Doping can be defined as the occurrence of one or more anti-doping code violations and is usually observed by the presence of a prohibited substance, its metabolites or markers in a biological specimen from an athlete (Sajber et al., 2013). Doping usage in sport is known to be related to negative health-related consequences and even death (Honour, 2016; Mazanov et al., 2012). Additionally, doping corrupts the main essence of sport and fair play and is therefore considered non-ethical behavior (Ljungqvist et al., 2008). As a result, the global fight against doping is highly prioritized in all organized sport societies (Ljungqvist, 2014).

The World Anti-Doping Agency (WADA), a global governing body for anti-doping in sports, has put special efforts into the development and application of differentially targeted approaches in the fight against doping. Generally, two approaches can be recognized in an anti-doping campaign. The first approach includes the development of reliable and applicable measurement tools and protocols that allow the precise identification and consequent penalization of athletes who have resorted to doping (Kiss et al., 2013; Malm et al., 2016). The second approach in global anti-doping efforts is more “preventive” in its nature and includes the identification of the cultural and sport-specific factors that influence doping behavior in each sport society (Furjan Mandic et al., 2013; Morente-Sanchez et al., 2013; Rodek et al., 2013). The idea is to identify certain precipitating factors of doping behavior in sports and to evaluate the nature of their influence (i.e., risk or protective effects on doping behavior) (Kisaalita and Robinson, 2014).

Several factors have been studied to identify possible associations with doping in sports, including sport-specific factors, socio-demographic variables, socio-cognitive factors (i.e., variables identified throughout the self-determination theory), and motivational variables (Barkoukis et al., 2013; Chan et al., 2015; Kondric et al., 2011; Matosic et al., 2016; Zenic et al., 2010). However, the studies conducted to date suggest that the factors associated with doping behavior (either actual or potential behavior) in one group of athletes (sport, gender, socio-cultural environment) could be differentially associated with doping behavior in other sport-specific groups (Rodek et al., 2013).

The specific influence of precipitating factors on doping behavior in different sports and societies is even more aggravated by the fact that the prevalence of doping is very different in relation to sport, gender, age and athlete (Lentillon-Kaestner and Ohl, 2011). Moreover, the doping intentions of athletes are influenced by distal influences (e.g., self-determination, sportpersonship orientations, and achievement goals), and proximal influences (e.g., situational temptation and perceived behavioral control, descriptive and subjective norms, and attitudes) (Barkoukis et al., 2013; Ntoumanis et al., 2017). Therefore, specific analyses of different sports and socio-cultural environments are needed.

It is widely accepted that characteristic relationships that exist between coaches and athletes should be observed as important determinant of athletes’ attitudes toward doping, and the importance of coaches as potential agents in the prevention of doping amongst athletes has been repeatedly emphasized (Backhouse and McKenna, 2012; Barkoukis et al., 2013; Lonsdale et al., 2017). In studies where type of coaching (i.e. style of coaching) was observed as a determinant of athletes’ doping-behavior, the authors used the self-determination theory to evidence coaches’ personal behavior, and observed the influence of the contextual climate that the coach creates on athletes’ motivation and pro- and antisocial behavior, including doping attitudes (Chen et al., 2017; Hodge and Lonsdale, 2011). Indeed, the coaching style (i.e., coaching behavior: autonomy supportive vs. controlling) is an important determinant that can modulate athletes’ attitudes (Chen et al., 2017; Hodge et al., 2013). However, behavioral characteristics are partially inherited but are mostly shaped throughout maturation-experience interactions, and each individual (i.e., coach) develops distinct behavioral characteristics as a result of the specific maturational-experimental interactions that influence him/her (Lerner, 2013). Therefore, the possible influence of coaching behavior on athletes’ doping tendencies can be used to identify at-risk athletes.

Meanwhile, coaching strategy and training methodology (CS&TM) have not been studied as factors potentially related to doping behavior in athletes. In short, CS&TM can be defined as a set of methods that coaches use throughout the sport training process to improve the athlete’s physiological capacities and sport-specific skills. The CS&TM includes (but is not limited to) the application of different types of training, training regimes and methodological/didactical approaches in sport training. Of particular importance is the fact that the CS&TM is modifiable and adaptable (Bompa and Haff, 2009). Therefore, if some aspects of the CS&TM are identified as being related to athletes’ doping susceptibility, it would implicate its applicability in anti-doping efforts, either by detecting those athletes who are at certain risk for doping behavior, or throughout instructing the coaches and informing them that certain type of CS&TM is recognized as a risk factor for PDB in athletes.

