Dementia diagnosis disparities by race and ethnicity

Predicting Risk of Alzheimer's Diseases and Related Dementias with AI Foundation Model on Electronic Health Records

Early identification of Alzheimer's disease (AD) and AD-related dementias (ADRD) has high clinical significance, both because of the potential to slow decline through initiating FDA-approved therapies and managing modifiable risk factors, and to help persons living with dementia and their families to plan before cognitive loss makes doing so challenging. However, substantial racial and ethnic disparities in early diagnosis currently lead to additional inequities in care, urging accurate and inclusive risk assessment programs. In this study, we trained an artificial intelligence foundation model to represent the electronic health records (EHR) data with a vast cohort of 1.2 million patients within a large health system. Building upon this foundation EHR model, we developed a predictive Transformer model, named TRADE, capable of identifying risks for AD/ADRD and mild cognitive impairment (MCI), by analyzing the past sequential visit records. Amongst individuals 65 and older, our model was able to generate risk predictions for various future timeframes. On the held-out validation set, our model achieved an area under the receiver operating characteristic (AUROC) of 0.772 (95% CI: 0.770, 0.773) for identifying the AD/ADRD/MCI risks in 1 year, and AUROC of 0.735 (95% CI: 0.734, 0.736) in 5 years. The positive predictive values (PPV) in 5 years among individuals with top 1% and 5% highest estimated risks were 39.2% and 27.8%, respectively. These results demonstrate significant improvements upon the current EHR-based AD/ADRD/MCI risk assessment models, paving the way for better prognosis and management of AD/ADRD/MCI at scale.

Services for Latino Family Dementia Caregivers: Is the Workforce Prepared?

The American healthcare system is ill-fitted to meet the needs of the growing population with Alzheimer's and Related Dementias (ADRD) and their caregivers, which disproportionately affects Latinos. This paper describes the caregiving process for Latino patients with ADRD from the perspective of caregivers and service providers to reveal service areas in need of improvement. Ten providers and 24 Latino caregivers were interviewed through the Miami-Dade County Aging Network. Interviews were analyzed using a grounded theory approach. Five themes emerged: lack of social service connectivity, case management challenges, unmet mental health needs, language barriers, and the need for alternative service models. Our analysis indicates insufficient training and the need to improve dementia competencies, support strategies, and understanding of cultural nuances among service providers. These findings can help inform the development of a required ADRD provider education curriculum that incorporates cultural competency training to improve the quality and effectiveness of care.

Evaluating Large Language Models in Extracting Cognitive Exam Dates and Scores

Importance

Large language models (LLMs) are crucial for medical tasks. Ensuring their reliability is vital to avoid false results. Our study assesses two state-of-the-art LLMs (ChatGPT and LlaMA-2) for extracting clinical information, focusing on cognitive tests like MMSE and CDR.

Objective

Evaluate ChatGPT and LlaMA-2 performance in extracting MMSE and CDR scores, including their associated dates.

Methods

Our data consisted of 135,307 clinical notes (Jan 12th, 2010 to May 24th, 2023) mentioning MMSE, CDR, or MoCA. After applying inclusion criteria 34,465 notes remained, of which 765 underwent ChatGPT (GPT-4) and LlaMA-2, and 22 experts reviewed the responses. ChatGPT successfully extracted MMSE and CDR instances with dates from 742 notes. We used 20 notes for fine-tuning and training the reviewers. The remaining 722 were assigned to reviewers, with 309 each assigned to two reviewers simultaneously. Inter-rater-agreement (Fleiss' Kappa), precision, recall, true/false negative rates, and accuracy were calculated. Our study follows TRIPOD reporting guidelines for model validation.

Results

For MMSE information extraction, ChatGPT (vs. LlaMA-2) achieved accuracy of 83% (vs. 66.4%), sensitivity of 89.7% (vs. 69.9%), true-negative rates of 96% (vs 60.0%), and precision of 82.7% (vs 62.2%). For CDR the results were lower overall, with accuracy of 87.1% (vs. 74.5%), sensitivity of 84.3% (vs. 39.7%), true-negative rates of 99.8% (98.4%), and precision of 48.3% (vs. 16.1%). We qualitatively evaluated the MMSE errors of ChatGPT and LlaMA-2 on double-reviewed notes. LlaMA-2 errors included 27 cases of total hallucination, 19 cases of reporting other scores instead of MMSE, 25 missed scores, and 23 cases of reporting only the wrong date. In comparison, ChatGPT's errors included only 3 cases of total hallucination, 17 cases of wrong test reported instead of MMSE, and 19 cases of reporting a wrong date.

