Introduction

Several systematic reviews have demonstrated that social isolation (the lack of social contacts and having few people to interact with regularly) is associated with depressive symptoms [1, 2], which, in turn, are correlated with unhealthy behaviors and reduced access to material resources [3]. Loneliness (the distressing feeling of being alone or separated) can lead to mental illness such as depression, alcohol abuse, sleep problems, Alzheimer’s disease, and to physical disorders like diabetes, autoimmune disorders and cardiovascular diseases, physiological aging, cancer, poor hearing and poor health [4]. This is even more problematic for individuals with addictive disorders, which are at greater risk of being isolated and feeling lonely than healthy individuals. The literature has shown significant positive associations between loneliness and diverse type of addictions, such as alcohol [5], internet [6], Facebook [7], smartphone [8, 9], gambling [10] and food [11,12,13]. Without tackling social isolation and loneliness of individuals with addictive disorders, the vicious circle of addiction, loneliness, mental and physical illness may not end. In the case of drug addicts, Atadokht et al. [14] demonstrated a significant negative correlation between perceived social support and the frequency of relapse (r = − 0.34, P = 0.001).

One solution arises from self-help groups, that is, a supportive, educational, generally change-oriented group that addresses a specific life problem shared by all its member [15]. Involvement in a self-help group has proved efficient in reducing loneliness, social isolation [16, 17] and stress [18], as well as in improving meaning in life [19, 20], hope and health-promoting behaviors [21]. It also facilitates abstinence maintenance and symptoms reduction [22,23,24].

The more widely spread kind of self-help groups are twelve-step mutual help (TSMH) groups. TSMH are available for Alcoholic use disorder (alcoholic Anonymous; AA), substance use disorder (Narcotics Anonymous; NA), pathological gambling (Gamblers Anonymous; GA), eating disorder (Overeaters Anonymous; OA), dual diagnosis (Double Trouble in Recovery; DTR), compulsive sexual behavior disorder (CSBD; e.g. Sexaholics Anonymous; SA) and several other addictive behaviors (see Appendix A for a list of 28 TSMH groups). The 12-steps underlying the recovery culture of AA, NA, GA, OA, SA and DTR are described in Appendix B. Most of TSMH groups are born in the USA, and nowadays North America represents 75%, 68% and 57% of global face-to-face meetings in NA, OA and GA, respectively (see Appendix C for the breakdown of face-to-face NA, OA and GA meetings by geographical area).

Among all TSMH groups, AA is the first in history, the most widely spread worldwide, and the most studied by the literature (see Appendix D for detailed information on AA, NA, OA, GA, SA and DTR). Evidences attesting the efficacy and cost-efficiency of AA are robust. A meta-analysis of 27 studies, containing a total of 10,565 participants, demonstrated that participation in AA/TSMH for alcohol use disorder performed at least as well as established active comparison treatments (e.g. CBT) on all outcomes except for abstinence where it often outperformed other treatments [25]. Humphreys and Moos [26] found that patients treated in cognitive behavioral treatment (CBT) programs had 64% higher annual healthcare costs (p = 0.001), compared to patients in AA/TSMH programs. Psychiatric and substance use outcomes were comparable across treatments, except that AA/TSMH participants had higher abstinence rates (45.7% AA/TSMH versus 36.2% in CBT; P = 0.001). Mundt et al. [27] found that each additional AA meeting attended was associated with an incremental medical cost reduction of 4.7% during 7-year follow-up. Recovery in AA may stem from the ability of the group processes to augment self-efficacy, coping skills, and motivation, and by helping people build supportive and pro-social networks [28].

The large literature on AA is yet to be compared to the scarcer literature on other TSMH groups. In particular, no systematic review and meta-analysis have been performed for TSMH group other than for alcohol use disorder. The aim of this article is to fill-up this gap by providing a systematic review of TSMH groups other than AA, and whatever the addiction targeted by the group is, hence contributing to the growth of dimensional psychiatry. Both qualitative and quantitative studies were included in the review. The four meta-analyses performed were based on those studies which allowed evaluating Pearson correlation between (i) Duration or involvement in TSMH groups; and (ii) Severity of symptoms or quality of life. Since all addictions follow a similar pattern [29], we expected results similar to those demonstrated by Kelly et al. [25] regarding AA/TSMH for alcohol use disorder. That is, significant and negative (respectively, positive) association between higher duration or involvement in TSMH and lower severity of symptoms (respectively, higher quality of life).

