Improving Care for Late-Life Depression Through Partnerships With Community-Based Organizations: Results From the Care Partners Project.

OBJECTIVES: Collaborative care (CC) has demonstrated effectiveness for improving late-life depression in primary care, but clinics offering this service can find it challenging to address unmet social needs that may be contributing to their patients' depression. Clinics may benefit from better coordination and communication with community-based organizations (CBO) to strengthen depression treatment and to address unmet social needs. We evaluated the feasibility of adding a CBO to enhance standard collaborative care and the impact of such partnered care on older adults. DESIGN: Multisite, prepost evaluation. SETTING: Eight (n = 8) partnerships between primary care clinics and community-based organizations in California. PARTICIPANTS: A total of 707 depressed older adults (60 years or older) as evidenced by having a score of 10 or more on the Patient Health Questionnaire (PHQ-9) received care under the Care Partners project. INTERVENTION: A CBO partner was added to augment CC for late-life depression in primary care. MEASUREMENTS: The PHQ-9 was used to identify depressed older adults and to monitor depression symptom severity during a course of care. RESULTS: At baseline, the average PHQ-9 depression score across the partnerships was 15, indicating moderate depression severity. Participating patients saw an average 7-point reduction in their PHQ-9 score, baseline to last score assessed, with nearly half of all participants (48.4%) experiencing a 50% or greater improvement from their baseline score. CONCLUSIONS: Our findings suggest that partnering with a community-based organization is a feasible and effective way for primary care clinics to address late-life depression in their patients.

Trends in firearm injury in a southern California health care system from 2010 to 2020.

BACKGROUND: Firearm injury is a significant public health concern in the United States. METHODS: Data on fatal and nonfatal firearm injuries were obtained from a cohort of N = 7,473,650 members of Kaiser Permanente Southern California, a large integrated healthcare system between 2010 and 2020. Age-adjusted rates of combined fatal and nonfatal firearm injury per 100,000 members were calculated by year, with the 2010 US census as the reference population. Trends were evaluated using Poisson or negative binomial regression. RESULTS: There was an increasing trend in overall firearm injuries between 2010 and 2020 among adults in this large integrated healthcare system (p < .0001), primarily driven by non-self-inflicted firearm injuries (p < .0001). Self-inflicted injuries decreased during this time (p = .01). Injuries among youth showed no significant change. CONCLUSION: There was an increasing trend in firearm injuries between 2010 and 2020 among adults in this large integrated healthcare system, primarily driven by non-self-inflicted firearm injuries; however, self-inflicted injuries decreased during this time. Injuries among youth showed no significant change.

Beyond clinical outcomes: Case control study of the role of race in disruptive life events for people with serious mental illness.

OBJECTIVE: To understand how race and serious mental illness (SMI) interact for disruptive life events defined as financial (bankruptcy and judgement filings), and non-financial (arrests). METHODS: Patients were adults with schizophrenia (SCZ; N = 16,159) or bipolar I disorder (BPI; N = 30,008) matched 1:1 to patients without SMI (non-SMI) from health systems in Michigan and Southern California during 1/1/2007 through 12/31/2018. The main exposure was self-reported race, and the outcome was disruptive life events aggregated by Transunion. We hypothesized that Black patients with SCZ or BPI would be the most likely to experience a disruptive life event when compared to Black patients without SMI, and all White or Asian patients regardless of mental illness. RESULTS: Black patients with SCZ had the least likelihood (37% lower) and Asian patients with BPI had the greatest likelihood (2.25 times higher) of experiencing a financial disruptive life event among all patients in the study. There was no interaction of race with either SCZ or BPI for experiencing an arrest. The findings did not support our hypotheses for patients with SCZ and partially supported them for patients with BPI. CONCLUSIONS: Clinical initiatives to assess social determinants of health should consider a focus on Asian patients with BPI.

Using natural language processing to identify the status of homelessness and housing instability among serious illness patients from clinical notes in an integrated healthcare system.

