Clinical decision support for bipolar depression using large language models.

Management of depressive episodes in bipolar disorder remains challenging for clinicians despite the availability of treatment guidelines. In other contexts, large language models have yielded promising results for supporting clinical decisionmaking. We developed 50 sets of clinical vignettes reflecting bipolar depression and presented them to experts in bipolar disorder, who were asked to identify 5 optimal next-step pharmacotherapies and 5 poor or contraindicated choices. The same vignettes were then presented to a large language model (GPT4-turbo; gpt-4-1106-preview), with or without augmentation by prompting with recent bipolar treatment guidelines, and asked to identify the optimal next-step pharmacotherapy. Overlap between model output and gold standard was estimated. The augmented model prioritized the expert-designated optimal choice for 508/1000 vignettes (50.8%, 95% CI 47.7-53.9%; Cohen's kappa = 0.31, 95% CI 0.28-0.35). For 120 vignettes (12.0%), at least one model choice was among the poor or contraindicated treatments. Results were not meaningfully different when gender or race of the vignette was permuted to examine risk for bias. By comparison, an un-augmented model identified the optimal treatment for 234 (23.0%, 95% CI 20.8-26.0%; McNemar's p < 0.001 versus augmented model) of the vignettes. A sample of community clinicians scoring the same vignettes identified the optimal choice for 23.1% (95% CI 15.7-30.5%) of vignettes, on average; McNemar's p < 0.001 versus augmented model. Large language models prompted with evidence-based guidelines represent a promising, scalable strategy for clinical decision support. In addition to prospective studies of efficacy, strategies to avoid clinician overreliance on such models, and address the possibility of bias, will be needed.

Problematic alcohol use and its impact on liver disease quality of life in a multicenter study of patients with cirrhosis.

BACKGROUND: Management of cirrhosis is challenging and has been complicated by the COVID-19 pandemic due to decreased access to care, increased psychological distress, and alcohol misuse. Recently, The National Institute on Alcohol Abuse and Alcoholism has broadened the definition of recovery from alcohol use disorder to include quality of life (QoL) as an indicator of recovery. This study examined the associations of alcohol-associated cirrhosis etiology and problematic drinking with liver disease QoL (LDQoL). METHODS: Patients with cirrhosis (N=329) were recruited from 3 sites (63% from 2 Veterans Affairs Health Care Systems and 37% from 1 safety net hospital) serving populations that are economically or socially marginalized. Cirrhosis etiology was ascertained by chart review of medical records. Problematic drinking was defined by ≥8 on the Alcohol Use Disorders Identification Test. Multivariable general linear modeling adjusting for age, sex, race/ethnicity, site, pandemic-related stress, and history of anxiety/depressive disorder were conducted. Sensitivity analyses further adjusted for indicators of liver disease severity. RESULTS: Participants were on average 64.6 years old, 17% female, 58% non-White, 44% with alcohol-associated cirrhosis, and 17% with problematic drinking. Problematic drinking was significantly associated with worse LDQoL scores in the overall scale and in the memory/concentration and health distress subscales. These associations remained significant after adjusting for indicators of liver disease severity, including Model for End-Stage Liver Disease-Sodium score and decompensated cirrhosis status. CONCLUSIONS: Among patients with cirrhosis, problematic drinking was associated with worse LDQoL, especially in the domains of memory/concentration and health distress. Assessment and awareness of cognitive deficits and negative emotionality within the context of cirrhosis and problematic drinking may help clinicians provide better integrated care for this population.

Lithium plus antipsychotics or anticonvulsants for bipolar disorder: Comparing clinical response and metabolic changes.

