Background

Athletes have traditionally been portrayed as individuals who possess an abundance of human strength and mental toughness [1]. The reality is, however, that athletes struggle with mental health needs in a manner similar to the general population [2]. In some cases, higher prevalence rates of mental disorders are evident when athletes are going through periods of transition or adversity through injury, deselection, burnout, or prolonged competetion periods spent away from family and friends. Sport coaches are often viewed as a conduit to, or gatekeepers of, athlete mental health promotion [3]. However, coaches show similar mental health disorder prevalence rates as athletes [4] and, indeed, report the existence of unique  stressors mostly related to an imbalance between the coach’s ability to self-manage their mental health and meet the demands of the role [5]. Moreover, sports officials (e.g. referees, judges) face substantial adversity and harmful stressors which can include verbal abuse and aggression from athletes, fans and the media, that few feel equipped to deal with [4]. Therefore, mental health awareness training and the provision of suitable psychological and psychiatric support services are required for athletes, coaches and officials [6, 7].

In 2017, we published a systematic review of mental health awareness interventions delivered to sport clubs [8]. Ten interventions programmes were included that aimed to increase mental health literacy and support athletes, coaches and officials experiencing a mental health problem. While some support was found for the effectiveness of programmes in enhancing mental health awareness and help-seeking, few showed rigorous methodological quality, and most suffered a high risk of bias. None of the studies followed standards for reporting trials or referred to the Medical Research Council process evaluation framework [9]. Furthermore, few studies were underpinned and/or tested with a psychological theory of behaviour change and health. Michie et al. [10] have presented strong evidence that the integration of theory provides a clearer understanding of the causal assumptions underpinning intervention outcomes and provides a systematic evaluation framework to understand how and why interventions are effective in practice. Hence, our recommendation was for programme designers to give due consideration to the integration of behaviour change theory in the development and analysis of programmes. Moreover, we concluded that longitudinal studies are required with larger sample sizes of males and females, wherein randomisation to groups is blinded, and outcomes are measured with validated measurement tools [8].

Since the initial search, there has been a proliferation of research in the area of mental health and sport, evident in systematic reviews [2], peer-reviewed journal special issues (Journal of Physical Education Review, 2020) textbooks [7, 11, 12] and mental health–themed conferences (European Congress in Sport Science, 2017, British Psychological Society, Division of Sport and Exercise Psychology, 2018, 2020). Some Government and leading sport associations have developed mental health and wellbeing action plans or consensus statements to safeguard athlete mental health [2, 13, 14]. More recently, an emphasis was placed on supporting all those participating in sport, through a call to action to move beyond only supporting the elite athlete [15]. An international consensus statement was also published describing that mental health awareness programmes should be available for all involved in sport (i.e. athletes, coaches, officials, parents), that programme content should be theory-based and evidence-informed and include robust evaluation [6].

With respect to such developments in the mental health in sport field of study, Garner and colleagues [16] have outlined that newly identified studies can potentially change conclusions and recommendations of a previous review. Given systematic reviews are central to healthcare science, and inform practitioners, intervention and policy development, those involved in design and implementation are not fully informed by the latest research. Furthermore, outdated reviews do not capture novel theoretical developments and/or topical issues where further research may be imminently needed. Given the upsurge in research in mental health awareness raising, our first aim was to update and extend our original systematic review conducted determining the effect of mental health awareness programmes to improve mental health knowledge and help-seeking among sports coaches, athletes and officials [8]. Using Garner and colleagues [16] consensus and checklist for updating systematic reviews, we replicated and extended the original review to include athlete parents. Interventions targeted at improving the mental health awareness of parents are important as they target three key aspects of mental health literacy that will allow the parent to provide the optimal support for their children who are participating in sport [17]. Namely, enhanced symptom recognition can allow the parent to recognise ‘warning signs’ of key mental health disorders to allow them to provide support for their child should they become affected. The ability for a parent to effectively help their child with a mental health issue has been shown to be greatly influenced by their attitudes toward mental health. Efforts to reduce any stigma among parents of athletes could be greatly beneficial for reducing barriers to help-seeking in their children. Finally, knowing how and where to seek appropriate information on mental health disorders and treatment options [17]. Not knowing what to do, or where to turn for help has been identified by parents as the most common barrier to facilitating help-seeking behaviour and highlights how improved mental health literacy knowledge and signposting could be of particular importance [18]. In reviews to date, the role of the parent has not been included. The second aim was to review newly retrieved study quality and report on the validity of measures that were used to determine the effectiveness of programmes. A description of intervention programmes delivered are provided and recommendations for those in the process of designing and evaluating mental health programmes for athletes, coaches, officials and parents are proposed.

