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 interview-based 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 between-site 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.

Characteristics of depressive and bipolar disorder patients with mixed features.

OBJECTIVE: To assess differences between subjects with vs. without mixed features in major affective disorders. METHODS: In 3099 out-patient subjects with DSM-5 major depressive disorder (MDD, n = 1921) or bipolar disorders (BD, n = 1178), we compared those with (Mx) vs. without (Non-Mx) mixed features (agitated-irritable depression or dysphoric [hypo]mania) in an index episode. RESULTS: Prevalence of Mx averaged 21.9% [CI: 20.5-23.4] overall, ranking: BD-II > BD-I > MDD, and in BD depression ≥ [hypo]mania > MDD. Mx subjects were significantly more likely than Non-Mx cases to (i) have other mixed episodes, (ii) have higher irritable and agitated ratings, (iii) have more substance abuse, (iv) switch into mixed episodes, (v) have more suicide attempts and higher suicidal ratings, (vi) change diagnosis from depression to BD, (vii) have higher hypomania scores when depressed or depression scores when [hypo]manic, (viii) be unmarried or separated with fewer children and siblings, (ix) be diagnosed more with BD than MDD, (x) be unemployed, (xi) have BD, suicide and divorce among first-degree relatives, (xii) be female, (xiii) be younger at illness-onset. Both BD and MDD Mx subjects also received antidepressants less, but antipsychotics and mood-stabilizers more, alone and in combination with antidepressants. CONCLUSIONS: Mood disorder subjects with agitated-irritable depression or dysphoric [hypo]mania differed from those without such mixed features, including having a less favorable clinical course and repeated mixed episodes. They may represent a distinct and prevalent, syndromal clinical subtype with prognostic and therapeutic significance.

Suicide attempts in bipolar disorders: comprehensive review of 101 reports.

OBJECTIVE: Assess reported risk of suicide attempts by patients with bipolar disorder (BD). METHOD: Systematic searching yielded 101 reports from 22 countries (79 937 subjects). We analyzed for risk (%) and incidence rates (%/year) of attempts, comparing sex and diagnostic types, including by meta-analysis. RESULTS: Attempt risk averaged 31.1% [CI: 27.9-34.3] of subjects, or 4.24 [3.78-4.70]%/year. In BD-I (43 studies) and BD-II subjects (30 studies), risks (29.9%, 31.4%) and incidence rates (4.01, 4.11%/year) were similar and not different by meta-analysis. Among women vs. men, risks (33.7% vs. 25.5%) and incidence (4.50 vs. 3.21%/year) were greater (also supported by meta-analysis: RR = 1.35 [CI: 1.25-1.45], P < 0.0001). Neither measure was related to reporting year, % women/study, or to onset or current age. Risks were greater with longer exposure, whereas incidence rates decreased with longer time at risk, possibly through 'dilution' by longer exposure. CONCLUSION: This systematic update of international experience underscores high risks of suicide attempts among patients with BD (BD-I = BD-II; women > men). Future studies should routinely include exposure times and incidence rates by diagnostic type and sex for those who attempt suicide or not.

Comparison of psychotic bipolar disorder, schizoaffective disorder, and schizophrenia: an international, multisite study.

OBJECTIVE: Nosological distinctions among schizoaffective disorder (SA), bipolar I disorder with psychotic features (BDp), and schizophrenia (SZ) remain unresolved. METHOD: We compared 2269 subjects with psychotic features in DSM-IV-TR diagnoses (1435 BDp, 463 SZ, 371 SA) from 8 collaborating international sites, by 12 sociodemographic and clinical measures, all between diagnostic pairs. RESULTS: In bivariate comparisons, SA was consistently intermediate between BDp and SZ for 11/12 features (except onset stressors), and SZ vs. BDp differed in all 12 factors. SA differed from both BDp and SZ in 9/12 factors: SA and BDp were similar in education and suicidal ideation or acts; SA and SZ were similar in education, onset stressors, and substance abuse. Meta-analytic comparisons of diagnostic pairs for 10 categorical factors indicated similar differences of SA from both SZ and BDp. Multivariate modeling indicated significantly independent differences between BDp and SZ (8 factors), SA vs. SZ (5), and BDp vs. SA (3). Measurement variance was similar for all diagnoses. CONCLUSION: SA was consistently intermediate between BDp and SZ. The three diagnostic groups ranked: BDp > SA > SZ related to lesser morbidity or disability. The findings are not consistent with a dyadic Kraepelinian categorization, although the considerable overlap among the three DSM-IV diagnostic groups indicates uncertain boundaries if they represent distinct disorders.

