicity data), as identified in the CLAS standards. The inclusion of these infrastructure
characteristics, indicative of organizational cultural competence, in future studies would be
informative.
NIH-PA Author Manuscript
Third, the sample size may have underpowered the study to detect relationships of provider
and clinic characteristics with provider cultural competence. A number of provider and clinic
characteristics were close to but did not meet the criteria for significance of p < .05. Although
our sample size was small, it did exceed the sample size of comparable studies examining
provider cultural competence in practice settings
NIH-PA Author Manuscript
In conclusion, we believe that the recommendations of the Institute of Medicine and the CLAS
Standards to take a multilevel approach to reducing health care disparities by improving
cultural competence at the clinic and the provider levels are prudent. Our study indicates that
an interrelationship between provider and clinic cultural competence exists. Healthcare
organizations can support clinicians in better caring for their patients of color by adopting
practices that increase provider and staff awareness of the cultural needs of patients and
integrate cultural competence into the daily work. Healthcare organizations must recognize
that brief cultural diversity training programs are unlikely to induce changes in attitude or
behavior without the implementation of additional organizational supports. Because they often
have influence over their practice environments, providers should be taught not only ways of
improving their own care to minority patients but also what ongoing support they should expect
from the organizations where they work. Cultural competence evolves from a concerted effort
by both clinicians and healthcare organizations to identify and remove the barriers impeding
quality care to vulnerable populations.
Fourth, the cross-sectional design of this study limits our ability to draw causal inferences.
Quantitative research is needed to understand in greater depth the nature of the relationships
that were found. A randomized trial or longitudinal study might help to determine the temporal
direction of the relationships between provider and clinic cultural competence.
Fifth, the generlizability of study findings to primary care providers working in small group
and solo practices, and in rural settings is not known. The perspectives and experiences of
providers choosing to work in these settings may differ from providers choosing to practice in
urban and suburban locations, larger group practices or clinics with a principal mission of
serving the disadvantaged. Differences in provider and clinic characteristics could alter the
association between provider and clinic cultural competence.
Soc Sci Med. Author manuscript; available in PMC 2009 March 1.
Paez et al.
Page 11
Acknowledgements
NIH-PA Author Manuscript
Financial support for this study was provided by grants from Agency for Healthcare Quality and Research
(R01HS013645), National Heart Lung and Blood Institute (R01HL069403 and K24HL083113), National Institute of
Nursing Research (FR31NR009889-01, T32NR07968), and the Fetzer Foundation
Reference List
NIH-PA Author Manuscript
NIH-PA Author Manuscript
Alpers RR, Zoucha R. Comparison of cultural competence and cultural confidence of senior nursing
students in a private southern university. Journal of Cultural Diversity 1996;3(1):9–15. [PubMed:
8788835]
Beach MC, Price EG, Gary TL, Robinson KA, Gozu A, Palacio A, Smarth C, Jenckes MW, Feuerstein
C, Bass EB, Powe NR, Cooper LA. Cultural competence: A systematic review of health care provider
educational interventions. Medical Care 2005;43(4):356–373. [PubMed: 15778639]
Bernal H, Froman R. The confidence of community health nurses in caring for ethnically diverse
populations. Image: Journal of Nursing Scholarship 1987;19(4):201–203.
Campinha-Bacote J. The process of cultural competence in the delivery of healthcare services: a model
of care. Journal of Transcultural Nursing 2002;13(3):181–184. [PubMed: 12113146]
Cauffman JG, Forsyth RA, Clark VA, Foster JP, Martin KJ, Lapsys FX, Davis DA. Randomized
controlled trials of continuing medical education: what makes them most effective? Journal of
Continuing Education in the Health Professions 2002;22(4):214–221. [PubMed: 12613056]
Chrisman NJ. Extending cultural competence through systems change: academic, hospital, and
community partnerships. J Transcult Nurs 2007;18(Suppl 1):68S–76S. 68S–76S. [PubMed:
17204815]
Crandall SJ, George G, Marion GS, Davis S. Applying theory to the design of cultural competency
training for medical students: a case study. Academic Medicine 2003;78(6):588–594. [PubMed:
12805037]
Crandall SJ, Reboussin BA, Michielutte R, Anthony JE, Naughton MJ. Medical students’ attitudes toward
underserved patients: a longitudinal comparison of problem-based and traditional medical curricula.