Methods

Protocol

All methods of data analysis and reporting followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [19]. Amendments to the original PROSPERO protocol (International database of prospectively registered systematic reviews in health and social care) were included in December 2021 and can be accessed (Registration number: CRD42016040178). A PRISMA checklist is provided as a supplementary file.

Eligibility criteria

Types of studies

Randomised or clustered randomised controlled trials and quasi-experimental studies that did not use a pre-specified randomisation processes when selecting the treatment and comparator condition [20] were included. Studies comparing the treatment with a comparison group, more than one intervention group or within subjects across time (i.e. pre-post testing) were included. Studies were required to have been published in the English language. The decision was taken to restrict our inclusion criteria to only peer-reviewed literature as grey literature (e.g. dissertations, reports, policy documents) is heterogeneous, and little methodological guidance exists for the systematic retrieval, analyses and reproducibility of such work [21].

Types of participants

Participants were children, adolescents or adults who are considered an athlete, leader, coach, parent, official or member (e.g. service provider) within a professional, semi-professional or amateur community sporting club or organisation.

Types of interventions

Mental health interventions that took a general approach to improving awareness of mental health (e.g. help-seeking, knowledge of disorders, literacy), or interventions tailored to improve mental well-being (e.g. positive affect, life satisfaction), or reduce symptoms of mental ill-being (e.g. depression, anxiety) were included. While eating disorders are a relevant topic for mental health awareness programmes, we decided to exclude these studies because several recent systematic reviews focus on the specific nature and implementation of eating disorder prevention initiatives for athletes [22, 23]. The mode of delivery was individual, group or web-based. To be eligible for inclusion, interventions had to take place within a sport setting (i.e. sport club, sport environment), or be focused for athletes, coaches, officials, parents or service providers. As many definitions of sport exist, we applied Rejeski and Brawley’s [24] definition for consistency: structured physical activity that is competitive, rule-governed and characterised by strategy, prowess and chance. Exclusion criteria applied to interventions that were considered being outside the domain of sport (e.g. physical activity, exercise, leisure, art, dance and music).

Types of outcome measures

Studies needed to include at least one outcome measure which we categorised as related to mental health attitudes (e.g. stigma, prejudice), knowledge of mental health (e.g. disorder/symptom recognition), or behaviour regarding mental health (e.g. intended/ actual help-seeking for oneself or others); mental health–promoting competencies/skills (e.g. mindfulness, coping), or specific mental health (e.g. anxiety, depressive symptoms) and/or well-being (i.e. subjective/psychological well-being domains, life satisfaction) outcomes. Only quantitative studies were included as it would be difficult to assume a level of generalisability between quantitative and qualitative outcomes. Furthermore, a qualitative review could be reported as a separate article.

Information sources and search strategy

We used electronic databases and also manually checked reference lists of articles. Five electronic databases were searched: PsychInfo, Medline (OVID interface), Scopus, Cochrane and CINAHL. Each database was searched (see Table 1) from its year of inception to July 2020. Search terms used keywords, truncation and MeSH terms as appropriate for each database’s indexing reference [25]. The search was stratified into four categories: sport, participants, setting and method of treatment. Search terms were the same as the original review and chosen based on previous research, theory and practice. The first category used sport as a single term as sport is central to the objective of the review. As with previous systematic reviews in sport [26], the second category used descriptors associated with participation or membership within sport. The third category depicted broadly cited sport settings in sport development literature [27] and also included internet-based terms to account for recent developments of online mental health interventions [28]. Lastly, search terms in the fourth category were applicable terms to constructs associated with mental health and well-being [29], mental health knowledge [30] and competence strategies appropriate for mental health interventions [31]. A full electronic search of the Psycinfo search is uploaded as a supplementary file. Reference lists of included articles were also searched.

