7/29/2019 Depress Anxiety 2006 23 496-501

1/6

DEPRESSION AND ANXIETY 23:496501 (2006)

Brief Report

PROBLEM-SOLVING ABILITY AND COMORBID

PERSONALITY DISORDERS IN DEPRESSED OUTPATIENTS

Rebecca Harley, Ph.D., Timothy Petersen, Ph.D., Margaret Scalia, B.A., George I. Papakostas, M.D.,Amy Farabaugh, Ph.D., and Maurizio Fava, M.D.

Major depressive disorder (MDD) is associated with poor problem-solvingabilities. In addition, certain personality disorders (PDs) that are commonamong patients with MDD are also associated with limited problem-solvingskills. Attempts to understand the relationship between PDs and problem solvingcan be complicated by the presence of acute MDD. Our objective in this study was

to investigate the relationships between PDs, problem-solving skills, andresponse to treatment among outpatients with MDD. We enrolled 312outpatients with MDD in an open, fixed-dose, 8-week fluoxetine trial. PDdiagnoses were ascertained via structured clinical interview before and afterfluoxetine treatment. Subjects completed the Problem-Solving Inventory (PSI)at both time points. We used analyses of covariance (ANCOVAs) to assessrelationships between PD diagnoses and PSI scores prior to treatment. Subjectswere divided into three groups: those with PD diagnoses that remained stableafter fluoxetine treatment (N591), those who no longer met PD criteria afterfluoxetine treatment (N5 119), and those who did not meet criteria for a PD atany time point in the study (N5 95). We used multiple v2 analyses to comparerates of MDD response and remission between the three PD groups. ANCOVAwas also used to compare posttreatment PSI scores between PD groups. Prior tofluoxetine treatment, patients with avoidant, dependent, narcissistic, andborderline PDs reported significantly worse problem-solving ability than didpatients without any PDs. Only subjects with dependent PD remained associatedwith poorer baseline problem-solving reports after the effects of baselinedepression severity were controlled. Patients with stable PD diagnoses hadsignificantly lower rates of MDD remission. Across PD groups, problem solvingimproved as MDD improved. No significant differences in posttreatmentproblem-solving were found between PD groups after controlling for baselinedepression severity, baseline PSI score, and response to treatment. Treatmentwith fluoxetine is less likely to lead to remission of MDD in patients with stablePDs. More study is needed to investigate causal links between PDs, problemsolving, and MDD treatment response. Depression and Anxiety 23:496501,

2006. & 2006 Wiley-Liss, Inc.

INTRODUCTIONThere is good evidence to suggest that majordepressive disorder (MDD) is associated with poorproblem-solving ability. Individuals with depressionshow deficits in their abilities to generate alternativesand effectively resolve interpersonal problems[Goddard et al., 1996; Watkins and Baracaia, 2002].

Published online 14 July 2006 in Wiley InterScience (www.

interscience.wiley.com).

DOI 10.1002/da.20194

Received for publication 10 August 2005; Revised 28 September

2005; Accepted 8 February 2006

Correspondence to: Rebecca Harley, Ph.D., Massachusetts

General Hospital, 15 Parkman Street, WAC 812, Boston, MA

02114. E-mail: rharley@partners.org

Depression Clinical and Research Program, Massachusetts

General Hospital, Boston, Massachusetts

2006 Wiley-Liss, Inc.

7/29/2019 Depress Anxiety 2006 23 496-501

2/6

In addition to demonstrating deficits on objectivemeasures of problem-solving ability, patients withdepression rate themselves on self-report measures asless effective problem solvers than do their nonde-pressed counterparts [Heppner et al., 1985]. Ineffective

problem solving may directly predict depression,hopelessness, and suicidal ideation [Cannon et al.,1999; Dixon et al., 1991; Nezu and Ronan, 1985].

Alternatively, or in addition, problem solving may serveas a moderator between negative life stress anddepressed mood [Priester and Clum, 1993].