Advances in Health Sciences Education, Theory and Practice 2007;12(1):71–86. [PubMed: 17041814]
Cross, TL.; Bazron, BJ.; Dennis, KW.; Isaacs, MR. Toward a culturally competent system of care.
Washington, D.C.: Georgetown University Child Development Center; 1989.
Davis DA, Thomson MA, Oxman AD, Haynes RB. Changing physician performance. A systematic
review of the effect of continuing medical education strategies. Journal of the American Medical
Association 1995;274(9):700–705. [PubMed: 7650822]
Dogra N. The development and evaluation of a programme to teach cultural diversity to medical
undergraduate students. Medical Education 2001;35(3):232–241. [PubMed: 11260446]
Dogra N, Stretch D. Developing a questionnaire to assess student awareness of the need to be culturally
aware in clinical practice. Medical Teacher 2001;23(1):59–64. [PubMed: 11260742]
Doorenbos AZ, Schim SM, Benkert R, Borse NN. Psychometric evaluation of the cultural competence
assessment instrument among healthcare providers. Nursing Research 2005;54(5):324–331.
[PubMed: 16224318]
Fernandez A, Schillinger D, Grumbach K, Rosenthal A, Stewart AL, Wang F, Perez-Stable EJ. Physician
language ability and cultural competence. An exploratory study of communication with Spanishspeaking patients. Journal of General Internal Medicine 2004;19(2):167–174. [PubMed: 15009796]
Field, A. Discovering Statistics Using SPSS. London: Sage; 2005.
Flores G, Gee D, Kastner B. The teaching of cultural issues in U.S. and Canadian medical schools.
Academic Medicine 2000;75(5):451–455. [PubMed: 10824769]
Gaertner SL. Helping behavior and racial discrimination among liberals and conservatives. Journal of
Personality and Social Psychology 1973;25(3):335–341.
Godkin MA, Savageau JA. The effect of a global multiculturalism track on cultural competence of
preclinical medical students. Family Medicine 2001;33(3):178–186. [PubMed: 11302510]
Gozu A, Beach MC, Price EG, Gary TL, Robinson K, Palacio A, Smarth C, Jenckes M, Feuerstein C,
Bass EB, Powe NR, Cooper LA. Self-administered instruments to measure cultural competence of
Soc Sci Med. Author manuscript; available in PMC 2009 March 1.
Paez et al.
Page 12
NIH-PA Author Manuscript
NIH-PA Author Manuscript
NIH-PA Author Manuscript
health professionals: a systematic review. Teaching and Learning in Medicine 2007;19(2):180–190.
[PubMed: 17564547]
Gurung RA, Mehta V. Relating ethnic identity, acculturation, and attitudes toward treating minority
clients. Cultural Diversity and Ethnic Minority Psychology 2001;7(2):139–151. [PubMed:
11381816]
Hutcheson, GD.; Sofroniou, N. The Multivariate Social Scientist. London: Sage; 1999. Factor Analysis;
p. 224-225.
Kairys JA, Like RC. Caring for diverse populations: do academic family medicine practices have CLAS?
Family Medicine 2006;38(3):196–205. [PubMed: 16518738]
Kerlinger, F. Liberalism and Conservatism: The Nature and Structure of Social Attitudes. Hillsdale, NJ:
Lawrence Erlbaum Associates, Inc; 1984. Liberalism and conservatism; p. 13-25.
Martinez IL, Carter-Pokras O. Assessing health concerns and barriers in a heterogeneous Latino
community. Journal of Health Care for the Poor and Underserved 2006;17(4):899–909. [PubMed:
17242537]
Mazor SS, Hampers LC, Chande VT, Krug SE. Teaching Spanish to pediatric emergency physicians:
effects on patient satisfaction. Archives of Pediatric and Adolescent Medicine 2002;156(7):693–695.
Merriam-Webster. Merriam-Webster’s Third New International Dictionary, Unabridged. USA: MerriamWebster; 2002.