Conclusions

In this diagnostic/prognostic study of ChatGPT and LlaMA-2 for extracting cognitive exam dates and scores from clinical notes, ChatGPT exhibited high accuracy, with better performance compared to LlaMA-2. The use of LLMs could benefit dementia research and clinical care, by identifying eligible patients for treatments initialization or clinical trial enrollments. Rigorous evaluation of LLMs is crucial to understanding their capabilities and limitations.

Bilingual neurocognitive resiliency, vulnerability, and Alzheimer's disease biomarker correlates in Latino older adults enrolled in the Health and Aging Brain Study - Health Disparities (HABS-HD)

INTRODUCTION

The effects of bilingualism on neuropsychological test performance in bilinguals with and without cognitive impairment are not well-understood and are relatively limited by small sample sizes of Latinos.

METHODS

Using analysis of covariance (ANCOVA), we explored patterns of cognitive performance and impairment across a large sample of community-dwelling bilingual and monolingual Latino older adults with (n = 180) and without (n = 643) mild cognitive impairment (MCI) enrolled in HABS-HD.

RESULTS

Bilinguals demonstrated cognitive resiliency in the form of significantly better performance on the Trail Making Test and Digit Symbol Substitution Test, observed across the cognitively unimpaired and MCI groups. In contrast, bilinguals demonstrated cognitive vulnerability in the form of significantly poorer performance and higher impairment rates on phonemic fluency in the MCI phase, only. Follow-up analyses revealed less balanced bilinguals demonstrated poorer performance and higher impairment rates on this measure, supported by lower levels of plasma Aβ 42/40.

DISCUSSION

Patterns of cognitive performance and impairment differ as a function of bilingualism. Bilingualism must be considered when evaluating cognitive and biomarker outcomes in Latino older adults.

Highlights

Latino bilinguals perform better on measures of processing speed and coding.Latino bilinguals with MCI demonstrate cognitive vulnerability in verbal fluency.Less balanced bilinguals demonstrate greatest vulnerability anchored by Aβ 42/40.

The (in)visible Brazilians: A perspective review on the need for brain health and dementia research with Brazilian immigrants in the United States

Introduction

The Brazilian population in the United States (U.S.), a Latinx subgroup, is rapidly growing and aging but remains underrepresented in U.S. health research. In addition to group-specific genetic and environmental risks, Brazilian immigrants and their offspring in the U.S. likely have cumulative risks for health inequities.It is estimated that 71% of Brazilian immigrants in the U.S. are undocumented, which may limit healthcare access/utilization. Furthermore, mental health is reported as a health priority by Brazilian immigrants in the U.S., and there is a lack of research on Alzheimer's disease and related dementia (AD/ADRD) in this population.

Methods

We reviewed the scientific literature using traditional (e.g., PubMed) sources and databases generated by U.S. and Brazilian governments, as well as international organizations, and press articles.

Results

This perspective review lists recommendations for researchers, health providers, and policymakers to promote greater inclusion of U.S. Brazilian populations in health research and care. The review identifies research areas in need of attention to address health inequities and promote mental/brain health in Brazilian immigrants and their offspring living in the U.S. These research areas are: 1) epidemiological studies to map the prevalence and incidence of mental/brain health conditions; 2) research on aging and AD/ADRD risk factors among Brazilian populations in the U.S.; and 3) the need for greater representation of U.S-residing Brazilian population in other relevant research areas involving genetics, neuropathology, and clinical trials.

Conclusions

The recommendation and research efforts proposed should help to pave the way for the development of community-engagement research and to promote mental/brain health education, improvement of mental/brain health and AD/ADRD services, and the development of culturally-informed intervention to the U.S.-residing Brazilian communities.

