Autumn 2023 Tier 1 pilot awardees report final project findings

Investigators
Amy J. Youngbloom, Department of Epidemiology
Stephen J. Mooney, Department of Epidemiology
Anjum Hajat, Department of Epidemiology
Isaac Rhew, Department of Psychiatry & Behavioral Sciences
Rebecca Walter, Runstad Department of Real Estate

Project summary
The primary goals of the project titled, “Housing affordability and chronic stress in the US: Does affordability modify the effect of neighborhoods on health?” were to assess the impact of housing cost burden and housing cost assistance on stress biomarkers. With this project we hope to better understand whether experiencing housing cost burden may negatively affect health by increasing stress, and whether receiving housing assistance can help mitigate the negative effects of high housing cost on health.

Additionally, we sought to understand how the effects of neighborhood characteristics on stress biomarkers differ for those who are housing cost burden versus not housing cost burden and for those receiving housing assistance versus those on a pseudo-waitlist. We hypothesized that the positive health effects of neighborhood walkability, greenspace, and park access would be diminished for those individuals experiencing housing cost burden. For individuals receiving housing assistance, we hypothesized that higher walkability, more greenspace, more park area, and lower neighborhood deprivation would have a stronger negative association with stress biomarkers compared to those on a pseudo-waitlist.

We accessed participant data from Housing and Urban Development Data (HUD) that was with the National Health and Nutrition Examination Survey and Exam (NHANES) for this project, which allowed us to link individual history of housing assistance with stress biomarker measures, and allowed for further geographic linking of individuals with neighborhood characteristics.

We originally planned to link neighborhood characteristics using buffer-based weighted interpolation. However, because all data linking must be completed by an assigned CDC analyst using SAS, we were not able to use the existing R code for geographic interpolation, and it was not reproducible in SAS. We modified our approach to link neighborhood characteristics— including walkability, park access, greenspace and neighborhood deprivation—at the census tract and block-group level.

The linking process also posed many challenges, as we could not assess the quality of the merge until we accessed the data in the RDC. Due to this process, we had to request a re merge by our analyst twice. First, to address GEOID formatting issues that led to a large amount of missing merged data, and secondly to access missing weight variables and incorrectly merged Park data. Each re-merge required months of review prior to being able to access the data again in the RDC.

As of July 2026, we have a final, complete and accurate dataset and are currently conducting analyses for this project. We are assessing the effect of housing cost burden on stress biomarkers using the full NHANES dataset of adults from 1999–2019. Individual housing costs were estimated using the linked tract median housing cost for renters or homeowners for the year that NHANES was completed. Individuals were categorized as (1) not housing cost burdened, if their estimated housing costs were less than 30% of their income, (2) moderately housing cost burdened if the estimated housing cost were 30-50% of their income, or (3) severely housing cost burdened if the estimated housing costs were more than 50% of their income. Chronic stress biomarkers included diastolic blood pressure, systolic blood pressure, pulse, BMI, HDL cholesterol, Total cholesterol, Triglycerides, White blood cell, and albumin creatinine ratio. We created a composite measure of stress biomarkers by summing the number of biomarkers for which an individual was in the highest quartiles (lowest quartile for HDL; highest or lowest for white blood cell count), a method that has been utilized and validated in prior studies (Merkin et al. 2009). We are also conducting analyses stratified by housing tenure (owner v renter) and by income level (low, middle, high).

Using the same composite measure of stress biomarkers, we are assessing the effect of housing cost assistance using the NHANES-HUD linked dataset from 1999-2018. We included linked individuals who were (1) concurrently enrolled in housing assistance when they complete the NHANES exam, or who were (2) enrolled in a housing assistance program within the two years following their NHANES exam (i.e., pseudo-waitlist). We additionally looked at a stratified analysis to assess differences in the association for Multifamily, Housing Choice Voucher, or Public Housing participants.

In our second aim, we are using geographically linked neighborhood measures to assess the association between neighborhood measures and stress biomarkers for the full NHANES and the NHANES-HUD linked datasets, and whether this association differed by housing cost burden status or by housing assistance status.

