Organizations in science and medicine must hold each other accountable for discriminatory practices.

Many organizations persist in working with others that engage in known, remediable structural discrimination. We name this practice interorganizational structural discrimination (ISD) and argue it is a pivotal contributor to inequities in science and medicine. We urge organizations to leverage their relationships and demand progress from collaborators.

Developing an individualized treatment rule for Veterans with major depressive disorder using electronic health records.

Efforts to develop an individualized treatment rule (ITR) to optimize major depressive disorder (MDD) treatment with antidepressant medication (ADM), psychotherapy, or combined ADM-psychotherapy have been hampered by small samples, small predictor sets, and suboptimal analysis methods. Analyses of large administrative databases designed to approximate experiments followed iteratively by pragmatic trials hold promise for resolving these problems. The current report presents a proof-of-concept study using electronic health records (EHR) of n = 43,470 outpatients beginning MDD treatment in Veterans Health Administration Primary Care Mental Health Integration (PC-MHI) clinics, which offer access not only to ADMs but also psychotherapy and combined ADM-psychotherapy. EHR and geospatial databases were used to generate an extensive baseline predictor set (5,865 variables). The outcome was a composite measure of at least one serious negative event (suicide attempt, psychiatric emergency department visit, psychiatric hospitalization, suicide death) over the next 12 months. Best-practices methods were used to adjust for nonrandom treatment assignment and to estimate a preliminary ITR in a 70% training sample and to evaluate the ITR in the 30% test sample. Statistically significant aggregate variation was found in overall probability of the outcome related to baseline predictors (AU-ROC = 0.68, S.E. = 0.01), with test sample outcome prevalence of 32.6% among the 5% of patients having highest predicted risk compared to 7.1% in the remainder of the test sample. The ITR found that psychotherapy-only was the optimal treatment for 56.0% of patients (roughly 20% lower risk of the outcome than if receiving one of the other treatments) and that treatment type was unrelated to outcome risk among other patients. Change in aggregate treatment costs of implementing this ITR would be negligible, as 16.1% fewer patients would be prescribed ADMs and 2.9% more would receive psychotherapy. A pragmatic trial would be needed to confirm the accuracy of the ITR.

Development of a model to predict psychotherapy response for depression among Veterans.

BACKGROUND: Fewer than half of patients with major depressive disorder (MDD) respond to psychotherapy. Pre-emptively informing patients of their likelihood of responding could be useful as part of a patient-centered treatment decision-support plan. METHODS: This prospective observational study examined a national sample of 807 patients beginning psychotherapy for MDD at the Veterans Health Administration. Patients completed a self-report survey at baseline and 3-months follow-up (data collected 2018-2020). We developed a machine learning (ML) model to predict psychotherapy response at 3 months using baseline survey, administrative, and geospatial variables in a 70% training sample. Model performance was then evaluated in the 30% test sample. RESULTS: 32.0% of patients responded to treatment after 3 months. The best ML model had an AUC (SE) of 0.652 (0.038) in the test sample. Among the one-third of patients ranked by the model as most likely to respond, 50.0% in the test sample responded to psychotherapy. In comparison, among the remaining two-thirds of patients, <25% responded to psychotherapy. The model selected 43 predictors, of which nearly all were self-report variables. CONCLUSIONS: Patients with MDD could pre-emptively be informed of their likelihood of responding to psychotherapy using a prediction tool based on self-report data. This tool could meaningfully help patients and providers in shared decision-making, although parallel information about the likelihood of responding to alternative treatments would be needed to inform decision-making across multiple treatments.

Development of a model to predict antidepressant treatment response for depression among Veterans.