HIGHLIGHTS

The Brazilian population in the United States is growing but is underrepresented in U.S. health research.Approximately 71% of Brazilian immigrants in the United States are undocumented, with an increased risk for health inequities.Mental health is reported as a central health priority by Brazilian immigrants in the United States.There is a lack of research on Alzheimer's disease and other dementias (ADRD) in Brazilian immigrants in the United States.Epidemiological research is needed to map the prevalence/incidence of mental health conditions and ADRD risk factors among Brazilian immigrants in the United States.

Geographic Patterns of Dementia in the United States: Variation by Place of Residence, Place of Birth, and Subpopulation

OBJECTIVES

The prevalence of dementia varies geographically in the United States. However, the extent to which this variation reflects contemporary place-based experiences versus embodied exposures from earlier in the life course remains unclear, and little is known regarding the intersection of place and subpopulation. This study, therefore, evaluates whether and how risk for assessed dementia varies by place of residence and birth, overall and by race/ethnicity and education.

METHODS

We pool data from the 2000 to 2016 waves of the Health and Retirement Study, a nationally representative panel survey of older U.S. adults (n = 96,848 observations). We estimate the standardized prevalence of dementia by Census division of residence and birth. We then fit logistic regression models of dementia on region of residence and birth, adjusting for sociodemographic characteristics, and examine interactions between region and subpopulation.

RESULTS

The standardized prevalence of dementia ranges from 7.1% to 13.6% by division of residence and from 6.6% to 14.7% by division of birth, with rates highest throughout the South and lowest in the Northeast and Midwest. In models accounting for region of residence, region of birth, and sociodemographic covariates, Southern birth remains significantly associated with dementia. Adverse relationships between Southern residence or birth and dementia are generally largest for Black and less-educated older adults. As a result, sociodemographic disparities in predicted probabilities of dementia are largest for those residing or born in the South.

DISCUSSION

The sociospatial patterning of dementia suggests its development is a lifelong process involving cumulated and heterogeneous lived experiences embedded in place.

On gaps of clinical diagnosis of dementia subtypes: A study of Alzheimer's disease and Lewy body disease

Introduction

Alzheimer's disease (AD) and Lewy body disease (LBD) are the two most common neurodegenerative dementias and can occur in combination (AD+LBD). Due to overlapping biomarkers and symptoms, clinical differentiation of these subtypes could be difficult. However, it is unclear how the magnitude of diagnostic uncertainty varies across dementia spectra and demographic variables. We aimed to compare clinical diagnosis and post-mortem autopsy-confirmed pathological results to assess the clinical subtype diagnosis quality across these factors.

Methods

We studied data of 1,920 participants recorded by the National Alzheimer's Coordinating Center from 2005 to 2019. Selection criteria included autopsy-based neuropathological assessments for AD and LBD, and the initial visit with Clinical Dementia Rating (CDR) stage of normal, mild cognitive impairment, or mild dementia. Longitudinally, we analyzed the first visit at each subsequent CDR stage. This analysis included positive predictive values, specificity, sensitivity and false negative rates of clinical diagnosis, as well as disparities by sex, race, age, and education. If autopsy-confirmed AD and/or LBD was missed in the clinic, the alternative clinical diagnosis was analyzed.

Findings

In our findings, clinical diagnosis of AD+LBD had poor sensitivities. Over 61% of participants with autopsy-confirmed AD+LBD were diagnosed clinically as AD. Clinical diagnosis of AD had a low sensitivity at the early dementia stage and low specificities at all stages. Among participants diagnosed as AD in the clinic, over 32% had concurrent LBD neuropathology at autopsy. Among participants diagnosed as LBD, 32% to 54% revealed concurrent autopsy-confirmed AD pathology. When three subtypes were missed by clinicians, "No cognitive impairment" and "primary progressive aphasia or behavioral variant frontotemporal dementia" were the leading primary etiologic clinical diagnoses. With increasing dementia stages, the clinical diagnosis accuracy of black participants became significantly worse than other races, and diagnosis quality significantly improved for males but not females.

Discussion

These findings demonstrate that clinical diagnosis of AD, LBD, and AD+LBD are inaccurate and suffer from significant disparities on race and sex. They provide important implications for clinical management, anticipatory guidance, trial enrollment and applicability of potential therapies for AD, and promote research into better biomarker-based assessment of LBD pathology.

