Effects of a trauma-sensitive yoga intervention on mental health of young incarcerated men : a pilot study

Effects of a trauma-sensitive yoga intervention on mental

health of young incarcerated men: a pilot study

Josephine Jankowski – 10202587

Master’s in Healthcare Psychology (Gezondheidszorgpsychologie)

Specialization Clinical Psychology

Faculty of Social and Behavioural Science

University of Amsterdam

First Supervisor: Johan Verwoerd

External Supervisor: Kristin Skotnes Vikjord

Date: 10

_{of July 2017}

Author Note

Special thanks to my supervisors Dr. Johan Verwoerd at the University of Amsterdam and clinical

psychologist Kristin Skotnes Vikjord for their expertise and guidance, and to the Prison Yoga Project team for their enthusiasm and support in realizing this research project.

Table of Contents

Abstract

3

1. Introduction

4

1.1 Yoga and Trauma

4

1.2 Emotion Regulation and the Yoga Black Box

6

1.3 Prison populations

9

2. Methods

11

2.1 Design and Setting

11

2.2 Participants

11

2.3 The yoga intervention

12

2.4 Materials

13

2.5 Procedure

14

2.6 Data Analysis

15

3. Results

15

3.1 Treatment dropout and number of attended classes

15

3.2 Pre intervention scores

15

3.4 Preliminary intervention effects

17

3.5 Qualitative data; applicability and experience of the yoga intervention

20

4. Discussion

21

4.1 Qualitative data in the current study

23

4.2 Limitations & Future Directions

24

5. References

26

Appendix A - Intake Form Instructors (Dutch)

31

Appendix B – Qualitative Assessment

32

Abstract

Mindfulness-based practices like yoga have been associated with physical and psychological improvement, but as for now the literature on the benefits of yoga interventions is only little. It has been suggested that the implementation of yoga within forensic settings may be particularly favourable because of its applicability to trauma-related psychopathology. In collaboration with Prison Yoga Project Netherlands, this pilot study investigated the effects of a 12-week trauma-sensitive yoga intervention on psychological outcomes of nine incarcerated young men. A single group design with pre, mid and post intervention assessments was completed containinga battery of self-report questionnaires. Qualitative data was obtained at mid and post intervention. Descriptive analysis of quantitative data revealed preliminary beneficial trends in scores for Trait Mindfulness, Emotion Regulation, Sleep problems, Hostility, Anxiety, Depression and PTSD-symptoms from pre to mid intervention; however, except for depression and PTSD-symptoms, these trends were not maintained until post intervention.Interview data demonstrated key themes including increased body awareness, stress relief and improvement in stress management. Results suggest that trauma-sensitive yoga is a feasible intervention for young male prisoners and is possibly beneficial for some aspects of

psychological functioning. This study was limited due to its small sample size and inability to use inferential statistics. Findings are promising and encourage further research into the application of yoga within forensic settings. Future studies should look at yoga programs of longer duration, and may benefit from continuity and group stability.

1. Introduction 1.1 Yoga and Trauma

Yoga, a comprehensive system that incorporates physical postures (asanas), breathwork (pranayama), and mindfulness or “the meditation practice of non-judgemental and moment-to-moment awareness” (Kabat-Zinn, 1994), has long been of interest as a therapeutic intervention. The term mindfulness derives from Buddhist meditation and has been brought into scientific perspective by authors like Kabat-Zinn (et al., 1985), Williams and Teasdale (Segal, Teasdale & Williams, 2004), who looked at the effects of mindfulness-based stress reduction (MBSR) and mindfulness-based cognitive therapy (MBCT) on physical and mental health. In accordance with the growing popularity of Eastern spiritual practices in Western modern societies, healthcare research on mindfulness and yoga has gained ground in Western scientific community. Over the past decades the empirical literature has suggested a

number of areas where yoga and other mindfulness-based interventions may be beneficial, either as additional or as primary treatment. These areas include depression, substance abuse, anxiety, stress, and chronic pain or fatigue (Büssing, Michalsen, Khalsa, Telles & Sherman, 2012; Baer, 2003).

The development of mental health problems results from a complex interplay between an individual’s susceptibility (expressed in biological and socio-psychological factors), adverse life events, and coping capabilities. Within this framework of complex underlying mechanisms and triggers, associations between trauma and psychopathology have been well established (e.g. Pine & Cohen, 2002). One amongst many definitions of trauma states that it concerns events that are extremely upsetting or stressful and at least temporarily overwhelming to the individual’s inner resources (Briere & Jordan, 2004, as cited in Telles, Singh and Balksirhna, 2012). Examples of such events include child maltreatment, exposure to warfare, robbery or physical abuse. Studies on adverse life events have estimated that in young adult and adult populations 43% to 89% have been exposed to at least one potentially traumatic event (Kilpatrick et al., 2013; Smyth, Hockemeyer, Heron, Wonderlich &

Pennebaker, 2008; see De Vries & Olff, 2009 for estimations in regards to Dutch populations). One has to note that there is a thin semantic line between the words adverse and traumatic, and that the events used in these studies do not necessarily lead to psychotraumatization or clinical diagnoses like PTSD. In fact, the literature shows that large numbers of people manage to endure the distress of adverse events remarkably well, implying that resilience in general populations is quite high (Bonanno, 2004).

Nonetheless, these percentages are worrisome and should be taken into consideration when it comes to mental health diagnostics and mental health care.

Relatively high rates of trauma are found in forensic populations where patients do not only report a higher number of traumatizing experiences compared to non-clinical populations, they also have a higher risk ofdeveloping psychological disorders such as PTSD and depression or the comorbid occurrence of different psychopathologies (Fazel & Danesh, 2002; Henrichs & Bogaerts, 2012;

Kristiansson, Sumelius & Sondergaard, 2004). Even though it is important to keep in mind that some pathologies (e.g. behavioural issues) can already be present before the trauma occurs, the consequence

of trauma on mental health may be manifested in both internalizing problems like depressive moods, social isolation, somatization, fear or inhibition, and externalizing problems like aggression, anger or disobedience (APA, 2013). In line with the above, studies have shown that early exposure to trauma can cause an individual to engage in high-risk behaviour, and that childhood trauma is associated with delinquency, substance abuse and criminality later in life (Fairbank, Putnam & Harris, 2007;Kaplan, Pelcovitz & Labruna, 1999). Furthermore, correctional incarceration inherently brings about traumatizing experiences (e.g. social isolation and confinement), which means traumas are often cumulative and complex in prisoners. Thus, due to this compounding prison environment, mental health problems increase in complexity, which influences treatment response negatively (Courtois, 2008). Accordingly, treating mental heath problems within forensic settings is found to be relatively difficult (Lamb, Weinberger & Gross, 1999), and the need for supplementary treatment options in forensic settings has already been identified by numerous authors (Duncan, Nicole, Ager, Dalgleish, 2006).

