CBT interventions for caregivers and their effects on psychophysiological

Regarding CBT for caregiver’s stress, the literature is vast, particularly studies addressing caregivers of patients with progressive brain disorders such as Alzheimer’s disease. These studies agree that interventions reduce psychological distress. However,

25

caregivers of patients with brain disorders present psychophysiological alterations due to chronic stress, in particular excessive cortisol release (Bauer et al., 2000). Increased release of cortisol may lead to endocrine problems and decreased immune activity, which may subsequently impair the capacity of caregivers to keep up with their caregiving role. We were therefore interested in assessing effects of a CBT group intervention on cortisol patterns. Two previous studies performed similar research (Vedhara et al., 2003; V. P. Williams et al., 2010), but no significant changes were observed in salivary cortisol following the CBT treatment. However, in the study of Williams and collaborators (2010) the intervention was delivered through videos providing coping skills, which may not have the same effects as interactive interventions. Vedhara (2003) used a CBT group approach, but assessment of salivary cortisol did not take account of individual circadian variations with respect to time of awakening, which is of particular importance when examining patterns related to stress.

Therefore, in our study 4 we compared the effects of a CBT group intervention to a structured psychoeducation program on self-perceived a psychophysiological parameters of stress, by respecting circadian patterns of cortisol level.

26 III. RESULTS (Research articles)

Four different studies constitute this thesis. Three of them focus on patients with brain injury and one on caregiver’s stress.

# Study Design Status

(July 2015)

1

Emotion regulation after traumatic brain injury: distinct patterns of sympathetic activity during anger expression and recognition

Experimental Published

2 Feasibility and initial efficacy of a cognitive-behavioural group programme for managing anger and aggressiveness after traumatic brain injury

Clinical trial Published

3

A controlled trial assessing the effects of an anger management program following acquired brain injury

Clinical trial Under review

4

Cognitive-behavioural group therapy improves a psychophysiological marker of stress in caregivers of patients with Alzheimer’s disease

Clinical trial Published

Emotion regulation after traumatic brain injury: distinct patterns of sympathetic activity during anger expression and recognition

Abstract

Objective: To assess psychological and psychophysiological correlates of emotion recognition and anger experience in participants with traumatic brain injury (TBI).

Participants and design: 20 participants with TBI presenting anger problems and 22 healthy controls participated in tasks assessing emotion recognition (The French Evaluation Task) and anger expression (Anger regulation task). The latter was designed to elicit and modulate anger feelings through verbal recall of a self-experienced event. It involved four recall conditions that followed a resting period: neutral, uninstructed anger recall, anger rumination and anger reappraisal. Measures: Skin conductance levels during recall and a self-report anger questionnaire between each condition. Results: In the TBI and control groups, self-reported anger was similarly modulated across emotion regulation conditions. However, only in the TBI group skin conductance levels significantly increased between neutral and uninstructed anger recall conditions. Conclusions: Impaired emotion regulation in TBI participants could be related to increased levels of autonomic system activity during emotional experience.

However, anger feelings in these participants can also be modulated with the use of emotion regulation strategies, including adaptive strategies such as reappraisal. Thus, promoting awareness and management of physiological activation, and encouraging cognitive restructuring can be recommended for interventions targeting emotion regulation in TBI patients.

Keywords: Anger, Traumatic Brain Injury, Emotion regulation, STAXI, Skin Conductance

*This study is a reprint of the article: Aboulafia-Brakha, T.; Allain, P.; Ptak, R. (2015) Emotion regulation after traumatic brain injury: distinct patterns of sympathetic activity during anger expression and recognition. Journal of Head Trauma Rehabilitation (in press).

28 1. Introduction

Patients with traumatic brain injury (TBI) of varied severity frequently present impaired emotion recognition and emotional dysregulation (Dikmen et al., 2010; McDonald, 2013; Metting, Spikman, Rodiger, & van der Naalt, 2014; Prigatano, 1992). Both factors are predictors of poor social and professional integration (Alderman, 2003; Knox & Douglas, 2009; O'Connor et al., 2005). While in the past decade emotion recognition has increasingly been examined in TBI patients (McDonald, 2013), psychological and psychophysiological correlates of emotion regulation have been addressed to a lesser extent.