Despite the fact that the variables of CS&TM could influence doping susceptibility in different sports, this problem is particularly important in individual, mono-structural, cyclic sports, which are performed in systematically controlled environment, such as swimming. Namely, to improve the competitive achievement (sport-result) in swimming, the variables which are controlled throughout the training process are training-volume, training-intensity, and mastering of the specific swimming-skills (i.e. swimming-technique) (Colwin, 2014). While doping in sport is mostly used in order to enhance athletes’ physiological capacities (i.e. to overcome physiological stress induced by training volume and intensity, and to boost the mechanism of supercompensation) (Colwin, 2014; Rodek et al., 2013), it is reasonable to expect that the influence of CS&TM on doping susceptibility in swimmers is higher than influence of CS&TM on doping susceptibility in athletes involved in “multifaceted” sports (i.e. sport games, team sports).

The aims of this study were to identify the prevalence of potential doping behavior (PDB) in high-level swimmers and to identify the factors associated with their PDB. We were mainly focused on variables of CS&TM but also studied those factors that were previously reported as being potentially important determinants of PDB in sports The increased knowledge on a problem will allow the development of a meaningful and accurate anti-doping strategy in swimming, Initially, we hypothesized certain associations between CS&TM variables and PDB in swimmers. As a methodological remark, it must be emphasized that this study included practically the whole population of high-level competitive swimmers older than 18 years in a country (see Methods for more details) and therefore allows a substantial generalization of the findings.

The original sample in this study comprised 97 swimmers from Slovenia (35 females; 19.7 ± 2.3 years of age; 11.3 ± 3.1 years of experience in swimming sport). All participants were older than 18 years and were tested during the 2017 National Championship. An invitation to participate in the study was sent by the national swimming federation, and none of the athletes refused to participate; therefore, all swimmers who participated in the championship were included. The study was originally initiated and approved by the national swimming federation, complied with all ethical guidelines and received approval from the Institutional Ethics Review Board at the corresponding author’s institution (EBO 10/09/2014-1).

The previously validated questionnaire on substance use (QSU) was used to test the athletes (Zenic et al., 2010). Additionally, participants were questioned about CS&TM they were exposed to.

The QSU included questions on socio-demographics (age [in years], and gender), sports factors, doping factors and questions on CS&TM. Sport factors were assessed by questions on the (i) athlete’s experience in swimming (in years) and (ii) competitive results achieved in (iia) non-Olympic events (25-m pool) and (iia) Olympic events (50-m pool) (“Regional-level medalist”, “National championship - finals”, “National championship - medal”, “European and World Championship - finals”, “European and World Championship – medalist”, “Olympics”), (iii) preferred style of swimming (i.e., front crawl, butterfly, breaststroke, backstroke, medley), and (iv) competitive discipline in which they mostly compete (i.e., short distance, middle distance, long distance).

Doping-related factors were assessed by asking participants their opinions about (i) the occurrence of doping in swimming (“I don’t think doping is used in swimming”, “Not sure about it”, “Occurs, but rarely”, “Doping is often”), (ii) the number of doping tests (“Never tested on doping”, “Once or twice”, “Three times or more”), and (iii) PDB. The PDB was tested on scale which included four possible answers (“I would engage in doping if it would help me”, “I would engage in doping if it would help me with no negative health consequences”, “Not sure” and “I do not intend to engage in doping in the future”), but for the purpose of logistic regression analysis the responses on PDB were specifically clustered (see later text on Statistical analyses). This scale was found to be valid in evaluation of PDB in different sports, including tennis, synchronized swimming, and various team sports (Furjan Mandic et al., 2013; Kondric et al., 2013; Sekulic et al., 2014; Sekulic et al., 2016). What is also important, recent investigation done on team sport athletes of both genders confirmed high correlation between PDB and another commonly used measurement tool (Performance Enhancement Attitude Scale - PEAS) (Morente-Sanchez et al., 2014; Sekulic et al., 2016).

