Effects of Mindfulness-Based Stress Reduction on employees’ mental health: A systematic review

Objectives The purpose of this exploratory study was to obtain greater insight into the effects of Mindfulness-Based Stress Reduction (MBSR) and Mindfulness-Based Cognitive Therapy (MBCT) on the mental health of employees. Methods Using PsycINFO, PubMed, and CINAHL, we performed a systematic review in October 2015 of studies investigating the effects of MBSR and MBCT on various aspects of employees’ mental health. Studies with a pre-post design (i.e. without a control group) were excluded. Results 24 articles were identified, describing 23 studies: 22 on the effects of MBSR and 1 on the effects of MBSR in combination with some aspects of MBCT. Since no study focused exclusively on MBCT, its effects are not described in this systematic review. Of the 23 studies, 2 were of high methodological quality, 15 were of medium quality and 6 were of low quality. A meta-analysis was not performed due to the emergent and relatively uncharted nature of the topic of investigation, the exploratory character of this study, and the diversity of outcomes in the studies reviewed. Based on our analysis, the strongest outcomes were reduced levels of emotional exhaustion (a dimension of burnout), stress, psychological distress, depression, anxiety, and occupational stress. Improvements were found in terms of mindfulness, personal accomplishment (a dimension of burnout), (occupational) self-compassion, quality of sleep, and relaxation. Conclusion The results of this systematic review suggest that MBSR may help to improve psychological functioning in employees.


Introduction
Given their potential benefits for physical and mental health as well as social relations [1,2], interest is increasing internationally in mindfulness interventions in the workplace [3]. This is also true for the Netherlands, where many work environments are characterized by high productivity targets, overtime, high work pressure, customer aggression, temporary employment contracts, continuous organizational changes, job uncertainty, employee shortages, and little autonomy [4].
Mindfulness is related to meditation, but the terms are not synonymous. Moreover Mindfulness consists of formal meditation exercises (e.g., paying attention to the body, lying on the ground, or walking slowly with a sense of awareness of one's surroundings) as well as informal exercises (e.g., paying full attention to what one is doing or experiencing at a certain moment) [5].
Mindfulness interventions vary in delivery mode (face-to-face, online) and target population (clinical populations with major depression, anxiety disorders, borderline personality disorders, chronic pain, or eating disorders [6] and non-clinical populations such as students and employees seeking to enhance their subjective well-being). They can range in duration from long term (e.g., eight weekly 2.5-hour sessions, eight-hour daylong retreats, one 2.5-hour session per month for 10 months [7]) to short term (e.g., four weekly 30-minute sessions [8]). The intensity of interventions can vary too, from high dose (e.g., eight weekly 3-hour sessions and 45 minutes of daily mindfulness practice [9]) to low dose (e.g., 30-minute sessions and 15-20 minutes of daily mindfulness practice [8]).
Mindfulness interventions in the workplace target workplace functioning: reducing stress and improving decision-making, productivity, resilience, interpersonal communication, organizational relationships, perspective-taking, and self-care [10]. This great diversity in mindfulness interventions makes it difficult to compare the efficacy of such interventions.
There are many hypotheses about the mechanisms underlying mindfulness practice that lead to different mental health outcomes. One hypothesis is that exposure to or willingness to experience difficult emotions (e.g., anxiety, distress, anger), awareness of these emotions, and observation of these emotions allow people to dis-identify with and better regulate difficult emotions [5,11]. Another is that awareness of thoughts, awareness of bodily sensations, and self-compassion help people to deal with stress [5].
This systematic review will evaluate the effects of two types of group-based mindfulness interventions-Mindfulness-Based Stress Reduction (MBSR) and Mindfulness-Based Cognitive Therapy (MBCT)-on employees' mental health.

Definition of mindfulness
The concept of mindfulness has existed in Buddhist traditions for 2,500 years. Mindfulness meditation is an attitude and a method for reducing personal suffering and developing insight, compassion, and wisdom [12]. In contemporary psychology, mindfulness is seen as a means of increasing awareness and responding optimally to mental processes that contribute to emotional distress and maladaptive behavior [6].
Many definitions of mindfulness have been posited in the psychological literature [6,[13][14][15]. Marlatt and Kristeller [13] described mindfulness as "bringing one's complete attention to the present experience on a moment-to-moment basis." According to Brown and Ryan [14], "mindfulness can be considered an enhanced attention to and awareness of current experience or present reality." 'Awareness' refers to the individual's consciousness of what they are experiencing, without those experiences being the center of attention. 'Attention' is the process of focusing conscious awareness on specific experiences.
Bishop et al. [6] proposed a working definition of mindfulness with two components. The first concerns the self-regulation of attention, which is focused on the immediate experience in the present. The second involves having an open, curious, and accepting attitude towards that experience. Kabat-Zinn [15] described mindfulness as "the awareness that emerges through paying attention, on purpose, in the present moment, and non-judgmentally to the unfolding of experience moment by moment" (p. 145). This awareness can be aimed at internal experiences (bodily sensations, feelings/emotions, and thoughts) and external experiences (what one sees, hears, smells, tastes, and touches).
Shapiro and Carlson [16] defined mindfulness as "the awareness that arises through intentionally attending in an open, caring, and discerning way." This definition contains three interrelated elements. The first element-intention-involves reflecting on one's personal goals and values, and paying attention to the most important things in accomplishing and upholding them. The second element, attention (i.e., attending to experiences in the here and now) is a prerequisite for seeing clearly. Finally, while intention refers to why we are paying attention, the third element-attitude-relates to how we pay attention. It refers not to an attempt to change things, but to an effort to relate to them in a non-judgmental way, with curiosity and compassion [10].
In this review article, we adopt Shapiro and Carlson's [16] definition, which is an elaboration of the commonly accepted definition by Kabat-Zinn [15].