BACKGROUND: Only a limited number of patients with major depressive disorder (MDD) respond to a first course of antidepressant medication (ADM). We investigated the feasibility of creating a baseline model to determine which of these would be among patients beginning ADM treatment in the US Veterans Health Administration (VHA). METHODS: A 2018-2020 national sample of n = 660 VHA patients receiving ADM treatment for MDD completed an extensive baseline self-report assessment near the beginning of treatment and a 3-month self-report follow-up assessment. Using baseline self-report data along with administrative and geospatial data, an ensemble machine learning method was used to develop a model for 3-month treatment response defined by the Quick Inventory of Depression Symptomatology Self-Report and a modified Sheehan Disability Scale. The model was developed in a 70% training sample and tested in the remaining 30% test sample. RESULTS: In total, 35.7% of patients responded to treatment. The prediction model had an area under the ROC curve (s.e.) of 0.66 (0.04) in the test sample. A strong gradient in probability (s.e.) of treatment response was found across three subsamples of the test sample using training sample thresholds for high [45.6% (5.5)], intermediate [34.5% (7.6)], and low [11.1% (4.9)] probabilities of response. Baseline symptom severity, comorbidity, treatment characteristics (expectations, history, and aspects of current treatment), and protective/resilience factors were the most important predictors. CONCLUSIONS: Although these results are promising, parallel models to predict response to alternative treatments based on data collected before initiating treatment would be needed for such models to help guide treatment selection.

Are Veterans Getting Their Preferred Depression Treatment? A National Observational Study in the Veterans Health Administration.

BACKGROUND: Physician responsiveness to patient preferences for depression treatment may improve treatment adherence and clinical outcomes. OBJECTIVE: To examine associations of patient treatment preferences with types of depression treatment received and treatment adherence among Veterans initiating depression treatment. DESIGN: Patient self-report surveys at treatment initiation linked to medical records. SETTING: Veterans Health Administration (VA) clinics nationally, 2018-2020. PARTICIPANTS: A total of 2582 patients (76.7% male, mean age 48.7 years, 62.3% Non-Hispanic White) MAIN MEASURES: Patient self-reported preferences for medication and psychotherapy on 0-10 self-anchoring visual analog scales (0="completely unwilling"; 10="completely willing"). Treatment receipt and adherence (refilling medications; attending 3+ psychotherapy sessions) over 3 months. Logistic regression models controlled for socio-demographics and geographic variables. KEY RESULTS: More patients reported strong preferences (10/10) for psychotherapy than medication (51.2% versus 36.7%, McNemar χ21=175.3, p<0.001). A total of 32.1% of patients who preferred (7-10/10) medication and 21.8% who preferred psychotherapy did not receive these treatments. Patients who strongly preferred medication were substantially more likely to receive medication than those who had strong negative preferences (odds ratios [OR]=17.5; 95% confidence interval [CI]=12.5-24.5). Compared with patients who had strong negative psychotherapy preferences, those with strong psychotherapy preferences were about twice as likely to receive psychotherapy (OR=1.9; 95% CI=1.0-3.5). Patients who strongly preferred psychotherapy were more likely to adhere to psychotherapy than those with strong negative preferences (OR=3.3; 95% CI=1.4-7.4). Treatment preferences were not associated with medication or combined treatment adherence. Patients in primary care settings had lower odds of receiving (but not adhering to) psychotherapy than patients in specialty mental health settings. Depression severity was not associated with treatment receipt or adherence. CONCLUSIONS: Mismatches between treatment preferences and treatment type received were common and associated with worse treatment adherence for psychotherapy. Future research could examine ways to decrease mismatch between patient preferences and treatments received and potential effects on patient outcomes.

Stereotactic radiotherapy for hepatocellular carcinoma: Expanding the multidisciplinary armamentarium.

