Systematic Review of Studies on Group Mindfulness–Based Interventions for Individuals Suffering from Schizophrenia Spectrum Disorders

The search strategy included the following steps: (1) exploration of research databases based on key words and exclusion of meta-analyses, book chapters, conference reports, and letters; (2) automatic removal of duplicates; (3) screening of articles based on acknowledgement of relevant titles and inclusion if they met the criteria; (4) identification of abstracts which fulfilled the criteria of inclusion performed by two independent researchers (an agreement on eligibility was required). CADIMA and ABSTRACKr applications were used to technically support the selection process. Additionally, a literature search was conducted using the bibliography of the included papers and a manual search of databases was performed in order to compare the search results in other reviews and meta-analyses. Searches were conducted from November 2021 to February 2022 and included papers available between 2005 and 2021. An additional manual search was conducted in September 2024 to update the review and included the period from 2021 to 2024 with the main criteria.

The assessment of quality and credibility of clinical studies was carried out using the Cochrane Handbook procedure to estimate the risk of bias (systematic error) and the modified Jadad scale. The choice of both methods was dictated by their widespread use and the complementarity of these procedures (Jadad et al., 1996; Lazzari et al., 2022). The modified Jadad scale was used because the original scale has been criticized for not assessing inclusion and exclusion criteria, adverse effects, and statistical methods. This scale ranges from 0 to 8 points, with the range of 5.5–8 representing high-quality research. Consequently, the following studies received 6.5 points: Langer et al. (2012), Davis et al. (2015), Chadwick et al. (2016), Böge et al. (2021), López-Navarro et al. (2015, 2022), Chien and Lee (2013), Chien and Thompson (2014), Lam et al. (2020). Two analyzed papers were classified below this level, receiving 5 points: Chadwick et al. (2009) and Kun-Hua Lee (2019). Detailed analysis is presented in Table 2 (Appendix, Table 5).

The assessment of bias was performed using the Cochrane Handbook procedure. The process of randomization: low risk was noted, in all presented studies the sequences of distribution of subjects to groups allowed to generate initially comparable groups. The code of allocation: low risk, effective concealment was observed. The blinding method: assessed as low risk; due to the nature of the intervention, and participants and researchers conducting the intervention were not blinded; however, researchers assessing the intervention’s effects were blinded; in two studies (Chadwick et al., 2009; Lee et al., 2019), high risk was noted due to the lack of blinding of researches performing assessments (Chadwick et al., 2009; Lee et al., 2019). Selective reporting: low risk: the results were explained by the circumstances which were considered independent of the implemented procedure and researcher preferences (death, hospitalization, relapse into drug use, other life event, withdrawal due to participants choice). The authors did not consider the effect of withdrawal from the study on the results of the procedure, nor did they implement any controls to address this issue, which is a significant limitation.

In both procedures, the results were consistent. Of the 11 studies selected for the review, 8 were of high quality and low risk of bias, and 2 were of medium quality and higher risk of bias (Chadwick et al., 2009; Kung-Hua Lee, 2019). All the studies were qualified for further analysis.

General data of all studies are presented in Table 1. In all studies, the sample sizes were small, ranging from 22 to 48 in the experimental groups. Additionally, some procedures included the use of a waiting list and cross-over design (Chadwick et al., 2009; Langer et al., 2012). The average age range was 25 to 54 years; in one study, patients were only eligible if the duration of their illness did not exceed 5 years (Chien & Thompson, 2014).

Inclusion criteria varied regarding the level of cognitive functioning with some requiring basic reading and writing abilities in the language in which intervention was implemented (Lee, 2019; López-Navarro et al., 2015, 2020) or understanding thereof (Chien & Lee, 2013; Lam et al., 2020); others also required a lack of neurological disorders/active substance use disorder (other than nicotine) (Chadwick’s addendum to studies from 2016 — in 2009 one drop-out was due to subject relapsing on heroin). Some exclusion criteria encompassed simultaneous participation in other cognitive or body-awareness therapies (Lam et al., 2020). The recruitment in several instances was partially based on researchers’ subjective assessment (Böge et al., 2021; Chadwick et al., 2009; Langer et al., 2012; López-Navarro et al., 2015, 2020); others relied mainly on objectified measures. The collected studies were published between 2009 and 2021.

In patients with psychosis, standard MBI/MBP procedures (Appendix, Table 6) are modified (Ridler et al., 2024). Chadwick et al. (2005) suggest that single MBIs sessions should last no more than 10 min. Additionally, active narrative exercises are resumed every 30–60 s to prevent patients from excessively engaging with arising symptoms. These symptoms are addressed directly in a normalizing way, without assigning them extra significance compared to other psychological experiences. The homework/own work was considered optional. Some of the presented papers did partially conform to this framework (see Table 2).

