Interventions for improving mental health in athletes: a scoping review

The aims of this scoping review were to map the current literature on interventions for improving mental health in athletes, identify knowledge gaps, and generate future research questions. The Preferred Reporting Items for Systematic Review and Meta-Analyses Extension for Scoping Reviews (PRISMA-ScR) guided this review. A systematic literature search was conducted in SPORTDiscus, PsycINFO, and SCOPUS and 44 intervention studies met the inclusion criteria. Results showed that 22 studies (50%) implemented cognitive behavioural principles, and the majority of these studies were in ﬂ uenced by various mindfulness programmes. Most studies (93%) included healthy athlete samples, and athletes aged 15 – 19 were the most examined age group (43%). Only three studies used clinical criteria in their sampling of participants and mediators were examined in two studies. The scarcity of studies examining mediators and subclinical or clinical samples revealed critical knowledge gaps in the literature. Furthermore, the critical appraisal showed that regardless of study design, most studies demonstrated low internal validity. We propose the use of high-quality single-case studies with athletes who experience subclinical or clinical mental health issues, and further investigation of mechanisms of change linking intervention components to outcomes of interest.


Introduction
The number of published studies focusing on athletes' mental health has steadily increased in recent years (Kegelaers et al., 2022;Ströhle, 2019), and several position and consensus statements have recently been published on the topic (Prior et al., 2022;Vella et al., 2021).Moreover, the results of prevalence studies on mental health, mental health problems, and mental disorders in athlete populations in the past decade (e.g.Gouttebarge et al., 2019), have highlighted athletes' need for effective mental health interventions.
Researchers have, however, questioned the existing evidence on mental health interventions for athletes and argued that the current evidence is insufficient to inform practice or policy (Prior et al., 2022).In light of this criticism, there is a need for a comprehensive mapping of the literature on psychological interventions for mental health in athletes to gain an in-depth understanding of the current knowledge base.Furthermore, a comprehensive mapping of the literature will identify critical knowledge gaps, which will guide future research and advance the knowledge base about mental health interventions for athletes.

Mental health, mental health problems, and mental disorders
Mental health is complex, and many different definitions grounded in different theoretical perspectives have been put forward in the literature (Lundqvist & Andersson, 2021).A frequently referred to definition of mental health is one proposed by the World Health Organisation (WHO).WHO defines mental health as 'a state of well-being that enables people to cope with the stresses of life, to realise their abilities, to learn well and work well, and to contribute to their communities' (WHO, 2022, p. 8).This definition emphasise functionality in life and that mental health is more than just the absence of mental ill-health.
Regarding theoretical perspectives, single-continuum models suggest that individuals move back and forth along a mental health-mental illness continuum (Chen et al., 2020).In the sport psychology literature, however, arguments against single-continuum models have been raised, because continuum models do not provide guidance on how to interpret symptoms of mental illness and whether these symptoms are natural reactions to adverse events in life or sport, or if the symptoms are indications of a clinical mental disorder (Lundqvist & Andersson, 2021;Lundqvist et al., 2022).Natural reactions in a sport setting, for example, could be a passing experience of performance anxiety before a competition or psychological distress during intensive training periods.To pose an alternative, Keyes (2002Keyes ( , 2005) ) argued that mental health and mental illness are not opposite ends of a single continuum but rather two distinct-but-related dimensions existing on two separate continua.By adopting the two continua model, we assume that athletes could experience good mental health and symptoms of mental illness simultaneously (Uphill et al., 2016).Although Keyes' two continua model of mental health (2002Keyes' two continua model of mental health ( , 2005) ) provides a more comprehensive view of mental health than single-continuum models, the dual-continuum model also adds complexity and imposes challenges when interpreting the mental health of athletes (Lundqvist & Andersson, 2021).It is thus important to differentiate between natural, expected mental health states and symptoms of mental illness in need of treatment.To help differentiate expected fluctuations, problems, and disorders of mental health, two classification systems are in place: the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5-TR; American Psychiatric Association, 2022) and the 11th revision of the International Statistical Classification of Diseases and Related Health Problems (ICD-11;World Health Organization, 2019).A mental disorder is a clinically significant disturbance in psychological functioning and is associated with psychological distress or impairment in social, occupational, or other essential activities (American Psychiatric Association, 2022).Mental health problems, on the other hand, refers to subclinical psychological ill-being that may not necessarily develop into or fulfil the criteria of a clinical mental disorder.
For competitive athletes specifically, the differentiation between natural, expected mental health states and symptoms of mental illness might become a challenge due to the context in which they operate.What can be seen as functional in athletes' performance environment (e.g.excessive training, weight-control) might be considered dysfunctional in other settings and vice versa (Lebrun & Collins, 2017).Additionally, competitive athletes face sport specific stressors (e.g.fear of failure, deselection; Küttel & Larsen, 2020), organisational stressors (e.g.issues related to leadership, culture and team, logistics and environment, and performance; Arnold & Fletcher, 2012), physical and psychological demands (e.g.pressure to perform; Rice et al., 2016), and operate in performance environments to an extent that most non-competitive athletes (i.e.recreational athletes, general population) do not.For these reasons, one must be careful to avoid pathologising natural human experiences, due to the potential consequences of falsely diagnosing normal suffering and potentially overtreating or stigmatising someone.At the same time, one must not disregard psychiatric issues and thereby withhold appropriate treatment from someone who needs it (Wakefield & First, 2013).

