Advertisement

Trait-Related Decision-Making Impairment in the Three Phases of Bipolar Disorder

  • Marc Adida
    Correspondence
    Address correspondence to Marc Adida, M.D., Ph.D., Sainte-Marguerite Hospital, Department of Psychiatry, Mediterranean University, 270 Bd de Sainte-Marguerite, 13009 Marseille, France
    Affiliations
    Mediterranean Institute of Cognitive Neurosciences, Department of Pharmacology and Neuropsychology of Emotions Related to Risk Taking and Reward, National Research Scientific Centre, Sainte-Marguerite Hospital, Mediterranean University, Marseille, France

    Department of Psychiatry, Sainte-Marguerite Hospital, Mediterranean University, Marseille, France
    Search for articles by this author
  • Fabrice Jollant
    Affiliations
    Mental Health University Institute Douglas and McGill Group for Suicide Studies, McGill University, Montréal, Canada
    Search for articles by this author
  • Luke Clark
    Affiliations
    Department of Experimental Psychology, University of Cambridge, Cambridge, United Kingdom
    Search for articles by this author
  • Nathalie Besnier
    Affiliations
    Mediterranean Institute of Cognitive Neurosciences, Department of Pharmacology and Neuropsychology of Emotions Related to Risk Taking and Reward, National Research Scientific Centre, Sainte-Marguerite Hospital, Mediterranean University, Marseille, France

    Department of Psychiatry, Sainte-Marguerite Hospital, Mediterranean University, Marseille, France
    Search for articles by this author
  • Sébastien Guillaume
    Affiliations
    University Montpellier I, Montpellier, France

    Emergency Department of Academic Hospital Montpellier, Montpellier, France

    National Institute of Health and Medical Research U888, Montpellier, France
    Search for articles by this author
  • Arthur Kaladjian
    Affiliations
    Centre Hospitalier Régional Universitaire Hôpital R. Debré, Reims, France
    Search for articles by this author
  • Pascale Mazzola-Pomietto
    Affiliations
    Mediterranean Institute of Cognitive Neurosciences, Department of Pharmacology and Neuropsychology of Emotions Related to Risk Taking and Reward, National Research Scientific Centre, Sainte-Marguerite Hospital, Mediterranean University, Marseille, France
    Search for articles by this author
  • Régine Jeanningros
    Affiliations
    Mediterranean Institute of Cognitive Neurosciences, Department of Pharmacology and Neuropsychology of Emotions Related to Risk Taking and Reward, National Research Scientific Centre, Sainte-Marguerite Hospital, Mediterranean University, Marseille, France
    Search for articles by this author
  • Guy M. Goodwin
    Affiliations
    University Department of Psychiatry, Warneford Hospital, Oxford, United Kingdom
    Search for articles by this author
  • Jean-Michel Azorin
    Affiliations
    Mediterranean Institute of Cognitive Neurosciences, Department of Pharmacology and Neuropsychology of Emotions Related to Risk Taking and Reward, National Research Scientific Centre, Sainte-Marguerite Hospital, Mediterranean University, Marseille, France

    Department of Psychiatry, Sainte-Marguerite Hospital, Mediterranean University, Marseille, France
    Search for articles by this author
  • Philippe Courtet
    Affiliations
    University Montpellier I, Montpellier, France

    Emergency Department of Academic Hospital Montpellier, Montpellier, France

    National Institute of Health and Medical Research U888, Montpellier, France
    Search for articles by this author

      Background

      In bipolar disorder (BD), little is known about how deficits in neurocognitive functions such as decision-making are related to phase of illness. We predicted that manic, depressed, and euthymic bipolar patients (BPs) would display impaired decision-making, and we tested whether clinical characteristics could predict patients' decision-making performance.

      Methods

      Subjects (N = 317; age range: 18–65 years) including 167 BPs (45 manic and 32 depressed inpatients, and 90 euthymic outpatients) and 150 age-, IQ-, and gender-matched healthy control (HC) participants, were included within three university psychiatric hospitals using a cross-sectional design. The relationship between predictor variables and decision-making was assessed by one-step multivariate analysis. The main outcome measures were overall decision-making ability on the Iowa Gambling Task (IGT) and an index of sensitivity to punishment frequency.

      Results

      Manic, depressed, and euthymic BPs selected significantly more cards from the risky decks than HCs (p < .001, p < .01, and p < .05, respectively), with no significant differences between the three BD groups. However, like HCs, BPs preferred decks that yielded infrequent penalties over those yielding frequent penalties. In multivariate analysis, decision-making impairment was significantly (p < .001) predicted by low level of education, high depressive scores, family history of BD, use of benzodiazepines, and nonuse of serotonin and norepinephrine reuptake inhibitor (SNRI) antidepressants.

      Conclusions

      BPs have a trait-related impairment in decision-making that does not vary across illness phase. However, some subtle differences between the BD groups in the individual deck analyses may point to subtle state influences on reinforcement mechanisms, in addition to a more fundamental trait impairment in risk-sensitive decision making.

      Key Words

      It is now accepted that bipolar disorder (BD) is associated with substantial alterations in neuropsychologic function. Whereas early studies focused on attentional, mnemonic, and executive domains, recent studies have highlighted the link between simple tests of risky decision-making and the manic phase of the illness (
      • Adida M.
      • Clark L.
      • Pomietto P.
      • Kaladjian A.
      • Besnier N.
      • Azorin J.M.
      • et al.
      Lack of insight may predict impaired decision making in manic patients.
      ,
      • Minassian A.
      • Paulus M.P.
      • Perry W.
      Increased sensitivity to error during decision-making in bipolar disorder patients with acute mania.
      ). Although trait-related cognitive impairments have been reported in BD patients (BPs) (
      • Wingo A.P.
      • Harvey P.D.
      • Baldessarini R.J.
      Neurocognitive impairment in bipolar disorder patients: Functional implications.
      ,
      • Torres I.J.
      • Boudreau V.G.
      • Yatham L.N.
      Neuropsychological functioning in euthymic bipolar disorder: A meta-analysis.
      ), the nature and extent of decision-making dysfunction across the phases of the illness remain unclear. Some studies have shown that patients have impaired decision-making in both the manic (
      • Adida M.
      • Clark L.
      • Pomietto P.
      • Kaladjian A.
      • Besnier N.
      • Azorin J.M.
      • et al.
      Lack of insight may predict impaired decision making in manic patients.
      ,
      • Minassian A.
      • Paulus M.P.
      • Perry W.
      Increased sensitivity to error during decision-making in bipolar disorder patients with acute mania.
      ,
      • Clark L.
      • Iversen S.D.
      • Goodwin G.M.
      A neuropsychological investigation of prefrontal cortex involvement in acute mania.
      ,
      • Murphy F.C.
      • Rubinsztein J.S.
      • Michael A.
      • Rogers R.D.
      • Robbins T.W.
      • Paykel E.S.
      • Sahakian B.J.
      Decision-making cognition in mania and depression.
      ,
      • Rubinsztein J.S.
      • Fletcher P.C.
      • Rogers R.D.
      • Ho L.W.
      • Aigbirhio F.I.
      • Paykel E.S.
      • et al.
      Decision-making in mania: A PET study.
      ) and depressed (
      • Rubinsztein J.S.
      • Michael A.
      • Underwood B.R.
      • Tempest M.
      • Sahakian B.J.
      Impaired cognition and decision-making in bipolar depression but no “affective bias” evident.
      ) states, whereas others have reported conflicting results in patients in remission (
      • Jollant F.
      • Guillaume S.
      • Jaussent I.
      • Bellivier F.
      • Leboyer M.
      • Castelnau D.
      • et al.
      Psychiatric diagnoses and personality traits associated with disadvantageous decision-making.
      ,
      • Christodoulou T.
      • Lewis M.
      • Ploubidis G.B.
      • Frangou S.
      The relationship of impulsivity to response inhibition and decision-making in remitted patients with bipolar disorder.
      ,
      • Rubinsztein J.S.
      • Michael A.
      • Paykel E.S.
      • Sahakian B.J.
      Cognitive impairment in remission in bipolar affective disorder.
      ,
      • Clark L.
      • Iversen S.D.
      • Goodwin G.M.
      Sustained attention deficit in bipolar disorder.
      ,
      • Yechiam E.
      • Hayden E.P.
      • Bodkins M.
      • O'Donnell B.F.
      • Hetrick W.P.
      Decision making in bipolar disorder: A cognitive modeling approach.
      ). It is also likely that other illness variables, such as number of episodes, severity of acute symptoms, type of medication, and family history of BD have an impact on decision-making cognition. To our knowledge, Yechiam et al. (
      • Yechiam E.
      • Hayden E.P.
      • Bodkins M.
      • O'Donnell B.F.
      • Hetrick W.P.
      Decision making in bipolar disorder: A cognitive modeling approach.
      ) are the only group that has used the same task to assess decision-making in both the acute and remitted state of BD. However, their study was limited by small group sizes and lack of power.
      Decision-making occurs when the individual has to select between multiple options associated with uncertain consequences. Laboratory tasks have been devised to assess competency in real-world decision-making and dissect some of cognitive processes involved. This study used the Iowa Gambling Task (IGT), a clinically sensitive tool that emulates real-world financial decision-making. Each choice leads to monetary gains or losses. Differences in IGT performance are seen in individuals with neuropsychiatric disorders characterized by problems in impulse control and emotional regulation (
      • Jollant F.
      • Bellivier F.
      • Leboyer M.
      • Astruc B.
      • Torres S.
      • Verdier R.
      • et al.
      Impaired decision making in suicide attempters.
      ,
      • Bar-On R.
      • Tranel D.
      • Denburg N.L.
      • Bechara A.
      Exploring the neurological substrate of emotional and social intelligence.
      ,
      • Bolla K.I.
      • Eldreth D.A.
      • Matochik J.A.
      • Cadet J.L.
      Neural substrates of faulty decision-making in abstinent marijuana users.
      ,
      • Ernst M.
      • Bolla K.
      • Mouratidis M.
      • Contoreggi C.
      • Matochik J.A.
      • Kurian V.
      • et al.
      Decision-making in a risk-taking task: A PET study.
      ). Functional imaging (
      • Ernst M.
      • Bolla K.
      • Mouratidis M.
      • Contoreggi C.
      • Matochik J.A.
      • Kurian V.
      • et al.
      Decision-making in a risk-taking task: A PET study.
      ,
      • Fukui H.
      • Murai T.
      • Fukuyama H.
      • Hayashi T.
      • Hanakawa T.
      Functional activity related to risk anticipation during performance of the Iowa Gambling Task.
      ) and brain lesion studies have implicated distributed neural circuitry in supporting successful decision-making on the IGT, including the ventromedial (ventromedial prefrontal cortex [VMPFC]) (
      • Bechara A.
      • Damasio H.
      • Tranel D.
      • Damasio A.R.
      Deciding advantageously before knowing the advantageous strategy.
      ) and ventrolateral prefrontal cortex (
      • Lawrence N.S.
      • Jollant F.
      • O'Daly O.
      • Zelaya F.
      • Phillips M.L.
      Distinct roles of prefrontal cortical subregions in the Iowa Gambling Task.
      ), and amygdala (
      • Bechara A.
      • Damasio H.
      • Damasio A.R.
      • Lee G.P.
      Different contributions of the human amygdala and ventromedial prefrontal cortex to decision-making.
      ), areas that have been associated with BD (
      • Blumberg H.P.
      • Leung H.C.
      • Skudlarski P.
      • Lacadie C.M.
      • Fredericks C.A.
      • Harris B.C.
      • et al.
      A functional magnetic resonance imaging study of bipolar disorder: State- and trait-related dysfunction in ventral prefrontal cortices.
      ).
      On the basis of these findings, we predict that decision-making is impaired in the acute phases but also in remission of BD. The IGT results were further analyzed in relation to sociodemographic and clinical variables in BD patients.

