Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Jun 10.
Published in final edited form as: Biomark Neuropsychiatry. 2025 Sep 29;13:100134. doi: 10.1016/j.bionps.2025.100134

Associations of social and genetic background variables to neuro-cognitive biomarkers of psychosis

Parker V Jarrett a, Kiernan L O’Mara a, Audrey L Berardi a, Kelsey E Roberts a, John T Greco a, Rebekah L Trotti b, David A Parker a,c,i, Sarah K Keedy d, Godfrey D Pearlson e, Matcheri S Keshavan b, S Kristian Hill f, Elena I Ivleva g, Carol A Tamminga g, Jennifer E McDowell a, Ezra Susser h, Brett A Clementz a,i,*
PMCID: PMC13249451  NIHMSID: NIHMS2129733  PMID: 42272918

Abstract

Socioeconomic (SES) and ethno-racial factors may tilt psychosis diagnoses for persons from different backgrounds. Clinical diagnoses depend on patient and informant reports and are suspected of being susceptible to unintended bias. Diagnoses using laboratory tests are thought to be objective. Social disadvantages, however, alter brain functions related to psychosis. Race and class bias in laboratory medical diagnostics is an area of concern. To probe these issues for psychosis diagnoses, we describe relationships of SES, race/ethnicity, and ancestral genetic background to 11 integrated laboratory bio-factors that are associated with DSM categories and distinguish B-SNIP Biotypes. A series of analyses evaluated relationships of social factors and ancestry-related genetic background to those bio-factors: (i) canonical correlation revealed that SES and race (a social construct) are strongly associated (r2=.305) with cognitive performance and measures of brain physiology (prominently ERP magnitudes); genetic background neither significantly added to nor altered the structure of those associations; (ii) regression models illustrated that cognitive performance, intrinsic brain activity, and ERP magnitudes are substantially to modestly predicted by SES/race/genetic background, with SES/race accounting for the most variance on cognitive performance (approximately 25 %); (iii) regardless of including SES/race in differential diagnosis models, group differences between psychosis Biotypes were largely (85 %) preserved on bio-factor scores. For DSM diagnoses, less than 11 % of psychosis group differences were preserved. These outcomes illustrate that social factors are associated with psychosis-related laboratory tests. Nevertheless, SES/race did not substantially modify differential diagnosis of psychosis Biotypes. Using laboratory tests for psychosis differential diagnosis may facilitate the usefulness of stratification approaches, aid investigations of psychosis neurobiology and environmental risk, and improve treatment selections and approaches for all persons suffering with idiopathic psychosis.

1. Introduction

Diagnosis aided by neurobiology may improve the objectivity of psychosis diagnosis and facilitate the stratification of patients to subgroups with specific treatment targets. The Bipolar-Schizophrenia Network for Intermediate Phenotypes (B-SNIP) used biomarkers from laboratory tests to identify neurobiologically distinctive psychosis subgroups termed “Biotypes.” B-SNIP psychosis Biotypes have characteristic biomarker patterns, replicate, cross-validate, and have unique treatment targets that are not easily seen with diagnostic approaches using clinical features alone. The utility of those treatment targets is an active area of investigation (Clementz et al., 2020).1 Even without biomarkers, B-SNIP Biotypes can be estimated using an adaptive clinical evaluation algorithm (Clementz et al., 2023). This algorithm, called Algorithmic Diagnostics for Efficient Prescription of Treatments (ADEPT), extends the possibility of Biotype etiological and treatment investigations to clinical and research settings without laboratory infrastructure.

Laboratory measures are typically considered more objective than clinical evaluation for medical diagnosis. Measures of brain structure and function are influenced by constitution, including genetics, but also may be influenced by the environment within which a person develops and lives. Social disadvantage may impact even laboratory-based diagnostic practices and clinical outcomes (Benjamin et al., 2024; Turcotte et al., 2024). Sociodemographic disadvantage can shape neurobiology through multiple pathways including chronic stress, differential exposure to environmental toxins, and limited access to cognitive enrichment and educational opportunities (Hackman et al., 2010; Yu et al., 2024). Lower childhood SES is linked to altered cognitive development and cognitive performance later in life, a finding observed in diverse global contexts (Sheehy-Skeffington, 2020; Bertola et al., 2021). Inequalities driven by structural racism disproportionately funnel these adversities toward certain ethno-racial groups, potentially biasing measurements anchored in biology (Benjamin et al., 2024). It is critical to investigate whether laboratory-based diagnostic biomarkers reflect these social disparities, which could confound their interpretation and application. This report explores possible associations between social and genetic background features and B-SNIP Biotype and DSM diagnoses among persons with psychosis.

Socioeconomic (SES) and ethno-racial groups are differentially exposed to modifiers of disease risks (Anglin et al., 2023; van der Ven & Susser, 2023). In the United States, psychosis risk is increased for persons racialized as Black or experiencing social class disadvantages, like fewer educational opportunities, housing and health care uncertainty, or exposure to race- and environment-related traumas (Choi et al., 2012; Jester et al., 2023). Black and White persons differ in health care access and neighborhood of residence, which influence types and timing of treatments received (Misra et al., 2022; van der Ven & Susser, 2023). Black persons have fewer desirable pathways to health care and less favorable psychiatric outcomes (Li et al., 2011). Individuals from lower SES families, regardless of skin color, also fare worse than their financially stable counterparts (Merritt-Davis & Keshavan, 2006).

Van Der Ven & Susser (van der Ven & Susser, 2023) propose that in the United States, structural racism contributes to a higher incidence of schizophrenia among self-identified Black than White individuals. Two epidemiologic studies examined Black-White differences in the treated incidence of schizophrenia, and both found higher rates among Black people (Akinhanmi et al., 2018; Bresnahan et al., 2007; Schwartz & Blankenship, 2014). These incidence findings are indirect indicators of higher diagnostic rates of schizophrenia among Black Americans, who also have a higher prevalence of schizophrenia (Nagendra et al., 2020). Higher prevalence reflects duration and incidence and may be related to the Black-White differences in treatment after onset. Given reports of more psychotic-like experiences among Black than White people, higher incidence may account for higher prevalence (Misra et al., 2022), despite the influence of duration on prevalence.

Structural racism also may influence the results of incidence studies. DSM-type diagnoses are based on patient and informant report, and the interpretation of those reports by clinical staff, which leaves room for interpretation. Diagnostic inflation could occur if clinicians are more likely to diagnose schizophrenia in Black than White persons due to stereotypes, implicit bias, or difficulties empathizing with persons from different cultural, ethno-racial, and socioeconomic backgrounds (Faber et al., 2023). Diagnostic deflation could occur if Black people with psychosis are less likely to be identified as having schizophrenia than White people, making it more difficult for Black people to access treatment centers (Delphin-Rittmon et al., 2015). Furthermore, the wide-spread misdiagnosis of bipolar disorder as schizophrenia in Black patients likely contributes to the confounding of DSM diagnoses with racial identity (Medina et al., 2024).

