Personality and the Expression of Symptomatology in Schizophrenia and Bipolar Disorder

Abstract

Researchers and clinicians have begun using dimensions rather than categories to classify psychopathology with a reliance on personality questionnaires to tap traits that can inform dimensional characterizations. A neglected concern is whether in severe psychopathology questionnaire-based assessments of personality reflect a lifetime propensity toward a diagnosis, as some personality-psychopathology models posit, or reflect the transient effects of current symptoms, as a complication model of personality-psychopathology would suggest. Accurate characterization of psychopathology is necessary to understand etiology and prescribe clinical care. We studied 127 adults with schizophrenia, schizoaffective, or bipolar disorder who completed well-validated measures of personality, current symptomatology, and lifetime psychopathology. We found that normative personality traits were related to current symptoms but unrelated to lifetime symptomatology, whereas the schizotypal trait of cognitive-perceptual distortions predicted lifetime psychosis severity. Questionnaire-based assessments of normative personality are likely affected by current symptom states and may fail to yield a stable characterization of psychopathology.

Introduction

Theoretical models of how personality is related to the expression of a mental disorder often cast deviations in normative personality dimensions (e.g., Positive Emotionality [PEM], Negative Emotionality [NEM], and Disinhibition) as diatheses for psychopathology and as a possible explanation for high rates of comorbid diagnoses (e.g., Clark, 2005; Kotov et al., 2017). These personality-psychopathology models are embodied in a recent paradigm shift by the National Institute of Mental Health, which focuses on investigating symptom dimensions across diagnostic categories and conceptualizing psychopathology as a dysfunction of core neurobehavioral systems (Insel et al., 2010). The core neurobehavioral systems are presumably reflected in the dimensional traits that make up normative personality (e.g., Depue and Collins, 1999). There is strong empirical support for a link between normative personality and common mental disorders (for review and meta-analysis see Kotov et al., 2010). Studies of normative personality-psychopathology associations are also beginning to address the role of personality in the symptom course of severe forms of psychopathology, such as schizophrenia (SCZ) and bipolar disorders (BD) – referred to broadly as severe mental illness (SMI) (Camisa et al., 2005; Horan et al., 2008; Quilty et al., 2009; Sparding et al., 2017). However, investigations of SMI have failed to adequately account for the likely effects of the current symptom states when examining how personality is related to the expression of psychopathology over a lifetime. To better understand the relationship of personality to severe psychopathology given the potential confound of current symptomatology, we examined whether lifetime symptomatology was related to personality after accounting for the effects of current clinical states.

For several reasons it is important to understand how deviations in normative personality may be relevant to severe psychopathology. SMIs exhibit higher rates of heritability (Lichtenstein et al., 2009; Kieseppa et al., 2004; Sullivan et al., 2003) as well as more severe functional impairment (Addington et al., 2001; Karpov et al., 2017; Goldberg and Harrow, 2011) and cognitive deficits (Reichenberg et al., 2009) than other mental disorders. SMIs also have specific symptom presentations (e.g., hallucinations, manic euphoria) rarely observed in common forms of psychopathology. These differences between SMI and common mental disorders may be reflected in unique relationships of personality deviations with SMI. Moreover, inclusion of SMI in prior studies of psychopathology has led to re-evaluation of an overarching dimensional structure of psychopathology (e.g., Caspi et al., 2014; Wright et al., 2013)—from the two-dimensions (Internalizing versus Externalizing; Krueger et al., 1998) to three dimensions (SMI loading on a third Psychotic Experiences factor; Wright et al., 2013) and even proposals of a single dimension (Caspi et al., 2014). Therefore, it is important to test the limits of models of personality-psychopathology associations through inclusion of highly heritable, severe, and impairing forms of psychopathology with less commonly observed symptoms—SMI.

Little is known about the role of the normative personality domains, such as positive and negative emotionality (PEM, NEM) and Constraint (CON: high self-control or the inverse – disinhibition) in SMI. Many studies focus on measures of Five Factor Model of normative personality (e.g., Camisa et al., 2005; Quilty et al., 2009). Studies point to increased NEM in individuals with SMI compared to healthy controls (Bagby et al., 1997; Horan et al., 2005; Wilson and Sponheim, 2014), whereas the few studies of CON revealed no associations of the trait with SMI (e.g., Lönnqvist et al., 2009). Investigations of PEM in SMI have yielded more mixed findings—from normal range PEM (Bagby et al., 1997) to abnormally low levels (Wilson and Sponheim, 2014). Given theoretical models positing positive emotion persistence as a core dysregulation in some forms of SMI, such as BD (Gruber, 2011), and the mixed findings, additional investigations are required to clarify PEM associations with SMI.

