Automated analysis of low-field brain MRI in cerebral malaria.

A central challenge of medical imaging studies is to extract biomarkers that characterize disease pathology or outcomes. Modern automated approaches have found tremendous success in high-resolution, high-quality magnetic resonance images. These methods, however, may not translate to low-resolution images acquired on magnetic resonance imaging (MRI) scanners with lower magnetic field strength. In low-resource settings where low-field scanners are more common and there is a shortage of radiologists to manually interpret MRI scans, it is critical to develop automated methods that can augment or replace manual interpretation, while accommodating reduced image quality. We present a fully automated framework for translating radiological diagnostic criteria into image-based biomarkers, inspired by a project in which children with cerebral malaria (CM) were imaged using low-field 0.35 Tesla MRI. We integrate multiatlas label fusion, which leverages high-resolution images from another sample as prior spatial information, with parametric Gaussian hidden Markov models based on image intensities, to create a robust method for determining ventricular cerebrospinal fluid volume. We also propose normalized image intensity and texture measurements to determine the loss of gray-to-white matter tissue differentiation and sulcal effacement. These integrated biomarkers have excellent classification performance for determining severe brain swelling due to CM.

Demographic and Symptom Correlates of Initial Idiopathic Psychiatric Diagnosis in Frontotemporal Dementia.

INTRODUCTION: This study aims to measure frequency and correlates of initial idiopathic psychiatric diagnosis in a cohort of 147 patients with Frontotemporal Dementia (FTD)-spectrum disorders. METHODS: Participants were evaluated at the National Institutes of Health in Bethesda, Maryland. Initial participant diagnoses were determined by chart review and patient and informant interviews. Logistic regression was used to assess the relationships between diagnosis and age of symptom onset, gender, education, family history of psychiatric illness, and family history of dementia. Additional exploratory analyses investigated patients' first symptom type. RESULTS: 25% (n=43) of all the patients reviewed were initially misdiagnosed with an idiopathic psychiatric illness, which is less than half the commonly cited 50% rate.3 Depression was the most common misdiagnosis (46.5%). Family history of dementia, family history of mental illness and an exploratory analysis of behavioral first symptoms suggested significant association with a greater likelihood of initial idiopathic psychiatric diagnosis in FTD patients. DISCUSSION: This data confirms patterns of initial idiopathic psychiatric diagnosis in FTD and elucidates potential factors underlying misdiagnosis. Potential implications for patient outcomes, caregiver burden and healthcare costs are discussed.

Early life stress, prenatal secondhand smoke exposure, and the development of internalizing symptoms across childhood.

BACKGROUND: Prior findings relating secondhand tobacco smoke (SHS) exposure and internalizing problems, characterized by heightened anxiety and depression symptoms, have been equivocal; effects of SHS on neurodevelopment may depend on the presence of other neurotoxicants. Early life stress (ELS) is a known risk factor for internalizing symptoms and is also often concurrent with SHS exposure. To date the interactive effects of ELS and SHS on children's internalizing symptoms are unknown. We hypothesize that children with higher exposure to both prenatal SHS and ELS will have the most internalizing symptoms during the preschool period and the slowest reductions in symptoms over time. METHODS: The present study leveraged a prospective, longitudinal birth cohort of 564 Black and Latinx mothers and their children, recruited between 1998 and 2006. Cotinine extracted from cord and maternal blood at birth served as a biomarker of prenatal SHS exposure. Parent-reported Child Behavior Checklist (CBCL) scores were examined at four timepoints between preschool and eleven years-old. ELS exposure was measured as a composite of six domains of maternal stress reported at child age five. Latent growth models examined associations between SHS, ELS, and their interaction term with trajectories of children's internalizing symptoms. In follow-up analyses, weighted quintile sum regression examined contributions of components of the ELS mixture to children's internalizing symptoms at each time point. RESULTS: ELS interacted with SHS exposure such that higher levels of ELS and SHS exposure were associated with more internalizing symptoms during the preschool period (ß = 0.14, p = 0.03). The interaction between ELS and SHS was also associated with a less negative rate of change in internalizing symptoms over time (ß=-0.02, p = 0.01). Weighted quintile sum regression revealed significant contributions of maternal demoralization and other components of the stress mixture to children's internalizing problems at each age point (e.g., age 11 WQS ß = 0.26, p < 0.01). CONCLUSIONS: Our results suggest that prior inconsistencies in studies of SHS on behavior may derive from unmeasured factors that also influence behavior and co-occur with exposure, specifically maternal stress during children's early life. Findings point to modifiable targets for personalized prevention.