Interventions for mental health
Interventions, in psychological practice, refers to assessment, diagnosis, prevention, treatment, psychotherapy, and consultation (American Psychological Association [APA], 2006), and several empirically supported interventions for mental health are available (see APA Division 12 (n.d.) for an overview).Interventions for improving athletes' mental health may therefore include promotion of mental health (e.g.Birrer et al., 2021), prevention strategies (e.g.Becker et al., 2012), or treatment (i.e.clinical interventions, psychotherapy) of mental health problems and disorders (e.g.Lundqvist, 2020).
A few previous studies have attempted to synthesise the literature on mental health interventions for athletes.Kegelaers et al. (2022) conducted a systematic scoping review on studies of the mental health of student-athletes, which also included intervention studies.The results from the intervention studies were, however, merely described and not examined or discussed in depth.Moreover, the review provided no future directions of research or considerations specifically about mental health interventions for student-athletes.Sutcliffe et al. (2021) performed a systematic review and meta-analysis of sport-based interventions focused on mental health and mental health literacy in non-elite sport (i.e.athletes, coaches, or parents).However, the sport-based interventions included both psychological and non-psychological interventions (e.g.acupuncture; Luetmer et al., 2019), and the synthesis was conducted on reported outcomes of mental health problems and mental health literacy, not intervention characteristics specifically.The meta-analyses showed small or undetectable effects of interventions on anxiety, psychological distress, and well-being, whereas no statistically significant effects of the interventions on depression were found (Sutcliffe et al., 2021).Prior et al. (2022), on the other hand, focused on competitive sport in their systematic scoping review on interventions, recommendations, and policy concerning athletes' mental health.Although interventions are reported in the review, the authors did not provide information about how mental health outcomes were operationalised, map theories underpinning the interventions, or specify whether mediators were included (Prior et al., 2022), which are key pieces of information to understand the current evidence for mental health interventions for athletes (Herzog et al., 2022).
Though previous studies (e.g.Kegelaers et al., 2022;Prior et al., 2022;Sutcliffe et al., 2021), in different ways, have provided valuable insights and increased our understanding of mental health interventions in athletes, questions remain concerning the theories underpinning interventions, the effective mechanisms of change, and the study quality.Hence, a scoping review targeting psychological interventions for improving mental health in athletes (regardless of competition level) will add to the synthesis of knowledge about mental health interventions for athletes, highlight current knowledge gaps, and provide several directions for future intervention research.

The current study
Researchers have indicated the limited availability of studies on the effectiveness of interventions for athletes' mental health (e.g.Ekelund et al., 2022;Stillman et al., 2019).For this reason, a comprehensive scoping review would benefit from aiming to capture the whole spectrum of intervention research rather than limiting its scope to one specific subgroup of athletes or excluding key aspects of interventions (e.g.theory, mediators, operationalisation).A scoping review is most suitable for such a comprehensive mapping and summary (Arksey & O'Malley, 2005;Grant & Booth, 2009) aimed at identifying knowledge gaps and generating future research questions rather than informing policy (Munn et al., 2018).
The main aim of this scoping review was to map existing evidence of interventions (i.e.promotion, prevention, or treatment) for improving athletes' mental health and to identify knowledge gaps to facilitate new questions that will guide future research.The review's secondary aims were to outline: (a) the theories/models on which the mental health interventions for athletes were based; (b) the type of treatment used in such interventions; (c) the way in which mental health was defined and operationalised/measured in such interventions; (d) the mediators examined in the studies reviewed; and (e) the athlete populations most commonly researched.Finally, each study was critically appraised to evaluate the quality of the current literature.