Hepatocellular carcinoma (HCC) is the fifth most common malignancy worldwide and the third most common cause of cancer-related death. Long-term prognosis remains poor with treatment options frequently limited by advanced tumor stage, tumor location, or underlying liver dysfunction. Stereotactic ablative body radiotherapy (SABR) utilizes technological advances to deliver highly precise, tumoricidal doses of radiation. There is an emerging body of literature on SABR in HCC demonstrating high rates of local control in the order of 80-90% at 3 years. SABR is associated with a low risk of radiation-induced liver disease or decompensation in appropriately selected HCC patients with compensated liver function and is now being incorporated into guidelines as an additional treatment option. This review outlines the emerging role of SABR in the multidisciplinary management of HCC and summarizes the current evidence for its use as an alternative ablative option for early-stage disease, as a bridge to transplant, and as palliation for advanced-stage disease. We outline specific considerations regarding patient selection, toxicities, and response assessment. Finally, we compare current international guidelines and recommendations for the use of SABR and summarize ongoing studies.

Immune checkpoint inhibitors in advanced nasopharyngeal carcinoma: Beyond an era of chemoradiation?

Now is an exciting era of development in immunotherapy checkpoint inhibitors and their effect on the treatment of NPC. While the general prognosis of R/M disease is poor, immunotherapy offers some promise in a malignancy associated with EBV and characterized by a peritumoural immune infiltrate. Our study aims to review past and on-going clinical trials of monoclonal antibody therapies against the checkpoint inhibitors (e.g. PD1 and CTLA-4), in R/M NPC. All randomized and nonrandomized controlled trials involving immune checkpoint inhibitor interventions for treatment of NPC were included in the study. We utilized a validated "risk of bias" tool to assess study quality. Four separate Phase I-II trials report the potential of PD1 inhibitor treatment for patients with NPC. Within the observed groups, camrelizumab combined with chemotherapy achieved an objective response in 91% of patients as first-line treatment for metastatic NPC (PFS 68% at 1-year) but this was associated with a high rate of grade >3 adverse events (87%; CTCAE version 4.03). The remaining three studies focused on recurrent NPC disease in patients who had received at least one line of prior chemotherapy. Within this group, camrelizumab monotherapy achieved an objective response in 34% of patients (PFS 27% at 1-year; range across all three studies 20.5-34%). No NPC trial has yet reported on specific outcomes for non-PD1 checkpoint inhibitors but 11 on-going studies include alternative targets (e.g. PD-L1/CTLA-4) as combination or monotherapy treatments. In considering checkpoint immunotherapies for NPC, initial results show promise for anti-PD1 interventions. Further phase I-III trials are in progress to clarify clinical outcomes, fully determine safety profiles, and optimize drug combinations and administration schedules.

Associations of Lifetime Traumatic Brain Injury Characteristics With Prospective Suicide Attempt Among Deployed US Army Soldiers.

OBJECTIVE: To estimate associations of lifetime traumatic brain injury (TBI) characteristics with prospective suicide attempt among US Army soldiers. METHOD: The Army STARRS (Study to Assess Risk and Resilience in Servicemembers) Pre/Post Deployment Study surveyed 3 Brigade Combat Teams that were deployed to Afghanistan in 2012. Lifetime TBI and past-month postconcussive/post-TBI symptoms were evaluated at predeployment baseline. Recency and number of TBIs were quantified, and TBI severity was classified on the basis of reports of alteration/loss of consciousness and memory lapse. Suicide attempt data came from administrative records and surveys administered after return from deployment. Logistic regression models estimated associations of TBI characteristics with prospective suicide attempt among baseline respondents who were deployed (n = 7677), adjusting for other risk factors including lifetime mental disorder. RESULTS: One hundred three soldiers made a suicide attempt over a median follow-up period of 30 months (weighted prevalence = 1.31% [0.14%]). In the final model estimating joint associations of TBI severity/recency and past-month postconcussive/post-TBI symptoms, only postconcussive/post-TBI symptoms were associated with a higher risk of suicide attempt (per standard score increase: AOR [adjusted odds ratio] = 1.31; 95% CI, 1.05-1.63; P = .012). CONCLUSIONS: Among the lifetime TBI characteristics evaluated at predeployment baseline, only past-month postconcussive/post-TBI symptoms were prospectively associated with an increased risk of suicide attempt following deployment. Detection of postconcussive/post-TBI symptoms could facilitate targeting of Army suicide prevention programs.