The need for supplementary intervention is also apparent within non-forensic settings, as current treatments of PTSD including EMDR and cognitive behavioural approaches are not sufficient for many patients given the high rates of incomplete response. Bradley et al. (2005) reviewed the efficacy of various forms of psychotherapy including exposure-based therapies, cognitive behaviour therapy, exposure and EMDR on patients with PTSD. Even though the results showed that 56% to 67% of participants no longer met criteria for PTSD post treatment, the majority of patients continued to have substantial residual symptoms. Studies have shown that yoga may be a beneficial alternative for or addition to treating patients with mental health disorders resulting from trauma. Telles, Singh and Balksirhna (2012) reviewed the efficacy of yoga on patients with physical health problems, anxiety, depression and post-traumatic stress disorder (PTSD), and found that yoga has beneficial potential but that studies often lack methodological quality. A more solid randomized controlled trial (RCT) by Van der Kolk et al. (2014) showed that yoga is able to establish clinically significant changes in PTSD symptoms with effect sizes comparable to well-researched psychotherapeutic and

psychopharmacologic approaches. The study’s participants were screened on treatment

unresponsiveness, meaning “having had at least three years of prior treatment for PTSD”. Furthermore, they were required to continue their pre-existing supportive therapy or pharmacologic treatment during the yoga trial. The study found that, over a period of 10 weeks, a weekly 1-hour class of trauma-sensitive yoga was more effective at reducing participants’ PTSD symptomatology than the same intensity of supportive women’s health education. Other controlled trials on different types of yoga have yielded comparable results, showing that, compared to an alternative intervention or a wait list group, yoga established reductions in PTSD symptoms, general stress, mood and anxiety (Butler et al., 2008; Descilo et al., 2010; Jindani, Turner, Khalsa, 2015; Streeter et al., 2010). Even though these results seem promising, a number of controlled trials did not find significant differences in

improvement between yoga and control groups (e.g. Mitchell et al., 2014). Regardless, due to (large) differences between studies in terms of methodology, assessment and type of yoga intervention,

conclusions are difficult to make. Taken together, these preliminary findings suggest that yoga can be an appropriateand effective (adjunctive) intervention in treating trauma-related psychopathologies. However, more evidence is needed to support these notions.

1.2 Emotion Regulation and the Yoga Black Box

Several studies have described the co-occurrence of traumatization and disruption in emotion or impulse regulation. Ehring and Quack (2010) showed that emotion regulation difficulties are

significantly related to levels of PTSD symptoms and that the severity of the dysregulation of emotions increases according to the severity of trauma (i.e. early-onset chronic as opposed to late-onset). To illustrate this, someone with a severely traumatizing history may experience more or qualitatively worse symptoms (such as dissociation) than someone with a less severe history of trauma. In addition Briere (2006) mentioned that, even though not all trauma-exposed individuals experience dissociation, severe problems in affect regulation resulting from traumatic stress are possibly directly instigating this dissociation. Confirming these statements, Van der Kolk described in earlier works that chronic trauma exposure is associated with significant problems in affect and impulse regulation (Van der Kolk, 2005; 2006; Van der Kolk et al., 1997), and that these dysregulations result from unmanageably high levels of emotional (including physiological) arousal associated with the traumatic event and its aftermath of putting victims in a constant fight-or-flight modus. He states that as an attempt to counteract this chronic hyperarousal, traumatized people shut down, behaviourally (e.g. by avoiding stimuli reminding them of the trauma) as well as psychobiologically - which is shown by the emotional numbing PTSD patients often display. In everyday life, in traumatized people there is often an alternate manifestation of emotional hyperarousal and emotional numbing (which could manifest itself in dissociation).

In addition to the literature focusing on the sympathetic part of the autonomic nervous system (emotional hyperarousal; fight-flight response), thePolyvagal theory (Porges, 1995) provides a framework for the importance of the parasympathetic nervous system in the development of PTSD.

Polyvagal refers to the two branches of the vagus nerve, which connects the brain to important organs

in the body and is a primary component of the parasympathetic nervous system. The Polyvagal theory explains the biology of safety and danger based on the interplay between visceral experiences and the expressions of emotional and social behaviour (Porges, 2001, as cited in Van der Kolk, 2014). The theory describes that the ventral branch of the vagus nerve plays an important role in our social engagement system, and that when social engagement is activated a person feels safe and is capable of effective interpersonal interaction. According to the theory, activation of the parasympathetic allows the social engagement system to regulate activation in such a way that it feels more nuanced and playful and puts the person in a physical and mental state where psychological healing can take place. This playful activation is different from the sympathetic-induced fight-or-flight activation, which often becomes too dominant in people who have experienced trauma. Based on the Polyvagal theory, PTSD symptoms have been described as a product of disruption of the parasympathetic nervous system,

thereby underscoring the importance of focusing on the parasympathetic nervous system in treating traumatized people (Sahar, Shalev & Porges, 2001).

The functionality of mindfulness-based practices such as yoga is yet to be described

comprehensively, but many hypotheses have been posed within previous years. Baer (2003) suggested that the mechanisms of mindfulness in theirability to establish positive change may include exposure, cognitive change, self-management, relaxation and acceptance. In line with Baer’s review, Shapiro et al. (2006) attempted to describe a meta mechanism of mindfulness called reperceiving, which leads to changes in self-regulation, values clarification, cognitive, emotional and behavioural flexibility, and ensures exposure. These four factors, in turn, promote positive change (cf. Thompson, Arnkoff & Glass, 2011). An example of how the practice of mindfulness may promote emotion regulation is by

encouraging people to tolerate their emotions as opposed to avoiding them. Hölzel et al. (2011) proposed that, in addition to emotion regulation, attention regulation, change in self-perspective, another important functionality of mindfulness is increasing body awareness or interoception, which is the awareness of the subtle sensory, bodily sensations.

Yoga, which is a combination of movement, breath and mindfulness, can thus be hypothesized to enhance self-regulatory mechanisms by promoting reperceiving as well as reducing physiological arousal by activating the parasympathetic nervous system. Literature shows that some traumatized individuals develop maladaptive coping strategies to deal with the trauma-induced stress they are experiencing. Whittlesey (1999) describes that, even though coping strategies such as avoidance or distraction directly minimize stress, they prevent individuals from properly processing the traumatic experience. Some authors have postulated that it is dysfunctional coping strategies like these that may lead to problems such as substance abuse or (self)-destructive and aggressive behaviours, which in turn in combination with adverse environmental factors may lead to criminality (Ullman et al., 2005; Filipas & Ullman, 2006, as cited in Danielly & Silverthorne, 2016).

A theoretical concept of how people regulate their emotions derives from Gross and John (2003), who described two different strategies of emotion regulation within the event of an emotional experience, i.e. a situation involving both emotion-evoking antecedents and emotional responses (and consisting of physical and cognitive as well as subconscious and conscious processes). The first strategy, cognitive reappraisal, is “a form of cognitive change that involves construing a potentially emotion-eliciting situation in a way that changes its emotional impact”. The second strategy, expressive suppression, is “a form of response modulation that involves inhibiting ongoing emotion-expressive behaviour” (p. 2). Whereas expressive suppression comes in after the emotional response has taken place (i.e. experienced), cognitive reappraisal occurs antecedently by changing ideas or attitudes towards a certain situation. The authors have studied how the different strategies relate to different states of mind and found that e.g. reappraisal is associated to greater experience and expression of positive emotions than is suppression. Similarly, they found that individuals who habitually used reappraisal experienced and expressed less negative emotions than those who habitually used suppression. Interestingly, the authors also showed that reappraisal, as opposed to suppression, was

associated with fewer symptoms of depression and more life satisfaction, self-esteem and optimism (cf. Gross & John, 2003). As described before, mindfulness refers to the practise of moment-by-moment awareness of our thoughts, feelings, bodily sensations, and surrounding environment – by accepting them and without judging them. According to Gross and John’s perspective on emotion regulation, it can thus be hypothesized that mindfulness’ ability to reduce symptoms including those inherent to affective disorders is linked to the antecedent-focused reappraisal strategy rather than the response-focused suppression strategy.

