Exercise Addiction

Exercise addiction is characterized by too much exercise over which the person has lost control. Instead, a constant urge to exercise and craving sensations control the individual (Szabo & Demetrovics, 2022). Withdrawal feelings emerge if exercise is prevented (i.e., because of injury, bad weather, the closure of the usual training facility, an important commitment, or other factors) (Griffiths, 2005). Apart from withdrawal, its typical symptoms include salience, tolerance, inter and intra-individual conflict, use of exercise for mood modification, and relapse if one intends to control the behavior (Griffiths, 2005). Its negative consequences emerge in psychological, physical, and social damages (Juwono & Szabo, 2020). Exercise addiction is studied within the scholastic field of behavioral addictions and emerges via two pathways: (I) therapeutic and (II) mastery (Dinardi et al., 2021). The former is associated with coping with or escaping from stress and/or trauma. At the same time, the latter is connected to the failure or unwillingness to recognize one’s physical limits, pushing the training over barriers of pain, resulting in injury or even termination of one’s athletic career (Egorov & Szabo, 2013).

Objective

This paper aims to send a timely, thought-provoking message to scholars in behavioral addictions, especially those focusing on exercise addiction. It is timely because research in this area has proliferated (Szabo & Kovacsik, 2019). For example, on March 15, 2024, there were exactly 22,000 papers on Google Scholar containing anywhere within their text the exact terms “exercise addiction” (n = 6980), “exercise dependence” (n = 6410), “compulsive exercise” (n = 6540), and “obligatory exercise” (n = 2070; see Appendix). The first relatively evident issue worth contemplating is that these terms denote the same concept, such as overexercising in an unusual or “abnormal” pattern mirroring morbid exercise behavior (Alcaraz-Ibáñez et al., 2020; Szabo et al., 2018). Using different terms for similar or an identical concept yields ambiguity, confusion, misconception, irreproducible results, and faulty interpretations (Szabo et al., 2018). This is why there is a call for studying unhealthy exercise patterns under a single unified term, such as “morbid exercise behavior” (Alcaraz-Ibáñez et al., 2020; Szabo et al., 2018).

The second issue that merits contemplation is the increasing number of papers published in this area. However, is this growth a problem indeed? The answer is yes because the current research efforts lead nowhere, as will become evident later in this article. The stagnant status quo mirrors a general superficial understanding often blended with relaxed research strategies, probably due to a closer focus on publication quantity than quality (the “publish or perish” principle). However, why is such abundant research unproductive? This position paper seeks to pinpoint conceptual and methodological obstacles in research that impede progress in this field to foster a paradigm shift toward more effective research. This aim is accomplished by summarizing the current research’s leading issues and simultaneously providing several points for consideration and recommendations for future research.

How it Started

In a research report published in 1970, Baekeland studied the sleep patterns of 14 college students who exercised regularly. Measures were taken on two exercise days and four nights over 1 month, during which all sports and exercises had to be avoided. The researcher noticed disturbed sleep associated with an increase in anxiety during the exercise deprivation period. Furthermore, participants reported subjectively perceived sleep disturbance, sexual tension, and an elevated need to socialize. These objective and subjective manifestations were interpreted as withdrawal symptoms projecting the addictive nature of regular exercise.

Research Proliferation

While a few papers emerged during the first quarter of a century after Baekeland’s study was published, research started proliferating in the new millennium (Fig. 1). Based on Fig. 1, the term exercise addiction is used more often than exercise dependence, which might reflect the position that dependence is only one component of addiction, the other being compulsion, and addiction is characterized by both high dependence and high compulsion (Goodman, 1990; Szabo & Demetrovics, 2022). Neither of these components being high alone reflects addiction. For example, one can depend on exercise to maintain a healthy weight, muscle tone, post-infarct cardiovascular health, normal blood pressure levels, etc. At the same time, another individual may manifest compulsive exercise as an auxiliary means of weight control in an eating disorder (Meyer et al., 2011). While Fig. 1 only presents the compilation of papers, including “exercise addiction” (n = 617) and “exercise dependence” (n = 512) in their title, these terms surface in 13,390 full-text articles. Adding two additional terms, “compulsive exercise” and “obligatory exercise,” increases this number to 22,000. Therefore, it can be safely assumed that thousands of papers are published in the area. One can react to these statistics by asking: So, what? Well, here are two issues that are worth contemplating concerning these numbers.