Approach, protocol, and registration
For this scoping review, we adopted the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews checklist (PRISMA-ScR; Tricco et al., 2018).Additionally, in line with recommendations specifically for scoping reviews within sport and exercise psychology (Sabiston et al., 2022), a group of experts, consisting of three of the co-authors (CL, HG, and AI) was involved to assist in specifying the research questions, search terms, and eligibility criteria.A study protocol was developed using the PRISMA-ScR (Tricco et al., 2018).The original protocol was approved by all authors and preregistered on the Open Science Framework on 22 February 2022 (https:// osf.io/cbtgm/).Minor edits to the study protocol were conducted up until 12 December 2022.

Eligibility criteria
To be included in this review, articles needed to be peer-reviewed, written in English, and report either observational (i.e.pre/post design) or controlled studies (non-randomised and randomised trials).Additionally, the review included studies that involved competitive athletes and implemented an intervention focused on improving athletes' mental health (i.e.prevention or promotion) or treating their mental health problems/disorders, as defined by the authors of each individual study in their research aims.
Studies were excluded if they: (a) had no available full text; (b) focused solely on performance enhancement/had only performance enhancement as a dependent variable; (c) were grey literature; (d) investigated non-competitive athletes (i.e.recreational sport participants); (e) implemented interventions that indirectly targeted athletes (e.g.coach or parental education programmes, team building interventions); (f) were case studies; or (g) did not report quantitative data (e.g.studies relying solely on qualitative data).Finally, studies were excluded if their reported outcomes pertained to competitive states (e.g.performance anxiety, mental readiness, irrational beliefs) solely in relation to performance and not mental health.

Information sources and search strategy
Two electronic searches were conducted.The first search was conducted on 23 February and 10 March 2022, and a second complementary search was conducted on 15 February 2023, covering work published between February 2022 and 15 February 2023.Both searches included three electronic databases: SPORTDiscus, PsycINFO, and SCOPUS.A professional librarian was consulted throughout the processes of developing a search strategy.A broad search strategy was adopted, and database searches included three search blocks (mental health, context, and intervention) with the search terms ('mental health' OR 'mental illness' OR 'mental ill-being' OR 'mental health problems' OR 'mental disorders' OR 'well-being' OR 'psycho*') AND ('sport' OR 'athlete*') AND ('intervention*' OR 'treatment*' OR 'therap*').All electronic search results were imported to EndNote, where duplicates were removed.
The electronic searches were supplemented with a manual search of reference lists from the included studies in the current study and we also examined studies included in previously published reviews focused on mental health in athletes (i.e.Breslin et al., 2022;Kegelaers et al., 2022;Prior et al., 2022;Rice et al., 2016;Stillman et al., 2019;Sutcliffe et al., 2021).In addition, the Journal of Clinical Sport Psychology (vol. 1) and the Journal of Sport and Exercise Psychology (vols. 1 through 9) were searched manually because these two journals did not have all volumes indexed in any of the databases included in this review.