OBJECTIVE: Patients with bipolar disorder treated with lithium often require additional antipsychotics or anticonvulsants. However, the comparative effectiveness and safety of these agents as add-on to lithium has not been studied. METHODS: This secondary analysis combined two similar 24-week trials on outpatients with bipolar disorder randomized to lithium (target serum level 0.4-0.6 mEq/L). Guideline-based adjunctive antipsychotics (Li+AP) and anticonvulsants (Li+AC) could be used if clinically indicated and was assessed at every study visit. Response was measured on the Clinical Global Impression scale and we performed adjusted mixed effects linear regression analyses. Analysis of variance tests compared metabolic measures including a binary diagnosis of metabolic syndrome before and after 24 weeks of treatment. RESULTS: Among 379 outpatients (57% female, mean age 38 years, mean Clinical Global Impression 4.4), users of Li+AP (N = 50, primarily quetiapine and aripiprazole) improved to a similar degree (mean Clinical Global Impression improvement = 1.6, standard deviation = 1.5) as those using lithium-only (i.e. without adjunctive antipsychotics or anticonvulsants, N = 149, mean Clinical Global Impression improvement = 1.7, standard deviation = 1.4) (p = 0.59). Users of Li+AC (N = 107, primarily lamotrigine and valproate, mean Clinical Global Impression improvement = 1.2, standard deviation = 1.3) and users of Li+AP+AC (N = 73, mean Clinical Global Impression improvement = 1.1, standard deviation = 1.3) showed worse response compared to lithium-only users (all p < 0.01). When comparing Li+AP to Li+AC, users of Li+AP improved slightly better on general (p = 0.05) and manic symptoms (p = 0.01), but showed a worse development of glucose, triglycerides, and metabolic syndrome. CONCLUSION: Despite treatment-by-indication confounding, these findings are relevant for real-world treatment settings and emphasize the need for randomized trials on this clinically important topic.

Effect of Pharmacogenomic Testing for Drug-Gene Interactions on Medication Selection and Remission of Symptoms in Major Depressive Disorder: The PRIME Care Randomized Clinical Trial.

Importance: Selecting effective antidepressants for the treatment of major depressive disorder (MDD) is an imprecise practice, with remission rates of about 30% at the initial treatment. Objective: To determine whether pharmacogenomic testing affects antidepressant medication selection and whether such testing leads to better clinical outcomes. Design, Setting, and Participants: A pragmatic, randomized clinical trial that compared treatment guided by pharmacogenomic testing vs usual care. Participants included 676 clinicians and 1944 patients. Participants were enrolled from 22 Department of Veterans Affairs medical centers from July 2017 through February 2021, with follow-up ending November 2021. Eligible patients were those with MDD who were initiating or switching treatment with a single antidepressant. Exclusion criteria included an active substance use disorder, mania, psychosis, or concurrent treatment with a specified list of medications. Interventions: Results from a commercial pharmacogenomic test were given to clinicians in the pharmacogenomic-guided group (n = 966). The comparison group received usual care and access to pharmacogenomic results after 24 weeks (n = 978). Main Outcomes and Measures: The co-primary outcomes were the proportion of prescriptions with a predicted drug-gene interaction written in the 30 days after randomization and remission of depressive symptoms as measured by the Patient Health Questionnaire-9 (PHQ-9) (remission was defined as PHQ-9 ≤ 5). Remission was analyzed as a repeated measure across 24 weeks by blinded raters. Results: Among 1944 patients who were randomized (mean age, 48 years; 491 women [25%]), 1541 (79%) completed the 24-week assessment. The estimated risks for receiving an antidepressant with none, moderate, and substantial drug-gene interactions for the pharmacogenomic-guided group were 59.3%, 30.0%, and 10.7% compared with 25.7%, 54.6%, and 19.7% in the usual care group. The pharmacogenomic-guided group was more likely to receive a medication with a lower potential drug-gene interaction for no drug-gene vs moderate/substantial interaction (odds ratio [OR], 4.32 [95% CI, 3.47 to 5.39]; P < .001) and no/moderate vs substantial interaction (OR, 2.08 [95% CI, 1.52 to 2.84]; P = .005) (P < .001 for overall comparison). Remission rates over 24 weeks were higher among patients whose care was guided by pharmacogenomic testing than those in usual care (OR, 1.28 [95% CI, 1.05 to 1.57]; P = .02; risk difference, 2.8% [95% CI, 0.6% to 5.1%]) but were not significantly higher at week 24 when 130 patients in the pharmacogenomic-guided group and 126 patients in the usual care group were in remission (estimated risk difference, 1.5% [95% CI, -2.4% to 5.3%]; P = .45). Conclusions and Relevance: Among patients with MDD, provision of pharmacogenomic testing for drug-gene interactions reduced prescription of medications with predicted drug-gene interactions compared with usual care. Provision of test results had small nonpersistent effects on symptom remission. Trial Registration: ClinicalTrials.gov Identifier: NCT03170362.