Alternatively, B-SNIP psychosis Biotypes were developed based on laboratory tests. Are biases related to socioeconomic or ethno-racial factors related to such laboratory-based psychosis diagnoses? Neuroscience research shows the effects of social disadvantage on multiple brain measures of relevance to psychosis. Kirschner et al. (Kirschner et al., 2020) reported an association between lower SES and decreased brain volumes. Individuals with cognitive or social struggles are more likely to experience reduced neural reactivity when presented with emotional stimuli (Trotti et al., 2023). Developing in low SES environments modifies psychological well-being and cognitive and emotional progression (Hackman et al., 2010). Lower SES is also associated with less diverse educational opportunities (Hahn & Truman, 2015), environmental exposures that influence cognitive development and performance (Bertola et al., 2021; Sheehy-Skeffington, 2020; Sosu & Schmidt, 2022; Yu et al., 2024), and may influence cellular biology (Turcotte et al., 2024). The effects of low family SES on one’s neurobiology can extend beyond early life development (Tomassi et al., 2017). These effects disproportionately affect Black persons in the US, partially independent of SES (Delphin-Rittmon et al., 2015; Faber et al., 2023), and perhaps influence laboratory-based differential psychosis diagnosis.

We compared the relationship of different measures of social advantage/disadvantage to cognitive and physiological features that define psychosis Biotypes and are related to DSM-type psychosis diagnosis. SES disproportionately disadvantages certain racial and ethnic groups in the United States, so there can be a conflation of race (a social construct based on self-report and perception by others), ethnicity (shared culture) and SES (educational, occupational, and financial attainment), but each should be probed as contributors (Faber et al., 2023; Nagendra et al., 2020; Sarullo et al., 2024). The genetic background (cross-generational genetic similarity) upon which these sociodemographic and ethno-racial factors are manifested also should be considered. This project probes whether SES/self-identified race/ethnicity/genetic background are related to laboratory measures used to investigate etiological and neurobiological correlates of psychosis. We tested whether social characteristics account for or modify the ability to differentially distinguish laboratory-based (B-SNIP Biotypes) or (DSM-type) psychosis diagnoses.

2. Methods

2.1. Subjects

Subject recruitment, interviews, and laboratory data collection were completed at B-SNIP sites (Tamminga et al., 2013). Recruitment occurred in Athens, GA (University of Georgia), Baltimore, MD (Maryland Psychiatric Research Center), Boston, MA (Beth Israel Deaconess Medical Center), Chicago, IL (University of Illinois-Chicago and University of Chicago), Dallas, TX (UT Southwestern Medical Center), Detroit, MI (Wayne State University), and Hartford, CT (Institute of Living). All interviews and laboratory data collections were completed at those locations. The Institutional Review Board at those institutions approved the projects; participants provided informed consent prior to involvement.

Clinically stable outpatients were administered the Structured Clinical Interview for DSM diagnosis (DSM-IV-TR; (American-Psychiatric-Association, 2000). Psychosis cases were limited to schizophrenia, schizoaffective disorder, and bipolar I disorder with psychosis because these are diagnoses with the highest prevalence in most clinical settings. Healthy persons were administered the non-patient version of the SCID. Participants were rated on the Birchwood Social Functioning (Birchwood et al., 1990), Montgomery-Asberg Depression Rating (Montgomery & Asberg, 1979), Positive and Negative Syndrome (Kay et al., 1987), and Young Mania Rating (Young et al., 1978) scales.

Psychosis cases (a maximum n = 1489 were available for this project and included participants from both B-SNIP 1 & 2) were drawn from academic and community mental health centers, small towns with large universities, large cities, inner cities, rural regions, affluent and less affluent areas. Healthy participants (a maximum n = 685) were recruited from those same communities. Psychosis and healthy participants were not matched on propensity to access clinical care, nor on social or genetic background characteristics to the populations from which they were recruited. This imposes limits on interpretation (see Discussion).

Data were collected on individual and family-of-origin socioeconomic status (SES; measured by the Hollingshead two-factor index, which incorporates level of education and occupational success; high scores indicate lower SES). Self-identified race and ethnicity were available for these participants. Social assignment based on others’ perceptions and additional factors are important for understanding the relationships between race and health disparities (White et al., 2020), but such information was not available. Three self-identified race/ethnicity variables were constructed for subsequent analyses: Black (1) or not (0), Asian (1) or not (0), and Hispanic (1) or not (0), a customary approach when constructing dummy variables for multivariate models (Knapp, 1978) — the below analyses estimated their individual and combined contributions to the questions of interest. Table 1 shows demographic and clinical information by Biotype (Supplementary Table 1 shows this information by DSM diagnosis). We previously showed that medications do not account for group differences on biomarkers (Clementz et al., 2022; Parker et al., 2025).

Table 1.

Demographic and Clinical Characteristics By Biotype.

Characteristic Overall
N = 2174		 Biotype 1
N = 495		 Biotype 2
N = 480		 Biotype 3
N = 514
Mean Age (SD) 37 (12) 38 (13) 39 (12) 35 (12)
Mean SES (SD) 43 (16) 50 (14) 40 (12) 43 (15)
 Proband
Family 40 (16) 44 (16) 43 (16) 37 (16)
Mean Symptom Score (SD) 16 (6) 16 (6) 17 (6) 14 (6)
 PANSS Positive
PANSS Negative 15 (6) 15 (6) 16 (6) 14 (6)
MADRS 13 (10) 13 (10) 13 (10) 12 (9)
YMRS 9 (7) 9 (7) 10 (8) 9 (7)
Birchwood Social 133 (27) 121 (24) 119 (24) 130 (24)
Ethnicity 87 % 87 % 87 % 88 %
 Not Hispanic
Hispanic 13 % 13 % 13 % 12 %
Race 34.9 % 49.5 % 41.9 % 22.8 %
 Black
American Indian 0.3 % 1.0 % 0.2 % 0.2 %
Asian 4.6 % 3.6 % 1.5 % 3.7 %
White 52.8 % 38.8 % 48.1 % 65.2 %
Multiracial 4.5 % 4.6 % 5.2 % 4.7 %
Other 2.9 % 2.4 % 3.1 % 3.5 %
Open in a new tab

2.2. Genetic background quantification

Genotyping details are in the Supplementary Methods. As was described previously for B-SNIP data (Alliey-Rodriguez et al., 2019), principal component analysis (PCA) was performed on genotypes of the individuals. The scree plot of this PCA identified three possibly useful genetic background components across individuals. The genetic principal components are dimensional scales and were used as such in subsequent analyses (Dauda et al., 2023; Lewis et al., 2022). This PCA quantified axes of genetic variation that characterize continental ancestry (e.g., African, European, East Asian). These components represent continuous dimensions of genetic similarity based on frequency of genetic variants in relation to the 1000 Genomes reference panel.

2.3. Bio-factor scores and Biotypes determination

Papers on the individual laboratory tasks provide data collection and analysis details (Ethridge et al., 2014; Hill et al., 2013; Huang et al., 2022; Parker et al., 2020; Parker et al., 2021). Details of biomarker quantification and biotyping procedures are in Clementz et al. (2022) and Parker et al. (2025), including extensive supplementary methods. Laboratory measures used to obtain bio-factor scores (what we call integrated biomarker variables) were (i) the Brief Assessment of Cognition in Schizophrenia (BACS; (Keefe et al., 2004), (ii) Stop-Signal Task (SST; (Lipszyc & Schachar, 2010), (iii) pro- and anti-saccade tasks (saccades; (Hallett & Adams, 1980; McDowell & Clementz, 2001), (iv) auditory paired stimuli and oddball tasks (ERPs; (Freedman et al., 1987; Polich, 2007), and (v) the 9–10 s inter-pair interval of the paired stimuli task (intrinsic EEG activity or IEA; (Thomas et al., 2019).

