Applying a systems thinking lens to anti-doping: A systematic review identifying the contributory factors to doping in sport

ABSTRACT

The use of performance enhancing substances and methods (known as “doping”) in sport is an intractable issue, with current anti-doping strategies predominantly focused on the personal responsibility and strict liability of individual athletes. This is despite an emerging understanding that athletes exist as part of a broader complex sports system that includes governance, policymakers, media, sponsors, clubs, team members, and athlete support staff, to name a few. As such, there is a need to examine the broader systemic factors that influence doping in sport. The aim of this systematic review was to identify and synthesise the factors contributing to doping and doping behaviours, attitudes, and beliefs and the extent to which this knowledge extends beyond the athlete to consider broader sports systems. The review followed PRISMA guidelines with risk of bias and study quality assessed by the Mixed Methods Appraisal Tool, and identified contributory factors synthesised and mapped onto a systems thinking-based framework. Overall, the included studies were determined to be of high quality. Support personnel, the coach, and the coach-athlete relationship represent key influences on the athletes’ decisions to dope. From the evidence presented, doping is an emergent property of sport systems and represents a complex systemic problem that will require whole-of-system interventions. The implications for this and the focus of future research are discussed.

KEYWORDS:

• Doping
• sport
• systems thinking
• risk management framework
• AcciMap
• ActorMap

1. Introduction

The illicit use of performance enhancing substances (PES) and methods (a practice colloquially known as “doping”) in sport is an ongoing issue which has serious global implications for athlete health and well-being, and violates the spirit of sport (Momaya et al., 2015). Doping as defined by the World Anti-Doping Agency (WADA) is the occurrence of one or more anti-doping rule violations (ADRVs) as set out in Article 2 of the WADA Code (https://www.wada-ama.org/en/resources/world-anti-doping-code-and-international-standards/world-anti-doping-code). This list includes the presence of prohibited substance or its metabolites in an athletes sample, the use or attempted use of prohibited substances, evading or failing to submit to sample collection, whereabouts failures, tampering or attempted tampering with any part of doping control, possession of a prohibited substance or method, trafficking or attempted trafficking of a prohibited substance, and other violations. Various programmes of research have explored the health and performance impacts of doping, strategies for enforcement and prevention, and the perceptions of athletes and support personnel who have doped (Engelberg et al., 2015; Harris et al., 2021; Mazanov et al., 2014; Petróczi & Haugen, 2012). Whilst this body of work has advanced the doping knowledge base around doping and its prevention and management, the issue of doping continues to influence various forms of sport at both professional and amateur levels.

Despite advances in our understanding of factors which influence doping and the methods for PES detection, stringent testing protocols, deterrents, and severe punishments (including participation bans) for detected athletes and support personnel, doping in sport persists. While the World Anti-Doping Agency (WADA) reported figures for adverse analytical findings to PES of between 1.1% (2008) and 1.6% (2016), true doping prevalence is estimated to be as high as 57% in some cohorts (Ulrich et al., 2018), although this figure has been disputed in a recent large systematic review of 175 studies which noted that most studies report prevalence at less than 5% (Gleaves et al., 2021). It has been argued that merely advancing the sciences to detect PES in athletes is limited, and that education appears to be an important factor to include in prevention programmes (Gregory & Fitch, 2007; Stuart & Mottram, 2019). In addition, prevention programmes should be aimed at changing the broader societal norms around doping (Harcourt et al., 2012; Petróczi et al., 2017). The current approach to anti-doping is noticeably similar to the traditional approach to road safety whereby the focus is on the three E’s; education and enforcement of road users, and engineering of vehicles and infrastructure to improve road safety (Salmon et al., 2016). In the context of current anti-doping approaches, there are education programmes for athletes and support staff (Patterson et al., 2019), strict enforcement via bans and suspensions for anti-doping rule violations (ADRVs) (Kolliari-Turner et al., 2021), and constant engineering of new doping detection methods (Momaya et al., 2015; Anawalt 2018}. Though this approach has had some success in reducing road trauma, it is now achieving diminishing returns and it has been argued that it is no longer the only answer given the broader complex societal issues that permeates into road transport systems. This appears analogous to efforts in doping, where the current approach may have had some success in detecting and preventing doping (although this is contentious – Houlihan & Vidar Hanstad, 2019), yet has failed to deal with a set of more resilient doping issues.

Whilst research into doping has undoubtedly assisted to help reduce doping in world sport, the use of PES by athletes seeking to gain competitive advantages remains a persistent issue. Prior work exploring the personal and psychological predictors of doping in physical activity settings has identified positive attitudes towards doping, and perceived social norms as correlates of doping intention and behaviour (Ntoumanis et al., 2014). Conversely, perceived morality and self-efficacy provide the strongest protective correlates against doping. Backhouse et al. (2016) examined the psychosocial factors which influence doping in a mixed-evidence synthesis, with one key theme to emerge being that doping exists in a complex web of interactions between sociodemographic and psychosocial factors. Further, critical incidents (such as injuries or decrements in performance) are likely to increase doping vulnerability, as are perceptions of reference groups (including coach, family or peers), and the perceived legitimacy (or lack thereof) of the current detection-deterrence-punishment system policies (Backhouse et al., 2016). The existing evidence on the predictors of doping suggests that there is a complex interaction of individual (e.g., psychological factors) and broader systemic (e.g., coach and teammate dynamics, testing policies) factors which influence individual athletes’ decision to dope (McLean et al., 2023).