The medical staff leading mindfulness groups were psychiatrists, psychologists, psychotherapists, and nurses with experience working with individuals with mental illness. In the paper by Langer et al. (2012), the person conducting the interventions was a Mindfulness-Based Cognitive Therapy (MBCT) therapist with 2-year professional experience. In the case of the Davis et al. (2015) procedure, the leading therapist was a Mindfulness-Based Stress Reduction (MBSR) trainer with 3 years’ experience who was in the process of obtaining a trainer certificate. The co-therapists were doctors of psychological sciences experienced in MBSR training. The research team attended weekly supervision and performed mindfulness practice. In Boge’s work (Böge et al., 2021), the mindfulness sessions were led by a therapist with 10 years’ experience in MBCT, MBSR, Vipassana meditation, and transcendental meditation and also trained in humanistic approach and CBT. The co-therapist was a psychologist trained in CBT. Chien and Thompson (2014) reported the therapists attended a 3-day workshop conducted by researchers and psychotherapists and their work was supervised. López-Navarro and Al-Halabí (2022a, b) participated in 60% of sessions as a supervisor. In Lee’s (2019) procedure, the therapists (clinical psychologists with work experience) attended a 3-day workshop on MBI method. In the procedure by Chadwick et al. (2009), the second author was also a trainer who completed a 3-year MBCT course with work experience with psychotic patients. In the work by Chadwick et al. (2016), the main author supervised the trainers-psychologists experienced in CBT or mindfulness approaches. Lam et al. (2020) described that the groups were led by a mindfulness trainer experienced in working with psychotic patients and a nurse experienced in psychoeducation. A significant limitation of this study is the wide variation in mindfulness teaching competency.

In studies which provided the information, the number of patients in the mindfulness groups ranged from 8 to 12: Davis et al. (2015), 8; Chien and Thompson (2014), 11–12; Lopez-Navarro and Al-Halabí (2022a, b), 8–12. These numbers are lower than those observed in standard MBSR and MBCR programs (up to 20 participants). The presumed limit of participants per group is consistent with the recommendations for patients with psychotic disorders.

The recommendations homework varied from daily to weekly and were either repetitive or linked to the topic of the week. In the majority of articles, no systematic monitoring of own work was performed, i.e., Chien and Lee (2013) and Lee (2019) based their assessments on the feedback from the leading therapists who also relied on the feedback from participants. Chadwick et al. (2009, 2016) described that subjects reported verbally whether they completed the assignment as planned. Two procedures monitored homework regularly: Lam et al. (2020) obtained data noted in the diaries, and they observed 58.3% of participants engaged in own work (31 min a week on average SD= 17.34; 0–86.4 min a week); Davis et al. (2015) reported the implementation of own work was monitored via logs, and participants meditated for 54% of days (27.72 min on average) and mindfully completed everyday activities for 47% of days (27.68 min on average). The only procedure that did not include own work was reported by Böge et al. (2021) which was performed during inward hospitalization. The lack of control over the participants’ mindfulness practice time significantly weakens the ability to draw conclusions regarding the relationship between the amount of mindfulness training prescribed in the procedures and the outcomes achieved. Details of the mindfulness techniques used in a reviewed studies and global outcome are presented in Table 3.

Congruently with the inclusion criteria, studies with control group were taken into consideration. Depending on the study structure, the control group was a waiting list (Chadwick et al., 2009; Langer et al., 2012), and control condition Intensive Support (IS) (Davis et al., 2015), Integrated Rehabilitation Treatment (IRT) (López-Navarro et al., 2015; López-Navarro & Al-Halabí, 2022a, b), Conventional Psychoeducation Programme group (CPEP) (Chien & Thompson, 2014), or a control group of individuals who received treatment as usual (TAU).

In all study assessments, time-points were set prior to and after the completion of intervention, and in some, a follow-up after the end of the intervention: one study after 2 months (Böge et al., 2021), in two studies after 3 months (Chien & Thompson, 2014; Lam et al., 2020; Lee, 2019), and in others after 6 months (Chadwick et al., 2016), 18 months (Chien & Lee, 2013), or 24 months (Chien & Thompson, 2014).

