Do Patients’ Cognitive Therapy Skills Predict Personality Change During Treatment of Depression?

Abstract

Background

Psychological interventions can change personality, including increasing positive temperament (extraversion) and decreasing negative temperament (neuroticism), but why these changes occur is unclear. The current study tested the extent to which patients’ acquisition and use of skills taught in cognitive therapy (CT) correlated with changes in positive and negative temperament during treatment of depression.

Method

Outpatients (N = 351) with recurrent major depressive disorder (MDD) were enrolled in a 12-week CT protocol. Temperament (early and late in CT), patient skills (mid and late in CT), and depressive symptoms (early, mid, and late in CT) were measured repeatedly.

Results

Patients with greater acquisition and use of CT skills showed significantly larger increases in positive temperament and larger decreases in negative temperament in path analyses. Effect sizes were small, median standardized |beta| = .13. Models controlled depressive symptom levels and changes.

Conclusions

Skills taught in CT for recurrent depression correlate with personality change during this efficacious treatment. The absence of measures of CT skills at baseline and personality mid-CT allows several interpretations of the current findings. Future research is needed to clarify whether patients’ use of CT skills facilitates adaptive changes in personality during CT.

Personality may be conceptualized as relatively stable patterns of thought, behavior, and emotion. Nonetheless, a growing research literature suggests that personality can change over the lifespan, as well as during shorter periods of psychological intervention (Barlow et al., 2014; Roberts et al., 2017). The purpose of the current study was to test the extent to which personality change in a well-established intervention, cognitive therapy (CT) for depression, relates to skills that patients learn and practice during this treatment.

During a study of CT for recurrent depression (N = 351; Vittengl et al., 2014), all scales on the Schedule for Adaptive and Nonadaptive Personality-2^nd edition (SNAP-2; Clark, Simms, Wu, & Casillas, 2014) changed significantly. For example, negative temperament (d = 0.86) decreased whereas positive temperament (d = 0.76) increased. The relatively large changes in negative temperament (neuroticism) and positive temperament (extraversion) found by Vittengl et al. (2014) mirrored more general findings of a meta-analysis (Roberts et al., 2017). In the meta-analysis, personality traits, particularly neuroticism/negative temperament and extraversion/positive temperament changed robustly across a range of interventions and disorders. Personality changed more on average in interventions (e.g., medication, CT, psychodynamic psychotherapy) than in comparison (e.g., wait-list, placebo) groups, suggesting that observed personality change was at least partly the result of intervention.

Past research suggests that personality change during depression treatment is related to decreases in depressive symptoms. For example, compared to pill placebo, CT or medication for depression produced greater changes in neuroticism and extraversion, and changes in depressive symptoms at least partly accounted for changes in personality (Tang et al., 2009). Similarly, research on antidepressant medication (Mulder et al., 2010) and medication plus psychotherapy (Kool et al., 2003) for MDD showed that personality pathology decreased with treatment, both for treatment non-responders and more so for responders. These clinical findings are consistent with factor-analytic research showing that depressive symptom decreases align more closely with state than with trait variance in personality measures (Clark, Vittengl, Kraft, & Jarrett, 2003; Vittengl, Clark, Thase, & Jarrett, 2014).

Why personality changes during psychological intervention is unclear and may vary by type of intervention or treatment. In addition to changes in depressive symptoms, a possible mechanism of personality change in CT is the development and use of skills taught in this efficacious treatment. Skills developed in CT may help patients recognize and change relations among emotion, thought, and behavior to reduce emotional distress and depressive symptoms. For example, CT patients learn to test the validity of negative thoughts during affect shifts, replace unsupported negative thoughts with more realistic ones, initiate problem solving, and engage in reinforcing activities to decrease negative affect and to increase positive emotionality (Jarrett, Vittengl, Clark, & Thase, 2011; Strunk, DeRubeis, Chiu, & Alvarez, 2007). The skills that patients learn in CT may mediate changes in depressive symptoms and reduce the impact of adverse life events on treatment outcomes (Jarrett, Vittengl, Clark, & Thase, 2018; Vittengl, Stutzman, Atluru, & Jarrett, 2020).