Following quantification of the tasks and paradigms, as described in relevant publications, PCA was performed within each laboratory measurement. This process produced 11 variables called bio-factors (Clementz et al., 2022; Parker et al., 2025) — BACS, antisaccade, SST, intrinsic EEG activity, paired-stimuli ongoing activity, oddball ongoing activity, paired-stimuli ERPs, oddball ERPs, Frontal P300, paired-stimuli S2 response, and saccade latency. These bio-factors were used to obtain B-SNIP psychosis Biotype designations via numerical taxonomy procedures described in previous B-SNIP papers (Clementz et al., 2022; Parker et al., 2025). Psychosis cases have a Biotype (Biotype-1, Biotype-2, or Biotype-3) and a DSM diagnosis (schizophrenia, schizoaffective disorder, or bipolar disorder with psychosis).

2.4. Data Analyses

Three approaches were used to probe the relationships between SES-race-genetic background and bio-factor scores: (i) canonical correlation using all variables simultaneously, (ii) linear regressions between social/genetic background variables and individual bio-factors, and (iii) analysis of variance comparing the effects of social variables on differentiations of Biotype and DSM groups. This multi-step approach (i) identified the overall bi-directional multivariate relationships between variable sets using canonical correlation, (ii) quantified the specific contribution of individual predictors to individual outcomes using linear regression, and (iii) tested the robustness of the primary diagnostic groupings to these sociodemographic factors via ANOVA. Those procedures and their purposes are described below.

2.4.1. Canonical correlation

Canonical correlation (CCA) probed the bidirectional relationships between SES, self-identified race, ethnicity, and genetic background on one side of the equation to B-SNIP bio-factors on the other side of the equation2 CCA identifies the relationship between two sets of variables by maximizing correlations between linear combinations of the sets, one on each side of the equation (Levine, 1977; Wang et al., 2020). CCA is the general procedure for investigating the relationship between two variable sets, and for which other parametric significance tests are special cases (Knapp, 1978). CCA creates correlated pairs of latent variates. Each pair is independent (additional variate pairs are calculated on the remainder after accounting for variance of the preceding variates) and composed of weighted sums of one variable set that maximally correlate with the weighted sums of the other variable set. Like PCA, interpretation of the latent variates is through the loadings of individual measures with the latent structure. The loadings are correlations of each individual variable in a set with that set’s latent variate. This approach describes the association of each variable with its latent variate rather than interpreting the individual weights used to construct the latent variate (Levine, 1977).

Two CCAs were conducted: (i) using only SES (parental and participant separately) and self-identified race/ethnicity (Black or not, Asian or not, Hispanic or not), and (ii) another adding the first three genetic background PCA components to SES and self-identified race/ethnicity. This allowed for comparisons between component structures and canonical loadings both without and with genetic background information. On one side of the equation, the first CCA had 5 variables and the second CCA had 8 variables. Both CCAs had the 11 bio-factors on the other side of the equation.

The multivariate nature of CCA does not require multiple testing within a CCA analysis, although multiple testing across the two CCA analyses does and was accounted for by Bonferroni correction with the threshold for significance set at p = .005 (.01 divided by 2 CCA analyses). To evaluate the stability of the canonical variate structures, and any difference in component structures between the two CCA solutions, we implemented a jackknife procedure. We conducted analyses with a randomly selected 80 % of the sample and repeated this procedure 10,000 times. We then computed the 99 % confidence intervals across those samples for percent of variance accounted for by each CCA and for the component loadings to determine if adding genetic background changed the variance accounted for or component structures of the bivariate relationships. Smaller confidence intervals also demonstrate the stability of the CCAs.

2.4.2. Regression

The CCAs describe the bi-directional relationships between two sets of variables. They do not describe the individual relationships between the SES/race/genetic background and individual bio-factor scores. Based on the outcomes of the CCAs (Wang et al., 2020), regression models were constructed using the lm() function in R between SES/race/genetic background variables with beyond zero canonical loadings. These analyses describe the maximum shared variance between individual qualifying SES, self-identified race, and genetic background variables and the 11 individual bio-factors. For statistical significance, the Holm-Bonferroni procedure (Holm, 1979) was used to maintain the family-wise alpha at.01.

2.4.3. Analysis of Variance

The above analyses explore the relationships between social and genetic background information and bio-factors, but they do not assess how such relationships affect Biotype or DSM differentiations on the bio-factors. This is especially relevant for Biotypes, since the categories are created from the bio-factors. For every social variable that significantly contributed to bio-factor scores, we performed Biotype (Biotype-1, Biotype-2, Biotype-3) by DSM (schizophrenia, schizoaffective, bipolar) ANOVAs. These models also tested for possible interactions between Biotype and DSM diagnostic status and social factors. Effects were considered of possible relevance at p-values < .01, although actual p-values are displayed in the relevant table.

3. Results

3.1. Canonical Correlation Analyses

3.1.1. CCAs without and with genetic background

Fig. 1 displays the overall CCA outcomes, described in this and the next three subsections. Supplementary Table 2 shows the correlation matrix for all measures used in these analyses. Latent variate pairs were retained for analysis if their correlation was significant (p < .01 divided by 2, or p < .005). This criterion was met for the first two CCA components without genetic background: CCA1 F(55, 5331)= 9.36, p < .001, r2= .305; CCA2 F(40, 4370)= 1.86, p < .001, r2= .036. This criterion was also met for the first two CCA components with genetic background: CCA1 F(88, 7537)= 6.52, p < .001, r2 = .324; CCA2 F(70, 6707)= 1.50, p = .004, r2= .043.

Fig. 1.

Fig. 1.

Open in a new tab

Canonical correlations between B-SNIP bio-factors (lower right bar chart) and SES/race/genetic background (upper left bar chart). The loadings, indicated by deviations of the bars from 0 (with 99 % confidence intervals), indicate the strength of the correlation between every individual variable and the latent canonical variates. Genetic background PCs add 1.9 % to variance accounted for when predicting bio-factor scores. The canonical loading for all common variables is not appreciably altered by adding genetic background to the model. This is appreciated by comparing the heights of the solid gray bars to the heights of the corresponding blue-gray bars.

3.1.2. Stability of CCA solutions

Probes of CCA stabilities illustrate the robustness of especially the first canonical variates. For CCA1, across the 10,000 jackknifed samples, 100 % of the resampled outcomes were significant for models without or with genetic background. The range of variance accounted for (99 % confidence intervals of r2 values) by CCA1 was reasonably tight for models both without (.281–.333) and with (.299–.353) genetic background. For CCA2, across the resamples, only 83 % without and 57 % with genetic background met the significance threshold. The 99 % confidence intervals of variance accounted for by CCA2 were similarly distributed (without genetic background =.0275–.0545; with genetic background =.0342–.0616). Given CCA2’s low variance accounted for and questionable stability, it was not further considered.

3.1.3. Comparison of CCA solutions without and with generic background

The next issue is whether the CCA solutions without and with genetic background are structurally similar. Fig. 1 shows the CCA1 loadings for the two models. Adding genetic background increased r2 of CCA1 by 1.9 %, which was not a significant increment (the two r2 values are captured by the 99 % confidence intervals, provided in the above subsection). The canonical loadings are also statistically indistinguishable between models without and with genetic background (the magnitudes of the paired bars are nearly identical, and they are within the 99 % confidence intervals for every loading).

Every participant also has a score for both sides of the canonical correlation equation, that is, a score that is a weighted average of the SES/self-identified race side of the equation and a similar weighted score for the bio-factors side of the equation (see Fig. 2 for the bivariate distribution without genetic background). The intra-individual correlations between those corresponding CCA1 variate scores of models without and with genetic background models is.97 for the SES/self-identified race weighted scores and.99 for the bio-factor weighted scores. This outcome suggests that the models have the same structure regardless of genetic background information.

