Strengths and limitations
To our knowledge, this is the first review to focus solely on evidence for the effectiveness of group peer support interventions, delivered only by people with lived experience of mental health conditions. This reduced heterogeneity in methods of intervention delivery and statistical heterogeneity was low for the meta-analyses, suggesting relative consistency in intervention effects across studies [
67]. There was a distinction in focus between interventions that aimed to reduce self-stigma [
10] and those that aimed to improve self-management. Effectiveness for improving recovery may differ between intervention subtypes, however only one included anti-stigma intervention reported recovery [
62], so it was not possible to analyse these separately. Variation in participant characteristics was a source of clinical heterogeneitybetween studies [
68]. Main analyses of all outcomes except empowerment, however included only participants experiencing mental health conditions that were defined as severe, providing a specific evaluation of intervention effectiveness for these outcomes for people with these experiences. Full appraisal of the effectiveness of group peer support for people with other mental health conditions was not possible due to current limitations of the evidence-base.
Since the focus of this review was intervention effectiveness, we included only RCTs to enhance the potential for causal inference and reduce the influence of bias on the findings [
69]. Conversely, this may have limited the studies returned by the search and therefore, the scope of the meta-analyses. We also excluded cluster RCTs since we characterized group peer support as a discrete intervention, which can be randomised at the individual level. However, many mutual support and peer support programs have arisen out of user-led organizations [
70], which might more parsimoniously function as the unit of randomisation. Of our 12 methodological exclusions, only one of these was due to the study being a cluster RCT [
71]. However, the study did not meet other inclusion criteria, for example, the intervention included both group and one-to-one components [
72]. Therefore, although it is unlikely that our exclusion of cluster RCTs has altered the findings of this review, future reviews of group peer support may wish to include this study design within inclusion criteria in order to minimise the risk of missing relevant evidence.
We adopted strict and limited eligibility criteria for this review in order to present a comparable group of interventions for which group peer support was the active ingredient, and to enable valid comparisons of intervention effects. However, this approach may have led to relevant evidence being missed, which could provide interesting and important contributions to our current knowledge of group peer support interventions. For example, we excluded all interventions with any one-to-one support elements. This may have led to the exclusion of potentially helpful programs, which blended group and one-to-one approaches. Combined one-to-one and group peer support programs may be particularly beneficial for flexibly accommodating the diverse needs of people using peer support interventions and require evaluation and synthesis in future reviews.
Our adoption of strict eligibility criteria for the review attempted to address the heterogeneity peer support interventions, through focusing on the effectiveness of one narrowly defined sub-type. However, this may limit the generalisability of these findings to other peer support interventions. Only one included study met our definition of mutual support, which was delivered online, so findings may not be generalisable to face-to-face groups due to distinctive barriers to peer support utilisation delivered via technology [
73]. Therefore, the review findings are specifically generalisable to structured peer support groups. Studies were predominantly conducted in America, which may further limit the generalisability of the findings identified here. We also adopted a strict definition of peer support to exclude all health professional involvement. However, some group peer support interventions are often co-delivered with health professionals and maintain a non-diagnostic, recovery-orientated ethos, such as peer support groups provided internationally by the Hearing Voices Network [
74]. These groups may have many benefits for recovery and require independent evaluation. Similarly, we excluded all groups with any focus other than promoting recovery with mental health conditions. This was to enable us to report any impact on recovery outcomes as direct effect of the interventions, rather than as possible secondary benefits experienced through addressing other issues, such as bereavement or physical health conditions. Peer-led and delivered group interventions targeting experiences commonly experienced by people who experience mental health conditions may also have benefits for recovery. These require independent syntheses and may further contribute to the evidence-base for the effectiveness of group peer support interventions.
Of the 4277 papers returned by our search, only 11 met our eligibility criteria for inclusion, reporting findings of eight trials. However, we used intentionally broad search terms in order to collect a large number of papers and to ensure that no potentially eligible studies were missed (see Supplementary Material, Additional file
1 for full search strategy). A large number of papers were also excluded at the full text screening stage. We were conservative about retrieving full text studies and retained all papers with any evidence of relevancy for detailed consideration. There were some studies that proved problematic for eligibility decisions, included in the supplementary material (Additional file
1). If there was any doubt that a study met eligibility criteria it was excluded, in accordance with recommended procedures for systematic reviews [
32].
A methodological limitation of this review was the omission of terms related to “consumer” within the intervention terms of our search strategy, included in Appendix 1 of the Supplementary Material (Additional file
1). In North America, Australia and other countries outside of the UK, this term is often used to describe people who use mental health services. Our initial drafts of our search strategy did include a larger number of terms for peers, including the term “consumer”. However, when piloting our search terms we found that a simplified search, excluding some intervention terms, continued to pick up all our model papers and streamlined the results more closely to our inclusion criteria. In spite of these considerations, we cannot rule out the possibility that our reduced search strategy may have missed some relevant studies. This shortcoming highlights the difficulties of conducting reviews in fields where the language used is not well-defined and varies across study locations.