The frequent comorbidity of personality disorders(PDs) is another widely supported finding amongindividuals with major mood disorders. Though rateshave varied somewhat across studies, review of theliterature suggests that 5085% of outpatients withcurrent MDD also meet criteria for at least onecomorbid personality disorder diagnosis [Corrubleet al., 1996]. Cluster C diagnoses appear to be most

common among depressed outpatients [Fabrega et al.,1991; Fava et al., 2002]. Notably, the study of PDs maybe complicated by the presence of acute MDD, becausemood state can influence self-description of personalitytraits [Klein et al., 2002]. For some patients withdepression, PD diagnoses might be more accuratelydescribed as behavioral and cognitive manifestationsof MDD that resolve with antidepressant treatment[Fava et al., 1994]. A previous study by our groupsupported this possibility when we found that a subsetof depressed patients diagnosed with PDs at baselineno longer met PD criteria after their depressivesymptoms were treated [Fava et al., 2002].

It has been hypothesized that certain PDs areassociated with problem-solving deficits not attributa-ble to depressed mood alone. For example, Linehan[1993] has proposed that deficits in the skills setrequired for effective emotion regulation are a centralproblem for patients with borderline PD. Dialecticalbehavior therapy (DBT) teaches mindfulness, inter-personal effectiveness, emotion regulation, and distresstolerance skills to address these deficits. Similarly,effective social problem-solving and interpersonal skillsare thought to be lacking among patients withdependent PD [Huprich et al., 2004], leading tocognitive-behavioral models that incorporate pro-blem-solving training into their treatment protocols

[e.g., Overholser and Fine, 1994].Whereas PDs are known to be common among

outpatients with MDD, and poor problem-solvingskills are observable among both patients with depres-sion and those with PDs, little is known about thecomplex relationship between problem solving, PDs,and depression. Our aim in this study was to examinethese associations in a large sample of depressedoutpatients. In light of the results of our previousstudy, in which some patients no longer met PDcriteria after treatment of MDD, another aim of ourstudy was to examine potential differences in problem-solving ability, MDD treatment response, and MDD

remission between patients whose PD diagnoses did ordid not remain stable after antidepressant treatment.

Our hypotheses were as follows:

1. Outpatients with MDD who have a PD diagnosis

will describe themselves as poorer problem solversthan will patients without PD diagnoses at baseline.

2. Patients with stable PD diagnoses (i.e., patients whoreceived the same PD diagnoses before and after

MDD treatment) will have significantly lowerresponse and remission rates of MDD after treat-ment with fluoxetine than will patients withoutstable PD diagnoses.

3. Patients with stable PDs will report significantlypoorer problem-solving skills after fluoxetine treat-ment of MDD than will patients without stable PDdiagnoses.

METHODSSample: Our initial sample was drawn from 384

(55% female) outpatients (ages 1865) who met criteriafor a current major depressive episode (MDE) accord-ing to the Structured Clinical Interview for DSM-III-RPatient edition [SCID-P; Spitzer et al., 1989].Patients entering the study were medication-free forat least 2 weeks, with a baseline 17-item HamiltonDepression Rating Scale [HAM-D-17; Hamilton,1960] scoreZ16. Participants were enrolled at the

Massachusetts General Hospital (MGH) DepressionClinical and Research Program (DCRP) in an 8-week,

fixed-dose, open-label trial of 20 mg/day fluoxetine.The following criteria were exclusionary: pregnancyor lactation; serious risk for suicide; history ofneurological illness, including seizure disorder; seriousor unstable medical illness, organic mental disorders;psychosis; active alcohol or substance use disorders;bipolar disorder; antisocial PD; adverse drug reactionsor allergy to study drugs; current use of otherpsychotropic drugs; current depression-specific psy-chotherapy; or evidence of hypothyroidism.

Procedure: At the baseline visit, participants signeda written informed consent approved by the InstitutionalReview Board of MGH. In addition to the SCID-P andHAM-D-17 interviews, the Structured Clinical Inter-

view for DSM-III-R Personality Disorders [SCID-II,Spitzer et al., 1990] was also conducted. The SCID-IIis a structured, diagnostic, clinician-rated interview

yielding PD diagnoses. Patients also completed theProblem-Solving Inventory [PSI; Heppner and Petersen,1982], a 35-item, Likert-type self-report measure ofperceived problem-solving ability. Higher scores in-dicate more dysfunctional problem-solving behaviorand attitudes. Testretest reliabilities have been foundto be .89 for the total scaled score, and scores have beenfound to be unrelated to social desirability [Heppnerand Petersen, 1982]. Clinician-rated instruments at thebaseline and subsequent visits were conducted by