Methods

A systematic review and meta-analysis was conducted following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidance [30]. Sub-sections show, respectively, search strategy and selection criteria (2.1), data extraction (2.2), quality assessment (2.3) and statistical analysis (2.4).

Search strategy and selection criteria

Figure 1 details the flow of information through the different phases of the review. In the identification stage, two databases (MEDLINE, PsychInfo) and a register (ClinicalTrials) from inception through November 01 2022 (PROSPERO CRD42022342605), with no restrictions on language, were used to obtain peer-reviewed articles that would allow for an analysis of TSMH groups for disorders other than AA. We searched for (i) “12-step group” OR “12-step program” OR “12-step facilitation” OR “12-step approach” NOT alcohol; and (ii) all the “anonymous” TSMH groups listed in Appendix A. Both qualitative and quantitative studies were included in the review. Figure 1 shows the 8 exclusion criteria which, among other reasons, led to exclude studies that (i) mentioned TSMH group only in the introduction or discussion sections; (ii) Did not allow evaluating the efficacy of TSMH groups; (iii) were not based on face-to-face meetings; and (iv) involved the direct participation of a clinician in the meetings.

Fig. 1
figure 1

Flow of information through the different phases of the review.2.2. Data extraction

Full size image

Data extraction

In the identification stage, the lead author independently scanned the abstract, title or both of every record to determine which studies should be considered for inclusion. Doubts were discussed with other authors. The review team included researchers with specialist background in mental health, psychology, public health, epidemiology, qualitative research, mood and emotional disorders.

In the screening stage, full-text articles were independently evaluated for inclusion by the lead author. Microsoft Excel was used to screen, remove duplicate entries, and record reviewers’ decisions. On completion of database searching, additional records were identified through checking reference lists of each article collected after the screening stage. A data abstraction table (DAT) was designed and piloted. The lead author formed an initial conceptual framework presenting a preliminary synthesis of findings of included studies, the DAT was reviewed by the authors, and refined accordingly. The DAT included information about the name of the main TSMH group studied, author, country of study, participant demographics (age, gender, country, main disease, comorbidities, type and frequency of attendance to TSMH groups), study design, study measures and main results.

The data needed to perform meta-analyses was extracted by the lead author. The corresponding authors were contacted by email when data of relevance for the meta-analysis were insufficient in the published article. If we got no answer at first email, corresponding authors were contacted 2 more times.

Quality assessment

Two distinctive tools were applied to assess the risk of bias of quantitative studies (2.3.1) and qualitative studies (2.3.2).

Quantitative studies

Study quality of quantitative studies was examined adapting the tool used by Strahler et al. [31], a modified set of the quality criteria for primary research, as proposed in the Evidence Analysis Manual of the Academy of Nutrition and Dietetics [32]. Our final scoring sheet (see Appendix E) included 14 criteria (e.g. sampling adequately described and free from bias, clearly defined outcomes, reliability estimates of measures given, appropriate statistical procedures) with each criterion rated as “positive” if present (= 2 points), “neutral” if the presence is ambiguous or when research is not exceptionally strong on this issue (= 1 point), or “negative” if not present (= 0 point). We computed the quality score as the mean of responses across criteria that could be evaluated. Hence, scores could range between 0 and 2. For most articles, coding was independently performed by the first author. For the 9 studies used in meta-analyses, coding was discussed with the second author until 100% agreement was achieved.

Qualitative studies

Cochrane Qualitative and Implementation Methods Group recommendations [33] are to use a tool that takes the multi-dimensional nature of qualitative evidence into account. Guided by this perspective, the quality of included studies and risk of bias was assessed using the Critical Appraisals Skills Programme [34]. This tool is the most frequently recommended tool for qualitative studies [35]. The CASP tool focuses on three domains: study design, validity of results, and generalizability. Each domain is assessed using a set of questions. Based on the response to these questions the studies were marked as low, medium, or high quality. Studies which provided satisfactory information in all domains were marked as high quality, with missing or unsatisfactory information in one domain as medium quality, and with missing or unsatisfactory information in two or more domains as low quality.

