Clusters Based on Within-Treatment Symptom Trajectories as Predictors of Dropout in Treatment for Posttraumatic Stress Disorder and Substance Use Disorder.

Objective: Dropout rates are high in treatments for co-occurring posttraumatic stress disorder (PTSD) and substance use disorders (SUDs). We examined dropout predictors in PTSD-SUD treatment. Methods: Participants were 183 veterans receiving integrated or phased motivational enhancement therapy and prolonged exposure. Using survival models, we examined demographics and symptom trajectories as dropout predictors. Using latent trajectory analysis, we incorporated clusters based on symptom trajectories to improve dropout prediction. Results: Hispanic ethnicity (integrated arm), Black or African American race (phased arm), and younger age (phased arm) predicted dropout. Clusters based on PTSD and substance use trajectories improved dropout prediction. In integrated treatment, participants with consistently-high use and low-and-improving use had the highest dropout. In phased treatment, participants with the highest and lowest PTSD symptoms had lower dropout; participants with the lowest substance use had higher dropout. Conclusions: Identifying within-treatment symptom trajectories associated with dropout can help clinicians intervene to maximize outcomes. ClinicalTrials.gov Identifier: NCT01211106.

Longitudinal invariance of the Patient Health Questionnaire-9 among patients receiving pharmacotherapy for major depressive disorder: A secondary analysis of clinical trial data.

Comparing self-reported symptom scores across time requires longitudinal measurement invariance (LMI), a psychometric property that means the measure is functioning identically across all time points. Despite its prominence as a measure of depression symptom severity in both research and health care, LMI has yet to be firmly established for the Patient Health Questionnaire-9 depression module (PHQ-9), particularly over the course of antidepressant pharmacotherapy. Accordingly, the objective of this study was to assess for LMI of the PHQ-9 during pharmacotherapy for major depressive disorder. This was a secondary analysis of data collected during a randomized controlled trial. A total of 1,944 veterans began antidepressant monotherapy and completed the PHQ-9 six times over 24 weeks of treatment. LMI was assessed using a series of four confirmatory factor analysis models that included all six time points, with estimated parameters increasingly constrained across models to test for different aspects of invariance. Root-mean-square error of approximation of the chi-square difference test values below 0.06 indicated the presence of LMI. Exploratory LMI analyses were also performed for separate sex, age, and race subgroups. Root-mean-square error of approximation of the chi-square difference test showed minimal change in model fits during invariance testing (≤ 0.06 for all steps), supporting full LMI for the PHQ-9. LMI was also supported for all tested veteran subgroups. As such, PHQ-9 sum scores can be compared across extended pharmacotherapy treatment durations. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Three Lessons Learned About Power and Improving Recruitment of Underrepresented Populations in Clinical Trials.

This Open Forum is relevant for investigators who conduct research with historically understudied and marginalized populations. The authors introduce a U.S. Department of Veterans Affairs clinical trial that experienced challenges with recruitment of African American or Black veterans and was terminated for not achieving its recruitment goals. The role of power dynamics in clinical research is discussed, specifically how unequal distributions of power may create recruitment challenges. The authors summarize three lessons learned and offer recommendations for sharing power equitably between investigators and potential participants. By recounting these experiences, the authors seek to promote culturally sensitive, veteran-centered approaches to recruitment in future clinical trials.

Tools to implement measurement-based care (MBC) in the treatment of opioid use disorder (OUD): toward a consensus.

BACKGROUND: The prevalence and associated overdose death rates from opioid use disorder (OUD) have dramatically increased in the last decade. Despite more available treatments than 20 years ago, treatment access and high discontinuation rates are challenges, as are personalized medication dosing and making timely treatment changes when treatments fail. In other fields such as depression, brief measures to address these tasks combined with an action plan-so-called measurement-based care (MBC)-have been associated with better outcomes. This workgroup aimed to determine whether brief measures can be identified for using MBC for optimizing dosing or informing treatment decisions in OUD. METHODS: The National Institute on Drug Abuse Center for the Clinical Trials Network (NIDA CCTN) in 2022 convened a small workgroup to develop consensus about clinically usable measures to improve the quality of treatment delivery with MBC methods for OUD. Two clinical tasks were addressed: (1) to identify the optimal dose of medications for OUD for each patient and (2) to estimate the effectiveness of a treatment for a particular patient once implemented, in a more granular fashion than the binary categories of early or sustained remission or no remission found in The Diagnostic and Statistical Manual of Mental Disorders, fifth edition (DSM-5). DISCUSSION: Five parameters were recommended to personalize medication dose adjustment: withdrawal symptoms, opioid use, magnitude (severity and duration) of the subjective effects when opioids are used, craving, and side effects. A brief rating of each OUD-specific parameter to adjust dosing and a global assessment or verbal question for side-effects was viewed as sufficient. Whether these ratings produce better outcomes (e.g., treatment engagement and retention) in practice deserves study. There was consensus that core signs and symptoms of OUD based on some of the 5 DSM-5 domains (e.g., craving, withdrawal) should be the basis for assessing treatment outcome. No existing brief measure was found to meet all the consensus recommendations. Next steps would be to select, adapt or develop de novo items/brief scales to inform clinical decision-making about dose and treatment effectiveness. Psychometric testing, assessment of acceptability and whether the use of such scales produces better symptom control, quality of life (QoL), daily function or better prognosis as compared to treatment as usual deserves investigation.