In this context, we examined whether acquisition and use of CT skills correlates with change in personality during treatment for depression, using the same clinical trial dataset (Jarrett & Thase, 2010) as Vittengl et al. (2014). We hypothesized that patients with greater CT skills experience larger decreases in negative temperament and larger increases in positive temperament.

Method

Participants

Data were drawn from the non-randomized acute phase of a multi-phase clinical trial (Jarrett & Thase, 2010), during which all participants received cognitive therapy (CT). Institutional review boards at the University of [MASKED] and University of [MASKED] approved the clinical trial. Participants provided written informed consent for assessment and treatment. We focus here on the subset of patients (N = 351) who entered the trial after the SNAP-2 was added to the assessment battery. Inclusion criteria were (a) provided written informed consent for evaluation and treatment; (b) met DSM-IV criteria for recurrent MDD (American Psychiatric Association, 2000); (c) remitted between major depressive episodes (periods of ≥ 2 months with minimal or absent symptoms) or had dysthymic disorder preceding the onset of MDD; and (d) scored ≥ 14 on the 17-item Hamilton Rating Scale for Depression (HRSD; Hamilton, 1960).¹ Exclusion criteria were (a) severe or poorly controlled concurrent medical disorders that could cause depression, (b) psychotic, bipolar, or organic mental disorders, active substance dependence, or primary obsessive-compulsive or eating disorders, (c) could not complete questionnaires in English, (d) active suicide risk, (e) age < 18 or > 70 years old, (f) history of non-response to ≥ 8 weeks of CT or 6 weeks of fluoxetine, or (g) pregnant or planning to become pregnant during the first 11 months of the study. Psychiatric diagnoses were made using the Structured Clinical Interview for DSM-IV (First, Spitzer, Gibbon, & Williams, 1996). The 351 patients analyzed here were M = 43.3 (SD = 12.4) years old with M = 15.0 (SD = 3.0) years of education; 67.8% were women and 80.3% were white. Participants’ mean age of MDD onset was 21.3 (SD = 11.1) years, and they had experienced ≥ 2 (median = 4) major depressive episodes.

Cognitive Therapy

Participants discontinued any psychotropic medications with physician supervision before, and were not prescribed antidepressant medications during, the acute phase. Therapists (N = 16) provided acute-phase CT as described by Beck et al. (1979). Cognitive therapists achieved mean Cognitive Therapy Scale (CTS; Young & Beck, 1980) scores ≥ 40 to demonstrate competence, submitted session recordings for review, participated in weekly group supervision, and received feedback on strengths and weaknesses. The CT protocol was 12 weeks, with 2 additional weeks allowed for rescheduling, and included 16 or 20 sessions. Patients received 2 sessions/week for 4 weeks. Then, patients with ≥ 40% reduction in HRSD scores received 8 additional weekly sessions (16 total sessions), whereas patients with less improvement continued to have 2 sessions/week for 4 weeks followed by 4 weekly sessions (20 total sessions). Patients with less initial improvement received more sessions to increase their chances of response. Among the 351 patients analyzed here, 197 responded to acute-phase CT (no major depressive episode and final HRSD ≤ 12) and 154 did not respond. Both responders and non-responders were included in all of the current analyses.

Measures

The current analyses focused on measures taken during the first, seventh, and last weeks of the 12-week CT protocol. We term these periods early, mid, and late CT, respectively.

