Language Dominance and Education Considerations in the Neuropsychological Assessment of Southwestern American Indians Using the National Alzheimer Coordinating Center’s Uniform Data Set Version 3

Abstract

To address disparities in dementia diagnosis and care in American Indian and Alaska Native communities, it is crucial to understand how sociocultural factors, such as language dominance and education, impact performances on standardized neuropsychological assessments. We discuss sociocultural considerations that are important to consider when evaluating cognition in American Indians. We conducted t tests/Kruskal-Wallis tests and correlation analyses to evaluate the impact of language and education factors on performances on the National Alzheimer Coordinating Center’s Uniform Data Set Version 3 Neuropsychological assessments in a community of Southwestern American Indians. There were no significant differences in cognitive performances between the Zuni (Shiwi)-dominant and English-dominant individuals. Number of years of education had a greater effect on cognitive performances relative to language dominance, particularly for the common cognitive screening measure, the Montreal Cognitive Assessment. Our results highlight that education factors have a greater effect on cognitive performances relative to language dominance in this unique cohort. The associations with the Montreal Cognitive Assessment raise concerns for the use of this tool in this population, highlighting a need to develop culturally appropriate cognitive testing tools as well as ensuring comprehensive, culturally competent neuropsychological assessments are accessible.

Estimates of Alzheimer disease (AD) and related dementias (ADRD) in American Indian (AI) and Alaska Native (AN; AI/AN) communities vary, though increased incidence of both early-onset AD and AD in AI/ANs has been observed.^1–3 It is likely that the rates of ADRD in AI/AN may be even higher given persisting barriers to dementia diagnosis⁴ and increased rates of ADRD risk factors, including hypertension, cardiovascular disease, and diabetes. Yet, AI/ANs continue to be underrepresented in national studies of ADRD and studies focusing on AI/ANs often are limited in sample size and from single communities. As part of the New Mexico Exploratory Alzheimer Disease Research Center (ADRC), we aim to address barriers to dementia diagnosis and care in the AI communities of New Mexico in the context of an ADRC. In this manuscript, as a first step, we evaluate the patterns of bilingualism and education in our Zuni study participants and whether these factors impact performance on neuropsychological tests.

Enormous complexity exists in the interaction of sociocultural factors and performance on standardized assessments. Most standardized neuropsychological tests are developed for and normalized on White, well- and mainstream-educated, native English speaking, and middle- to upper-class samples⁵ and their appropriate use with racial/ethnic minorities is challenging.^6,7 Sociocultural factors including but not limited to education, language, and socioeconomic status may impact performances.^7–10 Importantly, failure to incorporate the impact of these factors on cognitive testing can result in misdiagnosis of mild cognitive impairment and dementia in these communities. For instance, the Montreal Cognitive Assessment (MoCA)¹¹ is one of the most widely used cognitive screening measures in primary care and neurology clinics. This measure is used to identify who may require additional neuropsychological assessment to diagnose those with mild cognitive impairment (MCI). In primarily non-Hispanic White cohorts, up to 80% of individuals with MCI progress to dementia within 5 years, providing an opportunity to identify individuals early in the disease process, an ideal opportunity for intervention.¹² However, the proposed cutoff score of < 26 as suggestive of MCI has led to a high rate of misdiagnosis in community-dwelling diverse samples.¹³ Further, it is likely that sociocultural factors influence performance on individual neuropsychological assessments and it is unknown to what extent language and education influence performances on individual cognitive assessments.