In examining the progress into doping prevention, underlying contributory factors have been recognised by WADA (e.g., World Anti-Doping Agency, 2013), wherein it was concluded that doping in sport relates to human, broader environmental, and political factors. Examples here include latent socioeconomic factors, cultural attitudes, and social pressures that promulgates the epicentre of doping behaviours in competitive sports. This suggests that doping is an emergent property of a complex and dynamic sociotechnical system comprising multiple stakeholders ranging from athletes, coaches and support personnel to sports clubs, governing bodies, doping authorities, and sponsors to name only a few (McLean et al., 2023). In recent times there have been continuing efforts to examine broader systemic factors related to doping including understanding the dopogenic environment (Backhouse et al., 2018), the conceptual model of systematic doping behaviour (Johnson, 2011), and influence of agents and sponsors (Shelley et al., 2023). From a complex systems thinking perspective, this suggests that doping cannot be understood or adequately prevented without first understanding what this system comprises and what factors interact to create the issue of doping (Ottino, 2003). This form of systems thinking is popular in other problem spaces (Nayak & Waterson, 2016; Salmon et al., 2010), and, despite currently receiving increasing attention in the sports injury context (Bittencourt et al., 2016; Hulme, Thompson, et al., 2019), does not seem to have been adequately applied in the area of doping.

In short, there is a need to better understand the relationship between the broader sport system and athlete that influence doping in sport. As a first step towards understanding the complex and interrelated set of factors which create the issue of doping in sport, the aim of this systematic review is to identify and synthesise the actors involved and factors contributing to doping and doping behaviours, attitudes, and beliefs reported in the peer reviewed literature and determine the extent to which they extend beyond the athlete to encapsulate factors across broader sport systems.

2. Methods

2.1. Protocol

The systematic review process followed the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) 2020 guidelines and flow chart (Page et al., 2021). PRISMA is a widely used standard that establishes the minimum set of evidence-based reporting guidelines that ensure the literature review process is systematic, comprehensive, transparent, and replicable. For each of the included studies, the following information was extracted:

• the type of study and year publication;

• the country and demographic characteristics (where applicable) of the participants investigated; and

• the factors which were found to contribute to doping, and positive doping behaviours, doping attitudes, or doping beliefs.

The literature search was performed across four databases (Table 1). To ensure that a rigorous standard for the evidence which was included, the systematic review incorporated only databases containing peer-reviewed, published journal articles (PubMed, Scopus, SportDiscuss, and Web Of Science).

Table 1. Keywords, initial search results, and search results following automated and manual duplicate removal for each database searched.

Database	Initial search results	Search results following duplicate removal
PubMed	595	478
Scopus	508	484
SportDiscus	3656	3222
Web of Science	2142	1634
Total	6901	5818

2.2. Electronic search

The initial search was completed on 1st of March 2022 by MN. Potentially eligible studies were published between 1 January 1999 and 1 March 2022, inclusive. This date range was chosen as WADA was established in 1999, and to ensure that the review was as comprehensive as possible. The year 1999 and the introduction of WADA represent an obvious starting point as prior to 1999 anti-doping regulations and sanctions were disparate, with each overarching sporting organisation having differing rules and different sanctions. To ensure compliance with PRISMA guidelines, citation software was used (Endnote X8, Clarivate Analytics, Pennsylvania, United States), and at each stage the inclusion or exclusion of studies was tracked via custom spreadsheet. Justification for the removal of ineligible items was noted during the full-text screen and final selection process.

Boolean search terms were generated and used which were derived from the primary research question and adjusted for each specific database (Table 1). These search terms were “anti-doping rule violations” OR “doping” OR “performance-enhancing drugs” OR “performance-enhancing*” OR “steroids” AND “sport*” OR “sport performance”. The specific search terms ensured that all relevant studies were included in the initial search before being refined. Results of the initial search and following removal of duplicate entries is presented below (Table 1).

2.3. Eligibility criteria

The authors defined inclusion and exclusion criteria a priori to support the identification of studies relevant to the current research aims and rationale. Accordingly, as this review aimed to identify contributory factors to doping and positive doping behaviours, attitudes, and beliefs, the search strategy and scope of inclusion was deliberately broad to capture factors at both systemic (e.g., team environment, medical, regulatory/policy) and the individual athlete levels of doping. Given the aim was to identify contributory factors at different levels of the sports system, studies which sampled other stakeholders (e.g., medical personnel, coaches, performance staff) alongside athletes (with respect to the doping-related criteria mentioned above) were eligible for inclusion.

