University of Southern California Dissertations and Theses

Crosscultural differences in dementia diagnosis and care-seeking in Hispanic and non-Hispanic white outpatients

(USC Thesis Other)

Crosscultural differences in dementia diagnosis and care-seeking in Hispanic and non-Hispanic white outpatients

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Running head: CROSSCULTURAL DIFFERENCES IN DEMENTIA 1

Crosscultural Differences in Dementia Diagnosis and Care-seeking in Hispanic and Non-
Hispanic White Outpatients

Philip Sayegh
Doctoral Dissertation
University of Southern California
Department of Psychology
December 2013

CROSSCULTURAL DIFFERENCES IN DEMENTIA 2
Table of Contents
Abstract 3
Introduction 4
Methods 28
Results 45
Discussion 72
References 94
Bibliography 109
Tables 110-136
Figures 137-155
Appendix 156-168

CROSSCULTURAL DIFFERENCES IN DEMENTIA 3
Abstract
Crosscultural differences in dementia diagnosis and care-seeking merit more attention as the
booming older adult population becomes increasingly diverse. I aimed to examine differences in
clinicians’ assessment of dementia as well as diagnostic delays and impairment levels across 444
Hispanic and 11,081 non-Hispanic White (NHW) outpatients diagnosed with normal cognition
or dementia at their initial Alzheimer’s Disease Research Centers evaluations. Results revealed
that informant reports of patients’ behavioral and psychological symptoms were significantly
associated with diagnosis in NHWs only. Informant-reported functional abilities may be more
strongly related to diagnosis in Hispanics as opposed to neuropsychological test performance
(NP) in NHWs. I also found evidence for the crosscultural factorial invariance of these two
informant-report scales and scalar invariance for one of them. Additionally, Hispanics with
dementia were significantly more impaired in NP than their NHW counterparts when using
combined ethnic-group norms, but the opposite pattern emerged with ethnic group-specific
norms. Finally, in a subsample of 22 Hispanic informants, lower literacy and education levels
were associated with less accurate Alzheimer’s disease (AD) knowledge, whereas stronger
Hispanic acculturation levels were related to more cultural beliefs about AD. In a subsample of
120 NHW informants, higher AD knowledge levels were related to poorer NP in patients. In
sum, these findings bear importance regarding crosscultural diagnostic validity as clinicians may
weigh certain NP tests and informant reports differentially across Hispanics and NHWs during
the diagnostic process. Moreover, both informant scales can be used meaningfully among these
groups, as they have similar factor structures and loadings across cultures. Finally, higher AD
knowledge levels among NHW informants may be related to patients’ poorer NP perhaps
because informants learn more about dementia as patients’ cognitive symptoms progress.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 4
Crosscultural Differences in Dementia Diagnosis and Care-seeking in Hispanic and Non-
Hispanic White Outpatients
As the elderly population in the United States (US) continues to surge, the number of
people with dementia is also likely to grow rapidly (Hebert, Scherr, Bienias, Bennett, & Evans,
2003). Moreover, the number of older adults from cultural minority groups is predicted to
increase at a significantly greater rate than that of non-Hispanic White (NHW) Americans (U.S.
Census Bureau, 2008). Given that increasing age is a leading risk factor for Alzheimer’s disease
(AD) and other dementias (Zarit & Zarit, 2007), it is anticipated that there will be an increasing
prevalence of dementia among older adults from racial and ethnic minority groups.
Currently, the largest and fastest growing ethnic minority group in the US is Hispanics
(U.S. Census Bureau, 2008). Furthermore, Hispanics comprise the most rapidly increasing
subpopulation of older adults in the US, with a predicted rise from two million in 2000 to over
13 million by 2050 (Federal Interagency Forum on Aging-related Statistics, 2000). It is
important to note that Hispanics are quite ethnically and racially heterogeneous. The U.S. Census
created the label Hispanic to refer to people of Spanish-speaking Latin American ancestry,
representing a total of 21 different countries (Suárez-Orozco & Páez, 2002). The number of
elderly Hispanics with AD or other dementias is expected to rise drastically from under 200,000
in 2000 to up to 1.3 million by 2050 (Alzheimer’s Association, 2004). As such, Valle and Lee
(2002) projected that Hispanics in the US will be at a disproportionately increased risk of AD
and other later-life dementias through the next 50 years. The reality is that the number of
Hispanic older adults will likely continue to increase at a significantly greater rate than that of
NHW Americans (U.S. Census Bureau, 2008), which points to the pressing need to better
CROSSCULTURAL DIFFERENCES IN DEMENTIA 5
understand various crosscultural differences associated with dementia diagnosis and care-
seeking.
The Use and Limitations of Neuropsychological Test Performance in Clinicians’ Diagnosis
of Dementia among Hispanics
The clinical diagnosis of dementia, especially among diverse populations, should ideally
involve many sources of information, such as medical and social history, demographic
characteristics, and assessments of neuropsychological test performance (NP), functional
abilities, and dementia-related behavioral and psychological symptoms (BPS). Two of the
primary sources of information for diagnosing dementia are direct evaluations of NP and
informant-based reports of cognitive abilities (Potter et al., 2009). Neuropsychological tests
assess numerous aspects of individuals’ cognitive functioning, including verbal and nonverbal
memory, executive function (e.g., mental tracking, reasoning, and inhibition), language,
attention, and visuospatial abilities.
Although NP remains central in the screening and diagnosis of dementia, it is burdened
with several limitations that call attention to the need for the use of additional approaches in the
diagnostic process, especially among racially and ethnically diverse individuals such as
Hispanics. For example, premorbid intellectual ability, level and quality of education, and
language and cultural factors can affect NP in that poorly educated patients may be misclassified
as having dementia, whereas dementia in well-educated patients may not be captured
(Fillenbaum, Heyman, Huber, Ganguli, & Unverzagt, 2001; Le Carret et al., 2003). Similarly, the
norms for frequently used neuropsychological tests are often drawn from patients evaluated at
tertiary-care medical centers and from control participants who may perform differently on these
tests than representative community samples, which tend to be more culturally and
CROSSCULTURAL DIFFERENCES IN DEMENTIA 6
socioeconomically diverse. Similarly, cultural biases among cognitive tests may hinder the
ability to assess for valid group differences in cognitive performance (e.g., López & Taussig,
1991).
However, Manly and Espino (2004) noted that the majority of prior studies of racial and
ethnic group differences in NP have demonstrated that differences in the scores across ethnic
groups remained even after matching groups on demographic variables, such as age, education,
gender, and socioeconomic status (SES). For example, scores on the Mini-Mental State
Examination (MMSE; Folstein, Folstein, & McHugh, 1975) have been shown to vary by
ethnicity (Escobar et al., 1986; Hohl, Grundman, Salmon, Thomas, & Thal, 1999) and language
of administration (Escobar et al., 1986). Such ethnic-group and other demographic variable
differences can lead to reduced specificity of both verbal and nonverbal neuropsychological tests
in addition to screening instruments like the MMSE such that individuals from cultural minority
groups (e.g., Hispanics) with normal cognitive functioning (NCF) are at an increased risk of
being misdiagnosed as cognitively impaired compared to NHWs (e.g., Artiola i Fortuny, Heaton,
& Hermosillo, 1998; Coffey, Marmol, Schock, & Adams, 2005; Jacobs et al., 1997).
The Use of Informant Reports in Clinicians’ Diagnosis of Dementia
Other widely used sources of information in the diagnosis of dementia are informant
reports, both verbal and written, on patients’ cognitive functioning, functional abilities, and BPS.
Clinicians frequently use informant-based reports in the diagnostic process as they have certain
advantages over NP in terms of diagnosing dementia. For example, informant reports are
comparatively less affected by patients’ premorbid ability and education level or proficiency in
the culture’s dominant language than patients’ NP (Jorm, 2004). Moreover, Malmstrom et al.
(2009) noted that informant-based questionnaires decrease possible bias from factors such as
CROSSCULTURAL DIFFERENCES IN DEMENTIA 7
education and cultural differences because they evaluate patients against their previous cognitive
functioning rather than using norms that may not be valid for that group. These authors also
noted that informant-based questionnaires also benefit from face validity because cognitive
abilities are evaluated with regard to the ability to carry out instrumental activities of daily living
(IADLs, which are more complex tasks [e.g., shopping, driving and/or using transportation, and
managing finances] associated with the capacity to live independently within the community that
tap abilities including organizing, planning, and executing; Lawton & Brody, 1969). Malmstrom
and colleagues added that informant-based reports have also been shown to perform at least as
well at screening for dementia or other cognitive decline as conventional cognitive screening
tests, such as the MMSE.
However, informant reports suffer from their own limitations as well. For example, they
may be contaminated by other factors, such as the emotional state of the informant and the
quality of the relationship between the patient and informant (Jorm, 2004) in addition to
informants’ level of education (Kemp, Brodaty, Pond, & Luscombe, 2002). Despite these
limitations, Jorm suggested that using both NP and informant reports in combination can
increase the accuracy of the screening and diagnosing of dementia because the informant reports
provide information that is complementary to NP. Jorm also noted that using both methods
provides more information than using one method alone. These suggestions bear added
importance for Hispanics for whom the use of NP as the primary basis for diagnosis may be less
valid. Thus, it appears plausible that clinicians may rely on such informant reports more strongly
when assessing dementia among Hispanics as compared to NHWs given their likely awareness
of the limitations of neuropsychological testing.

CROSSCULTURAL DIFFERENCES IN DEMENTIA 8
Informant Reports of Patients’ Memory Decline
One type of informant-based report assesses whether informants report a decline in
patients’ memory. Episodic memory difficulties are generally known as the hallmark symptom
of early AD. For example, family members of individuals in the early stages of AD often report
that their relative frequently engages in repetitive questioning and forgets recent events or
appointments (Storandt, 2008). In addition, clinicians can often detect cognitive deficits due to
the underlying AD process even before a patient meets criteria for a diagnosis of some form of
cognitive impairment through the use of information pertaining to changes from patients’ prior
level of cognitive functioning obtained from reliable informants (Storandt, Grant, Miller, &
Morris, 2006). Some patients themselves may lack insight into their cognitive decline and/or
deny any changes in their episodic memory functioning. Thus, clinicians frequently rely on
informant-reported memory decline (IRMD) when assessing for dementia, and reports of
patients’ cognitive changes from reliable informants may serve as useful sources of information
in the diagnostic process.
Informant Reports of Patients’ Functional Abilities
Certain noncognitive changes also occur in many individuals with underlying cognitive
impairment even before formally receiving a diagnosis of dementia. For example, Storandt
(2008) noted that in some cases, the first prominent signs of AD may be unrelated to episodic
memory and may manifest as difficulty with functioning in daily life. Family members may first
notice that their relative has a hard time carrying out familiar tasks of daily living, such as
driving a car or preparing meals. In addition, Tabert et al. (2002) found that informant-reported
functional deficits in patients with mild cognitive impairment (MCI), commonly viewed as a
precursor of dementia, strongly predicted a future diagnosis of AD among their sample of
CROSSCULTURAL DIFFERENCES IN DEMENTIA 9
outpatients evaluated at memory disorder centers, even after controlling for age, education, and
scores on the modified version of the MMSE (Mayeux, Stern, Rosen, & Leventhal, 1981).
Moreover, Erzigkeit et al. (2001) found that the Bayer-Activities of Daily Living Scale (B-ADL;
Hindmarch, Lehfeld, de Jong, & Erzigkeit, 1998) was as effective as or even superior to the
MMSE in terms of identifying individuals with clinically manifest dementia symptoms. Thus,
evidence points to the conclusion that changes in some individuals’ ability to function in daily
life may represent the earliest stages of a dementing disorder such as AD and can prove to be an
important and valuable area to assess through informants during the diagnostic process.
The Functional Assessment Questionnaire (FAQ; Pfeffer et al., 1981; Pfeffer, Kurosaki,
Harrah, Chance, & Filo, 1982) is a frequently used scale that measures patients’ ability to
conduct IADLs to assist clinicians in the diagnosis of dementia. Although clinicians or health
professionals complete this questionnaire, their responses to each item are based on information
provided directly by an informant. The FAQ assesses a total of 10 IADLs, with response options
ranging from normal to dependent. Pfeffer and colleagues (1981, 1982) reported that they
validated the FAQ on a sample of 195 older adults ages 61 to 91 years in a stable retirement
community of 22,000 people, with participants referred to their study by 33 physicians.
However, the authors provided no information regarding the racial and ethnic breakdown of their
sample, thus limiting the ability to extend the validation of this scale to diverse groups.
Informant Reports of Patients’ Behavioral and Psychological Symptoms
Family members or other informants also often recognize changes in patients’
personality, behavior, and/or mood due to underlying dementia before a diagnosis of dementia
has been assigned. For example, Balsis, Carpenter, and Storandt (2005) performed a prospective
longitudinal study of older adults without dementia and found that about half of their sample
CROSSCULTURAL DIFFERENCES IN DEMENTIA 10
who later developed AD had already demonstrated changes in personality according to
informant-based reports beginning at least one year before diagnosis. They reported that the most
common personality changes among their sample were increased rigidity, apathy, self-
centeredness, and emotional lability. Therefore, similar to functional abilities, BPS that may be
due to underlying dementia, even at earlier stages of the disease, may be reliably reported by
patients’ informants and can thus assist clinicians during the diagnostic process.
Other dementia-related BPS include depression, agitation, and psychosis (Cummings et
al., 1994). Though such symptoms may be shared with competing diagnoses, they are a very
common problem in dementia and have been found to occur in 80% to 90% of dementia patients
(Aalten et al., 2003; Steinberg et al., 2004). Some studies have reported that Hispanic older
adults with dementia had higher levels of dementia-related BPS than their NHW counterparts at
the time of evaluation (Chen, Borson, & Scanlan, 2000; Hinton, Haan, Geller, & Mungas, 2003;
Ortiz, Fitten, Cummings, Hwang, & Fonseca, 2006). Similarly, Sink, Covinsky, Newcomer, and
Yaffe (2004) reported that BPS such as combativeness, episodes of unreasonable anger,
wandering, and hallucinations were more common among Hispanic as compared to NHW
community-dwelling Medicare patients at eight sites across the US.
Despite these findings, it is important to consider whether these differences are indeed
reflective of greater BPS levels among Hispanics or perhaps due to differences in informants’
reactions to dementia-related symptoms and/or reporting styles. Some family members may
delay seeking a dementia evaluation for their relative with symptoms of dementia until certain
triggers other than memory decline (e.g., BPS) arise. For example, Valle (1994) reported that
Mexican American caregivers were more sensitive to dementia-related BPS than NHW
caregivers. Conversely, Hinton, Franz, and Friend (2004) found that NHW caregivers may be
CROSSCULTURAL DIFFERENCES IN DEMENTIA 11
more likely than ethnic minority caregivers to seek a dementia evaluation in light of cognitive
changes, such as episodic memory decline. Accordingly, Ortiz et al. (2006) speculated that
Hispanic caregivers may be triggered to seek a medical evaluation for their relatives by
behavioral changes more so than cognitive changes, with the opposite pattern among NHW
caregivers. These findings lend support to the conception that NHWs may be more likely to
present with and emphasize cognitive complaints, whereas Hispanics may tend to focus on and
perhaps have higher levels of BPS.
The Neuropsychiatric Inventory (NPI; Cummings et al., 1994) is a validated, informant-
based interview that is widely used in both clinical and research settings to evaluate these kinds
of changes (i.e., BPS) that have occurred within the past month among individuals with possible
or probable dementia. Such symptoms include hallucinations, motor disturbances, disinhibition,
and apathy. Kaufer et al. (2000) developed and validated a briefer version of the NPI called the
Neuropsychiatric Inventory Questionnaire (NPI-Q) that differs from the NPI in several ways. For
example, the NPI-Q is a self-administered questionnaire that informants can fill out without
involving an interview. The NPI-Q also measures symptom severity in addition to frequency,
thereby tapping more aspects of cognitive deterioration due to dementia. This scale was found to
have adequate test-retest reliability (Kaufer et al., 2000) and convergent validity with regard to
individual symptom domain scores and total scores as well as caregiver distress ratings from the
NPI (Kaufer et al., 1998). However, a limitation of the NPI-Q that the authors noted is that they
conducted their validation study in a university-based research clinic on a sample of primarily
highly educated NHW participants and their informants. Thus, similar to the FAQ, it remains
unclear how their findings would generalize to more diverse populations and settings.

CROSSCULTURAL DIFFERENCES IN DEMENTIA 12
Summary
The diagnosis of dementia among Hispanics can be complicated by a number of factors
that may affect both how clinicians weigh different pieces of diagnostic information when
assessing dementia and diagnostic validity. First, the reduced validity of NP among many
individuals from this diverse ethnic group may result in the misdiagnosis of dementia. Second,
clinicians’ awareness of this limitation may lead them to consequently rely more heavily among
informant reports (e.g., of patients’ cognitive and functional abilities and BPS) than NP among
Hispanics, and vice-versa among NHWs. Third, Hispanic informants may tend to emphasize and
be more sensitive to noncognitive symptoms such as BPS, whereas NHWs may focus more on
cognitive symptoms such as declines in episodic memory. As a result, the clinical diagnosis of
dementia may differ systematically across ethnic groups.
Another important aspect of the validity of the clinical diagnosis of dementia is the
crosscultural psychometric properties of dementia-related symptoms scales commonly used
during assessment. Below, I address this topic in further detail in relation to the use of the FAQ
and NPI-Q in the dementia diagnostic process among Hispanics and NHWs.
Crosscultural Factorial and Scalar Invariance of Informant-report Scales Used in
Dementia Diagnosis
Researchers have long challenged the crosscultural equivalence of standard scales and
their items for use among diverse populations (e.g., Hui & Triandis, 1985). Specifically, they
have expressed doubt about the ability to make meaningful comparisons across cultural groups
when using measures that were not normed on the groups to whom they are being administered
(Chapleski, Lamphere, Kaczynski, Lichtenberg, & Dwyer, 1997). Accordingly, more research is
CROSSCULTURAL DIFFERENCES IN DEMENTIA 13
needed that focuses on the suitability, reliability, and validity of measures being used to assess
informant-based reports regarding dementia symptoms across diverse ethnic groups.
Measurement invariance assesses whether a scale arrives at a similar measure of the same
attribute across different groups. If measurement invariance is demonstrated, then comparisons
across groups based on that particular attribute can be made. If there is no evidence of such
equivalence, then explanations of differences across groups can be called into question. Factorial
invariance is one well-supported test of measurement invariance (Horn & McArdle, 1992).
Assessment of factorial invariance helps to test the hypothesis that the same attribute is being
measured across groups or with individual groups under different conditions. One approach to
assessing the factorial invariance of a scale involves testing hypotheses about the pattern of
factor loadings (i.e., the identity of the construct being measured) and the cross-group
equivalence of the factorial structure underlying the construct (Byrne, 2010; Raykov &
Marcoulides, 2000). Factor analysis represents one method by which to represent hypothetical
constructs, or latent variables, through one or more observable indicators that can be explicitly
measured (Bollen & Long, 1993; Raykov & Marcoulides). Specifically, confirmatory factor
analysis (CFA) can be used to verify an already existing factor structure or, in some cases, a
theoretically hypothesized factor structure. A strength of the CFA approach is that it allows for
the testing of hypotheses concerning cross-group comparisons, i.e., it is able to statistically test a
priori hypotheses about the existence of underlying structures, including the pattern of factor
loadings, number of factors, and factor correlations (van de Vijer & Leung, 1997). Multigroup
structural equation modeling is a powerful means by which to verify whether factorial invariance
holds across different groups. This analytic strategy allows one to systematically test hypotheses
about the equivalence of the underlying number of factors and the pattern of factor loadings as
CROSSCULTURAL DIFFERENCES IN DEMENTIA 14
well as the invariance of factor variances/covariances for a given model across independent
groups (Bryant & Yarnold, 2000; Byrne, 2010). The suitability of the empirical model to the
sample data, or the invariance of the factor pattern, is then evaluated through empirical
goodness-of-fit indices.
In addition to factorial invariance, it is important to assess the scalar invariance of
measures across ethnic groups. Scalar invariance refers to both groups demonstrating equivalent
intercept values for the pertinent underlying construct in predicting item scores. A finding of
scalar invariance dictates whether group latent means on the scales and/or their factors can be
meaningfully compared (Gregorich, 2006; Meredith, 1993).
Malmstrom et al. (2009) argued that the creation, evaluation, and validation of an
informant-based measure such as the FAQ or NPI-Q should ideally be carried out in a
representative, population-based sample and that the measure’s validity and reliability should be
evaluated both within and across cultural groups. Without sufficient investigation and validation
of these kinds of measures across different groups, their findings may not appropriately portray
the characteristics of the populations being examined. Unfortunately, research regarding the
measurement models underlying some of the informant-based scales clinicians often use when
evaluating dementia is lacking, calling into question the ability to make valid comparisons of
scores on these scales and diagnoses based on them (at least in part) across diverse ethnic groups.
Measurement Properties of Informant-report Scales of Functional Abilities
No research to my knowledge has examined either the factor structure or factorial
invariance of the FAQ across different cultural groups. However, certain studies have examined
the underlying factor structure of other measures of functional abilities, which pointed toward a
single-factor model for measures of functional abilities. For example, Erzigkeit et al. (2001)
CROSSCULTURAL DIFFERENCES IN DEMENTIA 15
found a one-factor structure for the B-ADL for separate samples of individuals with dementia of
varying severity in three European countries (the United Kingdom, Germany, and Spain) despite
mean differences in scores. Specifically, all B-ADL items loaded on a factor they termed
dementia severity, and they were not related to age, sex, education, or country. Their findings
provided evidence for the factorial invariance of the one-factor structure of the B-ADL across
these three European countries and suggested that the mean differences in scores were likely
meaningful and not due to measurement artifact.
Measurement Properties of Informant-report Scales of Behavioral and Psychological
Symptoms
With regard to the measurement properties of dementia-related BPS scales, Lange, Hopp,
and Kang (2004) noted that factor analytic studies using the original English version of the NPI
are lacking. However, Kang, Ahn, Kim, and Kim (2010) examined the factor structure of the 12-
item Korean NPI using both exploratory and CFA in their sample of South Korean individuals
with either AD or probable AD. The authors found that a 10-item, four-factor model was the
best-fitting model for explaining subsyndromes of BPS. The four factors they found were:
hyperactivity (agitation, disinhibition, and irritability), affect (anxiety and depression), psychosis
(delusions and hallucinations), and apathy/vegetative symptoms (apathy, nighttime behavior, and
appetite). Aalten et al. (2007) also found a nearly identical four-factor structure for the 12-item
NPI in their sample involving dementia patients from 12 European countries, which was:
hyperactivity (agitation, disinhibition, irritability, and aberrant motor behavior), affective
(anxiety and depression), psychosis (delusions, hallucinations, and nighttime behavior), and
apathy (apathy and appetite). As can be seen, the only differences between this latter study and
the former were the addition of aberrant motor behavior in the hyperactivity factor and the
CROSSCULTURAL DIFFERENCES IN DEMENTIA 16
inclusion of nighttime behavior in the psychosis factor (as opposed to the apathy/vegetative
symptoms factor in the former study). Aalten and colleagues themselves noted that these findings
were unclear but reported that the aberrant motor behavior item loaded to a very similar degree
on the apathy factor and barely met their .400 factor loading cutoff for both factors (.432 for
hyperactivity and .412 for apathy). In addition, the nighttime behavior item also loaded strongly
on the apathy factor in their study, albeit to a lesser degree (.510 for psychosis and .431 for
apathy). Given these results and the strengths of Kang and colleagues’ combined exploratory and
CFA approach, it appears plausible that the NPI-Q may also be best represented by the same 10-
item, 4-factor structure reported by Kang and colleagues among Hispanics and NHWs.
These studies’ factor analyses of the NPI had in common the co-occurrence of similar
BPS syndromes even across different racial and ethnic populations. Fuh, Liu, Mega, Wang, and
Cummings (2001) suggested that these symptoms might share a common mechanism. However,
the slightly divergent findings from these studies suggest that the NPI may not necessarily be
factorially invariant across these ethnic groups. These findings call attention to the need to assess
for measurement invariance before using the NPI in crosscultural studies, as the factorial
invariance of scales like the NPI cannot simply be presumed. Despite the frequent use of the NPI
in both research and clinical settings, the NPI-Q has received no consideration in the research
literature to my knowledge regarding its psychometric properties across different cultural groups.
Summary
The assessment of the crosscultural psychometric properties of dementia-related scales is
an important issue that has lacked sufficient attention in the field. If the FAQ and NPI-Q are
found to be factorially invariant across different cultural groups, then meaningful comparisons
can be made when using these measures in the context of dementia diagnosis. Moreover, if scalar
CROSSCULTURAL DIFFERENCES IN DEMENTIA 17
invariance is found across cultural groups, then differences in the means of the latent factors
associated with these scales can be meaningfully interpreted and compared. If, however, factorial
invariance is not obtained for these scales across ethnic groups, then it can be concluded that
these scales have different underlying factor structures, which could ultimately affect how
clinicians assess dementia across groups and call diagnostic validity into question.
Overall Summary
In sum, evidence supports the conjecture that the diagnosis of dementia may differ
systematically across Hispanics and NHWs for a number of reasons. First, as a result of the well-
known limitations of NP among Hispanics, clinicians may tend to rely more heavily on
informant reports rather than NP when diagnosing dementia among this group, with the opposite
pattern perhaps occurring among NHWs. Second, cultural differences in informant reporting
styles of and reactions to patients’ dementia symptoms may affect clinicians’ assessment of
dementia and hence diagnostic accuracy. Third, it has yet to be determined whether there are
differences in the psychometric properties of scales commonly used in the diagnostic process
(i.e., the FAQ and NPI-Q). As a result, the dementia assessment and diagnostic process may
indeed differ across Hispanics and NHWs, bearing relevance in terms of diagnostic validity.
While such factors may influence dementia diagnostic validity, a number of other
culturally-influenced variables may affect both the presentation of symptoms at dementia
evaluations and informant reports of such symptoms. Specifically, certain culturally-influenced
barriers to dementia care-seeking may result in delays to diagnosis and, thus, higher levels of
impairment at initial evaluations (i.e., above and beyond the spuriously higher levels of
impairment that may be present in the context of NP among Hispanics). Additionally, certain
culturally-influenced beliefs about AD may affect informants’ reactions to and perceptions and
CROSSCULTURAL DIFFERENCES IN DEMENTIA 18
reporting of patients’ dementia symptoms. Therefore, it is similarly important to review these
cultural influences, as it may help call attention to the fine line between actual levels of dementia
symptoms versus differences in NP validity in addition to informant-reporting styles. Below, I
review the literature on crosscultural differences in delays to dementia care-seeking and
cognitive impairment, barriers to dementia care-seeking among Hispanics that may contribute to
greater impairment levels at initial evaluations, and cultural factors that may also affect
informant reports of patients’ symptoms during the diagnostic process.
Crosscultural Differences in Time to and Impairment Level at Initial Dementia
Evaluations
In general, several lines of evidence point to the conclusion that Hispanic caregivers face
barriers that may cause them to be less willing and/or able than their NHW counterparts to seek a
formal evaluation for the individual with dementia, even despite the high levels of distress
associated with caregiving (Valle, Yamada, & Barrio, 2004). In part for this reason, Hispanic
patients may present with greater impairment levels at the time of initial evaluation as compared
to NHW patients. Indeed, a recently published meta-analysis by Cooper, Tandy, Balamurali, and
Livingston (2010) reported both that ethnic minority people with dementia were more
cognitively impaired than NHWs and that Hispanics in particular reported a longer duration of
memory loss than NHWs at the time of referral to diagnostic dementia services. The authors
reported that these group differences remained even after controlling for premorbid level of
education, suggesting that the higher level of cognitive impairment among Hispanics was likely
due to presenting later for evaluation rather than education level differences.
These findings are clinically relevant, as obtaining a timely diagnosis of dementia has
both medical and practical benefits. With regard to medical benefits, medications that have been
CROSSCULTURAL DIFFERENCES IN DEMENTIA 19
approved for use among patients in the earlier stages of AD can decelerate the mental
deterioration associated with the disease (e.g., Knapp et al., 1994; Rogers et al., 1998).
Additionally, Boyle, Ismail, and Porsteinsson (2006) noted that a timely diagnosis can help
provide a clinical justification for dementia-related complaints, facilitate the provision of
educational information about dementia, and afford patients and families with ample time to
discuss and prepare for the future (e.g., institutionalization options and advance directives) while
the patient is still competent. In sum, obtaining a timely and correct dementia diagnosis may
reduce patients’ and relatives’ burden and improve their quality of life for as long as possible.
Culturally-influenced Barriers to Dementia Evaluations among Hispanics and Their
Effects on Informant Reports
As summarized above, several studies have provided evidence to support the notion that
Hispanic older adults may seek a clinical evaluation for dementia at later stages in the disease
process than NHWs and may in part for this reason be more impaired at the time of diagnosis.
Various barriers, both systemic and culturally-influenced, to obtaining a dementia evaluation
among older adults from ethnic minority groups including Hispanics may contribute to this later
time to diagnosis. In addition, many of these cultural variables may also affect informant reports
of patients’ dementia symptoms differentially across cultures. Below, I provide a review of some
key culturally-influenced barriers to dementia care-seeking among Hispanics in the US and how
they may influence informant reports of patients’ dementia symptoms.
Culturally-influenced Beliefs about Dementia
A study involving a Hispanic older adult sample from Southern California found that
even when controlling for age, education and acculturation levels, and ethnicity, the view that
memory decline is a normal part of aging rather than due to illness was prevalent and represented
CROSSCULTURAL DIFFERENCES IN DEMENTIA 20
a significant barrier to diagnosis even in the absence of physical or financial barriers (Ortiz &
Fitten, 2000). Similarly, Gelman (2003) found that the perception of dementia-related symptoms
as a normal part of aging was a reason for a delay to evaluation for Hispanic American family
caregivers. These personal beliefs about cognitive disorders such as AD may delay help-seeking
activities for these patients and their family caregivers (Connell, Roberts, & McLaughlin, 2007;
Gray, Jimenez, Cucciare, Tong, & Gallagher-Thompson, 2009; Mahoney, Cloutterbuck, Neary,
& Zhan, 2005). Additionally, viewing cognitive decline as a normal part of aging may lead
Hispanic informants to emphasize noncognitive symptoms (e.g., functional abilities and BPS)
rather than cognitive symptoms (e.g., declines in episodic memory) as some studies have found
(Ortiz et al., 2006; Valle, 1994), which may impact clinicians’ diagnostic accuracy.
Hispanics may be more likely than NHWs to endorse culturally-influenced beliefs that
emphasize perceiving dementia symptoms as due to mental disorders or insanity rather than
neurodegenerative brain diseases. Attributing dementia to ‘craziness’ has been reported in
studies involving Hispanic American older adults (e.g., Ayalon & Areán, 2004). In addition,
those with dementia may be viewed as ‘crazy’ or having ‘bad blood,’ a dishonor shared among
the entire family due to collectivistic, family-centered cultural values (Henderson & Gutierrez-
Mayka, 1992; Neary & Mahoney, 2005). Religious views can also affect how some Hispanics
perceive various symptoms associated with dementia. Certain religious beliefs held by some
Hispanic individuals can encourage the labeling of behavioral changes associated with dementia
as “evil” or resulting from possession by evil spirits, leading these family members to have
alternative views about what constitutes dementia (Dilworth-Anderson & Gibson, 1999). In sum,
perceiving dementia and its symptoms as such may serve to hinder Hispanics older adults from
seeking timely dementia care. In addition, such perceptions may influence how informants view
CROSSCULTURAL DIFFERENCES IN DEMENTIA 21
and report on patients’ dementia symptoms at the time of diagnosis in that the way symptoms are
communicated or portrayed may differ systematically across cultural groups.
Lack of Accurate Knowledge about Dementia
Relatedly, people’s knowledge about a medical problem is also associated with the
actions they take regarding it. In contrast to the previously discussed culturally-influenced beliefs
about dementia, knowledge about dementia refers to an individual’s accurate understanding of
dementia. Morhardt, Pereyra, and Iris (2010) noted that individuals try to understand aspects of
their medical problems based on the knowledge they have, which in turn affects whether they
choose to seek evaluation or treatment. Moreover, culture can affect knowledge of a medical
problem, such as AD (Jones, Chow, & Gatz, 2006).
Hispanics and their family members may be less likely to have information about AD and
more likely to have culturally-based alternative ideas and beliefs about AD, which could both
result in the postponement of families’ help-seeking behaviors (Ayalon & Areán, 2004; Gray et
al., 2009; Hinton, Franz, Yeo, & Levkoff, 2005; Hinton & Levkoff, 1999; Levkoff, Levy, &
Weitzman, 1999; Mahoney et al., 2005) and influence informant reports of patients’ dementia
symptoms. Indeed, a systematic review of both quantitative and qualitative studies revealed that
a lack of knowledge about dementia was a barrier to seeking care among Hispanic Americans in
several studies (Mukadam, Cooper, & Livingston, 2011). Dilworth-Anderson and Gibson (1999)
suggested that a lack of education and isolation from dementia-related information can shape the
process of interpretation and the range of alternative explanations for and meaning of the illness,
which could both result in diagnostic delays and influence informant reports. Similarly, Hinton et
al. (2005) reported that their Hispanic family dementia caregiver participants were significantly
more likely to conceptualize dementia in ways that differed from the widely accepted biomedical
CROSSCULTURAL DIFFERENCES IN DEMENTIA 22
model as compared to their NHW participants, even after controlling for education level. The
authors suggested that this lack of biomedical knowledge about dementia could serve as a barrier
to dementia-related help seeking. Moreover, given these cultural differences in how some
Hispanics view dementia and its symptoms, informant reports of patients’ symptoms may also
differ systematically across ethnic groups. NHWs, in contrast, are more likely to view dementia
as a brain-based disease process for which the opportunity to volunteer for medical research
treatment could potentially be of assistance (Castleman, Gallagher-Thompson, & Naythons,
2003; Gallagher-Thompson, 2006). In conclusion, a lack of accurate AD knowledge may result
in diagnostic delays and affect informants’ views and reports of patients’ symptoms.
Lower Levels of Acculturation
Sayegh and Knight (2012) reported that some studies have found that lower levels of
acculturation to the US may be associated with delays to a dementia diagnosis among individuals
from minority ethnic groups such as Hispanics. In their recently proposed Sociocultural Health
Belief Model, these authors posited that this association is due to the influence of acculturation
levels on culturally-influenced beliefs and accurate knowledge about AD, which may ultimately
affect the decision to obtain a dementia evaluation. Given the posited association of acculturation
with culturally-influenced beliefs and knowledge about AD, Hispanics with lower acculturation
levels may also perceive and/or report on patients’ dementia symptoms differently than NHWs
as previously noted, which could clearly impact diagnostic validity. Therefore, the assessment of
acculturation levels may bear importance regarding the dementia care-seeking process,
impairment levels, informant reports, and diagnosis.