Background: Efficiently identifying the social risks of patients with serious illnesses (SIs) is the critical first step in providing patient-centered and value-driven care for this medically vulnerable population. Objective: To apply and further hone an existing natural language process (NLP) algorithm that identifies patients who are homeless/at risk of homeless to a SI population. Methods: Patients diagnosed with SI between 2019 and 2020 were identified using an adapted list of diagnosis codes from the Center for Advance Palliative Care from the Kaiser Permanente Southern California electronic health record. Clinical notes associated with medical encounters within 6 months before and after the diagnosis date were processed by a previously developed NLP algorithm to identify patients who were homeless/at risk of homelessness. To improve the generalizability to the SI population, the algorithm was refined by multiple iterations of chart review and adjudication. The updated algorithm was then applied to the SI population. Results: Among 206 993 patients with a SI diagnosis, 1737 (0.84%) were identified as homeless/at risk of homelessness. These patients were more likely to be male (51.1%), age among 45-64 years (44.7%), and have one or more emergency visit (65.8%) within a year of their diagnosis date. Validation of the updated algorithm yielded a sensitivity of 100.0% and a positive predictive value of 93.8%. Conclusions: The improved NLP algorithm effectively identified patients with SI who were homeless/at risk of homelessness and can be used to target interventions for this vulnerable group.

Association of neighborhood gentrification and residential moves with hypertension and diabetes control in Los Angeles County, 2014-2019: A retrospective cohort study.

OBJECTIVE: To examine whether gentrification exposure is associated with future hypertension and diabetes control. METHODS: Linking records from an integrated health care system to census-tract characteristics, we identified adults with hypertension and/or diabetes residing in stably low-SES census tracts in 2014 (n = 69,524). We tested associations of census tract gentrification occurring between 2015 and 2019 with participants' disease control in 2019. Secondary analyses considered the role of residential moves (possible displacement), race and ethnicity, and age. RESULTS: Gentrification exposure was associated with improved odds of hypertension control (aOR: 1.08; 95% CI: 1.00, 1.17), especially among non-Hispanic Whites and adults >65 years. Gentrification was not associated with diabetes control overall, but control improved in the Hispanic subgroup. Disease control was similar regardless of residential moves in the overall sample, but disparate associations emerged in models stratified by race and ethnicity. CONCLUSIONS: Residents of newly gentrifying neighborhoods may experience modestly improved odds of hypertension and/or diabetes control, but associations may differ across population subgroups. POLICY IMPLICATIONS: Gentrification may support-or at least not harm-cardiometabolic health for some residents. City leaders and health systems could partner with impacted communities to ensure that neighborhood development meets the goals and health needs of all residents and does not exacerbate health disparities.

Air pollution and the sequelae of COVID-19 patients: A multistate analysis.

IMPORTANCE: Recent evidence links air pollution to the severity COVID-19 symptoms and to death from the disease. To date, however, few studies have assessed whether air pollution affects the sequelae to more severe states or recovery from COVID-19 in a cohort with individual data. OBJECTIVE: To assess how air pollution affects the transition to more severe COVID-19 states or to recovery from COVID-19 infection in a cohort with detailed patient information. DESIGN AND OUTCOMES: We used a cohort design that followed patients admitted to hospital in the Kaiser Permanente Southern California (KPSC) Health System, which has 4.7 million members with characteristics similar to the general population. Enrollment began on 06/01/2020 and ran until 01/30/2021 for all patients admitted to hospital while ill with COVID-19. All possible states of sequelae were considered, including deterioration to intensive care, to death, discharge to recovery, or discharge to death. Transition risks were estimated with a multistate model. We assessed exposure using chemical transport model that predicted ambient concentrations of nitrogen dioxide, ozone, and fine particulate matter (PM2.5) at a 1 km scale. RESULTS: Each increase in PM2.5 concentration equivalent to the interquartile range was associated with increased risk of deterioration to intensive care (HR of 1.16; 95% CI: 1.12-1.20) and deterioration to death (HR of 1.11; 95% CI: 1.04-1.17). Results for ozone were consistent with PM2.5 effects, but ozone also affected the transition from recovery to death: HR of 1.24 (95% CI: 1.01-1.51). NO2 had weaker effects but displayed some elevated risks. CONCLUSIONS: PM2.5 and ozone were significantly associated with transitions to more severe states while in hospital and to death after discharge from hospital. Reducing air pollution could therefore lead to improved prognosis for COVID-19 patients and a sustainable means of reducing the health impacts of coronaviruses now and in the future.