Investigators
Alison Cardinal, School of Interdisciplinary Arts and Sciences (UW Tacoma)
Michele Andrasik, Fred Hutch Cancer Center
Pedro Goincochea, Fred Hutch Cancer Center
Rafael Gonzalez, Fred Hutch Cancer Center

Project summary
Our project sought to advance two pillars of population health, human health and social and economic equity, by seeking to understand why participation of Hispanic/Latino/a individuals in HIV/AIDS clinical research in the United States (US) is low (<12%) and identify ways to increase it, especially among Spanish speakers with limited English proficiency (LEP). The Hispanic/Latino/a population in the US is the second most impacted group by the HIV epidemic, comprising nearly 30% of new HIV cases (CDC 2023). Research is essential to identify innovative ways to prevent the acquisition of HIV and develop new HIV therapeutics and a cure for HIV infection. Clinical research requires diverse participation in clinical trials, and enrollment in HIV research studies should reflect the composition of the population most affected by the epidemic. Increasing participation of Hispanic/Latino/a individuals in HIV research is therefore critical for the generalization and application of research results and for contributing to Ending the HIV Epidemic in America by 2030. The HIV/AIDS clinical trials networks of the National Institute of Allergy and Infectious Diseases (NIAID) are designed to address these scientific research priorities. Our US-based study sought to understand why participation of Hispanic/Latino/a individuals in HIV/AIDS clinical research is low, especially among monolingual Spanish speakers.

Methodology: We conducted formative qualitative research by interviewing HIV investigators, clinicians, community educators and recruiters from different clinical research sites conducting HIV vaccine research across the US. Our main goal was to understand the strategies and challenges clinical research sites face when involving Hispanic/Latino/a communities at every stage of the clinical research cycle, including education, recruitment and retention. We particularly focused on understanding the strategies and challenges sites used to engage monolingual Spanish speakers in clinical research.

Results: We interviewed 14 participants from 10 clinical research sites across the US. Participants’ experience working in clinical research ranged from 3 to 30 years. Findings revealed that barriers to participation include a lack of bilingual staff, a need to adapt clinical research documents and procedures for monolingual Spanish speakers, misinformation about HIV and HIV clinical research in the Hispanic/Latino/a community and a lack of sustained relationships with Hispanic/Latino/a-serving community groups. Some clinical research investigators and staff demonstrated a nuanced understanding of the diversity within the Hispanic/Latino/a community and pointed to the need for strategies that reflect this complexity. Clinical research staff and investigators underscored the need for institutional support, funding and network-driven initiatives to provide language services and cultural competency training.

Conclusion: Our findings suggest that ensuring diverse participation in clinical studies requires a holistic approach to language access that addresses cultural, linguistic and structural barriers to participation among Hispanic/Latino/a individuals on a national scale. Participants’ insights into their working environments indicate that institutions must prioritize and fund initiatives that build long-term community relationships, facilitate language access and reflect the cultural diversity of the populations they aim to serve. The findings from our study point to the need for collaborative and sustained efforts among research sites, community organizations that represent Latinx communities and HIV/AIDS clinical research networks. These collaborations can facilitate the development of inclusive community engagement strategies that are essential for advancing equitable clinical research, not only for multilingual communities but for diverse communities more broadly.

Investigators
Robin E. Klabbers, Departments of Global Health and Emergency Medicine
Kelli N. O’Laughlin, Departments of Emergency Medicine and Global Health
Jacob Oluma, Medical Teams International
Timothy Muwonge, Makerere University

Project summary
The overarching goal of the ART2FIT study was to generate evidence to inform the optimization of differentiated service delivery (DSD) models for HIV care in refugee settlement contexts, so that humanitarian populations can benefit from this evidence-based approach to improving retention on antiretroviral therapy (ART). To contribute to this goal, the study examined factors that influence participation in community ART delivery among people living with HIV (PLHIV), as well as the capacity of health systems and providers to implement this model effectively within refugee settlements in Uganda.

The study pursued three interconnected aims. First, it sought to identify and prioritize barriers to participation in and implementation of community ART delivery in refugee settlement settings. Second, it explored stakeholder perspectives on strategies to address these barriers and on what optimal community ART delivery would look like in this context. Third, it synthesized findings from the first two aims to generate a set of community ART delivery characteristics valued by PLHIV, which were subsequently used to inform the design of a discrete choice experiment (DCE).

Between October 2024 and April 2025, the study team conducted 34 in-depth interviews (IDIs) with PLHIV across five refugee settlements. In addition, between March and May 2025, seven focus group discussions (FGDs) were conducted with health workers and expert clients providing HIV care in this context. IDI participants were asked to identify barriers to participation in community ART delivery and to prioritize the most important barriers from those they described. FGD participants were asked about barriers to community ART delivery implementation and participated in structured barrier-ranking exercises. Data were analyzed using rapid qualitative methods. Implementation science frameworks—including the Consolidated Framework for Implementation Research (CFIR) 2.0 and the Expert Recommendations for Implementing Change (ERIC)—were applied to systematically categorize identified barriers and corresponding recommendations, enabling the findings to be situated within the broader implementation literature. Multiple dimensions of implementation barriers, including impact, duration, addressability, frequency and equity, were examined using go-zone plots.