      Methods and Materials

      Participants

      The study population comprised 167 BPs (98 women and 69 men; age range: 18–65 years) and 150 healthy volunteers (75 women and 75 men; age range: 19–64 years; see power analysis in Section 1 of Supplement 1). Diagnostic assessment of the patients was initially performed by an experienced psychiatrist and was confirmed using the Structural Clinical Interview for DSM-IV (
      • Spitzer R.L.
      • Williams J.B.W.
      • Gibbon M.
      • First M.B.
      SCID I, version 2.0 for DSM-IV.
      ). Exclusion criteria included a history of head injury or neurologic disease. All subjects had normal thyroid function. No patient had received electroconvulsive therapy or had a history of substance abuse within the previous 6 months. Control subjects had no psychiatric history, no first-degree relatives with BD, and were not taking any drugs that might affect cognition. The study was approved by the local ethics committees. After complete description of the study to the subjects, written informed consent was obtained.

      Manic Group

      Forty-five inpatients suffering from mania were included (30 from Marseille University Department of Psychiatry and 15 from Oxford University Department of Psychiatry). All patients met the DSM-IV criteria for bipolar I disorder, manic episode, with a score greater than 12 on the Young Mania Rating Scale (YMRS) (
      • Young R.C.
      • Biggs J.T.
      • Ziegler V.E.
      • Meyer D.A.
      A rating scale for mania: Reliability, validity and sensitivity.
      ) and less than 7 on the Hamilton Depression Rating Scale (HDRS, 17 items) (
      • Hamilton M.
      A rating scale for depression.
      ). The manic group comprised 22 women and 23 men (age range: 18–65 years). Thirty-two of the 45 patients were receiving antipsychotic drugs at the time of testing. No patients were receiving D2-agonist antipsychotics. Fifteen patients were receiving typical antipsychotics. Seventeen patients were receiving the atypical antipsychotics risperidone (n = 1), clozapine (n = 1), and olanzapine (n = 15). Fourteen of these patients were also receiving lithium, valproate, carbamazepine, or a combination of these drugs. Eleven patients were receiving lithium, valproate, carbamazepine, or a combination of these drugs without a neuroleptic. Twenty-three patients were receiving a benzodiazepine, typically diazepam or lorazepam.

      Depressed Group

      Thirty-two inpatients suffering from bipolar depression were included (16 from Marseille University Department of Psychiatry and 16 from Montpellier University Department of Psychiatry). All patients met the DSM-IV criteria for bipolar I disorder, depressed episode, with a score of greater than 12 on the HDRS) (
      • Hamilton M.
      A rating scale for depression.
      ) and less than 7 on the YMRS (
      • Young R.C.
      • Biggs J.T.
      • Ziegler V.E.
      • Meyer D.A.
      A rating scale for mania: Reliability, validity and sensitivity.
      ). The depressed group comprised 18 women and 14 men (age range: 22–63 years). Sixteen of the 32 patients were receiving antipsychotic drugs at the time of testing. No patients were receiving typical antipsychotics. Sixteen patients were receiving the atypical antipsychotics aripiprazole (n = 3), olanzapine (n = 7), risperidone (n = 4), amisulpride (n = 1), and clozapine (n = 1). Fourteen were also receiving lithium, valproate, carbamazepine, or a combination of these drugs. Thirteen patients were receiving lithium, valproate, carbamazepine, or a combination of these drugs without a neuroleptic. Sixteen patients were receiving lithium, valproate, carbamazepine, or a combination of these drugs with an antidepressant. Sixteen patients were receiving a benzodiazepine.

      Euthymic Group

      Ninety bipolar patients in clinical remission were included (60 from Montpellier University Department of Psychiatry and 30 from Oxford University Department of Psychiatry). All patients were euthymic at the time of testing, as defined by a score of less than 8 on the HDRS (
      • Hamilton M.
      A rating scale for depression.
      ) and less than 8 on the YMRS (
      • Young R.C.
      • Biggs J.T.
      • Ziegler V.E.
      • Meyer D.A.
      A rating scale for mania: Reliability, validity and sensitivity.
      ), and met the DSM-IV criteria for bipolar I disorder, euthymic state. The euthymic group comprised 58 women and 32 men (age range: 18–65 years). Fifty-two of the 90 patients were receiving lithium, valproate, carbamazepine, or a combination of these drugs. Twenty-two patients were also receiving antipsychotics. Six patients were receiving typical antipsychotics. Seventeen patients were receiving the atypical antipsychotics aripiprazole (n = 1), risperidone (n = 5), and olanzapine (n = 11). Thirty patients were receiving lithium, valproate, carbamazepine, or a combination of these drugs without an antipsychotic. Thirty-eight patients were receiving lithium, valproate, carbamazepine, or a combination of these drugs with an antidepressant. Twenty-three patients were receiving a benzodiazepine.

      Control Group

      One hundred fifty healthy volunteers were recruited as control subjects by advertisements in the three communities (15 from Marseille, 30 from Oxford, and 105 from Montpellier). Control subjects had no psychiatric or neurological history, no first-degree relatives with BD, and were not taking any drugs that might affect cognition.