Studies investigating how specific SMI symptom dimensions relate to normative personality traits have also yielded inconsistent findings. Preliminary evidence suggests that normative personality traits and SMI symptoms covary across time (Horan et al., 2005). In SMI, NEM is positively related to depression severity (Horan et al., 2005), as well as current positive (delusions and hallucinations) and negative (affective flattening, motor retardation, emotional withdrawal) psychotic symptoms (Wilson and Sponheim, 2014). However, no studies have examined links between lifetime propensity toward specific symptom domains and normative personality dimensions in SMI. It is also unclear how symptoms of severe psychopathology are explained by normative personality traits compared to aspects of personality typically associated with SMI, such as schizotypy (e.g., Debbane and Barrantes-Vidal, 2015; Kwapil et al., 2013). Specifically, it is unknown whether the Big Three normative personality traits (PEM, NEM, and CON) uniquely account for aspects of lifetime symptomatology in SMI that are unrelated to schizotypal traits. To address these questions, the present study examined domains of schizotypal personality and normative personality domains for associations with current and lifetime symptomatology, in individuals with SCZ, schizoaffective disorders, and BD.

There are several models for normative personality-psychopathology associations (Clark, 2005) that lead to contrasting hypotheses for the present study. According to the common factor model (Clark, 2005; Cox et al., 2015), lifetime symptom severity and normative personality traits are related through common etiological factors. Alternatively, the vulnerability model posits that personality is the vulnerability factor for SMI, whereas the scar model posits a reversed association in which psychopathology permanently alters personality traits in SMI. Despite different etiological mechanisms that are beyond the scope of the current analyses, all three models would predict strong cross-sectional associations between lifetime SMI symptoms severities and normative personality traits, even after accounting for associations of personality with current clinical symptoms in the present study. Consequently, based on dominant theoretical models of dysregulated response to rewards in BD (e.g., Urosevic et al., 2008; Johnson et al., 2012), we predict that lifetime severity of mania will be related to greater PEM, and lifetime severity of depression to lower PEM, regardless of current clinical states. Although some studies have failed to find elevated PEM in BD, perhaps this was due to the relative infrequency of manic episodes as compared to depressive phases (Judd et al., 2002; Judd et al., 2003) and diagnosis-categorical rather than symptom-dimensional analytic approaches. Based on prior studies of SCZ (Mote et al., 2014), we predict that greater negative psychotic symptomatology over lifetime will be associated with lower PEM. Although a relationship between CON and substance use has been revealed by studies of non-SMI samples (Chassin, Flora, & King, 2004; Conway, Swendsen, Rounsaville, & Merikangas, 2002), predictions for associations between CON and lifetime SMI symptom dimensions are less clear.

Unlike the other three models, the complication model of normative personality-psychopathology posits that current symptoms temporarily alter personality trait expression and leads to a contrasting set of hypotheses. If the complication model better fits the personality-SMI associations, we predict that current symptoms severities will explain most variance in normative personality traits in the present study. Moreover, lifetime SMI symptom severities will not be associated with normative personality traits in this sample of individuals with SMI. Finally, according to the long-standing vulnerability models related to schizotypy (e.g., Meehl, 1990) and empirical support for links of schizotypal traits to vulnerability and development of psychosis (e.g., Debbane and Barrantes-Vidal, 2015; Kwapil et al., 2013), one could reasonably expect that schizotypal traits would be associated with lifetime history of positive psychotic symptoms.

Methods

Participants

Participants (n = 127) included individuals with SCZ (n = 71; 55 with schizophrenia, 6 with schizoaffective disorder, depressive type) and BD (n = 56; 42 with bipolar I disorder, 1 with bipolar disorder not otherwise specified, and 13 with schizoaffective disorder, bipolar type) diagnoses. They were recruited from the Minneapolis VA Health Care System (VAHCS) and community outpatient programs and were part of a larger study of SMI with the following exclusion criteria: non-native English speaker, IQ < 70, current alcohol/drug abuse, past drug dependence, history of neurological condition/disease affecting the central nervous system, or history of significant head injury. Institutional Review Boards at the Minneapolis VAHCS and University of Minnesota approved all procedures and monitored the study protocols. Participants completed an informed consent process prior to study enrollment.

Measures

Diagnostic and clinical symptom assessments.

Diagnostic and Statistics Manual, 4^th edition (DSM-IV) diagnoses were confirmed through administration of the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-I; Spitzer et al., 2002) by experienced and supervised interviewers. All interviewers achieved interclass correlation > .80 for symptom ratings prior to training certification, attended weekly diagnostic meetings for supervision purposes, and maintained acceptable inter-rater reliability. All diagnoses were reviewed during an additional consensus process by clinical psychologists and advanced graduate students, independent from interviewers (Leckman et al., 1982). Consensus reviewers utilized all available information—clinical interviews, patient medical records, and information obtained from participants’ relatives—to assign final diagnoses.