In line with the notion that yoga affects the parasympathetic nervous system, Streeter et al. (2012) theorized that yoga restores the stress-induced imbalance of the autonomic nervous system by correcting underactivity of the parasympathetic nervous system, thereby providing possible

explanations of the presumable effectiveness of yoga in reducing trauma-induced symptoms (Van der Kolk et al., 2014, See Telles, Singh & Balkrishna, 2012 for a previous review). The authors describe that one way yoga has been found to increase activity of the parasympathetic nervous system is by means of breathwork, either alone or combined with asana practice,which in turn may for example alter

negative emotional states like anxiety and stress (Brown & Gerbarg 2005, as cited in Streeter et al., 2012). Other suggested yoga-induced neurobiological changes that temper the effect of stress are increased GABA activity and decreased HPA-axis activity (e.g. Ross & Thomas, 2010). Low GABA levels are associated with anxiety disorders and other high-stress illnesses (Brambilla et al., 2003, as cited in Streeter et. al, 2007) and may be a predictive factor in the development of PTSD (Vaiva et al., 2004). These disorders are currently effectively treated with pharmacological agents that excite the GABA system, implying that other interventions that increase the activity of the GABA system could be as effective. In two small pilot studies, Streeter et al. (2007; 2012) tentatively pointed out that yoga influences the GABA-ergic activity in the brain in non-clinical and clinical populations. They first showed that an acute yoga session (N=8) versus a reading session (N=11) increased GABA levels in healthy individuals. Secondly, they found that a 12-week yoga intervention led to improvement in mood, and increased GABA levels in patients with Major Depressive Disorder (N=2) towards GABA levels of normal subjects (N=19). Contrastingly, this study did not find increased GABA levels in their healthy participants, which may be due to the specific exclusion criteria that were used. However, another trial did yield GABA-related effects on non-clinical subjects. Namely, in order to distinguish yoga from conventional physical practise, these authors showed that, when compared to a 12-week metabolically matched walking exercise - meaning individually rated and matched with the physical demands of the yoga intervention - (N=15), a 12-week yoga intervention (N=19) was associated with relatively higher GABA levels and improvements in mood and anxiety (Streeter et al., 2010). Hence, firstly these studies indicate that yoga may actively alter GABA levels and thereby someone’s mood. Secondly, these results tentatively imply that the effects of yoga are similar to those of

psychopharmaceuticals. However, due to the small number of participants in these studies results should be interpreted carefully.

Thus, the combination of moving, breathing and mindfulness may be specifically effective at altering neurobiological processes and thereby reducing stress-induced mental health problems. Within a similar framework, other mind-body interventions, or movement therapies, have already shown preliminary promising results in treating (treatment-resistant) patients with PTSD (Vermetten et al., 2013; Gray, 2001; 2002). Furthermore, there are studies suggesting that both exercise and yoga interventions can improve a variety of stress-related health-outcomes, but that yoga may be of superior quality (Patel, Newstead & Ferrer, 2012; Ross & Thomas, 2010). Explanations for these findings are speculative by nature but Salmon et al. (2009, as cited in Spinazzola et al., 2011)suggested that the addition of movement (i.e. learning new motor skills) to mindfulness brings about enhanced sensory awareness, which makes the focusing of attention easier and thus the person relatively more mindful. 1.3 Prison populations

Previous studies focusing on incarcerated youth have shown that mindfulness-based interventions may be feasible and effective in reducing stress and increasing self-regulation abilities. To give an example, the pilot study conducted by Himelstein, Hastings, Shapiro and Heery (2012) assessed 32 incarcerated participants pre- and post intervention on trait mindfulness, self-regulation, and perceived stress. Their intervention followed a protocol called Mind Body Awareness, which was developed to accommodate incarcerated adolescents and which incorporates practices comparable to other mindfulness-based interventions such as sitting meditation, discussion topics and awareness-increasing exercises (Kabat-Zinn, 1990). The authors found no significant increase on trait mindfulness but they did find a

significant increase in healthy self-regulations and a significant decrease in perceived stress. In addition, a systematic review conducted by Auty, Cope and Liebling (2015) showed that participants who completed a yoga or meditation program in prison experienced an increase in psychological well-being and improvement in behavioural functioning. The review included studies with incarcerated offenders of different sex, institutions and age categories (e.g. women’s and young offender

institutions). Further, the review included studies whose participants completed a yoga or meditation program. Overall, results showed small effect sizes on psychological well-being and behavioural functioning (Cohen’s d of respectively .46 and .30) but nonetheless the evidence suggested that yoga has favourable effects. The finding that effects on behavioural functioning were lower may be explained by the notion that psychological change occurs prior to behavioural change.

Examples of projects that introduced yoga to prison populations are the Prison Phoenix Project in the United Kingdom and the Prison Yoga Project that originated in the United States. Initiator of the Prison Yoga Project is James Fox, who started offering yoga programs to prisoners within the

framework of restorative justice. From a similar philosophy and the same efforts to provide alternative treatments to prisoners, the Prison Yoga Project started to offer programs in the Netherlands as well. Within their current program, Prison Yoga Project Netherlands (PYPN) targets a group of habitual offenders called the ‘Top600’, whose members have gained their status because of high crime rates such as (armed) robberies, mugging and burglary, and negative (psychological) impacts on the victims

of these crimes. The majority of the Top600 is of young adult age, i.e. roughly between 18 and 28 years old, many of whom come from disadvantaged backgrounds. In general, these offenders are not only associated with high impact criminality but also with a wide variety of problems including intellectual disabilities, substance abuse and social, emotional or psychiatric issues

(

https://www.amsterdam.nl/bestuur-organisatie/organisatie/overige/acvz/top600/

So far, the project team has conducted three trials of their yoga program at P.I. Lelystad, a penitentiary institution in the Netherlands. Up to now, results from yoga programs in prison environments have substantially come from anecdotal reports, but a number of controlled studies have been done. Danielly and Silverthorne (2016) conducted a study on Prison Yoga in the US by looking at the effect of a 10-week trauma-sensitive yoga intervention on self-reported depression, perceived stress, anxiety, rumination, self-control and improved self-awareness on female prisoners. The results showed that participants in the yoga-group (n=33) improved from pre- to post-intervention in terms of

depressogenic symptoms, perceived stress and self-awareness compared to waitlist participants (n=17). No significant effect was reported on anxiety, rumination and self-control. Furthermore, Bilderbeck, Brazil and Farias (2015) looked at the effects of a 10-week yoga program on measures of affect, perceived stress, and performance on a cognitive-behavioural computer task called the Go/No-Go task which assesses executive functioning, in particular attentional capacity and behavioural

inhibition. The findings were that participants in the yoga group (N=45), compared to the control group (n=55), experienced more positive affect, and less stress from pre- to post intervention. The study also showed that participants in the yoga group performed relatively better at the cognitive-behavioural task that was done post intervention, implying that yoga leads to better behavioural response inhibition and sustained attention. These findings may provide tentative evidence that yoga brings about behavioural improvement, and may also provide a preliminary insight into the mechanismsof yoga.