Interventions based on Kabat-Zinn's mindfulness training
The most common form of secular mindfulness-based training is Mindfulness-Based Stress Reduction (MBSR) training [17], developed by Kabat-Zinn [18]. MBSR aims to alleviate suffering [19] and was originally developed for patients with chronic pain. It consists of eight 2.5-hour weekly sessions and one 7-hour day of silence. A very important part of the training is the homework: 45 minutes of daily exercise at home, 6 days a week, with the support of CDs and set tasks. MBSR includes: • the body scan (paying attention to what the body is feeling); • sitting meditation (paying attention to breathing, sounds, thoughts, bodily sensations, feelings/emotions); • simple movement exercises such as walking or standing meditation, or lying yoga exercises (paying attention to what the body is feeling; exploring and accepting borders); • informal meditation exercises: paying full attention to daily activities (e.g., brushing one's teeth, taking a shower, eating).
Mindfulness-Based Cognitive Therapy (MBCT), which was developed by the cognitive behavior therapists Segal, Williams, and Teasdale [20] on the basis of MBSR, is often used to prevent relapse into depression [21][22][23]. In workplace settings, the focus is on distressing emotional states rather than on clinical depression (e.g., see the pre-post study by Ruths et al. [24]). MBCT has become the most important adaptation of MBSR [25], to which it is closely allied, although there are a few differences. The CT component of the program includes psycho-education about the nature of thoughts as mental events rather than facts, which fosters a decentered attitude towards one's own thoughts: thoughts are not facts; I am not my thoughts. The link between thought and mood is made explicit in MBCT. Moreover, MBCT also introduces a 'mini-meditation', known as the three-minute breathing space.
Other interventions incorporate mindfulness training as well, including Dialectical Behavior Therapy for the treatment of clients with a borderline syndrome [26], Acceptance and Commitment Therapy [27], and Mindfulness-Based Relapse Prevention [28]. As mindfulness is only one component of the treatment in these interventions, they are not included in this review.
MBSR and MBCT interventions are typically modified to suit the context in which they are delivered (Table 1). Sessions can vary in number and duration; they may be face-to-face or online, and involve less or more homework; and the retreat day may be dropped. In this study, we include MBSR and MBCT interventions conducted in groups with at least four face-to-face 30-minute sessions.
[50] identified potential benefits for depression, but not for anxiety disorders.

Effects of mindfulness interventions on healthy people and professionals
Following on from the promising results for patients, MBSR and MBCT have recently also been used for healthy people [51] and for employees and managers in a healthcare setting [52]. A meta-analysis by Chiesa and Serretti [51] focusing on healthy participants (not explicitly on employees) identified ten studies on MBSR, most of them of low methodological quality. The most striking outcome was the reduction of stress levels.
Employees and managers in a healthcare setting are regularly confronted with stress in the form of physical and mental suffering as well as strong emotions (their own or those of their patients) [53]. Adequate stress management can improve the health of these professionals [54] and the quality of care they provide to patients [55].
Mindfulness training is assumed to also have potential for other demanding work environments, as a means of improving employees' health and work engagement and, consequently, the quality of services provided to clients. To date, however, there has been no systematic review of studies investigating the effects of mindfulness training on the mental health of employees across different occupational sectors.
The most commonly studied group of employees are healthcare professionals: four reviews [52, 56-58] and two reviews/meta-analyses [59,60] focused exclusively on healthcare professionals and students. Escuriex and Labbé [56] found no clear correlation between a therapist's mindfulness and treatment outcomes. Irving et al. [52] reported that MBSR benefits the physical and mental health of clinicians. Morgan et al.
[57] discussed 14 qualitative studies and concluded that the benefits of MBSR "ranged from increased personal well-being and selfcompassion to enhanced presence when relating to others, leading to enhanced compassion and a sense of shared humanity" (p. 744). Smith [58], who examined nurses, concluded that MBSR helps them deal with work-related stress. The meta-analysis by Regehr et al. [59] supports the idea that mindfulness interventions reduce stress, anxiety, and burnout among medical students and practicing physicians. Similarly, Burton et al.'s [60] meta-analysis suggests  (7) Significant improvements in all aspects, except for sleep efficiency and the use of medication to sleep, in the TG compared to the CG. (7) 3) Mean age TG and CG: 40.72 3) Moments of measurement: 1. at baseline; 2. post class 3) Self-compassion Scale (SCS). Subscales (self-kindness; selfjudging; common humanity; isolation; mindfulness; overidentification) and total selfcompassion (6) Significant improvements in all subscales, except for common humanity and isolation, in the TG compared to the CG.   Significant decrease in daily symptoms in TG compared to the CG.

5) Teachers' Sense of Efficacy
Questionnaire (TSES). Total score and three subscales: instructional strategies; student engagement; classroom management (14) Significant increase in total sense of self-efficacy, instructional strategies and student engagement in the TG compared to the CG. that mindfulness training can significantly decrease stress levels among healthcare professionals. Virgili [61] performed a meta-analysis involving three subgroups of employees: healthcare professionals, teachers, and general/office employees. The aim was to examine the effect of a mindfulness-based intervention on a single outcome: employees' psychological distress. Unfortunately, the composition of the outcome 'psychological distress' was not explained and no other outcomes, positive or negative, were considered [61].
Two other meta-analyses [51, 62] did not explicitly focus on employees. Grossman et al. [62] examined a relatively small number of clinical and stressed non-clinical participants and found possible benefits of MBSR. Finally, another recent review [63] provided strong evidence that mindfulness-based interventions can reduce occupational burnout among healthcare professionals and teachers.