Using Brief Psychiatric Rating Scale (BPRS; Lukoff et al., 1986), interviewers rated participants’ symptoms from the past month on a 7-point Likert scale for the 24 items to derive indices of current symptomatology. BPRS is a widely used measure of SMI symptoms with well-established psychometric properties (Lukoff et al. 1986). Five current symptom dimensions of positive symptoms, negative symptoms, disorganization, mania, and depression, were computed based on the BPRS factor structure reported by Wilson and Sponheim (2014). Each subscale consisted of the average scores of BPRS items highly loading (≥ .30) on a relevant factor.

Operational Criteria Checklist for Psychotic and Affective Illness (OPCRIT; McGuffin et al., 1991) is a 90-item clinician-rated assessment instrument of lifetime psychiatric symptoms. In the present study, consensus reviewers completed OPCRIT symptom ratings as part of the consensus process utilizing all available diagnostic information (see above). The OPCRIT shows excellent psychometric properties, such as high agreement with gold standard consensus lifetime diagnoses and good interrater reliability (Williams, 1996; Craddock et al., 1996; Azevedo et al., 1999). Previous studies of SMI using the OPCRIT support its utility. For example, Docherty et al. (2015) found relationships between gene variants and symptom factors derived from OPCRIT items in SCZ. Goghari and Sponheim (2008) also found relationships between genetic variation (i.e., COMT Val158Met polymorphism) and lifetime clinical phenotypes assessed by the OPCRIT, while there were no relationships with current symptomatology.

Given the episodic nature of mood disorders and consistent with the OPCRIT Item Guidelines and Ratings manual (McGuffin et al., 1991), OPCRIT lifetime depression and mania symptom ratings were assessed using information from the most severe episode of depression and mania during lifetime, respectively. Although these depression and mania ratings are for the single most severe episodes over the lifetime, evidence indicates that individuals with BD experience a similar constellation of symptoms across episodes (Perlis et al., 2009; Sato et al., 2003). OPCRIT lifetime psychosis symptom ratings assessed a psychotic symptom occurrence at any point during lifetime. The OPCRIT Item Guidelines and Ratings manual instructed raters on a severity threshold necessary in order to mark the symptom as “present” with subsequent ratings determining severity based on duration and/or impairment (McGuffin et al., 1991). For example, for item rating manic symptom of pressured speech, the manual indicates patient must be more talkative than usual to receive a positive rating, with scores of 1, 2, and 3 reflecting a symptom duration of 4 to 6 days, 7 to 13 days, and ≥ 14 days, respectively.

Five lifetime symptom dimensions of mania (11 items assessing symptoms such as pressured speech, elevated mood), depression (9 items assessing symptoms such as loss of pleasure, poor concentration), reality distortion (10 items assessing positive symptoms, such as auditory hallucinations, thought insertion), disorganization (11 items assessing symptoms such as incoherence, bizarre behavior, catatonia), and negative symptoms (4 items assessing symptoms such as blunted affect) were computed according to the OPCRIT factor structure reported by Dikeos et al. (2006). This factor structure was based on a large representative sample of individuals with SMI (Dikeos et al., 2006) and thus was chosen over alternative structures (e.g., Serretti and Olgiati, 2004; Cardno et al., 1996; Maciukiewicz et al., 2012). First, OPCRIT item scores were transformed to ‘0’ (symptom absence) and ‘1’ (symptom presence) to account for differences in rating scale ranges across individual items. For example, item 21 “Distractibility” is rated on a 3-point scale ranging from none to at least 4 weeks, whereas item 22 “Reduced need for sleep” is rated on a 3-point scale ranging from none to at least 2 weeks. Transformation to 0 vs. 1 rating was also consistent with previous studies (e.g., Brittain et al., 2013). Five lifetime symptom dimensions were derived by summing items highly loading on a relevant factor.

Personality assessment.

Participants completed the Multidimensional Personality Questionnaire-Brief Form (MPQ-BF; Patrick et al., 2002), a common and well-validated measure of Big Three model of normative personality. The MPQ-BF is a 155-item self-report assessment of normative personality with three higher-order personality dimensions – Positive Emotionality (PEM), Negative Emotionality (NEM), and Constraint (CON) and 11 primary trait scales (e.g., Social Potency). Additionally, four second-order personality dimensions are often derived (Hicks et al., 2004; Ciarleglio et al., 2008) and include: Agentic-PEM (related strongly to Well-Being, Social Potency, Achievement primary trait scales), Communal-PEM (related strongly to Well-being and Social Closeness primary trait scales), Agentic-NEM (related strongly to Stress Reactivity and Aggression primary trait scales), and Alienated-NEM (related strongly to Alienation primary trait scale) (Patrick et al., 2002). MPQ-BF higher-order and lower-order trait dimensions show high correlations (rs from .92 to .98) with trait dimensions on the original MPQ version and exhibit excellent reliability and construct validity (Patrick et al., 2002). MPQ-BF Absorption subscale does not load on any of the higher-order trait dimensions and was included in analyses on its own. The MPQ contains validity scales assessing consistency and reliability of responses and provides cut off points for inclusion in analyses (Patrick et al., 2002); only participants with valid and complete MPQ-BF responses are included in the current sample (i.e., excluded 4 participants with invalid and 2 participants with incomplete responses).