      Procedure

      Patients' mood was formally assessed using the YMRS and HDRS. Level of education and National Adult Reading Test (NART) (
      • Nelson H.E.
      National Adult Reading Test (NART) Test Manual.
      ) were used indirectly to assess premorbid intelligence level in the four groups (Table 1). NART Z scores were defined as the Z standardization scores of NART and fNART (French language adaptation of the NART) (
      • MacKinnon A.
      • Ritchie K.
      • Mulligan R.
      The measurement properties of a French language adaptation of the National Adult Reading Test.
      ) scores, for English and French participants, respectively. Descriptive data for the 30 manic patients from Marseille, 15 manic and 30 euthymic patients from Oxford, and 60 euthymic patients from Montpellier have been published previously (
      • Adida M.
      • Clark L.
      • Pomietto P.
      • Kaladjian A.
      • Besnier N.
      • Azorin J.M.
      • et al.
      Lack of insight may predict impaired decision making in manic patients.
      ,
      • Clark L.
      • Iversen S.D.
      • Goodwin G.M.
      A neuropsychological investigation of prefrontal cortex involvement in acute mania.
      ,
      • Clark L.
      • Iversen S.D.
      • Goodwin G.M.
      Sustained attention deficit in bipolar disorder.
      ,
      • Jollant F.
      • Bellivier F.
      • Leboyer M.
      • Astruc B.
      • Torres S.
      • Verdier R.
      • et al.
      Impaired decision making in suicide attempters.
      ).
      Table 1Demographic and Clinical Characteristics of Manic, Depressed, and Euthymic Bipolar Patients, and Healthy Control Subjects
      CharacteristicManic (n = 45)Depressed (n = 32)Euthymic (n = 90)Control (n = 150)Analysis
      MeanSDMeanSDMeanSDMeanSDFdfp
      Age (years)37.812.743.810.139.31238.810.62.003,313.11
      Level of Education (years)14.03.114.63.214.03.514.33.0.483,313.70
      NART (Z score)
      NART score assesses premorbid IQ. NART Z scores were defined as the Z standardization scores of NART and fNART (French language adaptation of the NART) scores, for English and French participants, respectively.
      –.11.2.1.9–.21.1.0.9.893,313.45
      Number of Manic Episodes4.23.13.72.43.33.0
      Number of Depressive Episodes3.82.75.63.24.13.8
      Age at Onset of Bipolar Disorder (Years)23.76.625.17.424.77
      YMRS Score21.65.51.441.83.92.8
      HDRS Score6.163.718.64.82.32.5
      N%N%N%N%χ2dfp
      Sex2.533,313.12
       Male2351.11443.83235.67550.0
       Female2248.91856.35864.47550.0
      Medication
       Antipsychotic (D1/D2/D3/D4/D5 Ki values)
      The Dj-inhibitor constant Ki of an antipsychotic is the concentration of that antipsychotic needed to inhibit Dj receptor (j = 1–5). Ki values are reported in nanomolar units. Classically, the potency of an antipsychotic is indicated by the D2 Ki. The higher is the Ki, the lower is the potency. An antipsychotic is said to be D2 specific when D2 Ki is higher than D1, D3, D4, and D5 Ki. For example, we may infer from the values between brackets, that amisulpride is D2- and D3-specific.
      ,
      Ki values determined by the National Institute of Mental Health (NIMH) Psychoactive Drug Screening Program. Ki values selected were those listed as NIMH Psychoactive Drug Screening Program (PDSP) assay–certified data, determined from assays using the cloned human receptors with drug of interest as test ligands. For Ki values for which PDSP certified assay data were not listed, the average Ki value from assay data compiled on the PDSP website using the cloned human receptors with drug of interest as the test ligand was used.
      3271.116502224.447.042,164<.001
       D2 Antagonist3271.11340.62224.46.512,164<.05
       Typical: Haloperidol (83/2/12/3.9/147)1533.30.066.751.062,164<.001
       Atypical1737.81340.61617.88.882,164<.05
       Risperidone (60/4.9/12.2/18. 6/16)12.2413.355.63.012,164.22
       Amisulpride (> 104/1.3/2.4/> 103 > 104)0.013.30.04.392,164.11
       Clozapine (189/431/646/39/235)12.213.30.02.682,164.26
       Olanzapine (58/72/63/17.1/90)1533.3721.91112.29.912,164<.01
      D2 Agonist
       Aripiprazole (387/0.95/4.9/514/> 103)0.031011.18.742,164<.05
       Lithium920.01031.32831.128.432,164<.05
       Anticonvulsant1740.52784.44044.417.672,164<.001
       Antidepressant0.01546.93943.329.572,164<.001
       SSRI0.0724.12527.815.222,164<.001
       SNRI0.0310.3910.04.902,164.09
       Tricyclic0.013.444.42.022,164.36
       Other AD0.022.226.93.562,164.17
       Benzodiazepine2354.81650.02730.09.742,164<.01
       Anticholinergic410412.522.26.232,164<.05
      AD, antidepressant; D2, dopamine receptor subtype 2; HDRS, Hamilton Depression Rating Scale; Ki, inhibitor dissociation constant; NART, National Adult Reading Test; SSRI, selective serotonin reuptake inhibitor; SNRI, serotonin and norepinephrine reuptake inhibitor; Euthymic; YMRS, Young Mania Rating Scale.
      a NART score assesses premorbid IQ. NART Z scores were defined as the Z standardization scores of NART and fNART (French language adaptation of the NART) scores, for English and French participants, respectively.
      b The Dj-inhibitor constant Ki of an antipsychotic is the concentration of that antipsychotic needed to inhibit Dj receptor (j = 1–5). Ki values are reported in nanomolar units. Classically, the potency of an antipsychotic is indicated by the D2 Ki. The higher is the Ki, the lower is the potency. An antipsychotic is said to be D2 specific when D2 Ki is higher than D1, D3, D4, and D5 Ki. For example, we may infer from the values between brackets, that amisulpride is D2- and D3-specific.
      c Ki values determined by the National Institute of Mental Health (NIMH) Psychoactive Drug Screening Program. Ki values selected were those listed as NIMH Psychoactive Drug Screening Program (PDSP) assay–certified data, determined from assays using the cloned human receptors with drug of interest as test ligands. For Ki values for which PDSP certified assay data were not listed, the average Ki value from assay data compiled on the PDSP website using the cloned human receptors with drug of interest as the test ligand was used.

      Iowa Gambling Task

      The computerized version of the IGT (
      • Bechara A.
      • Damasio H.
      • Tranel D.
      • Damasio A.R.
      Deciding advantageously before knowing the advantageous strategy.
      ) was used in which the participant plays for a pretend monetary reward. The participant is required to make a series of 100 choices from four decks of cards, labeled A, B, C, and D. Each card choice results in a monetary win, but occasional choices also result in monetary loss, and the four decks differ in the profile of wins and losses. At the start of the task, the participant has no information about the four decks and must learn to choose advantageously based on trial-by-trial feedback. Penalties begin after 15 picks of cards.
      Decks A and B are associated with high immediate wins ($100/choice) but occasionally larger penalties that result in a net loss over time. Decks C and D are associated with smaller immediate wins ($50/choice) but lower long-term losses, such that participants accumulate gradual profit from choosing these decks.
      Decks B and D provide low-frequency but high-magnitude penalties (with a ratio of total wins to total losses higher in deck D than in deck B, whereas decks A and C provide high-frequency but low-magnitude penalties (with a ratio of total wins to total losses higher in deck C than deck A). Thus, profitability of the decks (C + D vs. A + B) is orthogonalized from punishment frequency/magnitude (B + D vs. A + C).