Prevalence of young-onset dementia: nationwide analysis of routinely collected data

INTRODUCTION

Young-onset dementia prevalence is understudied internationally. Previous studies have been limited by low case numbers, reliance on single sources of routinely collected health data for case identification and inclusion of a limited age range. Our objective was to determine the 1-year period prevalence of diagnosed dementia in people aged 0-64 in the entire New Zealand population using routinely collected health data.

METHODS

A population-based descriptive study was carried out in New Zealand (population 4.8 million) using routinely collected deidentified health data from 2016 to 2020. Dementia cases in seven linked health datasets in the New Zealand Integrated Data Infrastructure were identified using diagnostic codes and/or use of antidementia medication. Prevalence for each of the four study years was calculated by age, sex and ethnicity.

RESULTS

From a total population of 4 027 332-4 169 754 individuals aged 0-64, we identified 3396-3474 cases of 'all-cause' dementia in each of the study years (prevalence crude range: 83-84/100 000 people aged 0-64; 139-141/100 000 people aged 30-64 years; 204-207/100 000 people aged 45-64 years). Age-standardised prevalence was higher in males than females. Age-standardised and sex-standardised prevalence was higher in Māori and Pacific People than European and Asian.

DISCUSSION

By using a large study population and multiple national health datasets, we have minimised selection bias and estimated the national prevalence of diagnosed young-onset dementia with precision. Young-onset dementia prevalence for the total New Zealand population was similar to reported global prevalence, validating previous estimates. Prevalence differed by ethnicity, which has important implications for service planning.

Predictors of Incident Mild Cognitive Impairment and Its Course in a Diverse Community-Based Population

BACKGROUND AND OBJECTIVES

To investigate sociodemographic and medical predictors of incident mild cognitive impairment (MCI) and subsequent course of MCI at follow-up, including sustained MCI diagnosis, classification as cognitively normal, and progression to dementia.

METHODS

Within a community-based cohort, diagnoses of MCI were made with a published algorithm. Diagnosis of dementia was based on clinical consensus. Cox regressions estimated hazard ratios of incident MCI associated with several predictors. Modified Poisson regressions estimated relative risks associated with predictors of diagnostic status at follow-up after incidence.

RESULTS

Among 2,903 cognitively normal participants at baseline, 752 developed MCI over an average of 6.3 (SD 4.5) years (incidence rate 56 per 1,000 person-years). Presence of APOE ε4 and higher medical burden increased risk of incident MCI, while more years of education, more leisure activities, and higher income decreased this risk. Of the incident MCI cases, after an average of 2.4 years of follow-up, 12.9% progressed to dementia, 9.6% declined in functioning and did not meet the algorithmic criteria for MCI but did not meet the clinical criteria for dementia, 29.6% continued to meet MCI criteria, and 47.9% no longer met MCI criteria. Multidomain MCI, presence of APOE ε4, depressive symptoms, and antidepressant use increased the risk of progression to dementia.

DISCUSSION

This community-based study showed that almost half of the individuals with incident MCI diagnoses were classified as cognitively normal at follow-up. Predictors of incident MCI demonstrably differed from those of subsequent MCI course; these findings can refine expectations for cognitive and functional course of those presenting with MCI.

Gender Differences in Neuropsychiatric Symptoms Among Community-Dwelling Mexican Americans Aged 80 and Older

Background and ObjectivesTo assess gender differences in prevalence of neuropsychiatric symptoms (NPS) among community-dwelling Mexican Americans ≥80 years. Research Design and Methods: Using data from Wave 7 (2010-2011) of the Hispanic Established Population for the Epidemiological Study of the Elderly, we analyzed the NPS of 914 participants as determined by the Neuropsychiatric Inventory (NPI) with assessments conducted by their caregivers. Multivariate logistic regression models were used to test the association of individual NPS with gender, adjusting for relevant characteristics. Results: The average age of our sample was 86.1 years, and 65.3% were women. Over 60% of participants had at least one informant/caregiver reported NPS. After adjustment, women had lower odds than men of agitation/aggression but higher odds of dysphoria/depression and anxiety. Discussion: Recognizing gender differences in NPS phenotype could help guide development of culturally appropriate NPS screening and treatment programs.