Participants also completed the 74-item self-report assessment of schizotypal personality, Schizotypal Personality Questionnaire (SPQ; Raine, 1991). The SPQ consists of yes/no questions summed in nine subscales originally derived to map onto the diagnostic criteria for schizotypal personality disorder (American Psychiatric Association, 2013). However, the SPQ is frequently used to assess traits of schizotypy in addition to schizotypal personality disorders (Gross et al., 2014). Factor analyses of the SPQ yield three second-order personality dimensions: Cognitive-Perceptual (related strongly to Ideas of Reference, Odd Beliefs/Magical Thinking, Unusual Perceptual Experiences, and Suspiciousness/Paranoid Ideation), Interpersonal (strongly related to Social Anxiety, No Close Friends, Constricted Affect, and Suspiciousness/Paranoid Ideation), and Disorganization (strongly related to Eccentric/Odd Behavior and Odd Speech) (Raine et al., 1994; Barron et al., 2015; Reynolds et al., 2000). This three-factor model is supported across diverse samples when considering gender, ethnicity, and psychopathology (Barron et al., 2015; Reynolds et al., 2000). Sums of relevant SPQ subscales yielded three SPQ trait dimensions.

Statistical analyses

Bivariate correlations were calculated to examine associations among measures for descriptive purposes, thus not corrected for multiple comparisons. Seven hierarchical regressions were conducted for the entire sample, one for each normative and schizotypal personality domain. Each hierarchical regression consisted of two steps. To first account for current symptomatology effects, in Step 1, the five BPRS current symptom dimensions were entered. In Step 2, the five OPCRIT lifetime symptom dimensions were entered together to examine effects of overall SMI lifetime symptom severity after accounting for current clinical state effects. Step 2 OPCRIT scale predictors also examined unique associations of each lifetime symptom dimension with each personality trait (i.e., while accounting for a personality trait’s associations with other lifetime symptom dimensions). Outcome variables were higher-order MPQ-BF personality dimensions—PEM, NEM, and CON, as well as MPQ-BF Absorption scale, and higher-order SPQ personality dimensions—Cognitive-Perceptual, Interpersonal, and Disorganization. Regressions showed no signs of high multicollinearity (all VIF < 5). Significant associations of current or lifetime symptom dimensions with MPQ scores were followed up by examining second-order personality dimensions (e.g., agentic and communal PEM).

Finally, Supplementary Materials describe results of analogous analyses performed separately in SCZ vs. BD groups of participants; however, additional analyses found no significant interaction effects of the diagnostic group by each symptom dimension on personality traits (i.e., no difference in personality-psychopathology associations by diagnostic group), further justifying the main text’s analyses that assess symptom-personality dimensions across diagnoses. Supplemental Materials also present regression analyses assessing unique associations of each personality dimension with each lifetime symptom dimension by alternatively specifying lifetime symptom dimensions as outcome variables and personality traits as predictors.

Results

Descriptive statistics and correlations

Table 1 provides demographic and descriptive statistics for each diagnostic group of participants separately. Supplemental materials include separate analyses for the SCZ and BD groups, while analyses of the entire sample are presented here. Table 2 summarizes correlations among current symptom dimensions, lifetime symptom dimensions, normative personality MPQ-BF higher-order traits, MPQ-BF Absorption, and SPQ trait dimensions in the whole sample. Overall, there were positive associations between current symptoms and analogous lifetime symptoms on each symptom dimension, except for disorganization. Only current symptoms exhibited significant associations with MPQ-BF personality traits. However, SPQ personality dimensions exhibited associations with symptoms assessed for lifetime (e.g., Cognitive-Perceptual dimension was associated positively with lifetime positive psychosis and disorganization symptom severities, and inversely with lifetime mania severity) and current clinical state (e.g., all three SPQ dimensions were positively associated with positive symptoms and inversely with depression). All three SPQ dimensions were positively associated with MPQ-BF NEM, MPQ-BF Absorption, and inversely related to MPQ-BF PEM and CON, except for the lack of association between CON and SPQ Interpersonal domain.

Table 1.