Fig. 2.

Fig. 2.

Open in a new tab

Plot of the SES/self-identified race (y-axis) versus the Bio-factors (x-axis) canonical variates. Biotype and healthy participant centroids with standard deviation ellipsoids are also displayed.

3.1.4. Nature of CCA associations

The direction and magnitude of the canonical loadings support interpretation of the latent variates. To aid interpretations, Supplementary Figure 1 (SES) and Supplementary Figure 2 (Black, Asian, and Hispanic) show social differences across B-SNIP bio-factors. Fig. 2 shows the scatterplot of the SES/self-identified race variate against the bio-factor variate. As derived from Fig. 1, lower scores on the former variate are most clearly associated with worse SES and being Black. Alternatively, higher scores on the bio-factor variate are most clearly associated with better cognitive performance (especially BACS and Antisaccade bio-factors) and larger event-related potential (ERP) magnitudes, which indicate better ability to respond to salient stimuli. The location of Biotypes and healthy centroids (and standard deviation ellipsoids) are also displayed. Supplementary Figure 3 shows this same plot stratified by DSM diagnoses.

3.2. Regression Models

These models evaluated the individual relationships between SES, self-identified race, or genetic background and every individual bio-factor. Individual and parental SES, Black and Asian self-identified race, and genetic PC1 were used in these models. The Hispanic, PC2, and PC3 variables made no significant contributions in the CCAs, so were not included (see (Wang et al., 2020)). Being the most general characteristics, individual and family SES were initially added to the social models, followed by the addition of self-identified race, although the results were the same if self-identified race was added first. Genetic background was analyzed separately. Fig. 3 summarizes the outcomes; Supplementary Table 3 summarizes relevant statistical outputs.

Fig. 3.

Fig. 3.

Open in a new tab

Regression models relating socioeconomic status (SES), self-identified race (Race), or genetic background (gPC) to individual bio-factors. Models were constructed to maximize the direct relationships between those variables and every bio-factor. The light gray bars show the SES associations, the dark gray bars show the SES plus self-identified race associations, and the light blue bars show the genetic background associations. Stars under bars along the x-axis indicate statistically significant associations after modified Bonferroni adjustment.

BACS, Antisaccade, stop signal task (SST), intrinsic EEG activity (IEA), and ERP magnitudes had statistically significant associations with SES, SES plus self-identified race, and PC1. The frontal P300 complex had significant associations only with SES and SES plus self-identified race. The ongoing EEG activity and latency bio-factors showed no significant associations in these models (see Fig. 3). None of the other bio-factors had large proportions of variance account for by any of the included variables. Only the BACS showed substantially more than 10 % of shared variance between social or genetic background variables and bio-factor scores. Paired ERP, Oddball ERP, and Antisaccade bio-factors had the next highest shared variances with social or genetic background variables (in the 3–10 % range). IEA, SST, Paired S2 ERP, and the Frontal P300 complex shared variance with social and genetic background variables in 0.5–4.5 % range. In general, social variables shared larger proportions of variance with bio-factors than did genetic background (see Fig. 3 and Supplementary Table 3).3

3.3. Biotypes and DSM Effects via ANOVAs

The above analyses do not address whether incorporating SES or self-identified race changes the differences between diagnostic groups on bio-factors. Patterns of differences on bio-factors by Biotype and DSM groups have appeared in other publications (Clementz et al., 2022; Parker et al., 2025). This section specifically addresses whether group differences are modified by the inclusion of SES or self-identified race in statistical models. We performed Biotype (Biotype-1, Biotype-2, Biotype-3) by DSM (schizophrenia, schizoaffective, bipolar) ANOVAs that included an additional socio-demographic factor: participant SES, family SES, or self-identified race (Black or Asian). See Supplementary Table 4 for number of participants in self-identified race by SES cells. The Hispanic variable was excluded because it made no contribution in the CCAs.

Previous comparisons involving healthy persons showed only main effects of social variables and Healthy versus Psychosis on bio-factors (see Fig. 1 and Supplementary Figures 1 and 2). There were no significant interactions. The following analyses did not include the healthy group but specifically evaluated whether (i) diagnostic group differences were maintained when including social variables in the ANOVA models, and (ii) there are interactions between social factors and diagnosis.

In the simple models without social variables, Biotypes differ on all bio-factors (see Table 2), and DSM diagnoses differ on all bio-factors except measures of background EEG activity and saccade latency. Social variables change the nature of these significant main effects, especially for DSM diagnoses (see Table 2). For Biotypes, 37 of a possible 44 diagnostic group differences (84.1 %) were maintained after including social variables in the models. Including Asian in the models had the most significant impact, with SST, Frontal P300, paired stimuli S2 response, and Latency no longer significant. Adding Personal or Family SES also led Frontal P300 (P-SES) and Latency (P-SES and F-SES) to no longer significantly differentiate groups. For DSM, only three of 44 diagnostic group differences (6.8 %) were significant after including social variables. If the total is restricted only to the bio-factors that differentiated in simple DSM models, three of 28 tests survived adding social variables (10.7 %). The two bio-factors that survived in DSM comparisons are BACS (SES-F and Black) and Antisaccade (SES-F). There was only one significant interaction out of 88 total, for Oddball ongoing activity in the DSM by Black model.

Table 2.

Biotype and DSM by Socio-Demographics Effects on Bio-factors.

Bio-factors Biotype Alone
p-value		 Biotype SES-P
p-value		 Biotype SES-F
p-value		 Biotype Black
p-value		 Biotype Asian
p-value		 DSM Alone
p-value		 DSM SES-P
p-value		 DSM SES-F
p-value		 DSM Black
p-value		 DSM Asian
p-value
BACS < .001 < .001 < .001 < .001 < .001 < .001 .013 < .001 < .001 .084
 Main — .657 .514 .177 .419 — .258 .816 .656 .488
 Inter
Antisaccade < .001 < .001 < .001 < .001 .003 < .001 .035 < .001 .021 .069
 Main — .904 .273 .990 .940 — .269 .067 .155 .071
 Inter
SST < .001 < .001 < .001 < .001 .544 .001 .545 .603 .646 .076
 Main — .565 .074 .093 .372 — .502 .286 .639 .121
 Inter
IEA < .001 < .001 < .001 < .001 < .001 .902 .245 .586 .782 .789
 Main — .508 .487 .140 .092 — .211 .314 .107 .688
 Inter
PS Ongoing < .001 < .001 < .001 < .001 < .001 .974 .722 .068 .192 .352
 Main — .324 .813 .045 .661 — .870 .288 .235 .297
 Inter
OB Ongoing < .001 < .001 < .001 < .001 < .001 .947 .797 .193 .683 .388
 Main — .845 .471 .144 .834 — .851 .712 .644 .435
 Inter
PS ERP < .001 < .001 < .001 < .001 < .001 .002 .898 .151 .194 .874
 Main — .974 .248 .140 .227 — .896 .947 .281 .569
 Inter
OB ERP < .001 < .001 < .001 < .001 < .001 .009 .256 .020 .211 .720
 Main — .863 .445 .403 .046 — .304 .314 .008 .435
 Inter
Frontal P300 < .001 .036 < .001 < .001 .015 < .001 .742 .687 .595 .984
 Main — .185 .072 .984 .367 — .947 .990 .235 .839
 Inter
PS S2 < .001 < .001 < .001 < .001 .041 < .001 .019 .113 .237 .004
 Main — .397 .818 .621 .226 — .192 .307 .028 .050
 Inter
Latency < .001 .077 .018 < .001 .190 .026 .345 .864 .194 .846
 Main — .862 .812 .325 .122 — .628 .479 .896 .062
 Inter
Open in a new tab

4. Discussion

We explored social and genetic background associations to B-SNIP bio-factor scores, which are used for generating psychosis Biotypes and are related to DSM psychosis diagnoses. SES and race are related to health disparities, influence medical diagnosis and treatment access, and correlate with brain measures linked to presence or risk for serious psychiatric conditions. There are three main outcomes of this project: (i) SES, self-identified race, and genetic background are related to bio-factor scores, especially cognitive performance and EEG/ERP magnitudes; (ii) genetic background adds little beyond social information to the prediction of bio-factors; and (iii) adding SES and self-identified race to ANOVA models largely maintains bio-factor differences between B-SNIP psychosis Biotypes, but largely eliminates such differences between DSM psychosis diagnoses.