  1. 1.

    Why do we study exercise addiction? (General research orientations, practical values)

  2. 2.

    How many people suffer from exercise addiction? (Justification of the abundant work)

Fig. 1
figure 1

Scholastic papers with “exercise addiction” (EA) and “exercise dependence” (ED) in their title. These papers were published between 1970 and the end of 2023 and traced via a Google Scholar year-by-year using “advanced search”

Full size image

The Principal Objective of Exercise Addiction Research

Since Baekeland (1970), withdrawal from exercise has been associated with deprivation feelings, equated to withdrawal symptoms in substance addictions (Weinstein et al., 2017). The general idea is that after experiencing the mood-augmenting and physically invigorating effects of exercise, one may adopt it for therapeutic purposes, such as coping with acute, unexpected, or chronic stress (Freimuth et al., 2011). This circle (Fig. 2) could lead to levels of exercise exceeding one’s psychological and physical limits and lead to physical, psychological, and social losses (Juwono and Szabo, (2020). Alternately, one may not recognize the physiological self-limits and push exercise over these limits, resulting in similar losses to those emerging via the therapeutic path (Dinardi et al., 2021). Hence, these two paths could lead to addiction, as explained by the pour phase model (Freimuth et al., 2011) and the interactional model (Dinardi et al., 2021; Egorov & Szabo, 2013). Therefore, researchers study exercise addiction as a form of dysfunctional exercise that harms the individual, and only its thorough understanding could lead to efficient methods of treatment and prevention.

Fig. 2
figure 2

The path to therapeutic and addictive exercise is based on the four-phase model proposed by Freimuth et al. (2011)

Full size image

Who Is Addicted to Exercise?

Based on over 20,000 papers in the area, the newcomer into the field would probably expect thousands of cases of exercise addiction. In contrast, Szabo and Demetrovics (2022) list only six (6) cases they could locate in scholastic literature. However, three were not clinically dysfunctional; two resulted in high sporting achievement, and one emerged as a solution to a life problem. Consequently, there were merely three (3) cases that the literature could categorize as meriting clinical attention. Outside academic works, however, there is a book with nine cases, including the case of the first author (Schreiber & Hausenblas, 2015). Furthermore, Juwono and Szabo (2020) searched the Internet until they located 100 testimonials or reports of cases that exhibited a minimum of one of the six symptoms listed by the components model of addiction (Griffiths, 2005) and the person suffered at least one physical, psychological, or social damage (i.e., some loss). However, these criteria are insufficient to presume clinical concern. Indeed, from Table 2 (pp. 1802–1805) in Juwono and Szabo’s work, one can see that out of 100 cases, only three (cases 48, 57, and 83) have reported experiencing the full spectrum of symptoms in the components model along with physical, psychological, and social losses. Thus, it seems that only very few regular exercisers might develop an addiction that demands clinical attention.

Research on Exercise Addiction

Given that regular exercise, planned physical activity, or sports should be incorporated into the increasingly sedentary contemporary lifestyle (Kapoor et al., 2022), the topic is widely researched in psychology, sports science, and medicine. Parallelly, the darker side of exercise behavior—resulting in negative consequences to the person—receives increasing research attention on the topics of exercise addiction and exercise dependence (Fig. 1). Most research in this area is cross-sectional. Such studies vary largely in their objectives and variables studied. Some look at secondary exercise dependence, which Szabo and Demetrovics (2022) define as instrumental exercise because exercise behavior is used to reach a non-exercise-related goal, such as weight loss (Godoy-Izquierdo et al., 2021). Others examine the connection between personality traits, such as perfectionism, and exercise addiction (Çakın et al., 2021). Moreover, some researchers suggest that exercise addiction should be studied along with passion and perfectionism (de la Vega et al., 2020). Then, many studies still look at exercise addiction in athletes despite being criticized on theoretical grounds, claiming that addiction is unlikely to occur in group or organized and scheduled environments (Juwono et al., 2021). Furthermore, a few neuropsychophysiological studies looking at the link between brain activities and exercise addiction are also reported (e.g., Gapin et al., 2009). Finally, numerous papers are theoretically or methodologically oriented (e.g., Griffiths et al., 2023; Szabo et al., 2015).