Table 1 Search terms used in Psycinfo search reflecting keywords, mesh terms and suffixes
Full size table

Study selection and data collection process

Study selection was completed in three phases. First, database searches were exported to Refworks software into a master folder. All titles and abstracts were screened by one researcher. Duplicates were removed and all abstracts were exported to a subfolder (i.e. included for follow-up or excluded). All relevant abstracts were printed and screened by a second and third researcher to assess their eligibility for full-text printing and screening. Second, to ensure inter-rater reliability two researchers independently screened 10% of all excluded titles and abstracts. Although a high level of agreement (>95%) was reached, potentially relevant abstracts were highlighted and subsequently screened by two authors using the inclusion criteria. They were found to be irrelevant and were excluded. Third, full-text eligibility assessment was performed independently in an un-blinded standardised manner by three researchers (GB, SS, MC) using the screening tool (see Fig. 1). The remaining included articles were divided between three researchers and all pre-defined data (see below) was extracted by one researcher and cross-verified by a second and third for the synthesis of results.

Fig. 1
figure 1

Screening tool for independent author screening

Full size image

Data items, summary measures, synthesis and analysis of results

Detailed descriptive information from each intervention including the author(s) and year of study; study design features (e.g. data collection points, inclusion of a control group or not); sample characteristics including age and gender; mode of delivery and mental health descriptor (i.e. increase knowledge, improve attitudes or reduce depressive symptoms) were extracted. For assessing the effect of the interventions we obtained the name of the outcome measure(s), reported value(s) for intervention effectiveness (e.g. p value, effect size) and, based on prior research [32], provided a narrative commentary on study design methods that may influence the generalisability of study effects. As all of the outcomes measured were derived from psychometric scales, we observed statistically significant quantitative effects on the basis of p < .05 [33], and a small, medium or large effect size as d = .2, .5 or .8, respectively [34]. We reported the effects of each study in Table 3. For combining and reporting the results, we inspected each study’s outcomes and categorised them in accordance with the following key mental health constructs [35]: stigma, mental health knowledge, referral efficacy/confidence, help-seeking intentions and behaviour, well-being and additional outcomes.

Risk of bias within and across studies

For profiling, the study quality and risk of bias the principles of the Cochrane Collaboration for assessing methodological quality in systematic reviews were adopted [20]. As included studies were either categorised as randomised or non-randomised designs, each study’s design was matched with an applicable assessment of bias tool. For randomised controlled trials we used the Cochrane Collaboration’s tool for assessing risk of bias [36]. The tool includes six domains of bias such as selection, detection and reporting bias. Each domain is coded as high, low or unclear for the relative risk of bias and an overall judgement is accumulated. For non-randomised studies, we used the ‘Quality Assessment Tool for Quantitative Studies’ (QATSQ) [37] that is recommended for use in systematic reviews [38]. The QATSQ tool is scored based on six domains of bias including selection bias, confounding bias and withdrawals and dropouts. Based on the pre-defined bias criteria, the domains were scored as either weak (3), moderate (2) or strong (1). Studies with no weak ratings and at least four strong were considered strong, while studies with fewer than four strong ratings and one weak rating were considered moderate, and studies with two or more weak ratings were considered weak [37]. Based on the Cochrane Collaboration’s recommendations [36] we reported on the risk of bias across studies by summarising the cumulative bias for each outcome in the Cochrane and QATSQ tools. To facilitate reporting of bias across the studies, additional rows and columns were added to the tools.