Only some procedures clearly described the minimal attendance criterion which would define the completion of the program. Among the described, the lowest required attendance criterion of 50% was reported by Langer et al. (2012) and Lee (2019). Despite the very low threshold, a high dropout rate of 36% in the experimental group (vs. 8% in the control group) was observed by Langer et al. (2012). In the active control groups, the drop-out levels were similar to those of experimental groups. It suggests that the activity in itself, rather than its particular form, is hard to complete by individuals with schizophrenia. This would be consistent with the general knowledge on this disorder. If the intervention was performed in the treatment or rehabilitation structure context, the attendance levels were significantly higher. In only one study, an additional motivation in the form of remuneration was offered which resulted in high 77% attendance, despite a very engaging program (the predicted level was 70%) (Davis et al., 2015). Böge et al. (2021) reported a high attendance of 95.8% which was noted as long as the sessions constituted a part of general psychiatric ward programs daycare; however, after discharge, the attendance dropped 50% (follow-up). In other studies performed as external part of some psychiatric facility structure such as, i.e., the López-Navarro and Al-Halabí (2022a, b) study subjects participated in 91.58% of sessions (no drop-outs other than in the active control IRC (2 of 26)) or Chien and Lee (2013) study in which the attendance reached 94% (6% of the whole study group dropped-out).

No tools were used to measure the safety of the procedures directly. In general, in most studies adverse events (AE) or serious adverse events (SAE) such as life-threatening AE or fatal occurrences were reported congruently with the World Health Organization (WHO) recommendations. Moreover, some studies reported indirect measures of safety, such as the dropout rate (Ellett & Chadwick, 2021) or incidents of mild psychological discomfort identified in conversations with participants (2 individuals; Lam et al., 2020). All procedures were considered safe. However, the outcomes of individuals who withdrew from the study were not monitored, making it unclear whether they experienced any short- or long-term adverse effects from the intervention or possessed specific characteristics that might have made them less likely to benefit from mindfulness training. This is a potentially important issue, as some studies suggest that practicing mindfulness may exacerbate negative effects, such as emotional dysregulation, anxiety, or depressive symptoms. These effects may be influenced not only by the intensity of practice but also by individual characteristics of the practitioner (Britton, 2019).

Lopez-Navarro observed a low retention rate of 93% (López-Navarro et al., 2015; López-Navarro & Al-Halabí, 2022a, b); drop-out rates were noted as 5% by Böge et al. (2021) and 16.7% by Davis et al. (2015). Chadwick found (Chadwick et al., 2016) that the retention rates reached 86% and 76% after 4-month follow-up, while his earlier studies (Chadwick et al., 2009) reported a high drop-out rate of 40% (this study allowed for the participation of subjects with substance use/neurological disorders). Chien and Thompson (2014) described a 8.3% drop-out rate after 1 year of observation. Kung Hua Lee (2019) observed a 30% drop-out rate; all were considered secondary to external factors (influenza, psychotic episode, occupational training), while in the TAU group no drop-out was reported. Lam et al. (2020) noted a 7% drop-out rate .

In the next section, some statistical values are reported with greater precision than two decimal points to ensure accuracy in interpreting effect sizes and statistical outcomes. This approach aligns with best practices in statistical reporting, ensuring transparency and reproducibility of findings.

The results indicated an improvement in the mindfulness skills (SMQ) after vs. prior to intervention in the following papers: Langer et al. (2012) (p=0.028); Chadwick et al. (2009) (p=0.037); Böge et al. (2021) (p<0.001) and in other measures FMI (p=0.03); CFQ (p=0.01); AAQ (p=0.02); however, no statistically significant changes were observed in Mindful Attention Awareness Scale (MAAS) (López-Navarro et al., 2015).

Furthermore, in the experimental group, depression levels decreased after the intervention compared to before in Chadwick et al. (2016) (HADS, p=0.023), persisting after 6 months of follow-up (HADS, p=0.037); López-Navarro and Al-Halabí (2022a, b) (PANSS-dep; p=0.026); Böge et al. (2021) (DAAS; p=0.02; CDS; p<0.001). Lam et al. (2020) reported a significant decrease in ruminations in the total results (SRRS; p=0.0016), which is in the depression symptomatology range. Lam et al. (2020) compared two groups (experimental vs. control) in three time-points (pre, post, 3 months of follow-up); significant increase in cognitive reappraisal between the groups was observed (ERQ; p=0.033; d=0.47). Kun-Hua Lee (2019) reporting of direct assessment of depression is inconsistent; after 3 months of follow-up, a significant decrease was noted in the TAU vs. experimental group (BDI-II; p=0.007).