Statistical analysis

The final effect size analyzed was Pearson correlation r. The four meta-analyses performed were based on the studies which allowed evaluating Pearson correlation between (i) duration or involvement in TSMH groups; and (ii) severity of symptoms or quality of life. No previously published protocol nor pre-registration exists for these meta-analyses. All included studies collected an informed consent; therefore, an ethical approval was not obtained. However, these meta-analyses complied with the most recent version of the Declaration of Helsinki.

Statistical analysis was conducted with the R Software and following the guidelines of Harrer et al. [36]. As we anticipated considerable between-study heterogeneity, a random-effects model was used to pool effect sizes. Inverse variance weighting was applied to increase estimates’ efficiency and give studies that have greater precision more weight. We used Knapp–Hartung adjustments [37] to calculate the confidence interval around the pooled effects.

To assess the heterogeneity and consistency of the estimated correlations, I2 is used along its confidence interval (CI). I2 indicate the percentage of variation caused by heterogeneity. 95% prediction intervals (PI) are also given for each pooled effect. Prediction intervals give a range into which we can expect the effects of future studies to fall based on present evidence. PI are calculated using the heterogeneity variance τ2. The restricted maximum likelihood estimator [38] was used to calculate the heterogeneity variance τ2. For each meta-analysis, I2 was computed when excluding one study (for all possible configurations; leave-one-out analysis), and the case with the lowest I2.When the total number of studies is small, detecting small-study effects is difficult. Egger's regression test is typically not advisable below 10 studies [39], as well as other methods to detect small-study effects/publication bias. In our case, the least worst method was P-curve analysis, which can be performed when between-study heterogeneity is low [36], and for meta-analyses of more than 3 studies.

Results

Sub-sections show, respectively, systematic review and meta-analysis of quantitative studies (3.1) and systematic review of qualitative studies (3.2). Articles using mixed-methods are discussed either in subSect. "Systematic review and meta-analysis of quantitative studies" or subSect. "Systematic review of qualitative studies", where thought to be most relevant. Figure 2 shows the breakdown of the included papers by type of TSMH group. There are a total of 55 included papers, corresponding to 47 distinctive studies.

Fig. 2
figure 2

Breakdown of the 55 included articles by type of TSMH group

Full size image

Systematic review and meta-analysis of quantitative studies

Sub-Sect. "Systematic review and meta-analysis of quantitative studies" depicts the results regarding the systematic review of quantitative studies (3.1.1) and meta-analysis (3.1.2).

Systematic review of quantitative studies

Twenty-three quantitative-only articles (7 cross-sectional studies, 7 cohort studies, 6 randomized control trials, 2 comparative effectiveness research, 1 non-randomized study with a separate control group) and 3 mixed-methods papers are presented in Table 1. All those 26 papers were rated to be of at least moderate quality; scores ranged from 1.25 to 1.83, with a mean of 1.56 (SD 5 0.17). These 26 papers correspond to a total of 22 distinctive studies. Most studies were conducted in North America (n = 15). The mean age is 43.6. Due to research protocol (see 2.1), all these studies aimed to test the impact of duration in TSMH or TSMH involvement on outcomes (e.g. severity of symptoms, quality of life, self-efficacy, self-esteem). They all showed a significant, therapeutical impact on at least one outcome.

Table 1 Overview of quantitative studies characteristics
Full size table

Meta-analysis

Studies measuring any severity of symptom measures in relation to duration in TSMH.

We computed the pooled correlation as effect size across the 8 studies examining any severity of symptom measure in relation to duration in TSMH (n = 1,209). The measures used in each study are given in Appendix F, Table 10. The participants’ average age was Mpooled = 40.5 years (SDpooled = 11.16). 81% were male.