4.1. SES, Race, Genetic Background, and Bio-factors

Bi-directional associations illustrate that combined social/genetic background information share about 31 % of their variance with the 11 bio-factor scores. The consistency of this estimate over 10,000 resamples, with the 99 % confidence intervals falling within ±8.4 % of the total sample outcome on average, supports its reliability. The most substantial contributors to these associations are family and individual SES, Black or Asian self-identified race, and genetic principal component PC1 (which is highly correlated but not synonymous with African genetic similarity in these and other data), BACS scores, antisaccade performance, and auditory ERP magnitudes. This pattern illustrates what is partially known, i.e., social disadvantage is related to worse and social advantage is related to better cognitive performance and stimulus registration (Dotson & Duarte, 2020; Letang et al., 2021; Sosu & Schmidt, 2022). This result adds to the literature that the pattern is neither peculiar to nor interacts with status as a person with psychosis; the effect was the same regardless of healthy versus psychosis status.

4.2. Sociodemographics Have Highest Bio-factor Associations

A belief in genetic determinism often fosters the notion that certain ancestral groups are constitutionally more susceptible to serious psychiatric conditions. The outcomes of this project at least indirectly address this issue. First, in canonical correlation models there is a nonsignificant and negligible increment (less than 2 %) in shared variance with bio-factor scores when genetic background is added to social features. Second, adding genetic background did not alter the canonical correlation model structures. This is indicated by models without and with genetic background having nearly identical loadings and bivariate regressions when related to bio-factors scores. Third, individual social features account for more variance than individual genetic background patterns in their relationships to individual bio-factors. Bio-factors with substantial influence for differentiating psychosis Biotypes have generally larger individual associations with social variables than with genetic background. This includes (i) cognitive performance as estimated by the BACS with ~25 % shared variance with social features, antisaccade (~7.5 %), and stop-signal performance (~3.5 %), and (ii) early sensory registration fidelity as estimated by auditory ERP magnitudes (3–10 %). Only for IEA, an estimate of background brain activity, did genetic background equal social features in proportion of shared variance (~4.5 %). These results illustrate that SES and self-identified race are individually important for understanding psychosis-related neurobiology, but do not alter the need to consider ancestral genetic background in other contexts (Feero et al., 2024).

It is critical to consider these findings not as evidence of inherent biological differences between racial groups, but rather as a reflection of societal inequalities (for instance, via population stratification, where environment mimics genes). First, self-identified racial categories such as “Black” and “Asian” are not monolithic and encompass immense cultural, genetic, and experiential diversity. Second, in the United States, disparities associated with Black self-identified race are consistent with the well-documented effects of structural racism, systemic disadvantage, and differential exposure to a wide range of adversities that can alter neurocognitive function (Hackman et al., 2010; Letang et al., 2021). These include, but are not limited to, chronic stress, unequal access to healthcare resources, and psychological toll of experiencing discrimination. Finally, the patterns associated with Asian self-identified race likely reflect different social and environmental contexts within this heterogeneous group. The term “Asian” encompasses vast genetic, cultural, and experiential diversity; its use in these analyses is a limitation of our demographic coding, not a suggestion of a biologically coherent category.

4.3. Sociodemographics Minimally Influence Psychosis Biotype Stratification

The above outcomes challenge whether B-SNIP psychosis Biotype stratification is free of sociodemographic bias. The concern with such bias is sharpened by the distribution of individuals along the sociodemographic by bio-factor functions of Fig. 2 (Biotypes) and Supplementary Figure 3 (DSM). The answer to this challenge is addressed by the final set of ANOVA models.

For psychosis Biotypes, there are no interactions between sociodemographic status and Biotype membership. With sociodemographic variables included in models, 84.1 % of bio-factor group differences are maintained for Biotypes. The other 15.9 % of bio-factor comparisons capture less important laboratory measures for psychosis Biotype stratification (Clementz et al., 2023; Clementz et al., 2022; Parker et al., 2025). Self-reported Asian identity tended to affect bio-factors differentiating features of Biotype-3, the least impaired psychosis group. This overall outcome supports the utility of B-SNIP psychosis Biotype stratifications despite sociodemographic associations with their defining laboratory measurements.

These analyses are different for DSM psychosis diagnoses. DSM diagnoses are not based on laboratory tests, but they are studied in relation to historically important neurobiological deviations. Only 7–11 % of DSM group differences on bio-factors were maintained after including socio-demographics. The two bio-factors surviving in DSM comparisons, BACS and Antisaccade, assess the most important differentiating characteristics and treatments targets for DSM psychosis diagnoses (cognitive performance, goal maintenance, inhibitory control). Nevertheless, these results illustrate that investigation of neurobiological deviations unique to a specific DSM psychosis diagnosis might usefully consider SES and self-identified race.

4.4. Considerations and Limitations

The overall pattern of results illustrates the importance of considering SES and self-identified race for probing the neurobiology of psychosis. Multiple complimentary analyses with consistent outcomes tend to support the veracity of this conclusion. Like any scientific product, there are still important issues worthy of future consideration. Probing those issues will improve our understanding of the relationships between class, race, ancestry, and the neurobiology of psychosis.

First, ERP magnitudes are related to SES and self-identified race. In the United States, it is often difficult to separate the effects of SES and race within individuals. Adding self-identified race to SES, however, at least doubled shared variance with IEA and ERP bio-factors. Perhaps they are related to implicit racial bias and accompanying chronic stress that deleteriously affect health and neural development not captured by SES alone.

Second, the B-SNIP sample has possible bias related to sample selection, since people with and without psychoses were not drawn from the same underlying population. Not all self-identified racial groups are sufficiently represented, especially as a function of socioeconomic status. B-SNIP also does not have a complete representation of the full genetic background spectrum, knowledge of the perception of participant racial identity by others (Albuja, 2018), or information on healthcare disparities experienced by the individuals. B-SNIP does not have sufficient data on the developmental, environmental, race-based traumas, or life stressors participants may have experienced. The effects of early life disadvantage and access to education and healthcare also may be usefully considered in future studies. Hence, the outcomes may not generalize to all persons across the full social and genetic background ranges. Nonetheless, the study numbers and broad geographical recruitment suggest that even allowing for some possible bias, the major results are relevant to persons ranging from early onset through midcourse to chronic idiopathic psychosis.

Supplementary Material

1
2
3
4
5

Appendix A. Supporting information

Supplementary data associated with this article can be found in the online version at doi:10.1016/j.bionps.2025.100134.