TBI patients show impaired identification of negative emotions (Babbage et al., 2011;

McDonald, 2013) and emotional prosody (McDonald, 2013; McDonald et al., 2013).

Compared to healthy controls they do less well when asked to adopt anger postures (Dethier et al., 2013) or to mimic angry facial expressions (McDonald, Li, et al., 2011). When observing stimuli portraying anger, they exhibit lower skin conductance levels (SCL) than controls, which is a potential physiological marker of poor emotion recognition (de Sousa et al., 2010; Hopkins et al., 2002). Tasks using stimuli that integrate dynamic facial expressions, prosody and contextual information facilitate emotion recognition, but TBI participants still perform worse than controls (McDonald et al., 2006; McDonald, Flanagan, Rollins, & Kinch, 2003). Impaired abilities to decode emotion in others not only preclude the generation of socially adapted behavioral responses in TBI patients, but may also compromise recognition of one’s own affective states, which is a first step for appropriate emotion regulation (Spikman et al., 2013).

Emotion regulation relates to the ability to modulate subjective experience and expression of emotions (J. J. Gross, 2002). An emotional experience such as anger can be influenced by the way a situation is perceived (cognitive appraisal). In healthy participants, anger rumination – repetitive thinking about an emotional event without changing its initial appraisal – amplifies psychological and physiological correlates of anger (J. J. Gross, 2002;

Ray et al., 2008). In contrast, modifying negative thoughts in order to take a different perspective towards the event (reappraisal) diminishes angry feelings and decreases physiological arousal (Denson, Moulds, & Grisham, 2012; J. J. Gross, 2002). Reappraisal is considered a powerful emotion regulation strategy (J. J. Gross, 2002). Intervention studies with participants with TBI showed that anger management can be improved following specific psychotherapeutic programs (Aboulafia-Brakha, Greber Buschbeck, Rochat, &

29

Annoni, 2013; Hart et al., 2012; Novaco, 1976b; Walker et al., 2010) and some of these include cognitive restructuring.

On the other hand, cognitive and behavioral manifestations of anger are strongly influenced by physiological responses (Denson, 2013) and monitoring of sympathetic arousal is also an essential part of anger treatment (Novaco, 1976a). In healthy participants, sympathetic activity increases during the observation of anger-inducing clips and anger provoking situations that are artificially arranged in the laboratory, or when they recall past events that made them angry (Fernandez et al., 2012; Gerin et al., 2006; Marci et al., 2007;

Ottaviani, Shapiro, & Fitzgerald, 2011; Stephens et al., 2010). In a previous study, TBI participants felt less affected by anger-inducing clips and showed reduced SCLs when watching these clips, while they showed normal activity during neutral or pleasant clips (de Sousa et al., 2012). However, the observation of emotional situations does not necessarily imply a personal involvement and may therefore fail to induce arousal in clinical samples (Dixon-Gordon, Yiu, & Chapman, 2013; Limberg, Barnow, Freyberger, & Hamm, 2011;

Schmahl et al., 2004). In contrast, eliciting emotions by triggering personally relevant emotional contents, such as asking participants to recall an anger-inducing event, is more effective in inducing modulations of valence and arousal (Foster & Webster, 2001; Jallais &

Gilet, 2010; Marci et al., 2007). This procedure was applied in a pilot study with five participants with TBI (D. R. Neumann, Hammond, Norton, & Blumenthal, 2011). The authors assessed eye blink and skin conductance response following startle stimuli during recall of different emotional situations. While skin conductance was similarly modulated across emotions, participants showed differentiated eye-blink responses during anger scripts compared to fear and neutral scripts. However, as this was a preliminary study, there was no comparison a control group sample.