First-episode types in bipolar disorder: predictive associations with later illness.

OBJECTIVE: Characteristics of initial illness in bipolar disorder (BD) may predict later morbidity. METHOD: We reviewed computerized clinical records and life charts of DSM-IV-TR BD-I or BD-II patients at affiliated mood-disorder centers to ascertain relationships of initial major illnesses to later morbidity and other clinical characteristics. RESULTS: Adult BD patient-subjects (N=1081; 59.8% BD-I; 58.1% women; 43% ever hospitalized) were followed 15.7±12.8 years after onsets ranking: depression (59%)>mania (13%)>psychosis (8.0%)≥anxiety (7.6%)≥hypomania (6.7%)>mixed states (5.5%). Onset types differed in clinical characteristics and strongly predicted later morbidity. By initial episode types, total time-ill ranked: mania≥hypomania≥mixed-states≥psychosis>depression>anxiety. Depression was most prevalent long-term, overall; its ratio to mania-like illness (D/M, by per cent-time-ill) ranked by onset type: anxiety (4.75)>depression (3.27)>mixed states (1.39)>others (all<1.00). The MDI (mania or hypomania-depression-euthymia interval) course-pattern was most common (34.4%) and associated with psychotic or manic onset; the depression before mania (DMI) pattern (25.0%) most often followed anxiety (38.8%), depression (30.8%), or mixed onsets (13.3%); both were predicted by initial mania depression sequences. CONCLUSION: First-lifetime illnesses and cycles predicted later morbidity patterns among BD patients, indicating value of early morbidity for prognosis and long-term planning.

Clinical responses to antidepressants among 1036 acutely depressed patients with bipolar or unipolar major affective disorders.

OBJECTIVE: Whether responses to antidepressants differ in bipolar and unipolar depression remains unresolved. METHOD: We analyzed patient characteristics and outcomes of antidepressant treatment of 1036 depressed patients with bipolar-I or bipolar-II disorder, or unipolar major depression, using bivariate and multivariate methods and survival analysis, testing the hypothesis that responses would be superior in unipolar depression. RESULTS: Antidepressants were given to 84.8% (878/1036) of depressed patients: 58.9% of 93 bipolar-I, 80.1% of 117 bipolar-II, and 91.3% of 668 unipolar disorder cases. The 158 not given antidepressants had more manias/year, spent more months in mania and depression, and were far more likely to receive mood stabilizers or antipsychotics long term. Improvement of HDRS21 depression ratings ranked: bipolar-II (69.6%) > bipolar-I (62.9%) > unipolar (57.9%; P < 0.0001), independent of initial illness severity. Responder rates (≥50% improved without switching) ranked: bipolar-II (77.0%) > bipolar-I (71.6%) > unipolar (61.7%; P < 0.0001). Remission rates (final HDRS < 7) ranked: 54.0%, 50.6%, and 40.8% respectively (P = 0.02); 67.5% remitted within 12 weeks of treatment. Survival-computed median time to remission (15.0 weeks, overall) was shortest for bipolar-II patients (10.7 weeks). The 3-month risk of switching into mania-hypomania ranked: bipolar-II (15.8%) > bipolar-I (8.60%) > unipolar (0.56%). Multivariate modeling found bipolar diagnosis, shorter latency to remission, more recent trial year, and fewer weeks depressed before treatment to be associated with greater percent improvement of HDRS ratings. CONCLUSION: Selective use of antidepressants with or without mood stabilizers in non-agitated, depressed bipolar disorder patients for short periods was effective with moderate risk of potentially dangerous, manic mood elevation.