However, intervention studies also exist in the field. They examine the efficacy of various treatments for exercise addiction (Weinstein & Weinstein, 2014). These studies generally examine the effectiveness of cognitive-behavioral therapy, substitution therapy, cue exposure therapy, systematic desensitization, acceptance and commitment therapy, meditation, pharmacological intervention, and a combination of the above. For a review of the studies and a new treatment plan, the reader is referred to Szabo and Demetrovics (2022).

The Phantom Dysfunction

Theoretically, research considers exercise addiction a mental dysfunction, but that may be inaccurate. However, many recognize the semantic issue and use the term risk of exercise addiction (REA), which is more appropriate (Szabo & Demetrovics, 2022) because risk does not imply dysfunction. Thus, apart from less than a handful of studies presenting clinical cases associated with exaggerated exercise behavior, the over 20,000 papers published in academic journals deal with the phantom of exercise addiction, specifically an 'assumed' risk expressed as a number on a screening instrument lacking information on whether this number is related—or if it is, in what proportion—to eventual clinical morbidity. Szabo and Demetrovics (2022), after a deeper analysis, concluded that “… most of the evidence points toward exercise addiction being a symptom of another disorder, which might be the reason why insufficient evidence could be obtained for its inclusion in the latest edition (5th) of the Diagnostic and Statistical Manual of Mental Disorders (American Psychiatric Association, 2013) as a distinct category of mental disorder” (p. 194). Indeed, exercise addiction is not a psychiatric disorder based on the 5th edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) released in 2013, which assumes the following position: “…groups of repetitive behaviors, which some term behavioral addictions, with such subcategories as ‘sex addiction,’ ‘exercise addiction,’ or ‘shopping addiction,’ are not included because at this time there is insufficient peer-reviewed evidence to establish the diagnostic criteria and course descriptions needed to identify these behaviors as mental disorders” (American Psychiatric Association, 2013, p. 481). Furthermore, while the International Classification of Diseases (ICD-11) published by the World Health Organization (WHO) categorizes some behavioral addictions under the umbrella term of impulse control disorders, exercise addiction is not included in this classification.

It would be easy to add a point to the end of the sentence here and assume that exercise addiction is only a symptom of another mental disorder. However, the status quo seems to be more complicated than that. A few years ago, Colledge et al. (2020) conducted a meta-review focusing on symptoms of exercise addiction and their association with gambling disorder, the only behavioral addiction listed in the DSM-5. Their review concluded that several symptoms reported by individuals engaged in excessive exercise resemble symptoms of other addictive disorders. Thus, their findings appear to suggest that there is considerable evidence to further explore the concept of exercise addiction within the framework of behavioral addictions. More importantly, their analysis revealed an inconsistency between the risk of exercise addiction as assessed through subjective questionnaires and symptoms based on deep interviews. Although College et al. concluded that excessive exercise is likely a form of behavioral addiction, they recommended that future studies examine the presence of other co-occurring mental disorders and use specific interviews mapping excessive exercise patterns. Therefore, again, we are back to square one; the possibility that exercise addiction is a symptom of co-morbidity cannot be ruled out. To make things more complicated, excessive exercise also emerges as “instrumental exercise” (Szabo & Demetrovics, 2022, p. 129) in other dysfunctions like muscle dysmorphia or various eating disorders, both being listed in the DSM-5. Excessive exercise is instrumental in these dysfunctions because it helps achieve a body shape-related objective, such as a slim or muscular body. Thus, instrumental exercise surfaces as a symptom of another morbidity.