Systematic evaluation of the individual’s linguistic proficiency in all languages spoken and education factors is necessary to ensure appropriate test interpretation. Language proficiency can be captured through self-reported questionnaires as well as objective neuropsychological assessment.¹⁴ Several measures have been used for objective assessment, including timed lexical retrieval tests to develop a language dominance index.¹⁵ Specifically, by administering the well-known phonemic verbal fluency task, Controlled Oral Word Association Test (COWAT),¹⁶ in 2 languages (most commonly Spanish and English),^{17 a} ratio of English words to total words produced in both languages can be calculated and serve as the language dominance index. The Zuni Pueblo has a unique, primarily spoken language, referred to as Zuni or Shiwi’ma, a language isolate that has no known relationship to other Native American languages. Unlike many indigenous languages, Zuni is still spoken by a significant proportion of the population, though it is common that individuals learn English from a young age. Phonemic fluency assessment is not appropriate within the Zuni language given its limited written use. To overcome this, we use a semantic retrieval test administered in English and Zuni as a potential means to quantify a language dominance index in this unique cohort, in addition to self-reported current language use and age at which each language was learned. As education can serve as a neuroprotective factor,^18,19 we also evaluated the impact of education in years.

This study was conducted in close partnership with the Zuni Tribe and included approval by the Tribe at each stage of the study. We evaluate the degree of bilingualism present in our initial cohort and whether the degree of bilingualism and education impacts neuropsychological performances.

METHODS

Participants

The current cohort of Zuni Indians was recruited either as part of a larger project examining diabetes and kidney functioning²⁰ or from the broader community. Zuni Community Health Representatives (CHRs) used the Zuni Health Initiative (ZHI) project’s clinical database for American Indian Chronic Renal Insufficiency Cohort (AICRIC) to identify potential participants. Second, individuals were recruited through visits by CHRs in Zuni households, presentations at tribal health programs and at the health care center, distribution of flyers at local businesses and the civic center, and through other health programs. The inclusion/exclusion criteria for the ancillary study included the following: between the ages of 25 and 80 years, people living with diabetes for longer than 5 years, HbA1c > 7.0% or FPG 126 mg/dL or random glucose > 200 mg/dL, and microalbuminuria. Exclusion criteria included: active infection /inflammation (AIDS, active hepatitis B), malignancy, history of chronic inflammatory disease (eg, lupus, rheumatoid arthritis), severe malnutrition (serum albumin < 2.5 mg/dL and BMI < 18), pregnancy, liver dysfunction, severe congestive heart failure, on experimental drug protocols, exclusion for MRI (pacemakers, metal implants, claustrophobia, etc.), self-reported history of stroke, seizures, traumatic brain injury, or other major brain illness, current antipsychotic or antiepileptic medications, severe visual or hearing impairment that would interfere with completion of the neuropsychological test battery, or current incarceration. The inclusion criteria for participants in the feasibility study conducted as part of the exploratory Alzheimer Disease Research Center are 50 years of age and older. Exclusion criteria include: inability to complete neuropsychological testing, or severe visual or hearing impairment that would prevent them from being able to complete forms or cognitive tests, inability to consent, current prisoner, and monolingual non-English speaking. These studies were approved by the University of New Mexico Health Sciences Center Human Research Review Committee. We adhered to the Declaration of Helsinki and all participants provided written consent and received monetary compensation as appreciation for sharing their time and expertise. The research team also gained approval from the Zuni tribal council for this research before funding and the tribe was consulted throughout the project (eg, evaluation of study procedures by the tribal advisory board, presentation of findings, including this manuscript, were approved by the tribal council before submission).

Neuropsychological Assessment

Participants underwent neuropsychological assessment, including the Version 3 of the Alzheimer Disease Centers’ Neuropsychological Test Battery in the Uniform Data Set²¹, which includes the following tests: Montreal Cognitive Assessment (MoCA; Vancouver Island Coastal First Nations version),^22,23 Craft Story 21 (immediate recall: Craft-I; delayed recall: Craft-D), Benson complex figure copy (Benson-I) and delayed recall (Benson-D), Number span test forwards (NS-F) and backwards (NS-B), Multilingual naming test (MINT), Phonemic test (F and L words; Fluency F+L), Animals and Vegetables List generation, and Trail making test parts A (Trails-A) and B (Trails-B). The administrator was Zuni-English bilingual and would administer the tests in English, though if there was any confusion regarding instructions, would clarify in Zuni as necessary. Verbal responses were recorded to facilitate independent scoring by 2 separate raters. A third rater conducted conflict resolutions whenever there was a discrepancy. Age-corrected normative scores were obtained and converted to t-scores for all subsequent analyses using published regression-based normative data.²¹