CROSSCULTURAL DIFFERENCES IN DEMENTIA 23
Summary
Various barriers to dementia evaluations and certain culturally-based beliefs may delay
early evaluation and treatment among Hispanics and affect informants’ perception and reports of
patients’ dementia symptoms. Moreover, lower acculturation levels (i.e., to the US) may be
negatively associated with accurate AD knowledge and positively with culturally-influenced
beliefs about AD. Taken together, these findings suggest that Hispanic older adults may present
for a dementia evaluation at a later stage in their disease process than NHWs. Therefore, they
may be more impaired in terms of NP and the ability to function independently in daily life as
well as exhibit more dementia-related BPS. Additionally, these cultural influences may be
driving ethnic-group differences in informant reports of patients’ symptoms that consequently
lead to systematic dissimilarities in the dementia diagnostic process.
Clinical Diagnosis Confirmation through the Use of Post-mortem Neuropathological
Autopsy Findings
It should be reiterated that the diagnosis of dementia may be less accurate among
Hispanic elders as compared to their NHW counterparts for several reasons that I previously
discussed. In addition, for all individuals, the only current method for definitively diagnosing AD
is through pathological confirmation via a post-mortem brain autopsy, which is considered to be
the “gold standard” for diagnosis (Sperling & Johnson, 2010). In order to address these potential
limitations, it may be useful to compare diagnoses to the neuropathological findings from brain
autopsies to assess diagnostic accuracy and whether it differs across these two ethnic groups.
Summary and Hypotheses
My hypotheses are divided into two sections based on the two samples of participants I
will be using for this project. The first section involves a larger, national existing dataset,
CROSSCULTURAL DIFFERENCES IN DEMENTIA 24
whereas the second section involves a smaller subsample of these participants from whom I
collected additional data (to be described in further detail in the Methods section).
Hypotheses Using the National Dataset
Time to and level of impairment at the time of initial dementia evaluation. Numerous
factors can potentially facilitate or hinder individuals from seeking a dementia evaluation, and
these factors may play out differently across Hispanics and NHWs. As a result, several lines of
evidence point to the general conclusion that Hispanics may present for a dementia evaluation
later and be more impaired at the time of diagnosis in terms of NP as well as functional abilities
and levels of BPS as compared to NHWs. Many reasons may account for these crosscultural
differences, including culturally-influenced beliefs and knowledge about AD, religious views,
and acculturation levels. In addition, many of the neuropsychological tests administered in the
context of a dementia evaluation may not be well-suited for Hispanics for whom English is not
their primary and/or native language and/or who were not educated in the US. Finally, Hispanic
informants may place greater emphasis (i.e., demonstrate an overendorsing reporting style) than
NHWs on behavioral disturbances, such as decreased functional abilities and BPS, due to various
cultural beliefs about dementia. Therefore, Hypothesis 1 predicted that Hispanics who were
ultimately assigned a clinical diagnosis of dementia at their initial evaluations would be more
impaired and have had a longer delay to their initial evaluations than their NHW counterparts.
Crosscultural factorial and scalar invariance of informant-report scales. As
discussed earlier, factorial invariance must be established in order to make meaningful
comparisons across groups when using informant-report scales. The underlying factor structure
of the scales completed by informants during dementia evaluations as well as the mean values
CROSSCULTURAL DIFFERENCES IN DEMENTIA 25
for these underlying factors may or may not differ across cultural groups, which bears
significance in terms of drawing valid conclusions based on group comparisons.
A prior study confirmed the factorial invariance of an informant-based measure of
functional abilities that is similar to the FAQ among different ethnic groups and found one factor
that they termed dementia severity (Erzigkeit et al., 2001). Therefore, Hypothesis 2 predicted that
the FAQ would also have a one-factor structure reflecting dementia severity that would
demonstrate configural, measurement, and scalar invariance across these two ethnic groups and
can thus be used to make meaningful comparisons regarding functional abilities across groups.
Hypothesis 3 predicted that the NPI-Q would have a 10-item, four-factor structure
(consistent with the findings from the Aalten et al., 2007 and Kang et al., 2010 studies that found
four factors using 10 of the 12 NPI items, the same number and types of items as the NPI-Q,
among samples from 12 European countries and Korea, respectively). I also hypothesized that
these four factors would be similar to those found in the Kang et al. study: hyperactivity
(agitation, disinhibition, and irritability), affect (anxiety and depression), psychosis (delusions
and hallucinations), and apathy/vegetative symptoms (apathy, sleep, and appetite). In addition, I
hypothesized that the NPI-Q would demonstrate configural, measurement, structural, and scalar
invariance across Hispanics and NHWs, as the NPI has been shown to have a similar factor
structure samples from 12 European countries and Korea.
Path models depicting variables associated with a dementia diagnosis across
cultural groups. What is lacking in the literature in this area is a clear understanding of the
mediating variables between background and demographic characteristics that are known to be
associated with dementia (e.g., age, sex, and education level) and a clinician’s formal diagnosis
of dementia or NCF. Because some of the primary sources of information on which clinicians
CROSSCULTURAL DIFFERENCES IN DEMENTIA 26
rely when diagnosing dementia are patients’ NP and informant-based reports of patients’
cognitive and functional abilities as well as BPS, a richer knowledge of how these variables may
interact with one another could provide insight into potential systematic differences in the ways
dementia is diagnosed across ethnic groups.
Based on both prior research and the hypothesized findings from this study, Hypothesis 4
is presented in terms of proposed path models depicting various factors that may affect seeking a
dementia evaluation as well as level of dementia-related impairment and diagnosis at the time of
initial evaluation for Hispanic and NHW participants. For both groups of patients, I predicted
that overall NP, IRMD, and informant reports on patients’ functional abilities and BPS would be
significantly associated with diagnosis (when controlling for patients’ age, sex, and primary
language in addition to both patients’ and informants’ education levels).
Given the reduced validity of cognitive tests among Hispanics of which clinicians are
likely aware, it is plausible that clinicians may tend to rely more on informant reports of changes
in patients’ memory, functional abilities, and BPS, as opposed to overall NP among NHWs,
when diagnosing dementia. In addition, Hispanic caregivers may be more sensitive to
noncognitive dementia symptoms such as BPS than NHWs (Ortiz et al., 2006; Valle, 1994) and
may consequently be more likely to endorse such symptoms as compared to NHWs. Therefore, I
also hypothesized that i) NP would be more strongly associated with a diagnosis of either of
dementia or NCF among NHW as compared to Hispanic patients and that ii) all three informant-
based reports would be more strongly associated with diagnosis among Hispanic as compared to
NHW patients. Figures 1 and 2 provide illustrations of the path models that I proposed to
evaluate the relations among these variables for each ethnic group.

CROSSCULTURAL DIFFERENCES IN DEMENTIA 27
Hypotheses Using the Subsample of Informants
Knowledge about Alzheimer’s disease. Hypothesis 5 predicted crosscultural differences
in mean levels of accurate AD knowledge as operationalized by an existing scale that measures
this construct. Specifically, I hypothesized that Hispanics as compared to NHWs would have
significantly lower scores on this measure, indicative of a smaller amount of such accurate
knowledge of AD. The role of this variable in the context of the other variables to be examined
in this subsample’s path models is discussed in Hypothesis 7.i.
Cultural beliefs about Alzheimer’s disease. Hypothesis 6 pertained to the development
of a scale to measure culturally-influenced beliefs about AD based on prior findings from both
qualitative and quantitative studies as well as review papers on this topic. I hypothesized that i)
this scale, called the Cultural Beliefs about AD Scale (CBADS), would demonstrate adequate
internal consistency and ii) Hispanic informants would have significantly higher scores on this
measure than NHW informants, reflecting a larger amount of culturally-based beliefs about AD.
The role of this variable in the context of the other variables to be investigated in this
subsample’s path models is also reviewed in Hypothesis 7.i.
The effects of knowledge and cultural beliefs about Alzheimer’s disease and
acculturation on diagnosis across cultural groups. Finally, Hypothesis 7 involved the
reassessment of the originally proposed path models for each ethnic group by including a scale
measuring accurate AD knowledge and one measuring culturally-influenced beliefs about AD
among a subsample of Hispanics and NHW informants and measures of acculturation among the
Hispanic informant subsample only. Based on my review of the literature on culturally-
influenced barriers to dementia care-seeking, I hypothesized that i) less knowledge about AD and
higher levels of culturally-influenced beliefs about AD, as well as ii) lower levels of
CROSSCULTURAL DIFFERENCES IN DEMENTIA 28
acculturation (among the Hispanic subsample), would be positively associated with clinicians’
diagnosis of dementia as partially mediated by informant-based reports. In addition, given
Hypotheses 5 and 6.ii that predicted less knowledge of and more culturally-influenced beliefs
about AD among Hispanics and cultural influences on informant reports of patients’ symptoms, I
also hypothesized that iii) these variables would have a stronger association with NP, IRMD, and
informant-reported functional abilities and BPS among Hispanics as compared to NHWs.
The main purpose of these additional path models was to evaluate the effects of the richer
cultural variables that I obtained from a subsample of informants on the dementia diagnostic
process, rather than simply relying on ethnicity to examine cultural differences. These variables
may tap key factors involved in obtaining a dementia evaluation that can ultimately affect
patients’ symptom levels, informants’ perceptions of and reporting on patients’ symptoms, and
diagnosis at the time of initial evaluation. Figures 3 and 4 provide illustrations of these path
models.
Methods
Participants and Procedure
Participants for the current study included outpatients and their informants enrolled in the
longitudinal National Alzheimer’s Coordinating Centers (NACC) Alzheimer’s Disease Research
Center (ADRC) and Alzheimer’s Disease Center (ADC) study at various centers nationwide. The
sample used in the current study includes data from 11,525 individuals (and their informants)
with a diagnosis of either NCF or dementia by a clinician at the time of initial evaluation.
Participants were excluded if they had diagnoses of stroke, age-associated memory impairment,
Parkinson’s disease, and MCI (i.e., no dementia associated with these diagnoses). Participants
(and their informants) completed the consent process before undergoing a standard
CROSSCULTURAL DIFFERENCES IN DEMENTIA 29
neuropsychological assessment with several tests (see below) assessing multiple domains of
cognitive functioning. The assessments were performed either in the research centers or in the
participants’ homes. These assessments, which normally lasted approximately one hour, were
generally conducted by a psychometrician with review by a neuropsychologist. In addition,
informants answered several questions about their background and responded to questions
pertaining to patients’ cognitive and functional abilities as well as BPS.
In addition to this existing data set, I also contacted a subsample of Hispanic and NHW
informants via mail to collect additional information. These participants were drawn from the
pool of active participants currently enrolled in the NACC study through the University of
Southern California (USC) and the Universities of California—Los Angeles (UCLA), Irvine
(UCI), and San Diego (UCSD). Participants enrolled in the NACC study agreed to being
contacted about future NACC-related research when they reviewed and signed their informed
consent forms at the time of enrollment. The additional information that I collected was in the
form of questionnaires that tap informants’ knowledge of AD, cultural beliefs about AD, and
acculturation, which I describe in further detail below.
Measures
Demographic and background variables. I examined several demographic and
background variables as part of this study. These variables included patients’ age, sex, ethnicity,
national origin (if of Hispanic descent), and primary language. I also examined both patients’ and
informants’ years of education, which were calculated based on the sum of the total number of
years of formal education. Because age, sex, and education level can potentially affect one’s
cognitive performance and/or risk for cognitive decline, these variables were included as
covariates in the analyses, as was having a language other than English as a primary language
CROSSCULTURAL DIFFERENCES IN DEMENTIA 30
among the Hispanic participants. Finally, I examined the potentially confounding roles of
informants’ relationship to and coresidency with patients.
Neuropsychological test measures. Neuropsychological tests were chosen for inclusion
into the NACC data set on the basis of their ability to assess several broad cognitive domains
(Morris et al., 2006). These tests are described below and are organized by cognitive domain.
General dementia screen. The MMSE is widely used as a brief screening test for
cognitive impairment. Thirty items assess memory, orientation, language, attention, and
visuospatial functioning. In general, scores greater than or equal to 25 indicate normal (i.e.,
intact) gross cognitive functioning. Though slightly different cutoff rules have been established,
severely impaired cognitive functioning is often indicated by a score less than or equal to 9
points, whereas scores between 10 and 20 often indicate moderately impaired cognitive
functioning and scores between 21 and 24 are often suggestive of mildly impaired cognitive
functioning (Mungas, 1991). However, a recent systematic review by Lonie, Tierney, and
Ebmeier (2009) reported that the widely used cutoff scores of 24 and 26 are indicative of
cognitive decline (i.e., MCI or dementia), whereas higher scores are indicative of NCF. Test-
retest reliability over a four-week period was nearly perfect for the dementia patients in Folstein
et al.’s sample (r = .99).
Immediate and delayed verbal episodic memory. Immediate and delayed recall of
structured verbal material was assessed by Story A of the Logical Memory subtest of the
Wechsler Memory Scale—Revised (WMS-R; Wechsler, 1987a). Participants were read a short
story and asked to orally recall as much of the story as possible. Recall was assessed both
immediately after presentation of the story as well as after a 20- to 30-minute delay. Participants
can score between 0 and 25 points on both immediate and delayed recall of this subtest based on
CROSSCULTURAL DIFFERENCES IN DEMENTIA 31
the total number of story units (i.e., chunks of information) recalled. Test-retest reliability for
Story A alone of Logical Memory has not been documented. However, Snow, Tierney, Zorzitto,
Fisher, and Reid (1989) reported a test-retest reliability coefficient of .68 for Stories A and B of
Logical Memory among a normal elderly sample after one year.
Attention. The Digit Span Forward and Backward subtests of the Wechsler Adult
Intelligence Scale—Revised (WAIS-R; Wechsler, 1987b) were used as measures of attention.
Digit Span Forward requires participants to recall strings of numbers that are orally presented at
a rate of approximately one number per second. The strings range in length from three to eight
numbers. Digit Span Backward requires participants to recite a string of orally presented
numbers in reverse order. The strings range in length from two to seven numbers. Participants
are scored both in terms of the number of correct trials (ranging from 0 to 12 on both Digit Span
Forward and Backward) as well as the longest span length (i.e., the length of the longest string of
correctly recalled numbers). Possible span length ranges from 0 to 8 for Digit Span Forward and
0 to 7 for Digit Span Backward. Test-retest reliability coefficients have ranged from .66 to .89
depending on interval length and participants’ ages (Matarazzo & Herman, 1984; Snow et al.,
1989).
Working memory. The WAIS-R Digit Span Backward subtest, described above, is also a
measure of working memory. Working memory is responsible for temporarily storing and
managing information required to carry out more complex tasks, such as mental tracking and the
manipulation of information (Buchwald & Rapp, 2010).
Processing speed. Two measures were used to assess processing speed. The WAIS-R
Digit Symbol Coding subtest is similar to a transcription task. Participants are given a key that
pairs numbers to simple symbols. The task is to fill in the appropriate symbol underneath series
CROSSCULTURAL DIFFERENCES IN DEMENTIA 32
of numbers. The participants are encouraged to work as quickly as they can without making
errors. Scores are based on the total number of correctly completed items within a 90-second
time limit. Possible scores range from 0 to 93. In addition to attention, scores on this task reflect
psychomotor speed, visuospatial functioning, and memory (Lezak et al., 2004). Test-retest
reliability for this subtest has tended to run high, with correlation coefficients ranging from .82 to
.88 (Matarazzo & Herman, 1984; Wechsler, 1981).
Processing speed was also assessed with the Trail Making Test Part A (Trails A; Reitan,
1958). This paper-and-pencil task requires participants to draw lines connecting numbered
circles. The participants are expected to connect the circles in order and to work as quickly as
possible. In addition to processing speed, scores on this test reflect simple attention skills, visual
tracking, and motor speed. Participants are allowed up to 150 seconds to complete the task and
are assigned the maximum score of 150 seconds if they do not complete the task. Thus, scores on
this test can range from 0 to 150 seconds, with higher scores indicative of poorer performance.
Trails A has had reliability coefficients ranging from above .60 to the .90s (Lezak et al., 2004).
Executive function. Various executive functions, including set shifting and complex
mental tracking, as well as processing speed, simple attention, visual tracking, and motor speed,
were measured by performance on the Trail Making Test Part B (Trails B; Reitan, 1958). Similar
to Trails A, this paper-and-pencil task requires participants to draw lines connecting circles as
quickly as possible. However, on Trails B, participants are expected to hold in mind two series
(numbers and letters) and alternate between them, in order. Participants are allowed up to 300
seconds to complete the task and are assigned the maximum score of 300 seconds if they do not
complete the task. Thus, scores on this test can range from 0 to 300 seconds. Higher scores are
CROSSCULTURAL DIFFERENCES IN DEMENTIA 33
indicative of poorer performance. Similar to Trails A, Trails B has had reliability coefficients
ranging from above .60 to the .90s (Lezak et al., 2004).
Language. Within the domain of language, verbal fluency (i.e., category fluency) was
measured by asking participants to orally generate items within the categories of animals and
vegetables. Participants were given 60 seconds for each task. Total scores consisted of the
number of correct responses and excluded intrusions and perseveration errors. Morris et al.
(1989) reported that the animal naming test had one-month test-retest reliability coefficients of
.76 for individuals with moderate dementia and .67 for individuals with NCF. They also found
that that this test loaded on a language factor.
Confrontation naming was assessed by asking participants to name line drawings of
objects. The 30 odd-numbered items of the Boston Naming Test (Kaplan, Goodglass, &
Weintraub, 1983) were presented and participants were allowed 15 seconds to orally output the
name the object. Semantic cues were provided to participants if they did not recognize the object
or clearly misrecognized the object as something else (e.g., calling a pretzel a snake). If
participants named the object correctly (with or without the semantic cue) within 15 seconds,
then a point was awarded for that item. Total scores on this test range from 0 to 30. This test has
been found to be moderately highly correlated with verbal ability tests such as the WAIS-R
Vocabulary subtest (r = .65; Killgore & Adams, 1999).
Overall neuropsychological test performance. To obtain a measure of overall NP, raw
scores from each of the individual tests were converted into standardized scores for each ethnic
group before creating a composite standardized NP score. I calculated the means and standard
deviations for each neuropsychological test among the Hispanic participants without dementia
and the NHW participants without dementia, which served as the basis for standardization. The
CROSSCULTURAL DIFFERENCES IN DEMENTIA 34
standardization process allowed each of the tests to have equal weight despite differences in the
ranges of possible raw scores such that they were based on the same scale. I also conducted
analyses using combined ethnic-group norms (CEGN) based on the combined sample of both
Hispanics and NHWs without dementia to compare findings based on these norms to the ethnic
group-specific norms (EGSN). Of note, there were no significant ethnic-group differences
among patients with NCF in terms of either IRMD or patient-reported memory decline.
Informant reports.
Informant-reported memory decline. At the time of the initial evaluation, clinicians
asked patients’ informants whether they believed that the patient has had a meaningful decline in
memory. Clinicians coded responses to this IRMD variable as either yes or no.
Functional abilities. The patients’ informants completed the FAQ (Pfeffer et al., 1981,
1982), which, as described earlier, measures patients’ abilities to carry out IADLs. The FAQ
assesses a total of 10 IADLs, with response options including 0 (normal), 1 (has difficulty, but
does by self), 2 (requires assistance), 3 (dependent), and 8 (not applicable (e.g., never did)).
Total scores range from 0 to 30, with higher scores representing more difficulty or requiring
assistance with carrying out IADLs among patients over the past four weeks. Please refer to
Table 23 in the Appendix for a list of the items on this measure. The Cronbach’s alpha values for
this scale in the national sample were α = .97 for both Hispanics and NHWs. In the subsample of
informants, these values were α = .97 for Hispanics and α = .96 for NHWs.
Behavioral and psychological symptoms. The patients’ informants completed the NPI-Q,
which, as previously described, measures the occurrences and severity of patients’ BPS. Total
scores regarding the occurrence of BPS range from 0 to 12, with higher scores being indicative
of a greater number of BPS that represent a change in patients in the past month. Total BPS
CROSSCULTURAL DIFFERENCES IN DEMENTIA 35
severity scores range from 0 to 36, with higher scores representing increased symptoms severity.
The NPI-Q was found to have adequate test-retest reliability and convergent validity with regard
to both individual symptom domain scores and total scores and caregiver distress ratings from
the NPI (Kaufer et al., 2000). Table 24 in the Appendix contains the items from this scale. The
Cronbach’s alpha values for this scale in the national sample were α = .78 for Hispanics and α =
.79 for NHWs. In the subsample of informants, these values were α = .88 for Hispanics and α =
.82 for NHWs.
Knowledge about Alzheimer’s disease. To measure informants’ knowledge about AD, I
asked a subset of informants to complete the Alzheimer’s Disease Knowledge Scale (ADKS;
Carpenter, Balsis, Otilingam, Hanson, & Gatz, 2009). This 30-item, true/false scale, which takes
approximately five to 10 minutes to complete, assesses various issues pertaining to dementia,
including risk factors, assessment and diagnosis, symptoms, course, life impact, caregiving, and
treatment and management. Carpenter and colleagues reported that this scale had adequate
reliability (test-retest: r = .81, p < .001; internal consistency: α = .71) and validity (content,
predictive, concurrent, and convergent). See Table 25 in the Appendix for the list of items
included in this scale. The values obtained from the Kuder-Richardson Formula 20 (KR-20) for
this scale in the subsample of informants were α = .74 for Hispanics and α = .64 for NHWs.
Culturally-influenced beliefs about Alzheimer’s disease. To measure culturally-
influenced beliefs about AD, I created the CBADS by reviewing the literature pertaining to
Hispanic cultural beliefs regarding AD and creating a total of 26 items based on these findings.
Items were reviewed with experts in the research field of cultural influences on health-related
issues and revised appropriately. The same subset of participants that was administered the
ADKS was also asked to complete this scale. Informants responded with either true or false to
CROSSCULTURAL DIFFERENCES IN DEMENTIA 36
each question and I calculated a total correct score, with higher scores reflecting more culturally-
based beliefs about dementia. See Table 26 in the Appendix for the items included in this scale. I
ultimately removed 6 items from this scale, because all participants endorsed the correct (i.e., did
not endorse the associated culturally-influenced belief about AD) responses for these items (#2,
8, 16, 17, 21, and 22), thereby resulting in zero variance. Consequently, the final CBADS scale
in this study consisted of 20 items. The values obtained from the KR-20 for this reduced version
of this scale in the subsample of informants were α = .65 for Hispanics and α = .62 for NHWs.
Acculturation. In order to incorporate acculturation into my path models, I measured this
construct among the same subsample of Hispanic informants using various measures. First, I
used two subscales of Scale 1 of the Acculturation Rating Scale for Mexican Americans-II
(ARSMA-II): Anglo Orientation and Mexican Orientation (Cuellar, Arnold, & Maldonado,
1995). Given that the Hispanic sample included individuals from Hispanic countries other than
Mexico, I asked those individuals to complete the questions with regard to their specific country
of origin, and I renamed the Mexican Orientation subscale to Hispanic Orientation for the
purposes of this project. The Anglo and Hispanic Orientation subscales are composed of 13 and
17 items respectively (see Table 27 in the Appendix). Respondents were instructed to rate on a
scale ranging from 1 (not at all) to 5 (much or extremely) their frequency of language use
(English and Spanish), ethnic identity, and ethnic interaction. Higher scores indicate a stronger
orientation toward a specific culture. Cuellar et al. (1995) suggested subtracting average scores
on the Hispanic Orientation subscale from the average scores on the Anglo Orientation subscale
to obtain a raw acculturation score. However, I also examined the two subscales independently to
evaluate their unique effects on the other independent variables of interest (knowledge and
cultural beliefs about AD and IRMD). The authors reported that the two orientation subscales
CROSSCULTURAL DIFFERENCES IN DEMENTIA 37
had high internal reliabilities (α = .86 for Anglo and .89 for Hispanic) and good construct
validity in their sample. The Cronbach’s alpha value for this scale in the subsample of Hispanic
informants was α = .80.
Second, because measures of acculturation as derived from scales may only or primarily
tap language use and other cultural practices (Cuellar et al., 1995; Stephenson, 2000;
Szapocznik, Kurtines, & Hernandez, 1980), I collected other more proximal measures of
acculturation that may be more directly related to the independent variables of interest from this
subsample of Hispanic informants. One such measure included literacy level as measured by the
Cloze method, which requires respondents to fill in words that have been removed from the ends
of 49 sentences (Block & Baldwin, 2010; see Table 28 in the Appendix). The creators of this
measure found it to be a valid and reliable measure of English-language reading comprehension.
Additionally, in order to obtain further descriptive information regarding proxy measures of
acculturation among the Hispanic subsample, I also collected data on whether informants were
born in the US and whether English was their first language.
Clinicians’ assessment of length of time between onset of cognitive decline and initial
evaluation. To calculate the length of time between patients’ onset of cognitive decline and their
initial evaluation among those diagnosed with dementia, I subtracted the patients’ age at which
the cognitive decline began from their age at the time of initial evaluation, as assessed by the
clinician. According to this study protocol’s coding guidebook as outlined by the NACC,
clinicians were instructed to rely on the information obtained from patients, informants, medical
records, and/or observation to assess the age at which the cognitive decline began. The coding
guidebook also specifies that clinicians not rely on either NP (with the sole exception of the
MMSE) or imaging results in determining the length of time since symptom onset. Clinicians
CROSSCULTURAL DIFFERENCES IN DEMENTIA 38
likely relied more if not solely on informants as opposed to patients if patients demonstrated
clear cognitive impairment and/or poor insight.
Clinicians’ diagnosis of cognitive functioning. Based on all of the information (e.g.,
NP, FAQ and NPI-Q scores, neurologic exams, and health history) available to clinicians for
each patient as part of this data set, clinicians, whether individually or through consensus,
provided an overall binary response judgment regarding whether each patient had NCF or
dementia. Specifically, they responded with either yes or no to questions asking if the patient had
NCF (i.e., no MCI, dementia, or other neurological condition resulting in cognitive impairment)
or: i) met criteria for dementia in accordance with standard criteria for either AD (based on the
National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s
Disease and Related Disorders Association (NINCDS/ADRDA) Alzheimer’s criteria; McKhann
et al., 1984) or ii) vascular dementia (based on the National Institute of Neurological Disorders
and Stroke and Association