Longitudinal Influence of Prescribed Antidepressants on Fecal and Oral Microbiomes Among Veterans With Major Depressive Disorder.

OBJECTIVE: The purpose of this study was to evaluate the influence of a new course of antidepressant monotherapy on gut and oral microbiomes and the relationship to depressive symptoms. METHODS: Longitudinal microbiome samples obtained from 10 U.S. veterans were analyzed. Baseline samples were taken before a new course of antidepressant monotherapy (either switching from a previous treatment or starting a new treatment). Targeted genomic sequencing of the microbiome samples was used to analyze changes in taxonomy and diversity across participants, medications, and medication class. Associations between these changes and Patient Health Questionnaire-9 (PHQ-9) scores were analyzed. RESULTS: Taxonomic variability was observed across participants, with the individual being the main microbial community driver. In terms of the fecal microbiome, antidepressants were associated with shifts toward Bacteroides being less abundant and Blautia, Pseudomonas, or Faecalibacterium being more abundant. Likewise, the composition of the oral microbiome was variable, with individual participants being the primary drivers of community composition. In the oral samples, the relative abundance of Haemophilus decreased after antidepressants were started. Increases in Blautia and decreases in Bacteroides were associated with lower PHQ-9 scores. CONCLUSIONS: Antidepressants were found to influence fecal and oral microbiomes such that a new course of antidepressant monotherapy was associated with taxonomic alterations toward healthier states in both fecal and oral microbiomes, which were associated with decreases in depressive symptoms. Additional longitudinal research is required to increase understanding of microbiomes and symptom-based changes, with a particular focus on potential differences between medication classes and underlying mechanisms.

Using electronic health record metadata to predict housing instability amongst veterans.

Housing instability is considered a significant life stressor and preemptive screening should be applied to identify those at risk for homelessness as early as possible so that they can be targeted for specialized care. We developed models to classify patient outcomes for an established VA Homelessness Screening Clinical Reminder (HSCR), which identifies housing instability, in the two months prior to its administration. Logistic Regression and Random Forest models were fit to classify responses using the last 18 months of document activity. We measure concentration of risk across stratifications of predicted probability and observe an enriched likelihood of finding confirmed false negative responses from veterans with diagnosed housing instability. Positive responses were 34 times more likely to be detected within the top 1 % of patients predicted at risk than from those randomly selected. There is a 1 in 4 chance of detecting false negatives within the top 1 % of predicted risk. Machine learning methods can classify between episodes of housing instability using a data-driven approach that does not rely on variables curated from domain experts. This method has the potential to improve clinicians' ability to identify veterans who are experiencing housing instability but are not captured by HSCR.

High dimensional predictions of suicide risk in 4.2 million US Veterans using ensemble transfer learning.

We present an ensemble transfer learning method to predict suicide from Veterans Affairs (VA) electronic medical records (EMR). A diverse set of base models was trained to predict a binary outcome constructed from reported suicide, suicide attempt, and overdose diagnoses with varying choices of study design and prediction methodology. Each model used twenty cross-sectional and 190 longitudinal variables observed in eight time intervals covering 7.5 years prior to the time of prediction. Ensembles of seven base models were created and fine-tuned with ten variables expected to change with study design and outcome definition in order to predict suicide and combined outcome in a prospective cohort. The ensemble models achieved c-statistics of 0.73 on 2-year suicide risk and 0.83 on the combined outcome when predicting on a prospective cohort of [Formula: see text] 4.2 M veterans. The ensembles rely on nonlinear base models trained using a matched retrospective nested case-control (Rcc) study cohort and show good calibration across a diversity of subgroups, including risk strata, age, sex, race, and level of healthcare utilization. In addition, a linear Rcc base model provided a rich set of biological predictors, including indicators of suicide, substance use disorder, mental health diagnoses and treatments, hypoxia and vascular damage, and demographics.