Aim of the study
The aim of this systematic review is to gain deeper insight into the effects of two mindfulness interventions-MBSR and MBCT training-on employees' mental health across different occupational sectors. Looking at different occupational sectors gives us an opportunity to draw general conclusions about the effects of the interventions on employees and to identify potential differences between sectors.
This review uses the World Health Organization's (WHO) definition of mental health as "a state of well-being in which the individual realizes his or her own abilities, can cope with the normal stresses of life, can work productively and fruitfully, and is able to make a contribution to his or her community" [64]. The WHO definition has three main components: emotional or subjective well-being [65], psychological well-being (referring to optimal functioning in work and life) [66], and social well-being (targeting optimal functioning in social groups and society) [67].

Method
We conducted a systematic search of the literature with three rounds of screening. The flow chart in Fig 1 outlines our review process and findings.

Search
All searches were performed in October 2015. The search in PsycINFO yielded 1790 articles.
The following procedure was used to screen them: 1. First screening of the titles/abstracts. Independent categorization (include, exclude, or unclear) by three authors, as follows: • The first author (MJ) screened all 1790 articles based on the titles and abstracts. • To check the quality of this screening, the second (YH) and third (WK) authors independently screened random samples from the 1790 articles; specifically, from every five articles, one abstract was randomly selected and screened. Any differences in opinion among the three authors were discussed until consensus was reached.
2. Second screening of articles whose relevance was classed as 'unclear'. All 32 articles were independently screened by the first three authors as follows: • Each author assessed each abstract and, if necessary, the entire article.
• Any differences in categorization were discussed until consensus was reached.
3. Third screening of all remaining articles. The first, second, and fifth authors (MJ/YH/JE) assessed the full texts of the 18 articles initially labelled 'include' as follows: • Each author independently assessed the articles.
• They then discussed each article until consensus was reached as to their in-/exclusion.
The same search string and screening procedure were used for the 451 and 223 articles obtained from PubMed and CINAHL, respectively.

Inclusion and exclusion criteria
The following inclusion criteria were used for articles in the first and second screenings: • Full text was available.
• Published in English.
• Mindfulness training (MBSR or MBCT) was operationalized as "moment to moment awareness to be cultivated with a nonjudgmental attitude; teaching of formal meditation techniques; and stressing the importance of daily and systematic practice" [16]. Once we had compiled a set of eligible studies, we searched their reference lists for additional relevant studies ('snowball' method). For articles that did not provide a definition of mindfulness training, two criteria had to be met: • References to relevant authors (e.g., Jon Kabat-Zinn, Ruth Baer, Mark Williams, John Teasdale, Zindel Segal).
• Presence of the following elements in the mindfulness training: a focus on the 'here and now', an emphasis on increasing consciousness, learning to use a nonjudgmental attitude, learning formal meditation techniques, and daily and systematic practice.
• Study population: employees or managers in workplace settings.
• Group-based rather than individual interventions (i.e., no individual therapy or coaching).
• At least four face-to-face 30-minute sessions.
• Qualitative and/or quantitative work-related outcomes, including psychological/mental factors.
The following exclusion criteria was used in the first and second screenings: • Dissertations, conference papers, reviews, chapters in handbooks, editorials.
• Studies on forms of meditation that do not emphasize mindfulness, such as transcendental meditation. In some meditation approaches, attention is directed towards one object or stimulus, such as a word (a mantra), sound, or sensation. In contrast, mindfulness meditation emphasizes the observation of constantly changing internal and external stimuli [29].
• Studies on the cognitive model of mindfulness developed by Ellen Langer [68,69], which involves working with external material such as information and includes active, goal-oriented cognitive tasks, such as problem solving. We consider her approach to be distinct from that used by other mindfulness studies.
• Studies on interventions that included mindfulness training as a component of a broader treatment program, such as Dialectical Behavior Therapy [26], Acceptance and Commitment Therapy [27], and Mindfulness-Based Relapse Prevention [28].
• Population: employees with a specific mental or physical disorder and students/trainees.
The inclusion criteria for the third screening were: • Description of qualitative and/or quantitative psychological/mental outcomes.
• Research design: • Randomized controlled trial (RCT), or • Quasi-experimental design with an intervention group and a control group, where the control group comprised those on a waiting list or those receiving treatment as usual (e.g., relaxation practices, Jacobson relaxation, or an attention placebo) (quasi-RCT).