The current study aimed to extend the literature on yoga programs in prison by studying the effects of yoga on mental health of incarcerated young men, who areknown to be at high-risk for traumatization (outside and within the prison environment) and the subsequent development of mental health problems. More specifically, this study intended to investigate the effects of a 12-week trauma sensitive yoga intervention on self-reported symptoms of sleep problems, depression, anxiety, hostility, and PTSD. Based on previous empirical literature on prison populations, it was hypothesized that prison yoga would yield positive effects on mood and psychological wellbeing. Furthermore, the intention of the current study was to investigate if yoga has favourable effects on trait mindfulness and emotion regulation skills, based on the notion that these factors play an important role in the

functionality of yoga. It was hypothesized that yoga would increase trait mindfulness, thereby helping inmates to regulate their emotionsmore functionally. We expected that improvement in emotion regulation abilities would be expressed in a greater use of cognitive reappraisal strategies as opposed to suppressive expression strategies.

2. Methods

2.1 Design and Setting

We completed a 12-week, single group study of a yoga intervention with pre, mid and post assessments containing a battery of self-report questionnaires and qualitative questions, hereby measuring change in mental health over time. The study was conducted at P.I. Lelystad and was part of Prison Yoga Project the Netherlands (PYPN), the project group that initiated the yoga program at this facility. Their program was specifically intended for Top600 detainees, a group of about 600 young male offenders from Amsterdam who currently reside in a separate department at the penitentiary institute (see Introduction for further details on Top600).Given the characteristics of PYPN (e.g. previous trials had small numbers of participants) and given the unpredictable nature of the prison environment, this study was intended and conducted as an exploratory pilot intervention.

2.2 Participants

Detainees were recruited for the program by the prison staff, after which they were asked to participate in the current study as well. Because this was a pilot study, a sample size calculation or a predefined primary endpoint was not taken into account; however, a sample size ranging from 10 to 20 participants was expected based on previous PYPN-trials at P.I. Lelystad. Recruitment efforts yielded a total number of 18 participants who applied for the program, and a total number of 12 participants who consented to participate and were eligible to join the current study. Eligibility initially depended on the following criteria: sufficient knowledge of Dutch language and minimum length of stay at the facility of 12 weeks. Because the number of participants that were able to attend the full program was unexpectedly low, this criteria was changed to 6 weeks (from pre to mid assessment). Of the initial 12 participants who consented, 4 dropped out within the first week; one participant was discharged early, one did not want to partake anymore and two were relocated to a different department. These

participants were excluded from data analysis. Furthermore, one participant was included at mid intervention, making the final sample consisting of nine participants. Based on previous literature on forensic populations (Kristiansson, Sumelius & Sondergaard, 2004) it was expected that a majority of our participants would have had traumatizing experiences. However, in the current study participants were not specifically screened for exposure to traumatic events.

The sample’s demographics are outlined in Table 1. Participants’ age ranged from 19 to 39 years old (M=27.11, SD=6.72). As their highest obtained education, 1 participant reported primary school, 3 participants reported secondary education and 5 participants reported a subsequent degree. Participants’ release dates ranged from 6 weeks to 5 years after initiation of the study (M=19.97;

SD=24.19 months).The majority of participants reported having had prior psychological treatment for problems including behavioural disorders, bipolar disorder and PTSD. None of the other participants reported to be currently engaged in any kind of conventional psychological treatment. Three

participants received ongoing pharmaceutical treatment, two for physical injuries and one for a psychological diagnosis.Apart from two detainees who partook in a previous PYPN trial at the facility, all participants were new to yoga. This yielded an average mean of 0.67 months (SD=1.32) of

experience with yoga.

Table 1. Study participants’ demographic characteristics (N=9)

Criteria _{M(SD)/ n Range} Age (years) 27.11(6.72) 19 - 39 Highest education obtained

Primary school Secondary school Higher education (MBO) Higher education (HBO)

1 3 4 1

Sentence time until release (months)* 19.97(24.19) 1.5 - 66 Prior psychological treatment

PTSD

Personality disorder Bipolar disorder Behavioural disorder

Current pharmaceutical treatment

Physical condition Psychological diagnosis

Prior experience to yoga (months)

1 1 1 3 2 1 0.67(1.32) 0 – 3

* 19.97 months ≈ 1.67 years; 24.19 ≈ 2.02 years; 1.5 ≈ 0.13 years; 71≈5.92 year

2.3 The yoga intervention

The yoga program consisted of twice weekly, 90 minute-long trauma-sensitive yoga classes, incorporating classic hatha yoga elements: asanas, breathwork and mindfulness. All classes were taught by two yoga instructors, who both had prior experience with forensic populations and were certified to teach trauma-sensitive classes. The classes were standardized and supervised by a clinical psychologist and held according to the principles of Prison Yoga, that was specifically developed to accommodate trauma-affected individuals and adapts and applies yoga techniques and practices in order to assist individuals facing bodily or mental health challenges. The classes try to make their participants feel safe at all times, which supports the notion that survivors of trauma need a safe environment for healing to take place (e.g. Levine, 1997). Furthermore, these classes hold the following principles: All postures are demonstrated and done simultaneously by the teachers, participants are not adjusted (i.e. touched) during class, and the pace of the class is generally slow and predictable.

Lastly, participants are encouraged to connect to their own bodies and to respect their own boundaries, thereby increasing mental and bodily awareness and stimulating better coping abilities (e.g. knowing how to better deal with unwanted and uncomfortable emotions arising from past negative

experiences).

All classes were taught at the sport facility of the penitentiary institute and were accompanied by two departmental guards, who sat in the back of the room. These guards also took care of the transport of participants, picking them up from their cells before class and enclosing them after the class took place. Prior to every class, the instructors assessed the current physical state of the participants by asking about (new) injuries or other pressing issues that may have come up. Over the course of the program, participants were encouraged to discuss any physical or emotional issues coming up to allow for appropriate modifications in the asanas to be made. All participants were issued props commonly used in a yoga class, i.e. yoga mats, blocks, blankets and pillows.

2.4 Materials

Mental health was determined by questionnaire scales measuring PTSD, depression, anxiety, hostility, and sleep problems. To explore hypothesized mechanisms of change in mental health, affect regulation was measured, and as a means to check if participants in fact became more mindful, trait mindfulness was assessed. Additionally, some socio-demographics were obtained, including age, final date of sentence, prior experience with yoga or mindfulness, current or past psychological treatment(s), and highest education completed. The amount of items assessed in the current study was comparable to earlier psychological research on prisoners (Danielly & Silverthorne, 2016; Sistig et al., 2015). All questionnaires were self-report instruments, and all psychological measures were assessed pre, mid and post intervention. In addition, qualitative information was obtained via instructor intakes (See Appendix A) at pre assessment and via short semi-structural interviews with the researcher during mid and post assessment (See Appendix B).

Five Facet Mindfulness Questionnaire-SF (FFMQ-SF; Bohlmeijer et al., 2011). The FFMQ-SF, developed in the Netherlands, is a self-report instrument consisting of 24 items measuring five

different aspects of mindfulness: observing, describing, acting with awareness, non-judging of inner experience, and non-reactivity to inner experience. Participants rate on a 5-point Likert scale to which degree the statement is true. Both the FFMQ-SF and the original FFMQ have been validated and found to be a reliable, comprehensive instrument for use in adults with non-clinically and clinically relevant symptoms (Baer et al., 2006; Bohlmeijer et al., 2011).