Nunnally, JC. Psychometric Theory. New York: McGraw Hill; 1967. Assessment of reliability; p. 226
Office of Minority Health. National Standards for Culturally and Linguistically Appropriate Services
(CLAS) in Health Care. Federal Register 2000;65(247):80865–80879.
Pacquiao D. The relationship between cultural competence education and increasing diversity in nursing
schools and practice settings. J Transcult Nurs 2007;18(Suppl 1):28S–37S. [PubMed: 17204813]
Pettigrew TF, Tropp LR. A Meta-Analytic Test of Intergroup Contact Theory. Journal of Personality and
Social Psychology 2006;90(5):751–783. [PubMed: 16737372]
Pratto F, Sidanius J, Stallworth LM, Malle BF. Social dominance orientation: A personality variable
predicting social and political attitudes. Journal of Personality and Social Psychology 1994;67(4):
741–763.
Provider’s Guide to Quality & Culture. 2004. Retrieved July 25, 2007, from Management Sciences for
Health website:
<http://erc.msh.org/mainpage.cfm?file=1.0.htm&module=provider&language=English>
Roter DL, Hall JA. Physician gender and patient-centered communication: a critical review of empirical
research. Annu Rev Public Health 2004;25:497–519. [PubMed: 15015932]
Schim S, Doorenbos AZ, Borse N. Cultural competence among Ontario and Michigan healthcare
providers. Journal of Nursing Scholarship 2005;37(4):354–360. [PubMed: 16396409]
Schim S, Doorenbos A, Miller J, Benkert R. Development of a cultural competence assessment
instrument. Journal of Nursing Measurement 2003;11(1):29–40. [PubMed: 15132010]
Schneider B, Goldstein HW, Smith DB. The ASA framework: an update. Personnel Psychology 1995;48
(4):743–773.
Sidanius J, Pratto F, Bobo L. Social dominance orientation and the political psychology of gender: A
case of invariance? Journal of Personality and Social Psychology 1994;67(6):998–1011.
Smedley, BD.; Stith, AY.; Nelson, R. Unequal treatment: Confronting racial and ethnic disparities in
healthcare. Washington, D.C.: Institute of Medicine, National Academies Press; 2003.
SPSS. Chicago, Ill: 2006.
STATA. College Station, TX: 2003.
Stephens GG. Family medicine as counterculture. 1979. Family Medicine 1998;30(9):629–636.
[PubMed: 9798125]
Tervalon M, Murray-Garcia J. Cultural humility versus cultural competence: a critical distinction in
defining physician training outcomes in multicultural education. Journal of Health Care for the Poor
and Underserved 1998;9(2):117–125. [PubMed: 10073197]
Wade P, Bernstein BL. Culture sensitivity training and counselor’s race: effects on Black female clients’
perceptions and attrition. Journal of Counseling Psychology 1991;38(1):9–15.
Soc Sci Med. Author manuscript; available in PMC 2009 March 1.
Paez et al.
Page 13
NIH-PA Author Manuscript
Wright B, Scott I, Woloschuk W, Brenneis F, Bradley J. Career choice of new medical students at three
Canadian universities: Family medicine versus specialty medicine. Canadian Medical Association
Journal 2004;170(13):1920–1924. [PubMed: 15210640]
Wyatt GE, Bass BA, Powell GJ. A survey of ethnic and sociocultural issues in medical school education.
Journal of Medical Education 1978;53(8):627–632. [PubMed: 682153]
NIH-PA Author Manuscript
NIH-PA Author Manuscript
Soc Sci Med. Author manuscript; available in PMC 2009 March 1.
Paez et al.
Page 14
Table 1
Cultural competence summary measures and individual items, mean scores,
internal consistency and reliability, n=49
NIH-PA Author Manuscript
Cultural Competence Measures
Attitudea
Cultural motivation
▪ I have the responsibility to learn about all the different groups of people that make up society.
▪ I should be aware of the different cultures that exist within my practice.
Power/assimilation
▪ Regardless of one’s race, anyone can succeed in the U.S. if they try hard enough.
▪ Being white carries significant advantages in our society.
▪ Racial minorities should conform to the customs and values of the racial majority.
Behaviorb
▪ When I see a patient from a culture unfamiliar to me, I seek information abut his/her culture.