Internationale pour la Recherche et l’Enseignement en Neurosciences
(NINDS/AIREN) vascular dementia criteria; Roman et al., 1993); or iii) demonstrated sufficient
evidence of other non-Alzheimer’s or vascular types of dementia.
Neuropathological findings. The neuropathological findings from patients’ post-mortem
brain autopsies are available for 828 participants (24 Hispanics and 804 NHWs).
Neuropathologists determined for each participant whether neuropathological findings supported
a diagnosis of some form of dementia (e.g., AD, vascular disease, and Lewy Body Dementia) as
opposed to NCF.
Clinical Dementia Rating Scale. The Clinical Dementia Rating (CDR) Scale (Morris,
1993; see Table 29 in the Appendix) is a widely used scale by clinicians that categorizes patients
in terms of presence and level of severity of cognitive impairment. The CDR measures a variety
CROSSCULTURAL DIFFERENCES IN DEMENTIA 39
of dementia-related domains, including memory, orientation, judgment, problem solving, and
personal care. Each domain is rated by the clinician using the following scale regarding level of
impairment: 0 (none), 0.5 (questionable), 1 (mild), 2 (moderate), and 3 (severe). The standard
global CDR score categorizes patients as follows: 0 (no cognitive impairment), 0.5 (very mild
dementia), 1 (mild dementia), 2 (moderate dementia), and 3 (severe dementia). I used the CDR
as a measure of severity of cognitive and functional impairment in post hoc analyses to
determine whether there are indeed ethnic group differences in terms of level of impairment at
the time of initial evaluation. I also conducted post hoc analyses examining the CDR as an
outcome variable (i.e., rather than diagnosis of NCF versus dementia) to determine whether it
provides more meaningful information about level of impairment.
Data Analysis
First, I evaluated the data for multivariate normality, namely kurtosis, which is a critically
important assumption of structural equation modeling (Byrne, 2010). Next, I performed
regression diagnostics (i.e., examined the tolerance and variance inflation factor values) to
determine whether multicollinearity was present. I then performed descriptive data analyses
using SPSS 17.0 to describe the entire sample of patients in terms of age (in years), level of
education (in years), ethnicity (Hispanic or NHW), sex (female or male), and diagnosis (NCF or
dementia), as well as informants’ years of education. Additionally, I reported descriptive
statistics for the subsample. I provided descriptive statistics (e.g., means, standard deviations,
and frequencies) for each of the cultural groups separately and by diagnosis on the demographic
and all other key variables (NP performance, IRMD, and total scores on informant-based
measures). In addition, I established the interrelationships between the variables using
correlation coefficients.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 40
To test Hypothesis 1, I assessed for significant cultural group differences among mean
levels of the key continuous variables using independent samples t tests, chi-square tests for
differences in proportions of the key binary variable (IRMD), and Mann-Whitney U tests for the
ordinal CDR variable for those patients diagnosed by a clinician with dementia. I also reported
any significant differences on these variables among patients diagnosed with NCF.
To test Hypotheses 2.i and 3.i, I performed CFAs on the sample of Hispanic and NHW
patients using AMOS 17.0. As described by Byrne (2010), I conducted tests for factorial
invariance of the scales in two stages. First, I created baseline models for each scale separately
for each ethnic group and evaluated the goodness-of-fit to the data before testing for factorial
invariance across the multigroup sample (Bryant & Yarnold, 2000; Byrne, 2010). Second, I
examined three assumptions based on the baseline models estimated for the two groups to assess
each of the models’ factorial invariance across the ethnic groups (Byrne, 2010). The first
assumption I tested, configural invariance, was that the number of factors hypothesized for each
scale was the same across ethnic groups. The second assumption that I tested, measurement
invariance, was that the factor structure, or pattern of factor loadings, was invariant across the
two samples. The third assumption I tested, structural invariance, was that the structural relations
(i.e., factorial covariances) were equivalent across the two ethnic groups. This assumption only
needed to be tested for the NPI-Q, as the FAQ was hypothesized as only having one factor.
Lastly, the factors from both scales were assessed for scalar (i.e., intercept) invariance across
ethnic groups before assessing for latent mean value differences.
To test the first assumption, which holds that the measures are best explained by a
hypothesized factor structure, I began with a concurrent examination of baseline models through
multigroup analysis without specifying equality constraints. I determined the acceptability of the
CROSSCULTURAL DIFFERENCES IN DEMENTIA 41
models’ assumed factor structure with goodness-of-fit statistics. A well-fitting model suggests
that the structure best fits the data across groups.
The second assumption assesses for the invariance of the pattern of factor loadings across
groups. To test for the invariance of factor loadings, models were re-specified with equality
constraints on all of the loadings. If the two models significantly differed in terms of fit, then I
would conclude that the models had different loadings across groups. This finding would lead
me to conduct post hoc model fitting to discover which particular loadings are invariant across
the two groups. Should the tests for the invariance of the factor structure result in a statistically
nonsignificant difference in fit, then I would assume invariance of factor loadings across the
Hispanic and NHW groups. With this outcome (i.e., assuming equivalent factor loadings), I
would then test the assumption of equivalent factor covariances across groups by keeping the
equality constraints of the second model (given that the pattern of the factor loadings was
deemed to be tenable) and testing a still more restrictive model with the covariance matrix
constrained to be equal across the two groups (Byrne, 2010). A statistically significant change in
model fit would suggest that the model is assumed to have different factor variances/covariances
across groups, whereas a nonsignificant change would point to invariant factor covariances.
Although each scale may or may not be invariant in terms of their underlying factor
structure, item mean scores may not be comparable if both ethnic groups are treated as a single
sample in the case that scalar invariance is indeed not found. Therefore, to test for scalar
invariance, I followed the steps outlined by Byrne (2010). My first step was to constrain the
intercepts to be equal across both groups. Then, I allowed the factor mean(s) for one of the
groups to be freely estimated while constraining the other group’s to zero. Next, I examined the
goodness-of-fit statistics for both models to determine if they fit the data well before examining
CROSSCULTURAL DIFFERENCES IN DEMENTIA 42
for statistically significant differences on the latent mean estimates. Critical ratio values greater
than 1.96 indicate statistically significant differences regarding latent mean values. It should be
noted that although Byrne stated that the Jöreskog tradition of testing for invariance historically
called for the testing of the equality of error variances and their covariances, it is now widely
accepted that doing so represents an excessively restrictive test of the data.
I conducted all initial analyses of the invariance of the measurement instruments based on
analysis of the covariance structures with CFA models based on maximum likelihood methods
using AMOS 17.0. To evaluate the criteria used to accept a model in the CFA, I examined
various statistics indicating goodness-of-fit for the models as a whole, including the Goodness-
of-Fit Indices (GFI), Comparative Fit Indices (CFI), Tucker-Lewis Indices (TLI), and the root
mean square error of approximation (RMSEA). Models with a better fit generally have higher
CFI and TLI and lower RMSEA values. Though the rules for establishing goodness-of-fit for
models vary, adequate fit is indicated by GFI and CFI values greater than .90, TLI values greater
than .85, and RMSEA values of up to .08 (Hair, Anderson, Tatham, & Black, 1998; Hu &
Bentler, 1995). However, Browne and Cudeck (1993) reported that RMSEA values of up to 0.10
signify reasonable approximation errors in the population. To assess for statistically significant
changes in model fit, I examined the ΔCFI values, which have been shown to be a reasonable
means by which to base invariance decisions. Specifically, Cheung and Rensvold (2002)
suggested that ΔCFI values of less than .01 point to invariance.
To Test Hypothesis 4, I tested the path models using AMOS 17.0 by examination of the
statistical significance of estimated path coefficients and statistics indicating goodness-of-fit for
the models as a whole, as described above with the sole exception of the use of the Expected
Cross-validation Index (ECVI) in lieu of the GFI as I planned to compare a series of models to
CROSSCULTURAL DIFFERENCES IN DEMENTIA 43
obtain a final well-fitting model (Byrne, 2010). Because ECVI values can take on any value,
there is no established appropriate range of values. However, ECVIs with the smallest values are
deemed to have the greatest potential for replications. Because the dependent (i.e., endogenous)
variable (diagnosis of either dementia or NCF) was binary, I used Markov chain Monte Carlo
methods for these path analyses. I estimated the disturbance term of the dependent variable
(clinicians’ diagnosis) to acknowledge that various factors other than the exogenous (i.e.,
independent) variables in my models accounted for this outcome. Correlations among the
exogenous variables were examined before restricting the covariances between them to zero. In
cases where fit statistics of the models indicated acceptable levels of model fit, final models were
estimated after nonsignificant paths were deleted from the originally proposed models in the
interest of parsimony. The fit statistics of the reduced models were then tested to see if the data
fit these models well and whether they were significantly different in comparison to the original
models. To test the sub-hypotheses regarding specific relations among the variables, the path
coefficients among these variables in both the first and reduced models (in the case of deleting
nonsignificant paths) were assessed for statistical significance and compared across groups.
Hypothesis 5 predicted significant mean-level differences in terms of knowledge about
AD, as measured by the ADKS, with Hispanics having significantly lower levels of knowledge
than NHWs. To test this hypothesis, I used an independent samples t test to assess for a
statistically significant difference in mean values on this variable across the ethnic groups.
To test Hypothesis 6.i, I examined the internal consistency reliability (KR-20) of the
CBADS for each ethnic group. Then, for Hypothesis 6.ii, I tested for a statistically significant
difference in mean values on this scale across ethnic groups using an independent samples t test.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 44
For Hypothesis 7, I used the same goodness-of-fit statistics and analytic methods to
evaluate model fit as previously described. Specifically, I assessed the path coefficients among
these variables for statistical significance and compared them across groups. In addition, I
compared the goodness-of-fit statistics of these path models to those from Hypothesis 4 to
examine whether these models fit the data similarly.
To compare clinicians’ diagnoses with neuropathological findings among the participants
for whom this information is available, I compared the clinical diagnoses of NCF versus
dementia to the neuropathological findings. Because the neuropathological findings included
different types of dementia (e.g., AD and frontotemporal lobar degeneration) as well as normal
brain findings based on the autopsies, I recoded the findings as either normal or dementia in
order to directly compare them with the clinical diagnosis. I then calculated Cohen’s kappa
values for the overall sample as well as separately for each ethnic group. To evaluate whether
diagnostic agreement (as measured by Cohen’s kappa) differed significantly across the two
ethnic groups, I calculated a z-statistic value using the kappa values and their standard errors. I
considered differences to be statistically significant if the p value was less than .05.
Statistical power and effect sizes. I used a cutoff value of p < .05 to assess for statistical
significance in all relevant analyses. Given the large sample of NHWs, especially as compared to
the Hispanics, there was high statistical power that could allow for the detection of statistical
significance for small effects. Thus, I reported effect sizes (e.g., Cohen’s d for group-mean
differences, r coefficient for the ordinal CDR variable, and standardized path coefficients for the
path models) to assess the meaningfulness of any findings that were found to be statistically
significant. Though rules vary, Cohen (1988) suggested that effect sizes of approximately 0.10
are small, 0.30 are medium, and 0.50 are large. For the purposes of this study’s analyses
CROSSCULTURAL DIFFERENCES IN DEMENTIA 45
involving the larger national dataset, I only considered effect sizes greater than 0.20 to be
meaningful given the large sample.
Results
I begin by presenting descriptive statistics and key results for the larger national sample
(Hypotheses 1-4). I then provide descriptive statistics and results for the subsample of informant
participants from whom I collected additional data (Hypotheses 5-7).
Descriptive Statistics for the National Dataset
Analyses revealed that the national dataset showed evidence of multivariate nonnormality
(i.e., kurtosis) for both ethnic groups. Therefore, structural equation modeling analyses were
based on asymptotic distribution-free estimation (Browne, 1984). Regression diagnostics did not
result in evidence of problematic multicollinearity as assessed by tolerance and variance inflation
factor values.
The national sample was composed of 444 Hispanic and 11,081 NHW patients and their
informants (all assessed in English). A total of 6,323 (54.9%) patients were diagnosed with
dementia and 5,202 patients were diagnosed with NCF at their initial evaluations. Of the total
sample, 6,365 patients (55.2%) were female, and the ages of the patients ranged from 20 to 104
years (M = 72.57, SD = 10.76), whereas patient education levels ranged from 0 to 29 years (M =
15.18, SD = 3.05). Of the 444 Hispanic patients, 239 (53.8%) were Mexican/Chicano/Mexican
American, 81 (18.2%) were Puerto Rican, 26 (5.9%) were South American, 19 (4.3%) were
Central American, 19 (4.3%) were Cuban, 8 (1.8%) were Dominican, 31 (7.0%) were classified
as other, and 21 (4.7%) were coded as unknown. The origins of the Hispanic informants were
similar to those of the patients in terms of frequency. Of the Hispanic patients, 243 (54.7%) were
CROSSCULTURAL DIFFERENCES IN DEMENTIA 46
diagnosed with dementia and 201 with NCF. Among the NHW patients, there were similar
proportions of diagnoses, with 6,080 (54.9%) diagnosed with dementia and 5,001 with NCF.
Table 1 provides additional descriptive information on the demographic and other key
variables in this study separated by both ethnicity (Hispanic and NHW) and diagnosis (combined
NCF and dementia, NCF only, and dementia only). There were a number of significant
crosscultural differences on certain variables among the combined NCF and dementia sample.
For example, both Hispanic patients and informants had significantly lower education levels than
their NHW counterparts, t(460.60) = 10.65, p < .001, and t(428.66) = 6.85, p < .001,
respectively. These differences remained significant (p < .001) regardless of diagnosis. The
ranges of years of education for participants were as follows: Hispanic patients, 0 to 27; Hispanic
informants, 3 to 22; NHW patients, 0 to 29; and NHW informants, 0 to 33. Regarding
informants’ relationships to patients, among Hispanics, 43.5% of informants were spouses,
33.6% were adult children, and the remainder fell into another category (e.g., other relative,
friend/neighbor, and paid caregiver/provider), whereas among NHWs, these respective figures
were 61.0% and 23.3%, χ
2
(6, N = 11,525) = 95.81, p < .001. This difference was significant (p <
.001) regardless of diagnosis. In addition, the percentage of informants who coresided with
patients was significantly higher among NHWs (65.2%) as compared to Hispanics (57.2%), χ
2
(1,
N = 11,525) = 11.92, p = .001. This difference remained significant among the patients with
dementia, χ
2
(1, N = 6,323) = 12.27, p < .001, but not among those diagnosed with NCF.
There was a significant between-diagnosis difference among Hispanic patients.
Specifically, Hispanics diagnosed with dementia at their initial evaluations were significantly
less likely to have endorsed having English as a primary language than Hispanics diagnosed with
NCF, χ
2
(1, N = 444) = 15.14, p < .001.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 47
Table 2 provides the correlation coefficients among the demographic and key variables
for the entire national sample. There were statistically significant correlations among several of
the variables. All covariates (age, sex, and patient and informant education) and independent
variables (overall NP, FAQ, NPI-Q frequency and severity, and IRMD) were significantly
correlated with clinicians’ diagnosis. However, the correlations of diagnosis with the covariates
were comparatively much lower than those of the key independent variables with diagnosis,
ranging from r = ± .07 to ± .17 for the covariates to r = ± .50 to ± .83 for the key independent
variables. Tables 3 and 4 show these same correlation coefficients separated by ethnicity
(Hispanic and NHW, respectively). Similarly, all covariates (with the sole exception of
informant education among Hispanics only) and independent variables were significantly
correlated with diagnosis. Among Hispanics, the correlation coefficients ranged from ± .17 to ±
.19 for the covariates and from ± .52 to ± .88 for the key independent variables. Among NHWs,
these values ranged from ± .07 to ± .17 and ± .50 to ± .83, respectively.
Hypothesis 1: Crosscultural Differences in Impairment at and Time to Initial Evaluation
Table 1 also provides information on the significance levels of the chi-square and t tests
that I used to test for significant differences in the demographic and other key variables.
Hypothesis 1 projected that Hispanics with dementia would be significantly more impaired and
have had a longer delay to their initial evaluations than their NHW counterparts. Results revealed
mixed support for this hypothesis. First, Hispanics with dementia indeed exhibited higher levels
of cognitive impairment, as measured by overall NP, than NHWs with dementia when using
CEGN, t(6,321) = 5.11, p < .001, d = -0.35. However, the opposite pattern emerged when using
EGSN in that NHWs with dementia were significantly more impaired than their Hispanic
counterparts, t(271.42) = -3.46, p = .001, d = 0.21. Second, I did not find evidence to support my
CROSSCULTURAL DIFFERENCES IN DEMENTIA 48
hypothesis that the length of time between the onset of cognitive decline and initial evaluation
(as assessed by the clinician) would differ significantly across ethnic groups (p = .471). Third, I
found that Hispanics with dementia had a greater likelihood of having IRMD than their NHW
counterparts in that 239 (98.4%) Hispanics had IRMD as compared to 5,820 (95.7%) NHWs,
t(6,319) = -2.31, p = .021, d = -0.06. Fourth, my independent samples t test revealed that
functional impairment levels (as measured by the FAQ) were significantly higher among
Hispanics compared to NHWs with dementia, t(6,321) = -2.27, p = .023, d = -0.06, Fifth, I found
that Hispanics with dementia had significantly higher levels of BPS severity (as measured by the
NPI-Q) than their NHW counterparts, t(6,321) = -2.93, p = .003, d = -0.07, but not BPS
frequency levels (p = .088). Finally, my Mann-Whitney U test for the CDR revealed a significant
crosscultural difference in that Hispanics diagnosed with dementia had significantly worse scores
on this global measure of cognitive and functional impairment than their NHW counterparts, z =
-3.78, p < .001, r = -0.05.
As previously noted, I only deemed statistically significant group differences as
meaningful if the effect sizes exceeded 0.20. Using this criterion, the only significant and
meaningful group differences revealed by my analyses were the aforementioned differences in
NP when using both CEGN and EGSN.
Hypothesis 2: Testing for the Crosscultural Factorial and Scalar Invariance of the
Functional Assessment Questionnaire
Hypothesis 2 pertained to the crosscultural factorial and scalar invariance of the FAQ. I
conducted tests for factorial invariance in two stages, as suggested by Byrne (2010). First, I
established baseline models separately for each group and evaluated for adequacy to the
Hispanic and NHW data. Next, based on the baseline models estimated for both groups, I
CROSSCULTURAL DIFFERENCES IN DEMENTIA 49
investigated two assumptions to test for the models’ invariance across ethnic groups (Bryant &
Yarnold, 2000; Byrne, 2010).
Establishing the baseline model. Hypothesis 2 predicted that the FAQ would have a
crossculturally invariant one-factor structure reflecting dementia severity. The baseline FAQ
model for both Hispanic and NHW participants is shown in Figure 5. The hypothesized structure
is composed of one latent variable representing dementia severity, with this latent variable
having 10 indicators.
For the Hispanic group, my CFA yielded a GFI value of .958, a CFI value of .916, a TLI
value of .892, and a RMSEA value of .087, 90% confidence interval (CI) [.073, .101]. All of
these values represent indicators of reasonable to good model fit. When I tested the model for the
NHW group, results yielded a GFI value of .949, a CFI value of .894, a TLI value of .864, and a
RMSEA value of .080, 90% CI [.078, .083]. Overall, this model also appeared to adequately fit
the data. The factor loadings for both models are listed in Table 5.
Testing for multigroup factorial invariance. Next, I tested for the multigroup factorial
invariance of the FAQ across the Hispanic and NHW groups in two steps. First, I tested whether
the one-factor structure was equivalent across groups. My CFA resulted in fit indices suggestive
of adequate fit: GFI = .950, CFI = .895, TLI = .865, and RMSEA = .057, 90% CI [055, .059].
These results suggested that the FAQ structure is indeed most appropriately described by a one-
factor (dementia severity) model for both Hispanics and NHWs in this sample. It does not
necessarily imply that the pattern of factor loadings is the same across the two ethnic groups.
Goodness-of-fit statistics pertinent to this model (Model 1) as well as all subsequent models are
presented in Table 6.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 50
Second, I assessed whether the pattern of factor loadings was crossculturally equivalent.
In testing the assumption of factor loading equivalency, I placed equality constraints on all factor
loadings in Model 1, with Item 1 fixed to a value of 1.0 for purposes of statistical identification
and scaling. Results from this test of Model 2 determined that the postulated equality of factor
loadings across the two groups was tenable, ΔCFI = .00. The ΔCFI < .01 represented a nominal
difference between Models 1 and 2, suggesting that pattern of factor loadings is indeed invariant
a the Hispanic and NHW groups.
In sum, as indicated by the goodness-of-fit statistics and the results summarized in Table
6, I found that the FAQ was adequately well-described by the hypothesized one-factor (dementia
severity) model for both Hispanics and NHWs in this study. In addition, I found that the
observed measures were operating equivalently for both ethnic groups. These findings suggest
that the FAQ is measuring the same underlying construct across these ethnic groups and that
meaningful interpretations can be made using this scale across these groups.
Testing for scalar invariance. To test for the crosscultural scalar (i.e., intercept)
invariance of the FAQ, I first constrained the intercepts to be equal across both groups and then
allowed the factor mean for one of the groups to be freely estimated while constraining the other
group’s to zero. Table 6 provides the goodness-of-fit statistics for this structured means model
(Model 3). Both the CFI value of .910 and the TLI value of .908 were suggestive of an
adequately well-fitting model. The RMSEA value of .107 (90% CI [.106, .109] slightly exceeded
the .10 cutoff suggested by Browne and Cudeck (1993) as a measure of adequate model fit. The
increases in the CFI value for this model when compared to both the configural model (Model 1)
and the measurement model (Model 2) were less than .01 (with rounding), indicative of a
nominal change in fit. In sum, this structured means model fit the data reasonably well (e.g., CFI
CROSSCULTURAL DIFFERENCES IN DEMENTIA 51
= .910 and TLI = .908) and demonstrated a nearly adequate approximation of the two ethnic
group populations (RMSEA = .107). Moreover, the changes in CFI values were nominal
(suggestive of invariance) when imposing equality constraints on both the factor loadings and
observed variable intercepts across groups. Given these findings, I was able to meaningfully
interpret and test for crosscultural differences in the latent mean estimates associated with the
current solution.
In light of the evidence I found for configural, measurement, and scalar invariance across
groups, I assessed for statistically significant differences on the latent mean estimates for the
dementia severity factor derived from the FAQ. Accordingly, I set the latent mean value to zero
in the NHW group and freely estimated the latent mean value for Hispanics. Critical ratio values
greater than 1.96 indicate statistically significant differences in latent mean values across groups
(Byrne, 2010). Results revealed that there was no significant latent mean difference across
cultures for the dementia severity factor of the FAQ, z = 1.41, p = .16. This result is consistent
with the independent samples t test comparing observed mean differences on the FAQ across
ethnicities for the combined dementia and NCF sample, t(475.55) = -1.28, p = .20. When
examining the dementia groups only, results revealed a significant latent mean difference across
Hispanics and NHWs, z = 2.51, p = .01, with Hispanics having significantly higher latent mean
scores by a value of .14. This result is also consistent with the independent samples t test
comparing observed mean differences on the FAQ across ethnicities among the dementia sample
only, t(6,321) = -2.27, p = .02.

CROSSCULTURAL DIFFERENCES IN DEMENTIA 52
Hypothesis 3: Testing for the Crosscultural Factorial and Scalar Invariance of the
Neuropsychiatric Inventory Questionnaire
Hypothesis 3 pertained to the crosscultural factorial and scalar invariance of the NPI-Q. I
conducted tests for factorial invariance in two stages, as suggested by Byrne (2010). First, I
established baseline models separately for each of the groups and evaluated for adequacy to the
Hispanic and NHW data. Next, based on the baseline models estimated for the two groups, I
investigated three assumptions to test for the models’ invariance across ethnic groups (Bryant &
Yarnold, 2000; Byrne, 2010).
Establishing the baseline model. Hypothesis 3 predicted that the NPI-Q would have a
crossculturally invariant four-factor structure. The baseline NPI-Q model for both Hispanic and
NHW participants is shown in Figure 6. The hypothesized structure is composed of four latent
variables representing hyperactivity, affect, psychosis, and apathy/vegetative symptoms, with
each latent variable having three, two, two, and three indicators, respectively. The four latent
variables were also set to be correlated with each other, because they are theoretically all areas of
dementia-related BPS.
For the Hispanic group, my CFA yielded a GFI value of .966, a CFI value of .921, a TLI
value of .877, and a RMSEA value of .038, 90% CI [.017, .057]. All of these values represent
indicators of adequate to good model fit. When I tested the model for the NHW group, results
yielded a GFI value of .984, a CFI value of .918, a TLI value of .872, and a RMSEA value of
.034, 90% CI [.031, .037]. Overall, this four-factor model also appeared to fit the data reasonably
well. The factor loadings and correlations for both models are listed in Tables 7 and 8,
respectively.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 53
Testing for multigroup factorial invariance. Next, I tested for the multigroup factorial
invariance of the NPI-Q across the Hispanic and NHW groups in three steps. First, I tested
whether the four-factor structure was equivalent across groups. My CFA resulted in the
following fit indices, which were indicative of adequate to good model fit: GFI = .983, CFI =
.918, TLI = .873, RMSEA = .024, 90% CI [.022, .026]. These results suggested that the NPI-Q
structure is indeed most appropriately described by a four-factor (hyperactivity, affect, psychosis,
and apathy/vegetative symptoms) model for both Hispanics and NHWs in this sample. It does not
necessarily imply that the pattern of factor loadings is the same across the two ethnic groups.
Goodness-of-fit statistics pertinent to this model (Model 1) as well as all subsequent models are
presented in Table 9.
Second, I assessed whether the pattern of factor loadings was crossculturally equivalent.
In testing the assumption of factor loading equivalency, I placed equality constraints on all factor
loadings in Model 1, with the first item from each factor fixed to a value of 1.0 for purposes of
statistical identification and scaling. Results from this test of Model 2 determined that the
postulated equality of factor loadings across the two groups was tenable, ΔCFI = .00. The ΔCFI
< .01 represented a nominal difference between Models 1 and 2, suggesting that pattern of factor
loadings was invariant for the Hispanic and NHW groups.
Third, I examined whether the structural relations (i.e., factor covariances) were
equivalent across groups. In testing for the invariance of factor covariances across groups, the
assumption of interest focused on the structure, or different facets, of the NPI-Q, namely
hyperactivity, affect, psychosis, and apathy/vegetative symptoms. As found in Model 2, that the
equality of the pattern of factor loadings was tenable, these constraints were also maintained for
Model 3. In addition, the entire covariance matrix was constrained equal across groups. As
CROSSCULTURAL DIFFERENCES IN DEMENTIA 54
reported in Table 9, results from the estimation of Model 3 yielded the following fit indices
suggestive to adequate to good model fit: GFI = .982, CFI = .918; TLI = .894; and RMSEA =
.022, 90% CI [.020, .024]. Because the ΔCFI of .00 was indicative of a nominal change in fit, I
accepted the assumption of crossculturally invariant factor covariances.
In summary, as indicated by the goodness-of-fit statistics and the results summarized in
Table 9, I confirmed my hypothesis that the NPI-Q structure was well-described by the four-
factor model composed of different facets of BPS (hyperactivity, affect, psychosis, and
apathy/vegetative symptoms) for both Hispanics and NHWs. In addition, I found that the
observed measures were operating equivalently for both groups. Lastly, my analyses revealed
that the structural relations among the NPI-Q facets were equal across groups. These findings
suggest that the NPI-Q is measuring the same underlying construct across both groups and that
meaningful interpretations can be made using this scale across groups.
Testing for scalar invariance. To test for the crosscultural scalar (i.e., intercept)
invariance of the NPI-Q, I first constrained the intercepts to be equal across both groups and then
allowed the factor mean for one of the groups to be freely estimated while constraining the other
group’s to zero. Table 9 provides the goodness-of-fit statistics for this structured means model
(Model 4). The fit indices were all suggestive of a well-fitting model (CFI = .963, TLI = .953,
and RMSEA = .030, 90% CI [.028, .031]). However, the ΔCFI value exceeded .01 for this model
compared to both the configural model (Model 1), ΔCFI = .045, and the measurement model
(Model 2), ΔCFI = .045. In sum, although this structured means model fit the data quite well, the
change in the CFI value did not provide evidence for full scalar invariance across groups.
Because I did not find evidence for full scalar invariance, my next step was to conduct
analyses to assess for partial scalar invariance, which is suggested as a compromise between full
CROSSCULTURAL DIFFERENCES IN DEMENTIA 55
and a lack of measurement invariance. A conservative method of implementing partial
invariance constraints is to use modification indices (MI) to determine which constraints to relax.
Therefore, I examined the MI and degree of expected parameter change to decide whether any
modifications to the structured means model were indicated (Steenkamp & Baumgartner, 1998).
I considered an MI value greater than 20 to be indicative of a significant scalar invariance
problem, though I did not find any intercept MI values that met this criterion. An alternative
approach to testing for partial scalar invariance is to detect noninvariant item intercepts by
sequentially removing equality constraints on each item’s measurement intercept and examining
the ΔCFI value until one identifies the noninvariant items (Van Lieshout, Cleverley, Jenkins, &
Georgiades, 2011). However, I still did not find evidence for partial scalar invariance using this
approach, as the CFI values were identical or very similar in all cases. Therefore, because I did
not find evidence of either full or partial scalar invariance for the NPI-Q across cultural groups, I
could not meaningfully test for statistically significant differences on the latent mean value
estimates for the four factors derived from the NPI-Q.
Hypothesis 4: Testing Path Models for Crosscultural Differences in Dementia Diagnosis
Hypothesis 4 pertained to posited crosscultural differences in the associations of NP,
IRMD, and informant-reported functional abilities and BPS with diagnosis, as depicted in path
models. Table 10 provides the goodness-of-fit statistics for the first and reduced path models for
Hispanics. Among this group, the first (i.e., hypothesized) model, presented with standardized
path coefficients in Figure 7, demonstrated adequate fit overall (CFI = .964, TLI = .897, RMSEA
= .092, ECVI = .413). Therefore, my next step was to examine the statistical significance of the
estimated path coefficients and delete nonsignificant paths in the interest of parsimony before
reevaluating model fit. The nine nonsignificant paths that I deleted were: sex to NP; age, sex, and
CROSSCULTURAL DIFFERENCES IN DEMENTIA 56
informant education to NPI-Q frequency; informant education to IRMD; and age, sex, primary
language, and NPI-Q to diagnosis. This reduced model #1 resulted in adequate to good model fit
(CFI = .959, TLI = .919, RMSEA = .082, ECVI = .418). In addition, though the ECVI increased
slightly, the change in the CFI value was less than .01, suggesting that this reduced model #1 did
not meaningfully differ from the first model. Because this reduced model resulted in two
exogenous variables (sex and informant education) as well as one endogenous variable (NPI-Q)
that were not associated with other variables in the model, I also removed these variables in the
interest of parsimony before reevaluating this reduced model #2 for model fit. This reduced
model resulted in adequate fit overall, particularly in terms of the CFI and TLI values (CFI =
.969, TLI = .890, RMSEA = .122, ECVI = .260). Moreover, the change in the CFI value was less
than .01, ΔCFI = .005, and the ECVI value decreased. Therefore, I accepted this more
parsimonious reduced model #2 as the final model.
The unstandardized path coefficients for both the first and reduced models (reduced
model #2) for Hispanics are presented in Table 11. Figure 8 presents the final reduced path
model (reduced model #2) with the standardized path coefficients for Hispanics. A closer
examination of the reduced model #2 for Hispanics provides partial support for my hypothesis.
Overall NP, IRMD, and informant reports on patients’ functional abilities were significantly
related to diagnosis. However, informant-reported BPS was not associated with diagnosis. As
evidenced by the estimated standardized path coefficients, IRMD had the strongest association
with diagnosis (β = .60), followed by the FAQ (β = .21), then overall NP (β = -.18). These
findings are partially consistent with my hypothesis in that of all of the key mediating variables
that remained in the final model, overall NP had the weakest association with diagnosis as
CROSSCULTURAL DIFFERENCES IN DEMENTIA 57
compared to the informant-report variables (including the NPI-Q, which was nonsignificant).
Nonetheless, NP was significantly associated with diagnosis among Hispanics.
Table 10 provides the goodness-of-fit statistics for the first and reduced path models for
NHWs. The first model, presented with standardized path coefficients in Figure 9, also
demonstrated adequate to good fit overall (CFI = .966, TLI = .881, RMSEA = .103, ECVI =
.145). The reduced model in which I deleted the sole nonsignificant path (from education to
diagnosis) also demonstrated adequate to good fit (CFI = .966, TLI = .889, RMSEA = .099,
ECVI = .145) and resulted in a nominal change in fit, ΔCFI = .00. Therefore, I accepted this
more parsimonious reduced model as my final path model for NHWs.
The unstandardized path coefficients for both the first and reduced models for NHWs are
presented in Table 12. Figure 10 illustrates the final reduced path model with the standardized
path coefficients for NHWs. The reduced model for NHWs provides partial support for my
hypothesis in that all four mediating variables were significantly related to diagnosis. The
estimated standardized path coefficients revealed that IRMD had the strongest association with
diagnosis (β = .52), followed by overall NP (β = -.24), the FAQ (β = .19), and NPI-Q frequency
(β = .06). These findings are partially consistent with my hypothesis in that aside from IRMD,
overall NP had the strongest association with diagnosis as compared to the other informant-
report variables. It should be noted that the effect size for the association of the NPI-Q with
diagnosis (β = .06) was small, suggesting that this result may not be meaningful.
I then evaluated Hypotheses 4.i and 4.ii by looking for crosscultural differences in the
relations of the four key mediating variables with diagnosis. Specifically, Hypothesis 4.i held that
overall NP would be more strongly associated with diagnosis among NHWs than Hispanics,
whereas Hypothesis 4.ii posited that IRMD and informant-reported functional abilities (FAQ)
CROSSCULTURAL DIFFERENCES IN DEMENTIA 58
and BPS (NPI-Q) would be more strongly associated with diagnosis among Hispanics than
NHWs. As can be seen by the estimated standardized path coefficients in the final (i.e., reduced)
path models for each group in Figures 8 and 10, these hypotheses received partial support in that
the ranks of the sizes of the coefficients for overall NP, FAQ, and IRMD were slightly different
across groups. Specifically, the FAQ was most strongly associated with diagnosis among
Hispanics (after IRMD) followed by overall NP, whereas this pattern was the opposite among
NHWs. Perhaps the most major unanticipated crosscultural difference that I found was that the
NPI-Q was only significantly associated with diagnosis in NHWs, in contrast to my hypothesis
that this association would be stronger in Hispanics.
I also assessed the potentially confounding roles of informants’ relationship to and
coresidency with patients, as these could presumably affect informants’ awareness of patients’
dementia symptoms and thus their reporting of such symptoms during evaluations. When
controlling for these two variables in the Hispanic sample, results revealed that the fit of the
models was significantly poorer compared to my originally proposed path models as assessed by
the changes in CFI values exceeding .01 (ΔCFI = .012 and ΔCFI = .023, respectively) and higher
ECVI values. In addition, neither variable was significantly associated with diagnosis (p = .881
and p = .376, respectively). Therefore, I determined that my originally proposed final models fit
significantly better and did not control for these variables. When controlling for these variables
in the NHW sample, although both variables had significant paths to diagnosis (p < .001), the fit
of these models was significantly worse than my originally proposed models as assessed by the
changes in the CFI values exceeding .01 (ΔCFI = .024 and ΔCFI = .034, respectively) and higher
ECVI values. Therefore, I determined that my originally proposed final models fit the data
significantly better and did not control for these variables.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 59
Path models using the combined ethnic-group neuropsychological test norms. As
discussed earlier, I conducted my path model analyses again using CEGN (rather than EGSN)
for the overall NP variable to assess for changes among the variables both within and across
groups. Table 13 provides the goodness-of-fit statistics for the first and reduced path models for
Hispanics. Among Hispanics, the first (i.e., hypothesized) model, presented with standardized
path coefficients in Figure 11, demonstrated adequate to good fit (CFI = .962, TLI = .890,
RMSEA = .093, ECVI = .415). After examining the statistical significance of the estimated path
coefficients and deleting the eight nonsignificant paths, I reevaluated the fit of this reduced
model. The eight nonsignificant paths that I deleted were: age, sex, and informant education to
NPI-Q frequency; informant education to IRMD; and age, sex, primary language, and NPI-Q to
diagnosis. This reduced model #1 resulted in adequate to good model fit (CFI = .960, TLI = .918,
RMSEA = .080, ECVI = .407), a nominal change in fit (ΔCFI = .002), and a smaller ECVI
coefficient. Because this reduced model resulted in one exogenous (informant education) and
one endogenous variable (NPI-Q) that were not significantly associated with other variables, I
removed these variables and reassessed the fit of this reduced model #2. Results revealed a
similar pattern of results in that this model demonstrated adequate to good fit overall (CFI =
.957, TLI = .888, RMSEA = .107, ECVI = .326), a nominal change in CFI (ΔCFI = .005), and a
smaller ECVI value, pointing toward the goodness-of-fit of this more parsimonious model for
Hispanics and a nominal difference in fit between this model and the first model. Therefore, I
accepted reduced model #2 as the final model.
The unstandardized path coefficients for both the first and reduced models (reduced
model #2) for Hispanics are presented in Table 14. Figure 12 illustrates the final reduced path
model (reduced model #2) with the standardized path coefficients for this group. When
CROSSCULTURAL DIFFERENCES IN DEMENTIA 60
comparing this model with CEGN to the reduced model with EGSN for Hispanics (Figure 8), the
only key change was that patients’ sex remained in the final model and was significantly
associated with overall NP (β = .10).
Table 13 provides the goodness-of-fit statistics for the first and reduced path models for
NHWs. The first model, presented with standardized path coefficients in Figure 13, also
demonstrated adequate to good fit overall (CFI = .964, TLI = .876, RMSEA = .103, ECVI =
.146). The reduced model in which I deleted the sole nonsignificant path (from patients’ sex to
diagnosis) also demonstrated adequate to good fit (CFI = .964, TLI = .885, RMSEA = .099,
ECVI = .146) and resulted in a nominal change in fit, ΔCFI = .00. Therefore, I accepted this
more parsimonious reduced model as my final path model for NHWs.
The unstandardized path coefficients for both the first and reduced models for NHWs are
presented in Table 15. Figure 14 illustrates the final reduced path model with the standardized
path coefficients for NHWs. When comparing this model with CEGN to the reduced model with
EGSN for NHWs (Figure 10), the only key changes were that patients’ sex was no longer
directly associated with diagnosis whereas education level was.
I then reexamined Hypotheses 4.i and 4.ii using CEGN for the overall NP variable.
Hypothesis 4.i held that NP would be more strongly associated with diagnosis among NHWs
than Hispanics, whereas Hypothesis 4.ii posited that IRMD and informant-reported functional
abilities (FAQ) and BPS (NPI-Q) would be more strongly associated with diagnosis among
Hispanics than NHWs. As evidenced by the estimated path coefficients in the final path models
for each group in Figures 12 and 14, the pattern of findings remained the same as when using
EGSN. Specifically, the ranks of the sizes of the coefficients for overall NP, FAQ, and IRMD
CROSSCULTURAL DIFFERENCES IN DEMENTIA 61
were the same across groups, yet NPI-Q scores were only significantly associated with diagnosis
in NHWs, in contrast to my hypothesis that this association would be stronger in Hispanics.
With regard to crosscultural differences in the covariates in these reduced models, first,
informants’ education level only remained in the reduced model for NHWs and was associated
with lower NPI-Q frequency scores (β = -.02) and decreased odds of IRMD (β = -.02). Second,
patients’ age was only associated with diagnosis (β = -.04) and the NPI-Q (β = -.04) in NHWs.
Finally, patients’ sex was only associated with the NPI-Q in NHWs (β = -.06).
Path models using the Clinical Dementia Rating Scale as the dependent variable. I
also reran my path model analyses using the CDR as my dependent variable to see if this five-
point, ordinal-scale measure of dementia severity would provide richer results than the binary
diagnosis variable. Table 16 provides the goodness-of-fit statistics for the first and reduced path
models for Hispanics. Among this group, the first (i.e., hypothesized) model, presented with
standardized path coefficients in Figure 15, demonstrated adequate to good fit (CFI = .965, TLI
= .900, RMSEA = .092, ECVI = .412). After examining the statistical significance of the
estimated path coefficients and deleting the nine nonsignificant paths, I reevaluated the fit of this
reduced model. The nine nonsignificant paths that I deleted were: age, sex, and informant
education to NPI-Q frequency; informant education to IRMD; and age, sex, education, primary
language, and NPI-Q to diagnosis. This reduced model #1 resulted in adequate to good model fit
(CFI = .966, TLI = .933, RMSEA = .076, ECVI = .390), a nominal change in fit compared to the
first model (ΔCFI = .001), and a smaller ECVI value. The path from sex to NP was no longer
significant in this model, so I subsequently deleted this path and again found a nominal change in
fit in this reduced model #2 (ΔCFI = .002). Because this reduced model resulted in two
exogenous (patients’ sex and informant education) and one endogenous variable (NPI-Q) that
CROSSCULTURAL DIFFERENCES IN DEMENTIA 62
were not significantly associated with other variables, I removed them and reassessed fit for this
reduced model #3. Results revealed that this further reduced model demonstrated adequate to
good fit overall (CFI = .973, TLI = .915, RMSEA = .110, ECVI = .246), a nominal change in
CFI (ΔCFI = .008), and a smaller ECVI coefficient, pointing toward the goodness-of-fit of this
more parsimonious model and a nominal change in fit. Therefore, I accepted reduced model #3
as the final model.
The unstandardized path coefficients for both the first and reduced models (reduced
model #3) for Hispanics are presented in Table 17. Figure 16 illustrates the final reduced path
model (reduced model #3) with the standardized path coefficients for this group. An inspection
of the reduced model #3 for Hispanics provides partial support for my hypothesis in that overall
NP, IRMD, and informant reports on patients’ functional abilities (FAQ) but not BPS (NPI-Q)
were significantly associated with diagnosis. As can be seen in the estimated standardized path
coefficients, the FAQ had the strongest association with diagnosis (β = .54), followed by IRMD
(β = .25), then overall NP (β = -.21). When comparing this model with the model using the
binary clinicians’ diagnosis variable as the outcome (Figure 8), the only key change was that the
FAQ was most strongly associated with CDR scores in the current model, whereas IRMD was in
the former model. Patient education was related to CDR scores, in contrast to diagnosis in the
prior models.
Table 16 provides the goodness-of-fit statistics for the first (i.e., hypothesized) path
model for NHWs. This first model, presented with standardized path coefficients in Figure 17,
demonstrated adequate to good fit overall (CFI = .968, TLI = .890, RMSEA = .102, ECVI =
.145). All of the paths in this model were statistically significant. Therefore, I accepted this first
model as my final path model for NHWs.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 63
The unstandardized path coefficients for this model for NHWs are presented in Table 18.
When comparing this model with the final model using the binary clinicians’ diagnosis variable
as the outcome (Figure 10), the only key changes were that education was significantly
associated with the CDR but not diagnosis and that the FAQ was most strongly associated with
the CDR in the current model, whereas IRMD was in the former model.
I then reexamined these models (i.e., using the CDR as the dependent variable) for each
group using CEGN for the overall NP variable rather than EGSN. As evidenced by the estimated
path coefficients in the final path models for each group in Figures 18 and 19, the pattern of key
findings remained the same as when using EGSN and the CDR as the outcome.
Agreement between Neuropathological Findings and Clinicians’ Diagnosis
To compare the agreement between clinical diagnoses and neuropathological findings, I
calculated Cohen’s kappa values for the overall sample as well as separately for each ethnic
group. Neuropathological data were available for a total of 744 participants. The kappa value for
the combined sample was .43 (SE = .06), representing moderate agreement (Landis & Koch,
1977). For the 724 NHW participants for whom neuropathological data were available, the kappa
value was also .43 (SE = .06). I was unable to calculate a kappa value for the 20 Hispanic
participants for whom neuropathological data were available, as all 20 participants fell into the
same diagnostic category (i.e., dementia) based on autopsy findings. Therefore, I was unable to
formally test for statistically significant crosscultural differences in diagnostic accuracy based on
the kappa values. It is worth noting that a total of 19 of these 20 Hispanic participants also
received a clinical diagnosis of dementia at their initial evaluations, suggestive of a high level of
agreement with neuropathological findings.

