A systematic review was conducted to identify all relevant publications using PsychINFO, MEDLINE, ScienceDirect. We complied with the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) (9). The search terms used were “bipolar disorder“ associated with “personal recovery” or with “mental health recovery”. Indeed, both of the latter two terms were found to describe the process of PR in published studies. These terms needed to appear in title, abstract or keywords. Duplicates were eliminated.

Figure 1 presents the flowchart illustrating the search and identification of relevant articles.

Relevant articles were selected based on the following inclusion criteria: peer-reviewed studies available in English or French with the full text available that were published prior to 1 May 2021 and that focused on PR in people suffering from BD. No backward time constraints were used. Three types of studies were selected: qualitative studies, interventional studies, and observational studies.

Articles that met the following exclusion criteria were excluded: first-person accounts and case reports, study protocols that had no results yet available, studies of multiple psychiatric disorders, and studies not specifically focusing on PR.

First, articles were selected based on their titles and abstracts. Second, the full texts of all of the included articles were read. The authors (MCE and MD) performed this work independently and blindly using the same bibliographic search criteria. In the event of disagreement, the relevant articles were discussed with a third author (BS). Additional publications were found by reviewing the citations of the included papers.

The articles were classified into three types: qualitative, interventional and observational studies.

Data extraction was differentiated according to the type of article's pre-established grids.

For all studies, data were extracted from the articles and reported by data type as well as country of the research group, type of study, sample size, scale assessing PR/description of PR, correlated dimensions or outcomes, and significant findings/main themes.

Thematic analysis was conducted through a line-by-line analysis of included papers coding relevant text. The analysis of the data was carried out using a mainly inductive approach, without a pre-established thematic grid.

The identified articles were analyzed using a thematic analysis that incorporated the following themes: identification of how PR was measured in the various works, a comprehensive approach to PR in BD, and how care may contribute to PR. Through these three themes, we presented the results of different types of work: qualitative studies that described this process, as well as a number of observational and interventional studies exploring the measurement of PR and the factors involved in this process. Additionally, we presented the various interventional studies that targeted PR as an outcome to better understand the evidence for care tools that might contribute to it.

For the content analysis of the qualitative studies, themes and subthemes were extracted from all studies.

The level of scientific evidence of the published studies was classified according to a decreasing level of scientific evidence, in accordance with the guide for the analysis of the literature and the gradation of recommendations, published by the Haute Autorité en Santé (HAS) in 2013 (10). This grid was adapted from the Sackett score (11).

Twenty-four articles met the inclusion criteria (flowchart in Figure 1). The study characteristics and main results are presented in tables according to the types of studies: qualitative studies (Table 1), observational studies (Table 2), and interventional studies (Table 3).

Even though no backward time constraints were used in the selection of articles, no articles published prior to 2010 met the selection criteria with these keywords.

In this review, we were interested in the measurement tools for PR used with people with BD. In total, we identified four scales assessing PR.

Only two of these PR scales were specifically validated for BD: the Bipolar Recovery Questionnaire (BRQ) (8) and the Questionnaire of Personal Recovery (QPR) (24).

The BRQ was the only scale that has been developed de novo from qualitative work among people suffering from BD (13). Despite a moderate sample size (60 subjects), the scale showed good internal consistency (alpha = 0.875). External construct validity was established by incremental differences in personal growth, functioning, mood symptoms and wellbeing. The validation study of the BRQ (8) showed good test-retest reliability at 1 month (r = 0.866, p < 0.001). The sample size did not permit the exploration of its factor structure and dimensions.

The QPR was developed from the CHIME framework. The scale was initially created and validated for PR in schizophrenia (36). It was then reduced from 22 to 15 items for a more robust version. Its validation for BD was later performed by Kraiss et al. in 2019 through a cross-sectional survey that included 102 people: the scale showed good internal consistency on a unidimensional scale (α = 0.92). Convergent validation measures assessed wellbeing, social role participation, and symptomatology (24).

Other psychometric scales, nonspecific to BD, were used in the studies. The Recovery Assessment Scale (RAS) was chosen by Grover et al. (21) in a psychometric study. The factor structure of the RAS scale was explored by Corrigan et al. (37) in a population of people suffering from SMIs. This scale had previously been used in an Australian study (38) with a population of severely mentally ill people, of whom only 38 were affected by BD. The small sample of BD patients did not allow reliable conclusions to be drawn in this first work. Similarly, the Chinese team of Tse et al. (19, 20) used the Stages of Recovery Scale (SRS), a scale previously developed with a population of people with SMIs in Taiwan. In these two studies, Tse et al. used the SRS in a sample of 75 people suffering from BD, in which Cronbach's alpha for the SRS was 0.95.

At the date of this review, all interventional studies, whether controlled or uncontrolled, have used the BRQ scale. As a result, the BRQ scale has the most evidence in terms of sensitivity to change.

Characteristics of psychometric scales assessing PR in BD are presented in Table 4.