      Statistical Analysis

      Choices in the IGT were analyzed for individual decks A, B, C, and D (Section 2 of Supplement 1), over five blocks of 20 trials (Figure 1) and over 100 picks of cards (Figure 2), and classified as advantageous (“safe”) for decks C and D and disadvantageous (“risky”) for decks A and B. The overall net score or decision-making ability is the difference between the total number of advantageous and disadvantageous choices. Net scores were calculated for each block of 20 trials (Figure 1), for the first 40 and the last 60 trials (Figure 3, Section 3 of Supplement 1). Data were also analyzed in terms of sensitivity to punishment frequency by calculating a difference score [(B + D) – (A + C)], where a positive score indicates a preference for low-frequency/high-magnitude penalties and a negative score a preference for high-frequency/low-magnitude penalties. The difference score [(B + D) – (A + C)] were calculated for the first 40 and the last 60 trials (Figure 4, Section 3 of Supplement 1).
      Figure thumbnail gr1
      Figure 1Number of cards selected in individual deck A, B, C, and D and decision-making ability on the Iowa Gambling Task (IGT), in the manic, depressed, and euthymic bipolar patients (BPs), and healthy controls (HCs), graphed as a function of trial block.a,b (A) In HCs, the number of cards selected changed significantly over the course of the task in deck A (one-way analysis of variance [ANOVA] for repeated measures F = 18.17, df = 4, p < .001), deck B (one-way ANOVA for repeated measures F = 7.40, df = 4, p < .001), deck C (one-way ANOVA for repeated measures F = 10.95, df = 4, p < .001), and D (one-way ANOVA for repeated measures F = 4.77, df = 4, p < .01). (B) In euthymic BPs, the number of cards selected changed significantly over the course of the task in deck A (one-way ANOVA for repeated measures F = 3.42, df = 4, p < .05), deck B (one-way ANOVA for repeated measures F = 7.57, df = 4, p < .001), deck C (one-way ANOVA for repeated measures F = 6.94, df = 4, p < .001), and deck D (one-way ANOVA for repeated measures F = 13.69, df = 4, p < .001). (C) In manic BPs, the number of cards selected changed significantly over the course of the task in deck C (one-way ANOVA for repeated measures F = 2.46, df = 4, p < .05) but not in deck A (one-way ANOVA for repeated measures F = 1.70, df = 4, p = .15), deck B (one-way ANOVA for repeated measures F = 1.19, df = 4, p = .32), and deck D (one-way ANOVA for repeated measures F = 1.31, df = 4, p = .27). (D) In depressed BPs, the number of cards selected changed significantly over the course of the task in deck D (one-way ANOVA for repeated measures F = 4.00, df = 4, p < .01) but not in deck A (one-way ANOVA for repeated measures F = 2.08, df = 4, p = .16), deck B (one-way ANOVA for repeated measures F = 2.50, df = 4, p = .06), and deck C (one-way ANOVA for repeated measures F = .27, df = 4, p = .6). (E) Net scores changed significantly over the course of the task in the HCs (one-way ANOVA for repeated measures F = 38.3, df = 4, p < .001) but not in the manic (one-way ANOVA for repeated measures F = 4.8, df = 4, p = .24), depressed (one-way ANOVA for repeated measures F = 3.1, df = 4, p = .13), and euthymic BPs (one-way ANOVA for repeated measures F = 1.4, df = 4, p = .06). a Choices in the IGT were analyzed for each individual deck A, B, C, and D over 100 picks of cards and for each block of 20 trials and classified as disadvantageous for decks A and B (the “risky” decks are associated with high immediate wins but occasionally larger penalties that result in a net loss over time) and as advantageous for decks C and D (the “safe” decks are associated with smaller immediate wins but negligible long-term losses that results in gradual profit). B and D are decks with low-frequency/high-magnitude penalties, and A and C are decks with high-frequency/low-magnitude penalties. Penalties begin after 10 picks of cards. b *p < .05, **p < .01, and ***p < .001.
      Figure thumbnail gr2
      Figure 2Number of cards selected in individual decks A, B, C and D, and decision-making ability on the Iowa Gambling Task (IGT) in the manic, depressed, and euthymic bipolar patients (BPs) and healthy control subjects (HCs), over 100 picks of cards.a,b,c,d (A) One-way analysis of variance (ANOVA) revealed significant differences in deck A overall score between the four groups [F(3,313) = 3.24, p < .05], and post hoc Tukey tests of honestly significant differences (HSD) revealed that manic BPs selected significantly more cards from the deck A than did HCs (p < .05; effect size [d] = .48), with no significant differences between depressed BPs and HCs (p = .99), euthymic BPs and HCs (p = .43), or depressed and euthymic BPs (p = .73). (B) One-way ANOVA revealed significant differences in deck B overall score between the four groups [F(3,313) = 7.50, p < .001] and post hoc Tukey tests of HSD revealed that manic (p < .01; d = .68), depressed (p = .01; D = .79), and euthymic (p < .05; d = .44) BPs selected significantly more cards from the deck B than HCs with no significant differences between manic and depressed (p = .94), manic and euthymic (p = .58), or depressed and euthymic (p = .28) BPs. (C) One-way ANOVA revealed no significant differences in deck C overall score between the four groups [F(3,313) = 2.45, p = .071]. (D) One-way ANOVA revealed significant differences in deck D overall score between the four groups [F(3,313) = 4.42, p < .01], and post hoc Tukey tests of HSD revealed that manic BPs selected significantly fewer cards from the deck D than HCs (p < .01; effect size d = .53) with no significant differences between depressed BPs and HCs (p = .40), euthymic BPs and HCs (p = .08), or depressed and euthymic (p = .99) BPs. One-way ANOVA revealed highly significant differences in decision-making ability between the four groups [F(3,313) = 8.1, p < .001]. Post hoc Tukey tests of HSD revealed that manic (p < .001; effect size d = .68), depressed (p < .01; d = .59), and euthymic (p < .05; D = .35) BPs selected significantly more cards from the risky decks than HCs with no significant differences between manic and depressed (p = .97), manic and euthymic (p = .23), or depressed and euthymic (p = .62) BPs.a (E) Choices in the IGT were analyzed for each individual deck (A, B, C, and D) over 100 picks of cards and for each block of 20 trials and classified as disadvantageous for decks A and B (the “risky” decks are associated with high immediate wins but occasionally larger penalties that result in a net loss over time) and as advantageous for decks C and D (the “safe” decks are associated with smaller immediate wins but negligible long-term losses that results in gradual profit). B and D are decks with low-frequency/high-magnitude penalties, whereas decks A and C are decks with high-frequency/low-magnitude penalties. Penalties begin after 10 picks of cards.b *p < .05, **p < .01, and ***p < .001.c Cohen's d effect size is calculated as the difference between each of the three patient groups and the control group. The value of d is typed at the top of each patient group's histogram.d Budescu (
      • Budescu D.V.
      The power of the F test in normal populations with heterogeneous variances.
      ) reported that for normally distributed populations with heterogeneous variances, substituting for σ, in the denominator of equation f = σm/σ, the square root of the ni-weighted population variance, results in good power approximations.
      Figure thumbnail gr3
      Figure 3Iowa Gambling Task (IGT) net score, in the manic, depressed, and euthymic bipolar patients (BPs) and healthy control subjects (HCs), over the first 40 and the last 60 picks of cards,a,b,c (A) One-way analysis of variance (ANOVA) revealed no significant differences in IGT net score between the four groups [F(3,313) = 1.61, p = .18] over the first 40 picks of cards. (B) One-way ANOVA revealed highly significant differences in IGT net score between the four groups [F(3,313) = 7.9, p < .001] over the last 60 picks of cards. Post hoc Tukey tests of honest significant difference revealed that manic (p < .001), depressed (p < .05), and euthymic (p < .01) BPs selected significantly more cards from the risky decks than HCs with no significant differences between manic and depressed (p = .86), manic and euthymic (p = .57), or depressed and euthymic (p = .99) BPs.a Choices in the IGT were analyzed for each individual deck (A, B, C, and D) over 100 picks of cards and for each block of 20 trials and classified as disadvantageous for decks A and B (the “risky” decks are associated with high immediate wins but occasionally larger penalties that result in a net loss over time) and as advantageous for decks C and D (the “safe” decks are associated with smaller immediate wins but negligible long-term losses that results in gradual profit). B and D are decks with low-frequency/high-magnitude penalties, whereas decks A and C are decks with high-frequency/low-magnitude penalties. Penalties begin after 10 picks of cards.b *p < .05, **p < .01, and ***p < .001.c Budescu (
      • Budescu D.V.
      The power of the F test in normal populations with heterogeneous variances.
      ) reported that for normally distributed populations with heterogeneous variances, substituting for σ, in the denominator of equation f = σm/σ, the square root of the ni-weighted population variance, results in good power approximations.
      Figure thumbnail gr4
      Figure 4Difference score [(B + D) – (A + C)] on the Iowa Gambling Task (IGT) in the manic, depressed, and euthymic bipolar patients (BPs) and healthy controls (HCs) over 100 picks of cards (sensitivity to punishment frequency), over the first 40, and over the last 60 picks of cards.a,b (A) All four groups preferred decks offering low-frequency/high-magnitude penalties (decks B and D) over those with high-frequency/low-magnitude penalties (decks A and C; t = 5.0, p < .001; t = 6.0, p < .001; t = 7.5, p < .001; and t = 7.4, p < .001 for manic, depressed, and euthymic BPs and HCs, respectively), with no significant differences between the four groups [one-way analysis of variance (ANOVA) F(3,313) = 1.5, p = .22]. (B) One-way ANOVA revealed no significant differences in [(B+D) - (A+C)] score between the four groups [F(3,313) = .53, p = .66] over the first 40 picks of cards. (C) One-way ANOVA revealed a significant difference in [(B + D) – (A + C)] score between the four groups [F(3,313) = 2.75, p < .05] over the last 60 picks of cards. Post hoc Tukey tests of honest significant difference revealed a significant difference between depressed (p < .05) but not manic (p = .98) or euthymic (p = .99) BPs and HCs, with a significant difference between depressed and euthymic (p < .05) but not between manic and depressed (p = .19) or between manic and euthymic (p = .98) BPs. a Choices in the IGT were analyzed for each individual deck (A, B, C, and D) over 100 picks of cards and for each block of 20 trials and classified as disadvantageous for decks A and B (the “risky” decks are associated with high immediate wins but occasionally larger penalties that result in a net loss over time) and as advantageous for decks C and D (the “safe” decks are associated with smaller immediate wins but negligible long-term losses that results in gradual profit). B and D are decks with low-frequency/high-magnitude penalties, whereas decks A and C are decks with high-frequency/low-magnitude penalties. Penalties begin after 10 picks of cards. b *p < .05.
      The demographic, clinical, and cognitive variables were normally distributed (as assessed by the Kolmogorov–Smirnov test and visual inspection) and were analyzed with parametric statistical tests with a threshold of p < .05 (two-tailed). Age, level of education, and NART Z score for the manic, depressed, and euthymic patients and for control subjects were compared by one-way ANOVA, while gender and medication received were compared using chi-squared tests.
      Individual deck (Section 2 of Supplement 1) and IGT summary measures were analyzed using analysis of variance (ANOVA) with group (manic, depressed, euthymic, control) as a between-subjects factor and block (1–20, 21–40, 41–60, 61–80, and 81–100) as a within-subjects factor. Sensitivity to punishment frequency scores were also compared with the value zero using single sample t tests. Post hoc Tukey tests of honestly significant differences (HSD) were performed to determine significant group differences and to control for Type I error (
      • Tukey J.W.
      • Brillinger D.R.
      The Collected Works of John W Tukey.
      ).
      Univariate analysis with Pearson's correlations was performed to test for associations between IGT net score for the 165 bipolar patients and demographic and clinical variables (gender, age, level of education, NART Z score, number of manic episodes, number of depressive episodes, total number of admissions, age at onset, family history of BD, HDRS score, YMRS score, alcohol misuse history, drug misuse history, suicide attempt history, use of selective serotonin reuptake inhibitor (SSRI) antidepressant, use of serotonin and norepinephrine reuptake inhibitor (SNRI) antidepressant, use of tricyclic antidepressant, use of other antidepressant, use of aripiprazole, use of haloperidol, use of olanzapine, use of clozapine, use of amisulpride, use of risperidone, use of anticonvulsant, use of lithium, use of anticholinergic, and use of benzodiazepine). A threshold of p = .2 (
      • Cohen J.
      • Cohen P.
      Applied Multiple Regression/Correlation Analysis for the Behavioural Sciences.
      ) was set to select variables for multivariate regression analysis. One-step multivariate linear regression analysis was then carried out to determine how much of the variation in IGT net score for bipolar patients could be explained and which demographic and clinical variables might predict variations in decision-making ability in BPs.

      Results

      Demographic and Clinical Characteristics

      The demographic and clinical characteristics of the manic, depressed, euthymic, and control groups are shown in Table 1. There was no significant between-group difference with respect to age, level of education, NART Z score, and sex ratio.
      For manic patients, mean YMRS score was 21.6 ± 5.5 and mean HDRS score was 6.16 ± 3.7. Nine patients were experiencing a first episode of mania, and 22 had a manic episode with psychotic features. The manic group had experienced an average of 4.2 ± 3.1 manic episodes (range: 1–12) and 3.8 ± 2.7 depressive episodes (range: 0−9), with a mean age at onset of 23.7 ± 6.6 years (range: 15–38).
      For depressed patients, mean YMRS score was 1.44 ± 1.8 and mean HDRS score was 18.6 ± 4.8. The depressed group had experienced an average of 3.7 ± 2.4 manic episodes (range: 1–10) and 5.6 ± 3.2 depressive episodes (range: 0–13), with a mean age at onset of 25.1 ± 7.4 years (range: 15–45).
      For euthymic patients, mean YMRS score was 3.9 ± 2.8 and mean HDRS score was 2.3 ± 2.5. The euthymic group had experienced an average of 3.3 ± 3.0 manic episodes (range: 0−9) and 4.1 ± 3.8 depressive episodes (range: 1–9), with a mean age at onset of 24.7 ± 7.0 years (range: 16–45).