CROSSCULTURAL DIFFERENCES IN DEMENTIA 64
Descriptive Statistics for the Subsample of Informants
The subsample of Hispanic and NHW informants from whom I collected data consisted
of 22 Hispanics and 120 NHWs. Combined ethnic-group response rates from each of the four
ADRC sites were as follows: USC, 48 (50.53%); UCLA, 39 (34.82%); UCSD, 8 (40.00%); and
UCI, 47 (57.32%). For the Hispanic informants, response rates were: USC, 7 (41.18%); UCLA,
3 (30.00%); UCSD, 8 (40.00%); and UCI, 4 (44.44%). The response rates for the NHW
informants were: USC, 41 (52.56%); UCLA, 36 (35.29%); and UCI, 43 (58.90%). I did not
receive permission to collect data from NHW informants from UCSD.
With regard to this subsample’s initial diagnoses of the patients associated with the
informants, a total of 60 (42.3%) patients were diagnosed with dementia and 82 were diagnosed
with NCF. Of this total subsample of informants, 97 (68.3%) were female, and their ages ranged
from 30 to 89 years (M = 64.34, SD = 13.00), whereas their education levels ranged from 6 to 28
years (M = 15.92, SD = 2.71). Of the 22 Hispanic informants in this subsample, 17 (77.3%) were
born in the US, and 15 (68.2%) reported that English was their first language. In addition, 13
(59.1%) were Mexican/Chicano/Mexican American, 1 (4.5%) was South American, 1 (4.5%)
was Central American, and 2 (9.1%) were classified as other. The origins of the Hispanic
patients in this subsample were similar to those of the informants in terms of their frequencies.
Of the Hispanic patients in this subsample, 12 (54.5%) were diagnosed with dementia and 10
with NCF. Among the NHWs, 48 (40.0%) were diagnosed with dementia and 72 with NCF.
Table 19 provides additional descriptive information on the demographic and other key
variables in this subsample separated by both ethnicity and diagnosis (combined NCF and
dementia, NCF only, and dementia only). There were several significant across-group
differences among the combined NCF and dementia sample that are worth noting. For example,
CROSSCULTURAL DIFFERENCES IN DEMENTIA 65
Hispanic informants were significantly less likely than NHW informants to have reported that
English was their first language, χ
2
(1, N = 139) = 20.50, p < .001, and that they were born in the
US, χ
2
(1, N = 139) = 4.62, p = .032. These differences remained significant (p < .01) regardless
of diagnosis with the sole exception of the country of birth variable, which was not significant in
the dementia group. Informant education level was significantly lower in Hispanics than NHWs
among the dementia only group, t(56) = 2.25, p = .005. The ranges of education levels for the
informants were 12 to 20 years for Hispanics and 6 to 28 years for NHWs. Regarding
informants’ relationships to patients, among Hispanics, 40.9% of informants were spouses,
27.3% were adult children, and the remainder fell into another category (e.g., other relative,
friend/neighbor, and paid caregiver/provider), whereas among NHWs, these respective figures
were 65.8% and 12.5%, χ
2
(7, N = 142) = 22.83, p = .002. This difference remained significant (p
< .05) in the dementia group but not the NCF group.
With regard to the patient characteristics of the combined NCF and dementia group
subsample, Hispanics had significantly lower education levels than their NHW counterparts,
t(140) = 3.18, p = .002, and were significantly less likely to have reported that English was their
primary language, χ
2
(1, N = 142) = 11.13, p = .001. These differences remained significant (p <
.01) among the dementia group but not the NCF group. The ranges of education levels for the
patients were 3 to 20 years for Hispanics and 5 to 21 years for NHWs. Regarding the variables
tapping patients’ dementia symptoms, overall NP was significantly worse among Hispanics than
NHWs when using CEGN in the NCF group only, t(71) = 2.23, p = .029. In addition, NPI-Q
severity was significantly higher among Hispanics in both the combined NCF and dementia
group, t(24.50) = -2.15, p = .042, and the dementia-only group, t(58) = -2.41, p = .019.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 66
There were a few between-diagnosis differences that are worth noting. For example,
coresidency percentage was significantly higher among NHW informants of patients diagnosed
with dementia than those with NCF, χ
2
(1, N = 120) = 4.20, p = .040. In addition, there were
significantly more female patients in the NHW NCF group than the NHW dementia group, χ
2
(1,
N = 120) = 4.68, p = .031. NPI-Q frequency was only significantly higher in the dementia group
compared to the NCF group among NHWs but not Hispanics (p = .078).
Table 20 provides the correlation coefficients among the demographic and key variables
for the entire subsample. There were statistically significant correlations between several of the
variables. Two of the covariates (patients’ sex and education), the ADKS (but not CBADS), and
all key independent variables (overall NP, FAQ, NPI-Q frequency and severity, and IRMD) were
significantly correlated with diagnosis. However, the correlations of the covariates with
diagnosis were comparatively much lower than those of the key independent variables with
diagnosis, ranging from r = ± .00 to ± .21 for the covariates to r = ± .37 to ± .85 for the
independent variables. Tables 21 and 22 show the correlation coefficients separated by ethnicity
(Hispanic and NHW, respectively). None of the covariates was significantly correlated with
diagnosis for either ethnic group except for patients’ sex among NHWs, r = -.20, p = .031. In
contrast, all key dementia symptom independent variables (except NPI-Q frequency among
Hispanics) were significantly correlated with diagnosis. With regard to the culturally-influenced
variables, among Hispanics, informants’ ARSMA-II raw acculturation scores were significantly
correlated with diagnosis among patients (r = .48, p = .031), but neither the ADKS nor CBADS
was. Among NHWs, informants’ ADKS scores were significantly correlated with diagnosis
among patients (r = .20, p = .027), but CBADS scores were not.