The CS&TM questions examined swimmers opinion on training methodology and coaching strategy he/she has been experiencing. Some questions were assessed on binomial scale (Yes – No), while some others were assessed on scales that included more possible answers. With regard to training methodology swimmers were asked on their perception about: (i) general characteristics of their training, (ii) attention paid on mastering of the swimming technique during training, (iii) training volume they were exposed to (i.e. swam distance), and (iv) characteristic training intensity.

The general characteristics of the training were evaluated by three statements on binomial (Yes-No) scale: “Swimming technique is an important part of my training”, “Training is monotonous and lacks diversity”, and “Training is mostly oriented toward volume (swam distance)”. The attention paid on swimming technique during training was asked by one question (“The swimming-technique is practiced …” ) and swimmers had to choose one of three responses (“… in less than 10% of training”, “… in 10-30% of training”, “… in more than one-third of training”). Swimmers self-reported their training volume on one question (“My average training volume is…”), and had to choose one of five possible answers (“… approximately 20-30 km per week”, “… 30-40 km per week”, “… 40-50 km per week”, “… 50-60 per week”, “… >60 km per week”). The question on self-estimated “Training intensity” included six possible responses (“Training is high in intensity when I have to swim >6 km per session”, “… >2 km in one sequence”, “… repeated sets of maximal intensity, regardless of distance”, “… different relays while being highly focused on stroke technique, speed and force”, “… some specific sets which I have never/rarely performed before”, “Intensity is high but with no specific reason”).

Coaching strategy was evaluated by the following “Yes-No” statements: (i) Coach frequently explains the training aims, (ii) Coach overviews and discusses the quality of (my) execution of specific tasks, (iii) Coach is very strict and rigid, (iv) Discipline is an important part of our training regime, (v) Coach pushes me very hard, and (vi) Sometimes, I don’t know what the Coach wants me to do in training.

Testing was conducted in the local language in groups of at least five athletes who were informed that the survey was anonymous, they could refuse to participate, they could leave some of the questions and/or the entire questionnaire unanswered and returning the completed questionnaire was considered consent to participate in the study. After testing, the questionnaires were placed in a sealed box that was opened on the day after the testing. For those athletes who participated in the testing, the response rate was high, and only three athletes returned the questionnaire unanswered. Therefore, the final sample comprised 94 athletes. The statements included in the CS&TM were originally suggested by the head coach of the national swimming federation and then checked and defined in their final forms through consultations among five swimming experts (two university teachers - former swimmers, two high-level coaches - officials of the national swimming federation, and one former swimmer - Olympic medalist), which contribute to ensure content validity of the CS&TM. Also, before this study, the reliability of the CS&TM was checked via a test-retest procedure. Briefly, 15 athletes (not included in this study) responded to the CS&TM questions twice in a 10-d time frame while using self-determined codes for identification purposes (i.e., they were advised to use the last three digits of their e-mail password as the identification code for easier recollection). While the CS&TM variables were in most cases ordinal in nature (“Yes-No” statements), the percentage of equally responded queries was calculated to establish reliability, as previously done in similar studies (Sekulic et al., 2014; Sekulic et al., 2016). The percentage of equally answered statements was 92% on average (from 85% for statement “Technique is practiced in 10-30% of training”, up to 98% for statement “Intensity is high but with no specific reason”), thus demonstrating the high reliability of the CS&TM.

The statistics included counts and frequencies (for categorical and ordinal variables), or means and standard deviations (for continuous variables). Multinomial logistic regression models were employed to examine how variables derived by the questionnaires were associated with the PDB (Negative PDB [Those who responded: “I don’t intend to engage in doping in the future”] – Neutral [“Not sure”] – Positive PDB [“I would engage in doping if it would help me”, and “I would engage in doping if it would help me with no negative health consequences”]). The negative PDB was set as the reference value. Previous studies frequently reported significant associations between sociodemographic variables and personal opinion on doping presence in sports and PDB (Sekulic et al., 2014; Sekulic et al., 2016). Therefore, we first evaluated associations between sociodemographic variables and personal belief about doping presence in swimming, and PDB. The analyses showed a significant association between Gender and PDB (Neutral-PDB: OR: 1.64; 95%CI: 1.32-5.01; Positive-PDB: 6.58; 1.01-9.12). Consequently, multinomial regression models were adjusted for Gender as a possible confounding factor. For all analyses, Statistica 13.0 (Dell, Tulsa, OK, USA) was used, and a p-level of 95% was applied.