A local group differences test for subject-level multivariate density neuroimaging outcomes.

A great deal of neuroimaging research focuses on voxel-wise analysis or segmentation of damaged tissue, yet many diseases are characterized by diffuse or non-regional neuropathology. In simple cases, these processes can be quantified using summary statistics of voxel intensities. However, the manifestation of a disease process in imaging data is often unknown, or appears as a complex and nonlinear relationship between the voxel intensities on various modalities. When the relevant pattern is unknown, summary statistics are often unable to capture differences between disease groups, and their use may encourage post hoc searches for the optimal summary measure. In this study, we introduce the multi-modal density testing (MMDT) framework for the naive discovery of group differences in voxel intensity profiles. MMDT operationalizes multi-modal magnetic resonance imaging (MRI) data as multivariate subject-level densities of voxel intensities and utilizes kernel density estimation to develop a local two-sample test for individual points within the density space. Through simulations, we show that this method controls type I error and recovers relevant differences when applied to a specified point. Additionally, we demonstrate the ability to maintain power while controlling the family-wise error rate and false discovery rate when applying the test over a grid of points within the density space. Finally, we apply this method to a study of subjects with either multiple sclerosis (MS) or conditions that tend to mimic MS on MRI, and find significant differences between the two groups in their voxel intensity profiles within the thalamus.

Summary metrics of memory subnetwork functional connectivity alterations in multiple sclerosis.

BACKGROUND: Memory dysfunction is common in multiple sclerosis (MS); mechanistic understanding of its causes is lacking. Large-scale network resting-state functional connectivity (RSFC) is sensitive to memory dysfunction. OBJECTIVE: We derived and tested summary metrics of memory network RSFC. METHODS: Cognitive data and 3T magnetic resonance imaging (MRI) scans were collected from 235 MS patients and 35 healthy controls (HCs). Index scores were calculated as RSFC within (anteriority index, AntI) and between (integration index, IntI) dorsomedial anterior temporal and medial temporal memory subnetworks. Group differences in index expression were evaluated. Associations between index scores and memory/non-memory cognition were evaluated; relationships between T2 lesion volume (T2LV) and index scores were assessed. RESULTS: Index scores were related to memory and T2LV in MS patients, who showed marginally elevated AntI relative to HC (p = 0.06); no group differences were found for IntI. Better memory was associated with higher AntI (ß = 0.15, p = 0.018) and IntI (ß = 0.16, p = 0.014). No associations were found for non-memory cognition. Higher T2LV was associated with higher AntI and IntI; exploratory mediation analysis revealed significant inconsistent mediation, that is, higher index scores partially suppressed the negative association between T2LV and memory. CONCLUSION: Summary, within-subject metrics permit replication and circumvent challenges of traditional (incommensurate) RSFC variables to advance development of mechanistic models of memory dysfunction in MS.

Noise complaint patterns in New York City from January 2010 through February 2021: Socioeconomic disparities and COVID-19 exacerbations.

BACKGROUND: Excessive environmental noise exposure and noise annoyance have been linked to adverse physical and mental health outcomes. Although socioeconomic disparities in acoustically measured and geospatially estimated noise have been established, less is known about disparities in noise complaints, one of the most common sources of distress reported to local municipalities. Furthermore, although some studies have posited urban quieting during the COVID-19 pandemic, little empirical work has probed this and probed noise complaints during the pandemic. OBJECTIVES: Using over 4 million noise complaints from the New York City (NYC) 311 database, we quantified census tract-level socioeconomic disparities in noise complaints since 2010 and examined how such disparities changed during the COVID-19 pandemic. METHODS: Using data from January 2010 through February 2020, we fit linear mixed-effects models, estimating monthly tract-level noise complaints by the proportion of residents who were low-income, time in months since January 2010, categorical month, their interactions, and potential confounds, such as total population and population density. To estimate COVID-19 pandemic effects, we included additional data from March 2020 through February 2021 and additional interactions between proportion low-income, month of year, and an indicator variable for COVID-19 pandemic onset in March 2020. RESULTS: Census tracts with a higher proportion of low-income residents reported more monthly noise complaints and this increased over time (time × month × proportion low-income interaction p-values < .0001 for all months), particularly in warmer months. Socioeconomic disparities in noise complaints were exacerbated during the COVID-19 pandemic (month × proportion low-income × pandemic era interaction p-values < .0001 for March through November), also in a seasonal manner. DISCUSSION: Since 2010, noise complaints have increased the most in the most economically distressed communities, particularly in warmer seasons. This disparity was particularly exacerbated during the COVID-19 pandemic, contrary to some theories of urban quieting. Community-based interventions to ameliorate noise and noise annoyance, both public health hazards, are needed in underserved communities.