497Brief Report: Problem Solving, Personality Disorders, and Depression

Depression and Anxiety DOI 10.1002/da

7/29/2019 Depress Anxiety 2006 23 496-501

3/6

experienced psychologists or psychiatrists affiliatedwith the MGH DCRP. In our group, training in theuse of these instruments is conducted via peer review of

videotaped interviews. Our interrater reliability for theuse of the SCID-P was recently estimated as k5 .80

[Fava et al., 2002].During 8 weeks of treatment with 20mg/dayfluoxetine, patients were seen every 2 weeks for safetyand efficacy assessments. At the end of this acute phaseof treatment, patients were readministered the SCID-Pmood disorder module, the SCID-II, the HAM-D-17,and the PSI. A treatment responder was defined ashaving a 50% or greater reduction in HAM-D-17 totalscore from baseline to end point. Remission wasdefined as a total HAM-D-17 score of 7 or less at endpoint.

Of the 384 subjects who entered the study, 312completed the HAM-D-17, PSI, and SCID-II at bothbaseline and the end of acute treatment. These patients

comprise the sample for this study.Statistical Analyses: We conducted analyses of

covariance (ANCOVAs) to assess the relationshipbetween PD diagnoses and PSI scores prior tofluoxetine treatment. To investigate differences be-tween patients whose PD diagnoses endured after acute

MDD treatment, patients who met PD criteria atbaseline, but not after acute MDD treatment, andpatients who never met PD criteria, we then dividedthe sample into three groups: (1) The stable PD group(N5 91), which met criteria for the same PD or PDsat baseline and after fluoxetine treatment; (2) theunstable PD group (N5 119), which met criteria for

one or more PDs at baseline but did not retain thesame diagnostic profile posttreatment; and (3) the no-PD group (N5 95), which did not meet criteria forany PD at baseline and posttreatment. A few patients(N5 7) did not meet PD criteria at baseline but didreceive one or more PD diagnoses at the posttreatmentassessment. These patients were excluded from furtheranalyses. A multiple w2 analysis was conducted tocompare rates of MDD response between PD groups,controlling for the effects of age, baseline depressionseverity, and baseline problem-solving ability. Multiplew2 analysis was also used to compare rates of MDD

remission between PD groups. Finally, posttreatmentPSI scores were compared between groups using

ANOVA and ANCOVA analyses.

RESULTSThe mean baseline PSI score for patients with no PD

diagnosis was 111.3, which is consistent with mean PSIscores found in other samples of outpatients withdepression [Cannon et al., 1999] and is approximatelyone standard deviation above mean scores found innormal samples [Heppner and Petersen, 1982]. Atbaseline, our first hypothesisthat depressed out-patients who have a PD diagnosis would describethemselves as poorer problem solvers than would

patients without PD diagnosesappeared to be sup-ported (see Table 1). Patients with avoidant (Po.01),dependent (Po.01), narcissistic (Po.05), and border-

line (Po.01) PDs reported poorer problem-solvingabilities than did patients without a PD diagnosis atbaseline. These associations remained significant afteradjusting for the effects of age and gender. However,only dependent PD (F5 4.15; adjusted P valueo.05)remained associated with poorer baseline problem-solving abilities after the effects of baseline depressionseverity were also controlled.

The demographic and clinical characteristics of oursample by PD group are summarized in Table 2. There

were no differences between groups on age, gender,education, marital status, number of episodes of MDD,length of current episode, age of onset of first MDE,

or diagnosis of melancholic depression. BaselineHAM-D-17 scores differed significantly betweengroups (overall ANOVA: F5 6.217, Po.01) in thatscores for both the unstable and stable PD groups weresignificantly higher than those of the no-PD group.Patients with no PDs were significantly less likely tohave a diagnosis of atypical depression (20.9%) than

were patients with stable (46.1%) or unstable PDs(45.1%; w25 16.428, Po.01).