The results presented in Table 2, 3 revealed a significant pooled correlation of r = − 0.20 (p < 0.01). The effect size across all studies is shown in Fig. 3. The between-study heterogeneity variance was estimated at τ2 = 0.0081 (95%CI 0.0000–0.0918), with an I2 value of 51% (95% CI 0–78%). The prediction interval ranged from g = − 0.42 to 0.04, indicating that positive correlation cannot be ruled out for future studies. When removing the study with the largest influence on I2, [58], n = 1,227, I2 is 22% (95% CI 0–65%), and prediction interval ranged from g = -0.29 to − 0.14.

Table 2 Results of the P-curve analysis concerning correlation between severity of symptoms and duration in TSMH. Omitting Wnuk and Charzyńska (2022) [58]
Full size table
Table 3 Main results regarding the 4 pooled effect size calculated
Full size table
Fig. 3
figure 3

Forest plot of all studies examining correlation between any severity of symptom measure and duration in TSMH

Full size image
Risk of bias across studies

The results of the p-curve analysis are reported in Table 2. When removing [58], I2 = 22% < 50%. Overall, these results indicate the presence of evidential value and that there is a true non-zero effect. We can still not rule out that publication bias has affected the results of our meta-analysis. But, based on p-curve’s results, we can conclude that the pooled effect found is not totally spurious.

Studies measuring any Severity of symptom measures in relation to TSMH involvement.

We computed the pooled correlation as effect size across the 5 studies examining any severity of symptom measure in relation to TSMH involvement (n = 1019). The measures used in each study are given in Appendix F, Table 11. The participants’ average age was Mpooled = 38.8 years (SDpooled = 11.3). 80% were male.

The results presented in Table 3 revealed a non-significant correlation of r = – 0.21 (p = 0.063). The effect size across all studies is shown in Fig. 4. The between-study heterogeneity variance was estimated at τ2 = 0.0237 (95% CI 0.0033; 0.2894), with an I2 value of 76% (95% CI 42–90%). The prediction interval ranged from g = − 0.64 to 0.33, indicating that positive correlation cannot be ruled out for future studies. When removing the study with the largest influence on I2, [45], the pooled correlation is even less significant (see Table 3). However, when removing [58], a significative correlation is found, of r = − 0.25 (p = 0.045). N = 757, I2 is 75% (95% CI 30–91%), and prediction interval ranged from g = − 0.73 to 0.39. No p-curve analysis were led here as I2 is high [36].

Fig. 4
figure 4

Forest plot of all studies examining correlation between any severity of symptom measure and TSMH involvement

Full size image
Studies measuring any Quality of life measures in relation to Duration in TSMH.

We computed the pooled correlation as effect size across the 3 studies examining any quality of life measure in relation to duration in TSMH (n = 259). The measures used in each study are given in Appendix F (Table 12). The participants’ average age was Mpooled = 37.4 years (SDpooled = 9.8). 80% were male.

The results presented in Table 3 revealed a correlation approaching significance of r = 0.10 (p = 0.052). The effect size across all studies is shown in Fig. 5. The between-study heterogeneity variance was estimated at τ2 = 0 (95% CI 0.0000; 0.0538), with an I2 value of 0% (95% CI 0–90%). The prediction interval ranged from g = − 0.61 to 0.72, indicating that positive correlation cannot be ruled out for future studies. No influential case was removed as there were only 3 studies pooled.

Fig. 5
figure 5

Forest plot of all studies examining correlation between any quality of life measure and TSMH involvement

Full size image
Studies measuring any Quality of life measures in relation to TSMH involvement.

We computed the pooled correlation as effect size across the 3 studies examining any quality of life measure in relation to TSMH involvement (n = 478). The measures used in each study are given in Appendix F (Table 13). The participants’ average age was Mpooled 36.5 years (SDpooled 5 8.7). 80% were male.

The results presented in Table 3 revealed a non-significative correlation of r = 0.28 (p = 0.09). The effect size across all studies is shown in Fig. 6. The between-study heterogeneity variance was estimated at τ2 = 0.0187 (95% CI 0.0000; 1.0119), with an I2 value of 70% (95%CI: 0–91%). The prediction interval ranged from g =  − 0.95 to 0.98, indicating that positive correlation cannot be ruled out for future studies. No influential case was removed as there were only 3 studies pooled.