Bilingualism Measures

We assessed language proficiency by self-report and objective assessment. Participants were first asked about self-report measures of language dominance, eg, which primary and secondary languages spoken and written, age at which individuals learned Zuni and English, and current percentage of time that the Zuni and English languages are used in daily life. Objective assessment included the administration of the semantic fluency subtest of the COWAT¹⁶, first asking the participants to name animals and vegetables in English and then in Zuni. The language dominance index was quantified following prior approaches¹⁵ as follows with the subtext corresponding to the language the test was administered in and the participant’s response in the corresponding language:

Language dominance index = (Animals_English + Vegetables_English)/(Animals_English + Vegetables_English + Animals_Zuni + Vegetables_Zuni). On the MoCA, we did not apply the 1-point correction for < 12 years of education, as we wanted to evaluate associations between education and performance.

Statistical Analyses

Descriptive statistics were calculated to summarize patient characteristics. For normally distributed variables, means and SD were calculated, and t tests were used to evaluate group differences. For data not normally distributed medians and interquartile ranges (IQR) were calculated and were compared across groups by the Kruskal-Wallis test. Frequencies and percentages were calculated for categorical variables and were compared with the χ² test.

We conducted t tests/Kruskal-Wallis tests (depending on distributions) to evaluate group differences in cognitive performances (MoCA, Trails-A, Trails-B, Craft-I, Craft-D, Benson-I, NS-F, NS-B, MINT, Fluency F+L, Animals, Vegetables) between English- and Zuni-dominant participants (self-reported primary language spoken). For continuous measures, we conducted correlation analyses (Pearson and Spearman depending on distributions) between each cognitive variable and (1) education, (2) age at which the individual learned English, and (3) the language dominance index.

RESULTS

Thirty-nine participants were enrolled in the Zuni cohort. Table 1 displays basic demographic and bilingualism variables. With the exception of 2 monolingual English participants, the sample consisted of Zuni-English bilinguals. There were no significant differences in the age, sex, or education of the self-reported Zuni-dominant (N = 18) and English-dominant (N = 21) groups. The self-reported English-dominant group learned English at a significantly earlier age (median = 3) than self-reported Zuni-dominant group (median = 5). Across the group, 43.6% of participants reported that they used English and Zuni equally (Fig. 1), though self-reported Zuni-dominant individuals reported that they used English 32.5% of the time, whereas English-dominant individuals reported that they used English 60.0% of the time, on average. The language dominance index, calculated from the objective assessment of semantic fluency in English and Zuni indicated that participants were always able to produce more responses in English versus Zuni. Self-reported English-dominant participants exhibited an increased language-dominant index (higher performances suggesting greater English dominance) relative to Zuni dominant. Notably, using published cutoffs to diagnose MCI based on performance on the MoCA indicated that 35.21% of the cohort would be classified as MCI across the groups. While not significant, the Zuni-dominant group exhibited a higher rate of MCI relative to the English-dominant group.

TABLE 1.

Demographics and Language Characteristics of the Cohort

	Zuni dominant (N = 18)	English dominant (N = 21)	Total (N = 39)	P
Age (y)	56.3 ± 9.9	51.5 ±11.0	53.7± 10.7	0.16
Male (%)	6 (33.33)	9 (42.86)	15 (38.46)	0.78
Education (Y)	12.0 [12.0;14.0]	12.0 [12.0;14.0]	12.0 [12.0;14.0]	1.00
Age learned Zuni	3.5 [3.0;4.0]	4.0 [3.0;6.0]	4.0 [3.0;5.0]	0.27
Age learned English	5.0 [4.5;6.0]	3.0 [3.0;5.0]	4.5 [3.0;6.0]	0.001
Language learned first				0.004
Zuni (%)	15 (83.33)	6 (31.58)	21 (56.76)
English (%)	3 (16.67)	13 (68.42)	16 (43.24)
% time Zuni used	67.5 [50.0;75.0]	40.0 [15.0;50.0]	50.0 [40.0;60.0]	< 0.001
% time English used	32.5 [25.0;50.0]	60.0 [50.0;85.0]	50.0 [40.0;60.0]	< 0.001
Language dominance Index	0.68 [0.64;0.72]	0.74 [0.69;0.82]	0.71 [0.66;0.78]	0.01