Overall, the study captured diverse perspectives on community ART delivery in refugee settlements and identified key barriers affecting initial uptake, sustained engagement and effective implementation. These barriers operated at the individual, community and health system level and reflected both challenges unique to refugee settlement contexts and challenges common to other settings. Some barriers were long-standing, while others had emerged more recently. Participants articulated concrete strategies to address several of the identified barriers and described features of an ideal community ART delivery model tailored to refugee settlement contexts. Detailed analyses and findings are currently being prepared for dissemination through peer-reviewed publications.

Investigators
Barclay T. Stewart, Department of Surgery
Colleen Chalmers, King County Regional Homelessness Authority
Tony Machacha, King County Regional Homelessness Authority
Caitlin Orton, Department of Surgery

Project summary
The EQUIP Toolkit contains three fire and burn injury prevention education materials (i.e., How to Safely Start a Fire, How to Extinguish a Fire, Propane Tank Safety) and three cold injury prevention education materials (i.e., What is a Cold Injury, How to Dress Warmly in Cold and Wet Environments, Cold Weather Car Safety). The Model System Knowledge Translation Center’s Consumer Testing Tool and the Agency for Healthcare Research and Quality’s Patient Education Materials Assessment Tool were used as frameworks for the designing and consumer testing of the toolkit.

The research team and community partners completed outreach in six distinct communities of Seattle. Forty people who were living unhoused participated in cognitive interviews during which they provided feedback on the acceptability, understandability, and utility of the educational materials. Participants also provided their thoughts on dissemination, preferred prevention resources (e.g., tents, clothing, sleeping bags, extinguishing equipment, improved containers for burning fuel) and prevention education topics (e.g., those presented as educational materials and more, to include how to burn hand sanitizer safely).

Overall, we found the EQUIP toolkit to be appropriate, usable, and understandable for a diverse range of reading levels and experiences of those living unhoused. In addition to being presented to KCRHA’s Severe Weather Team, the EQUIP Toolkit has been presented at the World Conference on Injury Prevention and Safety Promotion in New Delhi, India and will be presented at the American Burn Association Annual Meeting in Phoenix, AZ in April, 2025. A subset of the interviews, focusing on How to Safely Start a Fire, were analyzed and presented by a student of the Harborview Injury Prevention and Research Center’s 2024 INSIGHT Program.

Investigators
Anna Larsen, Department of Psychiatry & Behavioral Sciences
Dror Ben-Zeev, Department of Psychiatry & Behavioral Sciences
Ahmad Y. Bawaneh, International Medical Corps – Jordan

Project summary
Arabic-speaking refugees are one of the most rapidly expanding trauma-exposed populations globally, yet existing digital mental health interventions are not culturally or linguistically tailored for this group. We conducted a mixed methods user-centered codesign study among Syrian refugees in Jordan to identify preferences for a digital trauma intervention. This project had four aims:

1. Expand partnerships with local Arab refugees and key stakeholders (e.g., community leaders, service providers) in-country through collaboration with IMC and the establishment of a project-specific community advisory board in Jordan;
2. Conduct a user-centered mixed-methods needs assessment using a community-participatory approach to identify gaps and priorities for mHealth intervention to address mental health concerns of the Arab refugees living in Jordan;
3. Integrate our empirical findings with evidence-based theory to generate a model to guide our next steps to generate a proof-of-concept; and,
4. Evaluate our prospective model with community members and key stakeholders to assess its acceptability and feasibility using the IDEAS (Integrate, Design, Assess, and Share) framework to lay the groundwork for large internal and external grant application submissions to develop and test our proof-of-concept.

We anticipate that this project will set the stage for the first digital mental health intervention tailored for use with Arab refugees in the Middle East with the overall goal of improving health and social equity. Overall, our project achieved our stated aims. We have summarized key findings.

We screened Syrian refugees for trauma symptoms in community centers in Mafraq, Jordan in October 2024. Among those with trauma, we purposively sampled four digital codesign groups: 1) males 18-30 years, 2) males >30, 3) females 18-30, 4) females >30. Those providing informed consent were assessed in Arabic for demographics, trauma experiences (DSM-5 Life Events Checklist), and trauma symptom severity (PTSD Checklist for DSM-5). Codesign groups were conducted in Arabic, audio-recorded for translation. We presented refugees with options for potential digital intervention attributes (visual theme, content format, modality, and content deliverers) and provided Likert responses (1-5, strongly do not like—strongly like).