Methodological quality and levels of evidence
As indicated above, we included two types of research design-RCT and quasi-RCT-and excluded pre-post studies. In RCT and quasi-RCT studies, participants in the control group (CG) were compared with those in the treatment group (TG) both before and after the treatment (T1 vs. T2, between-group analysis). The methodological quality of the selected studies was assessed using the nine criteria outlined in a paper entitled 'Assessment of a randomized controlled trial' by the Dutch public health institute Kwaliteitsinstituut voor de Gezondheidszorg CBO [71]. The studies were classified as 'high quality' if they met ! 7 of the criteria, as 'medium quality' if they met 5 or 6 of the criteria, and as 'low quality' if they met 4 of the criteria.
The nine criteria were formulated as closed questions with three possible answers: (1) yes, (2) no, or (3) not enough information. The questions were as follows: 1. Were the participants randomly assigned to the intervention? 2. Was the researcher (or research assistant) who assigned the participants blinded to the randomization order?
3. Were the participants blinded to the treatment?
4. Were the practitioners blinded to the treatment? 5. Were the assessors who evaluated the effects (i.e., mental health outcomes) of the treatment blinded to the treatment?
6. Were the control and intervention groups comparable at the beginning of the trial?
7. Were enough participants available for follow-up after the intervention(s)?
8. Were the participants analyzed in the group in which they were randomized? 9. Were the groups treated in the same manner, except for the intervention(s)?
Domusmedica [72] outlines three levels for assessing the importance of outcomes. An outcome can be classified as Level 1 ('there is evidence for . . .' / 'it has been proven that . . .') if at least two high-quality RCTs show significance between groups (change in TG versus change in CG). An outcome can be classified as Level 2 ('it is plausible that . . .') if at least two mediumquality RCTs show significance between groups. Finally, an outcome can be classified as Level 3 ('there are indications that . . .') if at least one medium-quality RCT shows significance.
We used five levels of evidence based on the levels described above. Our definitions of Levels 1 and 2 were the same as the Domusmedica definitions, although we found no outcome that could be classified as Level 1 [72]. Our definition of Level 3 was slightly different: we classified an outcome as Level 3 if at least one medium-quality RCT or three low-quality experiments (RCT, quasi-RCT) showed significance between groups (change in TG versus change in CG). Level 4 ('the evidence is weak') outcomes refer to one or two low-quality experiments (RCT, quasi-RCT) with significance between groups (change in TG versus change in CG). Level 5 ('there is no evidence') indicates that there are no studies with significant results between groups for the relevant outcome.

Studies included
The results of the review process are shown in Fig 1. The PsycINFO search and three screenings yielded 16 articles. The PubMed and CINAHL searches and screenings resulted in five and three additional articles, respectively. We thus assessed 24 articles in total, describing 23 studies. Table 1 presents an overview of the 24 articles, including the number of participants and their characteristics, the research design, treatment method, assessment instruments, and outcomes. The 24 articles refer to 23 studies as two articles described the same study, presenting the quantitative [73] and qualitative [74] data separately.

The effects of MBSR on employees' mental health
This section describes the effects of MBSR on employees' mental health in the 23 studies reviewed. Several studies, e.g., Klatt et al. [78], Wolever et al. [82] and Geary and Rosenthal [86], also measured stress biomarkers (e.g., salivary cortisol, pulse rate and heart rate variability), but the effects of MBSR on such biomarkers fall outside the scope of our discussion. The mental health outcomes are presented in order of importance, taking into account two criteria. First and foremost is the level of evidence; the second criterion refers to the number of studies reporting a particular (significant or non-significant) result. When two effects have the same level of evidence from the same number of studies, they are described in alphabetical order.
Almost 35 mental/psychological outcomes were identified, some of them overlapping (e.g., stress and occupational stress, or mood and depression). Some outcomes were measured using different assessment instruments: e.g., stress level by the Perceived Stress Scale (PSS) and the Depression Anxiety Stress Scale (DASS); burnout by the Maslach Burnout Inventory (MBI) and the Professional Quality of Life (ProQOF) scale.
All results are presented in Table 3. Results classified as Levels 3, 4, or 5 and reported in a single study only are mentioned in Table 3, but not discussed further in the text.

Burnout.
Nine studies examined the effects of MBSR on burnout symptoms [8,53,73,76,79,83,85,88,90]. The three main symptoms of burnout are emotional exhaustion, (job-related) personal accomplishment, and depersonalization. Three of the studies [53, 85,88] only reported on burnout in general. Five [8,73,76,79,83] reported the effects of MBSR on the three symptoms, but not on burnout in general. One study [90] dealt both with burnout in general and with the individual symptoms. Table 2. Quality criteria and quality of the selected studies quality criteria. 1) Assignment intervention randomized; 2) Includer blinded for randomization order; 3) Employees blinded for treatment; 4) Practitioner blinded for treatment; 5) Assessor blinded for treatment (mental health outcomes); 6) Groups comparable at the start of the trial; 7) Follow up available of enough included employees at T1, etc. l ; 8) Included employees analyzed in randomized group; 9) Same treatment of groups except the intervention.  • Burnout in general: two studies reported a significant reduction in the TG [88,90]; two reported no significant outcome [53,85].

Quality
• Emotional exhaustion: two studies [8,73] reported a significant reduction in the TG and in the TG compared to the CG. Two studies [79,90] showed a significant reduction in the TG compared to the CG after the intervention. Two studies [76,83] reported no significant outcome.
• (Job-related) personal accomplishment: levels increased significantly within-group and between-groups in three studies [73,79,83]. One study identified a significant increase in the TG [90], while two studies [8,76] reported no significant differences.
• Depersonalization: one study [90] showed a significant reduction after the treatment in the TG and in the TG compared to the CG. In another study [79], depersonalization remained stable in the TG before and after the intervention (positive result), and increased significantly in the CG. Three other studies [73,76,83] reported no significant changes in depersonalization.
In summary, it is plausible that MBSR results in increased (job-related) personal accomplishment and decreased burnout in general and emotional exhaustion (evidence Level 2). There are indications that MBSR causes a decrease in depersonalization (cynicism and lack of empathy) (evidence Level 3).
3. Stress level. Stress level, measured as perceived stress (mostly by the PSS), was investigated in nine studies [53, 75-78, 80, 82, 86, 87]. Eight studies reported a significant reduction in stress level after the intervention in the TG compared to the CG; one study [77] found no significant outcome. In summary, it is plausible that MBSR helps to reduce stress levels (evidence Level 2).