      IGT Performance

      Decision-Making Ability

      Net score changed significantly over the course of the task in the healthy control group (one-way ANOVA for repeated measures F = 38.3, df = 4, p < 10–3) but not in the manic (F = 4.8, df = 4, p = .24), depressed (F = 3.1, df = 4, p = .13), and euthymic groups (F = 1.4, df = 4, p = .06; Figure 1E) (Figure 1A, 1B, 1C, and 1D, see comments in Section 2 of Supplement 1).
      When IGT performance was analyzed according to the decision-making ability, one-way ANOVA revealed highly significant differences in decision-making ability between the four groups ([F(3,313) = 8.1, p < .001], and post hoc Tukey tests of HSD revealed that manic (p < .001; effect size [d] = .68), depressed (p < .01; d = .59), and euthymic (p < .05; d = .35) BPs selected significantly more cards from the risky decks than healthy control subjects, with no significant differences between manic and depressed (p = .97), manic and euthymic (p = .23), or depressed and euthymic (p = .62) BPs (Figure 2E) (Figure 2A, 2B, 2C and 2D, see comments in Section 2 of Supplement 1).
      IGT performance was also analyzed according to the decision-making ability, over the first 40 picks and over the last 60 picks of cards (Figure 3, see comments in Section 3 of Supplement 1).

      Sensitivity to Punishment Frequency

      When IGT performance was analyzed according to sensitivity to punishment frequency, all four groups preferred decks offering low-frequency penalties (B + D) over those with high-frequency penalties (A + C; t = 5.0, p < .001; t = 6.0, p < .001; t = 7.5, p < .001; and t = 7.4, p < .001 for manic, depressed, and euthymic BPs and healthy volunteers, respectively), with no significant differences between the four groups [one-way ANOVA F(3,313) = 1.5, p = .22] (Figure 4A) (Figure 4B and 4C, see comments in Section 3 of Supplement 1).

      Association Between IGT Performance and Clinical Variables

      Univariate analysis of sociodemographic and clinical variables was carried out against IGT net score for the 167 BD patients. Gender (females performed worse; r = –.23, p < .01), level of education (r = .26, p = .001), NART Z score (r = .17, p < .05), total number of admissions (r = –.22, p < .05), HDRS score (r = –.25, p < .01), YMRS score (r = –.28, p < .01), use of benzodiazepine (r = –.29, p < .001), and family history of BD (r = –.22, p < .05) correlated with IGT net score at p < .05 and were subjected to multivariate analysis with age at onset and use of serotonin and norepinephrine reuptake inhibitor antidepressants (SNRIs; p < .2).
      In multivariate analysis, level of education (β = .26, t = 2.79, p < .01), HDRS score (β = –.24, t = 2.46, p < .05), family history of BD (β = –.23, t = -2.53, p < .05), use of benzodiazepine (β = –.19, t = –1.93, p < .05), and use of SNRI antidepressant (β = .21, t = 2.29, p < .05) were significant predictors of IGT net score in bipolar patients (Table 2).
      Table 2Multivariate Linear Regression Analysis of Demographic and Clinical Variables Associated with Decision-Making Ability in Patients with Bipolar Disorder
      The significance of the explained amount of variance (R2) was assessed by analysis of variance [F(7,92) = 6.37; p < .001].
      VariableAnalysis
      βtp
      Constant1.09.28
      Sex.03.28.78
      Level of Education (years).262.79<.01
      R2 = .41; adjusted R2 = .35.
      NART (Z score)
      NART score assesses premorbid IQ. NART Z scores were defined as the Z standardization scores of NART and fNART (French language version) scores, for English and French participants, respectively.
      –.14–1.56.12
      YMRS Score–.18–1.77.08
      HDRS Score–.24–2.46<.05
      R2 = .41; adjusted R2 = .35.
      Total No. Hospital Admissions–.10–1.06.29
      Age at Onset of BD (years).06.64.53
      Familial History of BD–.23–2.53<.05
      R2 = .41; adjusted R2 = .35.
      Medication Received
      SNRI AD.212.29<.05
      R2 = .41; adjusted R2 = .35.
      Benzodiazepine–.19–1.93<.05
      R2 = .41; adjusted R2 = .35.
      AD, antidepressant; BD, bipolar disorder; HDRS, Hamilton Depression Rating Scale; NART, National Adult Reading Test; SNRI, serotonin and norepinephrine reuptake inhibitor; YMRS, Young Mania Rating Scale.
      a The significance of the explained amount of variance (R2) was assessed by analysis of variance [F(7,92) = 6.37; p < .001].
      b R2 = .41; adjusted R2 = .35.
      c NART score assesses premorbid IQ. NART Z scores were defined as the Z standardization scores of NART and fNART (French language version) scores, for English and French participants, respectively.