COVID-19 and substance use disorders: a review of international guidelines for frontline healthcare workers of addiction services.

BACKGROUND: People with substance use disorders may be at a greater risk of contracting COVID-19 infection and developing medical complications. Several institutional and governmental health agencies across the world developed ad hoc guidance for substance use disorder services and care of individuals misusing substances. We aimed to synthesise the best available recommendations on management and care of people with or at risk of substance use disorders during the COVID-19 pandemic from existing guidelines published in UK, USA, Australia, Canada, New Zealand, and Singapore. METHODS: We systematically searched existing guidelines and websites from 28 international institutions and governmental bodies in the context of the COVID-19 pandemic (May 4th 2021). We summarized the extracted data as answers to specific clinical questions. RESULTS: We organised the available recommendations from 19 sources in three sections. First, we focused on general advice and recommendations for people who misuse alcohol or drugs during the COVID-19 pandemic, the design of contingency plans, safeguarding issues for children and families of service users and advice to the public, patients, and carers. Then, we summarised specific guidelines for people who use illicit drugs and related services, such as opioid substitution treatment and needle and syringe programmes. Finally, we provided a synthesis on specific recommendations for services supporting people who misuse alcohol and key topics in the field, such as management of alcohol detoxification and safe transition between supervised and unsupervised consumption. CONCLUSIONS: Available guidance reflected different approaches, ranging from being extremely cautious in providing recommendations other than generic statements to proposing adaptation of previously available guidelines to confront the challenges of the COVID-19 pandemic. After the early phase, guidance focused on reduction of infection transmission and service delivery. Guidance did not provide advice on infection prevention via vaccination programmes and service access strategies tailored to individuals with substance use disorders.

Childhood trauma and treatment outcomes during mood-stabilising treatment with lithium or quetiapine among outpatients with bipolar disorder.

BACKGROUND: Childhood trauma affects the course of mood disorders. Researchers are now considering childhood trauma as an influential factor in the treatment of mood disorders. However, the role of childhood trauma in the treatment of bipolar disorder remains understudied. METHODS: The effect of childhood trauma on treatment outcomes was evaluated among participants randomised to treatment with lithium or quetiapine in the Clinical and Health Outcomes Initiatives in Comparative Effectiveness for Bipolar Disorder (Bipolar CHOICE) study by clinician assessment. Mixed effects linear regression models were used to analyse rates of improvement in symptom severity (assessed with the Bipolar Inventory of Symptoms Scale and the Clinical Global Impression Scale for Bipolar Disorder) and functional impairment (assessed with the Longitudinal Interval Follow-up Evaluation-Range of Impaired Functioning Tool). RESULTS: A history of any childhood trauma was reported by 52.7% of the sample (N = 476). Although participants with a history of any childhood trauma presented with greater symptom severity and functional impairment at most study visits, participants with and without a history of any childhood trauma showed similar rates of improvement in symptom severity and functional impairment over the 24 weeks of treatment. CONCLUSION: This is the first study to explore the association between childhood trauma and treatment outcomes during treatment with lithium or quetiapine in the context of a randomised trial. In Bipolar CHOICE, a history of childhood trauma did not inhibit improvement in symptom severity or functional impairment. Nevertheless, these findings need replication across different settings.

Lithium Treatment in the Prevention of Repeat Suicide-Related Outcomes in Veterans With Major Depression or Bipolar Disorder: A Randomized Clinical Trial.