Predominant recurrence polarity among 928 adult international bipolar I disorder patients.

OBJECTIVE: To test the hypothesis that patients with bipolar disorder (BPD) differ demographically and clinically within subgroups based on the predominant-polarity of major recurrences. METHOD: We tested factors for association with predominantly (≥2 : 1) depressive vs. mania-like episodes with 928 DSM-IV type-I BPD subjects from five international sites. RESULTS: Factors preliminarily associated with predominant-depression included: electroconvulsive treatment, longer latency-to-BPD diagnosis, first episode depressive or mixed, more suicide attempts, more Axis-II comorbidity, ever having mixed-states, ever married, and female sex. Predominant-mania was associated with: initial manic or psychotic episodes, more drug abuse, more education, and more family psychiatric history. Of the 47.3% of subjects without polarity-predominance, risks for all factors considered were intermediate. Expanding the definition of polarity-predominance to ≥51% added little, but shifting mixed-states to 'predominant-depression' increased risk of suicidal acts from 2.4- to 4.5-fold excess over predominant-mania-hypomania, and suicidal risk was associated continuously with increasing proportions of depressive or mixed episodes. CONCLUSION: Subtyping by predominant-polarity yielded predictive associations, including the polarity of first episodes and risk of suicide attempts. Such subtyping may contribute to improve planning of clinical care and to biological studies of BPD.

Reproduction among 1975 Sardinian women and men diagnosed with major mood disorders.

OBJECTIVE: Disability varies in patients with major affective disorders [type I and II bipolar disorders (BPD) and recurrent unipolar major depressive disorder (UP-MDD)]. It may include reproductive functioning, which has rarely been studied systematically. METHOD: We compared information acquired over several years pertaining to marital/reproductive status among 1975 systematically evaluated, treated, and followed women (n = 1351) and men (n = 624) diagnosed with DSM-IV type I (n = 300) or II BPD (n = 223), or MDD (n = 1452). We compared factors between patients with vs. without children and associated with fertility rate (children/fertile years × 100), using standard bivariate methods followed by multivariate modeling. RESULTS: Childless patients were younger at illness onset, more likely men, diagnosed with type I BPD, more educated, and unmarried, but similar in many aspects of clinical history to those with children. Fertility rate ranked: BP-I < BP-II ≤ MDD, and men < women. Mood-disorder patients had 17% fewer children/person than in the comparable general population of Sardinia. Among mood-disorder patients, fertility appeared to decline in Sardinia in recent decades, more in men than women. CONCLUSION: Type I BPD was associated with lower fertility than BP-II or UP-MDD, consistent with their relatively high levels of other disabilities.

Comparison of antidepressant responses in patients with bipolar vs. unipolar depression: a meta-analytic review.

BACKGROUND: Since there is considerable uncertainty about therapeutic responses to antidepressants among depressed patients diagnosed with bipolar (BP) vs. unipolar (UP) mood disorders, we have reviewed available studies that compared both types of depressed patients. METHODS: Extensive computerized literature-searching identified reports of antidepressant studies involving both BP and UP depressed patients. We used random-effects meta-analysis to compare short-term drug responses by patient type, as well as meta-regression modeling for effects of selected covariates. RESULTS: We identified only 10 studies meeting even liberal inclusion criteria, and they varied greatly in size and design quality. The overall difference in antidepressant responses between BP (n=863) and UP (n=2 226) disorder patients was not significant (pooled RR=1.05; CI: 0.96-1.15; P=0.34). Based on meta-regression, we also found no difference in responses based on diagnosis or subtype, subjects/study, % women, average age, or length of treatment based on meta-regression. Risk of manic-switching averaged 2.50 vs. 0.275%/week among BP vs. UP disorder patients, including co-treatment with mood stabilizers in 70% of BP patients. COMMENTS: The findings suggest little difference in antidepressant responses by diagnostic type, sex, or other factors considered, but a substantial risk of mania and hypomania with BP disorders. However, data pertaining to the fundamental question of antidepressant response among BP vs. UP depressed patients were strikingly limited, and support only tentative conclusions. Additional, well-designed, prospective trials of matched BP and UP depression patients and controlled treatment are required.