PTSD Checklist for DSM-5 (PCL-5; Weathers et al., 2013). The PCL-5 is a self-report

questionnaire that assesses the 20 DSM-5 symptoms of PTSD. It consists of 20 items where participants rate possible symptoms within the previous month on a 5-point Likert scale. An example of an item is “in the past month, how much were you bothered by: “repeated, disturbing, and unwanted memories of the stressful experience?” The PCL-5 is not developed for diagnosis, but rather as a means to screen individuals for PTSD and to monitor symptom change during and after treatment. The psychometric

qualities of PCL-5 have not been studied extensively yet, however, there has been evidence suggesting that the PCL-5 is a psychometrically sound instrument than can be used effectively on trauma-affected populations (Bovin et al., 2016; Sveen, Bondjers & Willebrand, 2016). So far, studies that measure the psychometric qualities of the Dutch version of the PCL-5 are only planned in the future (Boeschoten et al., 2014).

Emotion Regulation Questionnaire (ERQ; Gross & John, 2003; Dutch translation Koole, 2004). The ERQ is a self-report instrument that assesses individual differences in emotion regulation strategies. It consists of 10 items measuring two proposed strategies of emotion regulation: cognitive reappraisal (6 items) and expressive suppression (4 items). An example of an item is “I keep my emotions to myself”. Items are answered on a 7-point Likert scales, ranging from strongly disagree to agree. The ERQ has been shown to be a reliable and adequately validated instrument (convergent and discriminant validity) (Gross & John, 2003).

Symptom Checklist – 90 - Revised (SCL-90-R; Derogatis, 1975; 1992). The SCL-90-R is a multidimensional self-report instrument that helps evaluate a broad range of psychological problems and (physical) symptoms of psychopathology expressed in 9 different dimensions and 90 items measured on a 5-point Likert scale. The instrument is found to be useful in measuring patient progress or treatment outcomes. The psychometric qualities of the SCL-90-R have been studied extensively, and convergent and divergent validity as well as internal consistency and test-retest reliability have been generally found to be of respectively good and adequate quality (Holi, 2003). The current study utilized the subscales Anxiety (ANX; 10 items), Depression (DEP; 16 items), Hostility (HOS; 6 items), and Sleep problems (3 items), amongst which only convergent and divergent validity of the subscales DEP, HOS and ANX have been studied separately. Results range from poor to excellent (e.g. Buckelew et al., 1988). Reliability in terms of internal consistency of the subscales has been described as adequate (Prinz et al., 2013). The SCL-90-R is normed for four different populations and is designed to be appropriate for use with clinical and non-clinical populations (Arrindell & Ettema, 1986).

2.5 Procedure

Top600 participants that showed interest in the yoga program attended an informative session that took place four days prior to intervention. Those who decided to participate underwent separate individual intake sessions with respectively the yoga instructors and the researcher. The intake with the instructors took place in order to obtain additional demographics and qualitative information about the participants. For example, during these intakes participants were asked about their current medical state (e.g. injuries and/or use of medication) so that the instructors could take that information into account during classes. Intakes with the instructors also served as a means to either include or exclude participants on eligibility criteria. Pre intervention assessments with the researcher took place

afterwards, and consisted of a battery of self-report questionnaires, which was also used during mid (after 6 weeks) and post intervention assessments (after 12 weeks). During mid and post sessions, qualitative data was obtained by the researcher as well (e.g. “Did you use any (illicit) substances during

the program”). All assessments were recorded with an audio recording device. In light of the relatively low attention spans and the illiteracy associated with forensic populations, the length of assessments was maintained reasonably low, that is to say ranging from 15 to 30 minutes per session. For similar reasons, to those participants who requested assistance, instructions and questionnaire items were read out loud by the researcher.

2.6 Data Analysis

Intervention effects. We included data of participants who completed at least pre and mid

assessments. Because the study’s final sample size was too limited, it was decided to use descriptive analysis rather than inferential statistics to report changes from pre/mid, mid/post and pre/post assessment. Overall questionnaire data were reported by means of mean scores and standard deviations as calculated with SPSS-Statistic. For a more comprehensive overview, we interpreted participants’ individual score patterns over time.

Primary outcomes. Group means and standard deviations on the FFMQ-SF, ERQ, SCL-90, and PCL-5 were reported for pre, mid and post assessment. We specifically looked at pre intervention scores and we compared scores over time with available norms. Further, participants’ individual score patterns on Sleep Problems, Hostility and the PLC-5 were reported.

Secondary outcomes. To further illustrate intervention effects, we analysed and reported qualitative data resulting from semi-structural interviews during mid and post assessments.

3. Results

3.1 Treatment dropout and number of attended classes

Over the 12 week-course of the intervention, 4 additional participants (44%) dropped out prior to post intervention assessments because of (early) release, and one participant was included at mid

assessments. Finally, this yielded 8 participants at pre assessment, 9 participants at mid assessments and 5 participants at post assessments. A total number of 23 classes were held; one class was cancelled due to last-minute circumstances on the teacher’s side and two classes were cancelled because of national holidays. Overall attendance rates ranged from 26% (6 classes) to 100% (23 classes) with participants attending an average of 16 classes (SD = 7.02).

3.2 Pre intervention scores

Pre assessment scores on the ERQ showed that, when compared to averages of male undergraduate students (reappraisal M=4.60; suppression M=3.64, Gross & John, 2003), participants scored low on both Cognitive Reappraisal and Expressive Suppression strategies at pre assessment (See Table 2). Due to the limited time available to conduct assessments, only three participants completed the FFMQ-SF at pre assessment. Hence, scores on the FFMQ-SF should be interpreted very carefully. We did not find comparable studies or postulated norms for the FFMQ-SF to analyse pre assessment scores separately

(See Bohlmeijer et al., 2011 for data on a predominantly female and adult sample with depressive symptomatology).

As previous literature has shown that the presence of psychopathology is relatively high amongst forensic populations (e.g. Fazel & Danesh, 2002), we expected this to reflect in the

participants’ pre assessment scores on the SCL-90 scales and the symptom checklist for PTSD (PCL-5). We found that, when compared to both clinical and general population norms, participants scored high to very high on Sleep problems, and average to above average on Hostility at pre assessment. Further, participants’ scores on both Depression and Anxiety ranged from low to below average, and scores on the PCL-5 ranged from very low to above (n=1) cut off score (Weathers et al., 2014). It should be noted that this cut-off score is preliminary and applies to veterans; cut-off scores may be set higher or lower for forensic populations or civilians. Thus, in contrary to what was expected, on average at pre assessment participants did not report elevated scores on measures of mental health, except for sleeping disorders and hostility (See Table 3).

Table 2. Mean scores and standard deviations on Emotion Regulation and Trait Mindfulness at pre (n=8),

mid (n=9) and post (n=5) intervention assessment.