      Discussion

      Impaired decision-making was detected in all three phases of BD. Risk-taking behavior is a diagnostic feature of mania, and depressed patients also exhibit difficulties in decision-making in their daily lives. The translation of these ecological observations to the laboratory is difficult. It is therefore relevant to report that, using an experimental paradigm to detect risky-choice preference, both manic and depressed patients showed lower scores than healthy volunteers. These findings in the acute phases of BD are broadly consistent with previous studies of decision-making deficits assessed with the IGT (
      • Adida M.
      • Clark L.
      • Pomietto P.
      • Kaladjian A.
      • Besnier N.
      • Azorin J.M.
      • et al.
      Lack of insight may predict impaired decision making in manic patients.
      ,
      • Clark L.
      • Iversen S.D.
      • Goodwin G.M.
      A neuropsychological investigation of prefrontal cortex involvement in acute mania.
      ,
      • Murphy F.C.
      • Rubinsztein J.S.
      • Michael A.
      • Rogers R.D.
      • Robbins T.W.
      • Paykel E.S.
      • Sahakian B.J.
      Decision-making cognition in mania and depression.
      ,
      • Rubinsztein J.S.
      • Fletcher P.C.
      • Rogers R.D.
      • Ho L.W.
      • Aigbirhio F.I.
      • Paykel E.S.
      • et al.
      Decision-making in mania: A PET study.
      ), with the Cambridge Gamble Task (
      • Murphy F.C.
      • Rubinsztein J.S.
      • Michael A.
      • Rogers R.D.
      • Robbins T.W.
      • Paykel E.S.
      • Sahakian B.J.
      Decision-making cognition in mania and depression.
      ,
      • Rubinsztein J.S.
      • Fletcher P.C.
      • Rogers R.D.
      • Ho L.W.
      • Aigbirhio F.I.
      • Paykel E.S.
      • et al.
      Decision-making in mania: A PET study.
      ,
      • Rubinsztein J.S.
      • Michael A.
      • Underwood B.R.
      • Tempest M.
      • Sahakian B.J.
      Impaired cognition and decision-making in bipolar depression but no “affective bias” evident.
      ), and in relation to an increased sensitivity to error during a two-choice prediction task (
      • Minassian A.
      • Paulus M.P.
      • Perry W.
      Increased sensitivity to error during decision-making in bipolar disorder patients with acute mania.
      ). In addition, we report impaired decision-making in euthymic patients, which is also in accordance with some previous studies of decision-making deficits assessed with the IGT. Yechiam et al. (
      • Yechiam E.
      • Hayden E.P.
      • Bodkins M.
      • O'Donnell B.F.
      • Hetrick W.P.
      Decision making in bipolar disorder: A cognitive modeling approach.
      ) reported a similar decision-making impairment in 14 euthymic patients assessed with the IGT. Christodoulou et al. (
      • Christodoulou T.
      • Lewis M.
      • Ploubidis G.B.
      • Frangou S.
      The relationship of impulsivity to response inhibition and decision-making in remitted patients with bipolar disorder.
      ) reported an IGT score close to 0 in a group of 25 euthymic subjects. These findings suggest that decision-making alterations could be considered as a trait abnormality in BD.
      The failure of the BD patients to select from the safe decks may reflect a classic “myopia for the future” (
      • Bechara A.
      • Damasio A.R.
      • Damasio H.
      • Anderson S.W.
      Insensitivity to future consequences following damage to human prefrontal cortex.
      ), that is, the inability to use outcome information to guide an advantageous long-term strategy, acting as a trait marker in BD. However, some subtle differences were observed between the BD groups in the individual deck analyses. First, all groups preferred the two decks that yielded infrequent penalties over those that yielded more frequent penalties. This indicates that bipolar patients appear to be normally sensitive to the impact of losses. Nevertheless, the depressed BD group exhibited a specific punishment-sensitive pattern of choice by selecting more cards in decks with low-frequency penalties (B, D) and fewer cards in decks with high-frequency penalties (A, C) than did other groups; this difference was statistically significant over the last 60 trials of the task. In contrast, manic bipolar patients were less likely to avoid low-magnitude/high-frequency punishments (A, C) and more likely to avoid high-magnitude/low-frequency (B, D), as if they were less sensitive than other bipolar groups to the impact of frequent and small losses. These differences may point to subtle state influences on reinforcement mechanisms that operate during decision-making, over and above a more fundamental trait impairment in risk-sensitive decision-making.
      This is also the first report of any associations between decision-making impairment in BD and clinical variables. Five independent variables were found to be associated with IGT overall score in the regression analysis: high depression ratings, low level of education, use of benzodiazepines, nonuse of SNRIs, and family history of BD. Several additional factors that were associated with IGT performance in the univariate, but not the multivariate, analysis, were probably intercorrelated with one or more of those predictors. Interestingly, a relationship was previously reported in BD between the number of episodes and the severity of depression, which could be explained by the presence of residual symptoms from prior depressive episodes (
      • Tohen M.
      • Vieta E.
      • Gonzalez-Pinto A.
      • Reed C.
      • Lin D.
      Baseline characteristics and outcomes in patients with first episode or multiple episodes of acute mania.
      ). Moreover, women with BD may experience more depressive episodes than men (
      • Angst J.
      • Felder W.
      • Frey R.
      • Stassen H.H.
      The course of affective disorders .I. Change of diagnosis of monopolar, unipolar, and bipolar illness.
      ). Thus, in our study, high levels of depression (HDRS score) might be correlated with both total number of admissions and gender. In support of this, our group of depressed patients exhibited a greater number of previous depressive episodes and a higher proportion of females in comparison with the manic group. Similarly, the association with level of education is consistent with previous reports of a positive correlation between performance in cognitive and executive function tests and level of education in BD (
      • Martinez-Aran A.
      • Vieta E.
      • Reinares M.
      • Colom F.
      • Torrent C.
      • Sanchez-Moreno J.
      • et al.
      Cognitive function across manic or hypomanic, depressed, and euthymic states in bipolar disorder.
      ). However, its association with decision-making is disputed (
      • Evans C.E.
      • Kemish K.
      • Turnbull O.H.
      Paradoxical effects of education on the Iowa Gambling Task.
      ).
      The association between the use of benzodiazepine and impaired decision-making is consistent with previous reports of these drugs to have an adverse effect on psychomotor ability and memory (
      • Stein R.A.
      • Strickland T.L.
      A review of the neuropsychological effects of commonly used prescription medications.
      ).
      The positive correlation between the use of SNRI antidepressants and decision-making ability in BD is consistent with the psychopharmacology of SNRI antidepressant action. Although SNRI antidepressants are commonly called dual-action serotonin-norepinephrine agents, they actually have a third action on dopamine in the prefrontal cortex, but not elsewhere in the brain: they enhance dopamine levels and increase dopamine's diffusion radius, probably enhancing the ability of dopamine to regulate cognition (
      • Stahl S.M.
      Stahl's Essential Psychopharmacology Neuroscientific Basis and Practical Application.
      ). Furthermore, a large increase of dopamine release may facilitate learning in corticostriatal systems (
      • Reynolds J.N.
      • Wickens J.R.
      Dopamine-dependent plasticity of corticostriatal synapses.
      ) when salient stimuli occurred. This salience may be produced by unexpected or high-magnitude rewarding or punishing events or by novelty (
      • Seamans J.K.
      • Yang C.R.
      The principal features and mechanisms of dopamine modulation in the prefrontal cortex.
      ,
      • Shizgal P.
      • Arvanitogiannis A.
      Neuroscience Gambling on dopamine.
      ). Thus, boosting dopamine with SNRIs may improve IGT performance; however, we assume that excessive tonic dopamine might lead to reduced punishment sensitivity.
      The association between family history of BD and impaired decision-making observed here also parallels the genetics of mood disorders in which bipolar patients with a family history of BD exhibit enrichment of genetic effects compared with bipolar patients without a family history of BD (
      • Craddock N.
      • Forty L.
      Genetics of affective (mood) disorders.
      ). One target for future research will be to determine whether impaired decision-making impairment might be a candidate endophenotype to BD. This is suggested by Lovallo et al. (
      • Lovallo W.R.
      • Yechiam E.
      • Sorocco K.H.
      • Vincent A.S.
      • Collins F.L.
      Working memory and decision-making biases in young adults with a family history of alcoholism: Studies from the Oklahoma family health patterns project.
      ) in alcoholic patients. According to Hasler et al. (
      • Hasler G.
      • Drevets W.C.
      • Gould T.D.
      • Gottesman II, M.H.K.
      Toward constructing an endophenotype strategy for bipolar disorders.
      ), a dysmodulation of motivation and reward might be a candidate endophenotype to BD. Thus, we would recommend that future studies test for other criteria for the identification of endophenotype: heritability, cosegregation with BD within families, and presence in unaffected relatives at a higher rate than in the general population (
      • Gottesman II, G.T.D.
      The endophenotype concept in psychiatry: Etymology and strategic intentions.
      ).
      Previous studies on decision-making in patients with psychiatric disorders suggest that age (
      • Denburg N.L.
      • Tranel D.
      • Bechara A.
      The ability to decide advantageously declines prematurely in some normal older persons.
      ), female sex (
      • Bolla K.I.
      • Eldreth D.A.
      • Matochik J.A.
      • Cadet J.L.
      Sex-Related Differences in a Gambling Task and its Neurological Correlates.
      ), drug misuse (
      • Dom G.
      • De Wilde B.
      • Hulstijn W.
      • van den Brink W.
      • Sabbe B.