Prospective relationships between stigma, mental health, and resilience in a multi-city cohort of transgender and nonbinary individuals in the United States, 2016-2019.

PURPOSE: Gender-based stigma is a fundamental cause of mental health disparities among transgender and non-binary (TGNB) individuals, while resilience factors may be protective. We examined prospective relationships between gender-based enacted stigma, psychological distress, and resilience factors among TGNB individuals. METHODS: Between 2016 and 2017, we enrolled 330 TGNB individuals in three metropolitan areas in the U.S. in a prospective cohort study focused on gender identity development, risk, and resilience across the lifespan. Using multilevel regression, we examined prospective associations between enacted gender-based stigma and psychological distress (measured by the Global Severity Index/BSI-18), and examined transgender pride and social support as moderators, adjusting for age, sex assigned at birth, race/ethnicity, education, and income. RESULTS: Our sample was diverse in age (M = 34.4, range 16-87) and race/ethnicity (56.4% non-White). Over 2 years of follow-up, there was a decrease in reported gender-based stigma (b = - 0.61, p < 0.001) and transgender pride (b = - 0.14, p = 0.003), increase in social support (b = 0.21, p < 0.001), and no change in psychological distress. In adjusted analyses, gender-based stigma was positively associated with psychological distress (b = 1.10, p < 0.001) and social support was negatively associated with psychological distress (b = - 2.60, p < 0.001). Transgender pride moderated the relationship between stigma and psychological distress (p < 0.01), such that the association was stronger for lower levels of transgender pride. CONCLUSIONS: Our study provides longitudinal evidence for the deleterious role of gender-based stigma among TGNB individuals. Future interventions should consider fostering transgender pride and social support to promote mental health and mitigate negative effects of gender-based stigma.

Maturity of gray matter structures and white matter connectomes, and their relationship with psychiatric symptoms in youth.

Brain predicted age difference, or BrainPAD, compares chronological age to an age estimate derived by applying machine learning (ML) to MRI brain data. BrainPAD studies in youth have been relatively limited, often using only a single MRI modality or a single ML algorithm. Here, we use multimodal MRI with a stacked ensemble ML approach that iteratively applies several ML algorithms (AutoML). Eligible participants in the Healthy Brain Network (N = 489) were split into training and test sets. Morphometry estimates, white matter connectomes, or both were entered into AutoML to develop BrainPAD models. The best model was then applied to a held-out evaluation dataset, and associations with psychometrics were estimated. Models using morphometry and connectomes together had a mean absolute error of 1.18 years, outperforming models using a single MRI modality. Lower BrainPAD values were associated with more symptoms on the CBCL (pcorr  = .012) and lower functioning on the Children's Global Assessment Scale (pcorr  = .012). Higher BrainPAD values were associated with better performance on the Flanker task (pcorr  = .008). Brain age prediction was more accurate using ComBat-harmonized brain data (MAE = 0.26). Associations with psychometric measures remained consistent after ComBat harmonization, though only the association with CGAS reached statistical significance in the reduced sample. Our findings suggest that BrainPAD scores derived from unharmonized multimodal MRI data using an ensemble ML approach may offer a clinically relevant indicator of psychiatric and cognitive functioning in youth.

Timing-specific associations between income-to-needs ratio and hippocampal and amygdala volumes in middle childhood: A preliminary study.

It is well known that financial disadvantage is associated with alterations in brain development in regions critical to socioemotional well-being such as the hippocampus and the amygdala. Yet little is known about whether family income at different points in development is differentially associated with these structures. Furthermore, little is known about which environmental factors statistically mediate associations between income and subcortical structure. Using a longitudinal birth cohort and linear mixed-effects models, we identified associations between income-to-needs ratio (INR) at 6 timepoints throughout childhood and hippocampal and amygdala volumes at age 7-9 years (n = 41; 236 INR measurements; 41 brain measurements). Mediation analysis identified environmental sequelae of income that statistically accounted for INR-brain associations. Lower INR prior to age 4 was associated with smaller hippocampal volumes, whereas lower INR prior to age 2 was associated with smaller right amygdala volume. These associations were mediated by unmet basic needs (e.g., food, housing). These findings delineate the temporal specificity of associations between income and hippocampal and amygdala structures.