Psychological distress.
Eight studies investigated the effects of MBSR on psychological distress [53, 73,80,81,[85][86][87]92]. Psychological distress was mostly measured by the Brief Symptom Inventory (BSI), which consists of 10 subscales reflecting different mood states (e.g., anxiety, depression, total mood disturbance). Three studies [80,86,87] showed a significant reduction in psychological distress in the TG and in the TG compared to CG; one study [81] mentioned a significant reduction in the TG. The other four studies reported no significant differences. In summary, it is plausible that MBSR results in a decrease of psychological distress (evidence Level 2). [82,83,88,90,91]. Two studies [88,91] found a significant decrease of depression in the TG and in the TG compared to CG, one study [90] reported a significant reduction in the TG, and the other two studies [82,83] mentioned no significant results. In summary, it is plausible that MBSR results in decreased levels of depression (evidence Level 2).

(Occupational) self-compassion.
Three studies addressed self-compassion, measured by the Self-Compassion Scale (SCS) [53, 75,85], and one study addressed occupational self-compassion, measured by a modification of the SCS for teachers [88]. All four studies reported a significant increase in self-compassion in the TG and in the TG compared to the CG. Cohen-Katz et al.'s [74] qualitative study described an improvement in self-care resulting from MBSR: "I'm worrying about my own needs first, and trying to take care of them" (p. 82). Self-care can lead to feelings of guilt [74]: "they want to fix everyone else in the group!", as one participant stated, "it's the nurse in me!" (p. 85). In summary, it is plausible that MBSR leads to a significant increase in self-compassion (evidence Level 2).

Quality of sleep.
Three studies [78,82,85] investigated quality of sleep using the Pittsburgh Sleep Quality Index (PSQI), which consists of seven subscales: general (subjective) sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of medication to sleep, and daytime sleep-related dysfunction. Frank et al. [85] reported significant improvements in the TG compared to the CG in all aspects except for sleep efficiency and the use of medication to sleep. Wolever et al. [82] reported significant reductions in sleep difficulty in the TG compared to the CG.
Klatt et al. [78] found a significant improvement in the TG in terms of subjective sleep quality, sleep latency, sleep disturbances, and daytime dysfunction (i.e., four of the seven components of sleep quality). Changes in global sleep scores (PSQI) were significant for both the TG and the CG. Klatt et al. [78] suggested that the improvements in subjective sleep quality may be the study's most important result.
In summary, it is plausible that MBSR gives rise to a significant increase in quality of sleep (evidence Level 2). However, it is not possible to provide insight into the different components of sleep quality. 8. Relaxation. Three studies measured relaxation [7,8,79] and all found significant changes in relaxation in the TG compared to the CG. Cohen-Katz et al.'s [74] qualitative data also showed an increase in relaxation/calmness: "I'm feeling a greater calm and peace" (p. 82). Thus, it is plausible that MBSR produces a significant increase in relaxation (evidence Level 2). 9. Anxiety. Two studies measured anxiety [88,91] and both showed a significant decrease in anxiety in the TG compared to the CG. It is therefore plausible that MBSR causes a significant decrease in anxiety (evidence Level 2).
10. Occupational stress. Two studies [88,89] specifically measured occupational stress (i.e., stress employees experience on the job). Both reported a significant reduction in occupational stress after the intervention in the TG compared to the CG. In summary, it is plausible that MBSR causes a reduction in occupational stress level (evidence Level 2).
11. Life satisfaction. Eight studies examined the effects of MBSR on general life satisfaction [8, 53, 75-77, 79, 86, 92]. Four [8,75,79,86] reported significant positive changes in life satisfaction or general well-being in the TG compared to the CG. By contrast, four studies [53, 76,77,92] reported no significant improvement in terms of life satisfaction. In summary, there are indications that MBSR leads to an increase in life satisfaction (evidence Level 3).
12. Mood. Mood was examined in two of the studies [83,90]. Martín-Asuero et al. [90] used a short version of the Profile of Mood States (POMS), which consists of five subscales that measure negative emotions. They reported a significant reduction of total mood disturbance in the TG and in the TG compared to the CG. Jennings et al. [83] used the Positive and Negative Affect Schedule (PANAS), which measures positive and negative effects, and found no significant outcomes.
One qualitative study [74] described an improved ability to deal with difficult emotions following MBSR. In summary, there are indications that MBSR has a positive effect on mood (evidence Level 3).
13. Efficacy in regulating emotions at work. Two studies [85,89] measured employees' perceived self-efficacy regarding their ability to regulate their emotions on the job. Frank et al. [85] reported a significant improvement in the TG compared to the CG; Taylor et al. [89] showed a significant increase in the TG. In summary, it is plausible that MBSR improves efficacy in regulating emotions at work (evidence Level 3). 14. Self-efficacy/locus of control. Self-efficacy was investigated in two studies [83,92]. Jennings et al. [83] mentioned a significant increase in the TG compared to the CG, while Walach et al. [92] reported no significant changes. In summary, there are indications that MBSR causes an increase in self-efficacy in general (evidence Level 3).
15. Work engagement. Work engagement was measured in two studies [9,84]. Leroy et al. [9] showed a significant increase in work engagement for the group as a whole between T1 and T2; the increase in the TG compared to the CG appeared to be significantly higher. Klatt et al. [84] reported a significant increase in work engagement in the TG, mostly induced by the vigor subscale (a subscale of the Utrecht Work Engagement Scale, UWES). In summary, there are indications that MBSR produces a significant increase in work engagement (evidence Level 3).