Table 2 also shows rates of MDD response andremission by PD group. No differences were foundbetween groups in MDD response rates. However,remission rates among the three groups were signifi-cantly different (w25 13.39, Po.01). Rates of remission

among patients with stable PDs (17.6%) were sig-nificantly lower than were remission rates for patients

with unstable (40.3%) or no PDs (36.8%). Thesedifferences remained significant after controlling forage, baseline depression severity, and baseline problem-solving ability (Po.01). Thus, our second hypothesis

was partially confirmed, in that patients with stablePDs showed no differences in rates of response to

MDD treatment but did reach remission significantlyless often than did patients with unstable or no-PDdiagnoses.

To investigate the possibility that Axis I comorbiditymight be significantly elevated among patients with

TABLE 1. Baseline PSI scores by Axis II diagnosis

M SD

No Axis II diagnosis 111.3 22.3Avoidant PD (n586) 128.2 27.2

Dependent PD (n5

37) 130.4 23.6Obsessivecompulsive PD (n590) 111.6 27.7Paranoid PD (n574) 118.7 25.2Schizotypal PD (n510) 121.9 37.4Schizoid PD (n513) 122.3 31.9Histrionic PD (n59) 110.3 14.8Narcissistic PD (n536) 123.3 24.8Borderline PD (n544) 130.3 24.9

498 Harley et al.

Depression and Anxiety DOI 10.1002/da

7/29/2019 Depress Anxiety 2006 23 496-501

4/6

stable PDs, post hoc w2 analyses were conductedcomparing Axis I diagnoses across PD groups (see

Table 3). SCID data weres available for Axis I diagnosesof dysthymia, panic disorder, agoraphobia, socialphobia, simple phobia, obsessivecompulsive disorder,generalized anxiety disorder, and somatization disor-der. Significant differences between PD groups werefound only for rates of social phobia and generalized

anxiety disorder. Specifically, the diagnosis of socialphobia was significantly more common for patientswith stable PDs (48.4%) than for patients with unstablePDs (9.5%) or no PD diagnosis (19.8%; overallw25 40.15, Po.01; stable vs. unstable: w25 18.95,

Po.01; stable vs. no PD: w25 34.48, Po.01; unstablevs. no PD: nonsignificant). In addition, patients withunstable PD diagnoses (4.2%) were significantly lesslikely to have a diagnosis of generalized anxietydisorder than were patients with a stable PD (16.5%)

or no PD diagnosis (16.4%; overall w25 8.83, P5.01;

stable vs. unstable: w25 9.36, Po.01; stable vs. no PD:nonsignificant; unstable vs. no PD: w25 7.96, Po.01).

If rates of multiple comorbid Axis I diagnoses had beenfound to be significantly higher for patients with stablePDs versus patients with unstable and no PDs, thismight suggest that degree of Axis I comorbidity wasrelated to the association we observed between reduced

MDD remission rates and stable PDs. These resultssuggest instead that whereas social phobia was sig-nificantly more common among patients with stablePDs, overall, Axis I comorbidity did not seem to besignificantly greater for the stable PD group than for

the other two groups.Our third hypothesis was that patients with stablePDs would report significantly poorer problem-solvingskills after fluoxetine treatment of MDD than wouldpatients without stable PD diagnoses. ANOVA results(F5 3.939, P5.02) indicated that patients in the stablePD group had significantly higher posttreatmentPSI scores (M5 114.7) than did patients in theno-PD group (M5 103.1), indicating poorer problemsolving. However, when baseline depression severity,baseline PSI score, and response to treatment wereincluded in the analysis as covariates, no significantdifferences on posttreatment PSI scores were foundbetween PD groups (F5 2.931; P5.06). In addition,

degree of change in PSI scores across acute phasetreatment was not significantly different betweengroups (F5 0.642, P5.53). Thus, our third hypothesis

was not supported. Notably, posttreatment PSI scoreswere significantly associated with response to anti-depressant treatment across groups (Pearsons correla-tion of post-PSI scores with change in HAM-D-17total score: Z5 3.635, Po.01).