Footnotes

CRediT authorship contribution statement

Sarah K. Keedy: Methodology, Data curation. Brett A. Clementz: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. John T. Greco: Writing – review & editing. Ezra Susser: Writing – review & editing, Writing – original draft, Validation, Conceptualization. Kelsey E. Roberts: Writing – review & editing, Visualization, Validation. Jennifer E. McDowell: Writing – review & editing, Methodology, Data curation. Audrey L. Berardi: Writing – review & editing, Visualization, Validation. Carol A. Tamminga: Methodology, Funding acquisition, Data curation. Kiernan L. O’Mara: Writing – review & editing, Visualization, Investigation, Formal analysis. Parker V. Jarrett: Writing – review & editing, Writing – original draft, Visualization, Project administration, Investigation, Formal analysis, Conceptualization. Elena I. Ivleva: Methodology, Data curation. S. Kristian Hill: Methodology, Data curation. Matcheri S. Keshavan: Methodology, Formal analysis, Data curation. Godfrey D. Pearlson: Methodology, Funding acquisition, Data curation. David A. Parker: Validation, Methodology, Funding acquisition, Formal analysis, Data curation. Rebekah L. Trotti: Methodology, Investigation, Funding acquisition, Data curation, Conceptualization.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

1

There are three iterations of B-SNIP Biotypes (Clementz et al., 2016; Clementz et al., 202; Parker et al., 2025). Iterations replicated psychosis Biotypes, corrected oversights in the initial formulation, and improved efficiency to facilitate application of B-SNIP Biotypes to the clinic. Those refinements supported the ADEPT algorithm for efficient Biotype diagnosis without needing to collect the extensive B-SNIP biomarker panel (Clementz et al., 2023; Clementz et al., 2025).

2

The CCA models require every participant to have every measure, in this case every participant must have a value for 19 different variables. Rather than interpolating missing values for these analyses, we only included participants in the CCAs with complete data. This reduced the psychosis sample to 772 and the healthy sample to 395. Interpolating missing data, however, has no effect on the conclusions.

3

We also computed mediation models for the direct relationship between genetic background and each individual bio-factor score, with SES and self-reported race as possible mediators. There are difficulties with assuming the assumptions of such models are reasonably met, so the mediation outcomes are not included in the main manuscript. Nevertheless, for completeness, a description of the models, limitations, and outcomes is in Supplementary Results.