Here, we report an exploratory study that investigates to what extent self-reported anger and psychophysiological arousal can be affected by the recall of an experienced anger episode in TBI participants with anger problems and healthy controls. Further, we examined whether TBI participants are able to modulate angry feelings and physiological arousal by applying emotion regulation strategies. Finally, we contrasted physiological arousal during anger recall and while watching emotional clips. Based on the literature reviewed above we predicted that TBI participants would display lower SCL's than controls during emotion recognition and anger expression. Furthermore, on the basis of psychotherapeutic studies with TBI patients, we hypothesized that subjective emotional states of these participants could be

30

modulated by the use of different anger regulation strategies, and we expected physiological arousal to vary accordingly.

2. Methods

2.1 Participants

TBI out-patients were selected from the database of the Divisions of Neurology and Neurorehabilitation of the Geneva University Hospitals, where they had previously been hospitalized. The study was approved by the Ethical Committee of the Geneva University Hospital. Inclusion criteria were a) At least three months post-injury so patients would be in a post-acute stage and usually more aware of impairments, b) At most five years post-injury in order to limit variability regarding chronicity, c) age between 18 and 60 years and d) increased difficulties in managing anger following brain injury reported by the patient or family members. Patients with antecedents of neurological or psychiatric disease before the TBI, or cognitive impairment precluding the understanding of instructions (e.g., severe aphasia, impairment of vision or major memory problems) were excluded.

Twenty participants with TBI (17 males) and 22 healthy controls¹ (12 males) gave written informed consent to participate to this study. Participants were blind with regard to the study goals and hypotheses. The two groups did not significantly differ regarding age (Controls: Mean: 33.4± 8.5, TBI: Mean: 37.4± 12.6; t (40) = -1.23, ns) or education level (X²= 4.88, ns), but the proportion of males was higher in the TBI group (X²= 4.55, p= 0.05).

TBI patients were tested between 3 months and 3.8 years following their accident (mean:

18.8± 12.01). Motor vehicle accident was the most common reason of injury (65%). The remaining participants had injury from different origins, including fall, sports and assault.

Glasgow Coma Scale value upon admission ranged between 3 and 15 (Mean: 7.8± 4.3).

Applying Teasdale & Jennett’s (1974) criteria, injury was severe for 12 patients, moderate for three and mild for five. Brain lesions varied across subjects and included focal contusions, subarachnoid hemorrhages and diffuse axonal injury. In all cases frontal lobes were involved as documented by CT-scan or MRI.

1 Hospital staff recruited by advertisement. They did not report any particular anger management problem.

31 2.2 Material and procedures

In a single experimental session, participants completed an anger regulation task followed by a task assessing emotion recognition. For both tasks we computed behavioral data and SCL, which was continuously recorded. In order to better characterize anger problems in our sample, at the end of the session, participants received self-report questionnaires assessing anger expression and regulation in everyday life that were completed at home and returned on the following day.²

Physiological recording apparatus

For SCL data acquisition and extraction we used the data acquisition unit IWX 214 of the iWorx Systems (Inc., Dover, NH) and the Labscribe software with settings configured for galvanic skin response acquisition (GSR-A LS2). Two Ag/AgCl electrodes were attached to the volar surface of the index and the annular fingers of the non-dominant hand and connected to an amplifier (GSR amplifier). Participants were asked to rest until a stable SCL was

reached. Units are expressed in microsiemens(μS).

Anger regulation Task

Task Description. This task is based on previous work with healthy subjects (Denson, 2009; Denson et al., 2012; Fabiansson et al., 2012). Following three minutes of rest, participants were asked to give verbal reports (during three minutes) in each of four recall conditions: neutral (description of their apartment), uninstructed anger recall (description of a social, anger-inducing event), anger rumination (description of the same event while focusing on its emotional aspects) and anger reappraisal (description of the same event while focusing on the point of view of the other involved person). Precise instructions for the recall conditions are given in Appendix 1. For half of the subjects the order of the latter two conditions was reversed. Outcome variables. a) Behavioral: Following each recall period participants filled out the STAXI-State questionnaire (description below) in order to measure their current subjective level of anger. They then rested for 30 seconds before being given instructions for the subsequent conditions. Further, in order to control for possible differences related to the intensity of anger during the anger-inducing event, participants were asked to

2Some participants did not return the questionnaires. Out of 20 TBI patients, 17 filled-in the AQ-12 and 15 the STAXI and the MARS. The analysis of missing data (MCAR) reveals that these were missing completely at random.