The impact of substance use on posttraumatic stress disorder symptoms and treatment discontinuation.

This study examined the impact of ongoing substance use during posttraumatic stress disorder (PTSD) and substance use disorder (SUD) treatment on PTSD symptoms and treatment discontinuation. The study represents a secondary analysis of U.S. military veterans (N = 183) who participated in a randomized clinical trial for the treatment of both PTSD and SUD. Veterans mostly identified as Black (53.8%) or White (41.9%) and male (92.4%). Substance use, PTSD symptoms, and treatment discontinuation were measured at 4-week intervals throughout treatment. Predictors were the percentage of days with alcohol, cannabis, and other substance use (primarily cocaine and opioids) and the average number of alcoholic drinks per drinking day. Outcomes were PTSD symptoms and treatment discontinuation at concurrent and prospective assessments. Multilevel models accounted for the nested structure of the longitudinal data. Alcohol, cannabis, and other substance use did not predict PTSD symptoms or treatment discontinuation prospectively. Concurrently, we observed that as a participant's percentage of drinking days increased by 34.7% (i.e., 1 standard deviation), PTSD symptoms during the same period were 0.07 standard deviations higher (i.e., 1 point on the PCL), B = 0.03, p = .033. No other substances were related to PTSD symptoms concurrently. The findings demonstrate that PTSD symptoms improved regardless of substance use during exposure-based PTSD and SUD treatment, and treatment discontinuation was not associated with substance use. This study suggests that substance use during treatment cannot directly explain the poorer treatment outcomes observed in the literature on comorbid PTSD/SUD compared to PTSD-only populations.

A computer-assisted telephone collaborative care intervention provided by lay providers for the treatment of comorbid depression and at-risk drinking: Analysis of a randomized controlled trial.

INTRODUCTION: Virtual collaborative care for people with comorbid depression and at-risk drinking lacks strong evidence. Our aim was to assess the impact of 12 months of telephone collaborative care (tCC) versus enhanced usual care (eUC) on depression and drinking. METHODS: We performed a secondary analysis of the Primary care Assessment and Research of a Telephone intervention for Neuropsychiatric conditions with Education and Resources study (PARTNERs), a blinded randomized controlled trial. We examined 144 participants with comorbid depression and at-risk drinking, of which 129 were from the original sample whose data have been published, and 15 were studied since the original report had been published. PARTNERs compared eUC consisting of usual care plus assessment of symptoms at baseline, and 4, 8, and 12 months later vs. tCC consisting of eUC plus telephone-based coaching and symptom monitoring provided by a lay mental health technician to patients supervised by a psychiatrist. The study assessed depression response and remission using logistic regression; we assessed trajectory of drinking using Generalized-estimating equations (GEE). Baseline factors associated with likelihood of not exceeding number of drinks at 12 months were identified using decision trees. RESULTS: tCC produced a faster decline in the number of drinks than eUC (Wald Χ2 = 9.47, p = 0.02). However, drinking and depression outcomes did not differ significantly between the two groups at the end of treatment. Higher alcohol consumption at baseline (≥18 standard drinks per week in the tCC group and ≥11 standard drinks per week in the eUC group) was associated with a higher likelihood of having at-risk drinking after 12 months of treatment. CONCLUSIONS: Our findings suggest that, compared to eUC, tCC may accelerate drinking reductions in patients with comorbid depression and at-risk drinking. Both treatments were equally effective at the end of treatment for both depression and drinking outcomes.

Treating individuals with suicidal ideation in primary care: Patient-level characteristics associated with follow-up in the Collaborative Care Model.

INTRODUCTION: The Collaborative Care Model (CoCM) is an evidence-based approach which embeds behavioral health providers (BHPs) into primary care. Whether patients with suicidal ideation (SI) are willing to engage in CoCM is unclear. METHODS: Using Patient Health Questionnaire-9 (PHQ-9) administrative data from primary care practices within an urban academic health system, we identified patients with and without SI who were referred to a CoCM BHP. We compared engagement, defined as attendance at ≥1 CoCM visit, across groups. RESULTS: Between 2018 and 2022, 7391 primary care patients were referred to a CoCM BHP. Eight hundred and ninety-two of these patients reported SI on the PHQ-9 (754 on "several days" during the previous 2 weeks and 138 on "more than half or most days"). Across groups, most patients engaged in CoCM. Patients reporting SI on several days engaged at a lower rate (61.4%) than those reporting SI on more than half or most days (65.9%). Both SI groups engaged at a lower rate than the 6499 patients who did not report SI (67.5%). CONCLUSION: Most patients referred to a CoCM BHP engaged in ≥1 visit. Rates were lower for patients with SI, with the lowest rate among those reporting SI on several days.