References

  1. Akinhanmi MO, Biernacka JM, Strakowski SM, McElroy SL, Balls Berry JE, Merikangas KR, Assari S, McInnis MG, Schulze TG, LeBoyer M, Tamminga C, Patten C, Frye MA, 2018. Racial disparities in bipolar disorder treatment and research: a call to action. Bipolar Disord. 20 (6), 506–514. 10.1111/bdi.12638. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Alliey-Rodriguez N, Grey TA, Shafee R, Asif H, Lutz O, Bolo NR, Padmanabhan J, Tandon N, Klinger M, Reis K, Spring J, Coppes L, Zeng V, Hegde RR, Hoang DT, Bannai D, Nawaz U, Henson P, Liu S, Gershon ES, 2019. NRXN1 is associated with enlargement of the temporal horns of the lateral ventricles in psychosis. Transl. Psychiatry 9 (1), 230. 10.1038/s41398-019-0564-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. American-Psychiatric-Association, 2000. Diagnostic criteria from DSM-IV-TR. American Psychiatric Association. [Google Scholar]
  4. Anglin DM, Espinosa A, Addington J, Cadenhead KS, Cannon TD, Cornblatt BA, Keshavan M, Mathalon DH, Perkins DO, Stone W, Tsuang M, Woods SW, Walker E, Bearden CE, Ku BS, 2023. Association of childhood Area-Level ethnic density and psychosis risk among ethnoracial minoritized individuals in the US. JAMA Psychiatry 80 (12), 1226–1234. 10.1001/jamapsychiatry.2023.2841. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Benjamin KJM, Chen Q, Eagles NJ, Huuki-Myers LA, Collado-Torres L, Stolz JM, Pertea G, Shin JH, Paquola ACM, Hyde TM, Kleinman JE, Jaffe AE, Han S, Weinberger DR, 2024. Analysis of gene expression in the postmortem brain of neurotypical black Americans reveals contributions of genetic ancestry. Nat. Neurosci 27 (6), 1064–1074. 10.1038/s41593-024-01636-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bertola L, Bensenor IM, Barreto SM, Giatti L, Moreno AB, Viana MC, Lotufo PA, Suemoto CK, 2021. Early life socioeconomic status predicts cognition regardless of education level. Eur. J. Neurol 28 (12), 3972–3978. 10.1111/ene.15042. [DOI] [PubMed] [Google Scholar]
  7. Birchwood M, Smith J, Cochrane R, Wetton S, Copestake S, 1990. The social functioning scale. The development and validation of a new scale of social adjustment for use in family intervention programmes with schizophrenic patients. Br. J. Psychiatry 157, 853–859. 〈https://www.ncbi.nlm.nih.gov/pubmed/2289094〉. [DOI] [PubMed] [Google Scholar]
  8. Bresnahan M, Begg MD, Brown A, Schaefer C, Sohler N, Insel B, Vella L, Susser E, 2007. Race and risk of schizophrenia in a US birth cohort: another example of health disparity? Int J. Epidemiol 36 (4), 751–758. 10.1093/ije/dym041. [DOI] [PubMed] [Google Scholar]
  9. Choi MR, Eun HJ, Yoo TP, Yun Y, Wood C, Kase M, Park JI, Yang JC, 2012. The effects of sociodemographic factors on psychiatric diagnosis. Psychiatry Invest. 9 (3), 199–208. 10.4306/pi.2012.9.3.199. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Clementz BA, Chattopadhyay I, Hill SK, McDowell JE, Keedy SK, Parker DA, Trotti RL, Ivleva EI, Keshavan MS, Gershon ES, Pearlson GD, Tamminga CA, Gibbons RD, 2025. Cognitive performance and differentiation of B-SNIP psychosis biotypes: algorithmic diagnostics for efficient prescription of treatments (ADEPT) - 2. Biomark. Neuropsychiatry 12, 100117. 10.1016/j.bionps.2024.100117. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Clementz BA, Chattopadhyay I, Trotti RL, Parker DA, Gershon ES, Hill SK, Ivleva EI, Keedy SK, Keshavan MS, McDowell JE, Pearlson GD, Tamminga CA, Gibbons RD, 2023. Clinical characterization and differentiation of B-SNIP psychosis biotypes: algorithmic diagnostics for efficient prescription of treatments (ADEPT)-1. Schizophr. Res 260, 143–151. 10.1016/j.schres.2023.08.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Clementz BA, Parker DA, Trotti RL, McDowell JE, Keedy SK, Keshavan MS, Pearlson GD, Gershon ES, Ivleva EI, Huang LY, Hill SK, Sweeney JA, Thomas O, Hudgens-Haney M, Gibbons RD, Tamminga CA, 2022. Psychosis biotypes: replication and validation from the B-SNIP consortium. Schizophr. Bull 48 (1), 56–68. 10.1093/schbul/sbab090. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Clementz BA, Trotti RL, Pearlson GD, Keshavan MS, Gershon ES, Keedy SK, Ivleva EI, McDowell JE, Tamminga CA, 2020. Testing psychosis phenotypes from Bipolar-Schizophrenia network for intermediate phenotypes for clinical application: biotype characteristics and targets. Biol. Psychiatry Cogn. Neurosci. Neuroimaging 5 (8), 808–818. 10.1016/j.bpsc.2020.03.011. [DOI] [PubMed] [Google Scholar]
  14. Dauda B, Molina SJ, Allen DS, Fuentes A, Ghosh N, Mauro M, Neale BM, Panofsky A, Sohail M, Zhang SR, Lewis ACF, 2023. Ancestry: how researchers use it and what they mean by it. Front Genet 14, 1044555. 10.3389/fgene.2023.1044555. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Delphin-Rittmon ME, Flanagan EH, Andres-Hyman R, Ortiz J, Amer MM, Davidson L, 2015. Racial-ethnic differences in access, diagnosis, and outcomes in public-sector inpatient mental health treatment. Psychol. Serv 12 (2), 158–166. 10.1037/a0038858. [DOI] [PubMed] [Google Scholar]
  16. Dotson VM, Duarte A, 2020. The importance of diversity in cognitive neuroscience. Ann. N. Y Acad. Sci 1464 (1), 181–191. 10.1111/nyas.14268. [DOI] [PubMed] [Google Scholar]
  17. Ethridge LE, Soilleux M, Nakonezny PA, Reilly JL, Hill SK, Keefe RS, Gershon ES, Pearlson GD, Tamminga CA, Keshavan MS, Sweeney JA, 2014. Behavioral response inhibition in psychotic disorders: diagnostic specificity, familiality and relation to generalized cognitive deficit. Schizophr. Res 159 (2–3), 491–498. 10.1016/j.schres.2014.08.025. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Faber SC, Khanna Roy A, Michaels TI, Williams MT, 2023. The weaponization of Medicine: early psychosis in the black community and the need for racially informed mental healthcare. Front Psychiatry 14, 1098292. 10.3389/fpsyt.2023.1098292. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Feero WG, Steiner RD, Slavotinek A, Faial T, Bamshad MJ, Austin J, Korf BR, Flanagin A, Bibbins-Domingo K, 2024. Guidance on use of race, ethnicity, and geographic origin as proxies for genetic ancestry groups in biomedical publications. Nat. Genet 56 (4), 555–556. 10.1038/s41588-024-01708-8. [DOI] [PubMed] [Google Scholar]
  20. Freedman R, Adler LE, Gerhardt GA, Waldo M, Baker N, Rose GM, Drebing C, Nagamoto H, Bickford-Wimer P, Franks R, 1987. Neurobiological studies of sensory gating in schizophrenia. Schizophr. Bull 13 (4), 669–678. 10.1093/schbul/13.4.669. [DOI] [PubMed] [Google Scholar]
  21. Hackman DA, Farah MJ, Meaney MJ, 2010. Socioeconomic status and the brain: mechanistic insights from human and animal research. Nat. Rev. Neurosci 11 (9), 651–659. 10.1038/nrn2897. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Hahn RA, Truman BI, 2015. Education improves public health and promotes health equity. Int J. Health Serv 45 (4), 657–678. 10.1177/0020731415585986. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Hallett PE, Adams BD, 1980. The predictability of saccadic latency in a novel voluntary oculomotor task. Vis. Res 20 (4), 329–339. 10.1016/0042-6989(80)90019-x. [DOI] [PubMed] [Google Scholar]
  24. Hill SK, Reilly JL, Keefe RS, Gold JM, Bishop JR, Gershon ES, Tamminga CA, Pearlson GD, Keshavan MS, Sweeney JA, 2013. Neuropsychological impairments in schizophrenia and psychotic bipolar disorder: findings from the Bipolar-Schizophrenia network on intermediate phenotypes (B-SNIP) study. Am. J. Psychiatry 170 (11), 1275–1284. 10.1176/appi.ajp.2013.12101298. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Huang LY, Jackson BS, Rodrigue AL, Tamminga CA, Gershon ES, Pearlson GD, Keshavan MS, Keedy SS, Hill SK, Sweeney JA, Clementz BA, McDowell JE, 2022. Antisaccade error rates and gap effects in psychosis syndromes from bipolar-schizophrenia network for intermediate phenotypes 2 (B-SNIP2). Psychol. Med 52 (13), 2692–2701. 10.1017/S003329172000478X. [DOI] [PubMed] [Google Scholar]
  26. Jester DJ, Thomas ML, Sturm ET, Harvey PD, Keshavan M, Davis BJ, Saxena S, Tampi R, Leutwyler H, Compton MT, Palmer BW, Jeste DV, 2023. Review of major social determinants of health in Schizophrenia-Spectrum psychotic disorders: I. Clinical outcomes. Schizophr. Bull 49 (4), 837–850. 10.1093/schbul/sbad023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Kay SR, Fiszbein A, Opler LA, 1987. The positive and negative syndrome scale (PANSS) for schizophrenia. Schizophr. Bull 13 (2), 261–276. 10.1093/schbul/13.2.261. [DOI] [PubMed] [Google Scholar]
  28. Keefe RS, Goldberg TE, Harvey PD, Gold JM, Poe MP, Coughenour L, 2004. The brief assessment of cognition in schizophrenia: reliability, sensitivity, and comparison with a standard neurocognitive battery. Schizophr. Res 68 (2–3), 283–297. 10.1016/j.schres.2003.09.011. [DOI] [PubMed] [Google Scholar]