Variable Pre intervention (N=8) M SD Mid intervention (N=9) M SD Post intervention (N=5) M SD Trait mindfulness (FFMQ-SF) 76.67* 9.29* 86.25 10.59 75.00 25.45

Cognitive Reappraisal (ERQ)** 25.25 11.21 28.22 11.27 23.40 13.16

Expressive Suppression (ERQ)*** 14.37 7.13 13.87 7.83 13.40 5.03

Emotion Regulation (ERQ)**** 39.62 16.70 42.00 17.85 36.80 17.77

Note. ERQ = Emotion Regulation Questionnaire; FFMQ-SF = Five Facet Mindfulness Questionnaire Short

Version. *N=3

**These scores represent mean total scores; the overall mean scores for reappraisal are respectively 2.5(sd=1.2); 2.7(1.1); 2.3(1.3) ***These scores represent mean total scores; the overall mean scores for suppression are respectively 1.4 (sd=0.7); 1.4 (0.8); 1.3(0.5) ****These scores represent mean total scores; the overall mean scores on the ERQ are respectively 4.0(sd=1.7); 4.2 (1.8); 3.7(1.8)

Table 3. Mean scores and standard deviations on Mental Health at pre (n=8), mid (n=9) and post (n=5) intervention assessment. Variable Pre intervention (N=8) M(SD) Mid intervention (N=9) M(SD) Post Norms intervention CP/GP (N=5)

M(SD) Very High High Average

Sleep problems (SCL-90-R) 10.50(3.96) 9.11(2.93) 9.20(3.27) ≥14/≥9 10-13/6-8 6-8/4-5 Depression (SCL-90-R) 24.62(6.02) 22.89(2.71) 22.00(1.87) ≥63/≥36 50-62/25-35 35-41/20-23 Anxiety (SCL-90-R) 13.00(2.88) 11.78(2.22) 12.40(1.14) ≥36/≥22 27-35/15-21 18-23/12-14 Hostility (SCL-90-R) 9.75(3.06) 7.67(1.32) 7.80(2.95) ≥17/≥11 12-16/9-10 8-11/7-8 PTSD (PCL-5) 19.50*(13.86) 14.44(9.61) 14.20(9.42) -** - -

Note. SCL-90-R = Symptom Checklist - 90 - Revised; PCL-5 = PTSD Checklist for DSM-5; CP = Clinical

Population; GP = General Population. *Range = 5 - 47

**A preliminary cut-off score of 38 is proposed to indicate PSTD; a reduction of 5 points has been suggested to reflect a reliable reduction in symptoms (Weathers et al., 2014)

3.4 Preliminary intervention effects

The current study’s limited and inconsistent sample, and inability to statistically test our hypotheses means that the intervention’s effects should be interpreted carefully. Furthermore, when interpreting results it should be considered that most reported effects represent minor changes in scores. Group mean scores on measurements over time are reported in Table 2 and 3. In line with our hypothesis regarding trait mindfulness, an increasing trend on FFMQ-SF scores was found from pre (M=76.67

SD=9.29) to mid intervention (M=86.25, SD=10.59), suggesting a substantial change in ability to be

mindful in daily life. We found that these changes were not maintained at post assessment (M=75.00,

SD=25.45). However, data analysis showed that there was one particularly low score at post

assessment (See Appendix C for individual score patterns on Trait Mindfulness). Deleting this outlier yielded M=86.25 (SD=4.5), indicating that the outlier reflected particular individual circumstances and suggesting that overall mindfulness remained improved until post intervention. Regardless, these results should be interpreted with extra caution since only three participants completed the FFMQ-SF at pre assessment and the different individual score patterns over time show inconsistency. An

expected trend was found for Emotion Regulation, as scores on Expressive Suppression decreased, and scores on Cognitive Reappraisal increased from pre (M=14.37, SD=7.13 resp. M=25.25, SD=11.21) to mid intervention (M=13.87, SD=7.83 resp. M=28.22, SD=11.21). This beneficial trend in reported

Expressive Suppression was maintained at post intervention (M=13.40, SD=5.03), but in contrast to our hypotheses, this was not found for Cognitive Reappraisal (M=23.40, SD=13.16).

Comparable trends were found for scores on the SCL-90 and the PCL-5. As hypothesized, reported sleep problems decreased from pre (M=10.5, SD=3.96), to mid intervention (M=9.11,

SD=2.93). Similarly, reported symptoms of depression and anxiety decreased from pre (M=24.62, SD=6.02 resp. M=13.00, SD=2.88) to mid intervention (M=22.89, SD=2.71 resp. M=11.78, SD=2.22).

Further, reported hostility and PTSD symptoms decreased from pre (M=9.75, SD=3.06 resp. M=19.50,

SD=13.86) to mid intervention scores (M=7.67, SD=1.32 resp. M=14.44, SD=9.61). Unexpectedly, from

mid to post intervention no further reductions were observed, apart from minor decreases on reported depression (M=22.00, SD=1.87) and PTSD symptoms(M=14.20, SD=9.42). In contrary to FFMQ-SF and ERQ scores, overall SCL-90 and PCL-5 scores decreased from pre to post assessment. Thus, overall results showed inconsistent patterns for Sleep problems, Anxiety and Hostility, and a small reduction in Depression over time, suggesting zero to small practical significance. A more substantial decrease was found for the PCL-5 suggesting larger practical significance for PTSD symptoms, given that a 5 point change is accepted as a threshold for determining whether an individual has responded to treatment (Weathers et al., 2014).

Individual participant data analysis confirmed that the intervention had little to no beneficial effect on the outcome measures from pre to post assessment. However, individual score patterns indicated more consistency in reductions for Sleep problems (See Figure 2 and Table 4.1) and Hostility (See Figure 3 and Table 4.2; For complete illustrations of individual score patterns on measurements see Appendix C). This is particularly interesting since participants only showed elevated baseline scores on these scales and on the PCL-5. Figure 4 and Table 4.3 depict individual score patterns on the PCL-5. As shown, the one participant that scored above cut-off at pre assessments decreased to below cut-off at mid assessments, suggesting large practical significance for this particular participant; however, one should keep in mind that the PCL-5’s cut-off score is only preliminary, and that post intervention data were not obtained for this participant. Table 4.3 also shows that participants with relatively high pre intervention scores made most progress, with Participant 4 reporting an

exceptionally large decrease in PTSD symptoms over time. Reasons for this exceptional progress are unclear but qualitative data demonstrated that this participant was highly motivated and focused during classes. Also, evidently, when symptom pressure is low there is less progress to be made. One of the participants (Participant 5) reported an unexpectedly large increase in PTSD symptoms over time. At intakes, this particular participant shared that he applied to receive psychological treatment at the facility for his PTSD symptoms. This suggests that his symptom pressure was already high at baseline but that scores did not reflect this because of e.g. social desirability. However, clearly this explanation remains strictly hypothetical.