Personality

Patients completed the Schedule for Non-adaptive and Adaptive Personality-2^nd Edition (SNAP-2; Clark et al., 2014) early and late in A-CT. The SNAP-2 contains 390 true-false items and measures 15 dimensions relevant to normal and disordered personality. The negative temperament, positive temperament, and disinhibition scales anchor the SNAP-2’s Big Three factor structure. The SNAP-2 scales have demonstrated good reliability (alpha internal consistency median =.79 to .92 in student, adult, and patient samples; retest r mean = .87 over 1 week to 4.5 months in normal adults; Clark et al., 2014). Multiple studies support the validity of the scales’ scores (e.g., Morey et al., 2012; Pryor, Miller, & Gaughan, 2009; Ready & Clark, 2002). The negative and positive temperament scales were the focus of the current hypothesis tests.

Cognitive therapy skills

Patients and their therapists completed the 8-item Skills of Cognitive Therapy scale (SoCT; Jarrett et al., 2011) mid and late in CT. The measure’s items assess CT skill acquisition and use. Items are rated on a 5-point frequency scale. In support of the reliability and validity of the SoCT, patient (.67) and therapist (.72) ratings demonstrated moderately high retest correlations from mid to late CT, converged with one another mid (.43) and late (.44) in CT, and predicted (rs .17-.35) categorical CT response (no major depressive episode and HRSD score ≤ 12 at exit) in the current dataset (Jarrett et al., 2011). The current analyses focused on an aggregate CT skills score, averaging across 16 total patient and therapist ratings to produce a reliable scale mid (alpha = .89) and late (alpha = .92) in CT.

Depressive symptoms

Patients completed the 21-item Beck Depression Inventory (BDI; Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) and 30-item Inventory of Depressive Symptomatology--Self-Report (IDS-SR; Rush, Gullion, Basco, Jarrett, & Trivedi, 1996), and clinicians completed the 17-item HRSD. These measures’ scores have well-established validity, and replicated research suggests that the HRSD, BDI, and IDS-SR mark the same depressive symptom severity construct during CT (Vittengl, Clark, Kraft, & Jarrett, 2005; Vittengl, Clark, Thase, & Jarrett, 2013). For example, the measures are highly intercorrelated both concurrently (median r = .81) and longitudinally (median r = .81) and show similar mean-change trajectories during treatment (Vittengl et al., 2013). Consequently, we formed a robust symptom composite by standardizing the measures (M = 50, SD = 10) using their distributions at study intake and then averaging them. Treating the three scales as items, alpha internal consistency for the composite was high (median = .94, range .87-.95) early, mid, and late in A-CT.

Statistical Analyses

Analyses were intent-to-treat and included all patients who entered the clinical trial after the SNAP-2 was added to the assessment battery (N = 351). Table 1 shows descriptive statistics for study measures. Path analyses using full information maximum likelihood included all observations and participants in hypothesis tests. Separate path models were estimated for positive and negative temperament (see Figures 1 and 2). In Figures 1 and 2, single-headed arrows represent forward-predictive paths, and double-headed arrows represent concurrent correlations. For example, in Figure 1, we predicted late-CT negative temperament from early-CT negative temperament, mid-CT depressive symptoms, and mid-CT skills, while controlling for concurrent relations with late-CT depressive symptoms and late-CT skills. Thus, this part of the model tested whether changes in negative temperament were related to mid-CT skills, separately from depressive symptom severity. The threshold for statistical significance was set at p < .05, two-tailed.

Table 1.

Descriptive Statistics for Study Variables

Scale	Period in CT	N	M	SD
Depressive symptom severity	Early	351	47.77	11.27
	Mid	293	27.14	13.87
	Late	264	20.01	13.64
Negative temperament	Early	324	62.37	7.84
	Late	229	56.20	9.77
Positive temperament	Early	324	33.65	9.96
	Late	229	38.86	11.60
Patient CT skills	Mid	296	3.22	0.60
	Late	273	3.45	0.69

Note. Early, mid, and late refer to approximately the first, seventh, and last week of the 12-week cognitive therapy (CT) protocol, respectively.

Figure 1.