DISCUSSIONOur sample of depressed outpatients with no PD

diagnoses reported problem-solving scores comparable

TABLE 2. Demographic and clinical characteristics

No PD Unstable PD Stable PD

Age 40.6 (9.8) 40.1 (10.3) 39.6 (11.1)% Female 64% 46% 46%

oCollege degree 41% 34% 47%% Married 53% 57% 45%Baseline HAM-D-17 18.6 (2.7) 19.9 (3.7) 20.2 (3.5)No. of prior MDEs 3.9 (7.0) 3.1 (5.0) 3.3 (4.8)Length of current episode (years) 2.2 (4.0) 4.4 (6.9) 3.4 (5.2)

Age at 1st MDE 28.3 (12.5) 25.3 (13.1) 24.2 (13.2)% Melancholic(DSM-III-R) 9% 10% 14%% Melancholic(DSM-IV) 35% 44% 34%% Atypical 20% 46% 45%% Responders 67% 63% 53%% Remitters 37% 40% 18%PSI prescores 111.3 (22.3) 113.7 (25.9) 122.1 (25.8)PSI postscores 103.1 (21.0) 107.7 (26.7) 114.7 (25.2)

TABLE 3. Current Axis I comorbidity

NoPD (%)

UnstablePD (%)

StablePD (%)

Dysthymia 19.0 22.1 31.1Panic disorder 3.4 3.2 8.8

Agoraphobia 3.4 2.1 5.5Social phobia 19.8 9.5 48.4Simple phobia 15.5 5.3 12.1Obsessivecompulsive disorder 2.6 1.1 8.8Generalized anxiety disorder 16.4 4.2 16.5

Somatization disorder 0.0 0.0 1.1

Po.01; Po.02.

499Brief Report: Problem Solving, Personality Disorders, and Depression

Depression and Anxiety DOI 10.1002/da

7/29/2019 Depress Anxiety 2006 23 496-501

5/6

to those of other depressed outpatient samples. Thesescores were approximately 1 SD higher than scorestypically reported by the normal population, indicatingthat our outpatients with depression perceived them-selves as poor problem solvers. Prior to treatment of

MDD, those outpatients with depression who metcriteria for avoidant, dependent, narcissistic, andborderline PDs reported significantly poorer pro-blem-solving ability than did depressed patients with-out PDs. These scores were approximately 2 SDsdeviations above the mean for the normal population,indicating very poor self-perceived problem-solvingabilities. However, after controlling for baselinedepression severity, only the dependent PD diagnosisremained significantly associated with poorer self-perceived problem-solving abilities.

The finding that PSI scores of patients with avoidant,narcissistic, and borderline PDs were no longersignificantly different from those of patients with no

PD diagnosis after controlling for levels of baselinedepression suggests that these patients very poorproblem-solving scores were related to the severityof their depressive symptoms. Taken alone, theseresults support the argument that poor perceivedproblem solving among patients with depressionwithor without PDsmay represent a state-dependentphenomenon rather than a trait vulnerability [Dixonet al., 1993]. However, the PSI is a measure ofproblem-solving self-appraisal. We cannot know

whether problem-solving deficits among patients withavoidant, narcissistic, and/or borderline PD wouldhave been observed to be significantly different from

the no-PD group had a measure of real-life problem-solving skills been used.In our sample of outpatients with MDD, patients

with dependent PD described themselves as poorerproblem solvers than did patients without a PDdiagnosis at baseline. This result remained significantafter controlling for demographic variables and base-line depression severity. Bornstein [1992] has proposedthat dependent personality organization comprisesfour components: motivational (i.e., a marked needfor support and guidance from others), cognitive (i.e., aself-perception of powerlessness and ineffectuality),affective (i.e., anxiety when one is required to functionautonomously), and behavioral (help- and reassurance-

seeking behaviors). This studys results support thevalidity of the cognitive component of Bornsteinsmodel.

In studies of personality pathology among patientswith major depression, the chosen method of PDmeasurement significantly affects rates and reliabilityof PD diagnoses [Zimmerman, 1994]. A review ofthe literature concluded that structured interview ispreferable to self-report [Mulder, 2002]. The use ofstructured interview to establish PD diagnoses not onlyat baseline but also after acute fluoxetine treatment is astrength of our study. Baseline and posttreatmentdiagnoses enabled us to split our large sample of

outpatients with depression into stable and unstablePD groups.

