Estimating the risk of developing bipolar disorder (BD) in children and adolescents (C&A) with depressive disorders is important to optimize prevention and early intervention efforts. We aimed to quantitatively examine the risk of developing BD from depressive disorders and identify factors which moderate this development.

In this systematic review and meta-analysis (PROSPERO:CRD42023431301), PubMed and Web-of-Science databases were searched for longitudinal studies reporting the percentage of C&A with ICD/DSM-defined depressive disorders who developed BD during follow-up. Data extraction, random-effects meta-analysis, between-study heterogeneity analysis, quality assessment, sub-group analyses, and meta-regressions were conducted.

Thirty-nine studies were included, including 72,371 individuals (mean age=13.9 years, 57.1% females); 14.7% of C&A with a depressive disorder developed BD after 20.4–288 months: 9.5% developed BD-I (95% CI=4.7 to 18.1); 7.7% developed BD-II (95% CI=3.2% to 17.3%); 19.8% (95% CI=9.9% to 35.6%) of C&A admitted into the hospital with a depressive disorder developed BD. Studies using the DSM (21.6%, 95% CI=20.2% to 23.1%) and studies evaluating C&A with a major depressive disorder only (19.8%, 95% CI=16.8% to 23.1%) found higher rates of development of BD. Younger age at baseline, a history of hospitalization and recruitment from specialized clinics were associated with an increased risk of developing BD at follow-up. Quality of included studies was good in 76.9% of studies.

There is a substantial risk of developing BD in C&A with depressive disorders. This is particularly the case for C&A with MDD, DSM-diagnosed depressive disorders, and C&A admitted into the hospital. Research exploring additional predictors and preventive interventions is crucial.

Cardiovascular disease (CVD) is twice as prevalent among individuals with mental illness compared to the general population. Prevention strategies exist but require accurate risk prediction. This study aimed to develop and validate a machine learning model for predicting incident CVD among patients with mental illness using routine clinical data from electronic health records.

A cohort study was conducted using data from 74,880 patients with 1.6 million psychiatric service contacts in the Central Denmark Region from 2013 to 2021. Two machine learning models (XGBoost and regularised logistic regression) were trained on 85% of the data from six hospitals using 234 potential predictors. The best-performing model was externally validated on the remaining 15% of patients from another three hospitals. CVD was defined as myocardial infarction, stroke, or peripheral arterial disease.

The best-performing model (hyperparameter-tuned XGBoost) demonstrated acceptable discrimination, with an area under the receiver operating characteristic curve of 0.84 on the training set and 0.74 on the validation set. It identified high-risk individuals 2.5 years before CVD events. For the psychiatric service contacts in the top 5% of predicted risk, the positive predictive value was 5%, and the negative predictive value was 99%. The model issued at least one positive prediction for 39% of patients who developed CVD.

A machine learning model can accurately predict CVD risk among patients with mental illness using routinely collected electronic health record data. A decision support system building on this approach may aid primary CVD prevention in this high-risk population.

Despite growing awareness of the mental health damage caused by air pollution, the epidemiologic evidence on impact of air pollutants on major mental disorders (MDs) remains limited. We aim to explore the impact of various air pollutants on the risk of major MD.

This prospective study analyzed data from 170 369 participants without depression, anxiety, bipolar disorder, and schizophrenia at baseline. The concentrations of particulate matter with aerodynamic diameter ≤ 2.5 μm (PM2.5), particulate matter with aerodynamic diameter > 2.5 μm, and ≤ 10 μm (PM2.5–10), nitrogen dioxide (NO2), and nitric oxide (NO) were estimated using land-use regression models. The association between air pollutants and incident MD was investigated by Cox proportional hazard model.

During a median follow-up of 10.6 years, 9 004 participants developed MD. Exposure to air pollution in the highest quartile significantly increased the risk of MD compared with the lowest quartile: PM2.5 (hazard ratio [HR]: 1.16, 95% CI: 1.09–1.23), NO2 (HR: 1.12, 95% CI: 1.05–1.19), and NO (HR: 1.10, 95% CI: 1.03–1.17). Subgroup analysis showed that participants with lower income were more likely to experience MD when exposed to air pollution. We also observed joint effects of socioeconomic status or genetic risk with air pollution on the MD risk. For instance, the HR of individuals with the highest genetic risk and highest quartiles of PM2.5 was 1.63 (95% CI: 1.46–1.81) compared to those with the lowest genetic risk and lowest quartiles of PM2.5.

Our findings highlight the importance of air pollution control in alleviating the burden of MD.

Hippocampal disruptions represent potential neuropathological biomarkers in depressed adolescents with cognitive dysfunctions. Given heterogeneous outcomes of whole-hippocampus analyses, we investigated subregional abnormalities in depressed adolescents and their associations with symptom severity and cognitive dysfunctions.

MethodsSeventy-nine first-episode depressive patients (ag = 15.54 ± 1.83) and 71 healthy controls (age = 16.18 ± 2.85) were included. All participants underwent T1 and T2 imaging, completed depressive severity assessments, and performed cognitive assessments on memory, emotional recognition, cognitive control, and attention. Freesurfer was used to segment each hippocampus into 12 subfields. Multivariable analyses of variance were performed to identify overall and disease severity-related abnormalities in patients. LASSO regression was also conducted to explore the associations between hippocampal subfields and patients’ cognitive abilities.

Depressed adolescents showed decreases in dentate gyrus, CA1, CA2/3, CA4, fimbria, tail, and molecular layer. Analyses of overall symptom severity, duration, self-harm behavior, and suicidality suggested that severity-related decreases mainly manifested in CA regions and involved surrounding subfields with disease severity increases. LASSO regression indicated that hippocampal subfield abnormalities had the strongest associations with memory impairments, with CA regions and dentate gyrus showing the highest weights.

Hippocampal abnormalities are widespread in depressed adolescents and such abnormalities may spread from CA regions to surrounding areas as the disease progresses. Abnormalities in CA regions and dentate gyrus among these subfields primarily link with memory impairments in patients. These results demonstrate that hippocampal subsections may serve as useful biomarkers of depression progression in adolescents, offering new directions for early clinical intervention.

Polygenic risk scores for educational attainment (PRSEA), cognitive reserve (CR), and clinical symptoms are associated with functioning in first-episode psychosis (FEP). Nevertheless, the mechanisms underlying their complex interaction are yet to be explored. This study assessed the mediating role of CR and clinical symptoms, both negative (NS) and positive (PS), on the interrelationship between PRSEA and functionality, one year after a FEP.

A total of 162 FEP patients underwent clinical, functional, and genetic assessments. Using genome-wide association study summary results, PRSEA were constructed for each individual. Two mediation models were performed. The parallel mediation model explored the relationship of PRSEA with functionality through CR and clinical symptoms. The serial mediation model tested a causal chain of the three mediators: CR, NS, and PS. Mediation analysis was performed using the PROCESS function V.4.1 in SPSS V.22.

A serial mediation model revealed a causal chain for PRSEA > CR > NS > Functionality (β = −0.35, 95%CI [−0.85, −0.04], p < 0.05). The model fit the data satisfactorily (CFI = 1.00; RMSEA = 0.00; SRMR = 7.2 × 10−7). Conversely, no parallel mediation was found between the three mediators, PRSEA and functionality and the model poorly fit the data (CFI = 0.30; RMSEA = 0.25; SRMR = 0.11).

Both CR and NS mediate the relationship between PRSEA and functionality at one-year follow-up, using serial mediation analysis. This may be relevant for prevention and personalized early intervention to reduce illness impact and improve functional outcomes in FEP patients.