Pharmacogenetic testing and monitoring of complete blood counts among Veterans newly prescribed thiopurine treatments: a retrospective cohort study.

Pharmacogenetic (PGx) testing before initiation of thiopurine treatment and CBC monitoring post-initiation helps avoid adverse events and ensure patient safety. This study aims to evaluate trends in PGx testing and CBC monitoring among Veterans prescribed azathioprine, thioguanine, or mercaptopurine to demonstrate VA's efforts to improve medication safety after an adverse event. To assess testing patterns, we used VA electronic health report data to identify 20,524 Veterans who first began thiopurine treatment between January 1, 2010, to December 31, 2021. Aggregate monthly counts of thiopurine prescriptions and associated lab tests were tabulated, and the trend in the proportion of patients tested was analyzed using the Mann-Kendall test. The proportion of patients undergoing PGx testing rose from 30.0% in 2010 to 47.5% in late 2014 (July-December). However, PGx testing and overall testing only increased slightly after the sentinel event, and orders levelled off over time at slightly lower levels than before the sentinel event. Very little change was seen in the overall proportion of individuals receiving any testing across all patients with new prescriptions from the time of the sentinel event in 2014 to the end of 2021. A large portion of patients prescribed thiopurine drugs did not receive testing that could help prevent the development of potential adverse events, leading to a predominantly reactive approach. Increased PGx testing may result in a more proactive approach to the prevention of adverse events due to genetic interaction.

Addressing the Increasing Mental Health Distress and Mental Illness Among Young Adults in the United States.

ABSTRACT: Recent surveys show rising numbers of young people who report anxiety and depression. Although much attention has focused on mental health of adolescent youth, less attention has been paid to young people as they transition into adulthood. Multiple factors may have contributed to this steady increase: greater exposure to social media, information, and distressing news via personal electronic devices; increased concerns regarding social determinants of health and climate change; and changing social norms due to increased mental health literacy and reduced stigma. The COVID-19 pandemic may have temporarily exacerbated symptoms and impacted treatment availability. Strategies to mitigate causal factors for depression and anxiety in young adults may include education and skills training for cognitive, behavioral, and social coping strategies, as well as healthier use of technology and social media. Policies must support the availability of health insurance and treatment, and clinicians can adapt interventions to encompass the specific concerns and needs of young adults.

GWAS Meta-Analysis of Suicide Attempt: Identification of 12 Genome-Wide Significant Loci and Implication of Genetic Risks for Specific Health Factors.

OBJECTIVE: Suicidal behavior is heritable and is a major cause of death worldwide. Two large-scale genome-wide association studies (GWASs) recently discovered and cross-validated genome-wide significant (GWS) loci for suicide attempt (SA). The present study leveraged the genetic cohorts from both studies to conduct the largest GWAS meta-analysis of SA to date. Multi-ancestry and admixture-specific meta-analyses were conducted within groups of significant African, East Asian, and European ancestry admixtures. METHODS: This study comprised 22 cohorts, including 43,871 SA cases and 915,025 ancestry-matched controls. Analytical methods across multi-ancestry and individual ancestry admixtures included inverse variance-weighted fixed-effects meta-analyses, followed by gene, gene-set, tissue-set, and drug-target enrichment, as well as summary-data-based Mendelian randomization with brain expression quantitative trait loci data, phenome-wide genetic correlation, and genetic causal proportion analyses. RESULTS: Multi-ancestry and European ancestry admixture GWAS meta-analyses identified 12 risk loci at p values <5×10-8. These loci were mostly intergenic and implicated DRD2, SLC6A9, FURIN, NLGN1, SOX5, PDE4B, and CACNG2. The multi-ancestry SNP-based heritability estimate of SA was 5.7% on the liability scale (SE=0.003, p=5.7×10-80). Significant brain tissue gene expression and drug set enrichment were observed. There was shared genetic variation of SA with attention deficit hyperactivity disorder, smoking, and risk tolerance after conditioning SA on both major depressive disorder and posttraumatic stress disorder. Genetic causal proportion analyses implicated shared genetic risk for specific health factors. CONCLUSIONS: This multi-ancestry analysis of suicide attempt identified several loci contributing to risk and establishes significant shared genetic covariation with clinical phenotypes. These findings provide insight into genetic factors associated with suicide attempt across ancestry admixture populations, in veteran and civilian populations, and in attempt versus death.

