BackgroundRates and risk factors for suicidal behaviour require updating and comparisons among mood disorders.AimsTo identify factors associated with suicidal risk in major mood disorders.MethodWe considered risk factors before, during and after intake assessments of 3284 adults with/without suicidal acts, overall and with bipolar disorder (BD) versus major depressive disorder (MDD), using bivariate comparisons, multivariable regression modelling and receiver operating characteristic (ROC) analysis.ResultsSuicidal prevalence was greater in BD versus MDD: ideation, 29.2 versus 17.3%; attempts, 18.8 versus 4.78%; suicide, 1.73 versus 0.48%; attempts/suicide ratio indicated similar lethality, 10.9 versus 9.96. Suicidal acts were associated with familial BD or suicide, being divorced/unmarried, fewer children, early abuse/trauma, unemployment, younger onset, longer illness, more dysthymic or cyclothymic temperament, attention-deficit hyperactivity disorder (ADHD), substance misuse, mixed features, hospital admission, percentage time unwell, less antidepressants and more antipsychotics and mood stabilisers. Logistic regression found five independent factors: hospital admission, more depression at intake, BD diagnosis, onset age ≤25 years and mixed features. These factors were more associated with suicidal acts in BD than MDD: percentage time depressed/ill, alcohol misuse, >4 pre-intake depressions, more dysthymic/cyclothymic temperament and prior abuse/trauma. ADHD and total years ill were similar in BD and MDD; other factors were more associated with MDD. By ROC analysis, area under the curve was 71.3%, with optimal sensitivity (76%) and specificity (55%) with any two factors.ConclusionsSuicidal risks were high in mood disorders: ideation was highest with BD type II, attempts and suicides (especially violent) with BD type I. Several risk factors for suicidal acts differed between BD versus MDD patients.Declaration of interestNo author or immediate family member has financial relationships with commercial entities that might appear to represent potential conflicts of interest with the information presented.

BackgroundConcerns about potential adverse effects of long-term exposure to lithium as a mood-stabilizing treatment notably include altered renal function. However, the incidence of severe renal dysfunction; rate of decline over time; effects of lithium dose, serum concentration, and duration of treatment; relative effects of lithium exposure vs. aging; and contributions of sex and other factors all remain unclear.MethodsAccordingly, we acquired data from 12 collaborating international sites and 312 bipolar disorder patients (6142 person-years, 2669 assays) treated with lithium carbonate for 8–48 (mean 18) years and aged 20–89 (mean 56) years. We evaluated changes of estimated glomerular filtration rate (eGFR) as well as serum creatinine, urea–nitrogen, and glucose concentrations, white blood cell count, and body-mass index, and tested associations of eGFR with selected factors, using standard bivariate contrasts and regression modeling.ResultsOverall, 29.5% of subjects experienced at least one low value of eGFR (<60 mL/min/1.73 m2), most after ≥15 years of treatment and age > 55; risk of ≥2 low values was 18.1%; none experienced end-stage renal failure. eGFR declined by 0.71%/year of age and 0.92%/year of treatment, both by 19% more among women than men. Mean serum creatinine increased from 0.87 to 1.17 mg/dL, BUN from 23.7 to 33.1 mg/dL, glucose from 88 to 122 mg/dL, and BMI from 25.9 to 26.6 kg/m2. By multivariate regression, risk factors for declining eGFR ranked: longer lithium treatment, lower lithium dose, higher serum lithium concentration, older age, and medical comorbidity. Later low eGFR was also predicted by lower initial eGFR, and starting lithium at age ≥ 40 years.LimitationsControl data for age-matched subjects not exposed to lithium were lacking.ConclusionsLong-term lithium treatment was associated with gradual decline of renal functioning (eGFR) by about 30% more than that was associated with aging alone. Risk of subnormal eGFR was from 18.1% (≥2 low values) to 29.5% (≥1 low value), requiring about 30 years of exposure. Additional risk factors for low eGFR were higher serum lithium level, longer lithium treatment, lower initial eGFR, and medical comorbidity, as well as older age.

Objective To compare characteristics of bipolar disorder patients diagnosed as DSM-5 types I (BD-1) vs. II (BD-2). Methods We compared descriptive, psychopathological, and treatment characteristics in a sample of 1377 consenting, closely and repeatedly evaluated adult BD patient-subjects from a specialty clinic, using bivariate methods and logistic multivariable modeling. Results Factors found more among BD-2 > BD-1 cases included: [a] descriptors (more familial affective disorder, older at onset, diagnosis and first-treatment, more education, employment and higher socioeconomic status, more marriage and children, and less obesity); [b] morbidity (more general medical diagnoses, less drug abuse and smoking, more initial depression and less [hypo]mania or psychosis, longer episodes, higher intake depression and anxiety ratings, less mood-switching with antidepressants, less seasonal mood-change, greater %-time depressed and less [hypo]manic, fewer hospitalizations, more depression-predominant polarity, DMI > MDI course-pattern, and less violent suicidal behavior); [c] specific item-scores with initial HDRS21 (higher scores for depression, guilt, suicidality, insomnia, anxiety, agitation, gastrointestinal symptoms, hypochondriasis and weight-loss, with less psychomotor retardation, depersonalization, or paranoia); and [d] treatment (less use of lithium or antipsychotics, more antidepressant and benzodiazepine treatment). Conclusions BD-2 was characterized by more prominent and longer depressions with some hypomania and mixed-features but not mania and rarely psychosis. BD-2 subjects had higher socioeconomic and functional status but also high levels of long-term morbidity and suicidal risk. Accordingly, BD-2 is dissimilar to, but not necessarily less severe than BD-1, consistent with being distinct syndromes.

Prediction of lithium response using clinical data.Objective: Promptly establishing maintenance therapy could reduce morbidity and mortality in patients with bipolar disorder. Using a machine learning approach, we sought to evaluate whether lithium responsiveness (LR) is predictable using clinical markers. Method: Our data are the largest existing sample of direct interviewbased clinical data from lithium-treated patients (n = 1266, 34.7% responders), collected across seven sites, internationally. We trained a random forest model to classify LR-as defined by the previously validated Alda scale-against 180 clinical predictors. Results: Under appropriate cross-validation procedures, LR was predictable in the pooled sample with an area under the receiver operating characteristic curve of 0.80 (95% CI 0.78-0.82) and a Cohen kappa of 0.46 (0.4-0.51). The model demonstrated a particularly low false-positive rate (specificity 0.91 [0.88-0.92]). Features related to clinical course and the absence of rapid cycling appeared consistently informative. Conclusion: Clinical data can inform out-of-sample LR prediction to a potentially clinically relevant degree. Despite the relevance of clinical course and the absence of rapid cycling, there was substantial betweensite heterogeneity with respect to feature importance. Future work must focus on improving classification of true positives, better characterizing between-and within-site heterogeneity, and further testing such models on new external datasets.• Lithium response can be predicted accurately (area under the receiver operating characteristic curve of 0.80) in previously unobserved subjects based on data collected by careful clinical interview.• Variables related to clinical course of illness and absence of rapid cycling are particularly informative. Limitations• Our data were retrospective in nature, although data of such as scope would be difficult to collect prospectively.