Discussion
To the best of our knowledge, this is the first review of the mental health of employees across different occupational sectors. We identified 24 articles representing 23 studies on the effects of MBSR (2 high quality, 15 medium quality, and 6 low quality) published before October 2015.
Demanding workplace challenges can produce stress and symptoms of burnout [93]. In the Netherlands, the professionals reporting the highest levels of work pressure and stress are teachers and healthcare providers [94], and teachers appear to have the highest burnout percentage in the Dutch workforce [95]. Burnout is a major cause of loss of engagement, disease, and disability [96].
There is a great need for useful, practical workplace interventions that could reduce stress and enhance work engagement. Kabat-Zinn [18], who introduced the concept of mindfulness, suggested its possible use as a person-centered intervention. This study was aimed at investigating empirically whether MBSR and MBCT indeed contribute to employees' mental health.

General findings
Measuring mindfulness. Fourteen of the 23 studies reviewed [7, 9, 73, 75-78, 80, 82, 83, 85, 88, 90, 91] measured levels of mindfulness; 9 did not. This is striking because, as Cohen-Katz et al. [73] points out, measuring mindfulness is important to determining whether an MBSR program has successfully taught what it was designed to teach. Mindfulness skills may be the mediating factor for individual outcomes such as stress reduction, relaxation, and empathy [97]. Indeed, Baer [29] highlighted several mechanisms that might account for how mindfulness skills can reduce symptoms and bring about behavioral change: exposure, cognitive change, self-management, relaxation, and acceptance.
Three different mindfulness measures were used in the studies we reviewed: five used the Mindfulness Attention Awareness Scale (MAAS), eight used the Five Facets Mindfulness Questionnaire (FFMQ) with five subscales, and one used the Cognitive and Affective Mindfulness Scale-Revised (CAMS-R). These three scales correspond with different definitions of mindfulness. In the MAAS, mindfulness is operationalized as a one-dimensional construct involving attention and awareness. The FFMQ, which encompasses five dimensions of mindfulness, also includes an attitudinal component. The CAMS-R, which assesses four elements of mindfulness, yields only one score [98]. The use of different ways of operationalizing mindfulness makes it difficult to compare its effects. In addition, mindfulness can be said to be both an outcome variable of interest and an important mechanism of the therapeutic agent.
The advantage of the FFMQ is the fact that Baer's [29] five subscales result in a sophisticated measurement of mindfulness. As it is based on a factor analysis of items from the five most widely used mindfulness questionnaires, it can also be considered particularly robust [99,100].
Other outcomes. Most of the variables were assessed using negative symptom-focused outcome measures (e.g., burnout, stress level, psychological distress, depression, anxiety), while some used positive symptom-focused outcome measures (e.g., quality of sleep, relaxation, work engagement, job satisfaction). There were hardly any symptom-focused outcome measures relating to work performance (e.g., caring efficacy, work behavior, work performance, workability).
Few variables, except for mindfulness skills (observing, describing, acting with awareness, non-judging, non-reactivity), were assessed using process-focused measures. They may have been more suitable for capturing the mechanisms by which mindfulness practice leads to specific outcomes (e.g., (occupational) self-compassion, (occupational) self-efficacy, sense of coherence, coping strategies, ruminating). Specifically, four of the studies [53, 75,85,88] reported a significant increase in self-compassion in the TG compared with the CG. In another study, Shapiro et al. [11] reported that MBSR weakens stimulus-response relationships, thereby reducing reflexive behavior. This was associated with increased levels of self-management, self-efficacy, and self-care.
Self-care can be challenging for healthcare professionals and others engaged in people-centered work (e.g., teachers), as their job is to care for others. A lack of self-care can eventually undermine health and, as a result, sustainable employability [101]. Thus, employees' self-care is a highly important topic that deserves more attention within organizations. Our findings suggest that it might be increased by means of mindfulness training.
Behavioral variables, such as assessments of the quality and quantity of formal and informal meditation (e.g., frequency and intensity of practice) were measured in only three studies [9,74,92].
Self-reported measures are inherently biased because participants who have invested a lot of time and energy in a treatment program, such as MBSR, are less likely to give negative evaluations [102]. Therefore, behavioral reports by relevant others may be a useful addition.
A few outcomes in our review refer to perceptions of work characteristics (e.g., work relationships, job control, job demands). However, many other possible work-related perceptions were not measured (e.g., work pressure, emotional load, feedback, autonomy, learning opportunities). Further empirical work on the effects of mindfulness on perceptions of work could be revealing.
Future research on mindfulness treatments for employees may benefit from assessing a diversity of outcomes: negative and positive symptom-focused measures of mental health (mood, recovery need, job satisfaction, work engagement) and work performance (positive and negative work behavior, absence from work); process-focused outcome measures; behavioral measures of (in)formal practices; behavioral reports by relevant others (e.g., colleagues); and outcomes on work-related perceptions. Finally, future research may include outcomes differentiating between elements of mindfulness interventions, as these remain something of a black box thus far in terms of their working mechanisms.
MBCT. Our search method, which included both MBSR and MBCT programs, uncovered only one study [75] exploring the effects of a modified MBSR program in combination with some aspects of MBCT on employees. MBCT is often used in clinical settings to prevent depression relapse [20]. However, it could also be useful for employees, especially those who experience non-productive or irrational thoughts, which can cause stress. Two examples of excluded pre-post studies focusing exclusively on the effects of MBCT on employees are Ruths et al. [24] and Schenström et al. [103]. Future research on the impact of MBCT on employees is needed to compare its efficacy to that of MBSR.
Contraindications. None of the studies we reviewed reported that MBSR training negatively affected employees. The absence of empirical evidence on potentially harmful effects of MBSR [104] does not mean it is good for everyone in every situation; instead, potentially harmful effects should be thoroughly evaluated in future work. Ruths et al. [24], who investigated the effects of MBCT on 27 mental-health professionals in a pre-post study (which was therefore excluded from our study), mentioned depersonalization and initial mood deterioration as potentially harmful effects of meditation. Participants with severe trauma or those at risk for psychosis might be at elevated risk [49]. Dobkin et al. [104] made several recommendations in this regard, including pre-program screening to "assess the patient's ability to: (1) contain affect; (2) listen and respond in the present; (3) utilize instructional audio tapes and follow classroom instruction; (4) remain in the classroom; (5) practice yoga or equivalent; and (6) organize thoughts, manage logistics, and time commitment" (p. 4). Nonetheless, there is little empirical research on and evidence for the harmful effects of mindfulness training [49].