An improvement in general psychopathology was noted in the experimental group after the training in Lopez-Navarro research (López-Navarro et al., 2015) (p=0.038). Chien and Lee (2013) reported that group comparisons after 18 months indicated an increase in insight (ITAQ, p<0.001) and reduced symptoms of illness (BPRS, p<0.01). These results persisted after a 2-year follow-up. In other paper by Chien and Thompson (2014), three groups were compared (MBPP, active, and passive control) at four time-points; statistically significant differences were noted in measures of insight (ITAQ, p=0.001) and symptoms of illness (BPRS, p=0.005) favoring the experimental group. Chadwick et al. (2009) observed an improvement of clinical functioning (CORE-OM (p = 0.013)). Kun-Hua Lee (2019) found a statistically significant increase in symptoms of illness in the experimental group vs. control group as assessed right after the training (PANSS total p=0.01, General PANSS p=0.01); given that the presentation of results was vague, they should be interpreted with caution. No significant differences between the groups were observed in major symptomatic dimensions of schizophrenic disorders (Langer et al., 2012).

Positive changes in the symptomatologic subscales in the clinical group following the training were reported by Chadwick only in his later studies (Chadwick et al., 2016). These improvements were observed in subscales measuring the severity of distress related to auditory experiences (PSYRATS; p = 0.021) and disturbances (CHOICE severity; p = 0.031).

A significant reduction of negative symptoms was observed in the clinical group by López-Navarro and Al-Halabí (2022a, b) (PANSS; p<0.001), but not in her previous publication (López-Navarro et al., 2015), and by Böge et al. (2021) in the experimental group vs. TAU (PANSS, p=0.02) as well as in the pre- vs. post-training assessments in the experimental group (PANSS, p<0.001). Kun-Hua Lee (2019) reporting of data is unclear; the results should be interpreted with caution. Two groups were compared in three time-points (pre, post, 3-month follow-up) and a significant difference in the level of negative symptoms was noted, favoring the MBI group in the assessment right after the intervention (SANS; p<0.001; PANSS; p=0.049).

Positive results were reported in the measures related to work persistence in the experimental group vs. TAU (IS) by Davis et al. (2015) in the number of hours worked (p=0.04; d=0.76), improvement in the quality of work (WBI; p=0.021; d=0.88), and auto-presentation (WBI; p=0.036; d=0.79), but there were no statistically significant changes in social skills and cooperativeness subscales. Similarly, there was no improvement in social relationships after the training in the clinical group, as reported by Böge et al. (2021) and López-Navarro et al. (2015) and social support in Chien’s research (Chien & Lee, 2013 and Chien & Thompson, 2014).

In the assessments of well-being and social functioning, improvement was noted in the experimental group after the intervention by Böge et al. (2021) in the total score (PSP, p<0.001) and subscales of physical health (WHO-QoL BREF, p=0.03), psychological functioning (WHO-QoL BREF, p=0.01), and environment (WHO-QoL BREF, p=0.01). Likewise, Chien and Lee (2013) and Chien and Thompson (2014) noted that the group comparisons after 18 and 24 months showed improvement in the general functioning (SLOP; p<0.05 and p=0.004), number of readmissions (p<0.01), and their duration (p<0.001) favoring the experimental group.

Participants reported no significant changes in anxiety across measurement points in the research by Böge et al. (2021). None of the authors reported any adverse changes.

A limitation of the presented results lies in the attendance levels ranging from 50% upwards in the case of interventions carried outside the setting of treatment facilities. On one hand, the presented results might suggest the effectiveness of the intervention, but on the other hand, its small impact on the motivation to maintain training. Introducing an additional motivation system could be beneficial. Davis et al. (2015) noted that extra motivation in the form of remuneration allowed for the achievement of attendance exceeding 90%. Motivational dialogue is one of the methods used to support treatment engagement, identify goals, and reduce ambivalence. An adaptation of this method was created for patients with schizophrenia (Rush et al., 2003) and recommended for use together with other therapeutic techniques aimed at the development of specific skills (Garety et al., 2000). Previous clinical studies on the motivational dialogue in individuals with schizophrenia showed an increase in adherence to ambulatory treatment after discharge from the hospital (Swanson et al., 1999). The inclusion of motivational dialogue into the mindfulness procedures research could positively influence the levels of adherence and outcomes.

These conclusions are consistent with previous data — structured MBI procedures allow for easy applicability with well-proven effects in other disorders (Baer, 2003). The available results confirm the possibility of adapting MBIs/MBPs — in terms of both time and meeting frequency, to match them to the institution performing the intervention (Hodann-Caudevilla et al., 2020). However, there is no standardized intervention protocol that would facilitate conclusions about the achieved results and there is no clear information on when therapy is most beneficial. Researchers assess different goals and present heterogeneous results. The studies were conducted on a limited sample size. Replication in a larger group would increase the reliability of the results.

