Supervised machine learning to predict reduced depression severity in people with epilepsy through epilepsy self-management intervention.

OBJECTIVE: To develop a classifier that predicts reductions in depression severity in people with epilepsy after participation in an epilepsy self-management intervention. METHODS: Ninety-three people with epilepsy from three epilepsy self-management randomized controlled trials from the Managing Epilepsy Well (MWE) Network integrated research database met the inclusion criteria. Supervised machine learning algorithms were utilized to develop prediction models for changes in self-reported depression symptom severity. Features considered by the machine learning classifiers include age, gender, race, ethnicity, education, study type, baseline quality of life, and baseline depression symptom severity. The models were trained and evaluated on their ability to predict clinically meaningful improvement (i.e., a reduction of greater than three points on the nine-item Patient Health Questionnaire (PHQ-9)) between baseline and follow-up (<=12 weeks) depression scores. Models tested were a Multilayer Perceptron (ML), Random Forest (RF), Support Vector Machine (SVM), Logistic Regression with Stochastic Gradient Descent (SGD), K-nearest Neighbors (KNN), and Gradient Boosting (GB). A separate, outside dataset of 41 people with epilepsy was used in a validation exercise to examine the top-performing model's generalizability and performance with external data. RESULTS: All six classifiers performed better than our baseline mode classifier. Support Vector Machine had the best overall performance (average area under the curve [AUC] = 0.754, highest subpopulation AUC = 0.963). Our analysis of the SVM features revealed that higher baseline depression symptom severity, study type (i.e., intervention program goals), higher baseline quality of life, and race had the strongest influence on increasing the likelihood that a subject would experience a clinically meaningful improvement in depression scores. From the validation exercise, our top-performing SVM model performed similarly or better than the average SVM model with the outside dataset (average AUC = 0.887). SIGNIFICANCE: We trained an SVM classifier that offers novel insight into subject-specific features that are important for predicting a clinically meaningful improvement in subjective depression scores after enrollment in a self-management program. We provide evidence for machine learning to select subjects that may benefit most from a self-management program and indicate important factors that self-management programs should collect to develop improved digital tools.

Depression and suicidality among Hispanics with epilepsy: Findings from the Managing Epilepsy Well (MEW) Network integrated database.

OBJECTIVE: Although psychiatric disorders are more common among people with epilepsy,2 depression and suicidal ideation among Hispanics with epilepsy remain understudied. We examined the prevalence and correlates of depression and suicidal ideation among Hispanic adults with epilepsy who participated in self-management studies in the Managing Epilepsy Well3 Network. METHODS: This cross-sectional analysis of pooled data from ten studies used the Patient Health Questionnaire-94 or Neurological Disease Depression Inventory-Epilepsy5 to examine the prevalence of elevated depressive symptoms (PHQ ≥ 10, NDDI-E ≥ 15) and suicidal ideation (PHQ-9 item 9 ≥ 1, NDDI-E item 4 ≥ 2). Multilevel mixed-effects logistic regression models examined associations between ethnicity, elevated depressive symptoms, and suicidal ideation among PWE. Secondary analyses examined correlates of elevated depressive symptoms and suicidal ideation among Hispanic PWE. RESULTS: Of 559 participants, 49.6% (n = 277) were Hispanic. Elevated depressive symptoms were endorsed by 38.1% (n = 213) of all participants (32.5% of Hispanics); suicidal ideation was endorsed by 18.4% (n = 103) of all participants (16.3% of Hispanics). After adjustment for sociodemographic and health attributes, Hispanic PWE had a 44% lower prevalence of elevated depressive symptoms (OR = 0.56, CI 0.37-0.84, p = 0.0056) compared to non-Hispanics but similar rates of suicidal ideation (OR = 0.84, CI 0.45-1.58, p = 0.59). Acculturation measures were available for 256 (92.4%) of Hispanic PWE: language preference was Spanish for 62.9%, 46.1% were foreign-born. Spanish-speaking Hispanics were less likely than English-speaking Hispanics to report elevated depressive symptoms (OR = 0.43, CI 0.19-0.97, p = 0.041); however, Hispanics who reported fair or poor health status had a four-fold higher depression prevalence compared to those who reported excellent or very good health status [reference group] (OR = 4.44, CI 1.50-13.18, p = 0.0071). Of the Hispanics who provided prior 30-day seizure data, ≥1 monthly seizure was independently associated with higher depression prevalence (OR = 3.11, CI 1.29-7.45, p = 0.01). Being foreign-born was not associated with elevated depressive symptoms or suicidal ideation prevalence. CONCLUSIONS: In a large, geographically diverse sample of PWE, elevated depressive symptoms were significantly lower in Hispanics compared to non-Hispanics. Spanish language preference was associated with a lower prevalence of elevated depressive symptoms among Hispanic PWE. Future studies should include acculturation data to better screen for depression and suicidal ideation risk and optimize interventions for Hispanic PWE.