Assessment of Disruptive Life Events for Individuals Diagnosed With Schizophrenia or Bipolar I Disorder Using Data From a Consumer Credit Reporting Agency.

Importance: There is a dearth of population-level data on major disruptive life events (defined here as arrests by a legal authority, address changes, bankruptcy, lien, and judgment filings) for patients with bipolar I disorder (BPI) or schizophrenia, which has limited studies on mental health and treatment outcomes. Objective: To conduct a population-level study on disruptive life events by using publicly available data on disruptive life events, aggregated by a consumer credit reporting agency in conjunction with electronic health record (EHR) data. Design, Setting, and Participants: This study used EHR data from 2 large, integrated health care systems, Kaiser Permanente Southern California and Henry Ford Health. Cohorts of patients diagnosed from 2007 to 2019 with BPI or schizophrenia were matched 1:1 by age at analysis, age at diagnosis (if applicable), sex, race and ethnicity, and Medicaid status to (1) an active comparison group with diagnoses of major depressive disorder (MDD) and (2) a general health (GH) cohort without diagnoses of BPI, schizophrenia, or MDD. Patients with diagnoses of BPI or schizophrenia and their respective comparison cohorts were matched to public records data aggregated by a consumer credit reporting agency (98% match rate). Analysis took place between November 2020 and December 2022. Main Outcomes and Measures: The differences in the occurrence of disruptive life events among patients with BPI or schizophrenia and their comparison groups. Results: Of 46 167 patients, 30 008 (65%) had BPI (mean [SD] age, 42.6 [14.2] years) and 16 159 (35%) had schizophrenia (mean [SD], 41.4 [15.1] years). The majoriy of patients were White (30 167 [65%]). In addition, 18 500 patients with BPI (62%) and 6552 patients with schizophrenia (41%) were female. Patients with BPI were more likely to change addresses than patients in either comparison cohort (with the incidence ratio being as high as 1.25 [95% CI, 1.23-1.28]) when compared with GH cohort. Patients with BPI were also more likely to experience any of the financial disruptive life events with odds ratio ranging from 1.15 [95% CI, 1.07-1.24] to 1.50 [95% CI, 1.42-1.58]). The largest differences in disruptive life events were seen in arrests of patients with either BPI or schizophrenia compared with GH peers (3.27 [95% CI, 2.84-3.78] and 3.04 [95% CI, 2.57-3.59], respectively). Patients with schizophrenia had fewer address changes and were less likely to experience a financial event than their matched comparison cohorts. Conclusions and Relevance: This study demonstrated that data aggregated by a consumer credit reporting agency can support population-level studies on disruptive life events among patients with BPI or schizophrenia.

Association of Cost-Driven Residential Moves With Health-Related Outcomes Among California Renters.