32

rate retrospectively how angry they felt during the event on a scale ranging from 0 (not angry at all) to 10 (extremely angry). Based on the study of Fabiansson and collaborators (Fabiansson et al., 2012) only events associated to angry feelings greater than four were considered (one participant was excluded from the study as he could not report an episode that reached the minimum accepted intensity). b) Physiological: SCL was continuously recorded and triggers were set in order to precisely identify the beginning and the end of each of the five conditions (including rest and the four recall periods). Mean SCL of each condition were used for analyses.

The French Emotion Evaluation task (FEET)

Task Description. This task is a French adaptation of the emotion recognition subtask of The Awareness of Social Inference Test (TASIT) (McDonald et al., 2006). It consists of 35 video clips (duration between 8 and 28 seconds) in which actors portray basic emotions (anger, surprise, disgust, happiness, sadness and fear) or no emotion (neutral clips). After each clip, participants select from a list the emotion that they believe was portrayed in the clip.

Outcome variables. a) Behavioral: number of correct answers for each emotion (maximum score 5). b) Physiological: SCL was continuously recorded and triggers were set in order to precisely identify the beginning and the end of each clip. We considered for analyses mean SCL averaged for each emotion.

In order to compare SCL values under anger expression and anger recognition we computed relative SCL values. These were calculated by subtracting mean SCL values during neutral recall from mean SCL values during uninstructed anger recall (Anger regulation task) and by subtracting mean SCL values during neutral clips from mean SCL values during anger clips (FEET).

Additional measures

The Aggression Questionnaire- AQ-12 (Bryant & Smith, 2001; Buss & Perry, 1992):

composed of 12 statements (e.g., ‘I can’t help getting into arguments when people disagree with me’) for which participants indicate on a six-point Likert scale the extent to which a described behavior or feeling describes them. The total score (ranging between 12 and 68 points) is calculated by summing the points given for each question. Higher scores indicate higher levels of self-reported aggressiveness. In order to characterize the impact of TBI on behavioral changes we asked participants to rate how well each item characterized them both before the accident and currently. The AQ-12 shows acceptable levels of internal consistency and good test-retest reliability, with a Cronbach’s alpha coefficient of .80 (French version).

33

We used this same questionnaire in a previous studies with TBI participants (Aboulafia-Brakha et al., 2013).

State-Trait Anger and Expression Inventory-2- STAXI-2 (Borteyrou, Bruchon-Schweitzer,

& Spielberger, 2008; Spielberger, Gorsuch, & Lushene, 1970): STAXI-State (15 items):

Assesses currently experienced anger feelings (e.g., ‘I feel irritated’ or ‘I feel like screaming’). Each item is rated on a four-point Likert scale ranging from ‘Not at all’ to

‘Absolutely’. The total score is calculated by summing the points given for each statement and may vary between 15 and 60; STAXI-Trait (10 items) evaluates anger as a stable characteristic in terms of how much and how often angry feelings are generally experienced by the participant in everyday life (e.g., ‘I get angry very easily’). Each item is scored on a four-point Likert scale and the total score (ranging between 10 and 40) is computed by summing the points; Anger Expression (32 items) characterizes how anger feelings or related behaviors are expressed and controlled. This scale has four subscales, for which total scores (sum of the points in the four-point Likert scale) are calculated separately: Anger Expression-Out refers to the extent to which emotional experience of anger is outwardly expressed in a negative manner; Anger Expression-In relates to anger feelings that are ‘held in’ (not shown), even if the person feels furious; Anger control-Out refers to the extent to which the person controls himself or herself to prevent explosive manifestations; Anger Control-In measures how often the person tries to implement relaxation strategies to calm down before losing control. We also adapted the scale (subscales trait and expression) so participants could rate to what extent each item characterized them before the TBI and also currently. STAXI-2 shows acceptable levels of internal consistency and good test-retest reliability and was previously used in studies with participants with TBI (Hart et al., 2012; Walker et al., 2010). Cronbach’s alpha coefficient for the French version ranges between .72 and .85 depending on the subscale.