Selection of sources of evidence
The open-source machine-learning tool ASReview was employed to support the title and abstract screening of the records identified in the electronic searches (see van de Schoot et al., 2021, for more information).In ASReview, articles are manually marked by the researcher as relevant or irrelevant.Based on these classifications, ASReview ranks articles in order of relevance.A mixed strategy (i.e. a combination of pre-determined strategy and data-driven strategy) for stopping criteria was used.A pre-determined strategy is when screening stops after a minimum percentage of articles are screened.With a data-driven strategy, screening stops after a certain amount of consecutive irrelevant articles.
For this review, the pre-determined stopping criterion was set to a minimum of 33% of articles, based on the recommendations of van de Schoot et al. (2021) and Howard et al. (2020).The data-driven strategy was set to 1% of the whole dataset (i.e.210 consecutive irrelevant articles in the first search and 22 consecutive irrelevant articles in the complementary search).Both criteria had to be fulfilled before the screening stopped.Once the stopping criteria were fulfilled, the remaining non-validated articles were classified as irrelevant automatically by ASReview (Ros et al., 2017).When 33% of the articles from the first search had been screened, the last 1,007 articles in the initial title and abstract screening were marked as irrelevant; in the complementary screening, it was the last 189 articles.
The first author (RE) performed the title and abstract screening in ASReview.Following this, the results from the screening in ASReview were downloaded in an Excel file, where articles were ranked in order of relevance by ASReview (see https://osf.io/cbtgm/ for the full details on screening in ASReview).When it became evident that the duplicate removal by EndNote was not complete, duplicates and clearly irrelevant papers among the articles marked as relevant were screened for in the Excel file.RE screened the titles among the articles marked as irrelevant in the Excel file, to check for potential missing relevant articles.No missing relevant articles were found.
Thereafter, relevant articles were imported into Rayyan (an online programme for managing systematic reviews; Ouzzani et al., 2016) for additional manual title and abstract screening.Two of the co-authors (RE and SH) independently conducted the manual title and abstract screening.Discrepancies (9.9% of the articles) were resolved by reading the abstract again, reaching a consensus decision based on the inclusion and exclusion criteria, and, when needed, further discussion with a third co-author (AS).At this stage, articles reporting on non-psychological interventions (e.g.ice baths, acupuncture, dietary modifications, massage therapies) were excluded.Following this, the first author completed a full text screening of articles that were deemed eligible.
In the manual search, the first author screened titles and abstracts, and completed a full text screening of articles.Thereafter, RE, SH, and AS discussed and reached consensus decisions regarding potentially eligible articles based on the inclusion and exclusion criteria.

Data charting process
Data from eligible studies were charted in a data-charting form.The form included relevant information on the key characteristics of each study.RE independently charted data from all potentially eligible studies while AS independently charted data from 30% of the potentially eligible studies.No disagreements were found.Finally, RE, SH, and AS critically discussed the findings and determined the final inclusion of studies.

Data items and collection process
We extracted data on article characteristics (i.e.year of publication and country of origin), characteristics of the study population (i.e.sport, gender, age, and competition level), study characteristics (i.e.design, type of intervention, measurement used, dependent variables, identification of theory, intervention facilitator, procedural reliability or treatment integrity, and mediators), and key findings.

Critical appraisal
Critical appraisal is a key feature in reviews that systematically assess the quality of the included studies and it helps readers determine the credibility of the results (Tod et al., 2022).Studies with a group-based design were critically appraised with the Mixed Methods Appraisal Tool (MMAT) Version 2018 (Hong et al., 2018).The MMAT is used to assess different methodological qualities of various study designs (e.g.quantitative randomised controlled trials [RCTs] and quantitative non-randomised studies) and each study design is judged by five specific criteria.Methodological quality criteria for RCTs are (a) Is randomisation appropriately performed?(b) Are the groups comparable at baseline?(c) Are there complete outcome data?(d) Are the outcome assessors blinded to the intervention provided?and (e) Did the participants adhere to the assigned intervention?For quantitative non-randomised studies following methodological quality criteria was judged (a) Are the participants representative of the target population?(b) Are measurements appropriate regarding both the outcome and intervention (or exposure)?(c) Are there complete outcome data?(d) Are the confounders accounted for in the design and analysis?and (e) During the study period, is the intervention administered (or exposure occurred) as intended?Each criterion of the MMAT is rated as 'Yes', 'No', or 'Can't tell'.
For single-case designs, the Risk of Bias in N-of-1 Trials Scale (RoBiNT; Tate et al., 2015) was used.The RoBiNT consist of two subscales, one to assess internal validity and the other to assess external validity.The internal validity subscale includes seven items (i.e.design with control, randomisation, sampling of behaviour, blinding of people involved in the intervention, blinding of assessor(s), interrater agreement, and treatment adherence), whereas the external validity subscale includes eight items (i.e.baseline characteristics, setting, dependent variable, independent variable, raw data record, data analysis, replication, and generalisation).Each item is rated on a 3point scale (0-2), hence, the possible maximum combined score when using the RoBiNT is 30 points.
RE and SH independently appraised all studies with a quasi-experimental design.Mixed-methods designs were critically appraised based on their quantitative method.All included mixed-methods studies were independently critically appraised as quasiexperimental designs by RE and SH.RE and AS independently appraised all RCTs and single-case studies.Disagreements between the authors resulted from a lack of clarity in reporting in the reviewed studies.In case of disagreement, the reviewed article was further examined by a pair of co-authors (RE and SH for quasi-experimental studies, RE and AS for RCT and single-case studies) to resolve ambiguity and reach consensus.