      Decision-making deficits in alcohol-dependent patients with and without comorbid personality disorder.
      ), and suicide attempt (
      • Jollant F.
      • Bellivier F.
      • Leboyer M.
      • Astruc B.
      • Torres S.
      • Verdier R.
      • et al.
      Impaired decision making in suicide attempters.
      ) are associated with disadvantageous decision-making. In this study, these variables were not related to IGT performance. Age may be a critical factor in adolescent (
      • Hooper C.J.
      • Luciana M.
      • Conklin H.M.
      • Yarger R.S.
      Adolescents' performance on the Iowa Gambling Task: Implications for the development of decision making and ventromedial prefrontal cortex.
      ) and in old-age individuals (
      • Denburg N.L.
      • Cole C.A.
      • Hernandez M.
      • Yamada T.H.
      • Tranel D.
      • Bechara A.
      • Wallace R.B.
      The orbitofrontal cortex, real-world decision-making, and normal aging.
      ) who are not represented in our study. Female sex was associated with a lower IGT net score in bipolar patients and control subjects, but the effect did not survive multivariate analysis. The lack of replication in drug misuse and suicide attempt may be related to a strong effect of BD leading to a ground effect in these patients.
      Brain lesion studies implicate distributed neural circuitry in supporting successful decision-making on the IGT, including the VMPFC (
      • Lawrence N.S.
      • Jollant F.
      • O'Daly O.
      • Zelaya F.
      • Phillips M.L.
      Distinct roles of prefrontal cortical subregions in the Iowa Gambling Task.
      ,
      • Clark L.
      • Manes F.
      • Antoun N.
      • Sahakian B.J.
      • Robbins T.W.
      The contributions of lesion laterality and lesion volume to decision-making impairment following frontal lobe damage.
      ) and amygdala (
      • Bechara A.
      • Damasio H.
      • Damasio A.R.
      • Lee G.P.
      Different contributions of the human amygdala and ventromedial prefrontal cortex to decision-making.
      ). Several functional imaging studies of manic (
      • Blumberg H.P.
      • Leung H.C.
      • Skudlarski P.
      • Lacadie C.M.
      • Fredericks C.A.
      • Harris B.C.
      • et al.
      A functional magnetic resonance imaging study of bipolar disorder: State- and trait-related dysfunction in ventral prefrontal cortices.
      ), depressed (
      • Blumberg H.P.
      • Leung H.C.
      • Skudlarski P.
      • Lacadie C.M.
      • Fredericks C.A.
      • Harris B.C.
      • et al.
      A functional magnetic resonance imaging study of bipolar disorder: State- and trait-related dysfunction in ventral prefrontal cortices.
      ), and euthymic (
      • Blumberg H.P.
      • Leung H.C.
      • Skudlarski P.
      • Lacadie C.M.
      • Fredericks C.A.
      • Harris B.C.
      • et al.
      A functional magnetic resonance imaging study of bipolar disorder: State- and trait-related dysfunction in ventral prefrontal cortices.
      ) patients have also implicated VMPFC dysfunction. Thus, VMPFC attenuation may represent a trait feature of the disorder (
      • Blumberg H.P.
      • Leung H.C.
      • Skudlarski P.
      • Lacadie C.M.
      • Fredericks C.A.
      • Harris B.C.
      • et al.
      A functional magnetic resonance imaging study of bipolar disorder: State- and trait-related dysfunction in ventral prefrontal cortices.
      ). In support of this hypothesis, Stanfield et al. (
      • Stanfield A.C.
      • Moorhead T.W.
      • Job D.E.
      • McKirdy J.
      • Sussmann J.E.
      • Hall J.
      • et al.
      Structural abnormalities of ventrolateral and orbitofrontal cortex in patients with familial bipolar disorder.
      ) reported gray matter deficits in the bilateral VMPFC in 66 bipolar patients in different phases of the disorder. Our findings are consistent with a model of VMPFC abnormalities in mania, bipolar depression, and euthymia, leading to impaired decision-making.
      This study has several limitations. First, all bipolar patients were receiving some type of medication. Although univariate analysis was unable to detect any differences between patients who were receiving (71% of manic, 50% of depressed and 24% of euthymic BPs) or not receiving antipsychotic medication, there is a strong case for reinforcement learning performance and decision-making ability to be sensitive to dopaminergic agents (
      • Sevy S.
      • Hassoun Y.
      • Bechara A.
      • Yechiam E.
      • Napolitano B.
      • Burdick K.
      • et al.
      Emotion-based decision-making in healthy subjects: Short-term effects of reducing dopamine levels.
      ,
      • Beninger R.J.
      • Wasserman J.
      • Zanibbi K.
      • Charbonneau D.
      • Mangels J.
      • Beninger B.V.
      Typical and atypical antipsychotic medications differentially affect two nondeclarative memory tasks in schizophrenic patients: A double dissociation.
      ,
      • Maia T.V.
      • McClelland J.L.
      A reexamination of the evidence for the somatic marker hypothesis: What participants really know in the Iowa gambling task.
      ). Thus, comparison of decision-making ability between BD subgroups might be confounded by this point. In addition, we detected differences between BD patients who were receiving or not receiving benzodiazepines. These drugs have been reported to have an adverse effect on psychomotor ability and memory (
      • Stein R.A.
      • Strickland T.L.
      A review of the neuropsychological effects of commonly used prescription medications.
      ) and for anticonvulsants on general cognitive functioning (
      • Devinsky O.
      Cognitive and behavioral effects of antiepileptic drugs.
      ), although the importance of these effects is disputed (
      • Stein R.A.
      • Strickland T.L.
      A review of the neuropsychological effects of commonly used prescription medications.
      ). Second, we did not use any complementary neuropsychological test with the IGT to characterize BD patients in this study. Fellows et al. (
      • Fellows L.K.
      The role of orbitofrontal cortex in decision making: A component process account.
      ) showed that deficit in reversal learning was an important mechanism underlying the difficulties in the IGT. Maia et al. (
      • Maia T.V.
      • McClelland J.L.
      A reexamination of the evidence for the somatic marker hypothesis: What participants really know in the Iowa gambling task.
      ) reported that IGT might assess not only pure implicit decision-making, on the basis of somatic markers inaccessible to consciousness, but both conscious and nonconscious knowledge. Thus, impaired decision-making in BPs may be associated with defects in reversal learning or explicit memory. From this study, we cannot determine whether deficits reported in BPs are specific abnormalities in decision-making (and what subprocesses may be implicated) or refer to a more global impairment in cognitive control processes. Thus, further research should include additional investigations. Third, the multicenter nature of the study might have introduced potential site differences in clinical characteristics and IGT performance of bipolar and control participants. The only difference detected here between IGT net scores was for the 30 British euthymic subjects compared with the 60 French euthymic patients but was explained by level of education. Fourth, the study did not assess smoking status, which should be considered in further studies. Fifth, there is a strong case for an ecologically valid support for a link between cognitive and psychosocial functioning in BD (
      • Burdick K.E.
      • Goldberg J.F.
      • Harrow M.
      Neurocognitive dysfunction and psychosocial outcome in patients with bipolar I disorder at 15-year follow-up.
      ,
      • Rosa A.R.
      • Reinares M.
      • Michalak E.E.
      • Bonnin C.M.
      • Sole B.
      • Franco C.
      • et al.
      Functional impairment and disability across mood states in bipolar disorder.
      ). Further research should test for an association between decision-making ability and psychosocial functioning in BD (
      • Goldberg J.F.
      • McLeod L.D.
      • Fehnel S.E.
      • Williams V.S.
      • Hamm L.R.
      • Gilchrist K.
      Development and psychometric evaluation of the Bipolar Functional Status Questionnaire (BFSQ).
      ).
      In conclusion, manic, depressed, and euthymic BD patients showed poor performance in a laboratory test (IGT) of decision-making. Impaired decision-making was correlated with high depression scores, low level of education, use of benzodiazepines, nonuse of SNRIs, and family history of BD. Thus, the decision-making capacity of patients with BD is modulated by state, education/IQ, and medication. The observation that trait-related deficits in BD are characterized by an objective failure to weigh costs versus benefit in a simple game is of clinical interest and may represent a future therapeutic target.
      This work was supported by National Research Scientific Centre (CNRS), Marseille, France; National Institute of Health and Medical Research, Montpellier, France; Academic Hospital (CHU) Lapeyronie, CHU Sainte-Marguerite, and Oxford University. We thank Professors C. Lançon and J. Naudin for their collaboration; Dr. Adéla Ionita and Mrs. Nadia Corréard, Nathalie Viglianese-Salmon, and Sara-Nora Elissalde for their help in assessment; Anderson Loundou from the Public Health Department, University of Marseille, for his help concerning statistics; and Professors A. Bechara and A. Damasio for the use of the Iowa Gambling Task.
      Philippe Courtet has accepted reimbursement for advice or participation in industry-supported symposia from most pharmaceutical companies with an interest in bipolar disorder or suicide in the past 5 years. Jean-Michel Azorin has accepted reimbursement for advice or participation in industry-supported symposia from most pharmaceutical companies with an interest in bipolar disorder in the past 5 years and holds grants from Lilly and Sanofi-Aventis. Luke Clark is a consultant for Cambridge Cognition. Guy M. Goodwin has accepted reimbursement for advice or participation in industry-supported symposia from most pharmaceutical companies with an interest in bipolar disorder in the past 5 years and holds grants from Sanofi-Aventis for the balance trial and from Servier. Marc Adida received grants from Lilly (Bourse Lilly en Santé Mentale 2008) and Servier (Bourses de l'Institut Servier 2008 et 2009) for his postdoctoral years in United Kingdom. Fabrice Jollant, Nathalie Besnier, Sébastien Guillaume, Arthur Kaladjian, Pascale Mazzola-Pomietto, and Régine Jeanningros report no biomedical financial interests or potential conflicts of interest.

      Supplementary data

      References

        • Adida M.
        • Clark L.
        • Pomietto P.
        • Kaladjian A.
        • Besnier N.
        • Azorin J.M.
        • et al.
        Lack of insight may predict impaired decision making in manic patients.
        Bipolar Disord. 2008; 10: 829
        • Minassian A.
        • Paulus M.P.
        • Perry W.
        Increased sensitivity to error during decision-making in bipolar disorder patients with acute mania.
        J Affect Disord. 2004; 82: 203-208
        • Wingo A.P.
        • Harvey P.D.
        • Baldessarini R.J.
        Neurocognitive impairment in bipolar disorder patients: Functional implications.
        Bipolar Disord. 2009; 11: 113-125
        • Torres I.J.
        • Boudreau V.G.
        • Yatham L.N.
        Neuropsychological functioning in euthymic bipolar disorder: A meta-analysis.
        Acta Psychiatr Scand Suppl. 2007; : 17-26
        • Clark L.
        • Iversen S.D.
        • Goodwin G.M.
        A neuropsychological investigation of prefrontal cortex involvement in acute mania.
        Am J Psychiatry. 2001; 158: 1605-1611
        • Murphy F.C.
        • Rubinsztein J.S.
        • Michael A.
        • Rogers R.D.
        • Robbins T.W.
        • Paykel E.S.
        • Sahakian B.J.
        Decision-making cognition in mania and depression.
        Psychol Med. 2001; 31: 679-693
        • Rubinsztein J.S.
        • Fletcher P.C.
        • Rogers R.D.
        • Ho L.W.
        • Aigbirhio F.I.
        • Paykel E.S.
        • et al.
        Decision-making in mania: A PET study.
        Brain. 2001; 124: 2550-2563
        • Rubinsztein J.S.
        • Michael A.
        • Underwood B.R.
        • Tempest M.
        • Sahakian B.J.
        Impaired cognition and decision-making in bipolar depression but no “affective bias” evident.
        Psychol Med. 2006; 36: 629-639
        • Jollant F.
        • Guillaume S.
        • Jaussent I.
        • Bellivier F.
        • Leboyer M.
        • Castelnau D.
        • et al.
        Psychiatric diagnoses and personality traits associated with disadvantageous decision-making.
        Eur Psychiatry. 2007; 22: 455-461
        • Christodoulou T.
        • Lewis M.
        • Ploubidis G.B.
        • Frangou S.
        The relationship of impulsivity to response inhibition and decision-making in remitted patients with bipolar disorder.
        Eur Psychiatry. 2006; 21: 270-273
        • Rubinsztein J.S.
        • Michael A.
        • Paykel E.S.
        • Sahakian B.J.
        Cognitive impairment in remission in bipolar affective disorder.
        Psychol Med. 2000; 30: 1025-1036
        • Clark L.
        • Iversen S.D.
        • Goodwin G.M.
        Sustained attention deficit in bipolar disorder.
        Br J Psychiatry. 2002; 180: 313-319
        • Yechiam E.
        • Hayden E.P.
        • Bodkins M.
        • O'Donnell B.F.
        • Hetrick W.P.
        Decision making in bipolar disorder: A cognitive modeling approach.
        Psychiatry Res. 2008; 161: 142-152
        • Jollant F.
        • Bellivier F.
        • Leboyer M.
        • Astruc B.
        • Torres S.
        • Verdier R.
        • et al.
        Impaired decision making in suicide attempters.
        Am J Psychiatry. 2005; 162: 304-310
        • Bar-On R.
        • Tranel D.
        • Denburg N.L.
        • Bechara A.
        Exploring the neurological substrate of emotional and social intelligence.
        Brain. 2003; 126: 1790-1800
        • Bolla K.I.
        • Eldreth D.A.
        • Matochik J.A.
        • Cadet J.L.
        Neural substrates of faulty decision-making in abstinent marijuana users.
        Neuroimage. 2005; 26: 480-492
        • Ernst M.
        • Bolla K.
        • Mouratidis M.
        • Contoreggi C.
        • Matochik J.A.
        • Kurian V.
        • et al.
        Decision-making in a risk-taking task: A PET study.
        Neuropsychopharmacology. 2002; 26: 682-691
        • Fukui H.
        • Murai T.
        • Fukuyama H.
        • Hayashi T.
        • Hanakawa T.
        Functional activity related to risk anticipation during performance of the Iowa Gambling Task.
        Neuroimage. 2005; 24: 253-259
        • Bechara A.
        • Damasio H.
        • Tranel D.
        • Damasio A.R.
        Deciding advantageously before knowing the advantageous strategy.
        Science. 1997; 275: 1293-1295
        • Lawrence N.S.
        • Jollant F.
        • O'Daly O.
        • Zelaya F.
        • Phillips M.L.
        Distinct roles of prefrontal cortical subregions in the Iowa Gambling Task.
        Cereb Cortex. 2009; 19: 1134-1143
        • Bechara A.
        • Damasio H.
        • Damasio A.R.
        • Lee G.P.
        Different contributions of the human amygdala and ventromedial prefrontal cortex to decision-making.
        J Neurosci. 1999; 19: 5473-5481
        • Blumberg H.P.
        • Leung H.C.
        • Skudlarski P.
        • Lacadie C.M.
        • Fredericks C.A.
        • Harris B.C.
        • et al.
        A functional magnetic resonance imaging study of bipolar disorder: State- and trait-related dysfunction in ventral prefrontal cortices.
        Arch Gen Psychiatry. 2003; 60: 601-609
        • Spitzer R.L.
        • Williams J.B.W.
        • Gibbon M.
        • First M.B.
        SCID I, version 2.0 for DSM-IV.
        Lilly Research Laboratories, Indianapolis, IN1996
        • Young R.C.
        • Biggs J.T.
        • Ziegler V.E.
        • Meyer D.A.
        A rating scale for mania: Reliability, validity and sensitivity.
        Br J Psychiatry. 1978; 133: 429-435
        • Hamilton M.
        A rating scale for depression.
        J Neurol Neurosurg, Psychiatry. 1960; 23: 56-62
        • Nelson H.E.
        National Adult Reading Test (NART) Test Manual.
        NFER-Nelson, Berkshire, UK1982
        • MacKinnon A.
        • Ritchie K.
        • Mulligan R.
        The measurement properties of a French language adaptation of the National Adult Reading Test.
        Int J Methods Psychiatr Res. 1999; 8: 27-38
        • Tukey J.W.
        • Brillinger D.R.
        The Collected Works of John W Tukey.
        Wadsworth Advanced Books & Software, Monterrey, CA1994
        • Cohen J.
        • Cohen P.
        Applied Multiple Regression/Correlation Analysis for the Behavioural Sciences.
        3rd ed. Erlbaum, Mahwah, NJ2003
        • Bechara A.
        • Damasio A.R.
        • Damasio H.
        • Anderson S.W.
        Insensitivity to future consequences following damage to human prefrontal cortex.
        Cognition. 1994; 50: 7-15
        • Tohen M.
        • Vieta E.
        • Gonzalez-Pinto A.
        • Reed C.
        • Lin D.
        Baseline characteristics and outcomes in patients with first episode or multiple episodes of acute mania.
        J Clin Psychiatry. 2010; 71: 255-261
        • Angst J.
        • Felder W.
        • Frey R.
        • Stassen H.H.
        The course of affective disorders .I. Change of diagnosis of monopolar, unipolar, and bipolar illness.
        Arch Psychiatr Nervenkr. 1978; 226: 57-64
        • Martinez-Aran A.
        • Vieta E.
        • Reinares M.
        • Colom F.
        • Torrent C.
        • Sanchez-Moreno J.
        • et al.
        Cognitive function across manic or hypomanic, depressed, and euthymic states in bipolar disorder.
        Am J Psychiatry. 2004; 161: 262-270
        • Evans C.E.
        • Kemish K.
        • Turnbull O.H.
        Paradoxical effects of education on the Iowa Gambling Task.
        Brain Cogn. 2004; 54: 240-244
        • Stein R.A.
        • Strickland T.L.
        A review of the neuropsychological effects of commonly used prescription medications.
        Arch Clin Neuropsychol. 1998; 13: 259-284
        • Stahl S.M.
        Stahl's Essential Psychopharmacology Neuroscientific Basis and Practical Application.
        3rd ed. Cambridge University Press, New York2008
        • Reynolds J.N.
        • Wickens J.R.
        Dopamine-dependent plasticity of corticostriatal synapses.
        Neural Netw. 2002; 15: 507-521
        • Seamans J.K.
        • Yang C.R.
        The principal features and mechanisms of dopamine modulation in the prefrontal cortex.
        Prog Neurobiol. 2004; 74: 1-58
        • Shizgal P.
        • Arvanitogiannis A.
        Neuroscience.
        Science. 2003; 299: 1856-1858
        • Craddock N.
        • Forty L.
        Genetics of affective (mood) disorders.
        Eur J Hum Genet. 2006; 14: 660-668
        • Lovallo W.R.
        • Yechiam E.
        • Sorocco K.H.
        • Vincent A.S.
        • Collins F.L.
        Working memory and decision-making biases in young adults with a family history of alcoholism: Studies from the Oklahoma family health patterns project.
        Alcohol Clin Exp Res. 2006; 30: 763-773
        • Hasler G.
        • Drevets W.C.
        • Gould T.D.
        • Gottesman II, M.H.K.
        Toward constructing an endophenotype strategy for bipolar disorders.
        Biol Psychiatry. 2006; 60: 93-105
        • Gottesman II, G.T.D.
        The endophenotype concept in psychiatry: Etymology and strategic intentions.
        Am J Psychiatry. 2003; 160: 636-645
        • Denburg N.L.
        • Tranel D.
        • Bechara A.
        The ability to decide advantageously declines prematurely in some normal older persons.
        Neuropsychologia. 2005; 43: 1099-1106
        • Bolla K.I.
        • Eldreth D.A.
        • Matochik J.A.
        • Cadet J.L.
        Sex-Related Differences in a Gambling Task and its Neurological Correlates.
        Cereb Cortex. 2004; 14: 1226-1232
        • Dom G.
        • De Wilde B.
        • Hulstijn W.
        • van den Brink W.
        • Sabbe B.
        Decision-making deficits in alcohol-dependent patients with and without comorbid personality disorder.
        Alcohol Clin Exp Res. 2006; 30: 1670-1677
        • Hooper C.J.
        • Luciana M.
        • Conklin H.M.
        • Yarger R.S.
        Adolescents' performance on the Iowa Gambling Task: Implications for the development of decision making and ventromedial prefrontal cortex.
        Dev Psychol. 2004; 40: 1148-1158
        • Denburg N.L.
        • Cole C.A.
        • Hernandez M.
        • Yamada T.H.
        • Tranel D.
        • Bechara A.
        • Wallace R.B.
        The orbitofrontal cortex, real-world decision-making, and normal aging.
        Ann N Y Acad Sci. 2007; 1121: 480-498
        • Clark L.
        • Manes F.
        • Antoun N.
        • Sahakian B.J.
        • Robbins T.W.
        The contributions of lesion laterality and lesion volume to decision-making impairment following frontal lobe damage.
        Neuropsychologia. 2003; 41: 1474-1483
        • Stanfield A.C.
        • Moorhead T.W.
        • Job D.E.
        • McKirdy J.
        • Sussmann J.E.
        • Hall J.
        • et al.
        Structural abnormalities of ventrolateral and orbitofrontal cortex in patients with familial bipolar disorder.
        Bipolar Disord. 2009; 11: 135-144
        • Sevy S.
        • Hassoun Y.
        • Bechara A.
        • Yechiam E.
        • Napolitano B.
        • Burdick K.
        • et al.
        Emotion-based decision-making in healthy subjects: Short-term effects of reducing dopamine levels.
        Psychopharmacology Berl. 2006; 188: 228-235
        • Beninger R.J.
        • Wasserman J.
        • Zanibbi K.
        • Charbonneau D.
        • Mangels J.
        • Beninger B.V.
        Typical and atypical antipsychotic medications differentially affect two nondeclarative memory tasks in schizophrenic patients: A double dissociation.
        Schizophr Res. 2003; 61: 281-292
        • Maia T.V.
        • McClelland J.L.
        A reexamination of the evidence for the somatic marker hypothesis: What participants really know in the Iowa gambling task.
        Proc Natl Acad Sci U S A. 2004; 101: 16075-16080
        • Devinsky O.
        Cognitive and behavioral effects of antiepileptic drugs.
        Epilepsia. 1995; 36: S46-S65
        • Fellows L.K.
        The role of orbitofrontal cortex in decision making: A component process account.
        Ann N Y Acad Sci. 2007; 1121: 421-430
        • Burdick K.E.
        • Goldberg J.F.
        • Harrow M.
        Neurocognitive dysfunction and psychosocial outcome in patients with bipolar I disorder at 15-year follow-up.
        Acta Psychiatr Scand. 2010; 122: 499-506
        • Rosa A.R.
        • Reinares M.
        • Michalak E.E.
        • Bonnin C.M.
        • Sole B.
        • Franco C.
        • et al.
        Functional impairment and disability across mood states in bipolar disorder.
        Value Health. 2010; 13: 984-988
        • Goldberg J.F.
        • McLeod L.D.
        • Fehnel S.E.
        • Williams V.S.
        • Hamm L.R.
        • Gilchrist K.
        Development and psychometric evaluation of the Bipolar Functional Status Questionnaire (BFSQ).
        Bipolar Disord. 2010; 12: 32-44
        • Budescu D.V.
        The power of the F test in normal populations with heterogeneous variances.
        Educ Psychol Meas. 1982; 42: 409-416