Outcome measures were also assessed for validity as they can influence the generalisability of study findings [33]. The study adapted criteria used in a recent systematic review of mental health interventions [28] and also used in the original review [8]. Scales were considered acceptable if they met one or more of the following: a Cronbach’s alpha value of above .7; reporting of acceptable goodness of fit indices using confirmatory factor analysis [39], test-retest, construct or concurrent validity assessments; and/or the authors referenced a previous study that validated the scales through the above methods.

Results

A total of 2048 titles and abstracts were reviewed (See Fig. 2). One further article was identified through a co-author’s knowledge of the area of research. After removal of duplicates (n = 188), 1861 titles and abstracts remained. Of these, 1665 were identified as not meeting the eligibility criteria and were excluded. A total of 196 articles were identified as eligible and therefore underwent a further detailed screening, 33 articles met the criteria for full-text screening by two researchers. Of the 33 articles, researchers agreed upon six articles to be excluded because they did not meet the inclusion criteria on at least one aspect. Two of the articles were related to a study that was already included in the quantitative synthesis [40] that included a trial registration and a book chapter providing a description of the aforesaid intervention delivered within the study. Three articles were removed on the basis that they were tailored toward sport performance–related outcomes (psychological skills training) rather than mental health awareness and therefore fell outside of the scope of the review [41, 42]. One study was deemed ineligible as the participants were not considered to be involved in sport [43]. The remaining 27 studies [8, 18, 44,45,46,47,48,49,50,51,52,53,54,55,56,57,58] achieved 100% researcher agreement for their inclusion for review, 10 of these articles were those included in the original systematic review [8] and underwent quantitative synthesis [59, 60, 62,63,64,65,66,67,68,69]. A further 15 references were identified upon hand-searching the reference lists of the 27 included articles, wherein one further article [70] met the inclusion criteria. However, 14 were excluded from any further data synthesis as they fell under the category of chapters in books, cross-sectional surveys or contained qualitative findings (see Fig. 2).

Fig. 2
figure 2

PRISMA 2009 flow diagram

Full size image

Study characteristics

Study characteristics are detailed in Table 3. Across the 28 studies, 4657 participants took part, consisting of 1234 males and 1113 females. Three studies did not detail subject gender, including 1004 [57], 995 [68] and 311 [65] participants respectively. The interventions were delivered to a variety of sports participants. Half of the studies within the original systematic review targeted coaches and service providers (n = 5), others focused on elite and non-elite athletes (n = 3), one was aimed at officials (n = 1) and another ‘at-risk’ children (n = 1). The newly synthesised articles shifted toward studies that specifically targeted athletes (n = 16), collectively making athletes the most researched subgroup (n = 19). Two studies focused on parents of sport participants exclusively (n = 2). One study detailed a multicomponent intervention that included adolescent athletes, parents of athletes and coaches (n = 1). No further studies were found that investigated interventions in officials. Studies deployed a variety of designs including intervention pre-post testing (n = 10), randomised control trial (n = 8), controlled trial (n = 7), quasi-experimental study (n = 2) and a descriptive case trial (n = 1). The mode of delivery for the majority of studies was via group setting (n = 23); however, many of these interventions also included a blend of online and home-based elements. An individual counselling format (n = 2), web-based (n = 3) and home (n = 1) settings made up the remainder.

Study results

The name of the author(s) who conducted the study, the year, the design, study duration, sample characteristics, mental health descriptor employed, mode of delivery, mental health outcome measure(s), main findings and general comments regarding each study are summarised in Table 2. Studies selected for inclusion were published between November 1999 and May 2020.