Path model of relations among negative temperament, depressive symptoms, and patient skills during acute-phase cognitive therapy (CT) for recurrent depression. Early, mid, and late CT refer to the first, seventh, and last weeks of the 12-week CT protocol, respectively. Standardized path coefficients relevant to hypotheses appear in bold type. Additional paths of early depressive symptoms to late negative temperament (−.06) to late patient skills (.13*) not depicted.

* p < .05, two-tailed.

Figure 2.

Path model of relations among positive temperament, depressive symptoms, and patient skills during acute-phase cognitive therapy (CT) for recurrent depression. Early, mid, and late CT refer to the first, seventh, and last weeks of the 12-week CT protocol, respectively. Standardized path coefficients relevant to hypotheses appear in bold type. Additional paths of early depressive symptoms to late positive temperament (.17*) to late patient skills (.13*) not depicted.

* p < .05, two-tailed.

Results

The hypothesis was tested with path analyses depicted in Figures 1 (negative temperament) and 2 (positive temperament). Changes in temperament from early to late in CT were modeled as a function of CT skills assessed mid and late in CT, while controlling for changes in depressive symptoms and possible contributions of early-CT/baseline symptoms and personality to the development of CT skills. The models fit well for negative and positive temperament, as reflected in goodness of fit, comparative fit, and non-normed fit index values of 1.00.

In Figures 1 and 2, the paths between mid- and late-CT patient skills and temperament are of primary interest for hypothesis tests. The standardized path coefficients in the Figures can be interpreted on the effect size r metric. In support of the hypothesis, patients with greater skills mid-CT had larger changes in negative (−.12) and positive (.13) temperament, ps < .05. Also consistent with the hypothesis, late-CT patient skills correlated with changes in negative (−.16) and positive (.13) temperament, ps < .05.^{2, 3, 4}

These models showed that patients with greater CT skills showed larger decreases in negative temperament and larger increases in positive temperament, from early to late in CT. The observed magnitude of potential effects of CT skills on personality change was small, median standardized |beta| = .13, range .12-.16.

Discussion

Personality change may be central to efficacious treatment of depression with psychotherapy (Barlow et al., 2014). The current study tested whether patients’ acquisition and use of CT skills correlates with personality change during treatment of recurrent MDD. As hypothesized, we found that patients with greater CT skills showed larger reductions in negative temperament/neuroticism and larger increases in positive temperament/extraversion. Analyses controlled for depressive symptom severity change to help rule out transient distress as a third variable linking CT skills with personality change. These findings suggest that use of skills taught in CT may be one reason why personality changes during psychological interventions (Roberts et al., 2017). Observed effects sizes were small, however, indicating that additional mechanisms of personality change during CT are also likely.

Although specific mechanisms linking CT skills with personality change were not identifiable in the current study, the process of CT and content of personality inventories hint at possible connections. For example, perhaps acquiring and using CT skills ameliorates maladaptive patterns of thought, behavior, and emotion related to patients’ self-view, as reported on personality inventories. Consequently, CT patients may find that altering their behavior (e.g., engaging in work and social activities with high potential for reinforcement or solving chronic problems) and thought (e.g., challenging and replacing negative views of the self, world, and future that do not match objectively evaluated evidence) increases positive and decreases negative affect (Beck et al., 1979), which are central to positive and negative temperament, respectively (Watson & Clark, 1992). Through their efforts, patients may also realize that emotion is changeable and manageable. Consequently, patients may no longer view themselves as “the kind of person who is depressed” but instead as persons with skills to help regulate their mood, modify their life circumstances, or perceptions thereof. Personality measures may tap these emotional and behavioral experiences, plus related views of self, that improve with implementation of CT skills. Of course, our speculations require future empirical clarification.