The landscape of NeuroImage-ing research.

As the field of neuroimaging grows, it can be difficult for scientists within the field to gain and maintain a detailed understanding of its ever-changing landscape. While collaboration and citation networks highlight important contributions within the field, the roles of and relations among specific areas of study can remain quite opaque. Here, we apply techniques from network science to map the landscape of neuroimaging research documented in the journal NeuroImage over the past decade. We create a network in which nodes represent research topics, and edges give the degree to which these topics tend to be covered in tandem. The network displays small-world architecture, with communities characterized by common imaging modalities and medical applications, and with hubs that integrate these distinct subfields. Using node-level analysis, we quantify the structural roles of individual topics within the neuroimaging landscape, and find high levels of clustering within the structural MRI subfield as well as increasing participation among topics related to psychiatry. The overall prevalence of a topic is unrelated to the prevalence of its neighbors, but the degree to which a topic becomes more or less popular over time is strongly related to changes in the prevalence of its neighbors. Finally, we incorporate data from PNAS to investigate whether it serves as a trend-setter for topics' use within NeuroImage. We find that popularity trends are correlated across the two journals, and that changes in popularity tend to occur earlier within PNAS among growing topics. Broadly, this work presents a cohesive model for understanding the emergent relationships and dynamics of research topics within NeuroImage.

Bayesian nonparametric generative models for causal inference with missing at random covariates.

We propose a general Bayesian nonparametric (BNP) approach to causal inference in the point treatment setting. The joint distribution of the observed data (outcome, treatment, and confounders) is modeled using an enriched Dirichlet process. The combination of the observed data model and causal assumptions allows us to identify any type of causal effect-differences, ratios, or quantile effects, either marginally or for subpopulations of interest. The proposed BNP model is well-suited for causal inference problems, as it does not require parametric assumptions about the distribution of confounders and naturally leads to a computationally efficient Gibbs sampling algorithm. By flexibly modeling the joint distribution, we are also able to impute (via data augmentation) values for missing covariates within the algorithm under an assumption of ignorable missingness, obviating the need to create separate imputed data sets. This approach for imputing the missing covariates has the additional advantage of guaranteeing congeniality between the imputation model and the analysis model, and because we use a BNP approach, parametric models are avoided for imputation. The performance of the method is assessed using simulation studies. The method is applied to data from a cohort study of human immunodeficiency virus/hepatitis C virus co-infected patients.

Sex differences in brain resilience of individuals with multiple sclerosis.

BACKGROUND: Brain resilience allows maintenance of neurocognitive function in the face of age or disease-related neural changes. OBJECTIVE: Test the hypothesis that women and men with MS differ in brain resilience. METHODS: This cross-sectional analysis of prospective cohort data included 11,297 patients. Linear mixed effects models predicted performance outcomes on tasks of fine motor dexterity and cognitive processing speed for MRI proxies of disease burden: brain parenchymal fraction (BPF), T2 lesion volume, volumes of deep gray, thalamus, white and cortical gray matter. Covariates were age, sex, age-by-sex, current disease-modifying therapy, disease phenotype, education, total brain volume, and total brain volume-by-sex. Sex-by-MRI metric terms tested primary hypothesis of differential brain-behavior relationships between men and women. RESULTS: Final sample included 10,286 participants. Lower BPF was associated with worse performance (p's<0.001) in men and women; association was smaller for women than men for processing speed (ßetaWomen-Men=-0.044, 95 % CI=[-0.087, -0.002], p = 0.041) and manual dexterity (ßetaWomen-Men=-0.073, 95 % CI=[-0.124, -0.023], p = 0.005). For each MRI variable, women demonstrated better neurocognitive function controlling for disease burden. DISCUSSION: Sex differences in brain metric-neurofunctional performance relationships of people with MS suggest women have higher resilience than men in the face of increased disease burden. Future work exploring mechanism is warranted.

Disparities in DMT treatment: Demographic and neurocognitive differences between MS patients currently treated versus not treated with disease-modifying therapies.