CROSSCULTURAL DIFFERENCES IN DEMENTIA 67
Hypothesis 5: Crosscultural Differences in the Alzheimer’s Disease Knowledge Scale Scores
Hypothesis 5 predicted that Hispanic informants would have significantly lower mean
scores on the ADKS than NHW informants among this subsample of participants. The mean
ADKS total score for Hispanics was 23.09 (SD = 3.85), and the mean score for NHWs was 24.08
(SD = 3.35). The results of my independent samples t test did not confirm this hypothesis, t(140)
= 1.24, p = .218, suggesting that levels of accurate AD knowledge did not differ significantly
across ethnic groups. Given the small Hispanic informant sample size (N = 22) in my subsample,
I conducted a post hoc power analysis to determine whether my analysis was underpowered.
When using an alpha level of .05, my observed effect size of 0.27, and a two-tailed test, results
revealed that my achieved power was low (0.22). Therefore, I lacked sufficient statistical power
to determine whether there were in fact meaningful ethnic-group differences in accurate AD
knowledge levels among this subsample of informants.
Hypothesis 6: Crosscultural Differences in Cultural Beliefs about Alzheimer’s Disease
Scores
Regarding Hypothesis 6.i, the CBADS demonstrated borderline adequate internal
consistency among the entire subsample of informants (α = .62) as well as the Hispanic (α = .65)
and NHW (α = .62) subsamples separately. Hypothesis 6.ii posited that Hispanic informants
would have significantly higher mean scores on the CBADS than their NHW counterparts. The
mean CBADS total score for Hispanics was 2.14 (SD = 2.08), and the mean for NHWs was 1.74
(SD = 1.86). The results of my independent samples t test did not confirm this hypothesis, t(138)
= -0.91, p = .366, suggesting that levels of culturally-influenced beliefs about AD did not differ
significantly across ethnic groups in this subsample of informants. I conducted a post hoc power
analysis to determine whether this analysis was also underpowered. When using an alpha level of
CROSSCULTURAL DIFFERENCES IN DEMENTIA 68
.05, my observed effect size of 0.20, and a two-tailed test, results revealed that my achieved
power was again low (0.14). Therefore, I lacked sufficient statistical power to determine whether
there were in fact meaningful ethnic-group differences in culturally-influenced beliefs about AD
among this subsample of informants. It should be noted that there may have been a floor effect
for both groups given the low mean values (out of a possible total of 20 points) and small range
of scores overall (between 0 and 6 points for Hispanics and 0 and 7 points for NHWs).
Hypothesis 7: Testing Path Models for Crosscultural Differences in Dementia Diagnosis
among the Subsample
Given my relatively small NHW and small Hispanic subsample size, I lacked sufficient
statistical power to test my proposed path models. Although there are no strict or established
guidelines for estimating sample sizes needed to obtain adequate statistical power in path model
analyses, some rules of thumb indicate that 10 observations per parameter produce sufficient
power (Schreiber, Nora, Stage, Barlow, & King, 2006). My proposed path model for the
Hispanic subsample had a total of 54 distinct parameters to be estimated, which would call for
sample size of 540. For NHWs, there were 47 distinct parameters to be estimated, which would
indicate a sample size of 470. Other less stringent guidelines suggest that 200 observations are
always necessary for structural equation modeling (Garver & Mentzer, 1999; Hoelter, 1983).
As an alternative, I attempted to run multivariate binary logistic regression analyses for
the NHW and Hispanic subsample participants separately. However, my post hoc power analyses
revealed that I also lacked sufficient power for these analyses. Specifically, among the Hispanic
subsample, when using an alpha level of .05, the observed effect size (i.e., odds ratio) of 0.08 for
the overall NP (using EGSN) variable, and a two-tailed test, results revealed that my achieved
power was very low (0.02). For the NHWs, when using an alpha level of .05, the observed effect
CROSSCULTURAL DIFFERENCES IN DEMENTIA 69
size (i.e., odds ratio) of 0.42 for the overall NP (using EGSN) variable, and a two-tailed test,
results revealed that my achieved power was insufficient (0.53). Indeed, Taylor, West, and Aiken
(2006) reported that in order to achieve a power level of .80, a binary logistic regression model
would generally require a sample size between 317 to 608 participants (depending on the shape
and distribution of the dependent variable).
Because logistic regression analyses generally require larger sample sizes than linear
regression analyses (e.g., Taylor et al., 2006), I decided to test for variables that were
significantly related to both total ADKS and CBADS scores among Hispanics using multivariate
linear regression, given the continuous nature of these two outcome variables. As can be seen in
my proposed path model for the Hispanic subsample in Figure 3, I had hypothesized that
informant acculturation and education levels would be significantly associated with these two
culturally-influenced variables. In addition, the correlation matrix for Hispanics in Table 21
shows that informant education, ARSMA-II raw acculturation scores, and literacy levels were
significantly positively correlated with total ADKS scores, whereas only ARSMA-II raw
acculturation scores were significantly negatively correlated with total CBADS scores.
I evaluated the assumptions of linear regression and found that they were adequately met
for all analyses in both ethnic groups. In addition, I did not find evidence of problematic
multicollinearity in any of my models. Results of my analysis of total ADKS scores regressed on
informant education, acculturation (ARSMA-II raw acculturation score), and literacy (Cloze total
score) levels revealed that these three independent variables accounted for over 60% of the
variance in total ADKS scores among Hispanics (R
2
= .63), which was highly significant,
F(3,15) = 8.35, p = .002. Both informant education (β = 0.53, p = .035) and literacy levels (β =
1.18, p = .037) demonstrated significant effects on total ADKS scores, but not acculturation
CROSSCULTURAL DIFFERENCES IN DEMENTIA 70
levels. When I reran the models using the ARSMA-II Hispanic and Anglo Orientation subscales
in lieu of raw acculturation, neither variable was significantly related to the ADKS. In sum,
higher levels of informant education and literacy (but not acculturation) were significantly
associated with higher total ADKS scores among the Hispanic subsample of informants.
Results from my multivariate linear regression analysis of total CBADS scores regressed
on informant education, acculturation (ARSMA-II raw acculturation score), and literacy (Cloze
total score) levels revealed that these three independent variables accounted for nearly half of the
variance in total CBADS scores among Hispanics (R
2
= .48), which was significant, F(3,15) =
4.60, p = .018. Only informant acculturation level (β = -0.06, p = .015) demonstrated a
significant effect on total CBADS scores. When I reran the model using the ARSMA-II Hispanic
Orientation subscale in lieu of raw acculturation, acculturation had a significant and positive
association with the CBADS (β = 0.08, p = .007). My analysis using the ARSMA-II Anglo
Orientation subscale resulted in a nonsignificant F value, p > .05. In sum, higher informant raw
acculturation levels were significantly associated with lower total CBADS scores among the
Hispanic subsample of informants, whereas the opposite pattern emerged with the ARSMA-II
Hispanic Orientation subscale.
Because I did not collect data on acculturation or literacy among the NHW subsample of
informants, I decided to evaluate whether total ADKS and CBADS scores were related to NHW
patient impairment levels (i.e., overall NP, FAQ, NPI-Q, and IRMD), given that a lack of
accurate AD knowledge and higher levels of culturally-influenced beliefs about AD among
informants were hypothesized as being associated with dementia care-seeking delays and
diagnosis. As can be seen in my proposed path model for the NHW subsample in Figure 4, I had
hypothesized that the ADKS and CBADS would have effects on the informant-report variables
CROSSCULTURAL DIFFERENCES IN DEMENTIA 71
(FAQ, NPI-Q, and IRMD). In addition, the correlation matrix for NHWs in Table 22 reveals that
the ADKS was significantly negatively correlated with overall NP (using EGSN) and positively
with the FAQ, whereas the CBADS was not significantly correlated with other variables. I
conducted two separate multivariate regression analyses based on my original hypotheses and
these correlational findings, which are summarized below.
Results of my multivariate linear regression analysis of overall NP (using EGSN)
regressed on patients’ sex and education as well as informants’ total ADKS scores revealed that
these three independent variables accounted for approximately 12% of the variance in overall NP
scores among the NHW subsample of patients (R
2
= .12), which was highly significant, F(3,106)
= 4.81, p = .004. Both patients’ sex (β = 9.41, p = .030) and education level (β = 2.12, p = .014)
as well as informants’ total ADKS scores (β = -1.52, p = .021) demonstrated significant effects
on overall NP. In sum, female sex and higher education levels among the NHW subsample of
patients were related to better overall NP, whereas higher total ADKS scores among their
informants were associated with worse overall NP among patients.
Results of my remaining multivariate linear regression analysis resulted in a
nonsignificant F value, p > .05. This analysis, which was based both on my originally proposed
hypothesized path models and correlational findings, was total FAQ scores regressed on patients’
age and informants’ total ADKS scores.
It should be noted that I was unable to conduct these analyses with my Hispanic
subsample due to the small sample size and resulting lack of statistical power. For example, my a
priori power analysis using an effect size of .10, an alpha level of .05, a power level of .80, and
three independent variables revealed that I required a sample size of 81. Using these same values
except for an effect size of .30, I would require 27 participants.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 72
Regarding the correlations of the culturally-influenced variables with diagnosis among
the Hispanic subsample, Table 21 shows that acculturation levels, as measured by the ARSMA-
II raw acculturation score, were significantly correlated with diagnosis, r = .48, p = .031. The
ARMSMA-II Anglo Orientation subscale was also significantly correlated with diagnosis, r =
.47, p = .035, but not the Hispanic Orientation subscale, r = -.42, p = .064. The other culturally-
influenced variables (e.g., ADKS, CBADS, literacy, and country of birth) were not significantly
correlated with diagnosis. Among the NHW subsample, Table 22 reveals that the ADKS was
significantly positively correlated with diagnosis, r = .20, p = .027, but not the CBADS. In sum,
higher levels of acculturation to US (i.e., Anglo American) culture among informants were
significantly correlated with a dementia diagnosis as compared to NCF among the Hispanic
subsample of patients, whereas higher levels of accurate AD knowledge among informants were
significantly correlated with a dementia diagnosis among the NHW subsample of patients.
Discussion
The current study examined crosscultural differences in dementia assessment and
diagnosis, care-seeking, and impairment levels among Hispanic and NHW outpatients, as prior
research has pointed to a number of culturally-influenced variables that may affect obtaining
timely dementia evaluations and informant reports of patients’ symptoms. This study expanded
on previous efforts by not simply reporting on crosscultural differences but also investigating the
cultural mechanisms underlying these processes. To do so, I collected data on various indicators
of acculturation and AD knowledge. I also created and administered the CBADS to a subsample
of participants, which is the first scale to my knowledge to directly measure culturally-influenced
beliefs about AD found to be common among Hispanics. In addition, this study is the first to
examine the psychometric properties of two commonly used scales that assess patients’
CROSSCULTURAL DIFFERENCES IN DEMENTIA 73
functional abilities and BPS to determine whether they are operating equivalently across ethnic
groups, bearing relevance in terms of crosscultural dementia diagnostic validity.
I begin by discussing the findings and their implications from the path models for each
ethnic group using the national dataset, followed by those of the CFAs that assessed for the
crosscultural invariance of both the FAQ and NPI-Q. I then discuss the results regarding
differences in impairment levels (NP, length of cognitive decline, IRMD, FAQ, NPI-Q frequency
and severity, and CDR) across Hispanics and NHW patients diagnosed with dementia at their
initial ADRC evaluations. I then focus on the variables that were significantly associated with
the ADKS and CBADS as well as the association of acculturation levels with diagnosis among
the Hispanic subsample of informants. Next, I discuss the effects of total ADKS and CBADS
scores on patient impairment levels and diagnosis among the NHW subsample. Lastly, I note the
limitations and strengths of this study as well as its contributions to the literature.
Path Models Illustrating Crosscultural Differences in Dementia Diagnosis
When using either EGSN or CEGN, I found that all four key independent variables with
the exception of informant-reported BPS (which was only significant among NHWs) were
significantly associated with diagnosis in both ethnic groups. Specifically, IRMD, poorer overall
NP, and higher FAQ scores were associated with a dementia diagnosis in both groups, and
higher levels of BPS frequency were associated with a dementia diagnosis in NHWs only,
though the effect size for this latter finding was rather small and likely not meaningful. However,
I only found partial support for between-group differences regarding the rank orders of the
strengths of associations of NP versus informant reports with diagnosis. Specifically, though the
rank order of the standardized path coefficient for NP to diagnosis was higher among NHWs
than Hispanics (second versus third, respectively), the significant and key association of NP with
CROSSCULTURAL DIFFERENCES IN DEMENTIA 74
diagnosis was present among both groups. This finding suggests that NP may play a stronger role
in the diagnostic process in the context of the other variables of interest in this study among
NHWs as compared to Hispanics, whereas the FAQ may play a stronger role among Hispanics.
Of note, this difference may be rather small given that the sizes of the path coefficients and their
rank orders differed minimally. In sum, it appears as though clinicians relied on NP during the
diagnostic process to a significant degree among Hispanics (as well as NHWs) despite the well-
known fact that neuropsychological tests with unrepresentative norms may not be well suited for
this group due various language, cultural, and education factors (e.g., Manly & Espino, 2004).
The finding that the association of NP with diagnosis did not differ significantly across
ethnic groups when using CEGN was particularly unanticipated in light of the limitations of NP
among Hispanics. Because these norms place Hispanics at a disadvantage (as evidenced by their
significantly lower mean NP scores in the NCF-only group compared to NHWs), it appeared
plausible to predict that NP would be much less strongly related to diagnosis among Hispanics,
given that CEGN are not taking ethnic-group differences into account. However, my results did
not support this supposition as there were no significant changes in the rank orders of the path
coefficients when using EGSN as compared to CEGN among either group. In other words, the
role of NP remained similar and significant in either scenario. Despite the limitations of NP
among Hispanics, it is possible that clinicians in this study may still have relied heavily on NP
when assigning a diagnosis because this variable represents a more objective measure of
cognitive functioning than informant-based reports. It is also possible that clinicians may have
informally adjusted their overall impression of Hispanic patients’ NP to account for the limited
validity of many of the tests (e.g., tests within the domain of language) among this ethnic group.
Evidence for this conjecture can be seen in the consistently lower overall NP mean values for
CROSSCULTURAL DIFFERENCES IN DEMENTIA 75
Hispanics versus NHWs when using CEGN (i.e., between five and six standardized points lower)
regardless of diagnostic category. This finding points to a systematic ethnic-group difference in
terms of overall NP about which clinicians may be aware on some level. In addition, because the
Hispanic patients in this sample spoke English (i.e., fluently enough to be assessed in English
rather than Spanish) and had a relatively high mean level of education (13.25 years, SD = 3.93
years), clinicians may have been more likely to rely on NP to a greater degree than they would
have if these patients lacked sufficient English proficiency and education. Unfortunately, I could
not directly test this hypothesis as the larger national dataset did not include variables to assess
potentially important, more specific factors pertaining to English language proficiency (e.g.,
literacy and number of years speaking English) and education (e.g., location or language of
education).
I also found that informant reports, namely IRMD and total FAQ scores, were most
strongly associated with diagnosis among Hispanics, but that the NPI-Q was not, as it did not
remain in my final path models (when using either EGSN or CEGN). This latter finding may
perhaps be the most striking from my path model analyses and suggests that clinicians may have
only relied on this variable to a statistically significant (but perhaps not meaningful) degree
among NHWs when assessing for dementia. Though I had hypothesized that this association
would be significant among both groups (albeit stronger among Hispanics), I did not anticipate
finding a statistically nonsignificant association among Hispanics. This finding appears in
contrast to the results of prior studies that have found BPS to be very prevalent among dementia
patients (Aalten et al., 2003; Steinberg et al., 2004), which would suggest that BPS may be a key
diagnostic factor on which clinicians rely regardless of ethnicity. Though a number of possible
reasons exist that may account for this finding, one plausible explanation is that some BPS such
CROSSCULTURAL DIFFERENCES IN DEMENTIA 76
as depression, anxiety, and apathy may be due to causes other than dementia, and so perhaps
clinicians did not consider them as related to dementia in Hispanics as much as in NHWs for
various reasons. For instance, it is possible that NHW informants may have been better able than
their Hispanic counterparts to communicate and express changes in patients’ BPS to the English-
speaking clinicians in this study. As such, these BPS may have appeared more attributable to
dementia-related changes rather than preexisting or new psychological conditions, for example.
BPS are also susceptible to more subjective interpretation than functional abilities, which may
have also in part accounted for this between-group difference. It is also plausible that clinicians
may have attributed Hispanic informant reports of patients’ BPS to culture rather than pathology,
given that it is widely believed that culture influences the presentation of psychopathology (e.g.,
Weisman et al., 2000). For instance, Simons and Hughes (1993) noted that Hispanics tend to be
more emotive and outwardly expressive of emotions in contrast to NHWs whose gestures and
facial expressions may be comparatively more constricted. As a result, clinicians may discount
certain BPS as pathological and instead view them as culturally-influenced patterns of emotional
expression. Finally, it may simply be that clinicians rely most heavily on other variables such as
IRMD, informant-reported functional abilities, and NP, all of which tap the key features of a
dementia diagnosis (i.e., decline in cognition and daily functioning; American Psychiatric
Association, 2000) in their evaluations. Therefore, the NPI-Q may become a less important
additional piece of information in diagnosing dementia in Hispanics, and perhaps NHWs as well.
The results of this study also revealed that the FAQ was significantly related to diagnosis
in both ethnic groups. This finding is consistent with those of other studies that have found NP to
be strongly associated with the ability to carry out IADLs, particularly in the earlier stages of
dementia (e.g., Monaci & Morris, 2012). However, the role of the FAQ did not appear to differ
CROSSCULTURAL DIFFERENCES IN DEMENTIA 77
to a great extent in terms of the sizes of its standardized path coefficients across groups, in
contrast with my hypothesis. This finding suggests that clinicians overall weighed informant
reports of patients’ functional abilities relatively equally across ethnic groups during evaluations.
It is possible that the ability to carry out IADLs may not be as sensitive to cultural influences on
informants’ perceptions of changes in patients’ abilities. In other words, patients’ ability to
manage finances or remember appointments, for example, may not be as sensitive to cultural
interpretations as BPS like delusions and hallucinations. In addition, informants’ reporting styles
may not be as influenced by cultural or linguistic differences given the relatively objective nature
of IADLs compared to BPS, which may be more difficult to effectively communicate across
cultures. Similarly, responses to the FAQ may be affected by noncognitive factors such as
physical disability, which may also be less sensitive to culture-specific interpretations.
When using the CDR as the outcome variable and EGSN, there were some key changes
in terms of the associations of the independent variables with the CDR. Among Hispanics, the
FAQ became most strongly associated with the CDR (β = .54), followed by IRMD (β = .25) and
NP (β = -.21). In addition, education was no longer directly associated with the dependent
variable (CDR). Among NHWs, the FAQ also had the strongest association with the CDR (β =
.53), followed by NP (β = -.27), IRMD (β = .19), and the NPI-Q (β = .03). Additionally,
education became directly associated with the CDR. The fact that the FAQ became most strongly
associated with the CDR among both ethnic groups is logical, given that four of the six
dementia-related domains that the CDR taps either directly or indirectly pertain to patients’
functional abilities (personal care, home and hobbies, community affairs, and judgment and
problem solving). The pattern of key findings remained similar when using CEGN.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 78
Implications. The finding that overall NP was associated with diagnosis to a relatively
equal extent across Hispanics and NHWs suggests that clinicians may be relying on this variable
to a similar degree in spite of the known limitations of neuropsychological tests among
Hispanics. Future research should directly assess clinicians’ perceptions of the validity and
utility of NP among Hispanics and whether they are indeed informally adjusting their overall
impressions of Hispanics’ NP and/or weighing certain tests differentially across groups to
account for the limitations of these tests. If this supposition were confirmed, clinicians should be
encouraged to use the most appropriate neuropsychological tests and norms for each individual
patient. Moreover, the field should strive to create more normative data for diverse groups of
individuals to improve diagnostic accuracy in both clinical and research contexts.
In addition, it appears as though IRMD followed by informant-reported functional
abilities were most strongly associated with diagnosis among Hispanics. In contrast, IRMD
followed by overall NP were most strongly related to diagnosis among NHWs. These findings
call attention to potential systematic differences in the diagnostic process across ethnic groups,
placing diagnostic validity into question. Inaccurate or biased diagnosis has clear implications in
both clinical and research settings. The field would benefit from a larger amount of
neuropathological data associated with clinical diagnoses to help determine the accuracy of
diagnosis across cultures and whether such differences in the variables associated with diagnosis
are driving group differences in the dementia assessment and diagnostic process.
The association of total NPI-Q frequency scores and diagnosis among NHWs was
statistically significant but likely not meaningful given the effect size of less than .10. Future
research is needed to determine whether this variable plays a meaningful role in the diagnosis of
dementia among this group. In addition, given that the NPI-Q was not statistically significantly
CROSSCULTURAL DIFFERENCES IN DEMENTIA 79
associated with diagnosis among Hispanics, it is important to note that the assessment of both
cognition and other dementia symptoms (e.g., BPS) among Hispanics often lacks consideration
of other important culturally-influenced factors. These factors include acculturation, quality of
education (i.e., rather than simply level of education), stereotype threat, literacy, and language
barriers between clinicians and patients and their families (Manly & Espino, 2004) that can
affect the reporting of symptoms and the weight clinicians give to them during the diagnostic
process. Aside from the obvious linguistic differences that may impede or impair the obtainment
of a reliable and valid assessment, the scarcity of culturally and linguistically valid measures and
related normative data may present other challenges during assessment of which clinicians
should be mindful. Future research should aim to develop culturally- and linguistically-
appropriate, valid, and reliable measures of dementia symptoms such as BPS among diverse
groups. In addition, the use of bilingual clinicians may serve to facilitate communication with
patients and thus improve diagnostic accuracy. Lastly, clinicians should obtain training in
cultural sensitivity, culturally-influenced beliefs about dementia, and the limitations of
evaluating dementia among Hispanics in order to improve the accuracy of dementia assessment
and diagnosis.
Measurement Properties of the Functional Assessment Questionnaire across Ethnic
Groups
As hypothesized, results from my multigroup CFAs revealed that the FAQ demonstrated
crosscultural factorial (i.e., configural and measurement) invariance, suggesting that this scale
has the same number (i.e., one) of factors and a similar pattern of factor loadings across
Hispanics and NHWs. The one-factor (dementia severity) structure I found for the FAQ is
consistent with the findings of a prior study that used a similar measure of functional abilities
CROSSCULTURAL DIFFERENCES IN DEMENTIA 80
and also found a comparable one-factor structure of dementia severity that was invariant across
participants from three European countries (Erzigkeit et al., 2001). Given that I found evidence
for the crosscultural factorial invariance of the FAQ, I can conclude that this scale was operating
similarly across both ethnic groups and was measuring the same latent construct of dementia
severity with regard to IADLs.
My analyses also revealed that the FAQ demonstrated crosscultural scalar invariance,
suggesting that item mean scores were similar across ethnic groups. Therefore, meaningful
comparisons of levels of the latent dementia severity factor could indeed be made across groups.
When I compared Hispanics and NHWs in terms of levels of this latent factor, I did not find a
significant across-group difference, as hypothesized. This finding suggests that there were no
systematic group differences in item mean values, or, in other words, that the intercept values for
each item of the dementia severity factor were invariant across Hispanics and NHWs.
Implications. The findings of this study suggest that the FAQ can continue to be used as
a useful measure of patients’ functional abilities among both Hispanics and NHWs. Indeed, it
proved to be crossculturally invariant in terms of both its factor structure and item intercept mean
values. The finding of invariance helps to confirm that comparisons made between Hispanics and
NHWs using the FAQ are valid, thereby providing evidence for measurement validity in clinical
and research settings. In addition, the hypothesized lack of significant difference in terms of
mean levels of the latent dementia severity factor across ethnic groups using the combined NCF
and dementia sample is consistent with the notion that functional abilities would not be expected
to vary systematically by ethnicity alone. Future research should examine whether the FAQ is
invariant across other ethnic groups.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 81
Measurement Properties of the Neuropsychiatric Inventory Questionnaire across Ethnic
Groups
I also found evidence to confirm my hypothesis regarding the crosscultural factorial
invariance of the NPI-Q. These findings suggest that this scale has the same number (i.e., four)
of factors and a similar pattern of both factor loadings and covariances across Hispanics and
NHWs. The hypothesized four-factor (hyperactivity, affect, psychosis, and apathy/vegetative
symptoms) structure of the NPI-Q that I confirmed for both ethnic groups is consistent with the
findings of prior studies using similar versions of this BPS measure that also found a similar
crossculturally invariant four-factor structure in 12 European countries (Aalten et al., 2007) and
South Korea (Kang et al., 2010). Given that I found evidence for the crosscultural factorial
invariance of the NPI-Q, I can conclude that this scale was operating similarly across both ethnic
groups and was measuring the same four latent constructs of dementia-related BPS.
Contrary to my hypothesis, the NPI-Q did not demonstrate crosscultural scalar
invariance, which means that differences in the means of the four latent factors associated with
this scale cannot be meaningfully interpreted and compared across Hispanics and NHWs. Item
intercept invariance is a requirement for the comparison of latent means, as it implies identical
intervals and zero points of the scale across groups. When scalar invariance is not tenable, the
comparison of latent means becomes equivocal as the between-group differences in latent means
are confounded with the scale and origin of the latent variable (Cheung & Rensvold, 2002). The
CFI difference tests for the structured means model (used to test for scalar invariance) revealed
that the fit of this model was significantly worse than both the configural and measurement
models, corroborating the nontenability of my scalar invariance hypothesis. Therefore, I was
CROSSCULTURAL DIFFERENCES IN DEMENTIA 82
unable to meaningfully test for significant ethnic-group differences in mean values of the four
NPI-Q factors.
Scalar noninvariance is believed to result from differential additive response bias and
suggests that issues unrelated to the pertinent constructs are influencing the presence of
systematically different scores on some or all items in one ethnic group compared to the other for
any number of reasons. For example, as discussed previously, language barriers and cultural
differences in how BPS are interpreted and communicated may have influenced ethnic-group
differences on responses to the NPI-Q items. In addition, the shame and stigma associated with
dementia and its BPS may be particularly salient among Hispanics (e.g., Sayegh & Knight,
2012), which could have systematically influenced response patterns among informants in this
ethnic group. Similarly, informant-reporting styles on patients’ BPS may have differed across
groups given that Hispanic caregivers may be more sensitive to dementia-related BPS than NHW
caregivers (Ortiz et al., 2006; Valle, 1994). Because my post hoc analyses failed to reveal any
particular item(s) contributing to the scalar noninvariance, the most plausible explanation is that
these differences were likely spread out across some or all items but may not have been
significant on an item-level basis. When comparing NPI-Q item-level mean differences across
groups, Hispanics had significantly higher scores than NHWs on both delusions, t(467.73) = -
2.91, p = .004, and depression, t(475.81) = -2.32, p = .021. However, the effect sizes associated
with these differences were rather small (d = .16 and .13, respectively) and fell below my .20
cutoff, suggesting that they may not be meaningful differences.
Implications. My multigroup CFA revealed that the NPI-Q can continue to be used as a
useful measure of patients’ functional abilities among both Hispanics and NHWs, as it
demonstrated crosscultural factorial invariance. The finding of factorial invariance helps to
CROSSCULTURAL DIFFERENCES IN DEMENTIA 83
confirm that this scale is operating similarly across these groups, supporting the measurement
validity of the NPI-Q in clinical and research contexts. In addition, my results suggested that this
scale can be readily and validly used among both groups to measure specific dimensions of BPS
beyond a more general operationalization of these symptoms.
However, the NPI-Q cannot be used to make meaningful comparisons across ethnic
groups in terms of the mean values of the four latent factors derived from this scale. Researchers
should be mindful of this limitation when considering comparing Hispanics and NHWs on latent
mean values of these factors. Aside from this limitation, both the overall usefulness of this scale
and its validity and specificity suggest that future research should examine whether the NPI-Q
demonstrates factorial and/or scalar invariance across other ethnic groups.
Crosscultural Differences in Time to and Impairment Level at Initial Evaluation
First, I did not find evidence for crosscultural differences in terms of length of cognitive
decline among patients diagnosed with dementia. This finding is inconsistent with my hypothesis
that Hispanics would have presented for their initial evaluations later than NHWs and, thus, have
had a longer time since the onset of their cognitive decline. Though prior research has suggested
that Hispanics face various culturally-influenced and systemic barriers to dementia care-seeking
that may delay timely diagnosis (see Sayegh & Knight, 2012, for a review), I did not find
evidence in support of this conclusion. This finding may in part be due to the difficulty
associated with obtaining a precise estimate regarding actual length of decline. Therefore, this
variable may not be the most reliable measure of this construct, which is truly difficult to
accurately operationalize. Additionally, a post hoc power analysis revealed that given the sample
sizes accounting for missing data (N = 232 for Hispanics, N = 5,842 for NHWs), my observed
effect size of 0.05, an alpha value of .05, and using a two-tailed test, my achieved level of
CROSSCULTURAL DIFFERENCES IN DEMENTIA 84
statistical power was low (β = .11). Therefore, the results from this analysis involving the length
of cognitive decline variable should be interpreted with caution. However, it is also plausible that
there was in fact no significant or meaningful group difference in the estimated length of
cognitive decline in this study because of the nature of the Hispanic sample in this study.
Specifically, Hispanic patients who present to and enroll in comprehensive longitudinal research
studies such as the NACC ADRC study may not be representative of the greater Hispanic
American patient population.
Second, when using CEGN, Hispanics diagnosed with dementia at their initial
evaluations performed significantly worse than their NHW counterparts in terms of overall NP,
as I had hypothesized. Given the limitations of neuropsychological tests among Hispanics, it
came as little surprise that Hispanics with dementia performed worse than NHWs when overall
NP was normed with regard to the entire NCF sample. However, when using EGSN, NHWs
performed significantly worse than Hispanics (despite no significant crosscultural difference
among the NCF-only groups), which was unanticipated. These starkly different findings again
highlight the importance of using appropriate normative data and standardization techniques
when evaluating NP across cultural groups. This result also suggests that Hispanics with
dementia may not in fact be presenting with more impairment at their initial evaluations as
assessed by overall NP alone. It is possible, as previously noted, that the Hispanics with
dementia who were ultimately evaluated and enrolled as part of this study may have differed
from Hispanics who did not present for evaluations or presented to non-research-based clinics.
Specifically, the Hispanics in this study may have had more cognitive reserve and thus
performed better on NP than both their Hispanic counterparts in non-research settings and the
NHWs with dementia in this study. Indeed, it has long been known that volunteers in
CROSSCULTURAL DIFFERENCES IN DEMENTIA 85
longitudinal research projects tend to be more educated than average (e.g., Streib, 1966), and this
finding may be especially pertinent to the Hispanic participants in this study. Another possible
reason for this finding is that there may have been a disproportionately greater amount of NHW
compared to Hispanic patients in the national sample who presented with more sudden cognitive
impairment due to non-AD disorders, such as traumatic brain injury or prion disease, that may
initially only result in impaired cognitive functioning and minimal changes in functional abilities
and BPS. However, given the relatively low prevalence of these types of disorders in this study, I
was unable to statistically confirm this supposition.
Third, IRMD as well as total FAQ, NPI-Q severity (but not frequency), and global CDR
scores were all present to a statistically significantly greater degree among Hispanic rather than
NHW outpatients diagnosed with dementia at their initial ADRC evaluations. However, the
effect sizes for these differences were small (i.e., below .10), suggesting that these differences
may not be meaningful. As noted earlier, the convenience sample of Hispanics who presented to
these research-oriented clinics and enrolled in this comprehensive longitudinal study may not be
representative of the larger Hispanic population in the US. As a result, these participants may
have presented for their dementia evaluations with less impairment than they otherwise would
have if they had faced more culturally-influenced barriers to dementia care.
Implications. Though I found evidence for statistically significant crosscultural
differences in terms of IRMD and FAQ, NPI-Q severity (but not frequency), and global CDR
scores, the meaningfulness of these findings is questionable given the small effect sizes. Future
studies should assess whether these findings are in fact meaningful in more representative
Hispanic samples using probability-based sampling methods, for example. Another interesting
and unanticipated finding was that only NPI-Q frequency but not severity scores differed
CROSSCULTURAL DIFFERENCES IN DEMENTIA 86
statistically significantly across ethnic groups, albeit it to a small degree. It may simply be that
ethnic-group differences in BPS among patients with dementia are present only in terms of
severity and not frequency, given that BPS have been found to be very common among all
dementia patients (Aalten et al., 2003; Steinberg et al., 2004). Chen et al. (2000) provided an
alternative explanation, which posited that ethnocultural factors and behavioral patterns molded
in earlier life may affect the presentation and manifestation of BPS associated with dementia in
later life. This explanation would suggest that cultural differences in BPS severity (but not
necessarily frequency) may in fact be shaped across the lifespan well before the onset of
dementia. Though this suggestion is intriguing and worthy of investigation, more research is
needed to elucidate the mechanisms driving crosscultural differences in BPS severity levels.
With regard to neuropsychological assessment, the conflicting findings regarding greater
impairment levels among Hispanics with dementia when using CEGN and the opposite pattern
when using EGSN again highlight the importance of using the most appropriate test norms
available given the unique sociodemographic characteristics of each patient. Doing so allows
clinicians to obtain the most accurate picture of their patients’ cognitive functioning possible.
Unfortunately, the answer to this problem is not as simple as creating separate test norms for
different cultural minority groups, languages, and educational backgrounds. Certainly, separate
norms may be able to assist with misdiagnosis among culturally diverse individuals to a certain
degree (Miller, Heaton, Kirson, & Grant, 1997). However, the wide range of cultural and
educational experiences within each ethnic group may decrease these norms’ accuracy, as they
would not account for the factors associated with race, culture, education, and so forth that lie
behind differences on NP across ethnic groups. Manly and Espino (2004) suggested that future
research directly investigate the effects of more meaningful and explanatory variables other than
CROSSCULTURAL DIFFERENCES IN DEMENTIA 87
ethnicity (e.g., SES, acculturation, and literacy) that are known to affect NP across cultural
groups, as it may ultimately result in more valid instruments for dementia diagnosis and, thus,
more accurate neuropsychological assessment.
Variables Associated with Knowledge and Culturally-influenced Beliefs about Alzheimer’s
Disease in the Hispanic Informant Subsample
In my subsample of Hispanic informants, I found that higher informant education and
literacy levels but not acculturation (as measured by the ARSMA-II) were significantly
associated with higher ADKS scores, indicative of greater amounts of accurate AD knowledge. I
had hypothesized that all of these variables would be significantly associated with the ADKS,
but my analyses failed to provide support for ARSMA-II acculturation levels. These findings
suggest that education and literacy may play a key role in accurate AD knowledge levels. In
addition, I found that lower ARSMA-II raw acculturation and higher ARSMA-II Hispanic
Orientation subscale scores were significantly related to higher total CBADS scores among my
Hispanic informant subsample, but informant education and literacy levels were not. These
results indicate that being less acculturated to Anglo American culture as compared to one’s
Hispanic culture of origin is associated with greater amounts of culturally-influenced beliefs
about AD, as hypothesized.
My correlational findings regarding the positive associations of both ARSMA-II raw
acculturation and Anglo Orientation subscale scores with a dementia diagnosis suggest that
Hispanic informants’ being more acculturated to the US is related to obtaining a dementia
diagnosis among patients. Indeed, the Hispanic informants from the NCF group had higher
Anglo and lower Hispanic Orientation subscale and raw acculturation scores than their
counterparts in the dementia group. These findings may be due to the likelihood that Hispanic
CROSSCULTURAL DIFFERENCES IN DEMENTIA 88
informants with higher acculturation levels may have had less difficulty seeking and accessing
dementia care for patients who were exhibiting dementia symptoms. However, given the
correlational nature of these findings, future research, in particular with larger and more
representative sample sizes, is needed to determine the actual relations among these variables.
It should be noted that the Cronbach’s alpha values for the CBADS for the combined
sample and separate ethnic groups were in the .60 range, which is borderline adequate at best.
Post hoc analyses failed to reveal any items whose deletion would have improved the alpha
values to a great degree among both groups. In addition, there appeared to be a floor effect given
the low mean and small range of values for this scale. It is possible that this scale would better
apply to Hispanics residing in their countries of origin or those with lower acculturation or
formal education levels, for example, as compared to those in this research study, though further
research is needed to corroborate this supposition.
Implications. The finding that both informant education and literacy levels were
associated with the ADKS suggests that health service providers and researchers should consider
assessing AD knowledge where appropriate among Hispanics with lower education and literacy
to determine whether limited AD knowledge may be thwarting timely dementia care. Similarly,
the finding that higher levels of orientation towards one’s Hispanic culture of origin were
associated with higher CBADS scores suggests that acculturation levels assessed in this fashion
may be an appropriate target for assessment as well. Working to increase accurate AD
knowledge in a culturally-competent fashion may serve to facilitate a timelier dementia diagnosis
and earlier access to appropriate treatments and services. Future research should evaluate
whether lower levels of accurate knowledge and higher levels of culturally-influenced beliefs
about AD among Hispanic informants are in fact significantly associated with a dementia
CROSSCULTURAL DIFFERENCES IN DEMENTIA 89
diagnosis among Hispanic patients and whether interventions aimed at increasing accurate AD
knowledge can facilitate timelier dementia diagnoses.
The discrepant findings obtained when using different measures of acculturation
highlight a key implication regarding the measurement of this construct. Schwartz, Unger,
Zamboanga, and Szapocznik (2010) argued that acculturation is best conceptualized as a
multidimensional construct involving the convergence of the practices, values, and identification
of both the culture of heritage and dominant American culture. It may be that literacy plays a
more important role in obtaining accurate AD knowledge for Hispanics in the US as compared to
more deep-seated culturally-influenced beliefs about AD, which may be more sensitive to one’s
orientation to a particular culture. Future research involving acculturation should similarly assess
this variable multidimensionally to best capture the varying effects of this multifaceted construct.
The Effects of Knowledge about Alzheimer’s Disease on Patient Impairment Levels in the
Non-Hispanic White Informant Subsample
In my NHW subsample, I found that higher ADKS scores among informants were
associated with worse overall NP (using EGSN) among patients when covarying for patients’ sex
and education level. I anticipated that lower levels of accurate AD knowledge would be
associated with poorer NP, as the cognitive symptoms of dementia may be normalized or
misinterpreted among informants with less AD knowledge. However, it is plausible that the
display of cognitive decline in the context of low AD knowledge may motivate patients and their
loved ones to seek an evaluation to better understand what may be going on with the patient.
Relatedly, informants may have had more of an opportunity to learn about AD as patients’
cognitive decline progressed. It is also possible that having higher levels of accurate AD
knowledge would result in a greater likelihood of recognition of the cognitive symptoms of
CROSSCULTURAL DIFFERENCES IN DEMENTIA 90
dementia as abnormal. As a result, these patients would be more likely to ultimately be enrolled
in and evaluated as part of this research study than the patients of informants with lower AD
knowledge levels who may be less likely to present for evaluations at research-based clinics.
Implications. Among the NHWs in this subsample, it appears as though higher levels of
accurate AD knowledge in informants were associated with poorer overall NP in patients. Future
studies should directly explore how culturally diverse patients and their loved ones with different
AD knowledge levels interpret and react to cognitive changes in patients, which may in turn
affect the decision to obtain a dementia evaluation. Qualitative studies may serve as an ideal
starting point to obtain richer information about these associations that may help guide future
quantitative study development. In addition, future research using larger, probability-based
samples (or at least smaller samples using more rigorous methodologies) is needed to assess the
relations between accurate AD knowledge and dementia diagnosis and care-seeking among
diverse ethnic groups.
Limitations and Strengths
This study has certain limitations that suggest that caution should be used in the
interpretation of these results. First, this study used a cross-sectional design, which precludes the
ability to draw causal inferences. Second, the generalizability of these findings may be somewhat
limited, as ADRC participants essentially represent a convenience sample composed of patients
and informants who presented to academic AD clinics and may not be fully representative of the
general population. Third, the neuropsychological test battery used in this study is somewhat
limited in scope and does not assess certain aspects of cognitive functioning, such as nonverbal
memory. Fourth, the national sample lacked enough statistical power to examine differences
across subgroups of the Hispanic outpatients (e.g., Mexican, Cuban, Puerto Rican, Dominican,
CROSSCULTURAL DIFFERENCES IN DEMENTIA 91
and South and Central American), which could have provided findings on differences across
more specific cultural groups. I also lacked statistical power in my subsample to conduct such
subgroup analyses as well as my proposed path model analyses. Future studies should use larger,
diverse Hispanic samples to obtain the most accurate and generalizable findings. Fifth, there was
a lack of neuropathological data from postmortem autopsies to test for crosscultural differences
in diagnostic validity within these data at this time. In addition, these data did not contain
information on clinicians’ perceptions of the utility of each diagnostic variable, or how much
weight they assigned to each variable across ethnicities when assessing dementia. Future
research should directly assess clinicians’ perceptions of the validity of NP and informant-report
measures among diverse individuals to assess their influences on dementia assessment. Such
information could help corroborate, for example, the supposition that clinicians may have been
informally adjusting their overall appraisals of Hispanics’ NP or weighing certain tests
differentially across groups to account for the limited validity of many neuropsychological tests
in Hispanics.
Despite these limitations, this study also has a number of strengths and makes a number
of important contributions to the literature on crosscultural differences in dementia diagnosis and
care-seeking. First, it used a nationwide, multisite dataset characterized by standardized methods,
which bolsters both the external and internal validity of this study. Second, the dataset included
both patient and informant-report data derived from validated measures and tests, also enhancing
the validity of this study’s findings. Third, I used structural equation modeling for a number of
my key analyses, which not only allowed me to assess for the crosscultural multigroup factorial
and scalar invariance of both the FAQ and NPI-Q but also the direct and indirect relations among
the variables in my path model analyses. Structural equation modeling also provides other
CROSSCULTURAL DIFFERENCES IN DEMENTIA 92
advantages over more traditional analytic methods (e.g., regression), such as the explicit
provision of estimates of error variance parameters and, thus, more accurate parameter estimates
(Byrne, 2010). Fourth, the inclusion of a relatively large and diverse Hispanic sample as well as
a large NHW sample in my national dataset is another strong point of this study, as it afforded
me with adequate statistical power for many of my analyses. Fifth, rather than simply reporting
on ethnic-group differences in dementia care-seeking, assessment, and diagnosis, I was able to
evaluate some of the underlying variables (e.g., NP, BPS, and culturally-influenced beliefs and
accurate knowledge about AD) driving these group differences. Sixth, I conducted my path
model analyses using both EGSN and CEGN, which allowed me to tease apart differences in
findings that may have been attributable to the limitations of NP among Hispanics. This
approach also highlighted an important finding regarding how clinicians’ may have still weighed
NP invariantly across cultures. Seventh, this study is the first to establish the factorial invariance
of both the FAQ and NPI-Q across Hispanics and NHWs as well as the scalar invariance of the
FAQ. These findings confirm that these scales can be meaningfully used across ethnic groups.
Lastly, the CBADS, which I created based on my review of the literature on cultural beliefs
about AD among Hispanics and in consultation with experts in the field, can be further assessed
for its validity and reliability and ultimately used in future studies in more appropriate
populations, such as individuals residing in Hispanic countries or recent U.S. immigrants.
Conclusion
In sum, the findings of this study bear several important clinical and research
implications regarding differences in dementia diagnosis and care-seeking among Hispanic and
NHW outpatients. First, it appears as though clinicians may have been aware of the limited
validity of many neuropsychological tests and thus informally adjusted their overall impressions
CROSSCULTURAL DIFFERENCES IN DEMENTIA 93
of NP and/or weighed certain tests differentially across groups to account for such differences in
an attempt to increase diagnostic validity. Second, Hispanic informants’ reports of patients’ BPS
were found to be inconsequential in the diagnostic process, highlighting a possible systematic
difference in the diagnostic process across ethnic groups. This finding may have perhaps been
due to cultural and/or linguistic differences in the ability to communicate such symptoms and/or
the way they were conceptualized. Third, informant-reported functional abilities were
comparatively more strongly associated with a dementia diagnosis among Hispanics than overall
NP among NHWs, which may also indicate crosscultural differences in diagnosis. Fourth, I
determined some of the variables that are associated with accurate knowledge and culturally-
influenced beliefs about AD among Hispanic informants. Specifically, higher education and
literacy levels were related to higher ADKS scores, whereas stronger acculturation toward one’s
Hispanic culture of origin was associated with higher CBADS scores. Fifth, higher levels of
accurate AD knowledge among NHW informants were related to poorer overall NP among
NHW patients, suggesting that AD knowledge among informants may affect patients’ cognitive
impairment levels among this group. Finally, both the FAQ and NPI-Q were found to
demonstrate factorial invariance across Hispanics and NHWs, suggesting that these scales are
measuring the same underlying constructs across groups. However, because I only found
evidence for crosscultural scalar invariance for the FAQ, the NPI-Q may not be used to make
meaningful comparisons based on latent mean estimates of this measure’s four factors.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 94
References
Aalten, P., de Vugt, M. F., Lousberg, R., Korten, E., Jaspers, N., Senden, B., et al. (2003).
Behavioral problems in dementia: A factor analysis of the neuropsychiatric inventory.
Dementia and Geriatric Cognitive Disorders, 15, 99-105.
Aalten, P., Verhey, F. R. J., Boziki, M., Bullock, R., Byrne, E. J., Camus, V., et al. (2007).
Neuropsychiatric syndromes in dementia. Results from the European Alzheimer Disease
Consortium: Part I. Dementia and Geriatric Cognitive Disorders, 24, 457-463.
Alzheimer’s Association (2004). Hispanics/Latinos and Alzheimer’s disease. Chicago: Author.
American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders
(4
th
ed., text revision). Washington: Author.
Artiola i Fortuny, L., Heaton, R. K., & Hermosillo, D. (1998). Neuropsychological comparisons
of Spanish-speaking participants from the U.S.–Mexico border region versus Spain.
Journal of the International Neuropsychological Society, 4, 363-379.
Ayalon, L., & Areán, P. A. (2004). Knowledge of Alzheimer’s disease in four ethnic groups of
older adults. International Journal of Geriatric Psychiatry, 19, 51-57.
Balsis, S., Carpenter, B. D., & Storandt, M. (2005). Personality change precedes clinical
diagnosis of dementia of the Alzheimer type. Journal of Gerontology: Psychological
Sciences, 60B, P98-P101.
Block, C. K., & Baldwin, C. L. (2010). Cloze probability and completion norms for 498
sentences: Behavioral and neural validation using event-related potentials. Behavior
Research Methods, 42, 665-670.
Bollen, K. A., & Long, J. S. (Eds.) (1993). Testing structural equation models. Newbury Park,
CA: Sage.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 95
Boyle, L. L., Ismail, M. S., & Porsteinsson, A. P. (2006). The dementia workup. In M. E.
Agronin & G. J. Maletta (Eds.), Principles and practice of geriatric psychiatry (pp. 137-
153). Philadelphia: Lippincott Williams & Wilkins.
Browne, M. W. (1984). Asymptotically distribution-free methods for the analysis of covariance
structures. British Journal of Mathematical and Statistical Psychology, 37, 62-83.
Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit in the analysis of
covariance structure. Psychological Bulletin, 88, 588-606.
Bryant, F. B., & Yarnold, P. R (2000). Principal-components analysis and exploratory and
confirmatory factor analysis. In L. G. Grimm & P. R. Yarnold (Eds.), Reading and
understanding multivariate statistics (pp.99-136). Washington: American Psychological
Association.
Buchwald, A., & Rapp, B. (2010). Distinctions between orthographic long-term memory and
working memory. Cognitive Neuropsychology, 26, 1-28.
Byrne, B. M. (2010). Structural equation modeling with AMOS: Basic concepts, applications,
and programming (2
nd
Ed.). New York: Taylor & Francis Group.
Carpenter, B. D., Balsis, S., Otilingam, P. G., Hanson, P. K., & Gatz, M. (2009). The
Alzheimer’s Disease Knowledge Scale: Development and psychometric properties. The
Gerontologist, 49, 236-247.
Castleman, M., Gallagher-Thompson, D., & Naythons, M. (2003). There’s still a person in there:
The complete guide to treating and coping with Alzheimer’s. New York: Putnam
Publishing.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 96
Chapleski, E., Lamphere, J. K., Kaczynski, R., Lichtenberg, P. A., & Dwyer, J. W. (1997).
Structure of a depression measure among American Indian elders: Confirmatory factor
analysis of the CES-D scale. Research on Aging, 19, 462-485.
Chen, J. C., Borson, S., & Scanlan, J. M. (2000). Stage-specific prevalence of behavioral
symptoms in Alzheimer’s disease in a multi-ethnic community sample. The American
Journal of Geriatric Psychiatry, 8, 123-133.
Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of-fit indexes for testing
measurement invariance. Structural Equation Modeling: A Multidisciplinary Journal, 9,
233-255.
Coffey, D. M., Marmol, L., Schock, L., & Adams, W. (2005). The influence of acculturation on
the Wisconsin Card Sorting Test by Mexican Americans. Archives of Clinical
Neuropsychology, 20, 795-803.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2
nd
ed.). Hillsdale, NJ:
Erlbaum.
Connell, C. M., Roberts, J. S., & McLaughlin, S. J. (2007). Public opinion about Alzheimer
disease among Blacks, Hispanics, and Whites: Results from a national survey. Alzheimer
Disease and Associated Disorders, 21, 232-240.
Cooper, C., Tandy, A. R., Balamurali, T. B. S., & Livingston, G. (2010). A systematic review
and meta-analysis of ethnic differences in use of dementia treatment, care, and research.
The American Journal of Geriatric Psychiatry, 18, 193-203.
Cuellar, J., Arnold, B., & Maldonado, R. (1995). Acculturation Rating Scale for Mexican-
Americans-II: A revision of the original ARSMA scale. Hispanic Journal of Behavioral
Sciences, 17, 275-304.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 97
Cummings, J. L., Mega, M., Gray, K., Rosenberg-Thompson, S., Carusi, D. A., & Gornbein, J.
(1994). The Neuropsychiatric Inventory: Comprehensive assessment of psychopathology
in dementia. Neurology, 44, 2308-2314.
Dilworth-Anderson, P., & Gibson, B. E. (1999). Ethnic minority perspectives on dementia,
family caregiving, and interventions. Generations, 23, 40-45.
Erzigkeit, H., Lehfeld, H., Peña-Casanova, J., Bieber, F., Yekrangi-Hartmann, C., Rupp, M., et
al. (2001). The Bayer-Activities of Daily Living Scale (B-ADL): Results from a
validation study in three European countries. Dementia and Geriatric Cognitive
Disorders, 12, 348-358.
Escobar, J. I., Burnam, A., Karno, M., Forsythe, A., Landsverk, J., & Golding, J. M. (1986). Use
of the Mini-Mental State Examination (MMSE) in a community population of mixed
ethnicity. Cultural and linguistic artifacts. The Journal of Nervous and Mental Disease,
174, 607-614.
Federal Interagency Forum on Aging-related Statistics (2000). Older Americans 2000: Key
indicators of well-being. Hyattsville, MD: Author.
Fillenbaum, G. G., Heyman, A., Huber, M. S., Ganguli, M., & Unverzagt, F. W. (2001).
Performance of elderly African American and White community residents on the
CERAD Neuropsychological Battery. Journal of the International Neuropsychological
Society, 7, 502-509.
Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975). “Mini-Mental State:” A practical
method for grading the cognitive state of patients for the clinician. Journal of Psychiatric
Research, 12, 189-198.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 98
Fuh, J., Liu, C., Mega, M. S., Wang, S., & Cummings, J. L. (2001). Behavioral disorders and
caregivers reaction in Taiwanese patients with Alzheimer’s disease. International
Psychogeriatrics, 13, 121-128.
Gallagher-Thompson, D. (2006). The family as the unit of assessment and treatment in work
with ethnically diverse older adults with dementia. In G. Yeo & D. Gallagher-Thompson
(Eds.), Ethnicity and the dementias (2
nd
ed., pp. 119-124). New York: Taylor & Francis.
Garver, M. S., & Mentzer, J. T. (1999). Logistics research methods: Employing structural
equation modeling to test for construct validity. Journal of Business Logistics, 20, 33-57.
Gelman, C. R. (2003). Learning from recruitment challenges: Barriers to diagnosis, treatment,
and research participation for Latinos with symptoms of Alzheimer’s disease. Journal of
Gerontological Social Work, 53, 94-113.
Gray, H. L., Jimenez, D. E., Cucciare, M. A., Tong, H., & Gallagher-Thompson, D. (2009).
Ethnic differences in beliefs regarding Alzheimer disease among dementia family
caregivers. The American Journal of Geriatric Psychiatry, 17, 925-933.
Gregorich, S. E. (2006). Do self-report instruments allow meaningful comparisons across diverse
population groups? Testing measurement invariance using the confirmatory factor
analysis framework. Medical Care, 44, S78-S94.
Hair, J. F., Anderson, R. E., Tatham, R. L., & Black, W. C. (1998). Multivariate data analysis
with readings (5
th
ed.). Englewood Cliffs, NJ: Prentice-Hall.
Hebert, L. E, Scherr, P. A., Bienias, J. L., Bennett, D. A., & Evans, D. A. (2003). Alzheimer
Disease in the U.S. population: Prevalence estimates using the 2000 Census. Archives of
Neurology, 60, 1119-1122.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 99
Henderson, N. J., & Gutierrez-Mayka, M. (1992). Ethnocultural themes in caregiving to
Alzheimer’s disease patients in Hispanic families. Clinical Gerontologist, 11, 59-74.
Hindmarch, I., Lehfeld, H., de Jong, P., & Erzigkeit, H. (1998). The Bayer Activities of Daily
Living (B-ADL) Scale. Dementia and Geriatric Cognitive Disorders, 9, 20-26.
Hinton, L., Franz, C. E., & Friend, J. (2004). Pathways to dementia diagnosis: Evidence for
cross-ethnic differences. Alzheimer Disease & Associated Disorders, 18, 134-144.
Hinton, L., Franz, C. E., Yeo, G., & Levkoff, S. E. (2005). Conceptions of dementia in a
multiethnic sample of family caregivers. Journal of the American Geriatrics Society, 53,
1405-1410.
Hinton, L., Haan, M., Geller, S., & Mungas, D. (2003). Neuropsychiatric symptoms in Latino
elders with dementia or cognitive impairment without dementia and factors that modify
their association with caregiver depression. The Gerontologist, 43, 669-677.
Hinton, W. L., & Levkoff, S. (1999). Constructing Alzheimer’s. Narratives of lost identities,
confusion, and loneliness in old age. Culture, Medicine and Psychiatry, 23, 453-475.
Hoelter, J. W. (1983). The analysis of covariance structures: Goodness-of-fit indices.
Sociological Methods & Research, 11, 325-344.
Hohl, U., Grundman, M., Salmon, D. P., Thomas, R. G., & Thal, L. J. (1999). Mini-Mental State
Examination and Mattis Dementia Rating Scale performance differs in Hispanic and non-
Hispanic Alzheimer’s disease patients. Journal of the International Neuropsychological
Society, 5, 301-307.
Horn, J. L., & McArdle, J. J. (1992). A practical and theoretical guide to measurement invariance
in aging research. Experimental Aging Research, 18, 117-144.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 100
Hu, L. T., & Bentler, P. M. (1995). Evaluating model fit. In R. H. Hoyle (Ed.), Structural
equation modeling: Concepts, issues and applications (pp. 76-99). Thousand Oaks, CA:
Sage.
Hui, C. H., & Triandis, H. C. (1985). Measurement in cross-cultural psychology: A review and
comparison of strategies. Journal of Cross-Cultural Psychology, 16, 131-152.
Jacobs, D. M., Sano, M., Albert, S., Schofield, P., Dooneief, G., & Stern, Y. (1997). Cross-
cultural neuropsychological assessment: A comparison of randomly selected,
demographically matched cohorts of English- and Spanish-speaking older adults. Journal
of Clinical and Experimental Neuropsychology, 19, 331-339.
Jones, R. S., Chow, T. W., & Gatz, M. (2006). Asian Americans and Alzheimer’s disease:
Assimilation, culture, and beliefs. Journal of Aging Studies, 20, 11-25.
Jorm, A. F. (2004). The Informant Questionnaire on Cognitive Decline in the Elderly
(IQCODE): A review. International Psychogeriatrics, 16, 275-293.
Kang, H. S., Ahn, I. S., Kim, J. H., & Kim, D. K. (2010). Neuropsychiatric symptoms in Korean
patients with Alzheimer’s disease: Exploratory factor analysis and confirmatory factor
analysis of the Neuropsychiatric Inventory. Dementia and Geriatric Cognitive Disorders,
29, 82-87.
Kaplan, E., Goodglass, H., & Weintraub, S. (1983). The Boston Naming Test. Philadelphia: Lea
and Febiger.
Kaufer, D. I., Cummings, J. L., Christine, D., Bray, T., Castellon, S., Masterman, D., et al
(1998). Assessing the impact of behavioral disturbances in Alzheimer’s disease: The
Neuropsychiatric Inventory Caregiver Distress Scale. Journal of the American Geriatrics
Society, 46, 210-215.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 101
Kaufer, D. I., Cummings, J. L., Ketchel, P., Smith, V., MacMillan, A., Shelley, T., et al. (2000).
Validation of the NPI-Q, a brief clinical form of the Neuropsychiatric Inventory. The
Journal of Neuropsychiatry and Clinical Neurosciences, 12, 233-239.
Kemp, N. M., Brodaty, H., Pond, D., & Luscombe, G. (2002). Diagnosing dementia in primary
care: The accuracy of informant reports. Alzheimer Disease and Associated Disorders,
16, 171-176.
Killgore, W. D., & Adams, R. L. (1999). Prediction of Boston Naming Test performance from
Vocabulary scores: Preliminary guidelines for interpretation. Perceptual and Motor
Skills, 89, 327-337.
Knapp, M. J., Knopman, D. S., Solomon, P. R., Pendlebury, W. W., Davis, C. S., & Gracon, S. I.
(1994). A 30-week randomized controlled trial of high-dose tacrine in patients with
Alzheimer’s disease. The Journal of the American Medical Association, 271, 985-991.
Landis, J. R., & Koch, G. G. (1977). The measurement of observer agreement for categorical
data. Biometrics, 33, 159-174.
Lange, R. T., Hopp, G. A., & Kang, N. (2004). Psychometric properties and factor structure of
the Neuropsychiatric Inventory – Nursing Home version in an elderly neuropsychiatric
population. International Journal of Geriatric Psychiatry, 19, 440-448.
Lawton, M. P., & Brody, E. M. (1969). Assessment of older people: Self-maintaining and
instrumental activities of daily living. Gerontologist, 9, 179-186.
Le Carret, N., Lafont, S., Letenneur, L., Dartigues, J-F., Mayo, W., & Fabrigoule, C. (2003). The
effect of education on cognitive performances and its implication for the constitution of
the cognitive reserve. Developmental Neuropsychology, 23, 317-337.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 102
Levkoff, S. U., Levy, B., & Weitzman, P. F. (1999). The role of religion and ethnicity in the help
seeking of family caregivers of elders with Alzheimer’s disease and related disorders.
Journal of Cross-Cultural Gerontology, 14, 335-356.
Lezak, M. D., Howieson, D. B., & Loring, D. W. (2004). Neuropsychological assessment (4
th