Table 2 Descriptive information for included studies
Full size table

Twenty-one studies utilised a control group [8, 18, 40, 44, 45, 47, 49,50,51,52,53,54,55,56,57, 60, 63, 64, 66, 69, 70] of which nine implemented randomisation methods [40, 44, 45, 47, 51, 54, 55, 63, 69]. Sample sizes significantly varied, from the extremely large (n = 1004) [57] to small (n = 3) [67]. Nine studies collected outcome measures pre- and post-intervention and obtained measurements at a further follow-up point in time, i.e. [18, 40, 44, 46, 47, 51, 62, 63, 66]. Three of the studies from the original review did not report a statistical test of significance [62, 67, 68], all further studies included values for statistical significance testing. A total of 13 studies ( [40, 49, 55, 57,58,59,60, 62, 65,66,67,68, 70] did not report effect sizes. Collectively, findings across the studies support the importance and efficacy of introducing knowledge-based mental health programmes in sport settings. Of the studies that included a follow-up, the majority maintained some of their effects (n = 8), while one did not (n = 1) [66], showcasing the potential long-term efficacy of such interventions. The impact of the interventions on the multitude of targeted outcomes is detailed below.

Effects on stigma

One intervention elicited a reduction in stigma around anxiety, yet stigma for depression remained unchanged [63]. Stigma for depression was significantly reduced in another study [59]. Two studies reported on stigma for socialising with individuals with a mental health disorder, one study showed a significant reduction [60] while the other did not [57]. However, those who completed the entirety of the Vella et al. programme reported a decrease in attitudes to stigma. The same research team in the Liddle et al. study examined stigmatising attitudes and showed that they were significantly reduced post-intervention, with changes sustained at follow-up [40]. In the Chow et al. [46] study, improvement in self-stigma of seeking help was improved and was sustained at follow-up, conversely, personal, public and implicit stigma did not differ significantly between pre- and post-intervention.

Effects on mental health knowledge

Eleven [8, 18, 40, 46, 57, 59, 60, 63, 65, 66, 69] of the twelve studies reporting on mental health knowledge, showcased a significant rise in aspects of mental health knowledge, attitudes toward mental health, disorder-specific recognition and mental health referral knowledge. The study by Hurley and colleagues [52] displayed no significant improvements for the intervention group in comparison to control in mental health literacy outcomes. Six studies deployed the Anxiety Literacy Questionnaire (A-LIT) [63] and Depression Literacy Questionnaire (D-LIT) [71]. There was substantial heterogeneity in the further six studies’ assessment and conceptualisation of mental health knowledge (see Table 3).

Table 3 Study outcome measures, main findings and comments on study
Full size table

Effects on referral efficacy/confidence to help someone with a mental health problem

Confidence to provide help or to successfully refer an individual suffering from a mental health issue was increased in six studies [18, 52, 59, 60, 65, 66]; however, each of these studies deployed measurement tools that have not been psychometrically validated. One study observed significant positive changes for mental health referral efficacy [69] and utilised a validated scale. One study [40] did not observe any significant change in confidence to provide help.

Effects on help-seeking intentions and behaviour

Three studies [8, 40, 60] observed an increase in intentions to offer help to those with a mental health problem, although in the Liddle et al. study, intentions were not shown to be sustained at a later follow-up period. While Gulliver et al. [63] did not see improvements in intentions to seek help for oneself, two studies reported personal help-seeking improvements [18, 46]. In one study, participants favoured formal help [52] while in another informal help was preferred [57]. In concordance with the findings in the original review, actual behaviour change was not reported in any of the studies.

Well-being and additional outcomes

Sixteen studies reported improvements in at least one well-being outcome, with six enhanced overall mental health and well-being, six leading to a decrease in stress and four reducing symptoms of anxiety. In contrast, null effects were reported for distress levels [18, 57, 58] well-being and resilience [8] and depression and quality of life outcomes [50]. Two of these studies did not report statistical tests for significance or effect sizes, thus, restricting the interpretation of findings [62, 68]. Other outcomes that were assessed showed improvements, such as emotional intelligence [45], coping [50, 53] and mindful awareness [54, 70], mental toughness [72], relationship domains [62] and substance abuse [62], but were not confirmed with statistical tests for significance.