Note: Language dominance based on self-report primary language spoken. Unless otherwise specified, mean ± SD for normally distributed data and median (interquartile range) for not normally distributed data are presented. P-values reported from either t tests or Kruskal-Wallis test for continuous data and from the χ² test with continuity correction for categorical data.

FIGURE 1.

Histogram depicting the number of participants that endorsed percentages of time either English or Zuni languages were used for oral communication.

Table 2 presents age-corrected neuropsychological test performances for the total group in addition to performances split by self-reported language dominance (with the exception of the MoCA). Figure 2 presents boxplots of the neuropsychological test performances by self-reported language dominance. There were no significant differences between self-reported language dominance groups, with effect sizes ranging from −0.37 (negative effect sizes indicate self-reported Zuni-dominant individuals performed better than self-reported English-dominant individuals) to 0.29 (positive effect sizes indicate relatively better performances in self-reported English-dominant participants). Table 3 presents the associations between the neuropsychological assessments, language dominance variables, age, and education with the language dominance index. Of the cognitive tests, only the MoCA was correlated with education, and the phonemic fluency was inversely associated with the language dominance index. Age in which individuals learned English was associated with the individuals age (eg, the older an individual is the later they learned English). Age in which individuals learned English was also inversely associated with the language dominance index. Finally, the current percentage of time that individuals used Zuni or English, respectively, were strongly correlated with the Language dominance Index.

TABLE 2.

Performances on Neuropsychological Assessments by Self-reported Current Primary Language Used for Oral Communication

Self-reported current primary language	Zuni (N = 18)	English (N = 21)	Total (N = 39)	t or χ2 score	P	Cohen d
MoCA*	25.5 [22.0;28.0]	27.0 [26.0;28.5]	27.0 [23.0;28.0]	1.06	0.30	0.33
Craft-I	36.3 ± 11.3	38.4 ± 10.7	37.5 ± 10.9	0.56	0.58	0.19
Craft-D	37.9 ± 10.9	39.5 ± 8.2	38.7 ± 9.5	0.53	0.60	0.17
Benson-I*	58.2 [57.3;59.2]	58.0 [50.2;58.6]	58.2 [50.5;59.0]	0.89	0.35	−0.32
Benson-D	46.3 ± 9.9	48.6 ± 11.1	47.5 ± 10.5	0.71	0.49	0.22
NS-F	40.3 ± 8.1	41.5 ± 8.6	40.9 ± 8.3	0.44	0.66	0.14
NS-B	40.7 ± 4.8	38.2 ± 8.0	39.4 ± 6.7	−1.16	0.26	−0.37
MINT	39.3 ± 7.6	41.9 ± 9.8	40.7 ± 8.8	0.91	0.37	0.29
Fluency F+L	37.9 ± 6.7	37.0 ± 8.0	37.4 ± 7.4	−0.38	0.71	−0.12
Animals	38.5 ± 8.6	36.3 ± 8.0	37.3 ± 8.3	−0.83	0.41	−0.27
Vegetables	38.8 ± 10.2	39.1 ± 6.4	39.0 ± 8.3	0.10	0.92	0.03
Trails-A*	37.9 [28.6;45.3]	39.9 [27.9;47.4]	38.5 [28.2;45.5]	0.19	0.66	0.16
Trails-B*	37.2 [32.5;40.1]	38.1 [26.7;47.4]	38.1 [29.1;42.4]	0.20	0.65	−0.05