Importance: Suicide and suicide attempts are persistent and increasing public health problems. Observational studies and meta-analyses of randomized clinical trials have suggested that lithium may prevent suicide in patients with bipolar disorder or depression. Objective: To assess whether lithium augmentation of usual care reduces the rate of repeated episodes of suicide-related events (repeated suicide attempts, interrupted attempts, hospitalizations to prevent suicide, and deaths from suicide) in participants with bipolar disorder or depression who have survived a recent event. Design, Setting, and Participants: This double-blind, placebo-controlled randomized clinical trial assessed lithium vs placebo augmentation of usual care in veterans with bipolar disorder or depression who had survived a recent suicide-related event. Veterans at 29 VA medical centers who had an episode of suicidal behavior or an inpatient admission to prevent suicide within 6 months were screened between July 1, 2015, and March 31, 2019. Interventions: Participants were randomized to receive extended-release lithium carbonate beginning at 600 mg/d or placebo. Main Outcomes and Measures: Time to the first repeated suicide-related event, including suicide attempts, interrupted attempts, hospitalizations specifically to prevent suicide, and deaths from suicide. Results: The trial was stopped for futility after 519 veterans (mean [SD] age, 42.8 [12.4] years; 437 [84.2%] male) were randomized: 255 to lithium and 264 to placebo. Mean lithium concentrations at 3 months were 0.54 mEq/L for patients with bipolar disorder and 0.46 mEq/L for patients with major depressive disorder. No overall difference in repeated suicide-related events between treatments was found (hazard ratio, 1.10; 95% CI, 0.77-1.55). No unanticipated safety concerns were observed. A total of 127 participants (24.5%) had suicide-related outcomes: 65 in the lithium group and 62 in the placebo group. One death occurred in the lithium group and 3 in the placebo group. Conclusions and Relevance: In this randomized clinical trial, the addition of lithium to usual Veterans Affairs mental health care did not reduce the incidence of suicide-related events in veterans with major depression or bipolar disorders who experienced a recent suicide event. Therefore, simply adding lithium to existing medication regimens is unlikely to be effective for preventing a broad range of suicide-related events in patients who are actively being treated for mood disorders and substantial comorbidities. Trial Registration: ClinicalTrials.gov Identifier: NCT01928446.

Cardiometabolic risk markers during mood-stabilizing treatment: Correlation with drug-specific effects, depressive symptoms and treatment response.

BACKGROUND: Patients with bipolar disorder have higher rates of cardiometabolic comorbidities and mortality. Although guidelines emphasize the importance of cardiovascular monitoring, few studies characterized the cardiometabolic risk profile during treatment and their relation to symptomatology and treatment response. METHODS: We analyzed data from two similar 24-weeks comparative effectiveness trials, with a combined sample of 770 participants randomized to two different lithium doses, quetiapine (300 mg/day), or standard treatment without lithium. Glucose, lipids and vital signs were measured before and after 24 weeks of treatment. We calculated several cardiovascular risk scores, assessed baseline correlations and compared the four treatment arms via multiple linear regression models. RESULTS: Higher cholesterol and LDL levels were associated with greater depression severity, showing differential correlations to specific symptoms, particularly agitation, low energy and suicidality. Those randomized to quetiapine showed a significant worsening of cardiometabolic markers during the 24-week trial. Neither baseline nor change in lipid levels correlated with differential treatment response. LIMITATIONS: Study duration was short from the perspective of cardiometabolic risk markers, and all treatment arms included patients taking adjunct antipsychotics. The trials compared quetiapine to lithium, but not to other medications known to affect similar risk factors. CONCLUSIONS: Treatment with 300 mg/day quetiapine for 24 weeks, representing a short and common dose course, resulted in increased cardiometabolic risk markers, emphasizing the importance of monitoring during mood-stabilizing treatment. The symptom-specific associations are in line with previous studies in unipolar depression, suggesting a cardiometabolic-depression link that needs to be further studied in bipolar depression.

Response and remission rates during 24 weeks of mood-stabilizing treatment for bipolar depression depending on early non-response.