▪ I ask patients to tell me about their own explanations of illness.
▪ I welcome feedback from co-workers about how to relate to patients from different cultures.
▪ I adapt my care to patient’s preferences.
▪ I remove barriers (e.g. lack of insurance, need for interpreter) that affect the quality of healthcare for patients of different cultures.
Knowledge, %c
(Items test knowledge of diverse groups including variations in patterns of disease, risk factors and treatment patterns, and underlying factors that lead to health disparities among raci
a
Attitude scores range from 1 (strongly disagree) to 5 (strongly agree).
b
Behavior measures percentage of time a behavior was performed. Response choices were in 25% increments ranging from 1 (0% to 25%) to 4 (76% to
100%).
NIH-PA Author Manuscript
c
Knowledge was assessed using 10 multiple choice questions. Perfect score equals 100%.
SD=standard deviation
NIH-PA Author Manuscript
Soc Sci Med. Author manuscript; available in PMC 2009 March 1.
Paez et al.
Page 15
Table 2
Provider characteristics, n=49
Characteristic
n (%)
NIH-PA Author Manuscript
Age (years), mean (SD) a
Gender
Female
Race
White
Black
Asian
East Indian
Hispanic
Provider typea
44.3 (8.1)
30 (61)
24 (49)
10 (20)
9 (18)
5 (10)
1 (2)
NIH-PA Author Manuscript
Medical doctor
Doctor of osteopathy
Nurse practitioner
Board certifieda
Specialty
Internal medicine
Family practice
International medical graduate status
U.S. graduate
International graduate
Years since completed residency training, mean (SD)
Years with organization, mean (SD)
Political Identitya
Conservative
Liberal
Attended cultural competence training
Provider rating of cultural competence training
Excellent
Very good
Good
Fair
Percent of patients in provider’s practice who differ from provider by race/ethnicity, mean (SD)
Confidence in caring for ethnic minorities
Somewhat or less than somewhat confident
Very confident
Confidence in caring for disadvantaged
Somewhat or less than somewhat confident
Very confident
a
Values do not add up to 49 due to missing data.
SD= Standard deviation
NIH-PA Author Manuscript
Soc Sci Med. Author manuscript; available in PMC 2009 March 1.
43 (90)
1 (2)
4 (8)
46 (98)
36 (73)
13 (27)
36 (73.5)
13 (26.5)
13.6 (7.6)
9.6 (6.0)
10 (27)
27 (73)
17 (35)
1 (6)
7 (41)
3 (18)
6 (35)
60.67 (33.47)
22 (45)
27 (55)
31 (63)
18 (37)
Paez et al.
Page 16
Table 3
Clinic cultural competence characteristics, n=49
Clinic Cultural Competence Characteristics
NIH-PA Author Manuscript
Provider diversity, mean (SD)
White
Black
Hispanic
Asian
Other
Staff diversity, mean (SD)
White
Black
Hispanic
Asian
Other
Clinic provides cultural diversity training to physicians and support staff
Clinic has patient education materials tailored to patients’ race/ethnicity or language
SD= Standard deviation
NIH-PA Author Manuscript
NIH-PA Author Manuscript
Soc Sci Med. Author manuscript; available in PMC 2009 March 1.
Percent
48.0 (28.9)
31.2 (27.6)
3.0 (6.3)
11.8 (16.8)
5.6 (10.5)
31.0 (29.2)
61.3 (30.0)
5.3 (15.0)
1.6 (6.0)
0.8 (3.4)
51
78
NIH-PA Author Manuscript
NIH-PA Author Manuscript
NIH-PA Author Manuscript
Table 4
Association of clinic cultural competence characteristics with provider cultural competence, simple linear regression
Provider diversity
% Nonwhite providers b
Staff Diversity
% Nonwhite staff b
Soc Sci Med. Author manuscript; available in PMC 2009 March 1.
Provides cultural diversity training to providers and staff
No
Yes
Provides patient education materials adapted to ethnicity/race or non-English language
No
Yes
Cultural motivation attitudes
Power/assimilation attitudes
Behavior
β
Knowledge
β
P
β
P
0.00
0.148
0.01
0.092
0.00
0.475
−0.48
0.471
0.06
0.005
0.00
0.991
0. 10
0.002
0.54
0.406
0.26
0.026
0.25
0.241
−0.03
0.876
3.03
0.417
0.33
0.015
0.26
0.308
0.45
0.049
4.73
0.297
P
β
P
Paez et al.