Synthesis of results
Interventions were categorised by the type of underpinning model or theory.More specifically, we categorised interventions as focusing either on cognitive behavioural therapy (CBT), stress, motivation, or other.Mindfulness and acceptance-based interventions were categorised in the CBT category, as these are strongly embedded in the CBT tradition (Hayes, 2004;Hayes & Hofmann, 2017).Note, however, that the categorisation based on the underpinning model or theory was purely data-driven (i.e.not based on an existing framework) and was intended to provide a broad overview of the main models and theories underpinning the included studies.

Characteristics of sources of evidence
Sources of evidence that were screened, assessed for eligibility, and included in the review can be found in Figure 1, as can any reasons for exclusion.The type of intervention, sample, design, outcomes, measures, mediators, and key findings of each study are summarised and presented in Table 1.The definition of mental health used in each study and the theory, framework, or model underpinning the interventions (as described and cited by the studies' authors) are summarised in Table 2.

Intervention characteristics
Face-to-face intervention delivery was used in 34 of the 44 intervention studies.Five studies used a hybrid mode of delivery that combined face-to-face elements with either structured, self-based work, or digital components.Three studies were conducted solely online or without face-to-face interaction, and the mode of intervention delivery was unclear     (Belz et al., 2020;Gabana et al., 2019) to 18 three-hour sessions over six weeks (Stranberg et al., 2019).Individual interventions ranged from one 30-minute session (with additional individual self-work; McCarthy et al., 2010) to 21 sessions over 16 weeks (Çakmakcı et al., 2020).For further details on intervention characteristics, see Table 1.
Theories, frameworks, and models All but four studies reported either a theory or model that was underpinning their intervention.Half of all identified studies (n = 22) based their intervention on some type of CBT.
Of the 22 CBT-based studies, 14 employed some type of mindfulness intervention (e.g.MBSR, MAC, MSPE).Four studies were categorised as a stress intervention, whereas two studies combined stress with either CBT or motivation theories.A motivation theory or model (i.e.self-determination theory, competence motivation theory, achievement goal theory, or personal goal management programme) was used in five studies.Additionally, seven studies could not be categorised and were guided by pedagogy (i.e.experiential learning), positive psychology, community-based participatory framework, or eudaimonic and psychological well-being.Finally, four studies could not be categorised because they reported no underpinning theory or framework.For details, see Table 2.
Mental health was also defined as a fluctuating affective state characterised by the presence of positive affect and the absence of negative affect (Laurin et al., 2008).Various aspects of anxiety were most frequently used to operationalise mental health, with 12 studies using this approach.Mental health was operationalised via measures of well-being, stress, and depression in 12 studies, whereas eight studies included mood and affect in their operationalisation, seven included psychological (in)flexibility or experiential avoidance, and five included outcomes related to eating disorders.Five studies included measures of psychological distress as an outcome.Four studies used a global measure of mental health, whereas self-compassion, burnout, and life satisfaction/quality of life were each included in three studies.Two studies included rumination as an indicator of mental health, and another two included global measures of mental illness or mental health problems as outcomes.Self-criticism, vitality, thought suppression, DSM diagnosis, and emotional regulation were included as outcomes in one study each.For details, see Table 1.Well-being as a fluctuating affective state 34 involving the presence of positive affect and the absence of negative affect 35, 36   Personal goal management programs approach 37, 38   Motivation Macdougall et al. (2019) Global well-being 39 , integrated with dimensions of hedonia and eudaimonia 40, 41, 20   Mindfulness acceptance commitment approach 25 , motivational interviewing 42 CBT (Continued)