To understand potential barriers and facilitators to uptake and scale-up of a digital mental health intervention, we interviewed 10 key stakeholders from the Jordanian Ministry of Health, international nongovernmental organizations, and psychiatric/psychological associations in Jordan. We used structured interview guides with topics including gaps in current mental health service provision to Arabic-speaking refugees, exploration of pathways to adoption and scale-up of digital mental health interventions, and perspectives on barriers and facilitators to digital intervention implementation among Arabic-speaking refugees in Jordan.

Overall, 41 Syrian refugees participated in codesign. Mean age was 33.0 years (standard deviation [sd]:11.6), and 54% were female. The majority (61%) were married; 47% had primary school education or below. Nearly two thirds (59%) of participants endorsed ever utilizing mental health services. Participants reported a wide range of trauma experiences including physical assault (56%), witnessing sudden violent death (37%), captivity (31%), combat (15%), and sexual assault (12%). Average trauma severity scores were 48.9 (sd:15.3 range: 12-78). Participants preferred visual themes depicting nature (mean:4.6[sd:0.9]) and Islamic imagery (4.3[1.1]), over themes of people, spirituality, or patterns. Ratings were similar (~4.4) across intervention content options (psychoeducation, psychological skill, story). Participants preferred video-based content (3.9[1.3]) over audio, interactive, and written modalities. When presented digital images depicting potential female characters as intervention deliverers, a medical provider was preferred (4.4[0.8]), followed by a female wearing traditional clothing (skirt/hijab: 3.2[1.1]) as opposed to other characters with varying garb and features. A male clinician was also preferred over other male options. Participants expressed enthusiasm for a potential digital mental health intervention. One summarized, “I prefer to find a way to deal with my stress…by learning something that will help me overcome any psychological problem I face on my own”. Key stakeholders (N=9) had a mean age of 43.9 (SD:5.7), 4 out of 9 were female, and they had been serving in their current role for an average of 6.5 years (SD:5.3).

Key stakeholders discussed lack of trained mental health service providers and stigma to seek mental health support as key gaps in service provision to Arabic-speaking refugees. Common themes about key attributes to consider for a digital mental health intervention included ensuring it presents mainly visual content (“text-light”) to facilitate engagement among non-literate populations, and general ease-of use. Ensuring alignment and integration with stakeholder groups already providing psychosocial services to Syrian refugees was raised as a key facilitator. Failing to generate buzz among Syrian refugees and lack of integration with existing services and practices within refugee camps and communities were cited as key barriers. Overall, key stakeholders generally supported development of a digital mental health intervention for Arabic-speaking refugees and often highlighted the high potential for benefit to Arabic-speaking refugees elsewhere in Jordan, the Middle East, and the world.

In our codesign study, Arabic-speaking refugees experiencing trauma most-preferred digital interventions combining psychoeducation and psychological skills training delivered via videos by medical providers utilizing culturally relevant nature and religious aesthetics. Findings inform digital interventions to alleviate trauma among Arabic-speaking refugees.

Investigators
Lianne Sheppard, Departments of Environmental & Occupational Health Sciences and Biostatistics
Anjum Hajat, Department of Epidemiology
Magali Blanco, Department of Environmental & Occupational Health Sciences
Amanda Gassett, Department of Environmental & Occupational Health Sciences
Cynthia Curl, Boise State University
Carly Hyland, University of California Berkeley
Edward Kasner, Department of Environmental & Occupational Health Sciences
Anne Reiderer, Department of Environmental & Occupational Health Sciences
Cecilia Martindale, Department of Environmental & Occupational Health Sciences

Project summary
Glyphosate is the most widely used herbicide globally, and growing evidence links it to adverse health outcomes. However, large-scale studies characterizing exposure levels in the general population remain limited. The goal of this project was to examine environmental, social, and dietary factors associated with glyphosate levels among children and adults in the U.S. Due to delays in accessing restricted data required to evaluate environmental factors, this report focuses the social and dietary analyses from the public use data.

Using publicly available data from the National Health and Nutrition Examination Survey (NHANES), we conducted a nationally representative analysis of 7,067 participants (3 years and older) with urinary glyphosate measurements collected between 2013 and 2018. We assessed whether race/ethnicity, SES and dietary intake of foods known to carry high glyphosate residue were associated with elevated glyphosate levels.