Risk of bias assessment

Risk of bias assessment for the two randomised studies is presented in Table 4. The two studies using randomisation methods demonstrated a low [63] and unclear [69] risk of bias. There was no high risk of bias scored for any of the domains across the two studies. Information was not provided on selection, performance and detection bias in [69], giving the design an overall judgement decision as unclear. Across the studies, bias was mixed for random sequence generation, allocation concealment and blinding of participants with [63] scoring low on those domains and [69] scoring unclear. Collectively, bias was unclear for blinding of outcome assessors and both demonstrated a low risk of bias for (a) missing data, (b) selective reporting and (c) other biases.

Table 4 Risk of bias for randomised studies using Cochrane risk of bias tool
Full size table

Risk-of-bias assessment for each of the randomised controlled trials is presented in Table 4. Three studies [40, 47, 63] demonstrated a low risk of bias. For three of the studies [44, 51, 69] the risk of bias was deemed to be unclear as there was insufficient information provided on selection, performance and detection bias. One study was deemed to be of a moderate risk of bias [55] as a lack of blinding raised the risk of performance and detection bias. An invalid method of random sequence generation and non-blinding forced one study [45] to be adjudged as high risk of bias. Collectively, the risk of selection, performance and detection bias findings were mixed. The risk of attrition and reporting bias was generally low.

Risk-of-bias for each of the non-randomised studies is depicted in Table 5. Twelve studies were judged to have a weak study quality. Nine studies were characterised as being of moderate quality as a result of having one domain that was considered to be weak. A lack of clarity on blinding of assessors and participants was weak in 5 of those moderate quality studies. High attrition rate or unclear disclosure of dropout rate accounted for the weak domain in the remainder of the moderate quality studies (n = 4). Each of the non-randomised studies was of strong or moderate quality in controlling for selection bias, study design and data collection methods. A mix of quality was seen in the confounder and withdrawal domains, with nine and seven weak studies respectively. Sixteen of the non-randomised studies were of weak quality for blinding while the remaining five had moderate ratings.

Table 5 Risk of bias for non-randomised studies using the Quality Assessment Tool for Quantitative Studies (QATSQ) tool
Full size table

Outcome measure validity assessment

Sixteen studies were assessed to have acceptable outcome measures as their scales used had adequate internal consistency or referenced studies that had validated the scales used previously. Eight studies used some scales that had displayed adequate validity and reliability, while others deployed tools that had not met the predefined criteria. Five studies were deemed to be unacceptable.

Discussion

This updated systematic review analysed the recent proliferation of published studies in the field of mental health and sport, thus providing a more inclusive and contemporary reflection of the evidence base for those involved in the design, implementation and receival of mental health awareness programmes. While the previous review articulated the increasing recognition that athletes, coaches and officials in sport settings can be vulnerable to mental health problems [1, 73, 74], the identification of eighteen further studies in this updated review mirrors the wider exposure of sport and mental health evident in the media and wider public health agenda [75]. Overall, support was maintained for improving mental health knowledge and help-seeking among coaches, athletes and officials, with extensions to multicomponent programmes that included parents, athletes and coaches. However, and despite some improvements in methodological quality of the field, issues persist such as a lack of theoretical input into both programme design and analyses, and lack of long-term follow-up data collection periods.

Effects of studies on awareness outcomes

The studies that examined the effects of intervention programmes on mental health stigma produced mixed outcomes. For instance, one study revealed a reduction in stigma surrounding anxiety, however depression stigma remained unchanged [63], another study showed a reduction in stigma about their own mental health help-seeking but no significant effect was shown for public stigmatisation [46]. Liddle et al. [40] showed that stigmatising attitudes were reduced and sustained at follow-up; however, this study along with three other studies [57, 59, 60] examining stigma did not include effect sizes, which makes it difficult to ascertain the extent of intervention impact on mental health stigma.

Eleven of the twelve studies within the review that examined mental health knowledge demonstrated an improvement in at least one aspect of mental health knowledge. However, aside from the six studies that used the A-LIT [63] and D-LIT [71] questionnaires, generalisability of the other studies is difficult due to substantial heterogeneity in measurement tools used. That being said, there is definite potential to improve mental health knowledge in sport participants through awareness interventions, going forward, efforts to increase mental health literacy could be more clearly demonstrated if equivalent measuring tools were deployed across studies.