Importance: Unaffordable housing is associated with adverse health-related outcomes, but little is known about the associations between moving due to unaffordable housing and health-related outcomes. Objective: To characterize the association of recent cost-driven residential moves with health-related outcomes. Design, Setting, and Participants: This cross-sectional study involved a weighted multivariable regression analysis of California Health Interview Survey data from January 1, 2011, to December 31, 2017. A population-based sample of 52 646 adult renters and other nonhomeowners in California were included. Data were analyzed from March 2, 2021, to January 6, 2023. Exposure: Cost-driven moves in the past 3 years relative to no move and to non-cost-driven moves. Main Outcomes and Measures: Five outcomes were assessed: psychological distress (low, moderate, or severe, as categorized by the 6-item Kessler Psychological Distress Scale), emergency department [ED] visits in the past year (any vs none), preventive care visits in the past year (any vs none), general health (poor or fair vs good, very good, or excellent), and walking for leisure in the past 7 days (in minutes). Results: Among 52 646 adult renters and other nonhomeowners, 50.3% were female, 85.2% were younger than 60 years, 45.3% were Hispanic, and 55.1% had income lower than 200% of the federal poverty level. Overall, 8.9% of renters reported making a recent cost-driven move, with higher prevalence among Hispanic (9.9%) and non-Hispanic Black (11.3%) renters compared with non-Hispanic White renters (7.2%). In multivariable models, compared with not moving, cost-driven moving was associated with a 4.2 (95% CI, 2.6-5.7) percentage point higher probability of experiencing moderate psychological distress; a 3.2 (95% CI, 1.9-4.5) percentage point higher probability of experiencing severe psychological distress; a 2.5 (95% CI, 0-4.9) percentage point higher probability of ED visits; a 5.1 (95% CI, 1.6-8.6) percentage point lower probability of having preventive care visits; a 3.7 (95% CI, 1.2-6.2) percentage point lower probability of having good, very good, or excellent general health; and 16.8 (95% CI, 6.9-26.6) fewer minutes of walking for leisure. General health, psychological distress, and walking for leisure were also worse with cost-driven moves relative to non-cost-driven moves, with a 3.2 (95% CI, 1.7-4.7) percentage point higher probability of experiencing moderate psychological distress; a 2.5 (95% CI, 1.2-3.9) percentage point higher probability of experiencing severe psychological distress; a 4.6 (95% CI, 2.1-7.2) percentage point lower probability of having good, very good, or excellent general health; and 13.0 (95% CI, 4.0-21.9) fewer minutes of walking for leisure. However, the incidence of preventive care and ED visits did not differ between those who made cost-driven vs non-cost-driven moves. Conclusions and Relevance: In this study, cost-driven moves were associated with adverse health-related outcomes relative to not moving and to non-cost-driven moves. These findings suggest that policies to improve housing affordability, prevent displacement, and increase access to health care for groups vulnerable to cost-driven moves may have the potential to improve population health equity, especially during the current national housing affordability crisis.

Associations of Physical Inactivity and COVID-19 Outcomes Among Subgroups.

INTRODUCTION: Physical activity before COVID-19 infection is associated with less severe outcomes. The study determined whether a dose‒response association was observed and whether the associations were consistent across demographic subgroups and chronic conditions. METHODS: A retrospective cohort study of Kaiser Permanente Southern California adult patients who had a positive COVID-19 diagnosis between January 1, 2020 and May 31, 2021 was created. The exposure was the median of at least 3 physical activity self-reports before diagnosis. Patients were categorized as follows: always inactive, all assessments at 10 minutes/week or less; mostly inactive, median of 0-60 minutes per week; some activity, median of 60-150 minutes per week; consistently active, median>150 minutes per week; and always active, all assessments>150 minutes per week. Outcomes were hospitalization, deterioration event, or death 90 days after a COVID-19 diagnosis. Data were analyzed in 2022. RESULTS: Of 194,191 adults with COVID-19 infection, 6.3% were hospitalized, 3.1% experienced a deterioration event, and 2.8% died within 90 days. Dose‒response effects were strong; for example, patients in the some activity category had higher odds of hospitalization (OR=1.43; 95% CI=1.26, 1.63), deterioration (OR=1.83; 95% CI=1.49, 2.25), and death (OR=1.92; 95% CI=1.48, 2.49) than those in the always active category. Results were generally consistent across sex, race and ethnicity, age, and BMI categories and for patients with cardiovascular disease or hypertension. CONCLUSIONS: There were protective associations of physical activity for adverse COVID-19 outcomes across demographic and clinical characteristics. Public health leaders should add physical activity to pandemic control strategies.

Air pollution and meteorology as risk factors for COVID-19 death in a cohort from Southern California.