Mania associated with antidepressant treatment: comprehensive meta-analytic review.

OBJECTIVE: To review available data pertaining to risk of mania-hypomania among bipolar (BPD) and major depressive disorder (MDD) patients with vs. without exposure to antidepressant drugs (ADs) and consider effects of mood stabilizers. METHOD: Computerized searching yielded 73 reports (109 trials, 114 521 adult patients); 35 were suitable for random effects meta-analysis, and multivariate-regression modeling included all available trials to test for effects of trial design, AD type, and mood-stabilizer use. RESULTS: The overall risk of mania with/without ADs averaged 12.5%/7.5%. The AD-associated mania was more frequent in BPD than MDD patients, but increased more in MDD cases. Tricyclic antidepressants were riskier than serotonin-reuptake inhibitors (SRIs); data for other types of ADs were inconclusive. Mood stabilizers had minor effects probably confounded by their preferential use in mania-prone patients. CONCLUSION: Use of ADs in adults with BPD or MDD was highly prevalent and moderately increased the risk of mania overall, with little protection by mood stabilizers.

Age at onset in 3014 Sardinian bipolar and major depressive disorder patients.

OBJECTIVE: To test if onset age in major affective illnesses is younger in bipolar disorder (BPD) than unipolar-major depressive disorder (UP-MDD), and is a useful measure. METHOD: We evaluated onset-age for DSM-IV-TR major illnesses in 3014 adults (18.5% BP-I, 12.5% BP-II, 69.0% UP-MDD; 64% women) at a mood-disorders center. RESULTS: Median and interquartile range (IQR) onset-age ranked: BP-I = 24 (19-32) < BP-II = 29 (20-40) < UP-MDD = 32 (23-47) years (P < 0.0001), and has remained stable since the 1970s. In BP-I patients, onset was latest for hypomania, and depression presented earlier than in BP-II or UP-MDD cases. Factors associated with younger onset included: i) being unmarried, ii) more education, iii) BPD-diagnosis, iv) family-history, v) being employed, vi) ever-suicidal, vii) substance-abuse and viii) ever-hospitalized. Onset-age distinguished BP-I from UP-MDD depressive onsets with weak sensitivity and specificity. CONCLUSION: Onset age was younger among BPD than MDD patients, and very early onset may distinguish BPD vs. UP-MDD with depressive-onset.

Suicidal Risks in 12 DSM-5 Psychiatric Disorders.

OBJECTIVE: As modern studies evaluating suicidal behaviors in large samples of major psychiatric disorder patients are rare, we compared suicidal risks associated with a variety of psychiatric diagnoses. METHODS: We quantified rates of intake suicidal ideation and lifetime attempts, suicides, and violent acts (attempts + suicides) in 6050 adult patients in a European psychiatric center, diagnosed with 12 prevalent, DSM-5 psychiatric disorders. RESULTS: Ideation ranged from 53.9% of subjects with bipolar disorder (BD) with mixed features, to 8.70% in anxiety disorders. Subjects making at least one suicide attempt were most prevalent in BD with mixed or psychotic features. Suicide rates ranked: substance abuse > BD with psychotic features > psychotic disorders ≥ BD-I > major depressive disorder (MDD). Suicidal intensity (acts/100 PEY) was highest with BD, psychotic disorders, and MDD; lethality (lower attempt/suicide ratio) was greatest with substance abuse, psychotic disorders, and BD with psychotic features. Rates of suicidal acts in BD vs. MDD were similarly high among ever-hospitalized subjects but much lower in never-hospitalized MDD subjects. Women had higher overall risks of ideation and attempts, but violent acts and suicide were more likely among men, whereas SMR for suicide was greater among women, presumably reflecting very low risks among women in the regional general population. CONCLUSIONS: Suicidal risks were particularly high in BD with psychotic or mixed features as well as with substance abuse and in severe MDD with hospitalization.