Manualized cognitive behavioral group therapy to treat vasomotor symptoms for women diagnosed with mood disorders.

OBJECTIVE: This 6-week, prospective, single-arm study examined the feasibility, acceptability, and preliminary efficacy of cognitive behavioral group therapy in peri- and postmenopausal women with mood disorders (major depression or bipolar) and problematic vasomotor menopausal symptoms. METHODS: 59 participants from an outpatient clinic with mood disorders and problematic vasomotor symptoms were enrolled. The primary outcomes were change from baseline to 6 weeks in Hot Flush Night Sweat Problem Rating, Hot Flash Related Daily Interference, and Quality of Life. Secondary outcomes were change in Hot Flush Frequency, depression, anxiety, perceived stress, anhedonia, beliefs and cognitive appraisals of menopause. ClinicalTrials.gov [identifier: NCT02860910]. RESULTS: On the Hot Flush Night Sweat Problem Rating, 39.3% improved 2 or more points, which was clinically relevant. Changes in Quality of Life (p = .001) and the Hot Flash Related Daily Interference Scale were also significant (p < .001). Significant results were found on most secondary outcomes (hot flush frequency on the Hot Flush Daily Diary, depression, anxiety, perceived stress (p < .001) and anhedonia (p = .001). One of six subscales (control subscale) on the cognitive appraisal of menopause significantly improved (p < .001). Three subscales on the beliefs measure did not change significantly (p = .05, p = .91, and p = .14). Six-week study retention was robust (N = 55, 93%) and 94.2% of individuals reported that cognitive behavioral group therapy sessions were useful. CONCLUSION: This exploratory study suggests that CBGT is acceptable, feasible, and efficacious in women with mood disorders and problematic menopause vasomotor symptoms. Further studies are needed using more rigorous and controlled methods.

Longitudinally Measured Changes in Somnolence Severity With a Visual Analog Scale in a Randomized Lithium Versus Quetiapine-IR Study in Bipolar Disorder.

OBJECTIVE: The aim of this study was to use a visual analog scale (VAS) longitudinally measuring somnolence severity in patients with bipolar disorder. METHODS: A data set of patients with bipolar spectrum disorders who were randomized to lithium or quetiapine-IR for 16 weeks was used. The somnolence severity was measured with a VAS from 0 to 100 (VAS based), and somnolence frequency was recorded according to incident report (incidence based) at each visit. The rates of VAS-based and incidence-based somnolence and changes in somnolence severity from baseline to the end of study were compared between the lithium and quetiapine groups. Longitudinal changes in somnolence severity were analyzed with linear regression analysis. RESULTS: Of 42 patients randomized, only 3 scored 0 on the VAS at baseline. The rates of incidence-based and VAS-based somnolence were similar in the lithium and quetiapine-IR groups. The VAS change scores from baseline to each visit varied in both groups with significant decreases at weeks 6 and 12 in the quetiapine-IR group only. The decrease at week 6 in the quetiapine-IR group was significantly different from that in the lithium group. Patterns of changes in somnolence severity were inconsistent in both groups. A significant interaction between time course and the decrease in VAS scores was observed in the quetiapine-IR group, but not in the lithium group. CONCLUSIONS: Baseline somnolence was highly prevalent in patients with bipolar disorder. The change in somnolence severity was different between lithium-treated and quetiapine-treated patients. Quantifying somnolence longitudinally is important in clinical trials and practice.

Double-blind, placebo-controlled trial of pioglitazone for bipolar depression.