Provider cultural competence
Clinic characteristicsa
a
Clinic cultural competence characteristics were collected by provider report.
b
Per 10% increase in nonwhite providers/staff
Page 17
NIH-PA Author Manuscript
NIH-PA Author Manuscript
NIH-PA Author Manuscript
Table 5
Association of provider characteristics with provider cultural competence, simpler linear regression
Provider characteristics
Cultural motivation attitudes
β
Soc Sci Med. Author manuscript; available in PMC 2009 March 1.
Age
Gender
Male
Female
Race
White
Black
Asian
Other
Politics
Conservative
Liberal
International medical graduate status
U.S. graduate
International graduate
Specialty
Internal medicine
Family practice
Cultural competence Training
No
Yes
Years with organization
Years since completed residency training
% Patients who differ from provider by race/
ethnicity
Confidence in caring for minorities
Somewhat or less than somewhat confident
Very confident
Confidence in caring for disadvantaged
Somewhat or less than somewhat confident
Very confident
P
Power/assimilation attitudes
β
P
Behavior
β
Knowledge
P
β
P
0.01
0.538
−0.01
0.326
0.00
0.998
−0.52
0.032
0.18
0.138
0.484
0.44
0.039
0.192
0.04
0.841
0.176
−0.65
0.866
0.257
0.14
−0.15
0.04
0.364
0.346
0.825
0.11
−0.54
0.06
0.680
0.060
0.865
0.51
−0.01
0.33
0.045
0.974
0.287
−0.70
−5.57
−10.38
0.887
0.278
0.086
−0.04
0.768
0.75
0.005
−0.43
0.063
6.04
0.199
−0.06
0.703
−0.48
0.071
0.06
0.813
−9.72
0.050
−0.08
0.313
0.47
0.043
−0.12
0.583
−4.01
0.340
0.03
0.01
0.00
−0.07
0.796
0.626
0.887
0.705
−0.01
−0.01
0.00
−0.28
0.967
0.728
0.865
0.383
−0.02
0.00
−0.01
−0.04
0.942
0.918
0.713
0.906
1.15
−0.41
−0.41
−8.58
0.771
0.212
0.092
0.131
−0.07
0.586
−0.02
0.907
0.27
0.160
−8.77.
0.012
0.23
0.054
0.43
0.044
0.25
0.209
−0.86
0.828
Paez et al.
Provider cultural competence
Page 18
Paez et al.
Page 19
Table 6
Association of clinic and provider characteristics with provider cultural competence, multivariate linear regression
modelsa
NIH-PA Author Manuscript
Provider cultural competencec
Clinic characteristicsb
Cultural motivation attitudes (N=48)
Staff diversityd
% Nonwhite staff
Provides cultural diversity training to providers and staff
No
Yes
Provides patient education materials adapted to ethnicity/
race or non-English language
No
Yes
Provider characteristics
Behavior (N=49)
β
P
β
P
0.04
0.026
0.09
0.009
0.08
0.439
0.23
0.093
0.25
0.263
Power/assimilation attitudes (N=37)
β
NIH-PA Author Manuscript
Age
Gender
Male
Female
Politics
Conservative
Liberal
International medical graduate status
U.S. graduate
International graduate
Specialty
Internal medicine
Family practice
Confidence in caring for minorities
Somewhat or less than somewhat confident
Very confident
Confidence in caring for disadvantaged
Somewhat or less than somewhat confident
Very confident
P
0.28
0.217
0.65
0.013
0.33
0.41
Knowledge (N=46)
β
P
−0.52
0.026
−3.64
0.460
−9.43
0.011
0.190
0.073
a
Multivariate model includes only those variables significant at p<.0.05 in univariate regression analyses.
b
Clinic cultural competence characteristics collected by provider report.
c
All models: p<0.01
d
Per 10% increase in nonwhite staff
NIH-PA Author Manuscript
Soc Sci Med. Author manuscript; available in PMC 2009 March 1.