Further research should primarily focus on structuring interventions and homework performance monitoring. In the analyzed studies, the general lack of systematic and formal monitoring (9/11 studies) of performing homework tasks hampers the identification of their impact on patients’ recovery or consolidating mindfulness skills. Literature shows that there is a link between performing homework and improvement in depressive symptoms as well as reducing the risk of depression recurrence; perhaps, a similar effect could be observed in people with schizophrenia (Chadwick et al., 2016). In a recent systematic review (Jacobsen et al., 2022), only 12% of studies provided data on the extent of home practice, and none explored the relationship between home practice and clinical outcomes. Mindfulness practice influences outcomes, largely due to the dose-response relationship (Britton, 2019), but there is currently no established unit of measurement that adequately encompasses practice time, frequency, or quality (Segal et al., 2019). This area requires further investigation.

Further research is needed to assess the safety of these procedures. To better evaluate the safety of these interventions, recommendations (Ellett & Chadwick, 2021) have been developed, emphasizing the presentation of results on the harmfulness of studies focused on eight aspects of the process: RCT study protocol; SAEs, hospitalization and rehospitalization rates; AEs; side effect rate; symptom severity increase (calculated per number of subjects); patient involvement in the process of assessing/operationalizing the harmfulness of research. These postulates were not fully implemented in the studies presented in this review, participants were not systematically included in the evaluation of the program, hospitalization and rehospitalization rates were not controlled (except (Chien & Lee, 2013; Chien & Thompson, 2014), and side effect rates were not recorded. Most studies emphasized the lack of relationship between the intervention itself and withdrawal and considered MBIs/BMR safe. However, the outcomes of individuals who withdrew from the study were not monitored, and it remains unclear whether they experienced any adverse effects from the intervention.

The analyzed results indicate the possibility of using MBIs/MBRs as support in achieving various therapeutic goals, maintaining work, its quality, self-presentation in the workplace (Davis et al., 2015), well-being, psychosocial functioning (Böge et al., 2021; Chien & Lee, 2013; Chien & Thompson, 2014), and emotional regulation. These are consistent with the assumption that when the symptoms of the disease become less troublesome, patients gain more resources to develop in other areas of life.

Mindfulness practice encompasses two essential experiential components: the practice itself and the immediate reflections on the experiences that arise during it (Ellett, 2024). Therefore, recognizing the need for a standardized and appropriate level of qualifications for mindfulness trainers is critical to ensuring consistent quality in the transfer of knowledge and skills to participants. The reviewed studies indicate a lack of uniformity in this regard, as each offers differing definitions of what constitutes adequate trainer preparation. A prudent approach would involve employing a team of certified mindfulness trainers with relevant experience working with individuals diagnosed with schizophrenia. Additionally, trainer competence is crucial when working with groups. However, preparation for training and monitoring of work quality varied across studies. Future research focused on the relationship between training effectiveness and the level of trainer competence, or a comparison between self-practice and trainer-led practice, could clarify the role of trainers, especially in the context of individuals with schizophrenia practicing mindfulness.

There is also an urgent need to acquire more information regarding the impact of various elements of mindfulness protocols, including how these elements interact and influence clinical outcomes. This knowledge could aid in the standardization of treatment protocols (Ellett, 2024; Britton, 2019).

The Kun-Hua Lee procedure (Lee, 2019) differs significantly from the other studies discussed; due to the lack of researcher blinding, it is a study with a high risk of bias with unclear presentation of the results. Developing interventions in this manner requires further research. Further analysis requires the study of healing factors and a thorough examination of the mechanism of action of mindfulness in schizophrenia spectrum disorders. The Lam protocol seems promising — it contains the experiences of other researchers mentioned in this review and examines the mechanisms of change at the level of cognitive changes. Its supplementation with results at the metacognitive level, i.e., at the level of intervention, would facilitate gaining insight into the mechanisms of mindfulness and this seems to be particularly important in the development of mindfulness research.

Due to the heterogeneity of the included studies (different inclusion criteria, various outcome measures, different therapeutic methods applied), we were not able to conduct quantitative synthesis of the data and perform meta-analysis. Other limitations are a wide variety of mindfulness protocols, a variety of teaching competency among mindfulness trainers, and small sample sizes.

This review indicates that when designing MBIs, it is essential to consider the higher effectiveness of group interventions, its integration into existing treatment and support structures, its long-term and cumulative effects over time, its positive impact on negative and depressive symptoms, its enhancement of quality of life, and its capacity to adapt programs to various therapeutic needs.