BACKGROUND: Objective of the present study was to conduct an 8-week double-blind, randomized, placebo-controlled trial to test the efficacy of pioglitazone in the treatment of bipolar depression. METHODS: 38 outpatients with bipolar disorder and current major depressive episode were randomized to pioglitazone (15-45 mg/day) or placebo. The use of concomitant mood stabilizers, antipsychotics, and antidepressants was permitted. The primary outcome measure was the 30-item Inventory of Depressive Symptomatology, Clinician Rated (IDS-C30) total score change from baseline to endpoint. Laboratory evaluations, including serum level of inflammatory and metabolic biomarkers, were conducted. RESULTS: 37 subjects were analyzed for efficacy (1 subject had no follow-up data). Mean reduction from baseline to week 8 in IDS-C30 score was-6.59 for pioglitazone and -11.63 for placebo. Mixed effects modeling indicated borderline statistically significant difference between the two groups (p = 0.056) in favor of placebo. On analysis of inflammatory and metabolic markers, a statistically significant negative correlation was noted between change in leptin levels and change in depression scores in the pioglitazone group (r = -0.61, p = 0.047) but not in the placebo group, the significance of which is unclear as the study failed to demonstrate antidepressant efficacy of pioglitazone over placebo. No serious adverse effects were reported, and pioglitazone was well-tolerated. LIMITATIONS: small sample size with inadequate power, concomitant use of other psychotropic medications, and lack of statistical adjustment for multiple testing. CONCLUSION: Current study does not support the antidepressant efficacy of pioglitazone in the treatment of bipolar depression. (240 words).

A Pilot Study of the Effectiveness of Lithium Versus Quetiapine Immediate Release Monotherapy in Patients With Bipolar Spectrum Disorders.

OBJECTIVE: The aim of this study was to compare the effectiveness of lithium versus quetiapine immediate release (IR) monotherapy in patients with bipolar I, II, or subthreshold bipolar disorder at any phase. METHODS: Eligible patients were randomized to lithium or quetiapine IR for 16 weeks. The difference in the time to discontinuation from study due to "all causes" between lithium and quetiapine IR groups and changes from baseline to 8 and 16 weeks in depression, mania, anxiety, quality of life (QOL), metabolic profiles, and proinflammatory markers were compared. RESULTS: Of the 42 patients randomized to lithium (n = 18) and quetiapine IR (n = 24), the median time to discontinuation due to "all causes" was 6 weeks (95% confidence interval, 2-12 weeks) in the lithium group and 8 weeks (95% confidence interval, 6 weeks to not calculable) in the quetiapine IR group. The mean time to discontinuation due to "all causes" was 7.7 ± 1.1 weeks for lithium versus 8.4 ± 0.8 weeks for quetiapine IR (P = 0.54). There was no significant difference between lithium and quetiapine IR in changes in the severity of depression, mania/hypomania, anxiety, and QOL as a whole or only in patients with depressive index episode. The decrease in total cholesterol was significantly larger with lithium than with quetiapine IR (P = 0.05) as a whole, but not only in patients with depression index episode. There was no other significant difference in changes in metabolic panels and inflammatory markers between the 2 groups. CONCLUSIONS: The difference in effectiveness between lithium and quetiapine IR monotherapy in a real-world bipolar population was minimal. Large-sample studies are needed to support or refute this finding.

A placebo controlled study of quetiapine-XR in bipolar depression accompanied by generalized anxiety with and without a recent history of alcohol and cannabis use.

OBJECTIVE: This study aims to compare treatment response in bipolar I or II depression and generalized anxiety disorder (GAD) with and without recent alcohol and/or cannabis use disorder (ALC/CAN) to quetiapine-XR (extended release) or placebo. METHODS: A randomized, double-blind, 8-week study of quetiapine-XR versus placebo in patients with bipolar I or II depression and GAD with or without a recent ALC/CAN was used to compare changes in Hamilton Depression Rating Scale-17, Hamilton Anxiety Rating Scale, the 16-item Quick Inventory of Depressive Symptomatology-Self-Report (QIDS-SR-16), Clinical Global Impression for Bipolar Disorder-Severity (CGI-BP-S), and Timeline Follow Back within and between groups. RESULTS: In the quetiapine-XR group, patients with a recent ALC/CAN (n = 22) had significant decreases in QIDS-SR-16 (-9.6 ± 1.6 vs. -3.7 ± 1.7) and CGI-BP-S (-1.6 ± 0.4 vs. -0.8 ± 0.03) than those without a recent ALC/CAN (n = 24). In the placebo group, both patients with a recent ALC/CAN (n = 23) and those without (n = 21) had similar reductions in these measures. The reduction of QIDS-SR-16 scores in patients with a recent ALC/CAN was also significantly different from that of their counterparts in the placebo group. Patients who received quetiapine-XR had larger decreases in the number of drinking days/week (p = 0.17) and number of cannabis joints/week (p = 0.09) compared to those who received placebo. CONCLUSION: Quetiapine-XR was superior to placebo in reducing QIDS-SR-16 total score in patients with a recent ALC/CAN. Patients taking quetiapine-XR used less alcohol and cannabis than patients on placebo, suggesting that quetiapine-XR may be of use in patients with bipolar disorder accompanied by GAD and other comorbidities.