Figure 2. Individual score patterns on Sleep Problems using the

SCL-90 Table 4.1 Participant Number T1 T2 T3 1 15 12 - 2 15 15 15 3 9 6 7 4 13 9 8* 5 5 7 8 6 5 10 - 7 11 6* - 8 11 9 - 9 - 8 8

Individual scores on Sleep Problems over time

Note. T1=pre intervention, T2=mid intervention, T3=post

intervention

*Suggesting small practical significance; Clinical Population average norm=6-8

Table 4.2

Individual scores on Hostility over time

Participant Number T1 T2 T3 1 16 9* - 2 12 9 13 3 10 7* 7* 4 8 6 6 5 7 7 6 6 7 6 - 7 10 9 - 8 8 7 - 9 - 9 7*

Figure 3. Individual score patterns on Hostility using the SCL-90 Note. T1=pre intervention, T2=mid intervention, T3=post

intervention

*Suggesting small practical significance; Clinical Population average norm=8-11; General Population average norm=7-8

3.5 Qualitative data; applicability and experience of the yoga intervention

At mid and post assessments participants were interviewed about their experience of the intervention using a semi-structured interview format (See Table 5). In general, participants reported that classes were both entertaining and beneficial. They emphasized that the breathing exercises and guided meditation practices were beneficial to them, rather than the physical practise by itself. Most

participants also used breathing techniques in their cells in order to fall asleep. In contrast, they rarely practised asanas outside of the class environment. Participants mentioned that they became more aware of physical sensations and thoughts and that it became easier to accept it when these sensations or thoughts were negative, indicating increased interoception and improvements in self-regulation skills. At pre-assessments intakes, all participants reported some form of physical discomfort, ranging from minor muscular stiffness to injuries on the knees, shoulders or lower back. Apart from increased flexibility, participants did not report significant physical improvement. In line with Figure 2 and Table 4.1, the majority of participants described that yoga made them fall asleep better, specifically on nights after a class. Furthermore, doing yoga increased their sense of inner peace. For some this meant being able to concentrate better during the day or having less troubling thoughts, and for others this meant feeling less tension in their bodies.

Participants reported feeling safe during classes, and that doing the postures became more comfortable with time, but it was also mentioned that during the second half of the program the yoga classes became less challenging. The latter might explain why the majority of scores did not continue to decrease from mid to post intervention. At times there was restlessness during classes but at the same time participants encouraged each other. This underscores the notion that participants were

intrinsically motivated for the program. Most participants did not directly notice behavioural changes Figure 4. Individual score patterns in symptoms of PTSD using

the PCL-5 Table 4.3 Participant Number T1 T2 T3 1 47 32 - 2 21 12 11 3 11 7 9 4 32 5 6 5 14 18 30 6 9 11 - 7 17 9 - 8 5 8 - 9 - 28 15

Individual scores on the PCL-5 over time.

Note. T1=pre intervention, T2= mid intervention, T3=post

in themselves outside of the yoga class-environment. However, a majority reported that there were periods during the program that they did not feel the need to smoke cigarettes or cannabis. Moreover, the fact that reported hostility (which includes hostile urges and thoughts) decreased over time suggests that behavioural changes may have occurred in the long term (See Figure 3 and Table 4.2). Table 5. Examples of semi-structured interview questions and responses

Interview Questions Answers/Remarks

How have the yoga classes been for you so far? “Good, I feel more inner peace”; “I’m sleeping better, especially after the class”

Have you noticed any difference since the start of the program?

“Yes, I feel calmer both physically and mentally”; “I feel muscles that I’ve never felt”; “I notice where I am breathing now; the breath goes everywhere in the body”

Can you name some of the effects the classes have had on you?

“Focusing on breathing taught me how to be more in the moment and better accept situations from the past”; “I feel more

concentrated during the day”; “By focusing on my breath and sitting still I got better at accepting things as they are”; “Before falling asleep I use the technique of making my body heavy, just like we do in class”.

Did you use any substances over the past 6 weeks?

“I do not like to smoke before classes because I want to feel fresh”; “I quit smoking cannabis for a few weeks during the program because I did not need it as much to get rid of my stress”.

4. Discussion

This pilot study investigated the effects of a 12-week trauma-sensitive yoga intervention on self-reported mental health of Dutch detainees. To our knowledge, this study hereby provided the first preliminary scientific findings on Prison Yoga in the Netherlands.

Although the study’s sample was too limited to report statistically meaningful differences, our findings suggest that Prison Yoga is a feasible and potentially beneficial intervention amongst young male offenders. Over the first 6 weeks of the intervention, we found slight group improvements in reported trait mindfulness and emotion regulation. Consistent with previous findings that mindfulness-based interventions decrease the suppression and avoidance of emotions, it was found that practicing yoga increased the use of cognitive antecedent-focused strategies of emotion regulation as opposed to

response-focused suppressive strategies (Dick et al., 2014; Kumar, Feldman & Hayes, 2008). However, our findings showed that the increase of trait mindfulness and the use of more functional emotion regulation strategies were not maintained until post intervention. Further, we found that self-reported measures of sleep problems, anxiety, depression, hostility and symptoms of PTSD improved over the first 6 weeks of the intervention, but that these benefits were only maintained for reported depression and PTSD symptoms. Finally, this resulted in minor progress on psychological outcomes over the 12 weeks of the yoga program. An exception to this was the finding that PTSD symptoms improved more substantially, implying that participants suffered significantly less from the psychological aftermath of past traumatic events. This is consistent with previous literature demonstrating that yoga can

accommodate trauma survivors and can improve symptoms of PTSD (cf. Mitchell et al., 2014; Emerson et al., 2009). When interpreting participants’ individual score patterns over the course of the

intervention, we also found indications of practically significant improvements in reported sleep problems and hostility, confirming previous preliminary evidence that yoga may improve subjective sleep (See Ross & Thomas, 2010) and decrease hostile-anger moods (Lavey et al., 2005). Nevertheless, it remains questionable whether the changes in our study were large enough to reflect clinically relevant changes.

Thus, overall we found preliminary beneficial effects on mental health outcomes over the first 6 weeks of the intervention. Afterwards, symptomatic improvements either remained the same or decreased towards baseline. Likewise, improvements in trait mindfulness and emotion regulatory skills were not maintained over the second half of the course. Hereby this study does not confirm prior findings that mindfulness-based practices can substantially improve mental health amongst forensic populations over longer periods of time (Shonin et al., 2013). Even though these findings are contrary to our hypotheses, the fact that mindfulness skills were acquired but not maintained could explain the parallel finding that psychological outcomes improved over the first 6 weeks of the intervention and declined afterwards. This idea is based on the hypothesis that mindfulness serves as the effective mechanism in yoga that brings about improvements in mental health (Jain et al., 2007). Further, since emotion regulation is described as one of the functional aspects of mindfulness (cf. Hölzel et al., 2011), it is not surprising that emotion regulation improved according to improvements in trait mindfulness i.e. cognitive reappraisal strategies increased over the first half of the intervention and decreased back to baseline scores later on. The notion that the absence of reductions in symptoms may be caused by a lack of mindfulness skills is further underscored by the findings that improvements in unhealthy emotion regulation reduce negative mental health outcomes (Martin & Dahlen, 2005).

Although research on the effects of yoga in forensic settings is in its infancy, preliminary promising results have been outlined (Muirhead & Fortune, 2016). The two most methodologically robust studies (N=167; N=64) have shown in randomized controlled trials that yoga can lead to improvements in PTSD symptoms, higher levels of positive affect and executive functioning, and lower levels of perceived stress and psychological distress (van der Kolk et al., 2014; Bilderbeck et al., 2013). The remaining studies lack methodological quality and are heterogeneous in assessment material and

type of yoga intervention. Although it was not in the scope of this study to randomize and include a control group, we aimed to contribute to the current literature by assessing trauma-related symptoms, based on the notion that yoga may be specifically beneficial when implemented within a trauma-informed framework. To do so, we used a standardized trauma-sensitive yoga intervention based on the principles of Prison Yoga. Further, even though benefits of yoga and mindfulness in young adults have been demonstrated (e.g. Woolery et al., 2004), so far studies have not specifically looked at the effects of yoga on this population in prisons.