Demographic, Clinical and Personality Characteristics in Each Diagnostic Group

Variable	SCZ Groupa (n=71)	BD Groupa (n=56)	t/Mann-Whitney U	p
Age, M (SD)	41.11 (11.26)	45.71 (12.40)	−2.19	.031
Male, n (%)	50 (70%)	40 (71%)	1968.00	.902
Income, M (SD)b	2.29 (1.26)	2.95 (1.46)	−2.66	.009
Caucasian, n (%)	39 (55%)	45 (80%)	7.16	.007
Number of Comorbid Axis I Dxsc	1.61 (0.98)	1.57 (1.26)	0.17c	.870
BPRS Factors
Positive Sxs	2.85 (1.32)	1.64 (0.88)	6.15c	< .001
Negative Sxs	1.99 (1.18)	1.49 (0.75)	2.89c	.005
Disorganization	2.01 (0.76)	1.82 (0.63)	1.53	.128
Mania	1.29 (0.70)	1.54 (0.82)	−1.82	.071
Depression/Anxiety	2.18 (0.96)	2.46 (1.19)	−1.50	.137
OPCRIT Factors
Reality Distortion	4.13 (2.16)	1.66 (2.11)	6.50	< .001
Negative Sxs	1.37 (1.19)	0.82 (0.88)	2.98c	.004
Disorganization	3.99 (2.02)	2.45 (2.48)	3.86	< .001
Mania	1.70 (2.98)	8.57 (1.45)	−17.02c	< .001
Depression	3.51 (3.29)	6.25 (2.76)	−5.11c	< .001
MPQ Factors
PEM	63.20 (14.00)	62.86 (18.56)	0.11c	.910
Agentic PEM	53.92 (12.93)	56.34 (15.00)	0.98	.330
Communal PEM	56.42 (15.06)	54.68 (18.94)	0.56c	.575
NEM	46.56 (19.82)	39.66 (17.38)	2.06	.042
Agentic NEM	44.90 (16.30)	42.39 (12.89)	0.97c	.335
Alienated NEM	55.41 (21.36)	47.23 (18.57)	2.31c	.023
Constraint (CON)	80.70 (11.68)	78.66 (12.90)	0.94	.352
SPQ Factors
Cognitive-Perceptual	14.87 (9.12)	9.39 (7.05)	3.82c	< .001
Interpersonal	15.79 (7.84)	13.05 (7.12)	2.03	.044
Disorganization	7.38 (4.98)	5.93 (4.33)	1.73	.087

^aBD group includes participants with bipolar disorder and schizoaffective disorder-bipolar type diagnoses, whereas SCZ group includes participants with schizophrenia and schizoaffective disorder-depressive type diagnoses.

^bIncome level of 2 = $10,000–20,000; 3 = $20,000–30,000;

^cMost common comorbid diagnosis was substance dependence, in remission;

BPRS=Brief Psychiatric Rating Scale; MPQBF=Multidimensional Personality Questionnaire—Brief Form; OPCRIT=Operational Criteria Checklist for Psychoses and Affective Illness; SPQ= Schizotypal Personality Questionnaire; Sxs=symptoms.

Table 2.

Associations between BPRS, OPCRIT, MPQ, and SPQ Scales across SCZ and BD diagnoses

	1.	2.	3.	4.	5.	6.	7.	8.	9.	10.	11.	12.	13.	14.	15.	16.	17.
1. OPCRIT Reality Distortion	--	.28 a	.49 b	−.41b	−.14	.55 b	.10	.07	−.16	−.06	−.03	.17	.10	.12	.42 b	.17	.14
2. OPCRIT Negative Symptoms		--	.28 a	−.16	0	.13	.56 b	.01	−.24b	.08	−.08	−.03	.08	−.08	.04	.14	.06
3. OPCRIT Disorganization			--	−.16	−.22a	.50 b	.18 a	.11	.13	−.02	.02	.05	.05	−.06	.19 a	−.03	.15
4. OPCRIT Mania				--	.36 b	−.33b	−.22a	−.07	.24 a	.09	−.01	−.16	−.03	−.12	−.26a	−.18a	−.17
5. OPCRIT Depression					--	−.21a	−.13	−.04	−.12	.27 a	−.10	.10	−.14	.00	−.06	−.03	−.13
6. BPRS Positive Symptoms						--	.26 a	.23 a	.01	.14	−.13	.32 b	0	.28 a	.49 b	.26 a	.28 b
7. BPRS Negative Symptoms							--	.18 a	−.32b	.16	−.20a	.09	0	.09	.18 a	.27 a	.14
8. BPRS Disorganization								--	.13	.34 b	−.22a	.24 b	−.04	.02	.17	.18 a	.11
9. BPRS Mania									--	.02	.10	−.06	−.04	−.09	−.05	−.07	.02
10. BPRS Depression/Anxiety										--	−.29b	.29 b	−.10	.20 a	.33 b	.41 b	.26 a
11. MPQ PEM											--	−.06	.06	.06	−.24a	−.47b	−.21a
12. MPQ NEM												--	−.20a	.53 b	.67 b	.53 b	.49 b
13. MPQ Constraint (CON)													--	−.15	−.19a	−.15	−.28b
14. MPQ Absorption														--	.56 b	.42 b	.45 b
15. SPQ Cognitive-Perceptual															--	.76 b	.62 b
16. SPQ Interpersonal																--	.59 b
17. SPQ Disorganization																	--

Note:

^ap< .05

^bp≤ .001; significant findings highlighted in bold; all Pearson’s r correlations. Current symptoms associations presented with gray background.