Males and females had equal experience in swimming (11.51 ± 3.4 and 11.20 ± 3.0 years, respectively; t-test: 0.01, p = 0.99), and were of similar age (19.9 ± 2.0 and 19.6±1.9 years, respectively; t-test: 1.05, p = 0.29). Almost all studied swimmers observed swimming as sport being contaminated with doping, with 43% who perceived that doping is common in their sport. Approximately 11% of swimmers declared positive PDB, and additional 14% reported neutral PDB (Table 1).

Data on CS&TM are presented in Table 2. More than 71% of swimmers stated that they frequently master swimming technique during their training, with 35% of swimmers who stated that swimming technique is practiced at more than one-third of all training sessions. Approximately, one-third of swimmers approximated their average training volume on 40-50 km per week. The 16% of swimmers stated that they “sometimes don’t know what does the Coach wants from them to do in training”.

The results of the multinomial regression analyses between studied predictors and PDB-criterion are summarized in Table 3 and Table 4. Those swimmers who achieved better competitive result in Olympic-pools (i.e. Olympic disciplines in 50-m pools) were less likely to report neutral-PDB (OR: 0.41; 95%CI: 0.17-0.98) (Table 3). The neutral-PDB was less frequent in those swimmers who stated that technique is an important part of their training regime (OR: 0.45; 95%CI: 0.12-0.78). Further, the positive-PDB was evidenced in those swimmers who perceived that their training was monotonous and lacked diversity (OR: 1.82; 95%CI: 1.41-5.11). Also, those swimmers who perceived that they were involved in training which was mostly oriented toward volume were more likely to declare positive-PDB (OR: 2.76; 95%CI: 1.11-7.12), and neutral-PDB (OR: 1.64; 95%CI: 1.09-3.12). The lower likelihood of neutral-PDB (OR: 0.14; 95%CI: 0.03-0.96), and positive-PDB (OR: 0.12; 95%CI: 0.01-0.81) was evidenced in those swimmers who declared more frequent practicing of the swimming-technique. Statement “Sometimes, I don’t know what does the Coach wants me to do in training” was positively related to negative-PDB (OR: 1.67; 95%CI: 1.01-4.21). Those who stated that training intensity was high in those situations when they had to be focused on stroke-technique, -speed and –force; were less oriented toward positive-PDB (OR: 0.18; 95%CI: 0.02-0.62). Also, swimmers who replied that their coach regularly provided the attention to explain the training aims, were less likely to declare positive-PDB (OR: 0.21; 95%CI: 0.04-0.81), and neutral-PDB (OR: 0.10; 95%CI: 0.01-0.67). Similarly, opinion that coach frequently reviewed and discussed the quality of the athlete’s execution of specific tasks, was negatively correlated with positive-PDB (OR: 0.41; 95%CI: 0.02-0.81), and neutral-PDB (OR: 0.19; 95%CI: 0.11-0.98) (Table 4).

There are several most important findings of this investigation. First, prevalence of PDB in swimmers was within expected values. Next, male swimmers were found to be more prone to PDB than were females. Additionally, opinions about doping presence in swimming were not associated with athletes’ doping susceptibility. Finally, several factors related to CS&TM were associated with PDB. Therefore, results support our initial hypothesis on significant association between CS&TM variables and PDB.

The prevalence of PDB in the swimmers was within the expected values, mostly because a previous study showed similar figures with regard to the tendency toward PDB in similar-level swimmers from Croatia (Sajber et al., 2013). Although the comparison between the current study and the previous Croatian study is biased to some extent (i.e., the studies observed swimmers from different countries), it seems that the trend of positive PDB increased slightly (i.e. 80% and 75% swimmers who declared negative-PDB in Croatia and Slovenia, respectively). However, increased percentage of swimmers with positive PDB could be a result of the recent doping scandals in domestic swimming and a consequent (higher) belief in the high doping prevalence in swimming, which also resulted in an increased positive PDB.