Suicidal thoughts and behaviors among college students and same-aged peers: results from the World Health Organization World Mental Health Surveys.

PURPOSE: The primary aims are to (1) obtain representative prevalence estimates of suicidal thoughts and behaviors (STB) among college students worldwide and (2) investigate whether STB is related to matriculation to and attrition from college. METHODS: Data from the WHO World Mental Health Surveys were analyzed, which include face-to-face interviews with 5750 young adults aged 18-22 spanning 21 countries (weighted mean response rate = 71.4%). Standardized STB prevalence estimates were calculated for four well-defined groups of same-aged peers: college students, college attriters (i.e., dropouts), secondary school graduates who never entered college, and secondary school non-graduates. Logistic regression assessed the association between STB and college entrance as well as attrition from college. RESULTS: Twelve-month STB in college students was 1.9%, a rate significantly lower than same-aged peers not in college (3.4%; OR 0.5; p < 0.01). Lifetime prevalence of STB with onset prior to age 18 among college entrants (i.e., college students or attriters) was 7.2%, a rate significantly lower than among non-college attenders (i.e., secondary school graduates or non-graduates; 8.2%; OR 0.7; p = 0.03). Pre-matriculation onset STB (but not post-matriculation onset STB) increased the odds of college attrition (OR 1.7; p < 0.01). CONCLUSION: STB with onset prior to age 18 is associated with reduced likelihood of college entrance as well as greater attrition from college. Future prospective research should investigate the causality of these associations and determine whether targeting onset and persistence of childhood-adolescent onset STB leads to improved educational attainment.

Childhood adversities and post-traumatic stress disorder: evidence for stress sensitisation in the World Mental Health Surveys.

BackgroundAlthough childhood adversities are known to predict increased risk of post-traumatic stress disorder (PTSD) after traumatic experiences, it is unclear whether this association varies by childhood adversity or traumatic experience types or by age.AimsTo examine variation in associations of childhood adversities with PTSD according to childhood adversity types, traumatic experience types and life-course stage.MethodEpidemiological data were analysed from the World Mental Health Surveys (n = 27 017).ResultsFour childhood adversities (physical and sexual abuse, neglect, parent psychopathology) were associated with similarly increased odds of PTSD following traumatic experiences (odds ratio (OR) = 1.8), whereas the other eight childhood adversities assessed did not predict PTSD. Childhood adversity-PTSD associations did not vary across traumatic experience types, but were stronger in childhood-adolescence and early-middle adulthood than later adulthood.ConclusionsChildhood adversities are differentially associated with PTSD, with the strongest associations in childhood-adolescence and early-middle adulthood. Consistency of associations across traumatic experience types suggests that childhood adversities are associated with generalised vulnerability to PTSD following traumatic experiences.

Long-term outcomes in advanced anterior skull base malignancy: a single quaternary institution experience.

BACKGROUND: Advanced skull base malignancies are a heterogenous subset of head and neck cancers, and management is often complex. In recent times, there has been a paradigm shift in surgical technique and the advent of novel systemic options. Our goal was to analyse the long-term outcomes of a single quaternary head and neck and skull base service. METHODS: A retrospective review of 127 patients with advanced anterior skull base malignancies that were treated at our institution between 1999 and 2015 was performed. Multiple variables were investigated to assess their significance on 5 and 10-year outcomes. RESULTS: The mean age was 60.9 (± 12.6 SD). Sixty-four percent were males and 36% were females. Ninety percent of patients had T4 disease. Median survival time was 133 months. The 5-year overall survival (OS) was 66.2%, disease-specific survival (DSS) was 74.7%, and recurrence-free survival (RFS) was 65.0%. The 10-year OS was 55.1%, DSS was 72.1%, and RFS was 53.4%. Histological type and margin status significantly affected OS & DSS. CONCLUSION: Surgical management of advanced skull base tumours has evolved over the last few decades at our institution with acceptable survival outcomes and complication rates. Histological diagnosis and margin status are the main predictors of survival. The addition of neoadjuvant systemic agents in current trials may improve outcomes.