Normative personality traits and symptom associations

Table 3 summarizes findings for two hierarchical regressions with MPQ-BF PEM scores or NEM scores as outcome measures, which test associations between lifetime symptom dimensions and normative personality traits, after accounting for effects of the current clinical states on personality. For both PEM and NEM, there were significant associations with current symptom states but not with lifetime symptom dimensions. Using lifetime symptom dimensions as the outcome variables yielded similar results (see Supplementary Materials).

Table 3.

Current and Lifetime Symptom Associations with MPQ-BF Positive and Negative Emotionality and SPQ Domains across SCZ and BD Diagnoses

Model	MPQ-BF Positive Emotionality (PEM)		MPQ-BF Negative Emotionality (NEM)		SPQ Cognitive-Perceptual Dimension		SPQ Interpersonal Dimension		SPQ Disorganization Dimension
	Change R2/ β	[95% CI]	Change R2/ β	[95% CI]	Change R2/ β	[95% CI]	Change R2/ β	[95% CI]	Change R2/ β	[95% CI]
Step 1 Model: Current Symptomatology	.13	[.03, .23]	.18	[.07, .29]	.31	[.18, .44]	.23	[.11, .35]	.13	[.03, .23]
BPRS Positive Sxs	−.05	[−.23, .13]	.28	[.11, .45]	.46	[.30, .61]	.17	[.002, .34]	.24	[.07, .42]
BPRS Negative Sxs	−.10	[−.29, .09]	−.07	[−.25, .11]	.01	[−.16, .18]	.16	[−.02, .34]	.06	[−.13, .25]
BPRS Disorganization	−.12	[−.31, .06]	.13	[−.05, .31]	−.03	[−.19, .14]	−.01	[−.18, .17]	−.04	[−.22, .15]
BPRS Mania	.08	[−.10, .27]	−.10	[−.28, .07]	−.05	[−.21, .11]	−.03	[−.20, .14]	.04	[−.15, .22]
BPRS Depression	−.22	[−.40, −.04]	.22	[.05, .39]	.27	[.11, .43]	.36	[.19, .53]	.23	[.05, .41]
Step 2 Model: Lifetime Symptomatology	.01	[−.02, .04]	.03	[−.03, .09]	.08	[0, .17]	.06	[−.02, .14]	.03	[−.03, .09]
OPCRIT Reality Distortion	−.04	[−.28, .21]	.06	[−.17, .29]	.36	[.15, .56]	.19	[−.03, .41]	.01	[−.23, .25]
OPCRIT Negative Sxs	.02	[−.21, .25]	−.11	[−.33, .10]	−.17	[−.36, .02]	.01	[−.20, .21]	−.02	[−.25, .20]
OPCRIT Disorganization	.09	[−.13, .31]	−.08	[−.29, .13]	−.11	[−.29, .08]	−.25	[−.45, −.05]	.02	[−.20, .24]
OPCRIT Mania	−.06	[−.27, .15]	−.12	[−.32, .08]	−.07	[−.25, .10]	−.07	[−.26, .11]	−.09	[−.30, .12]
OPCRIT Depression	−.05	[−.25, .15]	.14	[−.05, .33]	−.02	[−.19, .15]	−.09	[−.27, .10]	−.12	[−.32, .08]
Full Model (R2)	.14	[.04, .24]	.22	[.11, .34]	.39	[.27, .51]	.30	[.18, .42]	.16	[.05, .27]

Note: Values in bold denote statistical significance, p < .05. No significant symptom associations for MPQ Constraint (CON).

CI=confidence interval; MPQ-BF=Multidimensional Personality Questionnaire-Brief Form; SMI=severe mental illness; SPQ=Schizotypal Personality Questionnaire.

In a follow-up regression of Communal PEM scores’ variance, the model with five current symptom dimensions was significant, R² = .12, 95% CI [.02, .22], F(5, 121) = 3.43, p = .006, and revealed that BPRS Depression had a unique inverse association with Communal PEM scores, β= −.24, 95% CI [−.42, −.06], t = −2.67, p = .009. Similarly, for Agentic PEM scores, the model with five current symptom dimensions was significant, R² = .10, 95% CI [.01, .19], F(5, 121) = 2.72, p = .023; however, no BPRS symptom dimensions had a unique association.

In an analogous follow-up regression of Agentic NEM scores, the model with five current symptom dimensions was significant, R² = .09, 95% CI [.001, .18], F(5, 121) = 2.44, p = .038, but no BPRS symptom dimension scores showed a unique association. In a follow-up regression of Alienated NEM scores’ variance, the model with five BPRS/current symptom dimensions was significant, R² = .20, 95% CI [.08, .32], F(5, 121) = 5.94, p < .001; BPRS Positive Symptoms and BPRS Depression had unique positive associations, β = .31, 95% CI [.14, .48], t = 3.60, p < .001, and β = 0.21, 95% CI [.04, .39], t = 2.46, p = .016, respectively.

