I'm able to create a brief and cohesive presentation incorporating the first two CHR projects out of the CAPPS lab at UCLA (the quantitative/MRI project and the access/pathways to care project). The third study will be briefly mentioned. Unfortunately the projects can't be combined in abstract form, so I've attached three: 

Guiding Interests: I entered medical school with an interest in advocacy and improving access to care for unhoused youth and people experiencing psychosis. The second two projects are about access to care: decreasing disparities in access to child and adolescent mental health care through system-, provider-, and care-giver level interventions. Additionally, I've found that increasing access to and utilization of care often requires improvement in the care itself, which led me to basic research regarding psychosis, clinical high risk states, and brain mapping (project 1). Dr. Marfisee also helped me incorporate more direct advocacy, education, and service into my discovery year, which does not fit well into this document, but has been critical to my discovery year experience and goals.

Project 1: Morphometric Similarity Networks and Inter-Individual Cognitive Variation

Keywords: connectome, clinical high risk for psychosis, cognitive test

Background: Neuroimaging has opened a door to understanding the biological and structural etiology of psychosis and clinical high risk states, but the mechanisms underlying these conditions remain poorly understood. Psychosis has been linked to differences in brain structure, neural connectivity, and regional gene expression, as well as differential performance on cognitive tests. Morphometric similarity networks (MSNs) present a new approach to mapping using structural MRI data and may approximate regional connectivity.

Objective: Assess the ability of MSNs to account for inter-individual cognitive variation in control (HC) and clinical high risk (CHR) participants from the North Atlantic Prodrome Longitudinal Study (NAPLS2).

Methods: Participants with baseline MRI data and cognitive testing measures from the NAPLS2 were included (n = 738; age 12-35). MSNs were created using seven morphometric features measured at 308 brain regions defined by the Desikan-Killiany atlas. IQ scores and measures predictive of conversion to psychosis in the NAPLS2 study were used (attention and working memory: CPT-IP, BACS, and aCPT; declarative memory: HLVT, BVMT). Six multivariate PLS regression models were created using MSNs and the following cognitive measures, divided by diagnosis: HC IQ, CHR IQ, HC attention and working memory tasks, CHR attention and working memory tasks, HC declarative memory tasks, and CHR declarative memory tasks. Linear regression was used to adjust MSNs (for age, sex, age*sex, site, and eTIV, in addition to COMBAT site correction) and cognitive measures (for age, sex, age*sex, and eTIV) and the residuals were used in PLS. Regions were ranked and visualized based on their standardized weights to elucidate regional differences between first and second components. PLS results were cross-validated using leave-one-out sets. R version 4.3.1 was used for all analyses; the plsdepot package was used for PLS.

Results: The first and second PLS components explained 31.7% of the variation in IQ in HC (18.7% and 13.0%) and 18.0% in IQ in CHR (3.0% and 15.0%). For attention and working memory tasks, the first two components explained 22.3% of the variation in HC (11.1% and 11.1%) and 11.8% in CHR (2.3% and 9.6%). For declarative memory tasks, the first two components explained 28.3% of the variation in HC (8.7% and 19.6%) and 16.8% in CHR (2.5% and 14.2%). Frontal and temporal regions loaded most strongly onto the first PLS component for all four IQ and attention and working memory models and the CHR declarative memory model. Lingual and post-central regions loaded most strongly onto the first PLS component for the HC declarative memory model.

Conclusions: MSNs may be a useful tool in approximating regional connectivity. Further study to determine the utility of MSNs in prediction of conversion to psychosis is warranted.