At the stage of peer review, it was highlighted that the inclusion of social support as an outcome for appraising the effectiveness of group peer support may be problematic, since initiating an intervention involving contact with others may physically increase social support. Only one study included in the present review included social support as an outcome and found no evidence for an effect of the intervention. This issue of circularity is particularly pertinent with respect to studies that do not include follow-up measurements beyond the end of the duration of the intervention. Only one included study reported social support as an outcome, which was assessed during and at the end of the intervention but not at longer-term follow-up. However, the study reported no effect of the intervention on social support. In order to appraise the impact of group peer support interventions on social support, it may be necessary for future studies to consider follow-up points beyond the end of the intervention. If any change in the outcome is maintained, this would be a more reliable indicator of any effect of the intervention.
Interpretation and contribution to the evidence-base
The findings of this review contribute to the mixed evidence-base for the effectiveness of peer support interventions based on findings from RCTs. Similarly, to the earlier review by Lloyd-Evans and colleagues [
21], interventions categorised as peer support services were found to improve recovery but not empowerment. Previous reviews have found that group peer support may increase empowerment [
10] and hope [
25], however, not all studies included in these reviews met our eligibility criteria, often due to the involvement of non-peer professionals in the delivery or moderation of the intervention. Compared to the more recent review [
10], this may have reduced the power of the meta-analyses to detect a small effect across studies. Since empowerment is a component of recovery [
42] and the effect of group peer support on recovery is small, intervention effects on recovery components are less likely to be detected by smaller studies and meta-analyses.
The meaning of recovery may differ between different individuals as it is a personally defined process [
75] and since peer support is a complex intervention, it may also work in different ways for different individuals. Therefore, individual domains of recovery may change at different rates within the recovery process, though broader measures of recovery are more able to capture overall improvement within the short timeframe of most included RCTs. Although further high-quality studies are needed to fully rule out potential influences of bias on study findings [
21], the findings of this review are indicative of a positive effect of group peer support on recovery. Four of the five studies included in the quantitative synthesis were self-management interventions, which suggests this intervention-type may be effective for recovery. It is worth noting that sensitivity analyses using just TAU comparison groups did not alter findings for recovery, though only two studies [
52,
58] employed active comparator conditions involving non-peer clinicians, which tentatively suggests that structured peer-delivered self-management interventions may be comparably effective for enhancing recovery to those delivered by other providers. This supports the findings of a previous review [
76], which found no difference in the effectiveness of interventions delivered by peer and non-peer providers for improving recovery outcomes. All self-management interventions involved contributions of examples from the lived experience of group facilitators, and recovery-orientated education, suggesting that recovery may be exemplified through practical strategies suggested by facilitators and group members, which could contribute to experiential knowledge and intervention effectiveness [
77]. However, it is possible that within peer support interventions delivering a structured curriculum, the potential for the exchange of experiential knowledge developed through individual experience may be limited. Mutual support groups might offer the potential to increase recovery through the sharing of personalised experiential knowledge [
15] and coping strategies [
78] though the relative absence of these trials in the literature prohibited comparisons of these intervention types on recovery outcomes.
Previous reviews have found no evidence for an effect of group peer support on global symptoms [
25] and no difference in symptoms compared to TAU [
25], or to non-peer providers [
76], across peer support interventions. Interpretation of our findings for global symptoms as fully consistent with those of previous reviews is complicated by the small number of trials contributing to the meta-analysis and heterogeneity in trial design, since one study [
52] compared two self-management interventions. This may have reduced the relative effectiveness of group peer support for symptoms since self-management interventions, delivered by either peers or non-peers, were found to improve psychiatric symptoms by a recent review and meta-analysis [
79] and the study included in the present review found evidence for improvements within both groups [
52]. Previous reviews have also found more consistent evidence for peer-delivered self-management interventions than other forms of peer support [
9], though the present review found no evidence for an effect of group peer support on depressive symptoms. It has been suggested that recovery outcomes may be more appropriate than clinical outcomes for assessing the effectiveness of peer support [
26], since the aim of interventions are to improve recovery rather than to eliminate symptoms [
75], which may still be present throughout the process of reclaiming personal well-being and satisfaction in life [
40]. However, it was not possible to assess the impact of group peer support on other outcomes that may be important for recovery, such as quality of life or social outcomes [
30], as either no or few studies reported these. These outcomes may also have greater value to many individuals with lived experience of mental health conditions than traditional clinical outcomes [
80].