BACKGROUND: Recent evidence links ambient air pollution to COVID-19 incidence, severity, and death, but few studies have analyzed individual-level mortality data with high quality exposure models. METHODS: We sought to assess whether higher air pollution exposures led to greater risk of death during or after hospitalization in confirmed COVID-19 cases among patients who were members of the Kaiser Permanente Southern California (KPSC) healthcare system (N=21,415 between 06-01-2020 and 01-31-2022 of whom 99.85 % were unvaccinated during the study period). We used 1 km resolution chemical transport models to estimate ambient concentrations of several common air pollutants, including ozone, nitrogen dioxide, and fine particle matter (PM2.5). We also derived estimates of pollutant exposures from ultra-fine particulate matter (PM0.1), PM chemical species, and PM sources. We employed Cox proportional hazards models to assess associations between air pollution exposures and death from COVID-19 among hospitalized patients. FINDINGS: We found significant associations between COVID-19 death and several air pollution exposures, including: PM2.5 mass, PM0.1 mass, PM2.5 nitrates, PM2.5 elemental carbon, PM2.5 on-road diesel, and PM2.5 on-road gasoline. Based on the interquartile (IQR) exposure increment, effect sizes ranged from hazard ratios (HR) = 1.12 for PM2.5 mass and PM2.5 nitrate to HR âˆ¼ 1.06-1.07 for other species or source markers. Humidity and temperature in the month of diagnosis were also significant negative predictors of COVID-19 death and negative modifiers of the air pollution effects. INTERPRETATION: Air pollution exposures and meteorology were associated the risk of COVID-19 death in a cohort of patients from Southern California. These findings have implications for prevention of death from COVID-19 and for future pandemics.

Factors Associated With Firearm Injury Among Pediatric Members of a Large Integrated Healthcare System.

BACKGROUND AND OBJECTIVES: Few studies have tested multiple socio-ecological risk factors assocated with firearm injury among pediatric populations and distinguished self-inflicted from non-self-inflicted injury. To address this gap, the current study examined demographic, individual psychosocial, and neighborhood variables as risk factors for firearm injury among a large cohort of children and adolescents. METHODS: Retrospective cohort study. Data were obtained from the electronic health records of a large integrated healthcare system. The cohort included children <18 years with at least one clinical encounter between January 1, 2010 and December 31, 2018. Poisson regression was used to examine demographic (age, gender, race and ethnicity, Medicaid status), psychosocial (depression, substance use disorder, medical comorbidities), and neighborhood education variables as potential risk factors for non-self-inflicted and self-inflicted firearm injuries. RESULTS: For non-self-inflicted injury, the highest relative risk was found for children age 12-17 years old compared to 0-5 year olds (RR = 37.57); other risk factors included male gender, Black and Hispanic race and ethnicity (compared to White race), being a Medicaid recipient, lower neighborhood education, and substance use disorder diagnosis. For self-inflicted injury, only age 12-17 years old and male gender were associated with increased risk. CONCLUSIONS: These results reinforce the established higher risk for firearm injury among adolescent males, highlight differences between self-inflicted and non-self-inflicted injuries, and the need to consider demographic, psychosocial, and neighborhood variables as risk factors to inform interventions aimed to reduce firearm injuries among children and adolescents.

Development and validation of the COVID-19 Hospitalized Patient Deterioration Index.

OBJECTIVES: To develop a COVID-19-specific deterioration index for hospitalized patients: the COVID Hospitalized Patient Deterioration Index (COVID-HDI). This index builds on the proprietary Epic Deterioration Index, which was not developed for predicting respiratory deterioration events among patients with COVID-19. STUDY DESIGN: A retrospective observational cohort was used to develop and validate the COVID-HDI model to predict respiratory deterioration or death among hospitalized patients with COVID-19. Deterioration events were defined as death or requiring high-flow oxygen, bilevel positive airway pressure, mechanical ventilation, or intensive-level care within 72 hours of run time. The sample included hospitalized patients with COVID-19 diagnoses or positive tests at Kaiser Permanente Southern California between May 3, 2020, and October 17, 2020. METHODS: Machine learning models and 118 candidate predictors were used to generate benchmark performance. Logit regression with least absolute shrinkage and selection operator and physician input were used to finalize the model. Split-sample cross-validation was used to train and test the model. RESULTS: The area under the receiver operating curve was 0.83. COVID-HDI identifies patients at low risk (negative predictive value [NPV] > 98.5%) and borderline low risk (NPV > 95%) of an event. Of all patients, 74% were identified as being at low or borderline low risk at some point during their hospitalization and could be considered for discharge with or without home monitoring. A high-risk group with a positive predictive value of 51% included 12% of patients. Model performance remained high in a recent cohort of patients. CONCLUSIONS: COVID-HDI is a parsimonious, well-calibrated, and accurate model that may support clinical decision-making around discharge and escalation of care.