Included studies complied with the following criteria:

1. The study investigated doping or doping behaviours/attitudes/beliefs amongst athletes, and between athletes and related stakeholders (e.g., medical personnel, coaches, performance staff).

2. The study primarily focused on the individual, situational, environmental, and/or systemic factors and characteristics that contribute to doping or behaviours/attitudes/beliefs amongst athletes.

3. The study was published in the peer reviewed literature and had full-text versions which were available in English.

Studies were excluded based on the following criteria:

1. The study focused on doping prevalence, or the development, implementation, and/or adoption of specific doping analytical techniques.

2. The study focused on individual-athlete psychological, socioeconomic, or demographic factors in the absence of situational, environmental, and/or systemic factors and characteristics.

3. The study focused on aesthetic athletes (e.g., bodybuilders) or the general gym-attending public not subject to the WADA Code.

4. The study focused on protective factors against doping (e.g., education, enforcement, religion).

5. The study was an opinion paper, a theoretical or simulation-type paper, not peer reviewed (e.g., was published as conference article or in the grey literature), and/or only available in languages other than English.

Following the initial search and removal of duplicates (Table 1), a search of study titles and abstracts from all retrieved studies was performed against the eligibility criteria by MN (see above). Following this, the full-text of the remaining studies were screened by MN. Additional screening, including the quality appraisal of studies (see below), was undertaken independently by SM. The rationale for the removal of studies that were removed during the full-text screen was noted (Figure 2).

2.4. Quality appraisal of the included studies

Due to the mixed nature of the designs of the included articles (i.e., qualitative/quantitative, observational/experimental) it was determined that risk of bias and quality appraisal should be undertaken using the Mixed Methods Appraisal Tool (MMAT; version 2018) (Hong et al., 2018). The MMAT was developed to assess the quality studies that use different study designs (qualitative, quantitative, and mixed methods) in a single tool (Hong et al., 2018). An overall rating for each article is calculated by summing over the five criteria for the MMAT (1 = yes, 0 = no, or can’t tell), with the two initial screening questions (S1 and S2) not being scored. Quality appraisal using the MMAT was undertaken by the lead author (MN). To ensure that this process was accurate, and bias is minimised, co-author (SM) completed the MMAT on a randomly selected 10% subset of the included studies independently, with inter-rater reliability calculated as a percentage rate of agreement (Gwet, 2008).

2.5. Mapping contributory factors onto Rasmussen’s Risk Management Framework (RMF)

Rasmussen’s (1997) Risk Management Framework (RMF) (Rasmussen, 1997), and associated ActorMap and AcciMap methods are state-of-the-art complex systems thinking-based accident analysis theory and methods, and have been used extensively in safety critical domains (Hulme, Stanton, et al., 2019; Salmon, Hulme, et al., 2020; Waterson et al., 2017). For example, the RMF and AcciMap method have been used to better understand the systemic influences on cycling crash contributory factors (Salmon et al., 2022), community rugby concussions (Clacy et al., 2019), food supply contamination (Cassano-Piche et al., 2009; Nayak & Waterson, 2016), and outdoor activity incidents (McLean et al., 2022; Salmon et al., 2017). According to Rasmussen’s RMF, complex work systems consist of various hierarchical levels (e.g., governments, regulatory bodies, organisations, medical staff, management, athletes, and equipment and environment), which each containing stakeholders and actors who share responsibility for the performance of the system. Therefore, decisions and actions which are (or are not) taken by actors at all levels within a systems hierarchy influence the overall behaviour of a system.

The RMF also describes the critical need for communication and feedback across the system as well various pressures (e.g., financial, workload, performance) that influence behaviour and create migration of practices, both at the systems level and across all levels of the system. As such, as a framework for understanding issues through a complex systems thinking lens and applying the RMF to doping, the act of doping and ADRVs would be influenced by the decisions and actions of multiple actors within the system and not just the athletes and support staff committing ADVRs. Further, stakeholders at all levels of the system operate under various pressures; for example, athletes and coaches often perform under financial and performance pressures as do sport’s governing bodies and anti-doping agencies. According to Rasmussen (1997), poor communication and migration of practices will combine to create risks which if not properly identified, assessed, and managed, can lead to system failures. In the context of the anti-doping system and sport, this may lead to doping, and ADRVs. The key implication is that it is not possible to understand doping by focusing on the athlete using the banned method or substance only. Rather, it is the decisions and actions of actors at all levels within the broader doping/anti-doping system that contribute to doping and are therefore of interest. Given this assertion, and recent extensive reviews which have explored the psychological and behavioural predictors to doping (Backhouse et al., 2016; Ntoumanis et al., 2014), and doping intension and behaviour predictors (Blank et al., 2016), these within individual factors were not considered unless there was interaction with other actors within the RMF.