We found elevated urinary glyphosate levels among children, non-Hispanic White participants, individuals who consumed grains and oats, and samples collected between May 1st and October 31st. More specifically, children had higher urinary glyphosate levels than any other age group (0.45 µg/g creatinine for ages 0-19 vs 0.33 µg/g creatinine for ages 20+). Among children, non-Hispanic White participants had the highest levels (0.48 µg/g), while Mexican American participants had the lowest levels (0.38 µg/g). Among adults, non-Hispanic White participants again had the highest levels (0.36 µg/g), and non-Hispanic Black participants had the lowest (0.26 µg/g). Grain and oat consumption within the past 24 hours was significantly associated with higher glyphosate levels among adults (0.34 µg/g vs 0.24 µg/g for grains, 0.52 µg/g vs 0.33 µg/g for oats). Samples collected between May 1st and October 31st showed elevated glyphosate levels compared to those collected outside this period (0.49 µg/g vs 0.40 µg/g among children, 0.35 µg/g vs 0.31 µg/g among adults).

These differences in glyphosate levels may be influenced by several factors. Children consume more food per pound of body weight than adults, which can result in proportionally higher glyphosate exposure. They may also consume more cereals and grains – foods commonly treated with glyphosate. Biologically, children are still developing, and this can affect how their bodies absorb, metabolize, and excrete toxicants. Moreover, while adjusting urinary glyphosate concentrations for creatinine is a standard approach to account for hydration and urine dilution across samples, this should be interpreted with caution in children. Children generally have lower creatinine levels than adults such that adjusting for creatinine can artificially inflate these levels. Still, we observed similar patterns when comparing non-adjusted values. Lifestyle and dietary patterns across racial and ethnic groups also likely contribute, for example differences in grain and oat consumption. Seasonal variation in glyphosate levels may reflect patterns in herbicide application and changes in food residue levels (particularly since May through September is generally considered the high pesticide spray season) as well as seasonal shifts in dietary habits.

These findings will support future analyses using restricted access geolocation data that will evaluate associations between glyphosate levels and environmental factors, including proximity to agricultural fields, urbanicity and ambient temperature.

Investigators
Shan Liu, Department of Industrial & Systems Engineering
Lawrence S. Wissow, Seattle Children’s Hospital
Alysha Thompson, Seattle Children’s Hospital

Project summary
Pediatric mental health disorders such as depression, anxiety, ADHD and behavior disorder significantly impact the well-being of children and youth. In Washington state (WA), data has shown that more than one-third of people live in areas lacking mental health providers. Accurately mapping the geographic distribution of true demand in WA is crucial for designing an optimal continuum of care.

The goal of this project was to estimate the true demand for pediatric mental health services in WA using statistical needs estimation and diverse data sources. We obtained a WA IRB exemption and collected data from the Washington State Comprehensive Hospital Abstract Reporting System’s inpatient discharge (CHARS) dataset, WA census data from the American Community Survey (ACS), survey data from the National Survey of Children’s Health (NSCH), the WA Healthy Youth Survey (HYS), and a youth mental health legislative report from the WA State Health Care Authority (HCA).

First, we identified the relevant factors affecting youth mental health prevalence through a literature review of previous studies using the NSCH. Next, we analyzed the NSCH data by building six machine learning models to predict the prevalence of several mental health conditions. The final best-performing logistic regression model included independent variables such as age, sex, race, family structure, language, income, health insurance, education, caregiver employment and family income.

To estimate the prevalence of each mental health condition among youth at the county level, we constructed a marginal distributions constrained optimization model. The model recovers detailed demographic group proportions by harmonizing regression results from the NSCH data with estimates from other datasets such as the HYS and CHARS, under the constraints of marginal distributions found in the ACS. See Fig. 1a for one example result. Our approach facilitates the integration of national and local datasets to improve small-area estimation and supports the strategic placement of new mental health facilities and programs in downstream research. Overall, this study demonstrates effective data fusion to improve disease prevalence prediction quality and service planning.

We also analyzed hospital admissions for pediatric mental health and their relationship to patients’ geographic location using CHARS. We calculated hospital admission volume for patients with a primary or any mental health diagnosis code and overlaid these data with the total population of children and the percentage of children in each ZIP code. In addition, we examined inpatient admissions and their relationship to pediatric patients’ social vulnerability by geographic area to better understand disparities in mental health treatment access. We used the Social Vulnerability Index (SVI) for patients’ home residences and the Social Deprivation Index (SDI) for hospital locations.