Melancholic versus Nonmelancholic Major Depression Compared.

BACKGROUND: The concept of melancholia has been associated with psychiatric nosology for centuries. Nevertheless, its definition, relationship to the contemporary concept of Major Depressive Disorder, and clinical implications remain uncertain. METHODS: In a total sample of 3211 closely evaluated patient-subjects diagnosed with DSM-5 Major Depressive or Bipolar Disorder and meeting DSM-5 criteria for major depression with melancholic features or not at a European mood disorder center, we matched 1833 for depression severity (baseline HDRS21 score ≥18) and compared rates and ratings of characteristics of interest between the subgroups, using bivariate and multivariate methods. RESULTS: Observed prevalence of melancholic features was 35.2% in the 1833 subjects matched for severity, and 21.0% among all 3211 subjects. Diagnosis was highly dependent on depression-severity measured three ways. Very few clinical characteristics differed between melancholic and nonmelancholic subjects matched for illness-severity; more suicidal ideation with melancholic features was a notable exception. CONCLUSIONS: Study findings leave the distinction of melancholic features from depression-severity unclear and the potential clinical value of diagnosing melancholic features uncertain.

Suicidal status during antidepressant treatment in 789 Sardinian patients with major affective disorder.

OBJECTIVE: Relationships between antidepressant treatment and suicidality remain uncertain in major depressive disorder (MDD), and rarely evaluated in bipolar disorder (BPD). METHOD: We evaluated changes in suicidality ratings (Hamilton Depression Rating Scale item-3) at the start and after 3.59 +/- 2.57 months of sustained antidepressant treatment in a systematically assessed clinical sample (n = 789) of 605 patients with MDD, 103 patients with BPD-II and 81 patients with BPD-I (based on DSM-IV; 68.1% women; aged 44.3 +/- 16.1 years), comparing suicidal vs. non-suicidal and recovered vs. unrecovered initially suicidal patients. RESULTS: Suicidal patients (103/789, 16.5%; BPD/MDD risk: 2.2) were more depressed and were ill longer. During treatment, 81.5% of suicidal patients became non-suicidal; 0.46% of 656 initially non-suicidal patients reported new suicidal thoughts, with no new attempts. Becoming non-suicidal was associated with greater depression severity and greater improvement. CONCLUSION: Suicidal ideation was prevalent in patients with depressed major affective disorder, but most of the initially suicidal patients became non-suicidal with antidepressant treatment, independent of diagnosis, treatment type or dose.

Suicidal risks among 2826 Sardinian major affective disorder patients.

OBJECTIVE: We estimated risks of suicidal behaviors in 2826 mood-disorder patients evaluated and followed in a Sardinian mood disorders research center over the past 30 years. METHOD: We determined rates of suicidal ideation, attempts, and suicides, with associated risk factors, in men and women with DSM-IV bipolar I (BP-I; n = 529), BP-II; (n = 314), or major depressive disorders (MDD; n = 1983), at risk for an average of 11 years of illness. RESULTS: Observed rates (% of patients/year) of suicide ranked: BP-II (0.16) > or = BP-I (0.14) > MDD (0.05); attempts: BP-I (1.52) > BP-II (0.82) > MDD (0.48); ideation: BP-II (42.7) > MDD (33.8) > BP-I (22.7). The ratio of attempts/suicides (lethality index) ranked: BP-II (5.12) < MDD (9.60) < or = BP-I (10.8). Male/female risk-ratios were greater for suicide than attempts or ideation. One-third of all reported acts occurred within the first year of illness, and earliest among MDD patients. Factors associated independently with suicidal acts included BP diagnosis, hospitalizations/person, and early illness-onset; factors associated with suicidal ideation were having an affective temperament, BP-II diagnosis, and higher suicidality-corrected depression score at intake. CONCLUSION: Suicidal behaviors were more prevalent among BPD than MDD out-patients.