Workforce Diversity, Equity, and Inclusion: A Crucial Component of Professionalism in Psychiatry.

Documented disparities have profoundly impacted the training and careers of physicians from socially and historically marginalized groups, including women, people with disabilities, people who identify with racial and ethnic minority groups, and the lesbian, gay, bisexual, transgender, and queer or questioning+ community. Professionalism is a core component of medical training and practice, yet a focus on workforce diversity, equity, and inclusion is often absent. This report aims to encourage the adoption of workforce diversity, equity, and inclusion as a crucial component of professionalism, with an emphasis on the field of psychiatry.

Characterising B cell expression and prognostic significance in human papillomavirus positive oropharyngeal cancer.

OBJECTIVES: The incidence of human papillomavirus positive oropharyngeal cancer (HPV+OPC) is increasing, and new biomarkers are required to better define prognostic groups and guide treatment. Infiltrating T cells have been well studied in head and neck cancer, however the presence and role of B cells and tertiary lymphoid structures (TLS) in the tumor microenvironment has not, even though the interplay between T and B cells is increasingly being recognised. MATERIALS AND METHODS: Using CD20 immunohistochemistry (IHC) to identify B cells and TLS in a cohort of 159 HPV + OPC patients, we semi-quantitatively scored abundance and location (intra-tumoral or stromal) and correlated findings with patient survival. RESULTS: 32% (51/157) of patients had high intra-tumoral (IT) abundance of CD20+ B cells (≥5%) and this was prognostic for improved overall survival (OS) with an adjusted hazard ratio (HR) of 0.2 (95 % CI 0.0-0.7, p = 0.014). We validated our results in an independent cohort comprising 171 HPV + OPC where 14% (23/171) were IT CD20+ high, again showing improved survival with an adjusted HR for OS of 0.2 (95 % CI 0.0-1.4, p = 0.003). Neither stromal abundance nor the presence of TLS were prognostic in either cohort. B cells were subtyped by multispectral IHC, identifying CD20+CD27+ cells, consistent with memory B cells, as the predominant subtype. Combined with validated biomarker CD103, a marker of tissue-resident memory T cells, IT CD20+ B cells abundance was able to prognostically stratify patients further. CONCLUSIONS: CD20+ B cell abundance has the potential to be used as a biomarker to identify good and poor prognosis HPV + OPC patients.

Estimated Average Treatment Effect of Psychiatric Hospitalization in Patients With Suicidal Behaviors: A Precision Treatment Analysis.

Importance: Psychiatric hospitalization is the standard of care for patients presenting to an emergency department (ED) or urgent care (UC) with high suicide risk. However, the effect of hospitalization in reducing subsequent suicidal behaviors is poorly understood and likely heterogeneous. Objectives: To estimate the association of psychiatric hospitalization with subsequent suicidal behaviors using observational data and develop a preliminary predictive analytics individualized treatment rule accounting for heterogeneity in this association across patients. Design, Setting, and Participants: A machine learning analysis of retrospective data was conducted. All veterans presenting with suicidal ideation (SI) or suicide attempt (SA) from January 1, 2010, to December 31, 2015, were included. Data were analyzed from September 1, 2022, to March 10, 2023. Subgroups were defined by primary psychiatric diagnosis (nonaffective psychosis, bipolar disorder, major depressive disorder, and other) and suicidality (SI only, SA in past 2-7 days, and SA in past day). Models were trained in 70.0% of the training samples and tested in the remaining 30.0%. Exposures: Psychiatric hospitalization vs nonhospitalization. Main Outcomes and Measures: Fatal and nonfatal SAs within 12 months of ED/UC visits were identified in administrative records and the National Death Index. Baseline covariates were drawn from electronic health records and geospatial databases. Results: Of 196 610 visits (90.3% men; median [IQR] age, 53 [41-59] years), 71.5% resulted in hospitalization. The 12-month SA risk was 11.9% with hospitalization and 12.0% with nonhospitalization (difference, -0.1%; 95% CI, -0.4% to 0.2%). In patients with SI only or SA in the past 2 to 7 days, most hospitalization was not associated with subsequent SAs. For patients with SA in the past day, hospitalization was associated with risk reductions ranging from -6.9% to -9.6% across diagnoses. Accounting for heterogeneity, hospitalization was associated with reduced risk of subsequent SAs in 28.1% of the patients and increased risk in 24.0%. An individualized treatment rule based on these associations may reduce SAs by 16.0% and hospitalizations by 13.0% compared with current rates. Conclusions and Relevance: The findings of this study suggest that psychiatric hospitalization is associated with reduced average SA risk in the immediate aftermath of an SA but not after other recent SAs or SI only. Substantial heterogeneity exists in these associations across patients. An individualized treatment rule accounting for this heterogeneity could both reduce SAs and avert hospitalizations.