Our results showed a higher doping tendency in male swimmers. The differences between males and females with regard to doping attitudes and doping tendencies were already studied. Some epidemiological data suggest higher doping prevalence in male athletes, which could indirectly demonstrate higher positive doping tendencies in males (Nicholls et al., 2017; Zaletel et al., 2015). However, higher doping susceptibility in males was not always supported in studies where specific samples of athletes were investigated. For example, although males were more prone to PDB in basketball and handball, no significant gender differences toward PDB were established for volleyball and soccer (Sekulic et al., 2016). Similarly, gender was not associated to PDB in kick-boxing (Sekulic et al., 2017), and attitudes toward doping among athletes involved in different sports (Zucchetti et al., 2015). Even more, a recent study reported female soccer players as being at a higher risk for PDB than their male peers (Zvan et al., 2017). Meanwhile, this is one of the first studies to examine gender-differences in PDB for swimming. As a result, it seems that for the studied country and sport, male gender should be recognized as a risk factor for PDB.

Studies regularly found higher doping susceptibility in those athletes who perceive their sport as being doping-contaminated (Furjan Mandic et al., 2013; Kondric et al., 2013; Sekulic et al., 2014; Sekulic et al., 2016), and these findings have been elegantly explained by the socio-psychological theory of self-categorization (Oakes and Turner, 1986). In short, it is generally known that people adopt the norms, beliefs and behaviors of “their group” (Page et al., 2015). Consequently, if athletes perceive their sport as being doping-contaminated, it is more likely that he/she will engage in doping in the future (Rodek et al., 2009; Wiefferink et al., 2008). Meanwhile, our results showed a non-significant association between athletes’ opinions about doping presence in swimming and their PDB. Interestingly, this finding is in accordance with only study which examined this problem in swimmers (Sajber et al., 2013). It is important to note that none of the swimmers declare that “swimming is doping-free sport”, while only approximately 5% of the swimmers stated were “not sure about doping presence in swimming”, and these results may be observed as plausible because almost identical values were reported for Croatian swimmers several years ago (Sajber et al., 2013).

Our results showed several significant relationships between the variables explaining CS&TM and PDB. Since the discussion for each pair of established associations will extensively broaden this discussion, we have tried to cluster the variables that were correlated with PDB and discuss them accordingly. The first cluster points to a specific relationship between athletes’ perception of characteristics of their training (i.e., training methodology) with doping susceptibility. Briefly, swimmers were generally more susceptible to doping if they perceived that their training is: monotonous, strictly oriented toward volume, and lacked work on improvement and mastering of the swimming technique.

Basically, training volume, intensity, and load are modulated to improve an athlete’s metabolic capacities, which consequently lead to better performance such as better endurance-, and/or sprinting-capacity (Bompa and Haff, 2009; Drew and Finch, 2016; Toubekis et al., 2013). The performance-enhancing substances, including doping substances, are frequently used to enhance recovery of the metabolic demands of training and competition (Bahrke and Yesalis, 2002). Therefore, the athletes’ perception of a high training-volume and -load indirectly implies high metabolic demands of training, which altogether in our study result in doping vulnerability. On the other hand, self-estimated intensity of training is not found as significantly related to PDB, although this training parameter also influence overall training load. Although the profound interpretation of such relative inconsistency (i.e. training volume is “positively correlated”, while training intensity is not correlated to PDB) exceeds the aims and design of this study, the following explanation can be offered.

One of the most important issues in sport training is a problem of overtraining. Overtraining can be defined as an accumulation of training and/or non-training stress resulting in long-term decrement in performance capacity with or without related physiological and psychological signs and symptoms (Halson and Jeukendrup, 2004). In general, two types of overtraining are recognized: parasympathetic-, and sympathetic-overtraining. The first one (parasympathetic) is related to accumulation of training with high volume, and includes heavy fatigue, insomnia, no libido, chronic tiredness, low motivation, low resting heart rate, and low blood pressure. The second one (sympathetic) is mostly induced by training intensity and characterized by high levels of stress hormones (cortisol), and results in irritability, restlessness, poor sleep, weight loss, poor performance, and low libido. In swimming, volume is basically modified throughout swam distance (i.e. the longer is the distance - the higher is the volume). Meanwhile, intensity can be regulated by various mechanisms (e.g. stroke, specific exercises, sets vs. rests). Supportively, volume of training is already found to be associated to overtraining in swimming, with positive correlation between swam distance and psychological indicators of overtraining (Pierce Jr, 2002). Therefore, here evidenced association between self-evidenced training volume and PDB can be at least partially described on a basis of such influence (i.e. higher volume increases the risk of overtraining and consequently increases the risk for PDB).