Results showed that patients living in areas with higher SVI scores tended to receive care at hospitals located in areas with higher SDI scores in WA.

Investigators
Leslie Coney, Department of Human Centered Design & Engineering
Julie Kientz, Department of Human Centered Design & Engineering
Monica McLemore, Department of Child, Family, and Population Health Nursing
Jazmin Williams, BLKBRY

Project summary
Black birthing people face significantly high maternal mortality rates in the United States. This disparity is rooted in historical reproductive oppression and persistent obstetric racism. These issues create a pronounced power imbalance and foster deep distrust between Black communities and the healthcare system, underscoring the critical need for community-based maternal health interventions. These interventions provide culturally competent support and education. The purpose of this research was to understand technology’s role in accessing and using community maternal health resources (CMHRs) for Black birthing people in Washington State.

In pursuit of a community-centered approach, the team partnered with Jazmin Williams, founder of BLKBRY, a Black-owned community-based organization (CBO) that offers perinatal services and doula care for Black and Indigenous communities in Washington State. The team formalized a partnership through the creation of a memorandum of understanding (MOU), ensuring the work was built on mutual goals and respect. Williams played a significant role in study design and participant recruitment. We conducted semi-structured interviews with 12 Black mothers in Washington State. These mothers shared their definitions of community, the community resources they used and technology’s role in accessing those resources, how their racial and cultural identities shaped their birthing journeys, and how Black joy has manifested in pregnancy and parenthood.

While qualitative data analysis is still ongoing, initial insights point to two primary roles of technology in community support: connection and education. The mothers frequently described how technology enabled them to connect with other parents, family and friends, and maternal care providers such as doulas and other community health workers. Additionally, many mothers shared how they used technology to educate themselves throughout pregnancy, birth and parenthood. Not only did they seek information broadly through Google searches, they also sought experiential knowledge through parent support groups on platforms such as Facebook and Reddit.

In addition to supporting this research, BLKBRY helped plan and execute Cradles & Culture, a culturally responsive community baby shower. Cradles & Culture was heavily informed by the stories and experiences shared by the 12 mothers who participated in the study.

Investigators
Will von Geldern, Evans School of Public Policy & Governance
Karin Martin, Evans School of Public Policy & Governance
Rachel Fyall, Evans School of Public Policy & Governance
Will Beck, Tacomaprobono Housing Justice Project

Project summary
This study sought to inform policies that could affect one salient aspect of evictions: default judgments. Default judgments occur when evicted tenants do not participate in eviction proceedings and are thus unable to dispute or delay their eviction. When a tenant fails to respond to a civil summons and complaint or misses their court hearing, the court sides with the landlord in response to the defendant’s non-compliance with the process.

Our team recruited 23 recently evicted tenants who had experienced default judgments to complete interviews as a part of our study. In preliminary analysis of the data collected from our interviews, we have identified three types of barriers commonly described by participants that lead to default judgments. Tenants described informational, cognitive or psychological, and logistical barriers to participating in the eviction process and contacting a legal aid provider. While informational and cognitive/psychological barriers predominate, logistical barriers proved insurmountable when they arose.

Because of the budgetary adjustment approved by PHI and a funding match from the Washington Office of Civil Legal Aid (OCLA), we were also able to hire a Research Assistant to begin manually collecting data on ten quantitative case-level variables from local court records for approximately 1,000 cases from six counties that are the site of more than half of all evictions statewide. Preliminary analysis indicates that tenants were represented in approximately 45% of cases despite the existence of Washington’s right to counsel program. Landlords, on the other hand, were represented in more than 95% of cases.

We have shared these preliminary results with scholars and policymakers in Washington. First, lead co-PI von Geldern presented these preliminary findings to the West Coast Poverty Center’s Housing Policy Roundtable on October 31, 2024. Based on that presentation, von Geldern was invited to present findings to the Washington State Senate Housing Committee Work Session on December 13th as detailed in this article. We have also begun to disseminate these findings in academic settings. von Geldern presented a theoretical paper with some preliminary empirical data at the Interdisciplinary Association for Population Health Sciences (IAPHS) conference in September 2024, and subsequently prepared a manuscript entitled “Evictions, Legal Counsel, and Population Health,” currently under review at Social Science and Medicine.