Age at menarche predicts age at onset of major affective and anxiety disorders.

BACKGROUND: Menarche age has been associated inconsistently with the occurrence, timing or severity of major depressive disorder (MDD), but rarely studied in women with bipolar (BDs) or anxiety disorders. METHODS: We investigated women patients at a Sardinian mood disorder center for associations of age at menarche with age at illness onset for major affective or anxiety disorders, year of birth, and other selected factors, using bivariate comparisons and multivariate regression modeling. RESULTS: Among women (n=1139) with DSM-IV MDD (n=557), BD-I (n=223), BD-II (n=178), or anxiety disorders (n=181), born in 1904-1998, of mean age 42.9 years, menarche age averaged 12.8 [CI: 12.7-12.9] years. Illness onset age averaged 30.9 [30.1-31.8] years, ranking: BD-I, 25.8; anxiety disorders, 28.0; BD-II, 30.3; MDD, 34.1 years. Menarche age declined secularly over birth years, and was associated with younger illness-onset, having no or fewer siblings, more psychiatrically ill first-degree relatives, living in rural environments, being suicidal, substance abuse, and being unemployed. Earlier menarche and earlier illness-onset were significantly associated for onset age groups of ≤ 20, 20-39, and > 40 years. Menarche age versus diagnosis ranked: BD-II

Early clinical predictors and correlates of long-term morbidity in bipolar disorder.

OBJECTIVES: Identifying factors predictive of long-term morbidity should improve clinical planning limiting disability and mortality associated with bipolar disorder (BD). METHODS: We analyzed factors associated with total, depressive and mania-related long-term morbidity and their ratio D/M, as %-time ill between a first-lifetime major affective episode and last follow-up of 207 BD subjects. Bivariate comparisons were followed by multivariable linear regression modeling. RESULTS: Total % of months ill during follow-up was greater in 96 BD-II (40.2%) than 111 BD-I subjects (28.4%; P=0.001). Time in depression averaged 26.1% in BD-II and 14.3% in BD-I, whereas mania-related morbidity was similar in both, averaging 13.9%. Their ratio D/M was 3.7-fold greater in BD-II than BD-I (5.74 vs. 1.96; P<0.0001). Predictive factors independently associated with total %-time ill were: [a] BD-II diagnosis, [b] longer prodrome from antecedents to first affective episode, and [c] any psychiatric comorbidity. Associated with %-time depressed were: [a] BD-II diagnosis, [b] any antecedent psychiatric syndrome, [c] psychiatric comorbidity, and [d] agitated/psychotic depressive first affective episode. Associated with %-time in mania-like illness were: [a] fewer years ill and [b] (hypo)manic first affective episode. The long-term D/M morbidity ratio was associated with: [a] anxious temperament, [b] depressive first episode, and [c] BD-II diagnosis. CONCLUSIONS: Long-term depressive greatly exceeded mania-like morbidity in BD patients. BD-II subjects spent 42% more time ill overall, with a 3.7-times greater D/M morbidity ratio, than BD-I. More time depressed was predicted by agitated/psychotic initial depressive episodes, psychiatric comorbidity, and BD-II diagnosis. Longer prodrome and any antecedent psychiatric syndrome were respectively associated with total and depressive morbidity.

Comorbid substance-use in schizophrenia: relation to positive and negative symptoms.

As substance use disorders (SUD) are common in schizophrenia patients, we tested the hypothesis that comorbid patients (SUD[+]) have more positive vs. negative symptoms than non-comorbid (SUD[-]) patients. From reports identified by literature-searching we compared Positive and Negative Syndrome Scale (PANSS) ratings in schizophrenia patients with and without SUD using meta-analytic methods. Among 9 comparisons (N=725 subjects), SUD[+] patients were more often men, and abused alcohol>cannabis>cocaine. SUD[+] patients had very significantly higher PANSS-positive, and lower PANSS-negative scores. Comorbid SUD in schizophrenia patients was associated with male sex and higher PANSS positive to lower negative scores. Cause-effect relationships remain to be clarified.