In the current review, the RMF was modified to include an International level which is in line with prior work in road safety (McIlroy et al., 2019), healthcare (Salmon et al., 2021), running injury (Hulme et al., 2017), and rugby (Clacy et al., 2019), to fit with the anti-doping context. A truncated RMF example adapted from McLean et al. (2023) is presented in Figure 1, along with examples of actors at each of the hierarchical levels involved in the “anti-doping system”:

1. International influences. The highest level of the framework includes international bodies who have an influence on doping/anti-doping in sport, such as the World Anti-Doping Agency, International governing bodies (e.g., World Rugby, FIFA, FINA), and the International Olympic Committee (IOC).

2. Government and government bodies. The second level includes the government and other governing bodies who have a role to play in doping/anti-doping within sport, including the Therapeutic Goods Administration, national governing bodies, and national governments.

3. Regulatory bodies and associations. The third level includes regulatory bodies and associations who have a role to play in doping/anti-doping within sport, including the media (legacy and social), insurers, advocacy groups, and research groups.

4. Teams and organisations. The fourth level includes the team and organisational actors such as sporting teams/organisations, players union/associations, and sponsors.

5. Direct supervisors, management, medical and performance personnel. The fifth level includes actors whose role is to act in a supervisory manner to the athletes, including team doctors, sports scientists, dieticians and nutritionists, and team management.

6. Athlete, teammates, and opponents. The sixth level of the framework includes the training and competition situational environment and the decisions, actions, and characteristics of different actors within that environment (e.g., athletes, their opponents, teammates, and fans).

7. Equipment, substances, methods, and environment. The lowest level of the framework includes the immediate doping environment along with banned substances, methods, and whereabouts violations.

Figure 1. An example of Rasmussen’s Risk Management Framework (RMF) in anti-doping in the Australian team sport context (adapted from McLean et al., 2023).

2.6. Identification of contributory factors

Contributory factors are defined as factors “that, if it had not occurred or existed at the relevant time, then either the occurrence would probably not have occurred, adverse consequences associated with the occurrence would probably not have occurred or have been as serious, or another contributing safety factor would probably not have occurred or existed” (Australian Transport Safety Bureau, 2008, para. 9). Contributory factors were identified based on whether they increased or were associated with an increase in doping occurrences, the risk or likelihood of doping, or doping behaviours, doping attitudes, or doping beliefs. Contributory factors were extracted from the included articles and placed at the relevant level of the RMF with factors grouped thematically, where appropriate (Salmon et al., 2022). For example, contributory factors related to the coach and athlete relationship such as coach to athlete confrontation efficacy and coaching style (Boardley et al., 2019; Chen et al., 2017), were placed at the “Direct supervisors, management, medical and performance personnel” and “Athlete, teammates, and opponents” levels of the RMF, respectively (Figure 1). Factors and their associated actors which were identified as contributing to doping at the different levels of the RMF were mapped to produce an AcciMap and ActorMap, respectively. Relationships between the actors from across the broader system and the athlete were included in the ActorMap to represent the number of included articles identifying each given relationship. The relationships were extracted from the included studies when they were identified either quantitatively or qualitatively. For example, if a study identified that sponsorship influenced athlete doping, a link was made on the ActorMap between these two actors at their respective levels of the RMF, with the number between these two actors on the ActorMap reflecting the number of studies which identified this relationship.

Given the breadth of types of analysis used in the included studies (and associated data [e.g., surveys, case studies, interviews]), the identified relationship did not have to achieve statistical significance or any effect size threshold (e.g., Cohen’s d) for it to be included.

3. Results

3.1. Full-text selection

The initial search across the four databases yielded 6901 records, with 5818 remaining following duplicate removal. The titles and abstracts of these records were assessed against the eligibility criteria, and 370 records remained. Based on the full-text screening of these records, 317 were removed as they did not meet the criteria for inclusion (Figure 2). Thereafter 56 studies remained and these were included in the final synthesis (Supplementary Table S1) (Aubel & Ohl, 2014; Aubel et al., 2018, 2019; Backhouse et al., 2013; Bae et al., 2017; Barkoukis et al., 2015, 2020; Bilard et al., 2011; Bloodworth & McNamee, 2010; Boardley et al., 2015, 2019; Chan et al., 2014; Chen et al., 2017; Devcic et al., 2018; Didymus & Backhouse, 2020; Erickson et al., 2015; Fincoeur et al., 2018; García-Grimau et al., 2021; Harcourt et al., 2014; Harris et al., 2021; Hauw & Mohamed, 2015; Henning et al., 2021; Hurst et al., 2019, 2021; Huybers & Mazanov, 2012; Jalleh et al., 2014; Kabiri et al., 2020, 2021; Kegelaers et al., 2018; Kirby et al., 2014; Kondric et al., 2011; Lentillon‐Kaestner, 2013; Lentillon‐Kaestner & Carstairs, 2010; Lentillon‐Kaestner et al., 2012; Liposek et al., 2018; Lucidi et al., 2013; Matosic et al., 2016; Mazanov & Huybers, 2010; Mazanov et al., 2011; Morente-Sánchez et al., 2013; Murray et al., 2013; Nicholls et al., 2020; Ntoumanis et al., 2017; Pappa & Kennedy, 2013; Petrou et al., 2022; Sekulic et al., 2014; Smith, 2017; Smith & Stavros, 2020; Sullivan & Razavi, 2017; Vakhitova & Bell, 2018; Vidar Hanstad & Waddington, 2009; Whitaker & Backhouse, 2017; Whitaker et al., 2014; Zucchetti et al., 2015).