4.1 Qualitative data in the current study

Overall, key themes of semi-structural interview data revealed that yoga increased self-control and body awareness, and led to improved sleep and stress management. In accordance with a previous qualitative study on yoga amongst prisoners (Dezerotes, 2000, as cited in Muirhead & Fortune, 2016), we found that participants enjoyed the program and believed they had acquired useful skills to deal with stressful thoughts or situations. Additionally, our study confirmed the findings that yoga improved prisoners’ subjective concentration and self-control (e.g. accepting negative thoughts without

immediately reacting to them). Not worrying as much about past or future happenings was often described as finding more inner peace.

Furthermore, we found that breath and meditation practices were perceived as most beneficial to our participants’ mental state. This confirms prior qualitative findings on Prison Yoga by Viorst (2017), who says that prisoners often emphasized the benefits of non-physical practices because they “offered the opportunity to practice being focused and fully present with the self in moments of turmoil and tension” (p. 20).

Lastly, in the current study, it was found that participants did not notice significant behavioural changes in themselves. This finding may be explained by evidence that yoga and mindfulness more directly affect brain functioning (Zylowska et al., 2008), and that effects on behaviour require longer training. Similarly, even though we found indications of changes in hostile attitudes (on the SCL-90), participants did not report changes in hostile behaviour. However, previous studies have shown that mindfulness-based interventions are able to induce positive changes in behaviour (e.g. verbal aggression and behavioural avoidance; see Muirhead & Fortune, 2016; Keng, Smoski & Robins, 2003) and also within the current study we found indications of small behavioural change. Namely,

participants reported that they did not want to smoke cannabis before classes because it would negatively affect the practice, and did not need to smoke in order to deal with their stress. These findings are in line with previous preliminary evidence that yoga reduces the consumption of

substances (Chiesa & Serretti, 2014), and underline the hypothesis that yoga helps to cope with stress in a more functional matter. Though our participants’ reported changes in substance use are promising, it should be noted that these improvements were often temporarily. This, again, emphasizes that longer practice may be needed to bring about lasting change.

4.2 Limitations & Future Directions

This study had several limitations to it. Clearly, because this study’s sample was small there was a lack of power to detect statistically significant relationships between our variables, thereby missing small but potentially important practical improvements. In addition, the absence of a control condition impeded our ability to draw conclusions about the effectiveness of the intervention. Consequently, our results should be interpreted as preliminary and cannot easily be generalized. Further, the relatively low baseline scores on reported depression, anxiety and PTSD symptoms also reduced the probability in finding (large) effects.

Another limitation relates to the prison environment. In most cases, drop-out and adherence rates were a direct consequence of these prison policies i.e. most participants dropped out because they were either relocated to a different department within the prison or released from the facility. Likewise, class absence was often the consequence of sudden punitive measures such as last minute solitary lock-up. Thus, although this study suffered from drop-out and nonattendance, in most cases this was not due to a lack of motivation on the participant’s side. To illustrate, we also concluded that participants were intrinsically motivated based on their efforts during classes and their positive feedback during mid and post interviews. Further, participants often showed specific interest during classes e.g. asking about the origin of yoga or inquiring about specific breathing techniques.This information underscores the potential applicability of the project amongst our study’s population. Regardless of the fact that drop-out will always remain an issue, future research could focus on

controlling unpredictable prison environments as much as possible. For example, inclusion criteria can be modified e.g. increasing the minimum length of stay to ensure that participants can complete the full intervention.

The current study was also limited in its choice of assessment material. Apart from issues of social desirability related to self-report, we also experienced differences in participants’

comprehension of the questionnaire items. This led to a variety in time needed to complete the assessment, with some participants needing more assistance from the researcher (i.e. elaborating or reading questions out loud) than others. This in turn led to only three participants having sufficient time to complete the FFMQ-SF - all of them being relatively highly educated. Apart from the fact that results of these participants should be interpreted with caution, it should be noted that their

educational level possibly further biased the results on trait mindfulness. To counteract possible bias, it is desirable for further research to include more objective measures of mental health. There have already been studies providing preliminary insight into the benefits of mindfulness and yoga on executive functioning (e.g. using computer tests, see Bilderbeck et al., 2013) and (neuro)physiological measures of mental health. Examples of these physiological measures included heart rate variability or HRV (see Ross & Thomas, 2010), neurotransmitters (see Streeter et al., 2010) and stress hormones (Sieverdes et al., 2014). However, more research into objective measures of cognitive functioning or somatic markers of mental health is needed.

Furthermore, the current study looked into specific symptoms domains to measure mental health, but there is reason to believe that measures of more general well-being can more easily determine the effects of yoga on prisoners. To illustrate, those studies on yoga in prisons that yielded positive results used measures of general stress, general psychological distress or general moods (e.g. Bilderbeck et al., 2013). However, a number of studies did find positive effects on depression and anxiety (cf. Harner et al., 2010; Sistig et al., 2015) using the Beck Depression/Anxiety Inventory resp. the Hospital Anxiety and Depression Scale, suggesting that our instruments may have not been suitable enough for our study’s population.

Another limitation refers to the screening that was used in the current study. For example, we intended to obtain information about past or present psychological diagnoses and did so by asking participants during interviews. However, we found that participants’ reports of their psychiatric background were not informative enough. We suppose that low cognitive age and developmental problems prevented participants from providing comprehensive information on these matters, just as low attention spans may have negatively influenced assessments. It is already suggested that a much larger proportion of prisoners suffer from attention deficit/hyperacitivity disorder (ADD/ADHD) than the general population (Zylowska et al., 2008). Further, it should be taken into consideration that our sample consisted of many second or third generation immigrants, which possibly led to underreporting of mental health problems (Brown et al., 1999). Evidently, future studies would benefit from taking cultural aspects into consideration and screening their sample for developmental disorders, intelligence, and comorbidity e.g. by means of case files or proper briefing by the facility. The

importance of obtaining more complete patient information is emphasized by the unexpected fact that we found low symptom pressure at pre assessment. Moreover, we found that the majority of our participants were currently engaged in other semi-therapeutic programs running at the facility such as artistic groups. Because our sample size was too limited, we could not exclude participants on such “extra-curricular” criteria, but future studies may want to take this into consideration.

The fact that improvements were not maintained over time may be explained by the lack of continuity over the second half of the intervention. It can be hypothesized that individuals need continuing practice of mindfulness for its working mechanisms to establish remaining effects. In our study, three classes were cancelled over a short period of time just before post assessments took place, suggesting that the intervention’s effects may have already decreased at that time. This raises another relevant limitation; this study did not manage to do a follow-up and was limited in assessing the long-term effects of yoga interventions. It is therefore recommended for future studies to not only look at interventions with longer durations but to also assess participants’ well-being later on after the intervention has taken place.

Continuity may have also been undermined because it was allowed for newcomers to join in throughout the program. Because of this influx of new students, class’ instructions and theory had to be repeated on several occasions which negatively influenced the students’ learning curves. It is also probable that a lack of group stability jeopardized the effects of the intervention, given that group