We found no statistically significant differences influoxetine response rates between PD groups. Despitethis lack of statistical significance, the observed

response rates were in the expected direction: Thestable PD group experienced lower response rates thanpatients with no or unstable PDs. It is possible that

with longer exposure (e.g., 12 instead of 8 weeks) to thestudy drug fluoxetine, differences may have reachedstatistical significance. Alternatively, it may be that thepresence of one or more personality disorders does notimpact a patients response to standard antidepressanttreatments.

The results of this study do suggest that achievementof remission may be impacted by the presence of PDs.Rates of MDD remission after acute fluoxetinetreatment were significantly lower among patients withstable PDs than among patients with unstable PDs or

no PD diagnosis. Previous studies by our group andothers [Fava et al., 1994; Fava et al., 2002; Blais et al.,1998] have suggested that, for some patients, PDdiagnoses may represent behavioral and cognitivemanifestations of depression. These symptoms appearto resolve with antidepressant treatment and do notinterfere with remission of MDD. For another groupof patients, however, it appears that PD diagnoses mayindicate behavioral, emotional, and cognitive problemsthat are more than a manifestation of MDD. Theseproblems both fail to respond to antidepressanttreatment (in that PD diagnoses remain after acute

MDD treatment) and also may interfere with full

remission of MDD symptomatology. Thus, for patientswith long-standing personality disturbances, it may bethat, in order to attain full remission, standardpsychopharmacological interventions need to be aug-mented with psychosocial treatments that targetdeficits characteristic of each PD. To fully untanglethe complex relationship between PD and MDDsymptomatology, longitudinal follow-up data are ne-cessary to track stability of PD diagnoses as MDDsymptoms fluctuate or remit over time. Our studysmeasurement of PD diagnosis prior to and immediatelyfollowing depression treatment cannot, therefore, fullyclarify the nature of these complex relationships. It is tobe hoped that future studies will include longer-term

follow-up to address this issue.After controlling for potential mediating variables

such as baseline problem-solving ability and responseto fluoxetine treatment, we found no significantdifferences on posttreatment problem-solving abilitybetween stable PD, unstable PD and no-PD groups.

Thus, our third hypothesis was not supported. Instead,our results suggest that when MDD is successfullytreated, perceived problem-solving ability improvesregardless of PD status. This finding is consistent withthat of Watkins and Baracaia [2002], who found thatsocial problem solving among formerly depressed patientsdid not differ from that of never-depressed controls.

500 Harley et al.

Depression and Anxiety DOI 10.1002/da

7/29/2019 Depress Anxiety 2006 23 496-501

6/6

Our sample comprised outpatients participating ina treatment study, and patients with comorbid substanceabuse and psychotic features were excluded. Thus, thissample may not be generalizable to other psychiatricpopulations. In addition, our study did not include a

placebo group, so we are unable to rule out regressiontoward the mean in PD diagnoses and problem-solvingscores in interpreting our results. As noted earlier,another limitation of the study involves potentialdifferences between measurement of self-appraisal ofproblem-solving skills (i.e., a cognitive variable relatedto belief in ones ability to solve problems effectively)and measurement of actual problem-solving skills level(i.e., a behavioral variable related to ones real-lifesolutions to problems). It is possible that greaterdifferences between PD groups would have beenobserved had this study used a measure of real-lifeperformance on problem-solving tasks, such as the

MeansEnds Problem-Solving Procedure [Platt and

Spivack, 1975]. Importantly, in this study, we wereunable to draw causal conclusions between problemsolving, PD status, and MDD treatment response.Future studies of depressed samples should examinethe associations between stable PDs, problem solving,and MDD treatment response via methods that candelineate cause-and-effect relationships between these

variables.

REFERENCESBlais MA, Matthews J, Schouten R, OKeefe SM, Summergrad P.

1998. Stability and predictive value of self-report personality traitspre-and post-electroconvulsive therapy: A preliminary study.

Compr Psychiatry 39:231235.Bornstein RF. 1992. The dependent personality: Developmental,

social, and clinical perspectives. Psychol Bull 112:323.Cannon B, Mulroy R, Otto MW, Rosenbaum JF, Fava M, Nierenberg

AA. 1999. Dysfunctional attitudes and poor problem solving skillspredict hopelessness in major depression. J Affect Disord 55:4549.