Fig. 6
figure 6

Forest plot of studies examining correlation between any quality of life measure and TSMH involvement

Full size image
Summary of findings of meta-analyses.

Table 3 summarizes the results that were obtained when pooling Pearson correlation r.

Systematic review of qualitative studies

Twenty-eight qualitative papers and 1 mixed study are presented in Table 4. Of the 28 articles, 12 (41%) were rated high quality, 11 (38%) medium and six (21%) low quality. These articles correspond to a total of 24 distinctive studies. Interviews were used by 14 studies (58%), participant observation by 7 (29%), focus groups by 5 (21%), survey-only by 2 (8%) and case report by 2 (8%). Studies were conducted in North America (n = 17), Middle East (n = 4) and Europe (n = 4). 20 of the 24 studies brought evidence on the factors influencing recovery, highlighting the following: Social (n = 15), emotional (n = 9), spiritual (n = 7), self-identification or psychological (n = 6).

Table 4 Overview of qualitative studies characteristics
Full size table

Discussion

Primary outcome was to strengthen knowledge on TSMH groups other than AA, which can be used by support services to inform the development of future research, policy, and practice within healthcare and other settings. Subsection 4.1 provides a summary of the knowledge gathered about the 5 types of TSMH group addressed in this article, each focusing on a specific addiction. The possibility to develop TSMH groups targeting ontological addiction, at the root of all others addictions, is then discussed in SubSect. “TSMH group for the root of all addictions: Ontological addiction?” . Sect. “Discussion” ends with the limitations of this paper.

Summary of the knowledge gathered regarding the five types of TSMH groups studied

Narcotics Anonymous. Over the 55 included papers, 22% deal with NA. However, this systematic review excludes papers which do not allow to distinguishing NA from AA members. The search in Pubmed of “Narcotics Anonymous” AND “Alcoholics Anonymous” gives 83 results (on the 22/12/2022), suggesting that many papers are dealing with both NA and AA without distinction (e.g. Andraka-Christou et al. [94]). Research on the effectiveness of NA show a robust level of evidence: One RCT [44], 2 cross-sectional studies [42, 45], one non-randomized study with a separate control group [41], one cohort study [40] and six qualitative studies [65,66,67,68,69,70]. These studies are generally supportive of NA’s effectiveness. Vederhus and Birkland [70] yet highlights that NA model do not fit all. Besides, NA members may stigmatized patients using buprenorphine [43] and methadone ([43, 95]) for treatment purposes.

Overeaters anonymous. OA is overlooked by the TSMH literature. Over the 55 included papers, there are 1.7 times more GA articles than OA articles. Worldwide, there are 2.4 times more OA than GA meetings (see Appendix D). OA is particularly underrepresented in quantitative studies (12% of total, versus 24% in qualitative studies). Research on the effectiveness of OA is limited to lower levels of evidence: a 2002 dissertation [48], one study comparing effectiveness of OA to weight watchers and multiple sclerosis mutual help groups ([46, 47]), two case reports ([71, 76]) and five qualitative studies ([72,73,74,75, 77, 96]). These studies generally support OA’s effectiveness, and highlight the need for higher quality research, including randomized controlled trials. Future research on OA should rely upon the good quality literature review of Bray et al. [97].

Gamblers anonymous. GA is over studied by the TSMH literature, when considering the number of GA meetings relative to others TSMH (see Appendix D). Over the 55 included papers, 31% deal with GA. Research on the effectiveness of GA show a robust level of evidence: two RCT (which led to 3 articles [51, 52, 55]:), four cross-sectional studies ([49, 50, 53, 54]) and seven qualitative studies (which led to 10 articles [49, 78, 79, 81,82,83,84,85,86,87]:). These studies are generally supportive of GA’s effectiveness. Schuler et al. [98] mitigate these results by indicating that larger RCT are needed to prove the effectiveness of GA either as a control condition or in conjunction with formal treatment or medication. Specificities of GA relative to other TSMH groups include (i) a focus on steps 4 and 9 (see Appendix B for the 12-steps); (ii) devoting much time and energy to counselling members on financial and special challenges; (iii) making direct comments during meetings (relative to others TSMH groups in which members avoid addressing one another directly); (iv) absolute assertion of identity as a ‘‘compulsive gambler’’. Future research on GA should rely upon the good quality scoping review of Schuler et al. [98].