This work is, to the best of our knowledge, the first systematic review collecting different types of studies centered on PR in BD, including qualitative works, observational studies, and interventional studies. A review by Murray et al. (41) focused on psychosocial approaches to supporting PR in BD but did not explore the process itself. In addition, Jagfeld et al. recently published a systematic review of only qualitative works centered on PR in BD (6).

This review highlighted several specific characteristics of PR in people suffering from BD. The results of the different studies collected were generally consistent: the studies using psychometric tools confirmed the data of the studies using a qualitative methodology. This observation allows us to reinforce the validity of our results.

We first observed that research on this topic and with this specific population is in its early stages. Our systematic review concerned all research works on this theme published before May 2021 and with no earlier cut-off date. Prior to 2010, in the field of BD research, the term ”recovery" was only used in the literature to refer to “clinical recovery” or “functional recovery”. Research interest in PR in BD has therefore lagged far behind that in the field of schizophrenia.

These observations are in contrast to the growing interest in PR in SMIs, shared by public authorities and mental health users. The COS published by Retzer et al. and selected in this review confirms this interest. A COS is a standardized collection of outcomes recommended to be reported in all controlled trials in a research area for community-based bipolar studies. This COS contained 11 outcomes including PR. PR is thus considered a central outcome for people with BD (18). Interventional research considering PR as a primary outcome in BD is still very limited: only five clinical studies, including two RCTs, were identified. However, five articles that were not selected for this current study presented study protocols on this subject (42–46).

The studies included in our review showed a significant negative correlation with current depression and anxiety, but not with mania. These results are consistent with those of the meta-analysis by Van Eck et al. (1) in schizophrenia, which showed a small to medium association between clinical and personal recovery. Depression would alter the feeling of PR, whereas PR could reduce symptoms and their thymic impact Best, 2020 (47). Among the subdomains of clinical recovery, affective symptoms were the most correlated with PR, and particularly depressive symptoms. Dubreucq et al. (48) longitudinally assessed the overlap and mutual distinctions between clinical and personal recovery, and found that clinical recovery and personal recovery, although distinct constructs, predict each other over time.

Van Eck et al. showed a positive but weak correlation between PR and functional recovery, while Dubreucq et al. put in evidence a mediating effect of quality of life, including various resources such as social support or autonomy. The studies included in our review focusing on BD, showed an important role for social factors, such as a valued role and the quality of social support.

Our results therefore confirm the reciprocal interplay of these different outcomes and the importance of taking them into account in care.

Some of the salient features of PR found here are similar to those described in other SMIs [for a review see (7)]: needing to find meaning in life and a fixed point in one's life despite fluctuations (14), taking responsibility for one's health and finding ways to self-manage one's disorders, making “the choice to get better” (12, 13), being employed and having a meaningful role (14, 20, 29, 39). In addition to other SMIs, but counterintuitively concerning mood disorders, PR does not seem to correlate very closely with the symptomatology itself, especially manic states (21, 22). The process seems to be more related to the relationship that people have with their thymic fluctuations than with the fluctuations themselves in a metacognitive way, including resilience (49).

Indeed, our findings provide additional knowledge on more specific elements of PR in BD: first, a new relationship with fluctuating mood states, which includes, on the one hand, an acceptance of this reality and, on the other hand, a decrease in hypervigilance and active struggle against these fluctuations, which may even include risk taking (12, 14, 22, 29, 39). This notion of acceptance had been found by Jagfeld's team in a review of qualitative studies on PR in BD (6). Acceptance concerned a form of ambivalence toward manic symptoms requiring a form of mourning, as well as acceptance of the repeated experience of losses (job, relationships) in connection with extreme mood states. Risk-taking includes inappropriate but potentially pleasurable activities, such as recreational drug use, binge drinking, and casual sex. It has been suggested that low-risk activities that are not associated with high levels of symptoms can be positive experiences and linked to better PR through experimenting and socializing more (39).

Moreover, our review highlights the fact that openness to others, including trust and closeness, seems to be a particularly important element of PR in BD (12, 14, 26). With regard to people with schizophrenia, Jose et al.'s 2015 review about PR in schizophrenia (50) found five themes of personal recovery specific to this disease: the process itself (non-linear, ongoing), self-orientation (e.g., understanding and accepting oneself, returning to a normal state), family relationships, social inclusion and connections, and recovery from illness (no symptoms, good functioning). The recent review by Leendertse (51) showed a strong correlation with the dimensions of hope, empowerment and meaning in life in PR and a lesser correlation with connectedness and identity. It would therefore seem that the notion of connectedness, linked to that of openness to others, is a PR lever that is rather specific to people living with BD compared to those living with schizophrenia. This result suggests that people suffering from BD are less concerned about the autistic symptoms and the social withdrawal present in the negative symptomatology of schizophrenia. Their need for quality social links would therefore be greater to improve their PR.