Refining stereotactic body radiation therapy as a bridge to transplantation for hepatocellular carcinoma: An institutional experience.

BACKGROUND: Stereotactic body radiotherapy (SBRT) has been established as a safe and effective treatment for hepatocellular carcinoma (HCC). Currently, there are no consensus guidelines to advise optimal patient selection and radiotherapy planning parameters to minimise the risk of surgical and medical complications after liver transplant (LT) in patients who have had prior SBRT for HCC, whilst optimising treatment benefit. METHODS: We performed a retrospective analysis of all adult patients who received liver SBRT as a bridge to LT at a tertiary institution between 2017 and 2019. RESULTS: Nine patients received SBRT as bridging therapy to LT. HCC location varied from peripheral to central/hilar regions and HCC diameter was 13-54 mm. Median time between SBRT and LT was 141 days (range 27-461 days). Median operating time was 360 min (range 270-480 min). Four patients (44%) had visible SBRT reaction or fibrosis at the time of LT. SBRT reaction resulted in clinical impact in one patient (11%) only, where vascular clamping of the IVC was required for 10 min. CONCLUSION: SBRT is a safe and effective treatment for HCC enabling patients to remain within LT criteria, even for lesions not amenable to other more conventional bridging therapies. We describe a preliminary decision pathway to guide the optimal use of SBRT as a bridge to LT developed in our institution.

A prospective observational cohort study examining the development of head and neck lymphedema from the time of diagnosis.

INTRODUCTION: Head and neck lymphedema can occur in the internal or external structures of the head and neck region. Little is known about the development of this condition over the course of treatment for head and neck cancer. This study aimed to observe the development of internal and external lymphedema from diagnosis to 12 weeks postacute treatment. METHODS: A single center, prospective observational cohort study assessed participants for external lymphedema, internal lymphedema, quality of life, and symptom burden. Assessments were conducted prior to starting radiotherapy (RT), at the end of RT, 6 and 12 weeks after RT. RESULTS: Forty-six participants were recruited. External lymphedema as measured by percentage water content, increased from 41.9 at baseline (95% CI: 39.3-44.4) to 50.4 (95% CI: 46.0-54.8) at 12 weeks following RT (p-value < .001). After adjusting for changes in weight and participant age at baseline, a general increase in tape measurements was observed over time with significant increases from baseline to 12 weeks post-RT for all measurement points. By 12 weeks post-RT, all participants had lymphedema present in eight of 13 internal sites assessed. CONCLUSIONS: Internal and external head and neck lymphedema was observed to increase from baseline to 12 weeks after completion of RT without abatement. People with head and neck cancer should be educated about the potentially extended duration of this treatment side effect. Further research is required to determine the point at which swelling symptoms recede.