TSMH group for compulsive sexual behavior disorder. Over the 55 included papers, 9% are dealing with a TSMH group for CSBD. Research on the effectiveness of those TSMH groups shows medium level of evidence, with two cross-sectional studies [57, 58], one cohort study [56] and one qualitative research [88]. These studies are generally supportive of TSMH’s effectiveness, by e.g. lowering the sexually related sense of helplessness and narrowing the repression away from sexual thoughts [57]. The tree studies hereafter worth to be noticed, even though not being included in this paper as they do not deal specifically with TSMH experience but rather focus on the recovery from compulsive sexual behaviors. Dhuffar-Pottiwal and Griffiths [99] analyzed recovery experiences of three Sex and Love Addicts Anonymous members from the UK, restricting their sample to female-only participants. Yamamoto [100] analyzed recovery experiences of four heterosexual men from unspecified sex groups. Antons et al. [101] performed a systematic review identifying 24 treatment studies on CSBD and problematic pornography use as well as treatment effects on symptom severity and behavior enactment.

More research (including RCT) on TSMH for CSBD is needed, as CSBD seem to be increasing in several countries ([57, 58, 89, 102]). In 2019, the World Health Organization (WHO) included the diagnosis of CSBD as an impulse control disorder in the eleventh revision of the International Classification of Diseases (ICD-11; WHO, 2019). In Iran, the number of SA meeting per week grew from 557 in 2016 to 1,246 in 2018 [102]. Future research could consider using the recently developed CSBD Scale (CSBD-19 [104];) that assesses CSBD based on ICD-11 diagnostic guideline. Practitioners may want to activate TSMH group meetings for CSBD in prisons, especially for prisoners of sexual crimes [102].

Double trouble in recovery. DTR is over represented in the TSMH literature. 18% of the 55 included papers deal with DTR. DTR however, has the fewest number of TSMH groups among those studied in this paper. In 2008, there were about 200 DTR groups in the USA across 14 states [61]. The authors are not aware of any DTR groups outside the US. PubMed. Thirteen papers on DTR were published between 2002 and 2015, among which 5 are based on the same study/sample ([21, 59, 60, 62, 105]), and authors are all connected through common institutions, universities or states. There are no published papers since 2015. The current activity level of DTR groups is unclear.

Studies on the effectiveness of DTR show a robust level of evidence: two RCT ([63, 64]), one large (n = 310) cohort study that led to four articles ([21, 59,60,61]), one other cohort study [62], one mixed-method [90] and two qualitative papers ([91, 92]). These studies are generally supportive of DTR’s effectiveness, by offering a place where persons with a dual diagnosis can support, share and educate one another on their comorbidity without fear of stigma [92]. Yet, even though DTR is not widely available, this should not necessarily inhibit healthcare professionals to encourage dual diagnosis patients to attend other TSMH groups, being aware that, “one size does not fit all” [70]. It is important however, that such TSMH groups will be receptive to dual diagnosed patients and will not stigmatize or in any way discriminate against them. As long as this principle is held, the type of TSMH group attended is of secondary importance on the positive therapeutic impact of TSMH ([70, 92, 106]). Following Rosenblum et al. [63], healthcare professionals offering an intensive referral and/or motivation enhancement component might facilitate better attendance at TSMH meetings.

TSMH group for the root of all addictions: ontological addiction?

We advocate that coupling TSMH group with 3rd wave CBT [109], which are specifically designed to cure ontological addiction, is a promising avenue for more efficient, transnosographic TSMH groups. According to ontological addiction theory [107], the root of the suffering of mental unsatisfaction is an addiction to try satisfying an incorrect self-concept. The incorrect self-concept (i) is perceived as intrinsically separated from its surroundings, and from experiences of well-being, safety and worth; and (ii) leads individuals to search for external sources to fill up this perceived lack. The 6 components inherent in any form of addiction [29] can be found in ontological addiction [108]. Significant correlations were found between ontological addiction and depressive symptoms (r = 0.537), anxiety symptoms (r = 0.565) and self-esteem (r = 0.426 [108];). Ontological addiction is at the root of the suffering of mental unsatisfaction, and thus at the root of the suffering experienced by subjects with psychiatric disorders and addictions. Only the form taken by the incorrect self-concept endorsing ontological addiction changes with the addiction.