In a hierarchical regression modeling variance in MPQ-BF CON, the Step 1 model including five current symptom dimensions and the Step 2 model including five lifetime symptom dimensions were not significant, R² = .01, 95% CI [−0.02, .04], F(5, 121) = 0.27, p = .927, and R² = .04, 95% CI [−0.02, .10], F(10, 116) = 0.52, p = .872, respectively.

Finally, in a hierarchical regression of MPQ-BF Absorption scores, the Step 1 model including five current symptom dimensions was significant, R² = .12, 95% CI [.02, .22], F(5, 121) = 3.27, p = .008. BPRS Positive Symptoms and BPRS Depression dimensions had unique associations with MPQ-BF Absorption subscale scores, β = .28, 95% CI [.10, .45], t = 3.05, p = .003, and, β = .20, 95% CI [.01, .38], t = 2.14, p = .034, respectively. The five lifetime symptom dimensions did not explain significant additional variance in Absorption scores. Still, the overall final model was significant, R² = .17, 95% CI [.06, .28], F(10, 116) = 2.44, p = .011.

Schizotypal personality traits and symptom associations

Table 3 also summarizes three hierarchical regressions examining association between lifetime symptomatology and SPQ Cognitive-Perceptual, Interpersonal, and Disorganization scores, after accounting for effects of current symptoms on schizotypal traits. SPQ Cognitive-Perceptual and Interpersonal dimensions were associated with lifetime severity of positive and disorganization symptoms, respectively, in these analyses. All three SPQ dimensions were also associated with current depression and positive symptom severities.

Finally, in analogous hierarchical regressions with MPQ and SPQ scores as outcomes and age and gender as additional covariates, the pattern of findings remained largely unchanged, with a few exceptions: 1) in Step 2 of relevant regressions, BPRS Disorganization had a significant association with MPQ NEM (β = .18, p = .048) and BPRS Depression with MPQ Agentic NEM (β = .19, p = .048), whereas BPRS Positive Symptoms’ association with SPQ Interpersonal Domain was no longer significant (β = .16, p = .061); and 2) in a regression predicting MPQ Absorption scores, despite the OPCRIT symptom dimensions overall not explaining a significant amount of additional variance after accounting for effects of current clinical states, a significant inverse association with OPCRIT Disorganization was present (β = −.24, p = .031).

Discussion

In the present study, personality traits were more related to current clinical symptoms than a lifetime prevalence of symptomatology in SCZ and BD. In a large group of mental health outpatients with SMI, variation in questionnaire-based assessments of the Big-Three personality traits (PEM, NEM, and CON) and schizotypal personality were most closely tied to clinical interviewer assessments of current symptoms, rather than clinician ratings of lifetime psychopathology. Specifically, current depression was related to lower PEM, higher NEM, and higher scores on all three schizotypal personality traits. Current positive symptoms of psychosis were related to higher NEM and higher scores on all the schizotypal personality traits. After accounting for the effect of current symptomatology, dimensions of lifetime symptomatology failed to be associated with normative personality characteristics among individuals with SMI. Lifetime severity of positive symptoms and disorganization were related to scores on schizotypal cognitive-perceptual and interpersonal domains, respectively. Lack of bivariate correlations between the MPQ traits and OPCRIT scales also indicates no viable model where normative personality contributes to lifetime SMI symptoms. Overall, results provide evidence for current clinical state being a dominant influence on self-report measures of personality.

The present findings are consistent with normative personality-psychopathology associations observed in common psychiatric disorders, but also identify unique challenges for extending this research to SMI. A meta-analysis of studies examining common mental disorders supports the relationship of depressive symptom severity with higher neuroticism (i.e., NEM) and lower extraversion, a trait related to PEM (Hakulinen et al., 2015). Supporting the effects of current symptoms on normative personality measures, another recent meta-analysis found that accounting for impact of baseline symptoms halved effect sizes for neuroticism and prospective symptom associations in common mental disorders (Jeronimus et al., 2016). The same meta-analysis revealed small associations with SMI (“thought disorder”) that failed to show reductions after adjustment for baseline symptoms (Jeronimus et al., 2016). In contrast, the present study extends this prior literature by providing novel support for the complication model of personality in SMI. According to the complication model, current SMI symptoms may lead to temporary effects on how individuals respond to personality measures, or to real temporary alterations of personality trait expressions, that resolve with remission of acute symptoms. In SMI, it appears critical to account for the impact of not just current depression, but also current positive symptoms, prior to examining lifetime symptomatology associations with personality traits.