Confidence in oneself to help or successfully refer someone with a mental health problem was improved in seven studies, only one study showed no significant effect [40]. Of methodological interest, only one study deployed a randomisation procedure, reported effect sizes, maintained longitudinal effects and provided evidence for validity of outcome measures [69]. Three studies showed an increase in intentions to offer help to a person experiencing mental health problems [8, 40, 60]; however, the intentions to support in Liddle et al. [40] were not sustained at follow-up. Further research is required to identify the favoured form of help-seeking for sport participants as one study [52] showed formal help-seeking was preferred, while another [57] indicated that help-seeking from informal sources was favoured. Similar to the previous review conclusion behaviour change was not reported in any of the studies, highlighting an area that requires further design and measurement consideration.

One of the 18 additional studies [56] (a mindfulness-based intervention) deployed a behaviour change model: Self-Determination Theory [76]. The study findings supported competence-promoting processes within the intervention, indicating that engaging with mindfulness practices can confer perceptions of competence in mental health self-management with subsequent indirect effects on stress regulation, mindfulness and overall well-being [56]. While the effect sizes were small, the findings showed that future programme design could benefit from the inclusion of behaviour change models.

To assess the long-term effects of the programmes on mental health and wellbeing improvement, the inclusion of post-intervention follow-up data is crucial. While seventeen studies displayed improvements in some aspect of well-being, only five studies included post-intervention follow-up [40, 44, 47, 51, 62]. Each of these longitudinal studies evidenced sustained improvements in depression, generalised anxiety and distress at follow-up.

Methodological quality of studies

Close inspection of the studies indicates various design limitations, these need to be overcome for future development of programmes. The previous review identified a lack of overall methodological rigour and a high risk of bias among the included studies. The authors described a need for further ‘well-designed controlled trials’ [60]. While some flaws remain in a number of the 18 additional articles included in the updated review, a trend toward higher quality studies with a lower risk of bias can be observed. Five of the ten studies within the previous review did not include a control group (50%), the percentage of studies without a control group in the additional 18 articles was greatly reduced to 3 studies [46, 48, 58]. Two studies within the previous review had a randomised control trial design; the total in the updated review is now eight. However, the generalisability of several of the studies is hindered by the small sample sizes included within the review; thus, the long-term significance of these studies remains unclear. Three studies did not report gender which prevents the review from providing any further insight into gender effects on help-seeking behaviours [77, 78].

Seven of the previous eight non-randomised studies were of weak quality and one moderate. Eight of the additional twelve non-randomised studies were of moderate quality and four were weak. Collectively, three of the randomised controlled trials displayed a low risk of bias, four were unclear and one was of high risk. Risk of selection, performance and detection bias was high in several studies due to a lack of randomisation and blinding measures in certain studies. Risk of attrition and reporting bias was low. While there were some signs of improvement in meeting the predefined criteria for acceptable psychometric measurement validity, significant heterogeneity remained present. The majority of referral efficacy tools lacked validation; therefore, it is not possible to take great confidence from the effects reported.

Overall, it is clear that a higher proportion of studies have adhered to methodological guidance in the design and reporting of interventions as was advised in the previous review. Examples from the current review show that research deemed to be of weak quality, deploying a ‘pre-post intervention’ design, with no or short follow-up periods [62], tend to be extended in more recent times to have greater methodological rigour, utilised a randomised control trial design and included 8 month follow-up periods, with low risk of bias [47]. These developments are indicative of an area of research that is improving as the recognition of mental health in sport grows internationally.

Intervention delivery methods

The content of each of the programmes varied and the attendees involved were from a variety of backgrounds within a sport setting, i.e. elite athletes, coaches, club leaders, student athletes, officials, parents and those athletes who had been referred after reporting substance misuse (see Table 2). Therefore, future reviews may want to consider limiting the search to a particular group only (i.e. athletes, coaches, officials, parents or athletes considered to be at high risk).