Impact of the collaborative care model on medical spending.

OBJECTIVES: The collaborative care model integrates mental health care into primary care. In 2017, CMS created new billing codes to reimburse collaborative care. We measured the impact of a program supported by these codes on medical spending. STUDY DESIGN: Quasi-experimental. METHODS: We identified a commercially insured and managed Medicare sample of 825 patients who received collaborative care services in 8 primary care practices. We used propensity score matching to match treated patients to potential controls, resulting in 569 patients per group. We performed a difference-in-differences regression analysis to evaluate the impact of collaborative care on total medical spending, including medical, psychiatric, and pharmaceutical claims. RESULTS: Collaborative care patients' mean total medical cost began to fall after a patient's third month in the program and fell below the mean cost of control patients at month 7. Difference-in-differences regressions indicate a nonsignificant savings in total medical cost of $29.35 per member per month for patients in collaborative care compared with matched controls (95% CI, -$226.52 to $167.82). Treated members incurred $34.11 (95% CI, $31.95-$36.27) higher primary care costs that were directly attributed to collaborative care, $19.91 (95% CI, $4.84-$34.98) higher costs for other mental or behavioral health care, and a nonsignificant reduction of $91.34 (95% CI, -$319.32 to $136.63) in inpatient costs. CONCLUSIONS: Modest spending on collaborative care services to address the behavioral health needs of patients did not increase overall health care costs. This is the first economic study of a collaborative care program supported by the new billing codes.

Deep sequential neural network models improve stratification of suicide attempt risk among US veterans.

OBJECTIVE: To apply deep neural networks (DNNs) to longitudinal EHR data in order to predict suicide attempt risk among veterans. Local explainability techniques were used to provide explanations for each prediction with the goal of ultimately improving outreach and intervention efforts. MATERIALS AND METHODS: The DNNs fused demographic information with diagnostic, prescription, and procedure codes. Models were trained and tested on EHR data of approximately 500 000 US veterans: all veterans with recorded suicide attempts from April 1, 2005, through January 1, 2016, each paired with 5 veterans of the same age who did not attempt suicide. Shapley Additive Explanation (SHAP) values were calculated to provide explanations of DNN predictions. RESULTS: The DNNs outperformed logistic and linear regression models in predicting suicide attempts. After adjusting for the sampling technique, the convolutional neural network (CNN) model achieved a positive predictive value (PPV) of 0.54 for suicide attempts within 12 months by veterans in the top 0.1% risk tier. Explainability methods identified meaningful subgroups of high-risk veterans as well as key determinants of suicide attempt risk at both the group and individual level. DISCUSSION AND CONCLUSION: The deep learning methods employed in the present study have the potential to significantly enhance existing suicide risk models for veterans. These methods can also provide important clues to explore the relative value of long-term and short-term intervention strategies. Furthermore, the explainability methods utilized here could also be used to communicate to clinicians the key features which increase specific veterans' risk for attempting suicide.

Towards shortening the Brief Addiction Monitor-Revised (BAM-R).

Introduction: The Brief Addiction Monitor-Revised (BAM-R) is a widely used, 17-item assessment of substance use, risk, and protective factors associated with recovery from substance use disorders. Despite wide adoption in the U.S. Department of Veterans Affairs (VA) and recommendations for use in measurement-based care (MBC), administration may not be feasible in many MBC settings due to time constraints. The purpose of this study was to derive a shortened version of the BAM-R for use in fast-paced healthcare settings. Methods: BAM-R data from 32,002 Veterans were obtained through the VA's Corporate Data Warehouse. We used logistic regression models to identify items for removal based on prediction of two clinical outcomes (90-day substance use disorder (SUD) treatment retention and 12-month mortality) and item-level sensitivity to change during substance use treatment. Results: Although no intake BAM-R items predicted SUD treatment retention or mortality, effect sizes for item-level sensitivity to change during substance use treatment varied from small to large. Seven items were judged as relevant for MBC of SUD. Among all BAM-R items, Heavy Alcohol Use, Self-Help, Drug Use, Craving, and Mood items demonstrated the greatest magnitude of sensitivity to change. Conclusions: Although additional research is recommended before a shortened BAM-R can be implemented in non-specialty MBC settings, we identified 5 BAM-R items with perceived clinical utility and scores that demonstrated evidence of sensitivity to change. Shortening the BAM-R increases feasibility of use, though more work is needed to optimize measurement for SUD MBC.