There are additional SMI-specific implications of the failure to find associations between normative personality traits with lifetime SMI symptom dimensions. The absence of associations is compelling given the observed associations between schizotypal personality domains and lifetime SMI symptom dimensions, even after taking into account current levels of symptomatology. This pattern of findings could be interpreted as failing to support the scar model—model specifying that current personality attributes are permanently altered by psychopathology symptoms across the lifetime (Akiskal et al., 1983; Enns and Cox, 1997)—for SMI-normative personality associations. In other words, if severity of SMI led to enduring changes on normative personality traits, significant associations between MPQ-BF personality measures and lifetime symptom severities would be expected but none were found. Furthermore, the current findings suggest that the MPQ-BF higher-order personality dimensions may not fully tap into etiological factors underlying SMI. Consistently, multiple lines of investigation have supported a dysregulation of behavioral approach system (BAS) as etiological factor contributing to BD (for reviews see Johnson et al., 2012; Urosevic et al., 2008) and that BAS may involve aspects of both positive emotions and negative emotions (e.g., anger; Harmon-Jones, 2003), which do not neatly map onto the PEM versus NEM dichotomy. Alternatively, the MPQ may tap into vulnerability factors for SMI, but these factors may not be sufficient to determine who develops more severe SMI symptoms over lifetime; other factors (e.g., environmental stressors) may interact with elements of vulnerability and personality to determine the severity of SMI symptoms over a lifetime. Finally, it could be that the MPQ-BF, like many other normative personality assessment tools, was developed and validated in healthy populations and potentially lacks the ability to fully capture the full range of personality profiles in populations with SMI.

It should be noted that the present study did not examine all possible normative personality traits, such as Big Five personality traits. Still, it is reasonable to expect a similar pattern of observations to emerge in relationships of other normative personality traits with SMI lifetime and current symptomatology. For example, the Big Five trait of openness shows strong links to psychosis and is reliably assessed by the MPQ subscale of Absorption (DeYoung et al., 2012). In the present study, the MPQ-BF Absorption scores exhibited similar pattern of findings to other MPQ-BF-assessed traits—associations with current depressive and positive symptom severities and an absence of associations with lifetime SMI symptoms. Other studies of the Big Five personality traits in individuals with SMI also support effects of current depression on self-report assessments of extraversion and neuroticism, while failing to find associations with lifetime markers of SMI course severity, such as number of psychotic episodes (Boyette et al., 2015). These findings for the Big Five personality traits are consistent with the present results.

The present study’s findings highlight that associations between psychopathology and normative personality evident in common mental disorders cannot be assumed to generalize to SMI. Prior studies have found that inclusion of SMI can lead to re-evaluation of the overall dimensional structure of psychopathology (e.g., Caspi et al., 2014; Wright et al., 2013). Thus, including individuals with SMI when examining the associations of psychiatric symptom dimensions with normative personality traits is important. Moreover, a productive theory of the relationship between psychopathology and personality needs to account for instances of breakdown in the predictive power of normative personality for lifetime psychopathology.

The current study has several strengths. The study utilized a large sample of individuals with SMI to assess associations between symptomatology and personality traits. These associations were examined using well-validated diagnostic and personality assessments—that relied on self-report of personality traits, interview-based ratings of current symptoms, and consensus ratings of lifetime symptoms. In other words, data stemmed from multiple sources of information assuring that associations were not driven by shared methodological variance or response biases that carried across instruments. Moreover, the MPQ-BF yielded validity scale scores and only participants with valid responses were included in analyses.

The present study also has limitations. Individuals prescribed psychotropic medications were not excluded from the study due to ethical considerations. Inclusion of individuals on psychotropic medications may have attenuated the severity of current SMI symptoms, consistent with BPRS-based ratings of low to moderate levels of current symptoms, or caused side effects (e.g., sedation, affective blunting) that may have also impacted symptom ratings. The fact that individuals were generally not in acute severe episodes of SMI, and the low to moderate severity of current symptoms still explained significant variance in their normative and schizotypal personality traits, makes these associations more compelling. Finally, the cross-sectional nature of the present study cannot fully address the vulnerability model of personality-psychopathology associations. Personality changes throughout adulthood are considered normative (Roberts et al., 2006) and additional longitudinal studies preceding illness onset are needed.

Conclusions

This study suggests that in SCZ and BD personality traits are more related to current clinical symptoms than a lifetime prevalence of symptomatology. Findings support the complication model of personality-psychopathology associations in severe psychopathology. Current levels of delusions and hallucinations (i.e., positive symptoms) and depression are reflected in personality assessments in individuals with SMI (SCZ and BD) and need to be considered when examining lifetime symptomatology-personality associations. Nevertheless, some schizotypal personality domains exhibit associations with lifetime psychopathology after accounting for the impact of current clinical states. The lack of associations between other personality traits and lifetime symptoms of severe psychopathology challenges models suggesting etiological contributions of normative personality traits to SMI. Still, longitudinal studies with repeated measures of symptoms and normative and schizotypal personality traits are needed to more completely understand the impact of current symptomatology on the assessment of trait-based characteristics in SCZ and BD, and to determine whether questionnaire-based personality measures can reflect important contributors to these forms of psychopathology.