ed.). New York: Oxford.
Lonie, J. A., Tierney, K. M., & Ebmeier, K. P. (2009). Screening for mild cognitive impairment:
A systematic review. International Journal of Geriatric Psychiatry, 24, 902-915.
López, S. R., & Taussig, I. M. (1991). Cognitive-intellectual functioning of Spanish-speaking
impaired and nonimpaired elderly: Implications for culturally sensitive assessment.
Psychological Assessment: A Journal of Consulting and Clinical Psychology, 3, 448-454.
Mahoney, D. F., Cloutterbuck, J., Neary, S., & Zhan, L. (2005). African American, Chinese, and
Latino family caregivers’ impressions of the onset and diagnosis of dementia: Cross-
cultural similarities and differences. The Gerontologist, 45, 783-792.
Malmstrom, T. K., Miller, D. K., Coats, M. A., Jackson, P., Miller, J. P., Morris, J. C., et al.
(2009). Informant-based dementia screening in a population-based sample of African
Americans. Alzheimer’s Disease and Associated Disorders, 23, 117-123.
Manly, J. J., & Espino, D. V. (2004). Cultural influences on dementia recognition and
management. Clinics in Geriatric Medicine, 20, 93-119.
Matarazzo, J. D., & Herman, D. O. (1984). Base rate data for the WAIS-R: Test-retest stability
and VIQ-PIQ differences. Journal of Clinical Neuropsychology, 6, 351-366.
Mayeux, R., Stern, Y., Rosen, J., & Leventhal, J. (1981). Depression, intellectual impairment,
and Parkinson disease. Neurology, 31, 645-650.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 103
McKhann, G., Drachman, D., Folstein, M., Katzman, R., Price, D., & Stadlan, E. M. (1984).
Clinical diagnosis of Alzheimer’s disease: Report of the NINCDS-ADRDA Work Group
under the auspices of Department of Health and Human Services Task Force on
Alzheimer’s Disease. Neurology, 34, 939-944.
Meredith, W. (1993). Measurement invariance, factor analysis, and factorial invariance.
Psychometrika, 58, 525-543.
Miller, S. W., Heaton, R. K., Kirson, D., & Grant, I. (1997). Neuropsychological (NP)
assessment of African Americans. Journal of the International Neuropsychological
Society, 3, 49.
Monaci, L., & Morris, R. G. (2012). Neuropsychological screening performance and the
association with activities of daily living and instrumental activities of daily living in
dementia: Baseline and 18- to 24-month follow-up. International Journal of Geriatric
Psychiatry, 27, 197-204.
Morhardt, D., Pereyra, M., & Iris, M. (2010). Seeking a diagnosis for memory problems. The
experiences of caregivers and families in 5 limited English proficiency communities.
Alzheimer Disease and Associated Disorders, 24, S42-S48.
Morris, J. C. (1993). The Clinical Dementia Rating (CDR): Current version and scoring rules.
Neurology, 43, 2412–2414.
Morris, J. C., Heyman, A., Mohs, R. C., Hughes, J. P., van Belle, G., Fillenbaum, G., et al.
(1989). The Consortium to Establish a Registry for Alzheimer’s Disease (CERAD). Part
I. Clinical and neuropsychological assessment of Alzheimer’s disease. Neurology, 39,
1159-1165.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 104
Morris, J. C., Weintraub, S., Chui, H. C., Cummings, J., DeCarli, C., Ferris, S., et al. (2006). The
Uniform Data Set (UDS): Clinical and cognitive variables and descriptive data from
Alzheimer Disease Centers. Alzheimer Disease and Associated Disorders, 20, 210-216.
Mukadam, N., Cooper, C., & Livingston, G. (2011). A systematic review of ethnicity and
pathways to care in dementia. International Journal of Geriatric Psychiatry, 26, 12-20.
Mungas, D. (1991). In-office mental status testing: A practical guide. Geriatrics, 46, 54-56.
Neary, S. R., & Mahoney, D. F. (2005). Dementia caregiving: The experiences of
Hispanic/Latino caregivers. Journal of Transcultural Nursing, 16, 163-170.
Ortiz, F., & Fitten, L. J. (2000). Barriers to healthcare access for cognitively impaired older
Hispanics. Alzheimer Disease and Associated Disorders, 14, 141-150.
Ortiz, F., Fitten, L. J., Cummings, J. L., Hwang, S., & Fonseca, M. (2006). Neuropsychiatric and
behavioral symptoms in a community sample of Hispanics with Alzheimer’s disease.
American Journal of Alzheimer’s Disease and Other Dementias, 21, 263-273.
Pfeffer, R. I., Kurosaki, T. T., Harrah, C., Chance, J. M., Bates, D., Detels, R., et al. (1981). A
survey diagnostic tool for senile dementia. American Journal of Epidemiology, 114, 515-
527.
Pfeffer, R. I., Kurosaki, T. T., Harrah, C. H., Chance, J. M., & Filos, S. (1982). Measurement of
functional activities in older adults in the community. Journal of Gerontology, 37, 323-
329.
Potter, G. G., Plassman, B. L., Burke, J. R., Kabeto, M. U., Langa, K. M., Llewellyn, D. J., et al.
(2009). Cognitive performance and informant reports in the diagnosis of cognitive
impairment and dementia in African Americans and whites. Alzheimer’s & Dementia, 5,
445-453.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 105
Raykov, T., & Marcoulides, G. A. (2000). A first course in structural equation modeling.
Mahwah, NJ: Erlbaum.
Reitan, R. M. (1958). Validity of the Trail Making test as an indicator of organic brain damage.
Perceptual Motor Skills, 8, 271-276.
Rogers, S. L., Farlow, M. R., Doody, R. S., Mohs, R., Friedhoff, L.T., & the Donepezil Study
Group (1998). A 24-week, double-blind, placebo-controlled trial of donepezil in patients
with Alzheimer’s disease. Neurology, 50, 136-145.
Roman, G. C., Tatemichi, T. K., Erkinjuntti, T., Cummings, J. L., Masdeu, J. C., Garcia, J. H., et
al. (1993). Vascular dementia: Diagnostic criteria for research studies: Report of the
NINDS-AIREN International Workshop. Neurology. 43, 250-260.
Sayegh, P., & Knight, B. G. (2012). Crosscultural differences in dementia: The Sociocultural
Health Belief Model. Manuscript submitted for publication.
Schreiber, J. B., Nora, A., Stage, F. K., Barlow, E. A., & King, J. (2006). Reporting structural
equation modeling and confirmatory factor analysis results: A review. The Journal of
Educational Research, 99, 323-337.
Schwartz, S. J., Unger, J. B., Zamboanga, B. L., & Szapocznik, J. (2010). Rethinking the concept
of acculturation: Implications for theory and research. American Psychologist, 65, 237-
251.
Simons, R. C., & Hughes, C. C. (1993). Culture-bound syndromes. In A. Gaw (Ed.), Culture,
ethnicity, and mental illness (pp. 75-99). Washington: American Psychiatric Press.
Sink, K. M., Covinsky, K. E., Newcomer, R., & Yaffe, K. (2004). Ethnic differences in the
prevalence and pattern of dementia-related behaviors. Journal of the American Geriatrics
Society, 52, 1277-1283.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 106
Snow, W. G., Tierney, M. C., Zorzitto, M. J., Fisher, R. H., & Reid, D. W. (1989). WAIS-R test-
retest reliability in a normal elderly sample. Journal of Clinical and Experimental
Neuropsychology, 11, 423-428.
Sperling, R., & Johnson, K. (2010). Pro: Can biomarkers be gold standards in Alzheimer’s
disease? Alzheimer’s Research & Therapy, 2, 17.
Steenkamp, J. E. M., & Baumgartner, H. (1998). Assessing measurement invariance in
cross ‐national consumer research. Journal of Consumer Research, 25, 78-107.
Steinberg, M., Tschanz, J. T., Corcoran, C., Steffens, D. C., Norton, M. C., Lyketsos, C. G., et al.
(2004). The persistence of neuropsychiatric symptoms in dementia: The Cache County
Study. International Journal of Geriatric Psychiatry, 19, 19-26.
Stephenson, M. (2000). Development and validation of the Stephenson Multigroup Acculturation
Scale (SMAS). Psychological Assessment, 12, 77-88.
Storandt, M. (2008). Cognitive deficits in the early stages of Alzheimer’s disease. Psychological
Science, 17, 198-202.
Storandt, M., Grant, E. A., Miller, J. P., & Morris, J. C. (2006). Longitudinal course and
neuropathological outcomes in original vs. revised MCI and in preMCI. Neurology, 67,
467-473.
Streib, G. F. (1966). Participants and drop-outs in a longitudinal study. Journal of Gerontology,
21, 200-209.
Suárez-Orozco, M. M., & Páez, M. M. (2002). Introduction. In M. M. Suárez-Orozco & M. M.
Páez (Eds.), Latinos: Remaking America (pp. 1-43). Berkeley: University of California
Press.
CROSSCULTURAL DIFFERENCES IN DEMENTIA 107
Szapocznik, J., Kurtines, W., & Fernandez, T. (1980). Bicultural involvement and adjustment in
Hispanic-American youths. International Journal of Intercultural Relations, 4, 353-365.
Tabert, M. H., Albert, S. M., Borukhova-Milov, L., Camacho, Y., Pelton, G., Liu, X., et al.
(2002). Functional deficits in patients with mild cognitive impairment: Prediction of AD.
Neurology, 58, 758-764.
Taylor, A. B., West, S. G., & Aiken, L. S. (2006). Loss of power in logistic, ordinal logistic, and
probit regression when an outcome variable is coarsely categorized. Educational and
Psychological Measurement, 66, 228-239.
U.S. Census Bureau. (2008). An older and more diverse nation by midcentury. Retrieved
October 1, 2010, from http://www.census.gov/newsroom/releases/archives/population
/cb08-123.html.
Valle, R. (1994). Culture-fair behavioral symptom differential assessment and intervention in
dementing illness. Alzheimer Disease and Associated Disorders, 8, 21-45.
Valle, R., & Lee, B. (2002). Research priorities in the evolving demographic landscape of
Alzheimer’s Disease and associated dementias. Alzheimer Disease and Associated
Disorders, 16, S64-S76.
Valle, R., Yamada, A. M., & Barrio, C. (2004). Ethnic differences in social network help-seeking
strategies among Latino and Euro-American dementia caregivers. Aging & Mental
Health, 8, 535-543.
van de Vijver, F., & Tanzer, N. K. (Eds.) (1997). Methods and data analysis for cross-cultural
research. Thousand Oaks, CA: Sage.

CROSSCULTURAL DIFFERENCES IN DEMENTIA 108
Van Lieshout, R. J., Cleverley, K., Jenkins, J. M., & Georgiades, K. (2011). Assessing the
measurement invariance of the Center for Epidemiologic Studies Depression Scale across
immigrant and non-immigrant women in the postpartum period. Archives of Women’s
Mental Health, 14, 413-423.
Wechsler, D. (1981). WAIS-R manual. New York: The Psychological Corporation.
Wechsler, D. (1987a). Wechsler Memory Scale—Revised Manual. San Antonio: The
Psychological Corporation.
Wechsler, D. (1987b). Wechsler Adult Intelligence Scale—Revised. San Antonio: The
Psychological Corporation.
Weisman, A. G., López, S. R., Ventura, J., Nuechterlein, K. H., Goldstein, M. J., & Hwang, S.
(2000). A comparison of psychiatric symptoms between Anglo-Americans and Mexican-
Americans with schizophrenia. Schizophrenia Bulletin, 26, 817-824.
Zarit, S. H., & Zarit, J. M. (2007). Mental disorders in older adults (2
nd
ed.). New York: The
Guilford Press.

CROSSCULTURAL DIFFERENCES IN DEMENTIA 109
Bibliography
No additional sources were consulted but not cited for this dissertation.

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 110
Table 1
Descriptive Statistics and P Values for t tests and Chi-square Statistics by Ethnicity and Diagnosis
Overall (N = 11,525),
Mean (SD)
NCF (N = 5,202),
Mean (SD)
Dementia (N = 6,323),
Mean (SD)
NCF vs. Dementia,
P value
Age (years)
Hispanic (N = 444) 71.51 (10.31) 69.47 (9.70) 73.20 (10.51) <.001
NHW (N = 11,080) 72.61 (10.78) 71.48 (10.97) 73.54 (10.53) <.001
Prob (H
0
) .034 .011 .615 ---
Patient education (years)
Hispanic (N = 441) 13.25 (3.93) 13.99 (3.65) 12.63 (4.05) <.001
NHW (N = 10,997) 15.26 (2.98) 15.83 (2.72) 14.79 (3.10) <.001
Prob (H
0
) <.001 <.001 <.001 ---
Informant education (years)
Hispanic (N = 406) 14.52 (3.10) 14.62 (3.08) 14.44 (3.12) .556
NHW (N = 10,517) 15.59 (2.68) 15.80 (2.68) 15.41 (2.66) <.001
Prob (H
0
) <.001 <.001 <.001 ---

Overall, N (%) NCF, N (%) Dementia, N (%)
NCF vs. Dementia,
P value
Women
Hispanic (N = 444) 261 (58.8) 139 (69.2) 122 (50.2) <.001
NHW (N = 11,081) 6,104 (55.1) 3,087 (61.7) 3,017 (49.6) <.001
Prob (H
0
) .124 .033 .858 ---
Coresidency
Hispanic (N = 444) 254 (57.2) 104 (51.7) 150 (61.7) .034
NHW (N = 11,081) 7,223 (65.2) 2,842 (56.8) 4,381 (72.1) <.001
Prob (H
0
) .001 .154 <.001 ---
English primary language
Hispanic (N = 444) 305 (68.7) 157 (78.1) 148 (60.9) <.001
NHW (N = 11,072) 10,927 (98.6) 4,935 (98.7) 5,992 (98.6) .363
Prob (H
0
) <.001 <.001 <.001 ---

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 111
Informant-reported memory decline
Hispanic (N = 439) 263 (59.2) 24 (11.9) 239 (98.4) <.001
NHW (N = 10,987) 6,481 (58.5) 661 (13.2) 5,820 (95.7) <.001
Prob (H
0
) .700 .604 .021 ---
Global CDR Score
Hispanic (N = 444) <.001
0 169 (38.1) 169 (84.1) 0 (0.00) ---
0.5 82 (18.5) 32 (15.9) 50 (20.6) ---
1 114 (25.7) 0 (0.00) 114 (46.9) ---
2 58 (13.1) 0 (0.00) 58 (23.9) ---
3 21 (4.7) 0 (0.00) 21 (8.6) ---
NHW (N = 11,081) <.001
0 4,548 (41.0) 4,502 (90.0) 46 (0.8) ---
0.5 2,343 (21.1) 493 (9.9) 1,850 (30.4) ---
1 2,723 (24.6) 6 (0.1) 2,717 (44.7) ---
2 968 (8.7) 0 (0.00) 968 (15.9) ---
3 499 (4.5) 0 (0.00) 499 (8.2) ---
Prob (H
0
) .024 .007 <.001 ---
Overall,
Mean (SD)
NCF,
Mean (SD)
Dementia,
Mean (SD)
NCF vs. Dementia,
P value
Overall NP, ethnic group-specific
norms (sum of scaled scores)

Hispanic (N = 444) -14.81 (17.61) 0.00 (8.28) -27.05 (13.34) <.001
NHW (N = 11,081) -16.53 (19.89) -0.02 (7.66) -30.10 (16.25) <.001
Prob (H
0
) .045 .976 .001 ---
Overall NP, combined ethnic-group
norms (sum of scaled scores)

Hispanic (N = 444) -21.71 (19.18) -5.73 (8.93) -34.93 (14.83) <.001
NHW (N = 11,081) -16.12 (19.69) 0.23 (7.59) -29.58 (16.06) <.001
Prob (H
0
) <.001 <.001 <.001 ---

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 112
Total FAQ
Hispanic (N = 444) 9.86 (10.92) 0.43 (1.47) 17.67 (9.03) <.001
NHW (N = 11,081) 9.19 (10.37) 0.48 (1.96) 16.34 (8.90) <.001
Prob (H
0
) .200 .699 .023 ---
Total NPI-Q Frequency
Hispanic (N = 444) 2.38 (2.54) 0.80 (1.36) 3.68 (2.55) <.001
NHW (N = 11,081) 2.18 (2.46) 0.69 (1.28) 3.40 (2.53) <.001
Prob (H
0
) .090 .216 .088 ---
Total NPI-Q Severity
Hispanic (N = 444) 3.93 (4.88) 1.16 (2.29) 6.23 (5.24) <.001
NHW (N = 11,081) 3.31 (4.41) 0.90 (1.93) 5.30 (4.86) <.001
Prob (H
0
) .009 .117 .003 ---
Length of cognitive decline (years)
Hispanic (N = 232) --- --- 4.77 (3.66) ---
NHW (N = 5,842) --- --- 4.94 (3.51) ---
Prob (H
0
) --- --- .471 ---
Note. NCF = normal cognitive functioning; NHW = non-Hispanic White; Prob (H
0
) = null hypothesis that group differences are not significant; NP =
neuropsychological test performance; FAQ = Functional Assessment Questionnaire; NPI-Q = Neuropsychiatric Inventory Questionnaire; CDR = Clinical
Dementia Rating Scale.