Questionnaires vs. Interviews

While several questionnaires assess REA, their reliability is questionable (for a detailed state-of-the-art analysis, refer to Alcaraz-Ibáñez et al., 2022b). In this paper, the examination of paper and pencil tools is avoided. Instead, it must be highlighted that none of them has a diagnostic value. Furthermore, they may not measure what they intend to measure. Indeed, a recent systematic review concluded that its findings raise doubts about (a) the accuracy and usefulness of the research evidence collected using these instruments and (b) whether it is productive to use them before future research yields more robust evidence on the measurement properties of the existing self-report instruments (Alcaraz-Ibáñez et al., 2022a).

Questionnaires are screening instruments; even the high scores should be followed up with interviews to gather insight into what is beyond the high score (Weinstein & Szabo, 2023). Questionnaires alone, with or without good psychometric properties or well-established cutoff scores, cannot be used for clinical diagnosis. However, researchers work with questionnaires and meet healthy exercisers who are unlikely to manifest any problems related to addictions. In contrast, if time permits, clinicians use interviews to determine the problem behind the mental dysfunction of a person seeking help. Here is the large gap overlooked in exercise addiction research. The thousands of academic papers rely on (mostly) healthy people, while the clinical cases remain inaccessible for researchers because clinicians are seldom interested in scholastic works; they do their job to treat people seeking help. However, these people are unlikely to surface in nomothetic exercise addiction research. If they do, their prevalence could be expected to be extremely small. In brief, research does not cross the road of clinical practice.

The above duality was long ago recognized by Szabo (2001). He proposed a model for collaboration between researchers and clinicians based on a method he termed the “Pyramid Approach.” According to the model, researchers can do surface screening at the population level using questionnaires (bottom layer of the pyramid). Subsequently, they can refer people with high REA to clinical practitioners (middle layer of the pyramid); however, note that in anonymous studies, this connecting step is not possible. At this phase, health professionals with clinical or medical training could follow up with the individuals at high REA using in-depth clinical interviews. As a result, they can separate those who exercise at very high volumes (like athletes) but maintain control over their exercise from those who have lost control over their exercise behavior and demonstrate maladaptive behavioral patterns. The latter can be examined for co-morbidities and then offered the established treatment for the dysfunction (top layer of the pyramid). The cases can then be published under pseudonyms with the patient’s consent. This way, the number of cases would probably accumulate (complementing the few published cases over the past 50+ years). These cases could project a better image concerning background dysfunctions or, alternately, the distinctiveness of exercise addiction. This method would clarify whether exercise addiction could be included in future clinical reference manuals.

Chasing the Phantom in Athletes

A very likely artifact of the literature is that the prevalence of the REA appears to be higher in athletes than in the general exercise population (Mónok et al., 2012). Juwono et al. (2021) conducted a literature review and found prevalence rates from 2.7 to 42.0% in 17 studies included in the systematic review. Despite their findings, the authors interpret their result as some artifact: “While non-athletes can engage in escape behaviors, including exercise when the urge arises, athletes train by following a rigidly (and often externally) controlled schedule that requires their personal life to be scheduled around their training regimen. Urges of addiction cannot be fulfilled on schedule. Therefore, these higher scores of risk of exercise addiction that were reported in the literature and are substantiated by the current literature review must reflect something else than pathological tendencies. Indeed, they may merely reflect keen passion, commitment, or dedication to the sport in which the athlete wants to excel, around which the athlete’s life revolves” (Juwono et al., 2021, p.3123). This argument was echoed by a recent position paper on exercise addiction in team-based exercises (Griffiths et al., 2023). The authors highlighted that addictions cannot be scheduled, and thus, athletics, team-based exercises, and all forms of workouts that are scheduled for an individual do not satisfy addictive needs. Still, addiction possibly cannot be ruled out in this context. However, if an individual in scheduled exercise or sports settings becomes addicted to exercise, then the person would need to succumb to the addiction above and beyond the scheduled training at an individual level, which is very unlikely in a high-level athletic environment characterized by long hours of hard training. Therefore, athletes are probably more likely to have other forms of addictions (if any) than exercise addiction.