Figure 2. PRISMA 2020 study selection flowchart.

3.2. Publication information

There were 56 articles, published between 2002–2021, were included in the final synthesis (Supplementary Table S1). In total, 53 of the included studies (95%) used an applied approach (e.g., surveys, interviews, focus groups, or mixed methods) to data collection, while 3 of the included studies (5%) were relevant reviews. Of the applied studies, 26 studies used a survey approach (49%), while 15 studies (28%) collected data via interviews (semi-structured or otherwise), 2 (4%) used focus groups, 1 (2%) was a case study, and the remaining 10 studies (19%) either used a mix of survey and interview methods and/or other approaches to data collection (eg., analysis of athlete transcripts, database analysis, statements, and/or biographies) (Supplementary Table S1).

3.3. Participant information

The included studies primarily examined male and female athletes or players as their participants, although coaches, doctors, sports scientists, strength and conditioning coaches, physiotherapists, dieticians, doping “experts”, and researchers were also sampled (Supplementary Table S1). These participants were drawn from a range of Olympic and non-Olympic professional and amateur sports including (but not limited to): track and field, soccer (football), cycling, rugby union, rugby league, mixed-martial arts (MMA), rowing, hockey, gymnastics, wrestling, volleyball, basketball, handball, and water polo (see Supplementary Table S1 for further details).

3.4. Geographical makeup

Figure 3 shows the count of the studies which included participants from a specific country. The United Kingdom (12 studies), Australia (11 studies), and the United States of America (10 studies) were the largest source of participants (Figure 3). When studies identified participants from multiple countries they were included for each of those various countries. See Supplementary Table S1 for further details.

Figure 3. The geographical origin for the participants of the included studies with the colour gradient reflecting the count of the included studies (range 1–12 studies).

3.5. Study quality appraisal

The included studies passed the initial appraisal questions (S1 and S2) of the MMAT, indicating that they were able to be considered as research items for appraisal. Using the classification of the MMAT, 28 (50%) of the 56 included studies were considered quantitative descriptive, 22 (39%) were qualitative, 5 (9%) were mixed-method, and 1 (2%) was a quantitative non-randomised trial. Results were compared between raters (MN and SM) and there was 100% agreement between the two raters. The overall average rating (± standard deviation [SD]) for the included studies was high (4.7 ± 0.5). The score for each study is presented in Supplementary Table 1.

3.6. Actors and contributory factors

An ActorMap showing the actors and organisations identified as having an influence on instances of doping is presented in Figure 4. Actors were identified across all levels of the sport system hierarchy, with the most populated levels including the “Athlete, teammates and opponents” and “Direct supervisors, management, medical, and performance personnel” levels. Given the central role that the athlete plays in doping, the identified actors predominantly interact directly with the athlete (i.e., they influence the athletes’ doping behaviours and/or doping attitudes and beliefs) (Figure 4).

Figure 4. Relationships between actors mapped onto an adapted version of Rasmussen’s Risk Management Framework (RMF) ActorMap. Lines connecting two nodes represents a described relationship between the contributory factors and numbers overlaid onto the line represent the number of studies to describe the relationship.

The contributory factors which are reported in the literature as having an influence on doping were mapped onto the RMF to produce an AcciMap (Figure 5). The identified contributory factors in Figure 5 are expanded on in further detail in Supplementary Table 2. As shown in Figure 5, contributory factors were identified at all levels of the sport system hierarchy, with the most populated levels including the “Athlete, teammates and opponents”, “Direct supervisors, management, medical, and performance personnel”, and “Teams and organisations” levels.

Figure 5. Identified contributory factors mapped onto an adapted version of Rasmussen’s Risk Management Framework (RMF) AcciMap. See Supplementary Table 2 for expanded details on these factors.