Mediators
Two of the 44 intervention studies examined mediators.Perna et al. (1998) investigated the mediating effect of fatigue and total mood disturbance on mood during heavy training in a cognitive behavioural stress management programme (CBSM).In Shannon et al. (2019), mindfulness, competence satisfaction, and stress were investigated as mechanisms of change in a mental health intervention that applied self-determination theory to promote well-being, reduce stress, and increase competence in mental health self-management.
Most studies (n = 21) recruited athletes from various sports.Seven studies recruited from soccer, three studies recruited from rugby, two studies recruited from rowing, and two from various parasports.The remaining studies included athletes competing in cycling, baseball, shooting, tennis, swimming, squash, basketball, and track and field, and one study did not report the type of sport practiced by its participants.
Three studies explicitly screened for and required clinical cut-off values as inclusion criteria.Sandgren et al. (2022) included participants who self-reported having mild symptoms of eating disorders, and a confirmed eating disorder diagnosis was required for inclusion in two studies (Çakmakçı et al., 2020;Stranberg et al., 2019).

Critical appraisal
Of the 44 studies identified, 24 were quasi-experimental, 10 were RCTs, five were mixedmethods, and five employed a single-case design.Raw scoring and details of the critical appraisal can be found in Appendix A.

Quasi-experimental studies
The critical appraisal of the 29 quasi-experimental studies (24 quasi-experimental and five mixed methods) showed sufficient quality of the representative sample of the target population as well as appropriate measurements of outcome and intervention for all studies.However, only 11 studies provided complete levels of outcome data, whereas nine studies did not, and it was not possible to assess levels of complete outcome data for nine studies.Furthermore, only one study accounted for confounding variables in their design and analysis, 25 did not, and it was unclear whether the remaining three studies included confounding variables.
Only two studies were determined to have administered their interventions as intended, without contamination or unplanned co-interventions.As for the remaining studies, 20 did not clearly indicate whether their interventions had been administered as intended, and seven studies did not deliver their interventions as intended (without contamination or unplanned co-interventions).

Randomised controlled trials
The critical appraisal of the 10 RCTs showed that four studies appropriately reported and performed randomisation.All RCTs but two had comparable groups at baseline, and only four studies reported complete outcome data.Two studies had blinded outcome assessors, three studies did not, and it was unclear in five studies whether outcome assessors were blinded to the intervention.Finally, three studies had an acceptable participant adherence to the intervention, two did not, and it was unclear whether adherence to the intervention had been sufficient in the remaining five studies.

Single-case studies
Three studies scored 1 out of 14 possible points on the internal validity scale and two studies scored 2 out of 14 points on the internal validity scale.The scores on the external validity scale ranged from 4 to 10 out of 16 possible points.The total score among the single-case studies ranged from 5 to 12 out of 30 points.

State of the knowledge base
Interventions for improving athletes' mental health generally seem to be effective.However, to understand the current evidence, it is important to consider what populations have been studied.Healthy athletes aged 15-19 were the most examined group of athletes, thus demonstrating a trend towards intervening for mental health, regardless of the presence of mental health issues or not.The three studies that did recruit clinical populations (i.e.athletes experiencing subclinical mental health problems and/or clinical mental disorders; Çakmakçı et al., 2020;Sandgren et al., 2022;Stranberg et al., 2019) all focused on eating disorders, and their initial results seem promising.
Regarding treatment studies, there is a general lack of evidence regarding the effectiveness of interventions on clinical levels of mental health outcomes, such as anxiety and depression as defined by the DSM-5-TR (American Psychiatric Association, 2022).The practice of targeting clinical outcomes that may not be present in the sample can be problematic.First, by doing so, the study's authors indirectly attribute to the participants a condition (e.g.depression) that might not need to be addressed to begin with.Second, an intervention that, for example, targets symptoms of depression in a sample where symptoms of depression are not of clinical relevance cannot indicate whether the intervention is effective for the treatment of depressive symptoms.Regarding promotion and prevention of mental health, it can be of interest to investigate the occurrence of mental health outcomes, although the sample may not be expected to show any greater levels of mental health problems.Hence, the targeting of appropriate samples, combined with a clear operationalisation of mental health for each intervention study, is crucial for the interpretation of the intervention's aim and effectiveness.
The findings of this review highlight the field of sport psychology's struggles with operationalisations and definitions of mental health in intervention studies.For example, seven different measures were used to assess anxiety.Similar findings of multiple operationalisations of the same construct have emerged elsewhere (Sutcliffe et al., 2021;Tahtinen et al., 2021), and researchers have argued that it is unlikely that the field of sport psychology will coalesce around a uniform definition and operationalisation of mental health outcomes (Lundqvist & Andersson, 2021).Consequently, potential confusion arises about the targeted constructs in intervention research when mental health is operationalised in many ways or is not clearly defined.
Moreover, this review shows that mental health interventions for athletes to date are strongly influenced by various cognitive behavioural approachesespecially elements of third wave CBT (e.g.mindfulness-and acceptance-based interventions).Third wave therapies aim to be transdiagnostic (i.e.targeting underlying mechanisms that are thought to drive and maintain ill-health rather than treating symptoms in isolation; Hayes & Hofmann, 2017) and have shown promising results in studies targeting common mental health issues, such as anxiety, depression, and eating disorders (Dimidjian et al., 2016).That these well-established approaches are guiding interventions for athletes' mental health is promising, and the field of sport psychology would benefit from continuing in this direction.