Passion or Addiction?

Exercise addiction might be confounded with an intense passion for exercise. Several studies have examined the connection between obsessive and harmonious passion and REA. They suggest that exercise intensity and obsessive and harmonious passion could significantly predict the REA. Obsessive passion might be the strongest predictor in individuals with a long exercise history (Kovacsik et al., 2018). However, if statistical control is exerted for passion (for example, if it is used as a covariate), the group differences in REA disappear between high and low-frequency exercisers. These results were also corroborated by the de la Vega et al. (2020) in a large international study. Consequently, it appears that passion and exercise addiction share a large proportion of variance to the extent that various group comparisons yield negative findings when control for passion is exerted in the data analyses. Alternately, such findings suggest that what many studies project to be a REA might mirror a passion for exercise.

Conclusion

There is extensive research on exercise addiction. Apart from treatment interventions, most research is cross-sectional. The number of papers published in the field is out of proportion compared to problematic (clinically attention-demanding) cases. However, problematic cases exist (Juwono & Szabo, 2020; Schreiber & Hausenblas, 2015). Still, because their routes of origin vary greatly, the nomothetic (generalist) approach is unlikely to advance knowledge. The latter is over-pathologizing exercise addiction based on prevalence rates derived from paper and pencil tools (Mónok et al., 2012). Based on individual-oriented clinical models, a paradigm shift is necessary to build inductive knowledge from genuine problematic or clinical cases or nomothetic research that expands group results with individually followed-up deep interviews or other self-reports. A close collaboration between researchers and clinicians could greatly facilitate this new approach.

Issues and Possible Solutions

Issue: The different instruments used to study exercise addiction are counterproductive.

Solution: Develop a highly sensitive and specific tool based on deep interviews.

Issue: The ratio of published research papers to known clinical cases is extremely high.

Solution: Focus more on clinical or problematic cases and derive inductive knowledge.

Issue: Due to a lack of evidence, the DSM-5 does not classify exercise addiction as a disorder.

Solution: Scientists must cooperate with clinicians to learn from clinically problematic cases and gather evidence for the dysfunction that permits inclusion in clinical reference manuals (obviously, only if the evidence supports that).

Issue: Without follow-up interviews, questionnaires may lead to false generalizations and hinder the advancement of knowledge about exercise addiction.

Solution: Complement questionnaire data with follow-up self-reports or interviews.

Issue: To date, most evidence suggests that exercise addiction is a symptom of some other dysfunction rather than being a unique dysfunction itself.

Solution: Examine more closely the co-morbidities of exercise addiction.

Issue: While many studies are published in the field, little new knowledge emerges.

Solution: Before researching exercise addiction, identify the literature gaps and plan how to fill them systematically.

Issue: Clinical models, like the four-phase and interactional models, predict that exercise addiction is idiosyncratic.

Solution: Examine the commonalities and differences between genuinely problematic cases and generate inductive knowledge.

Issue: Nomothetic studies may over-pathologize a certain level of risk—assessed with paper and pencil tools—which is unlikely to turn into clinical dysfunction.

Solution: Researchers should not interpret (high) questionnaire scores as problematic unless they can support it with qualitative data from self-reports or interviews.

Issue: The current research methods and frameworks do not advance the knowledge of exercise addiction.

Solution: A paradigm shift is urgently needed and must involve shifting the focus from pathologizing excessive exercise to a more nuanced understanding that includes both positive and negative aspects of intense exercise involvement and yields significantly clearer segregation of the two behavior patterns.

Issue: Commitment, passion, and some personality factors appear to be strong covariates of the risk of exercise addiction.

Solution: In research on exercise addiction, researchers should always control the known covariates of exercise addiction.