Cross-sectional Comparison of Disparities by Race Using White vs Hispanic as Reference Among Children and Youths With Developmental Disabilities Referred for Speech Therapy.

Importance: Health care research on racial disparities among children and youths has historically used the White race as a reference category with which other racial and ethnic groups are compared, which may inadvertently set up Whiteness as a standard for health. Objective: To compare 2 interpretations of an analysis of racial disparities in speech therapy receipt among children and youths with developmental disabilities: a traditional, White-referenced analysis and a Hispanic majority-referenced analysis. Design, Setting, and Participants: This cross-sectional study used multiple logistic regression to analyze speech therapy referrals for children, adolescents, and transition age youths in an integrated health care system in Southern California from 2017 to 2020. Eligible participants were children and youths up to age 26 years with 1 or more diagnosed intellectual or developmental disability (eg, autism spectrum disorder, speech or language delay, developmental delay, Down syndrome, and others). Exposures: Child or youth race and ethnicity as reported by parents or caregivers (Asian, Black and African American, Hispanic and Latinx, American Indian or Alaskan Native, Native Hawaiian or Pacific Islander, White, multiple, and other). Main Outcomes and Measures: Receipt of speech therapy within 1 year of referral. Results: A total 66 402 referrals were included; 65 833 referrals (99.1%) were for children under age 17 years, 47 323 (71.3%) were for boys, and 39 959 (60.2%) were commercially insured. A majority of participants were identified as Hispanic (36 705 [55.3%]); 6167 (9.3%) were identified as Asian, 4810 (7.2%) as Black, and 14 951 (22.5%) as White. In the traditional racial disparities model where the reference category was White, referrals of children and youths who identified as Hispanic, Black, Pacific Islander, and other had lower odds of actual receipt of speech therapy compared with referrals for White children and youths (Hispanic: OR, 0.79; 95% CI, 0.75-0.83; Black: OR, 0.72; 95% CI, 0.66-0.78; Pacific Islander: OR, 0.74; 95% CI, 0.57-0.98). When using the majority race group (Hispanic) as the reference category, referrals for children and youths who identified as White (OR, 1.26; 95% CI, 1.20-1.30), Asian (OR, 1.21; 95% CI, 1.12-1.30), and multiracial (OR, 1.35; 95% CI, 1.08-1.71) had higher odds of resulting in actual service receipt in comparison with referrals for Hispanic children and youths. Conclusions and Relevance: The cross-sectional study demonstrates the value of decentering Whiteness in interpreting racial disparities research and considering racial differences against multiple referents. Racial disparities researchers should consider investigating multiple between-group differences instead of exclusively using White as the default reference category.

Social Risk Factors and Desire for Assistance Among Patients Receiving Subsidized Health Care Insurance in a US-Based Integrated Delivery System.

PURPOSE: Because social conditions such as food insecurity and housing instability shape health outcomes, health systems are increasingly screening for and addressing patients' social risks. This study documented the prevalence of social risks and examined the desire for assistance in addressing those risks in a US-based integrated delivery system. METHODS: A survey was administered to Kaiser Permanente members on subsidized exchange health insurance plans (2018-2019). The survey included questions about 4 domains of social risks, desire for help, and attitudes. We conducted a descriptive analysis and estimated multivariate modified Poisson regression models. RESULTS: Of 438 participants, 212 (48%) reported at least 1 social risk factor. Housing instability was the most common (70%) factor reported. Members with social risks reported more discomfort being screened for social risks (14.2% vs 5.4%; P = .002) than those without risks, although 90% of participants believed that health systems should assist in addressing social risks. Among those with 1-2 social risks, however, only 27% desired assistance. Non-Hispanic Black participants who reported a social risk were more than twice as likely to desire assistance compared with non-Hispanic White participants (adjusted relative risk [RR] 2.2; 95% CI, 1.3-3.8). CONCLUSIONS: Athough most survey participants believed health systems have a role in addressing social risks, a minority of those reporting a risk wanted assistance and reported more discomfort being screened for risk factors than those without risks. Health systems should work to increase the comfort of patients in reporting risks, explore how to successfully assist them when desired, and offer resources to address these risks outside the health care sector.VISUAL ABSTRACT.