  29. Kirschner M, Shafiei G, Markello RD, Makowski C, Talpalaru A, Hodzic-Santor B, Devenyi GA, Paquola C, Bernhardt BC, Lepage M, Chakravarty MM, Dagher A, Misic B, 2020. Latent Clinical-Anatomical dimensions of schizophrenia. Schizophr. Bull 46 (6), 1426–1438. 10.1093/schbul/sbaa097. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Knapp TR, 1978. Canonical correlation analysis: a general parametric significance-testing system. Psychol. Bull 85 (2), 410–416. [Google Scholar]
  31. Letang SK, Lin SS, Parmelee PA, McDonough IM, 2021. Ethnoracial disparities in cognition are associated with multiple socioeconomic status-stress pathways. Cogn. Res Princ. Implic 6 (1), 64. 10.1186/s41235-021-00329-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Levine MS (1977). Canonical analysis and factor comparison. Sage Publications. Publisher description 〈http://www.loc.gov/catdir/enhancements/fy0660/77075941-d.html〉. [Google Scholar]
  33. Lewis ACF, Molina SJ, Appelbaum PS, Dauda B, Di Rienzo A, Fuentes A, Fullerton SM, Garrison NA, Ghosh N, Hammonds EM, Jones DS, Kenny EE, Kraft P, Lee SS, Mauro M, Novembre J, Panofsky A, Sohail M, Neale BM, Allen DS, 2022. Getting genetic ancestry right for science and society. Science 376 (6590), 250–252. 10.1126/science.abm7530. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Li H, Eack SM, Montrose DM, Miewald JM, Keshavan M, 2011. Longitudinal treatment outcome of African American and caucasian patients with first episode psychosis. Asian J. Psychiatr 4 (4), 266–271. 10.1016/j.ajp.2011.08.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Lipszyc J, Schachar R, 2010. Inhibitory control and psychopathology: a meta-analysis of studies using the stop signal task. J. Int Neuropsychol. Soc 16 (6), 1064–1076. 10.1017/S1355617710000895. [DOI] [PubMed] [Google Scholar]
  36. Medina C, Akinkunmi A, Bland N, et al. , 2024. Differences in schizophrenia treatments by race and ethnicity—analysis of electronic health records. Schizophr 10, 48. 10.1038/s41537-024-00470-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. McDowell JE, Clementz BA, 2001. Behavioral and brain imaging studies of saccadic performance in schizophrenia. Biol. Psychol 57 (1–3), 5–22. 〈https://www.ncbi.nlm.nih.gov/pubmed/11454432〉. [DOI] [PubMed] [Google Scholar]
  38. Merritt-Davis OB, Keshavan MS, 2006. Pathways to care for African Americans with early psychosis. Psychiatr. Serv 57 (7), 1043–1044. 10.1176/ps.2006.57.7.1043. [DOI] [PubMed] [Google Scholar]
  39. Misra S, Etkins OS, Yang LH, Williams DR, 2022. Structural racism and inequities in incidence, course of illness, and treatment of psychotic disorders among black Americans. Am. J. Public Health 112 (4), 624–632. 10.2105/AJPH.2021.306631. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Montgomery SA, Asberg M, 1979. A new depression scale designed to be sensitive to change. Br. J. Psychiatry 134, 382–389. 〈https://www.ncbi.nlm.nih.gov/pubmed/444788〉. [DOI] [PubMed] [Google Scholar]
  41. Nagendra A, Halverson TF, Pinkham AE, Harvey PD, Jarskog LF, Weisman de Mamani A, Penn DL, 2020. Neighborhood socioeconomic status and racial disparities in schizophrenia: an exploration of domains of functioning. Schizophr. Res 224, 95–101. 10.1016/j.schres.2020.09.020. [DOI] [PubMed] [Google Scholar]
  42. Parker D, Trotti R, McDowell J, Keedy S, Keshavan M, Pearlson G, Ivleva E, Huang L, Sauer K, Hill S, Sweeney J, Tamminga C, & Clementz B. (2025). Differentiating biomarker features and familial characteristics of B-SNIP psychosis Biotypes. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Parker DA, Trotti RL, McDowell JE, Keedy SK, Gershon ES, Ivleva EI, Pearlson GD, Keshavan MS, Tamminga CA, Sweeney JA, Clementz BA, 2020. Auditory paired-stimuli responses across the psychosis and bipolar spectrum and their relationship to clinical features. Biomark. Neuropsychiatry 3, 100014. 10.1016/j.bionps.2020.100014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Parker DA, Trotti RL, McDowell JE, Keedy SK, Hill SK, Gershon ES, Ivleva EI, Pearlson GD, Keshavan MS, Tamminga CA, Clementz BA, 2021. Auditory oddball responses across the Schizophrenia-Bipolar spectrum and their relationship to cognitive and clinical features. Am. J. Psychiatry 178 (10), 952–964. 10.1176/appi.ajp.2021.20071043. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Polich J, 2007. Updating P300: an integrative theory of P3a and P3b. Clin. Neurophysiol 118 (10), 2128–2148. 10.1016/j.clinph.2007.04.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Sarullo K, Barch DM, Smyser CD, Rogers C, Warner BB, Miller JP, England SK, Luby J, Swamidass SJ, 2024. Disentangling socioeconomic status and race in infant brain, birth weight, and gestational age at birth: a neural network analysis. Biol. Psychiatry Glob. Open Sci 4 (1), 135–144. 10.1016/j.bpsgos.2023.05.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Schwartz RC, Blankenship DM, 2014. Racial disparities in psychotic disorder diagnosis: a review of empirical literature. World J. Psychiatry 4 (4), 133–140. 10.5498/wjp.v4.i4.133. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Sheehy-Skeffington J, 2020. The effects of low socioeconomic status on decision-making processes. Curr. Opin. Psychol 33, 183–188. 10.1016/j.copsyc.2019.07.043. [DOI] [PubMed] [Google Scholar]
  49. Sosu EM, Schmidt P, 2022. Changes in cognitive outcomes in early childhood: the role of family income and volatility. Front Psychol. 13, 758082. 10.3389/fpsyg.2022.758082. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Tamminga CA, Ivleva EI, Keshavan MS, Pearlson GD, Clementz BA, Witte B, Morris DW, Bishop J, Thaker GK, Sweeney JA, 2013. Clinical phenotypes of psychosis in the Bipolar-Schizophrenia network on intermediate phenotypes (B-SNIP). Am. J. Psychiatry 170 (11), 1263–1274. 10.1176/appi.ajp.2013.12101339. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Thomas O, Parker D, Trotti R, McDowell J, Gershon E, Sweeney J, Keshavan MS, Keedy SK, Ivleva E, Tamminga CA, Pearlson GD, Clementz BA, 2019. Intrinsic neural activity differences in psychosis biotypes: findings from the Bipolar-Schizophrenia network on intermediate phenotypes (B-SNIP) consortium. Biomark. Neuropsychiatry 1, 100002. 10.1016/j.bionps.2019.100002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Tomassi S, Tosato S, Mondelli V, Faravelli C, Lasalvia A, Fioravanti G, Bonetto C, Fioritti A, Cremonese C, Lo Parrino R, De Santi K, Meneghelli A, Torresani S, De Girolamo G, Semrov E, Pratelli M, Cristofalo D, Ruggeri M, Group GU, 2017. Influence of childhood trauma on diagnosis and substance use in first-episode psychosis. Br. J. Psychiatry 211 (3), 151–156. 10.1192/bjp.bp.116.194019. [DOI] [PubMed] [Google Scholar]
  53. Trotti RL, Parker DA, Sabatinelli D, Keshavan MS, Keedy SK, Gershon ES, Pearlson GD, Hill SK, Tamminga CA, McDowell JE, Clementz BA, 2023. Emotional scene processing in biotypes of psychosis. Psychiatry Res 324, 115227. 10.1016/j.psychres.2023.115227. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Turcotte LM, Wang T, Beyer KM, Cole SW, Spellman SR, Allbee-Johnson M, Williams E, Zhou Y, Verneris MR, Rizzo JD, Knight JM, 2024. The health risk of social disadvantage is transplantable into a new host. Proc. Natl. Acad. Sci. USA 121 (30), e2404108121. 10.1073/pnas.2404108121. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. van der Ven E, Susser E, 2023. Structural racism and risk of schizophrenia. Am. J. Psychiatry 180 (11), 782–784. 10.1176/appi.ajp.20230733. [DOI] [PubMed] [Google Scholar]
  56. Wang HT, Smallwood J, Mourao-Miranda J, Xia CH, Satterthwaite TD, Bassett DS, Bzdok D, 2020. Finding the needle in a high-dimensional haystack: canonical correlation analysis for neuroscientists. Neuroimage 216, 116745. 10.1016/j.neuroimage.2020.116745. [DOI] [PubMed] [Google Scholar]
  57. White K, Lawrence JA, Tchangalova N, Huang SJ, Cummings JL, 2020. Socially-assigned race and health: a scoping review with global implications for population health equity. Int J. Equity Health 19 (1), 25. 10.1186/s12939-020-1137-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Young RC, Biggs JT, Ziegler VE, Meyer DA, 1978. A rating scale for mania: reliability, validity and sensitivity. Br. J. Psychiatry 133, 429–435. 〈https://www.ncbi.nlm.nih.gov/pubmed/728692〉. [DOI] [PubMed] [Google Scholar]
  59. Yu J, Haynie DL, Gilman SE, 2024. Patterns of adverse childhood experiences and neurocognitive development. JAMA Pedia 178 (7), 678–687. 10.1001/jamapediatrics.2024.1318. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS2129733-supplement-1.docx (1.1MB, docx)
2
NIHMS2129733-supplement-2.docx (80.9KB, docx)
3
NIHMS2129733-supplement-3.docx (13.9KB, docx)
4
NIHMS2129733-supplement-4.docx (63.9KB, docx)
5
NIHMS2129733-supplement-5.pptx (3.7MB, pptx)

RESOURCES