By opposition to an incorrect self-concept, a correct self-concept conceives of itself as inseparable from others, from its surroundings and from its own experiences of well-being, safety and worth. Curing one’s ontological addiction means to progressively change the self-concept from incorrect to correct. TSMH groups can be advantageous to that regards by providing a space in which (i) People can feel as a part of whole (feeling of connectedness with the group members); (ii) People can progressively start identifying themselves to a helper by providing emotional support and advises. Through the process of changing one’s self-concept from incorrect to correct, the perception one’s have of characteristics inherent to the psychiatric disorder goes from weakness to strength. If designed properly, TSMH groups can help individuals to use these characteristics both to serve's oneself and others instead of using them for self-destruction. Future research should evaluate the efficacy of TSMH groups coupled to 3rd wave CBT to reduce ontological addiction as well as more traditional kind of addictions.

Future TSMH groups could also get inspiration from the neurodiversity’s concept [110]. This concept has helped to favor a positive view of individuals with e.g. autism, attention deficit hyperactivity disorder, dyslexia, depression, anxiety, intellectual disabilities, schizophrenia [111], or stuttering [112]. In the case of individuals with autism, the ability to hyperfocus, attention to detail, good memory, and creativity, as well as honesty, loyalty, and empathy for animals or for other autistic people can be conductive to success for the whole work team, if they are appropriately balanced by consideration of relevant individual weaknesses ([113, 114]). As for individuals with borderline personality disorder, altruism, self-derision, creativity, enthusiasm and probity can be a great asset in any group, provided that the disorder is treated correctly [115].

Limitations

In addition of the limitations discussed previously in Sect. “Discussion”, the main limits of this study must be emphasized. They are caused by:

  • The method applied (see Sect. “Methods”). The eligibility criteria led to exclude papers focusing on (i) Mutual help groups which are not 12 step based, such as SMART ([116, 117]), Seeking Safety ([118, 119]) or Compassionate Friends [120]; (ii) Online or remote TSMH group meetings, which hold a large potential for improving healthcare continuity and access [121, 122] even though they may be less effective than face-to-face meetings in fostering solidarity and sense of belonging [128]. Besides, additional databases such as Web of Science or Science Direct could have been explored. Efforts were made to limit this bias by systematically checking the references of papers.

  • The meta-analysis objective (see Sect. “Quantitative studies”). The pooled effect is correlational, which precludes the ability for causal conclusions. In other words, it is unclear whether e.g. lower severity of symptoms is an outcome of duration in TSMH. Another possibility is that subjects with lower severity of symptoms fell less stigmatized and hence are more keen to seek help and attend TSMH meetings. One RCT and one cross-sectional study showed a causal relationship, from higher attendance to lower severity of symptoms ([53, 64]). More longitudinal studies are yet needed to bidirectional associations over time between TSMH participants.

  • The pooled samples. Among the 4 meta-analysis performed, pooled mean age ranges from 36.5 to 40.5, and 90–81% of pooled subjects were male. Many of the pooled subjects come from countries with a moderate to strong religious history. No study were found on TSMH of people living in Central and South America, while 13% of face-to-face GA, NA and OA meetings are hold there (see Fig. 7).

Conclusions

TSMH attendance and involvement were negatively correlated with severity of symptoms (high and medium levels of evidence) and positively correlated with quality of life (low levels of evidence). It is important to acknowledge that all TSMH have limitations and have been subject to criticism. The question is whether the benefits outweigh the potential risks. Our primary argument is that the answer to this question will depend on proper assessment and classification of the individual, as well as the type of TSMH engaged, and what other treatment services are being received. To reduce potential risks and improve efficacy, we advocate that TSMH group targeting ontological addiction, coupled with third wave CBT, is a promising direction.