Development and assessment of a natural language processing model to identify residential instability in electronic health records' unstructured data: a comparison of 3 integrated healthcare delivery systems.

OBJECTIVE: To evaluate whether a natural language processing (NLP) algorithm could be adapted to extract, with acceptable validity, markers of residential instability (ie, homelessness and housing insecurity) from electronic health records (EHRs) of 3 healthcare systems. MATERIALS AND METHODS: We included patients 18 years and older who received care at 1 of 3 healthcare systems from 2016 through 2020 and had at least 1 free-text note in the EHR during this period. We conducted the study independently; the NLP algorithm logic and method of validity assessment were identical across sites. The approach to the development of the gold standard for assessment of validity differed across sites. Using the EntityRuler module of spaCy 2.3 Python toolkit, we created a rule-based NLP system made up of expert-developed patterns indicating residential instability at the lead site and enriched the NLP system using insight gained from its application at the other 2 sites. We adapted the algorithm at each site then validated the algorithm using a split-sample approach. We assessed the performance of the algorithm by measures of positive predictive value (precision), sensitivity (recall), and specificity. RESULTS: The NLP algorithm performed with moderate precision (0.45, 0.73, and 1.0) at 3 sites. The sensitivity and specificity of the NLP algorithm varied across 3 sites (sensitivity: 0.68, 0.85, and 0.96; specificity: 0.69, 0.89, and 1.0). DISCUSSION: The performance of this NLP algorithm to identify residential instability in 3 different healthcare systems suggests the algorithm is generally valid and applicable in other healthcare systems with similar EHRs. CONCLUSION: The NLP approach developed in this project is adaptable and can be modified to extract types of social needs other than residential instability from EHRs across different healthcare systems.

Association Between Financial Distress with Patient and Caregiver Outcomes in Home-Based Palliative Care: A Secondary Analysis of a Clinical Trial.

BACKGROUND: Serious illness often causes financial hardship for patients and families. Home-based palliative care (HBPC) may partly address this. OBJECTIVE: Describe the prevalence and characteristics of patients and family caregivers with high financial distress at HBPC admission and examine the relationship between financial distress and patient and caregiver outcomes. DESIGN, SETTINGS, AND PARTICIPANTS: Data for this cohort study were drawn from a pragmatic comparative-effectiveness trial testing two models of HBPC in Kaiser Permanente. We included 779 patients and 438 caregivers from January 2019 to January 2020. MEASUREMENTS: Financial distress at admission to HBPC was measured using a global question (0-10-point scale: none=0; mild=1-5; moderate/severe=6+). Patient- (Edmonton Symptom Assessment Scale, distress thermometer, PROMIS-10) and caregiver (Preparedness for Caregiving, Zarit-12 Burden, PROMIS-10)-reported outcomes were measured at baseline and 1 month. Hospital utilization was captured using electronic medical records and claims. Mixed-effects adjusted models assessed survey measures and a proportional hazard competing risk model assessed hospital utilization. RESULTS: Half of the patients reported some level of financial distress with younger patients more likely to have moderate/severe financial distress. Patients with moderate/severe financial distress at HBPC admission reported worse symptoms, general distress, and quality of life (QoL), and caregivers reported worse preparedness, burden, and QoL (all, p<.001). Compared to patients with no financial distress, moderate/severe financial distress patients had more social work contacts, improved symptom burden at 1 month (ESAS total score: -4.39; 95% CI: -7.61, -1.17; p<.01), and no increase in hospital-based utilization (adjusted hazard ratio: 1.11; 95% CI: 0.87-1.40; p=.41); their caregivers had improved PROMIS-10 mental scores (+2.68; 95% CI: 0.20, 5.16; p=.03). No other group differences were evident in the caregiver preparedness, burden, and physical QoL change scores. CONCLUSION: These findings highlight the importance and need for routine assessments of financial distress and for provision of social supports required to help families receiving palliative care services.