Preliminary qualitative results suggest that informational barriers pose a significant challenge for most tenants. Tenants often do not understand the process or the documents they receive from the court, and are largely unaware of their right to an attorney. Quantitative results suggest that tenants’ behaviors in response to receiving an eviction summons (submitting a response or attending a hearing) are strong predictors of their take-up of legal counsel. These preliminary findings will motivate the final chapter of co-PI von Geldern’s dissertation, which will measure the effectiveness of a randomized controlled trial of a behaviorally-informed informational mailer intended to encourage tenants to attend eviction hearings. This intervention was funded by the U.S. Department of Health and Human Services (HHS) Administration for Children and Families (ACF) through the Behavioral Intervention Scholars program.

In contrast to the paperwork that tenants receive from the court, this mailer will contain a de-stigmatizing message (e.g., “You deserve the right to defend yourself”) along with clear directions to attend their hearing and information about the availability of legal assistance. This trial will begin in January 2025. We will also continue to disseminate our findings in academic outlets in the fields of socio-legal studies, public policy, and behavioral science as well as through our existing relationships with community-based organizations and policymakers in Washington

Investigators
Katherine T. Foster, Departments of Psychology and Global Health
Jennifer Forsyth, Department of Psychology
Matthew P. Conomos, Department of Biostatistics
Daniel J. O. Roche, University of Maryland

Project summary
The primary objective of this pilot was to incorporate a data collection pipeline for baseline genotyping into the ongoing PEARLY study (PI: Foster, RF1MH135458, converted from R01MH135458), which features high-dimensional clinical assessments of children and families (target n=400) using accelerated longitudinal cohort design with intensive smartphone-based ecological momentary assessment (EMA) and digital phenotyping (e.g., passive sensing of physiology, location, etc.).

To develop initial infrastructure for the pipeline we have successfully: (1) procured salivary DNA collection kits for the entire anticipated sample (n=400), (2) updated PEARLY protocols to include consent and support for DNA collection from participants following initial enrollment, (3) developed and refined standardized mailer prototypes for sending kits to participants by mail, (4) begun obtaining DNA collection consent from PEARLY participants, (5) mailed kits to enrolled participants providing consent and (5) established a partnership and a secured contract with the UW Center for Clinical Genomics (CCG) to conduct batched extraction for 300 returned samples. DNA collection and extraction is underway and, given the slow rate of PEARLY recruitment, is projected to conclude by March 2026.

The pipeline we have successfully begun establishing with the support of this award ensures that all future PEARLY participants will undergo baseline genetic testing using the same standardized protocol, creating a comprehensive multimodal dataset that integrates genetic biomarkers with intensive longitudinal phenotypic data.

Investigators
Mehmet Kurt, Department of Mechanical Engineering
Jacob Ruzevick, Department of of Neurological Surgery

Project summary
Our project’s main aim was to develop deep learning models for segmentation (or boundary detection) of glioma brain tumors from MRI images taken in low-resolution settings, with an emphasis on images taken in Sub-Saharan Africa. Furthermore, the broader scope of this project is to develop methods capable of generalized tumor segmentation of tumors beyond glioma and with the ability to adapt to variations in imaging protocols. These two aims address critical needs for the deployment of automated segmentation tools in clinical settings. The developed models are evaluated according to commonly used metrics such as Dice score and Hausdorff-Distance 95, which give a measure of accuracy in tumor segmentation.

Presently, we have experimented with various deep learning frameworks including traditional fine-tuning, fine tuning with frozen layers, and domain-adversarial training of neural networks. We are currently working on a novel unsupervised approach following insights gained from our previous experiments. We are also happy to announce that this work has been accepted for a poster presentation at the Medical Imaging with Deep Learning (MIDL) conference in Paris, France on July 3-6, 2024; an oral presentation at the SB3C Summer Bioengineering Conference on June 11-14, 2024; a poster presentation at the International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI) on October 6-10, 2024; and a poster presentation at the Lawrence Berkeley National Laboratory Summer Symposium on August 6, 2024.

In addition to developing custom deep learning model architectures, we have applied the current state-of-the-art architectures for tumor segmentation for generalized tumor segmentation. Previous research demonstrates high accuracy in specialized tumor segmentation using various model architectures, including nn-UNet, Swin-UNet and U-Mamba based approaches. To leverage these models, we evaluated a deep learning framework that ensembles these architectures by combining their produced results to produce refined tumor boundary predictions. We evaluated the ability of single and ensembled model architectures on the task of generalized tumor segmentation in our paper “An Ensemble Approach for Brain Tumor Segmentation and Synthesis.”