Methodological issues
Despite the promising findings of the studies in our review, our approach has some limitations which make it difficult to draw strong conclusions about the effects of MBSR on employees.
Meta-analysis versus systematic review. We chose to conduct a systematic review instead of a meta-analysis for four reasons. First, the quality of the studies conducted so far is limited; we only found 2 studies that could be classified as of high methodological quality, while 15 were of medium quality and 6 were of low quality (Table 2).
Second, mindfulness training is an emergent, relatively uncharted field of investigation, with a broad variety of outcome measures. Only 7 variables were examined in four or more studies, 8 in two or three studies, and 22 in one study only (Table 3). We intended to explore several negative and positive symptom-focused outcomes on mental health and work performance; in particular, positive and negative process-focused outcomes and positive and negative outcomes related to work characteristics. Furthermore, we wanted to use quantitative and qualitative data.
Third, the samples in 11 of the studies reviewed were relatively small, ranging from just 10 to 24 participants.
Finally, the 2 high-quality studies and 15 medium-quality studies appeared to have similar outcome variables in a few cases (as shown in Table 1).
The (lack of) clarity around the outcome variables. We identified many mental/psychological outcomes which sometimes partly overlapped (e.g., stress and occupational stress; mood and depression; efficacy in regulating emotions at work and self-efficacy/locus of control). Occupational stress refers to only one cause of stress, while depression is just one aspect of mood, measured by five subscales (tension-anxiety, depression-dejection, anger-hostility, vigor-activity, and fatigue-inertia) of the Profile of Mood States (POMS). In a similar vein, efficacy in regulating emotions at work appears to be a single dimension of self-efficacy/locus of control.
In some cases, outcomes were measured using different assessment instruments. For example, mood was measured by the POMS, which consists of five subscales, and by the Positive and Negative Affect Scale (PANAS), which consists of two corresponding subscales. Mindfulness was measured by the one-dimensional MAAS, the five-dimensional FFMQ or the CAMS-R, with its four elements of mindfulness.
Sometimes the distinction between outcomes was unclear. For example, there is no clear distinction between mood, measured by a short version of the POMS [90], and psychological distress, measured by the Brief Symptom Inventory (BSI). The BSI consists of 10 subscales reflecting different mood states, such as anxiety, depression, and total mood disturbance. The POMS consists of five subscales measuring negative emotions.
As already mentioned, three of the studies reviewed [53, 85,88] reported only on burnout in general. Five studies [8,73,76,79,83] reported on the three dimensions of burnout but not on burnout in general. One study [90] dealt with both types of outcomes.
There is a need for unambiguous, clear outcomes that are logically clustered (e.g., process and effect outcomes; mental-health and work-performance outcomes; positive and negative outcomes; work-related perceptions outcomes; individual and organizational outcomes) and measured by reliable and valid assessment instruments.
Publication bias. All the studies except one [77] reported statistically significant outcomes (Table 3). Three studies (12.9%) [76,81,92] mentioned only one significant outcome, three (12.9%) [7,53,84] mentioned two significant outcomes, and the other 16 studies (68.8%) reported at least three significant outcomes. The multitude of significant results ought to be seen in the context of the small sample sizes used in most of the studies, and the medium and low methodological quality of 15 and 6 studies, respectively, leaving them with relatively low statistical power. It may also reflect some form of publication bias, as studies with positive results tend to be published more easily than studies with negative results [105].
Short-and long-term effects. Little research has considered the long-term effects of mindfulness interventions. Fourteen of the studies reviewed measured short-term effects only [8, 53, 75-80, 82-85, 90, 91], as their final measurements were conducted immediately after the intervention. The other nine studies measured the effects over a longer period: eight weeks to four months post-intervention for six studies [9,73,[87][88][89]92], and almost one year postintervention in three studies [7,81,86]. Mindfulness training for employees should ideally result in sustainable long-term effects. Therefore, both short-and long-term effects of the MBSR program for employees need to be considered in future research.
Due to the different lengths of the programs, we cannot safely conclude anything about the effects of program length on the results. However, it is important to note that seven of the eight shortened MBSR studies (all except Pipe et al. [81]) only measured short-term effects. As Carmody and Baer [106] stated, "the effect of variation in class hours on outcomes has not been systematically studied" (p. 627). They did not find a significant correlation between the number of in-class hours and the mean effect size in clinical or non-clinical samples. Studies on very brief mindfulness interventions in lab settings found temporary effects on emotion [107], mood, and stress [108,109]. Increases tend to be found in state (present) mindfulness rather than trait (more permanent) mindfulness [5]. Bear, Carmody, and Hunsinger [110] revealed that structural changes in perceived stress did not occur until week four of the mindfulness training. Changing problematic, automatic patterns of thought and behavior takes time, and mindfulness is no quick solution.
Carmody and Bear suggested that "adaptations that include less class time may be worthwhile for populations for whom reduction of psychological distress is an important goal and for whom longer time commitment may be a barrier to their ability or willingness to participate" [106]. In Shapiro et al.'s [53] study, 8 of the 18 participants dropped out of the treatment group. The authors concluded that "adding a 2-hour intervention plus daily home practice to an already demanding schedule may not be feasible for a substantial number of health care professionals" (p. 172). Kabat-Zinn [111] emphasized the importance of flexibility in the MBSR program in different contexts. The original MBSR program with eight 2.5-hour sessions, a 7-hour day of silence, and 45 minutes of daily practice at home seems to be too demanding for busy workers [17]. Chaskalson et al. [17] indicated that although the number of organizations offering MBSR is increasing, they all use a format with fewer and shorter sessions. There is a growing need to adapt the duration and dose of the MBSR program to different workplace contexts, in order to better get to grips with its effectiveness.
