Modelling change in multivariate depression symptoms adjusting for gender and baseline temperament and character traits: a Latent Transition approach

Abstract

Personality is linked to mental illness. Data on 9 symptoms of psychological distress items of the Symptom Checklist (SCLs) (Derogatis, 1983) and 7 temperament and character traits (TCIs) of Cloninger (2008) were analysed from patients measured pre and post treatment from the NZ Christchurch Psychotherapy of Depression Study (Joyce et al. 2002; Turner, Hudson et al., 2003). Latent transition analysis (LTA) was used to investigate: How do depressive symptom patterns change over time pre and post treatment; is this change different across gender and how is it impacted by baseline temperament and character traits (TCIs). The LTA model found 4 latent classes based on the multivariate symptom response profiles of 9 SCLs. The 4 patient classes represent different levels of symptom risk whose interpretation varies across gender. Latent class 1 (C1), a high risk group was characterised by high probabilities of elevated scores on all 9 SCL items. Class 2 (C2) a low risk group was characterised by a low risk of high scores on all SCLs. In fact the likelihood of scoring above the median for depression, phobic anxiety and anxiety was zero for both sexes in C2. Class 4 males were characterised by very high depression, interpersonal sensitivity (IS), psychoticism (P) and OC, intermediate anxiety (A), phobic anxiety (PA) and paranoid ideation (PI) but low scores on somatisation (S). In contrast C4 females had a high risk of elevated anxiety and P, intermediate OC, depression and S, but low scores on IS, anger hostility (AH), PI and PA. Class 3 females exhibited very high IS, high P, intermediate AH, PI and depression risk, and low scores of (phobic) anxiety. Class 3 males had very high P and intermediate S risk, and like C3 females low scores on IS, AH and PA. Anxiety differentiated females across C3 and C4. Apart from the high risk class, C1, only C4 females had increased anxiety risk. The latent classes seem to represent a scale of distress risk, suggesting a continuum of low to high risk, as opposed to qualitatively different groups. LTA transition probabilities pre to post treatment showed the males had a higher chance of remaining in the high risk class than females. For both sexes at high risk pre-treatment, 0% transitioned from C1 to C4 posttreatment. For females at high risk pre-treatment, 37% moved to low risk, but only 20% males moved from high to low risk. Neither gender transitioned from C4 to C3. Three times more males in C3 remained in C3 (very high P, intermediate somatisation, low scores on IS, AH and PA) compared to females. The proportion of males remaining in C4 (high depression, IS, psychoticism and OC, and intermediate A, PA and PI) was over twice the C4 female rate (high anxiety and P, intermediate OC and depression). No males starting in the low risk class transitioned to the high risk class, compared to 6% of females, whereas of females starting in the low risk group a quarter transitioned to C4. Three times more males starting in C3 transitioned to the high risk class than for C3 females, and more than twice as many males moved from C3 to C4. A quarter of females starting in C4 transitioned to the high risk class, but no C4 males moved to the high risk class C1 post-treatment. Two character traits, self-directedness (SD) and self-transcendence (ST) were highly significant predictors of latent class status. In general higher ST was associated with higher distress and higher SD with lower distress. The impact of ST and SD varied across gender in class 4 and to a lesser extent in the low risk class. There was an enhanced benefit of increased SD for both sexes, especially for females in C4 characterised by a high risk of anxiety and P, intermediate OC, depression and S, but low scores on IS, AH, PI and PA, and for males in the low risk class. Increasing SD did not benefit the C4 males characterised by very high depression, IS, OC, and P, intermediate phobic/anxiety and PI, but low somatisation. Similar for both sexes, high ST was associated with higher odds of distress, except for males in C4. Low risk females were less likely to be in the low risk class (vs high) for a one unit increase in ST, and C2 males had a slightly less reduced odds of low risk vs ‘high’ risk status. This work extends Leigh et al. (2012) and shows that 4 discernible classes differed across gender, and transitions between classes pre to post treatment differed across gender (after baseline TCI adjustment). This study adds credence to the need for gender-specific scores of psychological distress risk to be developed.