Corruble E, Ginestet D, Guelfi JD. 1996. Comorbidity of personalitydisorders and unipolar major depression: A review. J Affect Disord37:157170.

Dixon WA, Heppner PP, Andersen WP. 1991. Problem-solvingappraisal, stress, hopelessness, and suicide ideation in a collegepopulation. J Counseling Psychol 38:5156.

Dixon WA, Heppner PP, Burnett JW, Anderson WP, Wood PK.1993. Distinguishing among antecedents, concomitants, andconsequences of problem-solving appraisal and depressive symp-toms. J Counseling Psychol 40:357364.

Fabrega H Jr, Ulrich R., Pilkonis P, Mezzich J. 1991. On thehomogeneity of personality disorder clusters. Compr Psychiatry 32:373386.

Fava M, Farabaugh AH, Sickinger AH, Wright E, Alpert JE,Sonawalla S, Nierenberg AA, Worthington JJ. 2002.Personality disorders and depression. Psychol Med 32:10491057.

Fava M, Bouffides E, Pava JA, McCarthy MK, Steingard RJ,Rosenbaum JF. 1994. Personality disorder comorbidity with major

depression and response to fluoxetine treatment. PsychotherPsychosom 62:160167.

Goddard L, Dritschel B, Burton A. 1996. Role of autobiographicalmemory in social problem solving and depression. J AbnormPsychol 105:609616.

Hamilton M. 1960. A rating scale for depression. J NeurolNeurosurg Psychiatry 23:5662.

Heppner PP, Baumgardner A, Jackson J. 1985. Problem-solvingself-appraisal, depression, and attributional style: Are they related?Cogn Ther Res 9:105113.

Heppner PP, Petersen CH. 1982. The development and implicationsof a personal problem-solving inventory. J Counseling Psychol29:6675.

Huprich SK, Clancy C, Bornstein RF, Nelson-Gray RO. 2004. Dodependency and social skills combine to predict depression?:

Linking two diatheses in mood disorder research. IndividualDifferences Res 2:216.

Klein DN, Durbin E, Shankman SA, Santiago NJ. 2002. Depressionand personality. In: Gotlib IH, Hammen CL, editors. Handbookof depression. New York: Guilford Press. p 115140.

Linehan MM. 1993. Cognitive-behavioral treatment of borderlinepersonality disorder. New York: Guilford Press.

Mulder RT. 2002. Personality pathology and treatmentoutcome in major depression: A review. Am J Psychiatry 159:359371.

Nezu AM, Ronan GF. 1985. Life stress, current problems, problem-solving and depression symptoms: An integrative model. J ConsultClin Psychol 53:693697.

Overholser JC, Fine MA. 1994. Cognitive behavioral treatment ofexcessive interpersonal dependency: A four-stage psychotherapy

model. J Cogn Psychother 8:5570.Platt JJ, Spivack G. 1975. Manual for the MeansEnds Problem-

Solving: A measure of interpersonal problem-solving skill.Philadelphia: Hahnemann Medical College and Hospital.

Priester MJ, Clum GA. 1993. Perceived problem-solving abilityas a predictor of depression, hopelessness, and suicide ideation ina college population. J Counseling Psychol 40:7985.

Spitzer RL, Williams JBW, Gibbon N, First MB. 1989. StructuredClinical Interview for DSM-III-RPatient Version (SCID-P).New York New York State Psychiatric Institute, BiometricsResearch.

Spitzer RL, Williams JBW, Gibbon N, First MB. 1990. StructuredClinical Interview for DSM-III-R Personality Disorders (SCID-II). New York: New York State Psychiatric Institute, BiometricsResearch.

Watkins E, Baracaia S. 2002. Rumination and social problem solvingin depression. Behav Res Ther 40:11791189.

Zimmerman M. 1994. Diagnosing personality disorders: A reviewof issues and research methods. Arch Gen Psychiatry 51:225245.

501Brief Report: Problem Solving, Personality Disorders, and Depression

Depression and Anxiety DOI 10.1002/da