BACKGROUND: Current treatment guidelines recommend consideration of disease-modifying therapy (DMT) for all multiple sclerosis (MS) patients, but barriers to access have begun to be identified. In particular, prior studies have found that people with higher education have better access to DMTs, perhaps explained by the association of higher education with higher income. And while the majority of people with MS are women, being male is also associated with higher income. These factors argue for the need to better understand whether there are differences in DMT uptake based on sex and education. Finally, in addition to well-documented benefits of DMTs for slowing disease progression, there is growing evidence to suggest benefits of DMTs for cognitive functioning. OBJECTIVE: Determine whether rates of DMT treatment differ based on education and sex. Secondarily, we investigate whether neurocognitive test performance differs in treated versus not treated groups. METHODS: In cross-sectional data, mixed effects linear regression evaluated differences in education and sex of those treated versus not treated with DMTs. Models included the following predictors: age, disease duration, MS subtype, sex/education, disability, atrophy, and T2 lesion volume. Propensity score weights were extracted to obtain unbiased estimates of the relationship between DMT status and each outcome of interest. The same models evaluated performance differences between groups on an iPad-based processing speed test (PST) and manual dexterity test (MDT). RESULTS: Controlling for covariates, individuals with less education (OR=1.09, 95 % CI=[1.03, 1.14], p = 0.003) and women (OR=0.80, 95 % CI=[0.72, 0.90], p < 0.001) were less likely to be currently treated with DMTs. Small effect size association was shown for DMT treatment with better performance on PST (beta=0.09, CI=[0.06, 0.13], p < 0.001) and MDT (beta=0.05, CI=[0.01,0.08], p = 0.011). CONCLUSIONS: Women and people with lower education had a lower likelihood of being currently treated with DMTs. After controlling for all relevant variables, an independent (small) association of DMT treatment to better performance on tests of processing speed and fine motor dexterity was found. Reasons for disparities remain to be investigated in future work, and may include employment status, health insurance coverage, or sex differences in risk tolerance.

Can the Montreal Cognitive Assessment and Mini-Mental State Examination detect cognitive decline in elderly patients with essential tremor?

Objective Given the association between essential tremor (ET) and higher rates of cognitive decline, assessing this decline is an important element of research and clinical care. The Mini-Mental State Examination (MMSE) and the Montreal Cognitive Assessment (MoCA) are two broad, brief measures that are widely used to monitor cognitive impairment in various neurological disorders. We sought to determine the relative ability of the MoCA and the MMSE to detect cognitive decline in elderly patients with ET. Methods We administered a neuropsychological battery to an ET cohort every 18 months over 4.5 years. We defined the gold standard for change in cognition as the change in neuropsychological test scores over consecutive evaluations and applied the Reliable Change Method to detect meaningful increases/decreases in test scores. We performed receiver operating characteristics (ROC) analysis to quantify the area under the curve (AUC) and compare the ability of the MoCA and the MMSE to detect cognitive decline. Results The AUCs for the MoCA and the MMSE did not differ significantly at any interval or when all intervals were pooled for analysis. Across all intervals, the ability of the MoCA and the MMSE to detect cognitive decline was consistently fair or poor. Conclusions We found that the ability of the MoCA and the MMSE to detect cognitive decline in ET patients over 18-month intervals is inadequate. Unchanged scores on the MoCA and the MMSE in ET over time should be approached with caution. We propose that these screening tools should be supplemented with additional neuropsychological tests.

Prevalence of substance use and mental health problems among transgender and cisgender U.S. adults: Results from a national probability sample.