Providers' Use of Pharmacogenetic Testing to Inform Antidepressant Prescribing: Results of Qualitative Interviews.

OBJECTIVE: Pharmacogenetic testing (PGx) for patients experiencing depression has been associated with modest improvements in symptoms. However, little is known about providers' use of PGx, including how and for whom providers use the test results in clinical decision making. In this article, results from qualitative interviews on the experience of providers participating in a pragmatic trial of PGx are described; implications of the providers' experiences are highlighted to inform future implementation of PGx. METHODS: Interviews were conducted with providers participating in the trial (N=61) who treated veterans who had depression. Questions were informed by the Consolidated Framework for Implementation Research. A rapid analytic approach was used. RESULTS: Two main themes were identified: perceptions regarding which patients would likely benefit from PGx and approaches to using the test results in prescribing. Providers generally expressed positive experiences with using PGx results. However, the providers varied in application of the test results to clinical decision making regarding medications, were uncertain about how much to rely on the results, and differed in perceptions about which patients would benefit from PGx. CONCLUSIONS: To support future implementation, policies and procedures are needed, as well as mechanisms to support ongoing provider education on PGx.

Using iterative random forest to find geospatial environmental and Sociodemographic predictors of suicide attempts.

Introduction: Despite a recent global decrease in suicide rates, death by suicide has increased in the United States. It is therefore imperative to identify the risk factors associated with suicide attempts to combat this growing epidemic. In this study, we aim to identify potential risk factors of suicide attempt using geospatial features in an Artificial intelligence framework. Methods: We use iterative Random Forest, an explainable artificial intelligence method, to predict suicide attempts using data from the Million Veteran Program. This cohort incorporated 405,540 patients with 391,409 controls and 14,131 attempts. Our predictive model incorporates multiple climatic features at ZIP-code-level geospatial resolution. We additionally consider demographic features from the American Community Survey as well as the number of firearms and alcohol vendors per 10,000 people to assess the contributions of proximal environment, access to means, and restraint decrease to suicide attempts. In total 1,784 features were included in the predictive model. Results: Our results show that geographic areas with higher concentrations of married males living with spouses are predictive of lower rates of suicide attempts, whereas geographic areas where males are more likely to live alone and to rent housing are predictive of higher rates of suicide attempts. We also identified climatic features that were associated with suicide attempt risk by age group. Additionally, we observed that firearms and alcohol vendors were associated with increased risk for suicide attempts irrespective of the age group examined, but that their effects were small in comparison to the top features. Discussion: Taken together, our findings highlight the importance of social determinants and environmental factors in understanding suicide risk among veterans.

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.

Validity and reliability of in-person and remote oral fluids drug testing compared to urine drug testing.

BACKGROUND: Increased telehealth use has led to greater interest in remote drug testing. The speed, acceptability, and ability to observe oral fluids testing makes it the best candidate for remote drug testing, but its validity and reliability compared to gold-standard urine drug testing have not been established. METHODS: Veterans (N = 99) recruited from mental health clinics completed in-person and remote oral fluids testing and in-person urine drug testing. The validity of oral fluids versus urine drug testing and reliability of in-person versus remote oral fluids testing were evaluated. RESULTS: Validity of oral fluids testing was similar for samples collected in-person and virtually. Oral fluids testing had good specificity (0.93-1.00) and negative predictive value (0.85-1.00), but lower sensitivity and positive predictive value. Sensitivity (0.21-0.93) was highest for methadone and oxycodone, followed by cocaine and then amphetamine and opiates. Positive predictive value (0.14-1.00) was highest for cocaine, opiates, and methadone, followed by oxycodone and then amphetamine. Validity for cannabis was low, likely because of differences in detection windows for oral fluids versus urine drug screens. Reliability of remote oral fluids testing was adequate for opiates, cocaine, and methadone, but not oxycodone, amphetamine, or cannabis. CONCLUSIONS: Oral fluids testing identifies most negative, but not most positive, drug test results. While oral fluids testing is appropriate in some circumstances, its limitations should be acknowledged. Remote drug testing addresses many barriers, but also generates new barriers related to self-administration and remote interpretation. Limitations include a small sample and low base rates for some drugs.