Project 2: Pathways to Care Timeline Analysis

Keywords: timeline, clinical high risk, family support

Background: Psychosis may be preceded by prodromal symptoms, or “clinical high risk” (CHR) states. Due to the young average age of onset (15-30), caregivers often assist CHR patients in care-seeking. Over the past 20 years, large multi-center studies, including NAPLS and ProNET, have been created to assess and monitor CHR participants using the Criteria of Prodromal Syndromes (COPS) [based on the Structured Interview for Prodromal Syndromes (SIPS)]. Patients enrolled in these studies have access to early intervention and care if conversion to psychosis occurs. Prior to enrollment, these patients typically have higher-than-average access to care and socio-economic status; this demographic distribution likely reflects disparities in access to care in the broader CHR population. Cascade-of-care frameworks are frequently used to determine factors in care utilization, but participant-created timelines are an infrequently utilized tool in these analyses.

Objective: Assess patient pathways to enrollment in an early psychosis intervention program using a cascade-of-care framework and novel timeline analysis.

Methods: Data is from a subset of caregivers for CHR participants (n = 19, age 13-25) in the UPLIFT (Understanding Prodromes and Lessening Illness in Family Therapy) study who were selected to take part in interviews and create timelines regarding their family member’s pathway to care. Timelines were coded for major events, including symptom onset and care utilization, and interview transcripts were reviewed to clarify timelines and rarely to add events. Assessment and enrollment in the CHR program (“program assessment”) was used as an endpoint. All timelines were coded by two separate researchers and consolidated in dedoose 9.2.7. Medical record data was used to clarify admission dates. Program intake interview data regarding symptomatology from the patient’s perspective was added and coded separately on each timeline. Coded timelines were visualized using Microsoft Excel 16.82 after which a novel inductive content analysis incorporating time-to-event data was utilized.

Results: CHR symptoms were highly variable across this sample, although positive symptoms were frequently identified as triggers for care-seeking. Inpatient psychiatric hospitalization often precipitated program assessment. Most timelines included “links,” or people outside of the caregiving/family unit who raised concern and led to care-seeking. Links were often associated with schools or community organizations. Caregiver-timeline symptom data was similar to the symptoms recorded in the patient interview; [results pending patient interview comparisons and creation of conceptual timeline model].

Conclusions: Participant-created timelines, when used in conjunction with interviews, are a valuable tool in qualitative cascade-of-care analyses. This data will be used to inform a quantitative study regarding access to care for a larger group of CHR participants. Further cascade-of-care analyses are needed to identify points of intervention to improve access to care for people with CHR for psychosis. 

Project 3: Primary Care Providers and Child and Adolescent Mental Health Care: A Focus Group Analysis 

Keywords: child and adolescent psychiatry portal, self-efficacy, primary care

Background: Screening, evaluation, and treatment of psychiatric disorders often falls on pediatric primary care providers (PCPs) due to a national shortage of child and adolescent psychiatrists. Child and adolescent mental health access portals (MAPs), including UCSF Cal-MAP, aim to reduce this burden by providing free pediatric psychiatry telephone consultation and training to PCPs. The experiences of PCPs delivering mental health care to children and adolescents are crucial to understanding how MAPs may better facilitate self-efficacy among PCPs.

Objective: Assess factors affecting PCPs providing mental health care to children and adolescents, with an interest in self-efficacy.

Methods: Focus groups were conducted with clinicians recruited from FQHC and non-FQHC pediatric primary care clinics throughout the San Francisco Bay Area as they were being enrolled in the UCSF Cal-MAP program. A total of 11 focus groups participated, composed of 48 providers from 5 FQHCs and 6 non-FQHCs. Focus group audio recordings were transcribed and coded using ATLAS.ti v8.4.4 and analyzed using inductive thematic analysis. Data was then re-assessed for themes related to self-efficacy.

Results: Factors influencing self-efficacy were identified at the patient- and provider-levels. PCPs also identified system-level factors that significantly affected their provision of care. Patient-level factors related to self-efficacy included patient complexity and family support. Provider-level factors included in-practice support, prior training, and experience with mental health care. System-level factors included referral process and payment structures. A conceptual model incorporating the self-efficacy factors and the related system-level factors was created. 

Conclusions: Factors at patient-, provider-, and system-levels influence PCP self-efficacy and thus the provision of pediatric mental health care. MAPs such as Cal-MAP can tailor interventions to address these factors.