Knowledge gaps
This review highlights critical knowledge gaps in current sport psychology research on interventions for athletes with subclinical or clinical mental health issues.These gaps align with previous findings showing that intervention studies with samples of athletes experiencing clinical mental health issues are scarce (Ekelund et al., 2022;Stillman et al., 2019).Untreated clinical issues such as depressive symptoms or major depressive disorder may result in personal suffering and struggles in the athletes' private lives and/or careers (e.g.performance issues or termination from sport; Wolanin et al., 2015).Hence, it is crucial to gather empirical evidence on how to best support and treat this group of athletes.
Despite efforts to rely on theories or existing established models, a major concern for intervention research in sport is the lack of knowledge about mediators that explain how and why interventions affect outcomes.The use of mediating variables is central to interventions designed to affect behaviour (Mackinnon, 2011) and evidence-based interventions should not solely target the troubled behaviour but also the underlying mechanisms (i.e.mediators) that drive the troubled behaviour (Kazdin, 2007).
Dispositional mindfulness (i.e.awareness and attention to our thoughts and feelings in the present moment) and emotional regulation (i.e.lack of emotional awareness, clarity, acceptance, and distress tolerance) are the putative mechanisms of the MAC approach (Josefsson et al., 2019;Moore, 2009).These mediators were, however, not examined in the studies that implemented MAC interventions (Goodman et al., 2014;Gross et al., 2018;Macdougall et al., 2019).Another mechanism underlying third wave therapies is experiential avoidance (i.e. the avoidance of undesirable thoughts, memories, and emotions; Hayes et al., 2006), which has been shown to mediate the relation between various symptom types and psychological constructs (Ruiz, 2010).However, in several studies, experiential avoidance (measured alongside psychological inflexibility; Bond et al., 2011) was assessed as an outcome variable rather than as a mediator.Consequently, current evidence provides limited insight into mechanisms that account for potential change in intervention outcomes.