Investigators
Lesley Steinman, Department of Health Systems and Population Health
Najma Mohammed, Neighborhood House

Project summary
Our project used anti-racist, community-engaged, human-centered design and implementation science methods to co-design mental health care to integrate into social service delivery. We convened Somali social service providers and community members to lead formative work to understand barriers to accessing mental health care, to identify usability issues and what adaptations are needed for evidence-based mental health care to improve cultural and linguistic relevance, to map the tasks and workflows of social service settings within which to embed mental health care; and to co-design mental health care based on learnings from formative work. Our ultimate goal is to improve Somali mental health equity by reducing barriers to access while addressing social drivers of mental health and challenges.

We established our co-design council with the Neighborhood House, Somali Family Safety Task Force, Somali Health Board, Seattle Housing Authority and Public Health – Seattle & King County​.

• We convened our co-design council in Feb, April, May, July, Aug 2024, guided by the Collaborative Intervention Planning Framework for co-design to tackle complex issues with community partners.
• We identified a mental health intervention (the WHO’s Problem Management Plus (PM+) that was developed to train lay health providers to support people facing adversity with behavioral approaches (problem-solving skills, social support, behavioral activation and stress management).
• We engaged with UW MPH student and Somali community member Samira Farah to conduct her practicum, reviewing Islamic healing principles, trauma-informed care and stigma reduction to adapt PM+ based on identified strengths and needs of the Somali community.
• We collaborated with a community partner to translate data collection tools from English into Somali and met with the co-design council to review and edit as needed for different dialects.
• We submitted and were approved for an IRB application for our discovery and design activities.
• We worked the Population Health Initiative to engage a summer Social Entrepreneurship Fellow Xinxin Feng, who brought co-design skills to plan for project sustainability post PHI-funding.
• We got a pilot research grant from UW Alacrity Center (UWAC) to expand the project to Feb 2025.
• We submitted a service grant in June 2025 to the Public Health Preservation and Development Authority to integrate a behavioral health response team into Neighborhood House’s services.
• We presented at the National Community Action Partnership conference in Aug 2024 in Seattle, WA to share lessons learned and best practices for community-academic partnerships for health equity.

Our proposal identified these measures of success to achieve our project aims:

• Formed the co-design council with at least 10 community members and staff: Our council was established in Feb 2024 and includes 9 Somali community members and staff from 5 community-based organizations (listed above) and 2 staff from UW. We also engaged with two local imams from the Muslim American Youth Foundation and the Omar Bin Al-Khattab Islamic Center.
• Identified and reviewed existing mental health tools: We selected the WHO’s PM+ which was developed to be done by lay health providers to engage people facing difficult life circumstances, and is effective at reducing anxiety, depression, PTSD symptoms and psychological distress. PM+ is designed to be adapted to local contexts. Training was recently piloted in NYC, in Somalia and Somaliland and expands our UW and Neighborhood House (NH) successful PEARLS partnership.
• Convened partners at 5 co-design meetings: We held five 1 ½-2 hour meetings in-person (except May) at an accessible location in the afternoon/evening before or after prayer times based on partner preferences. We shared food and $50 incentives per participant/meeting.
• Conducted problem analysis and updated needs assessment. In Feb and April we discussed barriers to mental health care which included stigma towards mental health as “taboo” (a personal deficit or lack of faith); need to prioritize basic needs over mental health, systemic barriers to accessing and navigating clinical care, fear of overmedication and inappropriateness of individualized care for multi-generational cultures. In May we discussed Somali coping strategies and explanatory models including “faith is the number one thing,” the value of both therapeutic and spiritual approaches to mental health care in safe, accessible spaces, the importance of addressing historical trauma and supporting the whole family as “when everyone is sick, everyone is involved to help them” and differences by age, gender, faith and acculturation. In May, CBO interviews with 8 NH staff identified appropriate client flow, engagement opportunities and safety mechanisms for PM+.
• Specified adaptations to existing interventions including community-defined solutions. Based on usability issues identified from surveys and group discussions, in July we co-designed adaptations to PM+ to reduce the cognitive load/complexity, increase cultural adaptation, and provide better scaffolding. We integrated Islamic healing practices and metaphors into the stress management session and simplified Friendship Bench guidance into the problem-solving session, and modified client stories and sample activities with diverse experiences affecting the local community (gun violence, immigration, single parent households). In Aug we conducted cognitive walkthroughs with the co-design council and Somali community members. Evaluations suggest adapted PM+ is more usable and appropriate than the original model. Future opportunities include testing with providers and participants with diverse faith and applying PM+ to specific problems rather than role plays.