BACKGROUND: We aimed to study the probability of bipolar depression response at 24 weeks given initial non-response. METHODS: We combined two multi-site, 24-week trials including similar populations following the same evidence-based guidelines randomizing patients to lithium or quetiapine. Additional mood-stabilizing treatment was possible if clinically indicated. We report cumulative proportions of response (>50% improvement in MADRS) and remission (MADRS<10). RESULTS: We included 592 participants with bipolar depression (mean 39 years, 59% female, mean MADRS 25). Among 393 (66%) participants without response after 2 weeks, 46% responded by 24 weeks; for 291 (49%) without response at 4 weeks, 40% responded and 33% remitted by 24 weeks; for 222 (38%) without a response at 6 weeks, 36% responded and 29% remitted by 24 weeks; for 185 (31%) without a response at 8 weeks, 29% responded and 24% remitted by 24 weeks. Rates were similar for participants who had started an additional mood-stabilizing drug during the first 6 or 8 weeks. CONCLUSIONS: Among patients with bipolar depression and non-response after 6 weeks treatment, representing an adequate bipolar depression trial, only one-third responded by 24 weeks. These results highlight the need for better treatment alternatives for non-responders to evidence-based treatments for bipolar depression.

Investigation of a Capnometry Guided Respiratory Intervention in the Treatment of Posttraumatic Stress Disorder.

Evidence-based treatments for posttraumatic stress disorder (PTSD), including psychotherapies and medications, have high dropout and nonresponse rates, suggesting that more acceptable and effective treatments for PTSD are needed. Capnometry Guided Respiratory Intervention (CGRI) is a digital therapeutic effective in panic disorder that measures and displays end-tidal carbon dioxide (EtCO2) and respiratory rate (RR) in real-time within a structured breathing protocol and may have benefit in PTSD by moderating breathing and EtCO2 levels. We conducted a single-arm study of a CGRI system, Freespira®, to treat symptoms of PTSD. Participants with PTSD (n = 55) were treated for four weeks with twice-daily, 17-min at-home CGRI sessions using a sensor and tablet with pre-loaded software. PTSD and associated symptoms were assessed at baseline, end-of treatment, 2-months and 6-months post-treatment. Primary efficacy outcome was 50% of participants having ≥ 6-point decrease in Clinician Administered PTSD Scale (CAPS-5) score at 2-month follow up. Tolerability, usability, safety, adherence and patient satisfaction were assessed. CGRI was well tolerated, with 88% [95% CI 74-96%] having ≥ 6-point decrease in CAPS-5 scores at 2-months post-treatment follow up. Mean CAPS-5 scores decreased from 49.5 [s.d. = 9.2] at baseline to 27.1 [s.d. = 17.8] at 2-months post-treatment follow up. Respiratory rate decreased and EtCO2 levels increased. Associated mental and physical health symptoms also improved. This CGRI intervention was safe, acceptable, and well-tolerated in improving symptoms in this study in PTSD. Further study against an appropriate comparator is warranted.Trial registration Clinicaltrials.gov NCT#03039231.

Moderators of the association between depressive, manic, and mixed mood symptoms and suicidal ideation and behavior: An analysis of the National Network of Depression Centers Mood Outcomes Program.

BACKGROUND: It has not been established that suicide risk with mixed symptoms is any greater than the depressive component or if there is synergy between depressive and manic symptoms in conveying suicide risk. METHODS: The National Network of Depression Centers Mood Outcomes Program collected data from measurement-based care for 17,179 visits from 6,105 unique individuals with clinically diagnosed mood disorders (998 bipolar disorder, 5,117 major depression). The Patient Health Questionaire-8 (PHQ-8) captured depressive symptoms and the Altman Self-Rating Mania scale (ASRM) measured hypomanic/manic symptoms. Generalized linear mixed models assessed associations between depressive symptoms, manic symptoms, and their interaction (to test for synergistic effects of mixed symptoms) on the primary outcome of suicidal ideation or behavior (secondarily suicidal behavior only) from the Columbia-Suicide Severity Rating Scale (C-SSRS). Moderation was assessed. RESULTS: PHQ-8 scores were strongly associated with suicide-related outcomes across diagnoses. ASRM scores showed no association with suicidal ideation/behavior in bipolar disorder and an inverse association in major depression. There was no evidence of synergy between depressive and manic symptoms. There was no moderation by sex, race, or mood disorder polarity. Those over 55 years of age showed a protective effect of manic symptoms, which was lost when depressive symptoms were also present (mixed symptoms). DISCUSSION: Mixed depressive and manic symptoms convey no excess risk of suicidal ideation or behavior beyond the risk conveyed by the depressive symptoms alone. Depressive symptoms are strongly linked to suicidal ideation and suicidal behavior and represent an important and potentially modifiable risk factor for suicide.