Our findings for group peer support broadly parallel those of the concurrent review by White and colleagues [
31] for the effectiveness of one-to-one peer support for improving outcomes for people using mental health services. The available evidence base for one-to-one peer support similarly suggests that interventions may be more likely to improve personal recovery than outcomes related to clinical recovery. Both reviews indicate a small positive effect for recovery, from a similar number of trials, indicating that this may be a consistent effect for peer support, irrespective of whether the intervention is delivered individually or in groups. Although our review does offer a tentative suggestion for a potential intervention effect on global symptoms, which could later be confirmed through expansions to the evidence-base, our more positive finding may be explained by the high representation of self-management interventions in the synthesis [
9] rather than by the format of delivery. In the case of both reviews, the use of lived experience within included interventions in relation to its hypothesised contribution to the mechanisms of effect is rarely described, which could be further specified in order to fully appraise the mechanisms of peer support. Comparably to the findings of the present review, White and colleagues also note the limited number of studies reporting each outcome and the continued presence of some risks of bias to included study findings, limiting interpretation of the available evidence base for both approaches and its utility for informing policy and service developments.
Research implications
The findings of this review highlight the current paucity of evidence from high quality trials of group support interventions needed to draw firm conclusions about effectiveness for a broad range of outcomes. As a result, many reviews of peer support have combined heterogeneous groups of interventions to attempt to appraise effectiveness [
26,
78]. The present findings suggest one distinction in terms of anti-stigma and self-management as subcategorizations within existing typologies, based on a limited number of included studies. The question of the most effective forms of peer support within different settings remains [
26] and cannot fully be addressed by meta-analytic approaches at present, due to an insufficient number of trials to group interventions appropriately [
78]. Future trials could clearly define the model of group peer support used and ensure people with mental health conditions adopt leadership roles in the design of the intervention, to ensure lived experience expertise is optimised [
81].
A more holistic appraisal of effectiveness for recovery would also be facilitated by the inclusion of a broader range of outcomes and service settings in order to expand the current evidence-base. In particular, there is a current lack of high-quality trials of mutual support group interventions, in spite of the high prevalence and uptake of this form of mental health support across the UK and the United States [
82] and the large body of qualitative literature detailing personal benefits derived through this form of intervention [
20]. Trials of group peer support interventions to improve outcomes for people diagnosed with common mental health conditions are virtually absent in the literature and these are also strongly encouraged. Expansions to the current evidence-base could establish more conclusive evidence for a positive effect of group peer support on recovery outcomes. Future reviews could then determine the specific effectiveness of structured and unstructured interventions, self-management and anti-stigma interventions, and for different clinical groups, to guide implementation within primary and secondary care settings. The present review found no evidence that small improvements in recovery were due to changes in hope or empowerment. Although these findings were based on a limited number of studies, this raises questions regarding causal mechanisms of existing group interventions. It is possible that increases in recovery could be caused by changes in component processes such as meaning or connectedness [
42], which were not reported by included studies and future studies could include measurements of these. Qualitative accounts of individuals participating in group peer support interventions, both as process evaluations embedded within trials and as independent studies could indicate the elements of the intervention that are helpful and mechanisms of effect [
83]. This may be particularly informative for determining whether self-management is an essential intervention component for improving recovery. Previous reviews [
11,
12,
84] have provided useful summaries of proposed mechanisms of effect for peer support interventions, which have also been identified in qualitative analysis [
18]. Future group peer support interventions need a clear theory of change and proposed mechanism of hypothesised effect as it is uncertain how any of the positive results presented were achieved from the included studies.
Policy and practice implications
The findings of this review and of other reviews that have included group peer support approaches [
8,
10,
25,
26] are promising with respect to the potential for group peer interventions to enhance recovery for people using mental health services. The current evidence base, however comprises a small number of trials of heterogeneous group interventions, often with considerable risks of bias to study findings. There is also limited available evidence to make conclusions about effectiveness for a broad range of outcomes that may be important for recovery, particularly social outcomes. This prohibits recommendations for the routine implementation of specific forms of group peer support across mainstream services at present. Some negative psychological outcomes have been reported previously by a trial of an online mutual support intervention for women with breast cancer [
85] and by a study included in this review [
59], in spite of high user satisfaction in both instances. If online mutual support group interventions are adopted by services, these may benefit from moderation, either by peer or non-peer professionals [
59], to guard against any potentially negative effects.
The findings of the present and previous reviews [
10,
25] suggest that where structured peer support groups are implemented locally, these may make small improvements to personal recovery for individuals accessing these services. International goals to implement recovery-orientated services within mental health systems [
86] may also be assisted by increasing implementation of interventionsdelivered by people with lived mental health conditions, ensuring individuals who use mental health services have had a lead role in the development of these [
81] in order to truly facilitate the integration of recovery principles and values [
87] and cultural change in working practices.