Future directions
Several opportunities are available for addressing the knowledge gaps identified by this review of research into interventions for athletes' mental health.First, researchers should aim to clarify the goals of their interventions and the contexts to which these are best suited (i.e.mental health promotion, prevention, or treatment), while also considering the athletes' issues and the severity levels of those issues.Moreover, intervention studies including athlete samples with both subclinical and clinical mental health problems and disorders are warranted.If intervention studies are to target subclinical or clinical mental health outcomes, they should also include subclinical or clinical samples.
Second, future mental health intervention studies should include mediators in their design to better discern the processes linking intervention components to targeted outcomes.Including mediators in intervention studies, however, requires a theoretical foundation of a putative mechanism to guide the intervention (Kazdin, 2007).One way to incorporate theoretical foundations more systematically is to employ methodological frameworks such as the Multiphase Optimisation Strategy (MOST; for a comprehensive overview of MOST, see Collins, 2018).MOST is a framework for developing, optimising, and evaluating behavioural interventions with the goal to 'empirically identify which intervention components work and which do not work, which ones work well together, and under which contextual characteristics' (Marques & Guastaferro, 2022, p. 794).The abovementioned questions are highly relevant in relation to interventions and certainly in mental health interventions for athletes, but are currently unanswered.A better understanding of mediators will help researchers (and practitioners) design more effective interventions, thereby also enhancing intervention success.
Finally, a call for high-quality intervention studies is not new, but specific suggestions for raising the quality of intervention research in sport psychology are rare.One suggestion has been the use of frameworks such as MOST (Collins, 2018).Another suggestion is to implement high-quality single-case experimental designs (SCED; Tate et al., 2013).SCEDs have previously been put forward as a promising approach for mental health intervention research in athletes (Ekelund et al., 2022).For psychological interventions targeting mental health specifically, a single-case multiple-baselines design (MBD) across participants offers many attractive and useful features (Kazdin, 2021).Single-case MBD across participants can achieve high internal validity and experimental control, and can incorporate randomisation to further reduce threats to internal validity (Kratochwill & Levin, 2014).
Typically, in an MBD across participants, randomisation is achieved by randomly assigning the order in which individuals receive the intervention.In a study with three individuals, for example, each participant would be randomly assigned to one of three staggered starting points.In other words, they would be randomly assigned to start the intervention first, second, or third (Kazdin, 2021).The intervention effect is thereby demonstrated by introducing the independent variable (i.e. the intervention) to different (multiple) baselines at different points in time (Kratochwill & Levin, 2014).
Furthermore, mediators could be incorporated into the design, allowing for close and continuous inspection of changes and fluctuations in the outcome variables of interest and creating an understanding of the process of change throughout the intervention.Hence, interventions using single-case experimental designs could be used to unlock the black box and provide a detailed outline of the link between intervention components, mechanisms of change, and the outcome variables.

Limitations
The current scoping review is not without limitations.First, the search was restricted to English language peer-reviewed papers and grey literature was excluded.Although restricting the language may have introduced bias in the sample of studies, the magnitude of the bias is unknown, and evidence suggests that language restriction may have little influence on the results (Dobrescu et al., 2021).Second, this review obtained a large number of articles in the electronic searches.The onerous task of screening over 26,000 articles was supported by a new machine-learning tool, ASReview.Although ASReview reduced the number of articles that needed to be screened, the system's error rates are impossible to evaluate without labelling the full dataset.Hence, the extent of the unintentional exclusion of articles is unknown.Nevertheless, ASReview, is an open-access, transparent, easily accessible timesaving tool that supports current and future reviews.The alternative, screening all articles by hand, would be immensely time-consuming and present a tangible risk of human errors.
Third, this review is also limited by the scattered results.Though only 44 studies were included, the challenging synthesis of their results reflects the current status of mental health interventions in the field.For this reason, this review cannot provide evidence-based recommendations for practice.Fourth, we aimed to answer the question of what mediators have been studied and, thus, identify mechanisms of change.However, this was not possible due to the almost complete absence of studies that included mediators.Consequently, we cannot make recommendations about effective mechanisms for inclusion in future interventions.Instead, we can only encourage future investigation of putative mediators and prompt researchers to address the critical knowledge gaps we have identified.

Conclusion
This review shines light on intervention research that has emerged mainly in the past 15 years and identifies several important issues for future consideration.The synthesis of findings suggest that the current literature on mental health interventions for athletes is inconclusive.Consequently, few evidence-based recommendations can be made and, to date, the effective components of mental health interventions for athletes are unknown.Although some promising themes have emerged (e.g. the use of established theories or models such as cognitive behavioural therapy), there is room for improvement, particularly regarding study quality; the operationalisation of mental health; and the proper use of promotion, prevention, or treatment of the target population's mental health.In conclusion, there is a pressing need for high-quality intervention studies, especially those that evaluate mediators in subclinical or clinical samples.As a result, the field of sport psychology will be more equipped to intervene for athletes' mental health when and where necessary.

Disclosure statement
No potential conflict of interest was reported by the author(s).
in one study.Furthermore, 24 interventions were group-based, 10 studies conducted interventions on individual bases, eight studies used a combination of group-based and individual work, and two studies did not clearly indicate whether their respective interventions were group-based or individual.Regarding intervention facilitator, a wide variety of actors were used (e.g.clinical psychologist, master level psychology student, doctoral students, mental performance consultant, resident physician, sport psychologist, athletic trainer).Regarding the length of the interventions, group-based interventions ranged from one 90-minute session

Table 2 .
Descriptive summary of definitions, theories, frameworks, and models.