^*Note: indicates variable is not normally distributed and Kruskal-Wallis test was used to evaluate group differences. MoCA indicates Montreal Cognitive Assessment; Craft-I, Craft Story 21 immediate recall; Craft-D, Craft Story 21 delayed recall; Benson-I, Benson complex figure copy; Benson-D, Benson complex figure delayed recall; NS-F, Number span test forward; NS-B, number span test backwards; MINT, multilingual naming test; Fluency F+L, phonemic test F and L words total; Animals, fluency animal list generation, Vegetables, Fluency vegetables list generation; Trails-A, trail making test part A; Trails-B, trail making test part B.

FIGURE 2.

Boxplots depicting neuropsychological performances by self-reported language dominance. MoCA indicates Montreal Cognitive Assessment; Craft-I, Craft Story 21 immediate recall; Craft-D, Craft Story 21 delayed recall, Benson-I, Benson complex figure copy; Benson-D, Benson complex figure delayed recall; NS-F, number span test forward; NS-B, number span test backwards; MINT, multilingual naming test; fluency F+L, phonemic test F and L words total; Animals, fluency animal list generation; Vegetables, fluency vegetables list generation; Trails-A, trail making test part A; Trails-B, trail making test part B.

TABLE 3.

Associations Between Neuropsychological Assessments and Years of Education, Age at Which Individuals Learned English, and Language Dominance Index

	Education		Age learned English		Language dominance index
Measure	r	P	r	P	r	P
MoCA*	0.33	0.04	−0.15	0.36	−0.08	0.65
Craft-I	−0.14	0.41	0.05	0.78	0.06	0.71
Craft-D	0.04	0.80	−0.11	0.52	0.13	0.42
Benson-I*	−0.10	0.53	0.20	0.22	−0.20	0.23
Benson-D	< 0.01	0.99	−0.08	0.62	0.06	0.73
NS-F	0.21	0.22	−0.11	0.53	−0.07	0.67
NS-B	0.20	0.24	−0.24	0.15	−0.10	0.55
MINT	0.24	0.14	−0.03	0.87	0.05	0.75
Fluency F+L	0.20	0.24	0.03	0.85	−0.36	0.03
Animals	0.31	0.05	0.07	0.67	−0.16	0.34
Vegetables	0.30	0.06	−0.03	0.85	0.28	0.08
Trails-A*	0.11	0.54	0.01	0.96	−0.06	0.74
Trails−B*	0.24	0.15	−0.20	0.23	−0.08	0.65
Age*	0.24	0.13	0.41	0.003	−0.25	0.11
Education*	NA	NA	−0.09	0.58	< 0.01	0.97
Age learned Zuni*	0.13	0.41	0.07	0.65	0.17	0.21
Age learned English*	−0.09	0.58	NA	NA	−0.46	0.003
Percentage of time use Zuni*	−0.01	0.94	0.70	< 0.001	−0.62	< 0.001
Percentage of time use English*	0.01	0.94	−0.70	< 0.001	0.62	< 0.001

^*Notes: indicates variable is not normally distributed and Spearman correlations were used. Bold indicates significant differences between groups. MoCA indicates Montreal Cognitive Assessment; Craft-I, Craft Story 21 immediate recall; Craft-D, Craft Story 21 delayed recall; Benson-I, Benson complex figure copy; Benson-D, Benson complex figure delayed recall; NS-F, Number span test forward; NS-B, number span test backwards; MINT, multilingual naming test; Fluency F+L, phonemic test F and L words total; Animals, fluency animal list generation, Vegetables, fluency vegetables list generation; Trails-A, trail making test part A; Trails-B, Trail making test part B.