The CHIME conceptual framework is currently the most widely used framework for describing the PR process (5). However, the data in Leamy's review mainly concerned schizophrenia. The CHIME framework has been criticized for not being sensitive to the specific characteristics of certain populations, especially people with mood disorders, and for omitting the public relations themes of coping and courage or risk taking (for a review, see van Weeghel et al., 2019). While Todd's work in 2012 (13) found dimensions close to those of the CHIME framework, other studies identified new components of PR in this population.

These new elements are of two types. First, the CHIME framework describes elements of PR but does not address the underlying psychological processes. The latter appears to be central in the mechanism of PR in BD, based on the studies in our review. Indeed, the modalities of mental relationships with mood fluctuations have been considered important levers of PR (12). Concordantly, it has been shown (29) that certain regulatory mechanisms of manic or hypomanic states, such as dampening, impede PR. Other authors are exploring cognitive models that may provide explanations for the PR process in cognitive ways. In particular, it seems that a lack of control over fluctuations, acceptance or even a high level of risk-taking may promote PR (39).

Furthermore, the notion of relationships with others is only slightly developed in the CHIME model. However, our results suggest the importance of relationships and quality connections to others (12). Indeed, personal meaning in relationships may be more important in BD recovery than in other SMIs.

Our findings are quite consistent with those of the qualitative literature review conducted by Jagfeld et al. (6). These authors emphasized vocational goals and self-management skills and emphasized the question of relationships with others. Indeed, importance was given to the need to address the tensions in particularly intimate relationships caused by mental health difficulties and to deal with the risk of stigmatization associated with openness. Jagfeld et al. added a “tension” dimension to the CHIME framework, which links the latter subdomain to the issue of the need for acceptance of vulnerability and personal limitations, as well as the need to handle ambivalence about mania. The authors thus proposed a POETIC (Purpose and meaning, Optimism and hope, Empowerment, Tensions, Identity, Connectedness) framework that is supposed to be more specific to PR in BD, according to qualitative data (6).

Our search strategy included three bibliographic databases, which may have limited the number of articles selected. As a result, it is possible that some data of interest could not be collected. We excluded all articles that were not specifically focused on people with BD. In this way, we were able to collect very specific information in a very complex area of research. This choice may also have limited the number of articles selected.

The literature search was limited to peer-reviewed studies available in English or French, which may not represent all the evidence and may have introduced a language bias.

The characteristics of PR in people with BD are in the early stages of exploration. The small number of articles to date may limit the scope of the evidence. Moreover, our work found that there is a lack of RCTs and works with a high level of scientific evidence. It is clear that future work will need to add to current scientific knowledge on the PR process in BD.

Our review has highlighted several possible avenues for future work.

The first avenue concerns the measurement tools of PR in BD. To date, the only scale specifically validated for the context of BD is the BRQ (8), which is the only tool that has been used in clinical research in specific bipolar populations. Our work highlighted that some new components of PR emerged from subsequent qualitative and observational studies. These elements could be taken into account in the construction of other measurement tools complementary to the BRQ.

The second avenue concerns the care tools and “recovery-oriented intervention” in BD. Our findings showed that a number of characteristics of health services supporting resilience, respect, hope and self-directed empowerment supported PR (19).

In 2017, Murray et al. proposed the use of mindfulness tools, such as mindfulness-based cognitive therapy (MBCT) programs or acceptance and commitment therapy (ACT) therapies. These tools would promote acceptance and decrease self-stigma (41). This proposal is consistent with our results; we can indeed hypothesize that these care tools are active in the cognitive relationships maintained with mood swings.

These principles are embodied not only in psychosocial interventions (41) but also in other fields, such as involvement in decision making regarding medication and constrained care. These issues are highly complex and require awareness and vigilance so that the development of self-management does not lead to a decrease in carers' commitment. In contrast, it is a question of finding and developing other, more humanistic, support methods that are a source of recovery. The results of the clinical trials collected in our review were consistent with these data. Recovery-focused care tools were already described (30, 31): they focused on self-determined goals and avoided stigmatizing discourses. This type of care helps people with BD understand and accept their disorders and develop self-care and a sense of agency. We did not find any clinical studies investigating the effects of pharmacological treatments with RA as the primary outcome.

Beyond the care tools themselves, it is also necessary to study communication modalities and the caregiving posture, as well as the partnership with peers in recovery-focused care, that could specifically facilitate PR. In the qualitative study conducted by Tse et al. in 2019 (17), the interviewed patients emphasized the value of learning from peers. Additionally, notions of empathic communication and hope instillation were mentioned. We can link this observation of Tse et al. to the previous results concerning the particular weight of connectedness and openness to others in the PR of people living with BD. Enabling people with BD to develop quality communication skills could be a central goal of care. The goal of recovery-focused care could first be embodied in the caregiving posture itself, through working with peers and valuing experiential knowledge. Additionally, the strengthening of the skills of openness to others, connectedness and bonding with others could be the focus of new care programs, as well as new research perspectives.