Development of a model to predict combined antidepressant medication and psychotherapy treatment response for depression among veterans.

BACKGROUND: Although research shows that more depressed patients respond to combined antidepressants (ADM) and psychotherapy than either alone, many patients do not respond even to combined treatment. A reliable prediction model for this could help treatment decision-making. We attempted to create such a model using machine learning methods among patients in the US Veterans Health Administration (VHA). METHODS: A 2018-2020 national sample of VHA patients beginning combined depression treatment completed self-report assessments at baseline and 3 months (n = 658). A learning model was developed using baseline self-report, administrative, and geospatial data to predict 3-month treatment response defined by reductions in the Quick Inventory of Depression Symptomatology Self-Report and/or in the Sheehan Disability Scale. The model was developed in a 70 % training sample and tested in the remaining 30 % test sample. RESULTS: 30.0 % of patients responded to treatment. The prediction model had a test sample AUC-ROC of 0.657. A strong gradient was found in probability of treatment response from 52.7 % in the highest predicted quintile to 14.4 % in the lowest predicted quintile. The most important predictors were episode characteristics (symptoms, comorbidities, history), personality/psychological resilience, recent stressors, and treatment characteristics. LIMITATIONS: Restrictions in sample definition, a low recruitment rate, and reliance on patient self-report rather than clinician assessments to determine treatment response limited the generalizability of results. CONCLUSIONS: A machine learning model could help depressed patients and providers predict likely response to combined ADM-psychotherapy. Parallel information about potential harms and costs of alternative treatments would be needed, though, to inform optimal treatment selection.

Evaluation of a Model to Target High-risk Psychiatric Inpatients for an Intensive Postdischarge Suicide Prevention Intervention.

Importance: The months after psychiatric hospital discharge are a time of high risk for suicide. Intensive postdischarge case management, although potentially effective in suicide prevention, is likely to be cost-effective only if targeted at high-risk patients. A previously developed machine learning (ML) model showed that postdischarge suicides can be predicted from electronic health records and geospatial data, but it is unknown if prediction could be improved by adding additional information. Objective: To determine whether model prediction could be improved by adding information extracted from clinical notes and public records. Design, Setting, and Participants: Models were trained to predict suicides in the 12 months after Veterans Health Administration (VHA) short-term (less than 365 days) psychiatric hospitalizations between the beginning of 2010 and September 1, 2012 (299 050 hospitalizations, with 916 hospitalizations followed within 12 months by suicides) and tested in the hospitalizations from September 2, 2012, to December 31, 2013 (149 738 hospitalizations, with 393 hospitalizations followed within 12 months by suicides). Validation focused on net benefit across a range of plausible decision thresholds. Predictor importance was assessed with Shapley additive explanations (SHAP) values. Data were analyzed from January to August 2022. Main Outcomes and Measures: Suicides were defined by the National Death Index. Base model predictors included VHA electronic health records and patient residential data. The expanded predictors came from natural language processing (NLP) of clinical notes and a social determinants of health (SDOH) public records database. Results: The model included 448 788 unique hospitalizations. Net benefit over risk horizons between 3 and 12 months was generally highest for the model that included both NLP and SDOH predictors (area under the receiver operating characteristic curve range, 0.747-0.780; area under the precision recall curve relative to the suicide rate range, 3.87-5.75). NLP and SDOH predictors also had the highest predictor class-level SHAP values (proportional SHAP = 64.0% and 49.3%, respectively), although the single highest positive variable-level SHAP value was for a count of medications classified by the US Food and Drug Administration as increasing suicide risk prescribed the year before hospitalization (proportional SHAP = 15.0%). Conclusions and Relevance: In this study, clinical notes and public records were found to improve ML model prediction of suicide after psychiatric hospitalization. The model had positive net benefit over 3-month to 12-month risk horizons for plausible decision thresholds. Although caution is needed in inferring causality based on predictor importance, several key predictors have potential intervention implications that should be investigated in future studies.