The mastering of the sport-specific technique improves an athlete’s sport-related skills (Bompa and Haff, 2009; Marinho et al., 2010). In swimming, this is mostly related to improvement of the stroke-technique, which is particularly beneficial with regard to long-term development and possibility of improvement in swimming results in later stages of sport career (Colwin, 2014). The fact that those who were susceptible to doping also reported that “less attention is paid to swimming technique” corresponds to a correlation between high volume and PDB, and indicates high doping susceptibility in those swimmers who were involved in training of high metabolic demands, which is discussed in previous paragraphs.

The second cluster of CS&TM variables correlated to PDB, consisted of variables related to coaching strategy. The coaching style is recognized as being an important determinant of athletes’ behavior, including athletes’ doping vulnerability (Barkoukis et al., 2013; Lonsdale et al., 2017). Investigators repeatedly evidenced coaching behavior (autonomy supportive vs. controlling) and contextual climate, and evaluated its influence on athletes’ doping attitudes (Chen et al., 2017; Hodge and Lonsdale, 2011). However, to the best of our knowledge, this is the first investigation where variables of coaching strategy were studied as possible determinants of doping susceptibility in athletes.

Briefly, variables explaining coaching strategy that were found to be associated with a positive doping attitude underlined the athletes’ perception on coaches’ indifference and nonchalance in regards to the athletes’ performance (i.e., coach does not explain training aims, coach does not overview the quality of work). Swimmers who experienced this type of coaching style were discontent and, consequently, were more prone toward PDB in the future. Although the explained causality is hypothetical to some extent, it is indirectly confirmed by the established relationship between achieved competitive results and doping attitude. Briefly, those swimmers with lower competitive achievement were more prone toward PDB, which is consistent with previous reports, in which similar correlations were observed in other sports (Furjan Mandic et al., 2013; Kondric et al., 2010; Sekulic et al., 2014).

One can argue that the previously described associations between CS&TM and PDB can be generated by “objective” or “subjective” perceptions. In other words, it is questionable whether specific characteristics of CS&TM truly occurred, or the athlete just subjectively perceived it as apparent. Indeed, while it is possible that the training and coaching truly were as judged by the athletes (i.e., “… strictly oriented toward volume and load”, “… lacked diversity”), it is also possible that athletes did not objectively evaluate CS&TM either because they lacked motivation, or they were not satisfied with the achieved results. Indeed, future studies should explore this dilemma more profoundly. However, regardless of the true origin of perception, it is clear that athletes’ opinions on CS&TM are related to their doping attitudes, and therefore these variables deserve attention as covariates of potential doping behavior.

Although the study was designed as anonymous, there is a certain possibility that athletes could lean toward socially desirable answers. However, we believe that the study design and our experience from previous studies decreased this possibility. Also, we studied “doping intentions”, and not “doping usage”, and therefore we may not speak about protective- and/or risk-factors without any doubt. Finally, although study involved large percentage of high-level swimmers (all senior age participants of the National Championship), the relatively small sample (i.e. less of 100 swimmers) should be emphasized as a certain limitation of this investigation. Namely, the number of subjects, together with multinomial statistical design (i.e. athletes were grouped into three clusters on a basis of PDB), probably decreased the possibility of reaching the appropriate statistical significance of some associations.

This study confirmed previous figures on the large percentage of swimmers who were convinced of the doping presence in their sport. Additionally, there is evidence that the tendency toward PDB in swimming has increased over the last several years. However, for a more profound analysis of this problem, longitudinal studies on athletes from the same country are needed. The results of this study support previous findings from swimming, where personal belief about doping presence in the sport was not found to be correlated with PDB in swimmers.

This study highlighted specific associations between CS&TM variables and PDB, with the following athletes being more susceptible to doping: (i) those who perceived their training as being mostly oriented toward training-volume (and not swimming technique) and (ii) those who perceived their coach as being indifferent to the evaluation of training goals and athletes’ achievement. While this is one of the first studies to observe CS&TM as possible covariates of PDB and to find several consistent associations explaining specific relationships among the variables, a similar approach is warranted in other sports.