Transgender individuals experience numerous health disparities relative to cisgender individuals. However, most transgender-health studies have focused on convenience samples with limited generalizability. This study utilized data from the 2016-2018 TransPop Study, the first national probability sample of transgender adults (n=274) with a cisgender comparison sample (n=1162). Using multivariable logistic regression, adjusted for demographics, we compared the prevalence of hazardous drinking, problematic drug use, serious psychological distress, suicidality, and non-suicidal self-injury between transgender and cisgender individuals and among transgender men (n=78), transgender women (n=120), and transgender nonbinary individuals (n=76). Among transgender individuals, 28.2% (95%CI 21.2-35.2) and 31.2% (95%CI 23.8-38.7) reported hazardous drinking and problematic drug use, respectively; 44.4% (95% CI 35.8-53.0) reported recent suicidal ideation, 6.9% (95% CI 2.3-11.5) reported a recent suicide attempt, and 21.4% (95% CI 14.5%-28.4%) reported recent non-suicidal self-injury. In their lifetime, 81.3% (95%CI 75.1-87.5) of transgender respondents had suicidal ideation, 42.0% (95%CI 34.2-49.8) had attempted suicide, and 56.0% (95% CI 48.2-63.8) reported non-suicidal self-injury. Most (81.5%; 95%CI 75.5-87.5) had utilized formal mental health care and 25.5% (95%CI 18.5-32.4) had sought informal mental health support. There were no differences in alcohol or drug-use outcomes between transgender and cisgender adults. Compared to cisgender adults, transgender adults had higher odds of serious psychological distress (aOR=3.1; 95%CI 1.7-5.7), suicidal ideation (recent: aOR=5.1, 95%CI 2.7-9.6); lifetime: aOR=6.7, 95%CI 3.8-11.7), lifetime suicide attempts (aOR=4.4, 95%CI 2.4-8.0), and non-suicidal self-injury (recent: aOR=13.0, 95%CI 4.8-35.1); lifetime: aOR=7.6, 95%CI 4.1-14.3). Transgender nonbinary adults had the highest odds for all outcomes, including substance use outcomes. Findings from these national probability samples support those of earlier convenience-sample studies showing mental health disparities among transgender adults relative to cisgender adults, with nonbinary individuals at highest risk. These findings also highlight variations in risk across sub-groups of transgender individuals.

Association of Subjective Cognitive Decline With Progression to Dementia in a Cognitively Unimpaired Multiracial Community Sample.

BACKGROUND AND OBJECTIVES: This prospective study seeks to examine the utility of subjective cognitive decline (SCD) as a marker of future progression to dementia in a community-based cohort of non-Latinx White, non-Latinx Black, and Latinx individuals. Debate surrounds the utility of SCD, the subjective perception of decline in one's cognition before such impairment is evident in traditional neuropsychological assessments, as an early indicator of impending Alzheimer disease. Unfortunately, most studies examining SCD have been conducted in non-Latinx White samples and commonly exclude groups of individuals shown to be most vulnerable to dementia. METHODS: Participants were enrolled into this cohort study from the Washington Heights-Inwood Columbia Aging Project if they were cognitively unimpaired, had baseline measurement of SCD, and self-identified as non-Latinx White, non-Latinx Black, or Latinx. SCD was measured as a continuous sum of 10 items assessing cognitive complaints. Competing risk models tested the main effects of baseline SCD on progression to dementia. Models were adjusted for age, sex/gender, years of education, medical comorbidity burden, enrollment cohort, and baseline memory test performance with death jointly modelled as a function of race/ethnicity. RESULTS: A total of 4,043 (1,063 non-Latinx White, 1,267 non-Latinx Black, and 1,713 Latinx) participants were selected for this study with a mean age of 75 years, 67% women, and with a mean follow-up of 5 years. Higher baseline SCD was associated with increased rates of incident dementia over time in the full sample (hazard ratio [HR] 1.085, CI 1.047-1.125, p < 0.001) and within Latinx (HR 1.084, CI 1.039-1.130, p < 0.001) and non-Latinx Black individuals (HR 1.099, CI 1.012-1.194, p = 0.024). DISCUSSION: Overall results of this study support SCD as a prodromal marker of dementia in a multiracial community sample, and in Latinx and non-Latinx Black individuals in particular. Because models examining the risk of dementia were adjusted for baseline memory test performance, the results support the idea that SCD, a subjective reflection of one's own current cognitive functioning, contributes information above and beyond standard memory testing. Current findings highlight the importance of carefully evaluating any memory concerns raised by older adults during routine visits and underscore the potential utility of screening older adults for SCD.

High school quality is associated with cognition 58 years later.

We leveraged a unique school-based longitudinal cohort-the Project Talent Aging Study-to examine whether attending higher quality schools is associated with cognitive performance among older adults in the United States (mean age = 74.8). Participants (n = 2,289) completed telephone neurocognitive testing. Six indicators of high school quality, reported by principals at the time of schooling, were predictors of respondents' cognitive function 58 years later. To account for school-clustering, multilevel linear and logistic models were applied. We found that attending schools with a higher number of teachers with graduate training was the clearest predictor of later-life cognition, and school quality mattered especially for language abilities. Importantly, Black respondents (n = 239; 10.5 percentage) were disproportionately exposed to low quality high schools. Therefore, increased investment in schools, especially those that serve Black children, could be a powerful strategy to improve later life cognitive health among older adults in the United States.