In synthesising the results of the included studies, a number of contributory factors were identified which were mapped onto the RMF ActorMap and AcciMap. Firstly, at the international level, policy makers play a pivotal role in influencing doping. The likelihood of doping is increased when there is perceived to be a low threat of being tested and when penalties are perceived to be lenient. Moreover, the perceived legitimacy of anti-doping laws and enforcement is low, with weak structural conditions permitting previous acts of confirmed state-sponsored doping. Predictable anti-doping testing schedules and low chances of being caught are other factors encouraging doping. Certain government structures and policy makers have, at times, fostered a culture where doping reached state-sponsored proportions, promoting a doping culture. Policy makers and media within regulatory bodies and associations are perceived to have influenced doping through infrequent testing, lack of effective education, and a view of complicity in the evolution of doping through inconsistent responses to doping events. Changes in anti-doping policies, and media pressures focused excessively on performance results have also contributed. Within teams and organisations, various factors including team culture, economic status, and management strategies have fostered a conducive environment for doping. Sponsors have influenced doping through financial gains associated with performance. A culture of silence (termed “Omerta”) within the team environment further fosters pressure to conform. At the direct supervision level, medical staff, coaches, and doctors have directly influenced doping through the direct prescription of substances, encouragement of doping attitudes and behaviours, and the provision of information on banned substances. The coaching style and strategies have also been directly and indirectly associated with fostering positive doping attitudes. Teammates encourage doping to increase performance levels, while opponents influence doping through competitive pressure to succeed, as well as the perception others in their sport are doping. Family and friends can either encourage or discourage doping based on their attitudes and behaviours. Finally, the use of dietary supplements has been associated with a higher prevalence of doping, with supplement users reporting stronger doping intentions and positive attitudes towards doping. These factors are described in further detail in Supplementary Table 2.

4. Discussion

Doping represents a continued threat to fairness in sport, and athlete health (Gerche & Brosnan, 2017; Perera et al., 2013). This review sought to identify the known actors and contributory factors that influence doping and doping behaviours, attitudes, and beliefs in sport. The intention was to determine the extent to which the current knowledge base extends beyond athletes and coaches to consider the broader sport system influences. Contributory factors were identified at each level of a systems thinking based framework. This supports the view that decisions and actions at every level of the anti-doping system interact to create the adverse event of doping. By identifying these contributory factors, this review has synthesised the knowledge base on doping causation, which will help to facilitate the development of more effective anti-doping strategies. Beyond this, the findings point to clear future research requirements that can be used to better understand and respond to the issue of doping. The key outcomes of this review and their implications are discussed below.

4.1. Doping as an emergent property of sport systems

The first and perhaps most important finding from this review is the confirmation that doping is an emergent property of sports systems and hence represents a complex systems issue. Specifically, the identification of actors and contributory factors across all levels of the sport system hierarchy indicates that doping is influenced by a diverse set of actors and contributory factors spanning multiple levels of sport systems. This is the first study to synthesis contributory factors to doping via a contemporary systems-thinking-based framework, and confirms that doping is a problem that is not confined to athletes, coaches, and support staff. Rather, there are multiple actors who contribute to, and share the responsibility for, doping in sport (McLean et al., 2023). As such, it is clear that there are potentially many different causal pathways which lead to incidents of doping. Indeed, the structural and systemic demands of competition may justify the self-interested decisions of people to dope (Heikkala, 1993). The implication of this is that doping prevention and management can only be achieved through system reform, and that targeting individual athletes will have only a minimal impact, as seen in other safety critical domains such as road safety (Salmon et al., 2016).

In taking a broader perspective across the evidence examined throughout this review, it is clear that doping and anti-doping are complex systems issues (Ottino, 2003). Among the different characteristic features of complex systems are that they contain multiple interacting components, contain interactions between components which are non-linear, they contain feedback loops of interactions, they are dynamic and open, contain emergent properties, and have a history (Dekker et al., 2011; Salmon & McLean, 2020). As a complex system, improvements in our collective understanding of doping would benefit from the application of systems thinking based methods that consider these features of complexity (Salmon & McLean, 2020). For example, when examining the performance of a complex system, sub-optimal system performance (i.e., the occurrence of doping in the anti-doping system) often results from inappropriate or insufficient feedback mechanisms from the lower levels back up the hierarchy, a process known as “vertical integration” (Cassano-Piche et al., 2009). Without appropriate vertical integration, actors at each level of the system are unable to assess how their decisions interact and influence the decisions of other actors at other levels, ultimately resulting in a loss of control and, in turn, to negative outcomes.

4.2. Influence of support personnel

The mapping of the doping contributory factors identified in the literature onto the RMF revealed that there is a strong body of evidence which has identified factors relating to the level of the system directly above the athlete (i.e., the direct supervisors, management, medical and performance personnel level). Specifically, at this level of the system (i.e., the direct supervisors, management, medical and performance personnel), there are several actors who are collectively termed “performance support personnel” who influence decisions about doping and doping behaviours, attitudes, and beliefs.

The coach and coach-athlete relationship was shown to have a powerful contributing influence both indirectly and directly to athlete doping. For example, coach behaviours (e.g., narcissism, controlling actions) and opinions (e.g., moral stance) towards doping can sway athlete doping attitudes and behaviours (Matosic et al., 2016; Nicholls et al., 2014, 2020), or coaches may have a more direct role in doping (Kirby et al., 2014). Given that coaches are often held in high esteem and reverence, they exert a strong influence on an athlete’s doping “worldview”. As such, the coach-athlete relationship represents a leverage point with which either doping or anti-doping behaviours can become normalised. Indeed, education programmes which are aimed at coaches (e.g., Coaches Tool Kit, CoachTrue) have become recommended as compulsory under the WADA Code (Patterson et al., 2016, 2019). Despite this, only a third of coaches actively engage in this formalised training, have little knowledge regarding anti-doping, and see themselves as ill-equipped to discuss doping matters.¹² It should be noted that education programmes have changed and evolved over time, and these programmes were recently harmonised as the International Standard for Education (ISE). Whilst previous education efforts have been critiqued for their lack of efficacy – e.g., Woolf (2020), the efficacy of ISE and the overall effect of this shift in education to a harmonised system has yet to be fully understood. The results of this review should be considered in light of the potential future improved education practices that follow these harmonisation efforts.