Another consideration regarding the treatment schedule is whether parts of the program are difficult to adhere to (e.g., regular class attendance, daily home practice, formal and informal meditation exercises). Only Cohen-Katz et al. [74] collected qualitative data to this end. More research is needed on this issue, and to determine whether a shortened version of the MBSR program would produce the same short-and long-term effects. Level of evidence. The level of evidence that we were able to obtain was based on the number of studies investigating a certain variable (see Table 3) and the quality of those studies (see Table 2). Only seven variables were examined in four or more studies, eight variables in two or three studies, and 22 variables in just one study (Table 3). We only found evidence that met the requirements for Levels 2 through 5; we found no evidence meeting the requirements for Level 1. Future studies should seek to generate Level 1 evidence from high-quality RCTs.
Small sample sizes. The sample sizes of 11 studies [7, 8, 53, 73, 76-79, 81, 85, 92] were relatively small: the group sizes (TG and CG) ranged from 10 to 24 non-clinical participants. As statistical significance is essential to the interpretation of findings, future empirical work should be conducted using larger sample sizes.
Future research may benefit from using more homogeneous samples (e.g., workers in healthcare, education, or finance), as these groups have specific demands and challenges with respect to the outcomes of MBSR treatment.
Employees. As previously mentioned, the samples in the studies reviewed largely consisted of healthcare professionals and elementary, secondary, and high school teachers. They do not represent the target population of employees in general. Therefore, more research across different occupational sectors is needed to generalize our conclusions. It may be that including terms such as 'business' or 'corporate' in our search string would have yielded additional references.
Self-selection bias. A major limitation of all the studies reviewed is self-selection, as voluntary participation by employees may result in somewhat biased samples. Specifically, the characteristics of employees who participate in such research (e.g., motivation, sensitivity to the MBSR program, discipline) may differ from those who do not. Self-selection bias thus complicates the evaluation of the MBSR program and the interpretation of results.
Quantitative and qualitative data. Future research may benefit from mixed-methods approaches that combine quantitative and quantitative data. To date such a combination is lacking in the literature.
Twenty of the studies reviewed contained mostly quantitative data, while only three [73,89,92] contained a substantial amount of qualitative data. We found one qualitative review, by Morgan et al. [57], which synthesized 14 qualitative studies on the experiences of health-care professionals and students with mindfulness training and discussed issues such as self-compassion, initial challenges to practice, training focus, and participant motivation.
Since mindfulness research on employees across occupational sectors is a relatively new phenomenon, qualitative data are needed to be able to investigate in depth whether an intervention significantly affects process-focused personal measures (e.g., mindfulness skills, coping strategies) and to capture the mechanisms by which mindfulness practice leads to specific outcomes. For example, unresolved or suppressed emotions may resurface during mindfulness training, contributing to self-consciousness. As Cohen-Katz et al. [74] write, "For the participants (i.e. nurses), becoming aware of the wound was both painful and ultimately extremely useful, helping them to move forward in their lives" (p. 86).
Qualitative data are also needed to thoroughly investigate key aspects of the mindfulness training program and to examine those factors that lead to successful implementation in an organization (e.g., allocating staff time to participate, support of superiors).
Additional organizational intervention. Stress is the result of a complex interaction between environmental factors (work and personal circumstances) and the individual [96,112]. Interventions designed to reduce stress have generally targeted either the relevant environmental factors or the individual, personal factors.
Person-centered interventions seem to be only partly effective in influencing mental health and well-being [113]. An integrated approach is needed that considers not only the person, but the work context as well [114]: for instance, by combining a person-centered intervention such as mindfulness with an additional organizational intervention. None of the studies reviewed used such an additional intervention, which may enhance the effects of the mindfulness intervention.
MBSR program and instructor skill. Specific characteristics of the programs used, such as the instructors' skill levels, may influence the efficacy of MBSR treatment. None of the 23 studies described the MBSR program in terms of its form, content, procedures, or material. Several studies described the instructors as "experienced," but the term was not well-defined. Future studies would benefit from a full explanation of both the MBSR program and the instructors' skill levels.

Conclusions
Our systematic literature search of PsycINFO, PubMed and CINAHL resulted in 24 articles representing 23 studies on the effects of MBSR and MBCT. Their methodological quality varied: 2 were of high quality, 15 of medium quality and 6 of low quality. Given the low number of studies and relatively low methodological quality, it is clear that research on the effects of mindfulness on employees remains a relatively uncharted area.
The outcomes of the studies reviewed suggest that MBSR may help to improve employees' psychological functioning. However, no firm conclusions can be drawn about the effects of specific mindfulness programs for different groups and/or under specific conditions. The strongest outcomes were decreased levels of emotional exhaustion (a dimension of burnout), stress, psychological distress, depression, anxiety, and occupational stress.
We also found a significant increase in mindfulness, personal accomplishment (a dimension of burnout), (occupational) self-compassion, quality of sleep, and relaxation.