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 113
Table 2
Correlation Coefficients among the Demographic and Key Variables for the Combined Hispanic and Non-Hispanic White National
Sample
Variable Age Sex
Patient
Education
Informant
Education
Overall
NP
FAQ
Total
NPI-Q
Frequency
NPI-Q
Severity
IRMD Diagnosis
Age 1.000

Sex -.002 1.000

Patient
Education
-.119
**
-.160
**
1.000

Informant
Education
-.008 .114
**
.368
**
1.000

Overall
NP
-.153
**
.087
**
.233
**
.102
**
1.000

FAQ Total .126
**
-.059
**
-.172
**
-.059
**
-.785
**
1.000

NPI-Q
Frequency
-.041
**
-.121
**
-.127
**
-.062
**
-.471
**
.582
**
1.000

NPI-Q
Severity
-.053
**
-.116
**
-.123
*
-.066
**
-.439
**
.559
**
.939
**
1.000
IRMD .112
**
-.138
**
-.144
**
-.070
**
-.648
**
.673
**
.510
**
.454
**
1.000
Diagnosis .098
**
-.124
**
-.172
**
-.072
**
-.753
**
.762
**
.549
**
.497
**
.833
**
1.000
Note. NP = neuropsychological test performance (using ethnic group-specific norms); FAQ = Functional Assessment Questionnaire; NPI-Q = Neuropsychiatric
Inventory Questionnaire; IRMD = informant-reported memory decline.
* p < .05, ** p < .01
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 114
Table 3
Correlation Coefficients among the Demographic and Key Variables for the Hispanic National Sample
Variable Age Sex
Patient
Education
English
Primary
Language
Informant
Education
Overall
NP
FAQ
Total
NPI-Q
Frequency
NPI-Q
Severity
IRMD Diagnosis
Age 1.000

Sex -.122
*
1.000

Patient
Education
-.254
**
-.011 1.000

English
Primary
Language
-.080 .017 .115
*
1.000

Informant
Education
-.073 .034 .425
**
.036 1.000

Overall
NP
-.220
**
.147
**
.355
**
.191
**
.120
*
1.000

FAQ
Total
.156
**
-.086 -.204
**
-.084 -.005 -.780
**
1.000

NPI-Q
Frequency
-.034 -.139
**
-.082 -.129
**
-.004 -.460
**
.570
**
1.000

NPI-Q
Severity
-.017 -.158
**
-.093 -.113
*
-.032 -.416
**
.535
**
.940
**
1.000
IRMD .163
**
-.185
**
-.186
**
-.170
**
-.075 -.714
**
.729
**
.558
**
.512
**
1.000
Diagnosis .180
**
-.192
**
-.173
**
-.185
**
-.029 -.766
**
.787
**
.566
**
.518
**
.879
**
1.000
Note. NP = neuropsychological test performance (using ethnic group-specific norms); FAQ = Functional Assessment Questionnaire; NPI-Q = Neuropsychiatric
Inventory Questionnaire; IRMD = informant-reported memory decline.
* p < .05, ** p < .01

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 115
Table 4
Correlation Coefficients among the Demographic and Key Variables for the Non-Hispanic White National Sample
Variable Age Sex
Patient
Education
Informant
Education
Overall
NP
FAQ
Total
NPI-Q
Frequency
NPI-Q
Severity
IRMD Diagnosis
Age 1.000

Sex .002 1.000

Patient
Education
-.117
**
-.168
**
1.000

Informant
Education
-.007 .119
**
.359
**
1.000

Overall
NP
-.151
**
.084
**
.232
**
.103
**
1.000

FAQ Total .125
**
-.058
**
-.170
**
-.061
**
-.786
**
1.000

NPI-Q
Frequency
-.041
**
-.121
**
-.128
**
-.064
**
-.472
**
.583
**
1.000

NPI-Q
Severity
-.054
**
-.115
**
-.123
**
-.065
**
-.441
**
.560
**
.939
**
1.000
IRMD .110
**
-.136
**
-.143
**
-.070
**
-.646
**
.671
**
.508
**
.452
**
1.000
Diagnosis .095
**
-.121
**
-.174
**
-.074
**
-.753
**
.761
**
.549
**
.496
**
.832
**
1.000
Note. NP = neuropsychological test performance (using ethnic group-specific norms); FAQ = Functional Assessment Questionnaire; NPI-Q = Neuropsychiatric
Inventory Questionnaire; IRMD = informant-reported memory decline.
* p < .05, ** p < .01
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 116
Table 5
Standardized Factor Loadings from the Functional Assessment Questionnaire Baseline Models
Hispanic Non-Hispanic White

Dementia
Severity
Dementia
Severity
Item 1 .98 .95
Item 2 .96 .92
Item 3 .95 .94
Item 4 .86 .84
Item 5 .89 .83
Item 6 .94 .89
Item 7 .94 .92
Item 8 .89 .87
Item 9 .92 .92
Item 10 .92 .91
Note. All factor loadings are significant at the <.001 level.
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 117
Table 6
Summary of Tests for Crosscultural Invariance of the Functional Assessment Questionnaire Measurement Model
GFI CFI TLI RMSEA
Model 1: Number of
factors (i.e., 1) invariant
.950 .895 .865 .057
Model 2: Model 1 with
pattern of factor loading
invariant
.950 .895 .880 .054
Model 3: Model 1 with
intercepts invariant and
latent mean freely
estimated in one group
--- .910 .908 .107
Note: CI = confidence interval.
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 118
Table 7
Standardized Factor Loadings from the Neuropsychiatric Inventory Questionnaire Baseline Models
Hispanic Non-Hispanic White
Hyperactivity Affect Psychosis
Apathy/
Vegetative
Hyperactivity Affect Psychosis
Apathy/
Vegetative
Agitation .75 .72
Disinhibition .60 .53
Irritability .63 .67
Anxiety .68 .64
Depression .60 .58
Delusions .77 .69
Hallucinations .51 .48
Apathy .73 .64
Sleep .38 .46
Appetite .49 .46
Note. All factor loadings are significant at the <.001 level.

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 119
Table 8
Factor Correlations for the Neuropsychiatric Inventory Questionnaire Baseline Models
Hispanic Non-Hispanic White

Hyperactivity Affect Psychosis
Apathy/
Vegetative
Hyperactivity Affect Psychosis
Apathy/
Vegetative
Hyperactivity -- --
Affect .70 -- .66 --
Psychosis .55 .51 -- .56 .38 --
Apathy/Vegetative .75 .83 .54 -- .71 .86 .35 --
Note. All correlations are significant at the .05 level (t > 1.96)

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 120
Table 9
Summary of Tests for Crosscultural Invariance of the Neuropsychiatric Inventory Questionnaire Measurement and Structural Models
GFI CFI TLI RMSEA
Model 1: Number of
factors (i.e., 4) invariant
.983 .918 .873 .024
Model 2: Model 1 with
pattern of factor loading
invariant
.982 .918 .884 .023
Model 3: Model 1 with
pattern of factor loading
invariant and factor
variances and covariances
invariant
.982 .918 .894 .022
Model 4: Model 3 with
intercepts invariant and
latent mean freely
estimated in one group
--- .963 .953 .030
Note: CI = confidence interval
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 121
Table 10
Goodness-of-fit Statistics for the First and Reduced Path Models for Hispanics and Non-Hispanic Whites Using Ethnic Group-
Specific Neuropsychological Test Norms
Hispanics CFI TLI RMSEA ECVI
First (i.e., hypothesized)
model
.964 .897 .092 .413
Reduced model #1 .959 .919 .082 .418
Reduced model #2 .969 .890 .122 .260
Non-Hispanic Whites CFI TLI RMSEA

First (i.e., hypothesized)
model
.966 .881 .103 .145
Reduced model .966 .889 .099 .145

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 122
Table 11
Unstandardized Regression Coefficients from Path Models Using Ethnic Group-Specific Norms for Hispanics
First Model
Endogenous variable
Independent variable Overall NP FAQ total NPI-Q total IRMD
Clinicians’
diagnosis
Exogenous
Age -0.28
***
0.17
***
-0.01 0.01
***
0.00
Sex 1.76 -- -0.34 -- -0.03
Education 0.84
***
-- -- -- 0.01
*

Primary language 3.02
**
-- -- -- -0.04
Informant education -- -- -0.01 -0.01 --
Mediating
Overall NP -- -- -- -- -0.01
***

FAQ -- -- -- -- 0.01
***

NPI-Q frequency -- -- -- -- 0.01
IRMD -- -- -- -- 0.58
***

Reduced Model #2
Endogenous variable
Independent variable Overall NP FAQ total NPI-Q total IRMD
Clinicians’
diagnosis
Exogenous
Age -0.29
***
0.17
***
-- 0.01
***
--
Education 0.79
***
-- -- -- 0.02
*

Primary language 2.86
**
-- -- -- --
Mediating
Overall NP -- -- -- -- -0.01
***

FAQ -- -- -- -- 0.01
***

IRMD -- -- -- -- 0.61
***

Note. NP = neuropsychological test performance (using ethnic group-specific norms); FAQ = Functional Assessment Questionnaire; NPI-Q = Neuropsychiatric
Inventory Questionnaire; IRMD = informant-reported memory decline.
* p < .05, ** p < .01, *** p < .001
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 123
Table 12
Unstandardized Regression Coefficients from Path Models Using Ethnic Group-Specific Norms for Non-Hispanic Whites
First Model
Endogenous variable
Independent variable Overall NP FAQ total NPI-Q total IRMD
Clinicians’
diagnosis
Exogenous
Age -0.26
***
0.12
***
-0.01
***
0.01
***
0.00
***

Sex 1.51
***
-- -0.32
***
-- -0.01
**

Education 0.67
***
-- -- -- 0.00
Informant education -- -- -0.02
**
-0.01
***
--
Mediating
Overall NP -- -- -- -- -0.01
***

FAQ -- -- -- -- 0.01
***

NPI-Q frequency -- -- -- -- 0.01
***

IRMD -- -- -- -- 0.52
***

Reduced Model
Endogenous variable
Independent variable Overall NP FAQ total NPI-Q total IRMD
Clinicians’
diagnosis
Exogenous
Age -0.26
***
0.12
***
-0.01
***
0.01
***
0.00
***

Sex 1.51
***
-- -0.31
***
-- -0.01
**

Education 0.67
***
-- -- -- --
Informant education -- -- -0.02
**
-0.01
***
--
Mediating
Overall NP -- -- -- -- -0.01
***

FAQ -- -- -- -- 0.01
***

NPI-Q frequency -- -- -- -- 0.01
***

IRMD -- -- -- -- 0.52
***

Note. NP = neuropsychological test performance (using ethnic group-specific norms); FAQ = Functional Assessment Questionnaire; NPI-Q = Neuropsychiatric
Inventory Questionnaire; IRMD = informant-reported memory decline.
** p < .01, *** p < .001
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 124
Table 13
Goodness-of-fit Statistics for the First and Reduced Path Models for Hispanics and Non-Hispanic Whites Using Combined Ethnic
Group Neuropsychological Test Norms
Hispanics CFI TLI RMSEA ECVI
First (i.e., hypothesized)
model
.962 .890 .093 .415
Reduced model #1 .960 .918 .080 .407
Reduced model #2 .957 .888 .107 .326
Non-Hispanic Whites CFI TLI RMSEA

First (i.e., hypothesized)
model
.964 .876 .103 .146
Reduced model .964 .885 .099 .146

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 125
Table 14
Unstandardized Regression Coefficients from Path Models Using Combined Ethnic Group Norms for Hispanics
First Model
Endogenous variable
Independent variable Overall NP FAQ total NPI-Q total IRMD
Clinicians’
diagnosis
Exogenous
Age -0.32
***
0.17
***
-0.01 0.01
***
0.00
Sex 3.51
***
-- -0.35 -- -0.02
Education 0.90
***
-- -- -- 0.01
**

Primary language 3.20
**
-- -- -- -0.03
Informant education -- -- 0.00 -0.01 --
Mediating
Overall NP -- -- -- -- -0.01
***

FAQ -- -- -- -- 0.01
***

NPI-Q frequency -- -- -- -- 0.01
IRMD -- -- -- -- 0.54
***

Reduced Model #2
Endogenous variable
Independent variable Overall NP FAQ total NPI-Q total IRMD
Clinicians’
diagnosis
Exogenous
Age -0.33
***
0.17
***
-- 0.01
***
--
Sex 3.51
***

Education 0.84
***
-- -- -- 0.01
**

Primary language 3.25
**
-- -- -- --
Mediating
Overall NP -- -- -- -- -0.01
***

FAQ -- -- -- -- 0.01
***

IRMD -- -- -- -- 0.56
***

Note. NP = neuropsychological test performance (using ethnic group-specific norms); FAQ = Functional Assessment Questionnaire; NPI-Q = Neuropsychiatric
Inventory Questionnaire; IRMD = informant-reported memory decline.
** p < .01, *** p < .001
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 126
Table 15
Unstandardized Regression Coefficients from Path Models Using Combined Ethnic Group Norms for Non-Hispanic Whites
First Model
Endogenous variable
Independent variable Overall NP FAQ total NPI-Q total IRMD
Clinicians’
diagnosis
Exogenous
Age -0.30
***
0.12
***
-0.01
***
0.01
***
0.00
***

Sex 1.58
***
-- -0.32
***
-- -0.01
Education 0.83
***
-- -- -- 0.00
***

Informant education -- -- -0.02
**
-0.00
***
--
Mediating
Overall NP -- -- -- -- -0.01
***

FAQ -- -- -- -- 0.01
***

NPI-Q frequency -- -- -- -- 0.01
***

IRMD -- -- -- -- 0.47
***

Reduced Model #2
Endogenous variable
Independent variable Overall NP FAQ total NPI-Q total IRMD
Clinicians’
diagnosis
Exogenous
Age -0.30
***
0.12
***
-0.01
***
0.01
***
0.00
***

Sex 1.61
***
-- -0.32
***
-- --
Education 0.83
***
-- -- -- 0.02
*

Informant Education -- -- -0.02
**
0.00
***
--
Mediating
Overall NP -- -- -- -- -0.01
***

FAQ -- -- -- -- 0.01
***

NPI-Q frequency -- -- -- -- 0.01
***

IRMD -- -- -- -- 0.48
***

Note. NP = neuropsychological test performance (using ethnic group-specific norms); FAQ = Functional Assessment Questionnaire; NPI-Q = Neuropsychiatric
Inventory Questionnaire; IRMD = informant-reported memory decline.
* p < .05, ** p < .01, *** p < .001
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 127
Table 16

Goodness-of-fit Statistics for the First and Reduced Path Models for Hispanics and Non-Hispanic Whites Using Ethnic Group-

Specific Neuropsychological Test Norms and the Clinical Dementia Rating Scale as the Dependent Variable
Hispanics CFI TLI RMSEA ECVI
First (i.e., hypothesized)
model
.965 .900 .092 .412
Reduced model #1 .966 .933 .076 .390
Reduced model #2 .964 .933 .076 .393
Reduced model #3 .973 .915 .110 .246
Non-Hispanic Whites CFI TLI RMSEA

First (i.e., hypothesized)
and final model
.968 .890 .102 .145

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 128
Table 17
Unstandardized Regression Coefficients from Path Models Using Ethnic Group-Specific Norms for Hispanics and the Clinical
Dementia Rating Scale as the Dependent Variable
First Model
Endogenous variable
Independent variable Overall NP FAQ total NPI-Q total IRMD CDR score
Exogenous
Age -0.28
***
0.17
***
-0.01 0.01
***
0.00
Sex 1.77 -- -0.35 -- -0.03
Education 0.83
***
-- -- -- 0.00
Primary language 3.05
**
-- -- -- -0.02
Informant education -- -- -0.01 -0.01 --
Mediating
Overall NP -- -- -- -- -0.02
***

FAQ -- -- -- -- 0.06
***

NPI-Q frequency -- -- -- -- 0.01
IRMD -- -- -- -- 0.60
***

Reduced Model #3
Endogenous variable
Independent variable Overall NP FAQ total NPI-Q total IRMD CDR score
Exogenous
Age -0.29
***
0.17
***
-- 0.01
***
--
Education 0.79
***
-- -- -- --
Primary language 2.91
**
-- -- -- --
Mediating
Overall NP -- -- -- -- -0.02
***

FAQ -- -- -- -- 0.06
***

IRMD -- -- -- -- 0.61
***

Note. NP = neuropsychological test performance (using ethnic group-specific norms); FAQ = Functional Assessment Questionnaire; NPI-Q = Neuropsychiatric
Inventory Questionnaire; IRMD = informant-reported memory decline; CDR = Clinical Dementia Rating Scale.
** p < .01, *** p < .001
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 129
Table 18

Unstandardized Regression Coefficients from Path Models Using Ethnic Group-Specific Norms for Non-Hispanic Whites and the

Clinical Dementia Rating Scale as the Dependent Variable
First Model
Endogenous variable
Independent variable Overall NP FAQ total NPI-Q total IRMD CDR score
Exogenous
Age -0.26
***
0.12
***
-0.01
***
0.01
***
0.00
**

Sex 1.51
***
-- -0.32
***
-- -0.02
*

Education 0.68
***
-- -- -- 0.01
***

Informant education -- -- -0.02
**
-0.01
***
--
Mediating
Overall NP -- -- -- -- -0.02
***

FAQ -- -- -- -- 0.06
***

NPI-Q frequency -- -- -- -- 0.01
***

IRMD -- -- -- -- 0.44
***

Note. NP = neuropsychological test performance (using ethnic group-specific norms); FAQ = Functional Assessment Questionnaire; NPI-Q = Neuropsychiatric
Inventory Questionnaire; IRMD = informant-reported memory decline; CDR = Clinical Dementia Rating Scale.
* p < .05, ** p < .01, *** p < .001

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 130
Table 19
Subsample Descriptive Statistics and P Values for t tests and Chi-square Statistics by Ethnicity and Diagnosis
Informant characteristics
Overall (N = 142),
Mean (SD)
NCF (N = 82),
Mean (SD)
Dementia (N = 60),
Mean (SD)
NCF vs. Dementia,
P value
Age (years)
Hispanic (N = 22) 63.45 (14.98) 66.10 (17.10) 61.25 (13.32) .463
NHW (N = 117) 64.50 (12.67) 65.87 (12.86) 62.39 (12.18) .147
Prob (H
0
) .730 .960 .778 ---
Education (years)
Hispanic (N = 21) 15.10 (2.74) 16.20 (2.86) 14.09 (2.30) .077
NHW (N = 112) 16.07 (2.69) 16.26 (2.96) 15.81 (2.28) .382
Prob (H
0
) .131 .951 .029 ---
Informant characteristics Overall, N (%) NCF, N (%) Dementia, N (%)
NCF vs. Dementia,
P value
Women
Hispanic (N = 22) 15 (68.2) 5 (50.0) 10 (83.3) .095
NHW (N = 120) 82 (68.3) 46 (63.9) 36 (75.0) .200
Prob (H
0
) .989 .396 .542 ---
English primary language
Hispanic (N = 22) 15 (68.2) 6 (60.0) 9 (75.0) .793
NHW (N = 120) 117 (97.5) 67 (93.1) 46 (95.8) .347
Prob (H
0
) <.001 <.001 .005 ---
Born in United States
Hispanic (N = 22) 17 (77.3) 6 (60.0) 11 (91.7) .190
NHW (N = 117) 108 (90.0) 64 (88.9) 44 (91.7) .493
Prob (H
0
) .032 .005 .810 ---
Coresidency with patient
Hispanic (N = 22) 12 (54.4) 5 (50.0) 7 (58.3) .746
NHW (N = 120) 87 (72.5) 47 (65.3) 40 (83.3) .040
Prob (H
0
) .092 .347 .060 ---

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 131
Patient characteristics
Overall,
Mean (SD)
NCF,
Mean (SD)
Dementia,
Mean (SD)
NCF vs. Dementia,
P value
Age (years)
Hispanic (N = 22) 73.91 (9.85) 69.50 (9.76) 77.58 (8.65) .053
NHW (N = 117) 71.70 (9.73) 72.38 (8.93) 70.69 (10.84) .354
Prob (H
0
) .330 .348 .046 ---
Education (years)
Hispanic (N = 21) 14.14 (3.52) 15.70 (3.16) 12.83 (3.38) .055
NHW (N = 112) 16.18 (2.61) 16.46 (2.39) 15.75 (2.87) .145
Prob (H
0
) .002 .370 .004 ---
Patient characteristics Overall, N (%) NCF, N (%) Dementia, N (%)
NCF vs. Dementia,
P value
Women
Hispanic (N = 22) 13 (59.1) 7 (70.0) 6 (50.0) .342
NHW (N = 120) 62 (51.7) 43 (59.7) 19 (39.6) .031
Prob (H
0
) .521 .532 .513 ---
English primary language
Hispanic (N = 22) 19 (86.4) 9 (90.0) 10 (83.3) .650
NHW (N = 120) 119 (99.2) 71 (98.6) 48 (100.0) .412
Prob (H
0
) .001 .098 .004 ---
Informant-reported memory decline
Hispanic (N = 22) 14 (63.6) 2 (20.0) 12 (100.0) <.001
NHW (N = 120) 59 (49.2) 11 (15.3) 48 (100.0) <.001
Prob (H
0
) .240 .731 --- ---
Global CDR Score
Hispanic (N = 22) <.001
0 8 (38.1) 8 (80.0) 0 (0.0) ---
0.5 6 (18.5) 2 (20.0) 4 (33.3) ---
1 5 (25.7) 0 (0.0) 5 (41.7) ---
2 3 (13.1) 0 (0.0) 3 (25.0) ---
3 0 (0.0) 0 (0.0) 0 (0.0) ---

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 132
NHW (N = 120) <.001
0 59 (49.2) 59 (81.9) 0 (0.0) ---
0.5 25 (20.8) 13 (18.1) 12 (25.0) ---
1 30 (25.0) 0 (0.0) 30 (62.5) ---
2 6 (5.0) 0 (0.0) 6 (12.5) ---
3 0 (0.0) 0 (0.0) 0 (0.0) ---
Prob (H
0
) .248 .882 .900 ---
Patient characteristics
Overall,
Mean (SD)
NCF,
Mean (SD)
Dementia,
Mean (SD)
NCF vs. Dementia,
P value
Overall NP, ethnic group-specific
norms (sum of scaled scores)

Hispanic (N = 22) -16.83 (20.14) 0.00 (8.03) -30.85 (15.81) <.001
NHW (N = 107) -16.07 (23.02) 0.16 (8.22) -39.31 (16.43) <.001
Prob (H
0
) .886 .953 .117 ---
Overall NP, combined ethnic-group
norms (sum of scaled scores)

Hispanic (N = 22) -21.03 (18.61) -5.21 (8.85) -34.20 (13.42) <.001
NHW (N = 107) -14.58 (21.99) 0.90 (7.94) -36.75 (15.70) <.001
Prob (H
0
) .202 .029 .610 ---
Total FAQ
Hispanic (N = 22) 10.05 (10.50) 0.20 (0.42) 18.25 (6.98) <.001
NHW (N = 120) 6.58 (9.33) 0.18 (0.57) 16.19 (7.92) <.001
Prob (H
0
) .119 .917 .413 ---
Total NPI-Q Frequency
Hispanic (N = 22) 2.18 (3.00) 1.00 (1.70) 3.17 (3.54) .078
NHW (N = 120) 1.55 (2.27) 0.40 (0.85) 3.27 (2.63) <.001
Prob (H
0
) .257 .301 .909 ---
Total NPI-Q Severity
Hispanic (N = 22) 1.91 (2.25) 0.70 (1.06) 2.92 (2.50) .014
NHW (N = 120) 0.84 (1.49) 0.46 (1.13) 1.42 (1.77) <.001
Prob (H
0
) .042 .524 .019 ---

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 133
Length of cognitive decline (years)
Hispanic (N = 12) --- --- 2.42 (0.70) ---
NHW (N = 47) --- --- 2.46 (0.36) ---
Prob (H
0
) --- --- .444 ---
Informant characteristics
Overall,
Mean (SD)
NCF,
Mean (SD)
Dementia,
Mean (SD)
NCF vs. Dementia,
P value
Total ADKS
Hispanic (N = 22) 23.09 (3.85) 22.20 (4.71) 23.83 (2.98) .334
NHW (N = 120) 24.08 (3.35) 23.50 (3.39) 24.94 (3.11) .020
Prob (H
0
) .218 .283 .272
Total CBADS
Hispanic (N = 22) 2.14 (2.08) 2.40 (2.17) 1.92 (2.07) .599
NHW (N = 119) 1.74 (1.86) 1.74 (1.87) 1.74 (1.91) .993
Prob (H
0
) .366 .305 .775
ARSMA-II raw acculturation score
Hispanic (N = 22) 7.77 (23.77) -5.60 (22.16) 18.92 (19.47) .012
ARSMA-II-Hispanic Orientation
Hispanic (N = 22) 47.70 (18.22) 56.90 (17.52) 40.04 (15.55) .027
ARSMA-II-Anglo Orientation
Hispanic (N = 22) 55.48 (7.83) 51.30 (8.03) 58.96 (5.94) .018
Total Cloze
Hispanic (N = 22) 47.86 (1.52) 47.30 (2.00) 48.33 (0.78) .114
Note. NCF = normal cognitive functioning; NHW = non-Hispanic White; Prob (H
0
) = null hypothesis that group differences are not significant; NP =
neuropsychological test performance; FAQ = Functional Assessment Questionnaire; NPI-Q = Neuropsychiatric Inventory Questionnaire; CDR = Clinical
Dementia Rating Scale; ADKS = Alzheimer’s Disease Knowledge Scale; CBADS = Cultural Beliefs about Alzheimer’s Disease Scale; ARSMA-II =
Acculturation Rating Scale for Mexican Americans-II.

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 134
Table 20
Correlation Coefficients among the Demographic and Key Variables for the Combined Hispanic and Non-Hispanic White Subsample
Variable
Patient
Age
Patient
Sex
Patient
Educ.
Inform.
Educ.
Overall
NP
FAQ
Total
NPI-Q
Freq.
NPI-Q
Severity
IRMD ADKS CBADS Diagnosis
Patient
Age
1.000
Patient
Sex
-.016 1.000
Patient
Educ.
-.020 -.092 1.000
Inform.
Educ.
.104 .166 .217
*
1.000
Overall
NP
-.068 .169 .273
**
.098 1.000
FAQ
Total
.006 -.150 -.191
*
-.124 -.794
**
1.000
NPI-Q
Freq.
-.081 -.204
*
-.100 -.054 -.462
**
.669
**
1.000
NPI-Q
Severity
.060 -.030 -.231
**
-.160 -.273
**
.485
**
.690
**
1.000
IRMD .019 -.290
**
-.133 -.124 -.743
**
.715
**
.501
**
.338
**
1.000
ADKS .082 .049 .038 .180* -.167 .144 .069 .010 .045 1.000
CBADS .008 -.121 .082 -.103 .057 .091 .105 .115 .049 -.170
*
1.000
Diagnosis .002 -.191
*
-.209
*
-.141 -.833
**
.851
**
.574
**
.366
**
.830
**
.199
*
-.011 1.000
Note. NP = neuropsychological test performance (using ethnic group-specific norms); FAQ = Functional Assessment Questionnaire; NPI-Q = Neuropsychiatric
Inventory Questionnaire; IRMD = informant-reported memory decline; ADKS = Alzheimer’s Disease Knowledge Scale; CBADS = Cultural Beliefs about
Alzheimer’s Disease Scale.
* p < .05, ** p < .01
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 135
Table 21
Correlation Coefficients among the Demographic and Key Variables for the Hispanic Subsample
Variable Age Sex
Patient
Educ.
Inform.
Educ.
Overall
NP
FAQ
Total
NPI-Q
Freq.
NPI-Q
Sev.
IRMD ADKS CBADS
ARSMA
-II
Cloze Diag.
Age 1.000

Sex .031 1.000

Patient
Educ.
-.291 -.128 1.000

Inform.
Educ.
.113 .299 .273 1.000

Overall
NP
-.199 .242 .581
**
.373 1.000

FAQ
Total
.273 -.294 -.363 -.517
*
-.813
**
1.000

NPI-Q
Freq.
-.368 -.169 -.106 -.264 -.456
*
.541
**
1.000

NPI-Q
Sev.
-.093 .008 -.239 -.219 -.508
*
.564
**
.879
**
1.000

IRMD .160 -.245 -.464
*
-.528
*
-.702
**
.731
**
.337 .486
*
1.000
ADKS .358 .056 -.109 .465
*
-.066 .056 -.016 .108 .138 1.000
CBADS -.090 -.257 .154 -.279 -.028 .085 .057 -.073 -.050 -.411 1.000
ARSMA
-II
.201 .158 -.272 .111 -.381 .300 .205 .382 .426 .581
**
-.684
**
1.000
Cloze .192 -.270 -.277 .117 -.200 .222 .059 .091 .503
*
.666
**
-.395 .618
**
1.000
Diag. .418 -.203 -.415 -.404 -.781
**
.876
**
.368 .503
*
.828
**
.280 -.016 .482
*
.382 1.000

Note. NP = neuropsychological test performance (using ethnic group-specific norms); FAQ = Functional Assessment Questionnaire; NPI-Q = Neuropsychiatric
Inventory Questionnaire; IRMD = informant-reported memory decline; ADKS = Alzheimer’s Disease Knowledge Scale; CBADS = Cultural Beliefs about Alzheimer’s
Disease Scale; ARMSA-II = Acculturation Rating Scale for Mexican Americans-II raw acculturation score; Diag. = diagnosis.
* p < .05, ** p < .01

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 136
Table 22
Correlation Coefficients among the Demographic and Key Variables for the Non-Hispanic White Subsample
Variable Age Sex
Patient
Educ.
Inform.
Educ.
Overall
NP
FAQ
Total
NPI-Q
Freq.
NPI-Q
Severity
IRMD ADKS CBADS Diagnosis
Age 1.000
Sex -.030 1.000
Patient
Educ.
.071 -.070 1.000
Inform.
Educ.
.116 .150 .166 1.000
Overall
NP
-.043 .158 .212
*
.051 1.000
FAQ
Total
-.061 -.131 -.111 -.032 -.801
**
1.000
NPI-Q
Freq.
-.025 -.222
*
-.069 -.015 -.472
**
.699
**
1.000
NPI-Q
Severity
.080 -.058 -.162 -.115 -.227
*
.448
**
.625
**
1.000
IRMD -.016 -.305
**
-.029 -.039 -.754
**
.709
**
.537
**
.290
**
1.000
ADKS .041 .056 .035 .103 -.191
*
.187
*
.108 .022 .043 1.000
CBADS .017 -.103 .114 -.041 .078 .072 .104 .139 .053 -.097 1.000
Diagnosis -.085 -.197
*
-.134 -.080 -.848
**
.844
**
.621
**
.317
**
.828
**
.202
*
-.028 1.000
Note. NP = neuropsychological test performance (using ethnic group-specific norms); FAQ = Functional Assessment Questionnaire; NPI-Q = Neuropsychiatric
Inventory Questionnaire; IRMD = informant-reported memory decline; ADKS = Alzheimer’s Disease Knowledge Scale; CBADS = Cultural Beliefs about
Alzheimer’s Disease Scale.
* p < .05, ** p < .01

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 137
Figure 1. Proposed path model for Hispanics with neuropsychological test performance, FAQ, NPI-Q, and informant reports of
patients’ memory decline predicting clinicians’ diagnosis of dementia, controlling for patients’ age, sex, education level, and primary
language, and informants’ education level.
Age
Education
Primary
Language
Informant
Education
NP Test
Perform.
FAQ
NPI-Q
Inf.-Report
of Decline
Clinicians'
Diagnosis
e5
e2
e3
e1
Sex
1
1
e4
1
1
1

Note. NP Test Perform. = Standardized overall neuropsychological test performance; FAQ = Functional Assessment Questionnaire; NPI-Q = Neuropsychiatric
Inventory Questionnaire; Inf.-Report of Decline = Informant-reported meaningful decline in patients’ memory (yes or no).

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 138
Figure 2. Proposed path model for non-Hispanic Whites with neuropsychological test performance, FAQ, NPI-Q, and informant
reports of patients’ memory decline predicting clinicians’ diagnosis of dementia, controlling for patients’ age, sex, and education
level, and informants’ education level.