The current findings further highlight that doctors, medical, and sports science staff are key influencers on athletes. The influence of doctors and medical staff on athletes comes as no surprise. Connor (2013) recognised that the athlete is connected to a network of support which includes (amongst others) doctors, nutritionists, physiotherapists, and biomechanists. Moreover, the environment within sport has become increasingly medicalised, as athletes and their network seek licit and elicit competitive advantage(s) (Barkoukis et al., 2015; Kegelaers et al., 2018). Clearly, the athlete exists in a network of potential medical and performance influences, and therefore, individualistic explanations for doping hide the contributory factors that exist around the “networked athlete” (Connor, 2013). Given the results of this review, there is a clear need to focus future research funding and efforts at the higher levels of the system, on investigating the influence of other actors within the system such as the athlete’s network (including support personnel), and how this network can be leveraged to facilitate anti-doping outcomes.

4.3. ‘Strict liability’ and the (lack of) interrelationships between higher level actors

The findings identified a lack of research examining the actors and associated contributory factors represented at the higher levels of the anti-doping system. In particular, the findings show that there are critical gaps in the knowledge base around the indirect influences on athletes doping behaviour. For example, higher system influences on coaches, doctors and medical staff, and teams. This reflects the current detection-deterrence system which places “strict liability” on the individual athlete with the sole responsibility for a positive doping test or ADRV (Kleiderman et al., 2020). Under the current Code, strict liability results in automatic sanctions as a result of an ADRV. Under strict liability, the decisions of the individual to dope have resulted in them being labelled as “cheats” who must be removed or banned from their sport to maintain a “clean” environment (Pappa & Kennedy, 2013). However, this eschews the understanding that the outcome of doping is a result of decisions, actions, and inadequacies at various levels within the system (including with the athlete), as identified within this review and elsewhere (e.g., Pappa & Kennedy, 2013). A conclusion from the review is that, though research has identified a wide range of actors and influences who play a role in doping, it has not extended beyond direct influences on athletes to consider the network of influences across sport systems.

This finding should be considered alongside advances in domains such as safety science which have seen a shift in thinking around prevention from focussing on sharp-end operators to consider the broader sociotechnical system (Dekker et al., 2011; Rasmussen, 1997). In road safety, for example, the traditional response to issues such as drink driving involved the “3 Es” of education, enforcement, and engineering, with the primary focus on educating and punishing drivers to prevent future occurrences. Whilst this approach was undoubtedly successful in reducing the road toll, a slowing of progress over the past decade has led to criticism, with systems thinking proponents pointing to the many factors outside of the driver, vehicle and road infrastructure that cannot be addressed through the 3 Es (Newnam & Goode, 2015; Salmon et al., 2012, 2016). As a result, driver focused interventions will have some impact; however, systemic road transport and societal issues will not be dealt with and so drivers will continue to migrate towards undesirable behaviours (Salmon & McLean, 2020). The findings from the present review suggest that, though doping is a systems issue with multiple contributory factors, the issue is currently being dealt with in a reductionist manner that is analogous to that previously seen in road safety. Despite strong enforcement, it can be inferred that doping will continue until broader sport system and societal interventions are introduced. Potential broader solutions might include a shift in examining interventions which focus at deeper leverage points which may include a shift to focus on medical supervision and harm reduction, as suggested below.

4.4. Limits of the current deterrence-detection-punishment paradigm

Acknowledging that doping is a systems issue will facilitate the development of more effective anti-doping interventions. For example, Meadows identified 12 leverage points within systems where interventions can be made to alter the systems behaviours and change outcomes (Meadows, 1999). Such leverage points within the system can be ordered by the potential for effectiveness to create systemic change from shallow (e.g., mechanistic parameters, strength of negative feedback loops) to deep leverage points (e.g., the rules of the system, the goals of the systems, the mindset/paradigm out of which the system arises) (Abson et al., 2017). Presently, the focus of many of the current interventions, such as new detection methods and harsher sanctions on the individual athlete, is at the shallow end of these potential leverage points, which have minimal potential to elicit systemic change. For effective system-wide change in doping, it is clearly necessary for anti-doping to refocus on directing greater effort and resources towards influencing deeper leverage points (where possible) including the incentives and constraints involved, and the goals of the system (Abson et al., 2017). One such suggestion in the literature is to allow and destigmatise doping practices in sport in conjunction with a shift in values to focus on the health of the athlete who may want to dope under medical supervision (Abson et al., 2017; Savulescu et al., 2013). For others, this suggestion goes against the fundamental “spirit of sport” fairness value of sport and includes potential negative health implications, which renders it untenable. This disagreement is unsurprising, given that this would necessitate changing the mindset out of which the current system arises (one of the deepest and most difficult to influence leverage points) away from deterrence, detection, and punishment. If harm minimisation becomes the goal of anti-doping, radical shifts to what the Code contains would then need to be necessitated, as the focus of the Code is currently on deterrence-detection-punishment. What an updated Code would contain in a harm minimisation framework is an open question. This also highlights how doping remains both a complex and “wicked” problem, in which potential solutions are ill-defined, there is no likely endpoint in which the problem is solved, and solutions likely require a shift in mindsets or approaches beyond what may have worked in the past (Petróczi & Boardley, 2022).