Adjunctive antidepressant treatment among 763 outpatients with bipolar disorder: Findings from the Bipolar CHOICE and LiTMUS trials.

BACKGROUND: Adjunctive antidepressants are frequently used for bipolar depression but their clinical efficacy has been studied in few trials and little is known about how co-occurring manic symptoms affect treatment response. METHODS: Bipolar Clinical Health Outcomes Initiative in Comparative Effectiveness (N = 482) and Lithium Treatment Moderate-Dose Use Study (N = 281) were similar comparative effectiveness trials on outpatients with bipolar disorder comparing four different randomized treatment arms with adjunctive personalized guideline-based treatment for 24 weeks. Adjunctive antidepressant treatment could be used if clinically indicated and was assessed at every study visit. Adjusted mixed effects linear regression analyses compared users of antidepressants to nonusers overall and in different subcohorts. RESULTS: Of the 763 patients, 282 (37.0%) used antidepressant drugs during the study. Antidepressant users had less improvement compared to nonusers on the Clinical Global Impression Scale for Bipolar Disorder and on measures of depression. This was particularly true among patients with co-occurring manic symptoms. Exclusion of individuals begun on antidepressants late in the study (potentially due to overall worse response) resulted in no differences between users and nonusers. We found no differences in treatment effects on mania scales. CONCLUSIONS: In this large cohort of outpatients with bipolar disorder, clinically indicated and guideline-based adjunctive antidepressant treatment was not associated with lower depressive symptoms or higher mania symptoms. The treatment-by-indication confounding due to the nonrandomized design of the trials complicates causal interpretations, but no analyses indicated better treatment effects of adjunctive antidepressants.

Familial severe psychiatric history in bipolar disorder and correlation with disease severity and treatment response.

BACKGROUND: Bipolar disorder is a heritable disorder, and we aimed to assess the impact of family history of mental disorders in first-degree relatives on the severity and course of bipolar disorder. METHODS: The Bipolar CHOICE (lithium versus quetiapine) and LiTMUS (optimized treatment with versus without lithium) comparative effectiveness studies were similar trials among bipolar disorder outpatients studying four different randomized treatment arms for 24 weeks. Patients self-reported on six severe mental disorders among first-degree relatives. We performed ANOVA and linear regression regarding disease severity measures, sociodemographic and cardiometabolic markers and mixed effects linear regression to evaluate treatment response. RESULTS: Among 757 patients, 644 (85.1%) reported at least one first-degree relative with a severe mental disorder (mean=2.8; standard deviation=2.2; range=0-13). Depression (67.1%), alcohol abuse (51.0%) and bipolar disorder (47.0%) were the most frequently reported disorders. Familial psychiatric history correlated with several disease severity measures (hospitalizations, suicide attempts, and earlier onset) and sociodemographic markers (lower education and household income) but not with cardiometabolic markers (e.g. cholesterol or waist circumference) or cardiovascular risk scores, e.g. the Framingham risk score. Patients with familial psychiatric history tended to require more psychopharmacological treatment (p=0.054) but responded similarly (all p>0.1) to all four treatment arms. CONCLUSIONS: Our findings indicate that familial psychiatric history is common among outpatients with bipolar disorder and correlates with disease severity and sociodemographic measures. Patients with a greater familial psychiatric load required more intense treatment but achieved similar treatment responses compared to patients without familial psychiatric history.