The clinical trial (Jarrett & Thase, 2010) providing data for the current analyses was not designed to test relations between CT skills and personality change and, consequently, several design features make the current findings preliminary. First, our sample had carefully diagnosed recurrent MDD and was treated by well-trained and supervised cognitive therapists in a research protocol. Generalization of the current findings to other patient populations, treatments, and contexts is uncertain.

Second, CT skills were not measured at the beginning of treatment, and personality was not measured at mid CT, to limit patients’ assessment burden. Thus, increases in skills from early to mid CT, and personality changes from mid to late CT, are unknown. The lack of full synchronization of assessments of personality (early and late CT) and skills (mid and late CT) also limited tests of potential reverse causality (i.e., personality change driving subsequent change in skills). An imperfect test using available assessment points did not find evidence for reverse causality, however. More broadly, the timing of personality and skills assessment did not allow us to examine the precise sequencing of skills development and personality change during CT. Future research with more frequent measurement might clarify when in CT greater personality change occurs and which skills are being learned and used at those times using, for example, bidirectional growth models.

Thus, the current results are consistent with alternative interpretations of the relations between CT skills and personality change. For example, CT skills may reduce depressive symptoms (cf. Clark et al., 2003; Vittengl et al., 2014) or depressive cognitive content (e.g., dysfunctional attitudes, automatic thoughts, hopelessness; Jarrett et al., 2011) which, in turn, changes self-views expressed in personality assessments. It is also possible that the directions of relations are reversed such that changes in CT skills are consequences rather than antecedents. That is, changes in depressive symptoms or personality may precede and drive subsequent changes in reported CT skills. Assessment of patients’ CT skills before treatment in future research would facilitate comparisons of how development of skills versus stable individual differences in skills relate to personality change.

Finally, personality was assessed with a well-established self-report inventory (Clark et al., 2014), but other methods of personality assessment are available (e.g., informant reports or clinician interviews). Past research suggests that self-report personality measures (e.g., of neuroticism/negative temperament and extraversion/positive temperament dimensions) often provide information that does not differ systematically from (e.g., Kim et al., 2019) and converges highly with (e.g. Oltmanns et al., 2019) informant reports. However, some self-reports may be less differentiated (i.e., suggest higher comorbidity of personality disorders) than structured interviews (Friborg et al., 2014). Moreover, personality assessment with self-reports, informant reports, and clinical interviews may each offer complementary information that is incrementally valid in some domains (e.g., prediction of depression; Galione & Oltmanns, 2013). Whether CT skills predict change in personality assessed with methods other than self-report is unknown and a topic worthy of future research. Similarly, the current depression symptom severity composite measure may not have captured all aspects of depression relevant to CT skills and personality change. Consequently, we cannot rule out the possibility that changes in depressive symptoms account for the relation of CT skills to personality change observed here. We do note, however, that the reported results were robust across three measures of depressive symptoms, including patient-report and clinician-report measures.

Future research is needed to identify specific mechanisms through which CT skills may influence personality change, as well as to discover or rule out unknown third variables (including what type of rater is assessing CT skills and personality) that may account for observed relations between CT skills and personality change. Effect sizes observed here for associations between personality change and mid-CT skills were small but similar to mid-CT skills’ prediction of treatment response (Jarrett et al., 2011), indicating that a number of important variables beyond skills measured by the SoCT may be operating during CT.

Both affect and personality change during psychotherapy. This study may be the first to establish that personality change during CT for recurrent MDD is related to acquisition and use of skills taught during this treatment. Thus, our findings suggest broadly that CT skills are a possible means by which personality changes during CT. Whether this potential mechanism can be clarified and amplified to improve theories of depression and treatment outcomes will be important topics for future research.

• Cognitive therapy (CT) for major depressive disorder (MDD) is often helpful.

• Patients who learn and use CT skills tend to have better CT outcomes.

• Patients’ positive and negative temperament may also change during CT.

• We tested whether patients’ CT skills correlated with favorable personality changes.

• Better skills correlated with increased positive and decreased negative temperament.