DISCUSSION

AI/AN communities face numerous barriers to AD diagnosis and care. As assessment tools were not designed for use with AI/AN communities, a major concern is the potential for misdiagnosis due to a lack of research evaluating the influence of sociocultural/environmental factors on neuropsychological test performance and the availability of appropriate normative data. We have partnered with the Zuni community to develop socio-culturally sensitive assessment procedures and address barriers to dementia diagnosis and care. The aim of this study is to first evaluate the degree of bilingualism and education in our study cohort and the impact of these factors on neuropsychological performances in the Zuni community. Our primary findings indicate that the Zuni community members that participated in our study have a high proportion of balanced Zuni and English bilinguals and more English-dominant bilinguals. Most likely because the cohort was at least equal bilingual or English-dominant, there was no evidence that language dominance influenced performance on neuropsychological assessments. However, education significantly impacted performance on the MoCA.

With the exception of 2 participants who were monolingual English speakers, our cohort was comprised of Zuni and English bilinguals. A total of 43.59% of the cohort endorsed using English and Zuni similar percentages of time for oral communication (ie, 50/50 split). Self-reported English-dominant participants learned English at a significantly earlier age. The language dominance index, calculated from the objective assessment of semantic fluency in English and Zuni, indicated that participants were consistently able to produce more responses in English versus Zuni. Self-reported English-dominant participants exhibited a significantly higher language-dominant index (higher performances suggesting more English dominance) relative to Zuni dominant. Performance on the MoCA, a general screening assessment routinely used in primary care and neurology, indicated that approximately a third of the cohort (34.21%) exhibited performances suggestive further evaluation is necessary, despite being a relatively young cohort. There were no significant differences in performances on the neuropsychological tests between Zuni- and English-dominant participants. Evaluation of factors as continuous measures indicated that primarily education was associated with performances, most evident on the MoCA. Effect sizes suggest that with the exception of phonemic fluency, education exhibited relatively greater effect sizes relative to age of English acquisition and language dominance index.

Typical approaches to evaluating appropriateness for English versions of neuropsychological tests include making decisions based on self-reported language dominance in addition to more detailed measures, such as age at which an individual learned English and quantitative assessment of language dominance. Individuals from the Zuni community are typically exposed to both English and Zuni from relatively young ages and continue to use both languages throughout their lives, which is consistent with what we observed in our cohort. It is likely that the combined early exposure and continued use of English are why we fail to observe differences in neuropsychological performances in self-reported Zuni relative to English-dominant individuals. This highlights that when individuals from this community are sufficiently exposed to English at a younger age, there appear to be minimal differences, and English testing is appropriate. However, there are likely many limitations pertaining to the representativeness of this sample and these results do not apply to individuals who are Zuni dominant, further discussed below.

We also evaluated whether the degree of language dominance based on objective assessment and education influenced neuropsychological performances. Both self-reported English- and Zuni-dominant cohorts’ performance on a language dominance index indicated that all participants were able to produce more words in English relative to Zuni. This provides further support indicating that this cohort is at least equally bilingual or English-dominant and therefore English-language assessment is appropriate in this cohort of Zuni/English bilinguals. However, it was surprising that even Zuni-dominant individuals that reported they used Zuni > 75% of the time still exhibited a language dominance score indicative of English dominance. Given the trend-level associations between education and both animals and vegetables (higher education correlated with higher performance on Animals and Vegetables), this pattern may suggest that community members are taught to categorize animals and vegetables through their English schooling, in which case this approach to organizing this type of information would be done primarily in English. The language dominance index was correlated with the age at which an individual learned English as well as the percentage of time an individual used English for oral communication. On the one hand, this validates our approach. However, it also raises the question of whether objective assessment is necessary or assessment of the age of language acquisition and current percentage of time language is used for oral communication are sufficient to quantify the degree of bilingualism in this community, though future work is necessary to identify alternatives.