Multidimensional Anger Reaction Scale, Reaction subscale (MARS; French version)(Recchia, Steffgen, Weber, & Kubiak, 2010): Composed of 28 items assesses anger regulation strategies commonly used following an anger-inducing event. It focuses on seven main reactions: Venting, Rumination, Submission, Feedback, Distraction, Putting in perspective and using Humour (e.g., a Rumination item, ‘I don’t easily forget other’s behavior. I think about it continuously’). Each of them is assessed through four different questions rated on a four-point Likert scale ranging from 1 (‘never’) to 4 (‘always’). Total

34

scores are calculated separately for each sub-scale and may vary between 4 and 16. This instrument shows acceptable levels of internal consistency and good test-retest reliability with Cronbach’s alphas ranging between .69 and .83.

2.3 Statistical Analysis

Between-group comparisons of demographic data were performed using the Chi-square test for frequency and ordinal data, and Student’s t-tests for continuous data.

Experimental outcomes: 1) Anger regulation task: behavioral data and mean SCL's were submitted to a 2 x 5 repeated measures ANOVA (rANOVA) with group (TBI, Control) and condition (rest, neutral, uninstructed, rumination and reappraisal) as factors. In order to control for gender effects we also performed the same analyses using gender as a covariate. F and p values were adjusted (Greenhouse-Geisser) if data sphericity was violated. Significant interactions were followed-up with independent t-tests for between-group comparisons and paired t-tests for within-group contrasts. Order effects were examined with a 2 x 2 x 2 rANOVA, with group, order (rumination first, reappraisal first) and condition (rumination, reappraisal) as factors. 2) FEET: Correct answers and SCL were analyzed with a 2 X 7 rANOVA, with group (TBI, Control) and emotion (anger, surprise, happiness, disgust, sadness, fear and neutral) as factors. In order to investigate the relations among anger coping style in everyday life, STAXI-State scores following uninstructed recall and emotion recognition abilities we performed bivariate correlational analyses for each group using Pearson’s correlation coefficients. For all cases significance levels were set to p < 0.05.

Post-hoc Power analyses on experimental data were run using G*Power 3.1. For rANOVAs, achieved power and effect sizes (Cohen’s f with 0.1= small effect, 0.25=

moderate and 0.5= large) were computed separately for between group effects, within group effects and interactions when results were statistically significant. All effect size computations considered the correlation between repeated measures and involved a correction of sphericity.

Effect sizes (Cohen’s d) were also computed for paired contrasts when appropriate.

35 3. Results

3.1 Self-reported anger in everyday life

Ratings for self-report questionnaires and statistical outcomes are provided in Table 1.

TBI participants reported current levels of anger and aggressiveness to be significantly higher than before the accident, as indicated by the AQ-12 and the STAXI-Trait. In addition, they reported significantly higher anger expression on the subscale anger-expression out and lower control of externalized anger in the subscale Anger-Control-Out. In contrast, no significant differences were reported for internalized anger feelings, whether for Anger-Expression-In or Anger-Control-In.

Self-reported levels of anger and aggressiveness were significantly higher in the TBI group than in controls, as revealed by the AQ-12 and the STAXI-Trait. Self-reported levels of

Table 1. Self-reported anger in everyday life

TBI TBI Controls

*p < 0.05, **p < 0.01, ^a Significance levels indicated by stars reflect within group differences

*p < 0.05, **p < 0.01, ^a Significance levels indicated by stars reflect within group differences