An alternative idea which would sustain the current system would be to expand sanctions to punish entire sports, governing bodies, or nations should doping become rife. Indeed, there are recent examples of this occurring including Russian athletes being banned from the 2016 Olympics due to state-sponsored doping practices, and the removal of weightlifting from the 2028 Olympics programme (Harris et al., 2021). Other less controversial options may include changing the incentives from detection and deterrence to focusing on health outcomes. One option would be to provide incentives for teams and players not to dope through nation, team, or individual athlete bonuses (financial or otherwise) for achieving certain specified non performance-related health outcomes. Similar incentives have been made in football (soccer) whereby national teams in UEFA who demonstrate “fair play” across criteria including frequency of yellow and red cards, positive play, and respect for the opponent and referee are awarded cash prizes (from 1995–2015 the award was given as extra places in UEFA affiliated competitions). Any such incentive structure change that may positively shift the anti-doping systems orientation needs to be seriously considered before implementation to ensure it does not become exploited and corrupted (i.e., Goodhart’s or Campbell’s Law) (Rodamar, 2018).

There are analagous safety critical systems to anti-doping in other domains which have attempted to shift the focus from shallow to deep leverage points. For example, research that explored the influence of policy, societal, and political factors on drink driving and road safety has identified the need for an integrated approach which includes (amongst other factors) rehabilitation, urban planning, and public health (Poznyak et al., 2005; Stephens et al., 2017; Stevenson et al., 2016). Importantly, research in road safety has identified that resistance from the community to proposed changes at the deeper leverage points will influence the effectiveness of any intervention. Therefore, any change needs to be clearly communicated, with the problem being addressed, the approach being taken, reasons for that approach, and previous evidence for success of that approach being rapidly communicated. Whilst a number of issues have been identified and suggestions have been made throughout the literature, and in this systematic review, the focus and content of the interventions necessary to positively influence the anti-doping systems orientation remain an open question. Examining this through a systems thinking lens should be considered as part of the progressive research agenda.

5. Conclusion

Doping remains prevalent throughout sport (Momaya et al., 2015). This systematic review of the contributory factors to doping provided a first-of-its-kind analysis of the systemic factors which contribute to doping and doping behaviours, attitudes, and beliefs in the peer reviewed literature. The findings show that there are contributory factors to doping behaviours and attitudes at each level of sport systems from international influences and government levels, through to the equipment and substances at the lower levels of the system. The coach and coach-athlete relationship are one particular aspect which seems to have a large influence on the doping “worldview” of athletes. The current deterrence-detection system appears to lack legitimacy to athletes as they perceive the penalties are low relative to the potential rewards on offer for doping-related improvements in performance. Given that doping has been persistent under the current anti-doping approach, research examining interventions to improve the anti-doping system should focus their efforts on higher system leverage points which have the potential to elicit whole of system change. The findings have implications for athlete health, as well as anti-doping authorities and policy makers as they seek to to minimise and, where possible, eliminate doping from sport.

The current findings point to the importance of understanding the complex network of organisations and actors who exert influence on athletes, and the importance of the coach-athlete and performance staff-athlete relationships which may be a leverage point with which doping attitudes and behaviours can be influenced. It is important for doping authorities to continue funding efforts which focus on different approaches to examining the complex dynamics involved in doping, particularly as the effects of the current education, enforcement, and engineering (of new tests) approach has reached somewhat of an impasse, with respect to eradicating doping. Given the complex, emergent nature of doping in the wider sporting system, future research investigating doping and anti-doping would benefit from the adoption of systems thinking methods. Indeed, recent research has supported the observation that majority of existing controls which in anti-doping are reactive, and there is potential to employ proactive measures (such as leading indicators) to prevent doping (McLean et al., 2023). A shift away from understanding doping through the current reactive and reductionist lens which focuses on the use of detection and enforcement (termed “Anti-doping-I”) to a more balanced view which integrates proactive and systemic strategies in a cohesive approach (termed “Anti-doping-II”) is needed (McLean et al., 2023).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online https://doi.org/10.1080/02640414.2024.2306056.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.