Age
Education
Informant
Education
NP Test
Perform.
FAQ
NPI-Q
Inf.-Report
of Decline
Clinicians'
Diagnosis
e5
e2
e3
e1
Sex
1
1
1
e4
1
1

Note. NP Test Perform. = Standardized overall neuropsychological test performance; FAQ = Functional Assessment Questionnaire; NPI-Q = Neuropsychiatric
Inventory Questionnaire; Inf.-Report of Decline = Informant-reported meaningful decline in patients’ memory (yes or no).
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 139
Figure 3. Proposed path model for Hispanics with neuropsychological test performance, FAQ, NPI-Q, informant reports of memory
decline, and the summed ADKS and CBADS scored predicting clinicians’ diagnosis of dementia, controlling for patients’ age, sex,
education level, and primary language, and informants’ education level.
Age
Education
Inf. Accul-
turation
Inf.
Education
NP Test
Perform.
FAQ
NPI-Q
Inf.-Report
of Decline
Clinicians'
Diagnosis
e6
e3
e4
e2
Sex
1
ADKS +
CBADS
e5
e1
Primary
Language
1
1
1
1
1

Note. Inf. Acculturation = Informants’ level of acculturation as measured by two subscales of Scale 1 of the Acculturation Rating Scale for Mexican Americans-
II (ARSMA-II): Anglo Orientation and Hispanic Orientation; ADKS + CBADS = Alzheimer’s Disease Knowledge Scale and Cultural Beliefs about Alzheimer’s
Disease Scale summed score; NP Test Perform. = Standardized overall neuropsychological test performance; FAQ = Functional Assessment Questionnaire; NPI-
Q = Neuropsychiatric Inventory Questionnaire; Inf.-Report of Decline = Informant-reported meaningful decline in patients’ memory (yes or no).
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 140
Figure 4. Proposed path model for non-Hispanic Whites with neuropsychological test performance, FAQ, NPI-Q, informant reports of
memory decline, the summed ADKS and CBADS scores predicting clinicians’ diagnosis of dementia, controlling for patients’ age,
sex, and education level, and informants’ education level.
Age
Education
Inf.
Education
NP Test
Perform.
FAQ
NPI-Q
Inf.-Report
of Decline
Clinicians'
Diagnosis
e6
e3
e4
e2
Sex
1
1
1
ADKS +
CBADS
e1
1
e5
1
1

Note. Inf. Acculturation = Informants’ level of acculturation as measured by two subscales of Scale 1 of the Acculturation Rating Scale for Mexican Americans-
II (ARSMA-II): Anglo Orientation and Hispanic Orientation; ADKS + CBADS = Alzheimer’s Disease Knowledge Scale and Cultural Beliefs about Alzheimer’s
Disease Scale summed score; NP Test Perform. = Standardized overall neuropsychological test performance; FAQ = Functional Assessment Questionnaire; NPI-
Q = Neuropsychiatric Inventory Questionnaire; Inf.-Report of Decline = Informant-reported meaningful decline in patients’ memory (yes or no).
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 141
Figure 5. Hypothesized one-factor baseline model for the Functional Assessment Questionnaire.

Dementia
Severity
Item 10 err10
1
Item 9 err9
1
Item 8 err8
1
Item 7 err7
1
Item 6 err6
1
Item 5 err5
1
Item 4 err4
1
Item 3 err3
1
Item 2 err2
1
Item 1 err1
1

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 142
Figure 6. Hypothesized four-factor baseline model for the Neuropsychiatric Inventory Questionnaire.
Hyperactivity
Irritability err3
1
Disinhibition err2
1
Agitation err1
1
Affect
Depression err5
1
Anxiety err4
1
Psychosis
Hallucinations err7
1
Delusions err6
1
Apathy/vegetative
symptoms
Appetite err10
1
Sleep err9
1
Apathy err8
1

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 143
Figure 7. First (i.e., hypothesized) path model for Hispanics with neuropsychological test performance (using ethnic group-specific
norms), FAQ, NPI-Q, and informant reports of patients’ memory decline predicting clinicians’ diagnosis of dementia, controlling for
patients’ age, sex, education level, and primary language, and informants’ education level (standardized path coefficients).

Age
Education
Informant
Education
NP Test
Performance
FAQ
NPI-Q
IRMD
Dementia
Diagnosis
-.17
.16
.19
.04
.58
e5
e2
e3
e1
Sex
-.06
.17
-.04
e4
.22
.05
-.17
-.07
-.01
-.47
-.78
.58
.72
-.70
.02
.05
Primary
Language
.08
.57
-.03
-.04
.10
.43
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 144
Figure 8. Reduced path model for Hispanics with neuropsychological test performance (using ethnic group-specific norms), FAQ, and
informant reports of patients’ memory decline predicting clinicians’ diagnosis of dementia, controlling for patients’ age, education
level, and primary language (standardized path coefficients).

Age
Education
NP Test
Performance
FAQ
IRMD
Dementia
Diagnosis
-.18
.16
.18
.60
e4
e2
e1
.17
e3
.21
-.18
-.78
.72
-.70
.05
Primary
Language
.08
.12
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 145
Figure 9. First (i.e., hypothesized) path model for non-Hispanic Whites with neuropsychological test performance (using ethnic group-
specific norms), FAQ, NPI-Q, and informant reports of patients’ memory decline predicting clinicians’ diagnosis of dementia,
controlling for patients’ age and sex and patients’ and informants’ education level (standardized path coefficients).

Age
Education
Informant
Education
NP Test
Performance
FAQ
NPI-Q
IRMD
Dementia
Diagnosis
-.14
.13
.10
.06
.52
e5
e2
e3
e1
Sex
-.03
.11
-.04
e4
.19
.04
-.24
-.06
-.02
.36
-.47
-.78
.59
.67
.51
-.64
-.01
-.02
-.01
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 146
Figure 10. Reduced path model for non-Hispanic Whites with neuropsychological test performance (using ethnic group-specific
norms), FAQ, NPI-Q, and informant reports of patients’ memory decline predicting clinicians’ diagnosis of dementia, controlling for
patients’ age and sex and patients’ and informants’ education level (standardized path coefficients).

Age
Education
Informant
Education
NP Test
Performance
FAQ
NPI-Q
IRMD
Dementia
Diagnosis
-.14
.13
.10
.06
.52
e5
e2
e3
e1
Sex
-.03
.11
-.04
e4
.19
.04
-.24
-.06
-.02
.36
-.47
-.78
.59
.67
.51
-.64
-.01
-.02
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 147
Figure 11. First (i.e., hypothesized) path model for Hispanics with neuropsychological test performance (using combined ethnic group
norms), FAQ, NPI-Q, and informant reports of patients’ memory decline predicting clinicians’ diagnosis of dementia, controlling for
patients’ age, sex, education level, and primary language, and informants’ education level (standardized path coefficients).

Age
Education
Informant
Education
NP Test
Performance
FAQ
NPI-Q
IRMD
Dementia
Diagnosis
-.20
.16
.21
.04
.54
e5
e2
e3
e1
Sex
-.05
.17
-.04
e4
.21
.10
-.22
-.07
.00
-.48
-.75
.58
.72
-.73
.01
.07
Primary
Language
.09
.57
-.02
-.03
.10
.43
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 148
Figure 12. Reduced path model for Hispanics with neuropsychological test performance (using combined ethnic group norms), FAQ,
and informant reports of patients’ memory decline predicting clinicians’ diagnosis of dementia, controlling for patients’ age, sex,
education level, and primary language (standardized path coefficients).

Age
Education
NP Test
Performance
FAQ
IRMD
Dementia
Diagnosis
-.20
.16
.20
.55
e4
e2
e1
Sex
.17
e3
.21
.10
-.24
-.75
.72
-.73
.07
Primary
Language
.09
.12
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 149
Figure 13. First (i.e., hypothesized) path model for non-Hispanic Whites with neuropsychological test performance (using combined
ethnic group norms), FAQ, NPI-Q, and informant reports of patients’ memory decline predicting clinicians’ diagnosis of dementia,
controlling for patients’ age and sex and patients’ and informants’ education level (standardized path coefficients).

Age
Education
Informant
Education
NP Test
Performance
FAQ
NPI-Q
IRMD
Dementia
Diagnosis
-.19
.13
.14
.06
.47
e5
e2
e3
e1
Sex
-.02
.11
-.04
e4
.16
.05
-.34
-.06
-.02
.36
-.48
-.77
.59
.67
.51
-.68
-.01
-.04
.02
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 150
Figure 14. Reduced path model for non-Hispanic Whites with neuropsychological test performance (using combined ethnic group
norms), FAQ, NPI-Q, and informant reports of patients’ memory decline predicting clinicians’ diagnosis of dementia, controlling for
patients’ age and sex and patients’ and informants’ education level (standardized path coefficients).

Age
Education
Informant
Education
NP Test
Performance
FAQ
NPI-Q
IRMD
Dementia
Diagnosis
-.19
.13
.14
.06
.47
e5
e2
e3
e1
Sex
-.02
.11
-.04
e4
.15
.05
-.34
-.06
-.02
.36
-.48
-.77
.59
.67
.51
-.68
-.04
.02
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 151
Figure 15. First (i.e., hypothesized) path model for Hispanics with neuropsychological test performance (using ethnic group-specific
norms), FAQ, NPI-Q, and informant reports of patients’ memory decline predicting Clinical Dementia Rating Scale scores, controlling
for patients’ age, sex, education level, and primary language, and informants’ education level (standardized path coefficients).

Age
Education
Informant
Education
NP Test
Performance
FAQ
NPI-Q
IRMD
CDR
-.17
.16
.19
.01
.24
e5
e2
e3
e1
Sex
-.06
.16
-.04
e4
.55
.05
-.21
-.07
-.01
-.47
-.78
.58
.72
-.69
.00
-.01
Primary
Language
.08
.57
-.01
-.01
.10
.43
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 152
Figure 16. Reduced path model for Hispanics with neuropsychological test performance (using ethnic group-specific norms), FAQ,
and informant reports of patients’ memory decline predicting Clinical Dementia Rating Scale scores, controlling for patients’ age,
education level, and primary language (standardized path coefficients).

Age
Education
NP Test
Performance
FAQ
IRMD
CDR
-.18
.16
.18
.25
e4
e2
e1
.17
e3
.54
-.21
-.78
.72
-.70
Primary
Language
.08
.12
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 153
Figure 17. First (i.e., hypothesized) and final path model for non-Hispanic Whites with neuropsychological test performance (using
ethnic group-specific norms), FAQ, NPI-Q, and informant reports of patients’ memory decline predicting Clinical Dementia Rating
Scale scores, controlling for patients’ age and sex and patients’ and informants’ education level (standardized path coefficients).

Age
Education
Informant
Education
NP Test
Performance
FAQ
NPI-Q
IRMD
CDR
-.14
.13
.10
.03
.19
e5
e2
e3
e1
Sex
-.03
.11
-.04
e4
.53
.04
-.27
-.06
-.02
.36
-.47
-.78
.59
.67
.51
-.63
-.01
-.01
.02
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 154
Figure 18. Reduced path model for Hispanics with neuropsychological test performance (using combined ethnic group norms), FAQ,
and informant reports of patients’ memory decline predicting Clinical Dementia Rating Scale scores, controlling for patients’ age, sex,
education level, and primary language (standardized path coefficients).

Age
Education
NP Test
Performance
FAQ
IRMD
CDR
-.19
.16
.20
.25
e4
e2
e1
Sex
.17
e3
.58
.10
-.16
-.78
.72
-.72
Primary
Language
.09
.12
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 155
Figure 19. Reduced path model for non-Hispanic Whites with neuropsychological test performance (using combined ethnic group
norms), FAQ, NPI-Q, and informant reports of patients’ memory decline predicting Clinical Dementia Rating Scale scores, controlling
for patients’ age and sex, and patients’ and informants’ education level (standardized path coefficients).

Age
Education
Informant
Education
NP Test
Performance
FAQ
NPI-Q
IRMD
CDR
-.18
.13
.13
.03
.17
e5
e2
e3
e1
Sex
-.02
.11
-.04
e4
.52
.04
-.29
-.06
-.02
.36
-.49
-.80
.59
.67
.51
-.67
-.02
.03
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 156
Appendix
Table 23
The Functional Assessment Questionnaire (FAQ)

Instructions (for clinician or other trained health professional per informant interview): In the past four weeks, did the subject have
any difficulty or need help with:
Not
applicable
(e.g., never
did) Normal
Has
difficulty,
but does
by self
Requires
assistance Dependent
1. Writing checks, paying bills, or balancing a checkbook. -- 0 1 2 3
2. Assembling tax records, business affairs, or other papers. -- 0 1 2 3
3. Shopping alone for clothes, household necessities, or
groceries.
-- 0 1 2 3
4. Playing a game of skill such as bridge or chess, working
on a hobby.
-- 0 1 2 3
5. Heating water, making a cup of coffee, turning off the
stove.
-- 0 1 2 3
6. Preparing a balanced meal. -- 0 1 2 3
7. Keeping track of current events. -- 0 1 2 3
8. Paying attention to and understanding a TV program,
book, or magazine.
-- 0 1 2 3
9. Remembering appointments, family occasions, holidays,
medications.
-- 0 1 2 3
10. Traveling out of the neighborhood, driving, or arranging
to take public transportation.
-- 0 1 2 3
Sources. Pfeffer, R. I., Kurosaki, T. T., Harrah, C., Chance, J. M., Bates, D., Detels, R., et al. (1981). A survey diagnostic tool for senile dementia. American
Journal of Epidemiology, 114, 515-527.
Pfeffer, R. I., Kurosaki, T. T., Harrah, C. H., Chance, J. M., & Filos, S. (1982). Measurement of functional activities in older adults in the community. Journal of
Gerontology, 37, 323-329.

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 157
Table 24
The Neuropsychiatric Inventory Questionnaire (NPI-Q)

Instructions (for clinician or other trained health professional per informant interview): Please ask the following questions based
upon changes. Indicate “yes” only if the symptom has been present in the past month; otherwise, indicate “no.” For each item marked
“yes,” rate the SEVERITY of the symptom (how it affects the patient): 1 = Mild (noticeable, but not a significant change); 2 =
Moderate (significant, but not a dramatic change); 3 = Severe (very marked or prominent; a dramatic change).
Yes No Severity
1. Delusions: Does the patient believe that others are stealing from him or her, or are planning to harm him or
her in some way?
1 0 1 2 3
2. Hallucinations: Does the patient act as if he or she hears voices? Does he or she talk to people who are not
there?
1 0 1 2 3
3. Agitation or aggression: Is the patient stubborn and resistive to help from others? 1 0 1 2 3
4. Depression or dysphoria: Does the patient act as if he or she is sad or in low spirits? Does he or she cry? 1 0 1 2 3
5. Anxiety: Does the patient become upset when separated from you? Does he or she have any other signs of
nervousness, such as shortness of breath, sighing, being unable to relax, or feeling excessively tense?
1 0 1 2 3
6. Elation or euphoria: Does the patient appear to feel too good or act excessively happy? 1 0 1 2 3
7. Apathy or indifference: Does the patient seem less interested in his or her usual activities and in the
activities and plans of others?
1 0 1 2 3
8. Disinhibition: Does the patient seem to act impulsively? For example, does the patient talk to strangers as if
he or she knows them, or does the patient say things that may hurt people’s feelings?
1 0 1 2 3
9. Irritability or lability: Is it patient impatient or cranky? Does he or she have difficulty coping with delays or
waiting for planned activities?
1 0 1 2 3
10. Motor disturbance: Does the patient engage in repetitive activities, such as pacing around the house,
handling buttons, wrapping string, or doing other things repeatedly?
1 0 1 2 3
11. Nighttime behavior: Does the patient awaken you during the night, rise too early in the morning, or take
excessive naps during the day?
1 0 1 2 3
12. Appetite and eating: Has the patient lost or gained weight, or had a change in the food he or she likes? 1 0 1 2 3
Source. Kaufer, D. I., Cummings, J. L., Ketchel, P., Smith, V., MacMillan, A., Shelley, T., et al. (2000). Validation of the NPI-Q, a brief clinical form of the
Neuropsychiatric Inventory. The Journal of Neuropsychiatry and Clinical Neurosciences, 12, 233-239.
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 158
Table 25
Alzheimer’s Disease Knowledge Scale (ADKS)
Instructions: Below are some statements about Alzheimer’s disease. Please read each statement carefully and circle whether you think
the statement is True or False. If you are not sure of the correct answer, please just make your best guess and circle either True or
False. It is important to circle an answer for every statement, even if you are not completely sure of the correct answer.
True / False 1. People with Alzheimer’s disease are particularly prone to depression.
True / False 2. It has been scientifically proven that mental exercise can prevent a person from getting Alzheimer’s disease.
True / False 3. After symptoms of Alzheimer’s disease appear, the average life expectancy is 6 to 12 years.
True / False 4. When a person with Alzheimer’s disease becomes agitated, a medical examination might reveal other health
problems that caused the agitation.
True / False 5. People with Alzheimer’s disease do best with simple, instructions given one step at a time.
True / False 6. When people with Alzheimer’s disease begin to have difficulty taking care of themselves, caregivers should
take over right away.
True / False 7. If a person with Alzheimer’s disease becomes alert and agitated at night, a good strategy is to try to make sure
that the person gets plenty of physical activity during the day.
True / False 8. In rare cases, people have recovered from Alzheimer’s disease.
True / False 9. People whose Alzheimer’s disease is not yet severe can benefit from psychotherapy for depression and anxiety.
True / False 10. If trouble with memory and confused thinking appears suddenly, it is likely due to Alzheimer’s disease.
True / False 11. Most people with Alzheimer’s disease live in nursing homes.
True / False 12. Poor nutrition can make the symptoms of Alzheimer’s disease worse.
True / False 13. People in their 30s can have Alzheimer’s disease.
True / False 14. A person with Alzheimer’s disease becomes increasingly likely to fall down as the disease gets worse.
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 159
True / False 15. When people with Alzheimer’s disease repeat the same question or story several times, it is helpful to remind
them that they are repeating themselves.
True / False 16. Once people have Alzheimer’s disease, they are no longer capable of making informed decisions about their
own care.
True / False 17. Eventually, a person with Alzheimer’s disease will need 24-hour supervision.
True / False 18. Having high cholesterol may increase a person’s risk of developing Alzheimer’s disease.
True / False 19. Tremor or shaking of the hands or arms is a common symptom in people with Alzheimer’s disease.
True / False 20. Symptoms of severe depression can be mistaken for symptoms of Alzheimer’s disease.
True / False 21. Alzheimer’s disease is one type of dementia.
True / False 22. Trouble handling money or paying bills is a common early symptom of Alzheimer’s disease.
True / False 23. One symptom that can occur with Alzheimer’s disease is believing that other people are stealing one’s things.
True / False 24. When a person has Alzheimer’s disease, using reminder notes is a crutch that can contribute to decline.
True / False 25. Prescription drugs that prevent Alzheimer’s disease are available.
True / False 26. Having high blood pressure may increase a person’s risk of developing Alzheimer’s disease.
True / False 27. Genes can only partially account for the development of Alzheimer’s disease.
True / False 28. It is safe for people with Alzheimer’s disease to drive, as long as they have a companion in the car at all times.
True / False 29. Alzheimer’s disease cannot be cured.
True / False 30. Most people with Alzheimer’s disease remember recent events better than things that happened in the past.
Source. Carpenter, B. D., Balsis, S., Otilingam, P. G., Hanson, P. K., & Gatz, M. (2009). The Alzheimer’s Disease Knowledge Scale: Development and
psychometric properties. The Gerontologist, 49, 236-247.

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 160
Table 26
Cultural Beliefs about Alzheimer’s Disease Scale (CBADS)
Instructions: Below are some statements about Alzheimer’s disease. Please read each statement carefully and circle whether you think
the statement is either True or False. If you are not sure of the correct answer, please just make your best guess and circle either True
or False. It is important to circle an answer for every statement even if you are not completely sure of the correct answer.
True / False 1. Losing your memory as you grow older is a normal part of aging.
True / False 2. Alzheimer’s disease is due to having bad blood.
True / False 3. A person can curse someone with Alzheimer’s disease if they dislike that person.
True / False 4. Alzheimer’s disease is embarrassing for the relatives of the person with the disease.
True / False 5. There is no reason to bring someone with Alzheimer’s disease to a doctor because they cannot do anything to
help.
True / False 6. Alzheimer’s disease is a form of insanity.
True / False 7. Having a relative who has Alzheimer’s disease is shameful for the family.
True / False 8. Alzheimer’s disease is due to a curse being placed on a person by someone else, or “the evil eye” (“mal de
ojo”).
True / False 9. Sometimes, the brain dries up as people grow older, which causes Alzheimer’s disease.
True / False 10. Alzheimer’s disease is a mental illness rather than a physical illness.
True / False 11. Only God can change or cure a person who has Alzheimer’s disease.
True / False 12. Alzheimer’s disease is a form of craziness.
True / False 13. Alzheimer’s disease can be caused by having experienced a difficult life, such as traumatic events or the death
of loved ones.
True / False 14. If someone develops Alzheimer’s disease, it is because it is God’s will.
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 161
True / False 15. All elderly people eventually suffer from serious memory loss.
True / False 16. Committing wrongs in one’s life can cause someone to have Alzheimer’s disease in later life.
True / False 17. People who have Alzheimer’s disease bring dishonor to their relatives.
True / False 18. There are currently no medical treatments such as medications for Alzheimer’s disease.
True / False 19. Alzheimer’s disease is a psychological disorder rather than a biological disease.
True / False 20. Alzheimer’s disease is the same thing as senility.
True / False 21. If someone is jealous of another person, he or she can wish ill will on the other person so that they develop
Alzheimer’s disease.
True / False 22. Alzheimer’s disease is a punishment from God for one’s past sins.
True / False 23. A stressful life can cause someone to develop Alzheimer’s disease in later life.
True / False 24. All people with Alzheimer’s disease exhibit childlike behaviors.
True / False 25. People with Alzheimer’s disease should be cared for by family members rather than outside of the home such
as in nursing homes.
True / False 26. Family members of people with Alzheimer’s disease should not tell others that their relative has Alzheimer’s
disease.

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 162
Table 27
Scale 1 of the Acculturation Rating Scale for Mexican Americans-II (ARSMA-II): Anglo Orientation and Mexican Orientation (with
the term Mexican substituted with the appropriate country or countries of origin)
Instructions: Circle a number between 1 and 5 next to each item that best applies.
1 2 3 4 5
Not at all Very little or
not very
often
Moderately Much or
very often
Extremely
often or
almost
always
1. I speak Spanish 1 2 3 4 5
2. I speak English. 1 2 3 4 5
3. I enjoy speaking Spanish. 1 2 3 4 5
4. I associate with Anglos. 1 2 3 4 5
5. I associate with Mexicans and/or Mexican Americans. 1 2 3 4 5
6. I enjoy listening to Spanish language music. 1 2 3 4 5
7. I enjoy listening to English language music. 1 2 3 4 5
8. I enjoy Spanish language TV. 1 2 3 4 5
9. I enjoy English language TV. 1 2 3 4 5
10. I enjoy English language movies. 1 2 3 4 5
11. I enjoy Spanish language movies. 1 2 3 4 5
12. I enjoy reading (e.g., books in Spanish). 1 2 3 4 5
13. I enjoy reading (e.g., books in English). 1 2 3 4 5
14. I write (e.g., letters in Spanish). 1 2 3 4 5
15. I write (e.g., letters in English). 1 2 3 4 5

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 163
1 2 3 4 5
Not at all Very little or
not very
often
Moderately Much or
very often
Extremely
often or
almost
always
16. My thinking is done in the English language. 1 2 3 4 5
17. My thinking is done in the Spanish language. 1 2 3 4 5
18. My contact with Mexico has been: 1 2 3 4 5
19. My contact with the USA has been: 1 2 3 4 5
20. My father identifies or identified himself as
‘Mexicano.’
1 2 3 4 5
21. My mother identifies or identified himself as
‘Mexicana.’
1 2 3 4 5
22. My friends, while I was growing up, were of
Mexican origin.
1 2 3 4 5
23. My friends, while I was growing up, were of
Anglo origin.
1 2 3 4 5
24. My family cooks Mexican foods. 1 2 3 4 5
25. My friends now are of Anglo origin. 1 2 3 4 5
26. My friends now are of Mexican origin. 1 2 3 4 5
27. I like to identify myself as an Anglo American. 1 2 3 4 5
28. I like to identify myself as a Mexican American. 1 2 3 4 5
29. I like to identify myself as a Mexican. 1 2 3 4 5
30. I like to identify myself as an American. 1 2 3 4 5
Source. Cuellar, J., Arnold, B., & Maldonado, R. (1995). Acculturation Rating Scale for Mexican-Americans-II: A revision of the original ARSMA scale.
Hispanic Journal of Behavioral Sciences, 17, 275-304.
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 164
Table 28

The Cloze Method: 49 Sentence Stems as a Measure of Literacy

Instructions: Please write in the correct word that belongs at the end of each of the following sentences. If you are not sure which
word belongs, just take your best guess. Please only write one response for each item.
Sentence stem Your response
1. She could tell he was mad by the tone of his

2. She went to the bakery for a loaf of

3. Bob proposed and gave her a diamond

4. The dentist recommends brushing your teeth twice a

5. He loosened the tie around his

6. Dan was asked to be the new coach of the

7. They paid for their meals but forgot to leave a

8. To pay for tuition she took out two student

9. She didn’t have her watch so she asked for the

10. Sherry had to read lips because she was

11. They sat together without speaking a single

12. After hitting the iceberg the ship began to

13. She went to the beauty parlor to perm her

14. Without her sunglasses the sun hurt Erika’s

15. Joe did not like his outfit and decided to

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 165
16. The limping horse was obviously in much

17. To prevent football injury all players must wear shoulder

18. To hang the picture Ted needed a hammer and

19. After the argument Ann went to her room and slammed the

20. Father carved the turkey with a

21. Water and sunshine help plants

22. When the two met, one of them held out his

23. He mailed the letter without a

24. She wore a colorful scarf around her

25. The cheap pen ran quickly out of

26. He had a long day and was in a bad

27. She graduated at the top of her

28. The athlete enjoying lifting weights at the

29. Expecting Jeff’s call she waited for the phone to

30. When she got out of the car she closed the

31. The package was sent through the

32. In the shower he washed his skin with

33. You would need a raincoat to avoid getting

34. They turned in their project on the date it was

CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 166
35. She lied about losing her report card to hide her bad

36. For his date Tom bought a long stemmed

37. The genie promised the man he would grant one

38. One year after her death Bill visited his mother’s

39. The children went outside to

40. The teacher wrote the problem on the

41. At night the old woman locked the

42. Her job was easy most of the

43. When you go to bed turn off the

44. After dinner he washed his hands with

45. In the quiet movie theater, Kim’s phone

46. Spring was Jo’s favorite season of the

47. It was dark in the room so she turned on the

48. The kitten played with the ball of

49. The farmer spend the morning milking his

Source. Block, C. K., & Baldwin, C. L. (2010). Cloze probability and completion norms for 498 sentences: Behavioral and neural validation using event-related
potentials. Behavior Research Methods, 42, 665-670.
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 167
Table 29

Clinical Dementia Rating (CDR) Scale
Impairment
Please enter
scores below
None
0
Questionable
0.5
Mild
1
Moderate
2
Severe
3
1. Memory

__.__
No memory loss, or
slight inconsistent
forgetfulness.
Consistent slight
forgetfulness;
partial recollection
of events; “benign”
forgetfulness.
Moderate memory
loss; more marked
for recent events;
defect interferes
with everyday
activities.
Severe memory
loss; only highly
learned material
retained; new
material rapidly
lost.
Severe memory
loss; only
fragments remain.
2. Orientation

__.__
Fully oriented. Fully oriented
except for slight
difficulty with time
relationships.
Moderate difficulty
with time
relationships;
oriented for place at
examination; may
have geographic
disorientation
elsewhere.
Severe difficulty
with time
relationships;
usually disoriented
to time, often to
place.
Oriented to person
only.
3. Judgment &
Problem Solving

__.__
Solves everyday
problems; handles
business and
financial affairs
well; judgment good
in relation to past
performance.
Slight impairment
in solving
problems,
similarities, and
differences.
Moderate difficulty
in handling
problems,
similarities, and
differences; social
judgment usually
maintained.
Severely impaired
in handling
problems,
similarities, and
differences; social
judgment usually
impaired.
Unable to make
judgments or solve
problems.
CROSSCULTURAL DIFFERENCES IN DEMENTIA DIAGNOSIS 168

4. Community
Affairs

__.__
Independent
function at usual
level in job,
shopping, volunteer,
and social groups.
Slight impairment
in these activities.
Unable to function
independently at
these activities,
although may still
be engaged in
some; appears
normal to casual
inspection.
No pretense of
independent
function outside the
home; appears well
enough to be taken
to functions outside
the family home.
No pretense of
independent
function outside
the home; appears
too ill to be taken
to functions
outside the family
home.
5. Home & Hobbies

__.__
Life at home,
hobbies, and
intellectual interests
well maintained.
Life at home,
hobbies, and
intellectual interests
slightly impaired.
Mild but definite
impairment of
function at home;
more difficult
chores abandoned;
more complicated
hobbies and
interests
abandoned.
Only simple chores
preserved; very
restricted interests,
poorly maintained.
No significant
function in the
home.
6. Personal Care

__.__
Fully capable of self-care (= 0). Needs prompting. Requires assistance
in dressing,
hygiene, keeping of
personal effects.
Requires much
help with personal
care; frequent
incontinence.
7. __ __.__ Standard CDR Sum of Boxes
8. __.__ Standard Global CDR
Source. Morris, J. C. (1993). The Clinical Dementia Rating (CDR): Current version and scoring rules. Neurology, 43, 2412–2414.

Abstract (if available)

Abstract Crosscultural differences in dementia diagnosis and care-seeking merit more attention as the booming older adult population becomes increasingly diverse. I aimed to examine differences in clinicians’ assessment of dementia as well as diagnostic delays and impairment levels across 444 Hispanic and 11,081 non-Hispanic White (NHW) outpatients diagnosed with normal cognition or dementia at their initial Alzheimer’s Disease Research Centers evaluations. Results revealed that informant reports of patients’ behavioral and psychological symptoms were significantly associated with diagnosis in NHWs only. Informant-reported functional abilities may be more strongly related to diagnosis in Hispanics as opposed to neuropsychological test performance (NP) in NHWs. I also found evidence for the crosscultural factorial invariance of these two informant-report scales and scalar invariance for one of them. Additionally, Hispanics with dementia were significantly more impaired in NP than their NHW counterparts when using combined ethnic-group norms, but the opposite pattern emerged with ethnic group-specific norms. Finally, in a subsample of 22 Hispanic informants, lower literacy and education levels were associated with less accurate Alzheimer’s disease (AD) knowledge, whereas stronger Hispanic acculturation levels were related to more cultural beliefs about AD. In a subsample of 120 NHW informants, higher AD knowledge levels were related to poorer NP in patients. In sum, these findings bear importance regarding crosscultural diagnostic validity as clinicians may weigh certain NP tests and informant reports differentially across Hispanics and NHWs during the diagnostic process. Moreover, both informant scales can be used meaningfully among these groups, as they have similar factor structures and loadings across cultures. Finally, higher AD knowledge levels among NHW informants may be related to patients’ poorer NP perhaps because informants learn more about dementia as patients’ cognitive symptoms progress.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Affective neuropsychiatric symptoms and neural connectivity in the early stages of Alzheimer’s disease

PDF

Accuray of subjective cognitive complaints in a longitudinal context: the effect of depression and dementia status

PDF

Adoption and implementation of innovative diagnostic tools for Alzheimer's Disease: challenges and barriers in primary care

PDF

Disparities in exposure to traffic-related pollution sources by self-identified and ancestral Hispanic descent in participants of the USC Children’s Health Study

PDF

Disease modifying treatments for Alzeimer's disease: modeling, value assessment and eligible patients

PDF

A series of longitudinal analyses of patient reported outcomes to further the understanding of care-management of comorbid diabetes and depression in a safety-net healthcare system

PDF

Carcinogenic exposures in racial/ethnic groups

Asset Metadata

Creator Sayegh, Philip S. (author)

Core Title Crosscultural differences in dementia diagnosis and care-seeking in Hispanic and non-Hispanic white outpatients

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Psychology

Publication Date 04/17/2014

Defense Date 07/24/2012

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Alzheimer's disease,assessment,confirmatory factor analysis,crosscultural differences,dementia,Diagnosis,Ethnicity,FAQ,Hispanic,informant reports,Latino/a,measurement invariance,neuropsychology,NPI-Q,OAI-PMH Harvest

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Knight, Bob G. (committee chair), McCleary, Carol (committee member), Monterosso, John R. (committee member), Silverstein, Merril (committee member)

Creator Email psayegh@gmail.com,psayegh@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-104895

Unique identifier UC11290304

Identifier usctheses-c3-104895 (legacy record id)

Legacy Identifier etd-SayeghPhil-1255.pdf

Dmrecord 104895

Document Type Dissertation

Rights Sayegh, Philip S.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA