Neurocognitive impairment in bipolar disorder patients: functional implications


  • Aliza P Wingo,

    1. Department of Psychiatry and Behavioral Sciences, Emory University, Atlanta, GA
    Search for more papers by this author
  • Philip D Harvey,

    1. Department of Psychiatry and Behavioral Sciences, Emory University, Atlanta, GA
    Search for more papers by this author
  • Ross J Baldessarini

    1. Department of Psychiatry and Neuroscience Program, Harvard Medical School, Boston
    2. Psychopharmacology Program and International Consortium for Bipolar Disorder Research, McLean Division of Massachusetts General Hospital, Belmont, MA, USA
    Search for more papers by this author

  • PDH has served as a consultant to or has received research support from AstraZeneca, Bristol-Myers Squibb, Eli Lilly & Co., Johnson & Johnson, Memory, Novartis, Pfizer, Solvay, Wyeth, and Sanofi-aventis. RJB is a consultant to or research collaborator with AstraZeneca, Auritec, Biotrofix, Janssen, JDS, Eli Lilly, Luitpold, NeuroHealing, Novartis, Pfizer, and SK-BioPharmaceuticals Corporations, but is not a member of pharmaceutical speakers bureaus, nor does he or any family member hold equity positions in biomedical or pharmaceutical corporations. APW has no relevant potential conflicts of interest.

Aliza P. Wingo, M.D., Department of Psychiatry, Emory University, 101 Woodruff Circle, NE, Suite 4000, Atlanta, GA 30322, USA. Fax: 404-727-3233; e-mail:


Background:  Functional recovery among treated bipolar disorder (BPD) patients is far less likely than syndromal and even symptomatic recovery. We hypothesized that increasingly well-documented aspects of cognitive impairment may contribute to poor functional outcomes in BPD patients, and reviewed the available research on the topic.

Methods:  Computerized literature searching identified 12 studies with 13 comparisons that simultaneously evaluated cognitive and functional status in euthymic (n = 8) or non-euthymic (n = 5 comparisons) adult BPD patients versus otherwise similar healthy controls.

Results:  In 6/8 studies of euthymic BPD patients and 5/5 studies of non-euthymic BPD patients, neurocognitive impairment was significantly associated with impaired psychosocial functioning, even after adjusting for residual mood symptoms and relevant demographic and clinical variables. Cognitive status was consistently assessed with standardized, performance-based neuropsychological tests, but functional status usually was based on subjective self-appraisals. Approximately 55% of BPD patients were unemployed.

Conclusions:  Available studies are limited by subjective assessments of functional status rather than objective, performance-based measures. Nevertheless, they support the hypothesis that enduring aspects of cognitive impairment found even in euthymic BPD patients are associated with inferior functioning. These findings encourage further studies with better assessment methods and greater rehabilitative efforts in BPD patients.

Growing evidence suggests that bipolar disorder (BPD) patients experience prominent neurocognitive impairment not only during acute mood episodes (1–3) but also during euthymia (1, 4–12). A recent meta-analytic review of 39 studies comparing neurocognitive functioning in 948 euthymic BPD patients and 1,128 normal controls, matched for age, sex, education, and estimated premorbid IQ, found medium-to-large effect sizes (ES, as group differences/variance) for impaired attention and processing speed (ES = 0.60–0.79), episodic memory (ES = 0.43–0.81), and executive functioning (ES = 0.47–0.71) in the BPD patients (11). Another meta-analytic review of 20 studies, also comparing neurocognitive performance in euthymic BPD patients to matched normal controls, found deficits of similar magnitude in measures of executive functions, verbal learning, immediate and delayed verbal memory, abstraction, sustained attention, and psychomotor speed in BPD patients (12). Neurocognitive impairment is observed in both bipolar I disorder (BP-I) and bipolar II disorder patients during euthymia, but usually is greater among BP-I patients (13, 14). Consistently, nearly two-thirds of BPD patients in clinical remission complain of subjective awareness of cognitive dysfunctions associated with the illness or its treatment (15).

Persistence of neurocognitive impairment during euthymia suggests that these deficits are an enduring component of the neuropsychopathology of BPD and not merely manifestations of acute illness. Neurocognitive impairment has been found to limit creativity, work performance, quality of life, self-esteem, and psychosocial functioning substantially in patients with other severe mental illnesses, especially schizophrenia (16–21). In BPD, there has been increasing recognition of more functional impairment than had been realized formerly, particularly with regard to independent living, personal relationships, and vocational success (22). In one recent review, only 19%–23% of adult BP-I patients were married (versus 60% of adults in the general population); 19%–58% were not living independently and usually resided with family members; 57%–65% were unemployed (versus 6% of the general population); and up to 80% were considered to have at least partial vocational disability (22). It is highly likely that sustained neurocognitive deficits critically contribute to major occupational, social, and other dysfunctions commonly observed in BPD patients (22, 23).

Specifically, we hypothesize that neurocognitive impairment in BPD patients adversely affects functional recovery, which is strikingly uncommon or delayed, even when euthymia is achieved with modern symptomatic treatments. Notably, the McLean-Harvard First Episode Mania study found that within two years of initial hospitalization for DSM-IV mania or mixed manic-depressive state, only 43% of BP-I patients achieved functional recovery, defined as regaining their own individual premorbid occupational and residential status, whereas 98% no longer met DSM syndromal criteria, and 72% had achieved symptomatic recovery, based on low scores on standard mania and depression ratings (24). Another U.S. study (25) of midcourse BP-I patients found that only 24% achieved similarly defined functional recovery within a year of an index hospitalization for mania or a mixed state. These findings indicate that 57%–75% of BP-I patients failed to attain functional recovery within 1–2 years after psychiatric hospitalization, even early in the illness course. These results are consistent with observations that BPD is often associated with disability, even early in the course of the illness, and suggest that current treatments are of limited value in facilitating functional recovery.

Factors associated with relatively poor functional outcomes among BPD patients not unexpectedly include poor premorbid functioning, as well as younger onset or more total years of illness, prominent major or minor depressive morbidity, psychotic features, more hospitalizations, limited education and training, drug or alcohol abuse, weak social supports, poverty, and others (22). However, in a recent review of 15 studies, Bauer and colleagues (26) found that, except for ongoing depressive symptoms, few of these factors were consistently associated with poor functional outcomes across studies. Neurocognitive functioning, remarkably, was not examined in most of the preceding studies of BPD patients, despite having been explored extensively in relation to functional disability in patients with schizophrenia and other neuropsychiatric conditions (17–21, 27).

Whether neurocognitive deficits adversely affect functional recovery in BPD patients, despite achievement of even sustained euthymia, is an important scientific and clinical question. We therefore reviewed published reports on relationships between neurocognitive performance and functional status among patients diagnosed with BPD of any type. Our principal hypothesis was that neurocognitive impairment would be associated with less favorable functional status in euthymic individuals with BPD.


We searched the Medline, Embase, and PsychInfo publication databases, considering reports from 1950 through April 2008, using key words: bipolar, mania, manic, cognitive, neurocognitive, neuropsychological, executive, verbal learning, memory, attention, concentration, processing speed, function, functioning, functional, employment, occupation, living status, and marriage. Computer searching was supplemented with bibliographic cross-referencing. Inclusion criteria included peer-reviewed reports on analyses of BPD patients with assessments of both cognitive and functional status. We excluded reviews, abstracts, letters, editorials, books, case reports, pediatric subjects, and studies that did not evaluate both neurocognitive and functional status in the same patients. We screened reports or their abstracts for potential relevance before considering full texts for systematic review and potential inclusion.


The search strategy identified 78 reports, of which 65 were excluded by consensus as not meeting inclusion criteria or representing overlapping publications. Of the 13 remaining reports, one was excluded for not exploring correlations between cognitive and functional domains, yielding 12 studies suitable for further analysis (1, 3, 4, 23, 28–35). These 12 studies provided 13 sets of correlations between cognitive and functional status, including eight studies involving euthymic BPD patients (3, 4, 29–34) and five involving BPD patients in non-euthymic or uncertain mood states (1, 3, 23, 28, 35). The latter five studies included BPD patients either recently hospitalized (23, 28) or still in major depression, mania, or hypomania (1, 3), or in unspecified mood states (35). One of the 12 studies (3) involved longitudinal follow-up through episodes of major depression, hypomania, and euthymia, and thus provided data involving both euthymia and non-euthymia. Since mood symptoms may confound cognitive and functional performance, results are reported separately for the euthymic and non-euthymic BPD patients.

Demographic and clinical variables

In the eight studies of euthymic patients, there were 316 BPD patients and 147 healthy controls, approximately matched in each study for sex, age, and educational levels. Euthymia was assessed as low scores on standard rating scales for mania [Young Mania Rating Scale (YMRS) (36) total scores currently at 5–10, or less] and depression [Hamilton Depression Rating Scale (HDRS) (37) scores of 5–10, or less], as well as sustained clinical stability for 1–6 months or longer (Table 1A). Of these BPD patients (type I or both types I and II), 51% were women, age averaged 45 ± 1.3 years, and education 13.3 ± 0.7 years (weighted by study sample size; Table 1A). Age at illness onset averaged 26.2 ± 1.7 years, with 15.7 ± 1.0 total years of illness with multiple recurrences, and variable rates of psychotic features, hospitalizations, and suicide attempts (Table 1A). Most patients were being prescribed one or more psychotropic medicines, usually including a mood stabilizer (typically, carbamazepine, lamotrigine, lithium, or valproate), with or without an antipsychotic, antidepressant, benzodiazepine, or other sedative (Table 1A).

Table 1.   Characteristics of adult patients with bipolar disorder (BPD) in 13 correlations of cognitive performance versus functional status
StudynBPD typesAge (years)% MenEducation (years) Clinical featuresPsychotropic treatmentsEuthymia criteria or mood state
  1. NP = not provided; FH = family history; Rxs = treatments; MS = mood stabilizer [lithium (Li) or anticonvulsant (AC)]; PANSS = Positive & Negative Syndrome Scale; AD = antidepressant; HDRS = Hamilton Depression Rating Scale; YMRS = Young Mania Rating Scale; HC = healthy controls; CBZ = carbamazepine; VPA = valproate; BP-I = bipolar I disorder; sed = sedative; Dx = diagnosis; anti-ACh = anticholinergic; AP = antipsychotic drug; NOS = not otherwise specified; MDE = major depressive episode.

  2. aPANSS mania = 0–5; depression = 0–9.

  3. bLongitudinal study of 25 BPD patients followed through depressive, hypomanic, and euthymic phases; 15 patients were assessed during euthymia.

  4. cParticipants had cognitive testing when ‘sufficiently stable’ after hospitalization; at one-year follow-up they were rated for functional status [Multidimensional Scale for Independent Living (40)].

A. Euthymic adult BPD patients
Atre-Vaidya et al. 1998 (29)36 BPDNP47 ± 177513.4 ± 2.2Onset: 29 ± 15 years
50% no psychiatric FH
39% had FH of major mental illness
Rxs: 17% Li, 17% other MSs, 25% MS + APD, 42% multiple agentsPANSSa mania: 0.4 ± 1.1 and depression: 1.4 ± 1.8, ≥6 weeks
Martínez-Arán et al. 2002 (30)49 BPDNP38 ± 104113.0 ± 3.4Onset: 24 ± 7 years
Time ill: 13 ± 9 years
First hospitalization: 28 ± 9 years Hospitalizations: 2.3 ± 2.0
Suicide attempts: 0.9 ± 1.6
Rxs: 92% MS (27%≥2), 16% MS + AD, 60% MS+APD; 8% noneHDRS ≤8 and YMRS ≤6, ≥6 months
Martínez-Arán et al. 2004 (4)40 BPD; 30 HCNP38 ± 104213.0 ± 3.5Onset: 24 ± 7 years
2.6 ± 2.6 hospitalizations
Time ill: 15 ± 8.4 years
Episodes: 13.6 ± 11.1
Rxs: 83% Li, 30% CBZ, 18% VPA, 28%≥2 MSs, 58% MS + APD, 20% MS + AD, 43% MS + sedHDRS ≤8 and YMRS ≤6, ≥6 months
Dittmann et al. 2007 (31)55 BPD; 17 HCI, II42 ± 134711.7 ± 1.7Onset: 26 ± 10
Number of episodes: 11.6 ± 15.6
Time ill: 17 ± 11 years
In remission 18.3 ± 36.8 months
Rxs: 91% MSs (22% Li, 26% AC, 20%≥2 MSs), 38% AP + MS, 16% ADs; 9% none
HDRS ≤5 and YMRS ≤5, ≥1 month
Malhi et al. 2007 (3)b15 BPD; 25 HCI39 ± 11NP14.7 ± 2.9Onset: 24 ± 8 years
Time ill: 15 ± 11 years
11.5 ± 8.2 depressive and 8.2 ± 7.2 manic episodes/ person
Rxs: 84%≥1 MS (no AD, APD ≥ 2 weeks); 16% noneHDRS ≤9 and YMRS <10, ≥1 month
Martínez-Arán et al. 2007 (32)77 BPD; 35 HCI, II39 ± 103913.0 ± 3.580% BP-I
Onset 24 ± 8 years
Time ill: 15 ± 8 years
11.4 ± 10.8 episodes and 2.5 ± 2.6 hospitalizations/ person
73% prior psychosis
Rxs: 80% Li, 27% AC, 28% AD, 52% APD; low-functioning: 2.6 ± 1.1, high-functioning: 1.7 ± 1.1 meds/personHDRS ≤8 and YMRS ≤6, ≥6 mo
Gildengers et al. 2007 (33)20 BPD; 40 HCI, II74 ± 84015.7 ± 3.4Illness history: NPRxs: 6 Li,11 ACs ± adjuncts (AD, modern APDs, seds); 3 noneHDRS ≤10 and YMRS ≤10 and euthymic, ≥4 weeks
Altshuler et al. 2008 (34)14 BPDI48 ± 2010016.3 ± 2.6Onset: 28 ± 14 years
Time ill: 21 ± 15 years
28% prior psychosis
No other Axis-I Dxs
Rxs: 100% Li, 14% CBZ, 28% APDs, 29% ADs, 29% seds, 14% anti-AChHDRS <7 and YMRS <6, ≥3 months
Total or Mean316 BPD; 147 HCI, II45 ± 1.34913.3 ± 0.7Onset: 26.2 ± 1.7 years
Time ill: 15.7 ± 1.0 years
100% MS ± (AD, AP or sed)Varied but euthymic
B. Adult BPD patients in non-euthymic or unspecified mood states
Dickerson et al. 2004 (35)117 BPDI, II41 ± 123014.3 ± 2.9Onset: 20 ± 9 years
Time ill: 21 ± 12 years
84% had prior hospitalization
Mean time to prior hospitalization 49.7 ± 69.8 months
Rxs: 28% Li, 88% MS (include Li), 57% APD, 63% AD, 5% anti-AChMood state: NP
Martínez-Arán et al. 2004 (1)64 BPD; 30 HCI, II43 ± 115011.4 ± 3.2Onset: 26 ± 10
Time ill: 16 ± 11 years
Number of total episodes: 16.1 ± 17.6
2.1 ± 2.7 hospitalizations
Rxs: 73% Li, 28% AC, 55% APD, 34% ADDepression: HDRS ≥17 Mania or hypomania: YMRS ≥12
Jaeger et al. 2007 (28)c78 BPDI, II, NOS36 ± 103414.1 ± 2.3Onset: 19 ± 8 years
5.6 ± 4.5 hospitalizations
Age of first treatment: 21.6 ± 8.2 years
NP‘Sufficiently stable’ after hospitalization. At one-year follow-up: MDE, mania, subsyndromal symptoms, euthymia
Altshuler et al. 2007 (23)213 BPDI, II43 ± 79194% had ≥12Onset: 20 ± 8 years
55% had psychosis at index hospitalization
42% had > 10 prior hospitalizations 5.2 ± 4.9 hospitalizations in past 5 years
73% had previous suicide attempts
Rxs: 71% taking ≤4 psychotropic meds; 28% taking >4 psychotropic meds; total number of psychotropic meds at admission: 7.3 ± 1.6Near discharge from hospitalization
Malhi et al. 2007 (3)25 BPD; 25 HCI39 ± 11NP14.7 ± 2.9Onset: 24 ± 8 years
Time ill: 15 ± 11 years
11.5 ± 8.2 depressive and 8.2 ± 7.2 manic episodes/person
Rxs: 84%≥1 MS (no AD, AP ≥2 weeks); 16% noneDepressed (HDRS >16) or hypomanic (YMRS >10)
Total or Mean497 BPD; 55 HCI, II, NOS41 ± 961≥ 12.9 ± 2.8Onset: 20.8 ± 8.5 yearsMS ± (AD, AP or seds)Non-euthymic or uncertain mood state

In the five studies of non-euthymic or uncertain mood-state patients (1, 3, 23, 28, 35), there were 497 BPD patients and 55 healthy controls; 61% were men; age averaged 41 ± 9 years; education averaged at least 12.9 ± 2.8 years; and illness began at 20.8 ± 8.5 years (weighted by study sample size; Table 1B). Mood states of BPD participants were described variously as sufficiently stable for hospital discharge, involving ongoing recovery from recent major depression, mania, or hypomania, or were not specified. Most patients had type I or II BPD, and only a small proportion in one study (28) had an unspecified (not otherwise specified) type. Patients of non-euthymic or uncertain mood states were prescribed psychotropic medications similar to those prescribed to euthymic subjects (Table 1A,B).

Neurocognitive performance

Neurocognitive performance of BPD patients and healthy controls was assessed by various combinations of standard neuropsychological tests (Table 2). Specifically, in studies of euthymic BPD patients (Table 2A), estimated premorbid IQ was assessed in 6/8 studies (3, 4, 29–32), executive function in six (3, 4, 30, 32–34), verbal learning and memory in six (3, 4, 29, 32–34), attention, concentration, mental tracking, information processing speed, and immediate memory in five (3, 4, 30, 31, 33), verbal fluency or language in four (4, 30, 31, 33), visuospatial performance in three (29, 31, 33), and psychomotor speed in two studies (3, 31). Euthymic BPD patients performed less well than matched healthy controls in these cognitive measures in 7/8 studies. Their impaired domains involved verbal learning and memory in four studies (4, 29, 32, 34), executive functioning in three (4, 33, 34), and attention, concentration, mental tracking, and information processing speed in three (31–33).

Table 2.   Cognitive performance versus functional status in bipolar disorder (BPD) patients
StudyCognitive domains (tests)Functional assessmentAssociated with functioning
  1. Cognitive assessment: WAIS-R = Wechsler Adult Intelligence Scale Revised (74); COWAT = Controlled Oral Word Association Test (75); CVLT = California Verbal Learning Test (76); WCST = Wisconsin Card Sorting Test (77); TMT = Trail Making Test (78); FAS = letter fluency test (75); SCWT = Stroop Color-Word Interference Test (79); RBAN = Repeatable Battery for the Assessment of Neuropsychological Status Form A (80); LNST = Letter-Number Sequencing Test (81); NART = National Adult Reading Test (82); SOC =Stockings of Cambridge (83); RAVLT = Rey Auditory Verbal Learning Test (84); EXIT = Executive Interview (85); SCOLP = Speed and Capacity of Language Processing (86); WMS-III = Wechsler Memory Scale-III (86); WMS-R = Wechsler Memory Scale Revised (87).

  2. Functional assessment: IRS = Impairment Rating Scale (29); SSIM = Structured and Scaled Interview for Maladjustment (88); GAF = Global Assessment of Functioning Scale (38); SAS = Social Adjustment Scale (39); IADL = Instrumental Activities of Daily Living; MSIF = Multidimensional Scale for Independent Living (40).

  3. aIRS (with some mood symptoms) or SSIM used in half of the subjects.

  4. bThirty-month study through euthymia, depression, hypomania; 15 assessed when euthymic.

A. Euthymic BPD patients
Atre-Vaidya et al. 1998 (29)aGeneral intelligence (WAIS-R vocabulary subtest)
Verbal fluency (COWAT)
Verbal memory & learning (CVLT)
Visuospatial ability (Modified Raven’s Matrices)
IRS, SSIMVerbal memory
Martínez-Arán et al. 2002 (30)Estimated premorbid IQ (WAIS vocabulary subtest)
Attention & immediate memory (WAIS digit subtest)
Executive function (WCST; TMT-A, B)
Verbal fluency (FAS test)
GAFWeakly with verbal fluency
Martínez-Arán et al. 2004 (4)Estimated premorbid IQ (WAIS vocabulary subtest)
Executive functions (WCST, SCWT, FAS)
Attention, concentration, & mental tracking (WAIS digit subtest; TMT-A, B)
Verbal learning & memory (CVLT)
GAFVerbal learning & memory
Dittmann et al. 2007 (31)Estimated premorbid IQ (WAIS-R vocabulary subtest)
Immediate memory, visuospatial/constructional abilities, language, attention, delayed memory (RBAN battery)
Psychomotor speed and set (TMT-A, B)
Information & working memory (WAIS-III LNST subtest)
SASWorking memory
Malhi et al. 2007 (3)bEstimated premorbid IQ (NART)
Executive functions (WCST, SOC, TMT-B, COWAT, animal naming)
Attention, concentration, mental tracking (SCWT, TMT-A, WAIS-III digit-span & digit-symbol)
Verbal learning and memory (RAVLT)
Psychomotor function (reaction time, Purdue, grooved pegboard)
Martínez-Arán et al. 2007 (32)Estimated premorbid IQ (WAIS vocabulary subtest)
Executive functions (WCST, SCWT, FAS)
Attention/concentration/mental tracking (WAIS digit subtest, TMT)
Verbal learning and memory (CVLT)
GAFVerbal memory
Executive functioning
Gildengers et al. 2007 (33)Information processing speed (WAIS-III digit-symbol, grooved pegboard, TMT-A)
Visuospatial ability (simple drawings, WAIS-III block design subtest, clock drawing, modified Rey-Österreich)
Executive functioning (EXIT interview, TMT-B, SCWT, WCST)
Language (Boston naming test, FAS, category fluency, spot-the-word from SCOLP)
Memory (WMS-III logical memory, modified Rey-Österreich, CVLT)
IADLInformation-processing speed
Executive functioning
Altshuler et al. 2008 (34)Verbal declarative memory (CVLT)
Executive functioning (WCST)
GAFVerbal memory
Executive functioning
B. BPD patients who were non-euthymic or in uncertain mood states
Dickerson et al. 2004 (35)Immediate memory, visuospatial/constructional abilities, language, attention, delayed memory (RBAN battery)
Working memory (WAIS-III LNST subtest)
Concentration and processing speed (TMT-A)
Employment statusImmediate memory
Martínez-Arán et al. 2004 (1)Estimated premorbid IQ (WAIS vocabulary subtest)
Executive function (WCST, SCWT, COWAT, animal naming)
Attention, concentration, mental tracking (WAIS digit subtest, TMT)
Verbal learning & memory (CVLT, WMS-R logical memory subtest)
Nonverbal learning & memory (WMS-R visual reproduction subtest)
GAFVerbal learning & memory
Jaeger et al. 2007 (28)Attention (SCWT, TMT-A, WAIS-R digit-symbol)
Working memory (WAIS-R digit-span & digit-symbol, LNS, arithmetic)
Ideational fluency (COWAT, Ruff-unique designs, animal naming)
Verbal knowledge (WAIS-R vocabulary, comprehension, similarities)
Nonverbal functions (WAIS-R block design, object assembly, picture completion, picture arrangement)
Learning (WMS-R ver pair I, II; vis pair I, II)
MSIFAttention Ideational fluency
Altshuler et al. 2007 (23)Executive function (15-min structured interview using 22-item Executive Interview)Employment statusExecutive function
Malhi et al. 2007 (3)Estimated premorbid IQ (NART)
Executive functions (WCST, SOC, TMT-B, COWAT, animal naming)
Attention, concentration, mental tracking (SCWT, TMT-A, WAIS-III digit-span & digit-symbol)
Verbal learning and memory (RAVLT)
Psychomotor function (reaction time, Purdue, grooved pegboard)
GAFIn depressed patients: attention, mental tracking, reaction time
In hypomanic patients:
 working memory, verbal learning

In studies involving non-euthymia or uncertain mood states (Table 2B), estimated premorbid IQ was assessed in only 2/5 studies (1, 3), executive functioning in three (1, 3, 23), verbal learning and memory in three (1, 3, 28), and attention, concentration, mental tracking, information processing speed, and immediate memory in four studies (1, 3, 28, 35). One study used a 15-minute bedside interview instead of a standardized neuropsychological test to assess executive functioning (23). Only two studies reported results of cognitive performance separately from the association between cognitive and functional performance. In them, depressed and manic or hypomanic BPD patients, compared to healthy controls, exhibited cognitive impairment in verbal memory and executive function (1, 3), or attention, concentration, and mental tracking (3).

Psychosocial functioning

To assess functional status in euthymic BPD patients, 5/8 studies (3, 4, 30, 32, 34) used the Global Assessment of Functioning (GAF) Scale (38); three of these (4, 32, 34) reported explicitly that only social and occupational functioning, and not mood symptoms, were considered, two relied on patient self-assessment on psychosocial functioning rating scales (29, 31), and only one used performance-based measures to assess the instrumental activities of daily living [IADL (33)]. Functional domains considered were vocational and educational activities, household maintenance, level of independent living, and quality of family and social relationships. Of studies reporting functional status separately from its correlation with cognitive status, mean GAF scores for euthymic BPD patients ranged from 68 to 86 (3, 4, 30) for a pooled average of 74.6 ± 11.4, or 19% lower than the value of 92 ± 2.9 in healthy controls (3). In one study where the Social Adjustment Scale (39) was used, 61% of euthymic BPD patients were considered as having good-to-very-good psychosocial functioning, 25% moderate, and 14% poor-to-very-poor functioning, compared to good-to-very-good psychosocial functioning in 100% of healthy controls (31).

In the five studies of non-euthymic BPD patients, employment status (23, 35), GAF scores (1, 3), or the Multidimensional Scale for Independent Living (28, 40) were the basis of functional assessments. Among the BPD patients, 51%–57% were unemployed, 21% working part time or volunteering, and 27%–43% were employed full time (23, 35).

Association between cognitive and functional performance

Among euthymic BPD patients, 6/8 studies found neurocognitive impairment to be significantly associated with lower psychosocial functioning, usually after adjusting for demographic and salient clinical variables, including residual depressive symptoms (4, 29, 31–34; Table 2A). Specific neurocognitive domains found to be associated with functional status included verbal memory (4, 29, 32, 34), executive functions (32–34), and attention, concentration, working memory, and information processing speed (31, 33). Of two studies that did not find significant correlations of cognitive measures and functional status, one found a weak association between verbal fluency and functional outcomes (30) and the other found no notable relationships between cognitive and functional status (3). The information provided in the available reports is not adequate to support a quantitative or meta-analytic summary of the findings.

Among BPD patients in uncertain or non-euthymic states, all five studies also found a significant correlation between cognitive performance and functional status, after statistically controlling for some relevant clinical factors including mood symptoms (1, 3, 23, 28, 35). The cognitive domains found to be associated with functional status included executive function (23), verbal learning and memory (1, 3), attention, concentration, immediate memory, working memory, and mental tracking (3, 28, 35).


A significant correlation between impaired neurocognitive performance and lower psychosocial functioning was found in euthymic BPD patients as well as those in uncertain mood states, usually after adjusting for relevant demographic and clinical variables, in 85% of the 13 studies reviewed. The two studies of euthymic BPD patients that did not find such association should be interpreted in light of their limitations. One involved only 15 euthymic BPD patients and so had low statistical power (3). The other included cognitive abilities such as attention and psychomotor speed within a complex group of ‘residual mood symptoms,’ which correlated significantly with functional outcomes but might be vulnerable to treatment effects, whereas verbal fluency was only weakly associated with functional status (30). One relatively large study involving 213 non-euthymic BPD patients used a 15-minute bedside interview rather than standardized neuropsychological methods to assess executive functions, raising a question of accuracy of the findings (23). Remarkably, unemployment rates averaged approximately 55% in BPD patients, despite more than 13 years of education, underscoring the striking occupational impairment in the majority of the BPD patients studied (23, 35).

Influence of mood symptoms on cognitive performance

Slightly different cutoff scores for depression and mania rating scales were used to define euthymia in the eight studies involving euthymic BPD patients; higher scores may allow residual affective morbidity to confound the observed associations of cognitive and functional measures. Nevertheless, five of these eight studies reported to have addressed this issue by statistically adjusting for residual affective symptom scores (3, 4, 30–32). Notably, the 12 reported studies of cognitive versus functional performance in BPD patients focused on the mood state of euthymia, major depression, mania, or hypomania, but not subsyndromal depression. As a phase, subsyndromal depression accounts for 50%–73% of the total weeks in depressive morbidity (41, 42), which in turn accounts for 68%–82% of the total weeks ill among BPD patients being treated by current standards, even early in the illness course (41–44).

Assessment of cognitive performance in BPD patients

Various objective and quantifiable, performance-based neuropsychological tests were used to assess cognitive performance in all except one (23) of the reviewed studies. However, it remains unclear which neuropsychological tests are most sensitive to cognitive impairment and reliably relate to functional performance in BPD patients, in contrast to far more extensive studies of patients diagnosed with schizophrenia. Notably, Green and colleagues (45) reviewed 37 studies examining relationships between various neurocognitive measures and indications of functional status in schizophrenia patients. They found that verbal learning and memory (in 13/18 studies), executive functioning (11/22 studies), attention, concentration, or mental tracking (9/17 studies), and immediate memory (5/7 studies) were particularly strongly or consistently associated with functional outcomes. Functioning was assessed as independent living, medication management, social competence, social problem solving, and occupational functioning (45). Additionally, the National Institute of Mental Health’s Measurement and Treatment Research to Improve Cognition in Schizophrenia initiative has developed a consensus cognitive battery for standardized evaluation of cognitive changes in schizophrenia patients, based on validity, test-retest reliability, practice effects, relationship to functional status, practicality, and tolerability (46). Similar studies are needed to identify cognitive constructs most related to functional abilities in BPD patients to identify targets for potential remediation.

Assessment of functional performance in BPD patients

Of the 12 studies reviewed, nine assessed functional performance by relying on either the GAF scale or patient self-assessment on functional rating scales. Only three studies used objective data, such as employment status, or performance-based measures of skills, such as the IADL scale. The GAF is a relatively crude clinical scale, and is essentially based on patients’ reports. Subjective self-appraisal of performance levels is liable to bias and inaccuracy, as demonstrated in studies of patients with schizophrenia or BPD, as well as in healthy subjects (47–50). For instance, scores of patients’ self-assessment of their functioning on the World Health Organization Disability Assessment Schedule correlated poorly with performance-based measures of functional capacity in a study of schizophrenia patients (49). Moreover, scores on a self-report scale of cognitive impairment correlated poorly with objectively tested cognitive deficits in a study of BPD patients (50). Such findings support the conclusion that most of the studies reviewed are limited by potential inaccuracy in their findings of functional status.

Furthermore, in detailed assessment of functional performance in mentally ill persons, there is an important distinction between functional capacity (competence) and functional outcomes (actual performance). Performance can be influenced by intrinsic capability (capacity), affective morbidity, adverse medication effects, or other environmental factors. For instance, a depressed BPD patient may have the capacity to pay bills (full capacity) but lack motivation to attend to the task, or ignore it due to excessively sedating medication (impaired performance), so as to yield discrepancies between functional capacity and functional performance. Objective evaluation of functional status with standardized, performance-based measures in a controlled setting is more likely to yield valid data on functional capacity. Employment, education, and residential and marital status, among others, can provide more objective measures of real-world functional performance. Although the influence of cognition and symptomatic morbidity on objectively measured functional capacity and real-world functional outcomes has been studied in schizophrenia patients (17, 18), such relationships have not been reported in BPD patients. Therefore, it remains unknown whether there are significant discrepancies between functional capacity and real-world functioning in BPD patients. If significant discrepancies are observed, efforts can be made to identify potential barriers to successful applications of skills that exist in a patient’s behavioral repertory, with the overall goal of improving functional recovery in BPD patients.

Studies using performance-based functional assessment instruments in BPD patients, except for the IADL, have not been reported. Performance-based instruments for assessing self-care, independent living, and social and occupational skills have been designed and studied in schizophrenia and HIV-infected patients. Among these are the University of California Performance-Based Skills Assessment (such as comprehension and planning, transportation and mobility subtests) (51), Everyday Functioning Battery (including medication management and advanced financial skills subtests) (52), Social Skills Performance Assessment (53, 54), and computerized COMPASS vocational assessment batteries (53). BPD patients share with schizophrenia and HIV-infected patients impairments in the domains of independent living, interpersonal relationships, and vocational success, though with less severity, as reviewed above. Whether such performance-based instruments are sensitive to the functional impairments of BPD patients, specifically, and whether ‘floor’ and ‘ceiling’ (too easy versus too difficult) effects pose a problem for these tests in BPD patients remain to be determined.

BPD course-of-illness factors and neurocognition

Neurocognition is influenced by many factors, including age, premorbid IQ, education, course-of-illness or chronicity factors, mood state, substance use, and adverse medication effects. The influence of onset age, illness duration, number of episodes, and time in euthymia on cognitive impairment in BPD patients was examined in a recent review by Robinson and Ferrier (55). A greater number of manic episodes was associated with greater deficiencies in verbal memory and executive function in 7/10 studies ( =  –0.42 to –0.90), and more depressive episodes were associated, though less robustly, with reduced performance in measures of executive function, verbal learning, and visual and spatial memory in 6/10 studies ( =  –0.43 to –0.63) (55). A greater total affective episode count was associated with impaired cognitive function in only 1/4 studies; younger age at onset of BPD was a risk factor for cognitive impairment in only 1/6 studies, more years with the illness correlated with reduced cognitive performance in only 5/11 studies, and time in remission was unrelated to cognition (55). Taken together, these findings suggest that recurrent mania may be a particularly important risk factor for cognitive and functional impairment (55). In addition, 3/4 studies found that a history of psychotic features in BPD was associated with impaired executive function and verbal memory (56–59).

Pathophysiology of cognitive impairment

Our additional systematic literature search using the databases identified above found only seven studies investigating the neurochemistry (n = 5), neuroanatomy (n = 1), and neurophysiology (n = 1) of cognitive impairment in euthymic BPD patients. Variously, elevated regional cerebral monoaminergic concentrations (60), abnormal glucocorticoid receptor function (61), and elevated serum homocysteine concentrations (31, 62) were associated with impaired cognitive impairment in such patients. One small study of 14 BPD patients found very little effect on cognition with acute reduction of L-tryptophan intake, which can lower cerebral serotonin levels and induce depression (63). Another study found that magnetic resonance imaging-measured white matter lesions did not differ between euthymic BPD patients with cognitive impairment and healthy matched controls (64). Finally, electroencephalographic abnormalities in euthymic BPD patients include increased power in all wave bands, particularly in the cerebral areas associated with impairment of visuospatial processing (65).

Psychopharmacologic effects on neurocognition

The impact of psychotropic drugs on cognitive impairment in euthymic BPD patients, particularly with growing reliance on combinations of psychotropic medicines (66), remains a complex and challenging issue. Of five studies (67–71) evaluating effects of medication on cognition of euthymic BPD patients, three found that lithium adversely affected cognitive performance (67, 68, 70), one reported that lithium and carbamazepine had no adverse effects on cognition (69), and another found lithium and valproate to have equivocal effects on cognition (71). More studies are needed to elucidate the extent of adverse psychopharmacologic effects on cognition of euthymic BPD patients as a contribution to assessing risk/benefit relationships.

Potential pharmacologic treatments for cognitive impairment in euthymic BPD patients remain to be studied. One potential lead is that mifepristone (RU-486), a corticosteroid receptor antagonist, was associated with 20% improvement in spatial working memory, with little change in ratings of depression, in a small, double-blind, crossover study in depressed BPD patients (72). Also of potential interest, a randomized, double-blind, placebo-controlled, crossover study reduced serum homocysteine concentrations in schizophrenia patients with oral folic acid, vitamin B12, and pyridoxine and found improvements in executive functioning as well as possibly confounding symptomatic status (73). These highly preliminary observations suggest that selective beneficial effect on some aspects of cognition, independent of mood-altering actions in BPD, may be feasible.


In the limited available research literature regarding BPD patients, particularly those in euthymia, cognitive impairment measures by reliable methods have been associated with impaired functional status, which usually was assessed by subjective self-appraisal rather than with objective and quantitative measures. Better studies are needed to assess relationships between cognitive impairment during verified euthymia and objectively measured, performance-based functional status to support efforts to improve the currently low functional recovery rates in treated BPD patients.


Supported in part by a Research Fellowship award from the American Psychiatric Institute of Research and Education (to APW), and by a grant from the Bruce J. Anderson Foundation and by the McLean Private Donors Research Fund (to RJB).