Generalizability of an Automatic Explanation Method for Machine Learning Prediction Results on Asthma-Related Hospital Visits in Patients With Asthma: Quantitative Analysis.

BACKGROUND: Asthma exerts a substantial burden on patients and health care systems. To facilitate preventive care for asthma management and improve patient outcomes, we recently developed two machine learning models, one on Intermountain Healthcare data and the other on Kaiser Permanente Southern California (KPSC) data, to forecast asthma-related hospital visits, including emergency department visits and hospitalizations, in the succeeding 12 months among patients with asthma. As is typical for machine learning approaches, these two models do not explain their forecasting results. To address the interpretability issue of black-box models, we designed an automatic method to offer rule format explanations for the forecasting results of any machine learning model on imbalanced tabular data and to suggest customized interventions with no accuracy loss. Our method worked well for explaining the forecasting results of our Intermountain Healthcare model, but its generalizability to other health care systems remains unknown. OBJECTIVE: The objective of this study is to evaluate the generalizability of our automatic explanation method to KPSC for forecasting asthma-related hospital visits. METHODS: Through a secondary analysis of 987,506 data instances from 2012 to 2017 at KPSC, we used our method to explain the forecasting results of our KPSC model and to suggest customized interventions. The patient cohort covered a random sample of 70% of patients with asthma who had a KPSC health plan for any period between 2015 and 2018. RESULTS: Our method explained the forecasting results for 97.57% (2204/2259) of the patients with asthma who were correctly forecasted to undergo asthma-related hospital visits in the succeeding 12 months. CONCLUSIONS: For forecasting asthma-related hospital visits, our automatic explanation method exhibited an acceptable generalizability to KPSC. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): RR2-10.2196/resprot.5039.

Longitudinal trajectories of adiposity-related measures from age 2-5 years in a population of low-income Hispanic children.

BACKGROUND: We estimated longitudinal trajectories of body mass index (BMI) z-score and percentile, weight for height (WFH) z-score and percentile, and percentage of the 95th BMI percentile (BMIp95) among low-income Hispanic children ages 2-5 years to provide normative data for this population and compare the behavior of different measures. METHODS: Longitudinal height and weight measurements obtained from 18,072 Hispanic children aged 2-5 years enrolled in the Special Supplemental Nutrition Program for Women, Infants and Children in Los Angeles County were analyzed. Trajectories of adiposity-related measures were estimated using mixed models, stratified by sex and BMI percentile at age 2 years. RESULTS: For children in the 5th-85th BMI percentile at age 2 years, all adiposity-related measures rose during ages 2-3.5 years; during ages 3.5-5 years, BMI-based measures increased, BMIp95 decreased, and WFH-based measures were stable. For children exceeding the 85th BMI percentile at age 2 years, measures generally trended downward during ages 2-5 years, except for BMIp95, which had variable trends. CONCLUSIONS: Adiposity measures changed at different rates as children grew during ages 2-3.5 years compared to ages 3.5-5 years, and different measures displayed different trends. Studies should consider examining multiple measures and focusing on change relative to a comparison group. IMPACT: To address the childhood obesity epidemic, information on normative trajectories of adiposity-related measures in at-risk populations of young children is needed. Longitudinal analysis of data collected from low-income Hispanic children during ages 2-5 years revealed different patterns for different adiposity measures and for ages 2-3.5 years versus 3.5-5 years. Child obesity studies should consider examining multiple adiposity measures and focus on change relative to a comparison group to avoid misinterpreting longitudinal patterns.