Education had a greater influence on cognitive performances relative to age at which an individual learned English and the Language Dominance Index. This was most apparent on the MoCA, with higher education correlating with MoCA performances. It is well-established that MoCA scores are associated with education.^24–26 In individuals with lower education, this may result in performance below cutoffs, but this is not a reflection of impairment or a decline from premorbid estimates. Approaches to account for this include adding a point to the total score for individuals who have < 12 years of education¹¹ or discounting certain item-level performances.²⁷ Our sample ranged from 12 to 14 years in education level. Future work is necessary to evaluate whether these adjustments minimize misdiagnosis in AI/ANs or whether it is appropriate to use the MoCA with this population. In addition, while years of education are helpful to assess education, they fail to incorporate the quality of education. The Wide Range Achievement Test—IV Reading subtest, which is often referred to as a literacy measure, can serve to approximate the quality of education the individual was exposed to.²⁸ AI/ANs have a long and often tragic history of education in the United States through assimilation-focus boarding and other Native schools. The quality of education in older AI/ANs may be quite variable. Accounting for the quality of education has been shown to diminish the effects of race/ethnicity on most neuropsychological measures in older Black/AAs²⁹; thus, quality of education may be a more informative measure than years of education in racial/ethnic minorities and will be incorporated in future studies.

Sociocultural factors beyond education and bilingualism, such as individual and structural social determinants of health, can also drive different pathways that either promote resilience or confer risk for cognitive decline.^30–32 The National Institute of Aging (NIA)³³ and the National Institute of Minority Health and Health Disparities (NIMHD)³⁴ have developed overlapping health disparities frameworks for evaluating the various domains of influence (eg, biological, behavioral, sociocultural, environmental, and health care system—the latter only included in NIMHD model) that can impact outcomes at the individual, interpersonal, community, or society level. Importantly, the framework emphasizes the importance of adopting a life course perspective,³⁵ which includes consideration of cumulative social and environmental exposures, of early adverse events, and transgenerational transmission of risk and resilience. For example, life course social forces, such as increased early life socioeconomic status and educational attainment can lead to increased cognitive reserve.^19,36–38 It is hypothesized that individuals with higher education and corresponding cognitive reserve may experience a greater degree of pathologic change while maintaining intact cognitive functioning.^39,40

There are several strengths of this study, including that assessment was conducted within the community and recruitment and study procedures were conducted by community members, overcoming barriers to research participation. However, despite this, there are larger barriers to research participation that include mistrust, logistical challenges, and prioritizing the needs of the community that may have resulted in a cohort that is not fully representative of the broader community. This is reflected in a cohort that has a higher rate of high school education than the broader community (23.5% of community members have not completed high school) and it is likely that Zuni-dominant individuals may not have participated in the research due to anticipated barriers or lack of access. While the data collected support continuing to conduct neuropsychological assessment in English, we had very few participants who were relatively more Zuni dominant (eg, speaking Zuni > 75% of the time). As such, these results cannot be extended to these individuals and future work is necessary.

We consulted with the Tribe at every stage of the process. In addition, while the study consists of a community-based sample, it is enriched with individuals with kidney disease identified through prior projects conducted within the community. We appreciate that kidney disease can impact cognition and could influence the rate of MCI diagnosis. However, in this cohort, our prior work indicated no significant associations between kidney functioning and cognition.²⁰ Finally, we failed to counterbalance administration order of the tests used for the language dominance index. Our future studies will be sure to counterbalance to evaluate whether we continue to see a consistent indication of better performance in English relative to Zuni across the cohort.

Taken together, this work highlights that the number of years of education has a greater effect on cognitive performances relative to language dominance, suggesting future efforts incorporate more detailed assessment of the quality of education in balanced or English-dominant bilinguals. Our results raise concerns for the use of the MoCA as a screening measure and highlights the need for culturally appropriate cognitive testing tools that are specifically designed for AI/AN populations. In addition, comprehensive neuropsychological assessments that incorporate the impact of language, education, and other cultural factors on test performance are critical. Together, this highlights the importance of a comprehensive history, family member and patient-reported concern for a decline from premorbid estimates, and longitudinal standardized assessments to evaluate MCI and dementia in this community.