Similarly, the frequency and duration of sessions for each programme varied greatly, for instance: an 8-h programme across three group sessions [59]; 12 separate group sessions each with a different topic; 12 individual one to one sessions; a programme that lasted 1.5 h initially then completing a home programme for 6 weeks [64]; to a 45-min one-off workshop [40]. Programmes were delivered online or delivered in groups by trained facilitators. Determining the most effective delivery method (i.e. online, one to one, in groups) and intervention duration and frequency is not possible from the current review, but could be considered a screening variable for future reviews.

Limitations and recommendations

The review was limited to studies published in English which could perhaps have forced the exclusion of data from certain parts of the world, narrowing the generalisability of the review.

As has been discussed, in spite of an overall trend toward a higher quality of research, there remain inconsistencies in the outcome measures. In addition to small sample sizes, these limitations negatively impact upon the long-term significance of study findings. This review is limited by its exclusion of grey literature. When the research was proposed, it planned for a full search of the grey literature, searching for programmes published by the government, sporting governing bodies such as the NCAA and national public health agencies. The decision was made to exclude grey literature due to the time constraints. Further research could expand on this update by incorporating insights from grey literature. Excluding any form of potentially valuable information goes against the nature of systematic reviews which aim to summarise the findings of all relevant studies [79]. Furthermore, qualitative data was not included as it would be difficult to assume a level of generalisability between quantitative and qualitative findings. Individual case study work of applied sport psychologists was identified but excluded from the review, and thus, the review may miss out on valuable expert insights.

There are several recommendations from the studies included in this review. While one study incorporated theory of behaviour change in design and evaluation of the intervention, the remaining studies did not. Application of behaviour change modalities can allow researchers to identify the motivational factors that influence the decision-making process to seeking help. With a collection of studies using theory, interventions can then be tailored to target these factors and in turn influence behaviour [80]. The inclusion of some more established theories of behaviour change have associated valid and reliable measurement tools [10], the incorporation of these tools would enhance future mental health awareness programme evaluation.

Further, the mental health of sports officials is a pressing concern. The lack of research aimed toward officials within this review provides further evidence for the ‘Call to action: the need for a mental health research agenda for sports match officials’ [4]. It is imperative that research is carried out within officiating to allow for evidence-based mental health interventions to be implemented among officials.

Initiatives that target multiple levels of influence (e.g peers, coaches, environment, systemic influences) have great potential to be more effective than more narrow research. Studies within this review have shown the potential benefits of targeting parents to create a supportive environment for their participating child [18, 52, 57]. Furthermore, coaches who manage stress effectively have been shown to be better equipped to support athletes in dealing with stress [81, 82]. Providing training for coaches in mental health awareness could contribute to changing the culture of help-seeking in sport, providing an environment in which athletes feel more comfortable to seek mental health support and services [66]. As has been alluded to, more longitudinal studies are required to assess the long-term impacts of the included interventions. Future research should seek to attain follow-up information where possible as this would be useful in future planning of intervention content.

Conclusions

This updated systematic review reaffirms the benefits and the urgent need for evidence and theory-based intervention programmes designed to increase mental health awareness to aid prevention and provide support for athletes, coaches, officials and parents who are suffering from a mental health problem. The contribution of the updated review has deduced that there has been a trend toward research of higher methodological quality and reduced risk of bias in the intervening time since the previous review was published. Therefore, greater value can be placed on the findings within studies in this updated review. However, there remains room for improvement in research quality. For instance, future longitudinal studies are required with larger sample sizes, randomisation to groups should be double-blinded and outcomes should be measured with externally validated measurement tools. Programme designers would benefit greatly from considering grounding programme content in relevant behaviour change theory to more effectively tailor programmes to the motivational needs of participants. To conclude, the findings within this review can aid the development of sport-specific programmes to increase the mental health awareness and well-being of the vulnerable, underserved sporting population and can contribute to reducing the overall burden of global mental health.