While athletic adaptation and the development of adequate physical fitness occur only under training stimuli that substantially stress the body’s functional systems, excessive exertion may exhaust these systems, resulting in maladaptation [1]. Therefore, to achieve high athletic performance, it is essential to maintain a balance between the external load and the functional capacity. In the scientific literature, conditions arising when workloads surpass this threshold are conventionally divided into several categories:

Brel et al. note that the use of the term “syndrome” in relation to the overtraining state is justified by its multifactorial nature. Among the key contributing factors, they highlight excessive training stress and the athlete’s individual biological, neurochemical, and hormonal characteristics [4]. According to the authors, the difference between overreaching and overtraining syndrome lies in the time required for performance recovery, rather than in the duration of the training load itself [4].

To date, a unified framework for the early diagnosis of athletic overtraining has not been established. Researchers note that overtraining syndrome can only be considered retrospectively, as a certain amount of time is required for this condition to fully manifest. Furthermore, the deliberate induction of this state is unacceptable, as it is unethical towards the athlete who strives to achieve high performance [5].

The findings of both retrospective and cross-sectional studies investigating overtraining often contradict each other. This inconsistency is due to differences in the applied methods, selected markers, and the characteristics of the studied athlete cohorts. To identify diagnostically informative metrics of overtraining while accounting for the methodological heterogeneity across existing studies, it is necessary to assess the frequency with which methods and markers are employed, as well as the consistency of shifts in these metrics in this condition.

Previously published systematic reviews on athletic overtraining have focused on specific aspects: biomarkers and diagnostic tools [2]; physiological and psychological changes [5]; cognitive impairments in elite athletes [6]; heart rate variability (HRV) in football players; and hormonal biomarkers [7]. The present study aims to provide a comprehensive analysis of the consistency of research findings across the main categories of overtraining markers (biochemical and endocrine; physiological; psychometric; morphological; and molecular genetic and immunological markers).

The study aims to assess the consistency of research findings on overtraining markers in athletes.

The systematic review was conducted in accordance with the PRISMA guidelines1 [6]. The study protocol was not prospectively registered in a systematic review registry.

Search strategy. The literature search was conducted using the Google Scholar search engine and the bibliographic databases PubMed/MEDLINE, RSCI (eLIBRARY), and CyberLeninka. The search covered all studies published between 1 January 2006 and 18 May 2026 in Russian and English. The following search terms were used:

The inclusion criteria for the systematic review were established based on the PICOS (Population, Intervention, Comparison, Outcomes, Study Design) framework2:

Exclusion criteria were as follows:

The inclusion and exclusion criteria were applied at all stages of the selection process: first during the screening of titles and abstracts, and subsequently during the review of full-text articles. If the abstract provided incomplete information, the full-text version of the article was screened.

Study selection procedure. Two authors independently screened all titles and abstracts in the pre-specified databases for compliance with the inclusion and exclusion criteria. Duplicate articles and studies not meeting these criteria were removed. At the second stage, both authors independently retrieved the full texts of the articles and assessed them for eligibility. Articles that did not meet the inclusion criteria were excluded from further consideration. In case of discrepancies or partial compliance of a publication with the eligibility criteria, a third author was consulted to resolve the issue, with the final decision reached by consensus among all three researchers. To minimize the risk of publication bias, the analysis included studies reporting both positive and negative results.

Synthesis of evidence. The study was conducted in accordance with the Synthesis Without a Meta-Analysis (SWiM) guidelines3. Due to the heterogeneity across study designs and evaluated sports, a meta-analysis was unfeasible; consequently, the vote-counting method was used. The direction of the effect was assessed, with results considered statistically significant at p < 0.05 and statistically non-significant at p > 0.05. For each marker, the statistically significant direction of change was recorded (increase, decrease, or no change; significance threshold: p < 0.05). Consistency was defined as the proportion of studies reporting a unidirectional statistically significant change in the marker during overtraining development, relative to the total number of studies that evaluated this marker. Studies reporting statistically non-significant changes (p > 0.05) were considered to indicate “no effect”. When counting the number of effect directions, studies with different sample sizes were assigned equal weight. This methodological limitation is inherent to the vote-counting approach and was taken into account during data interpretation.

To assess the heterogeneity of results, a qualitative analysis was performed: the directions of effects for each marker were compared, and, if discrepancies were observed, potential causes were investigated. The risk of bias was assessed using the ROBINS-I tool for non-randomized studies4. The selected articles were grouped by diagnostic method to evaluate the informativeness of each category: biochemical and endocrine markers, physiological markers, psychometric and subjective indicators, morphological markers, and molecular genetic and immunological markers. If a single study provided data for multiple categories, its results were included in each relevant category. To evaluate the certainty of evidence for each marker, the GRADE (Grading of Recommendations Assessment, Development and Evaluation) approach was applied5. Five factors that could reduce the certainty of evidence were considered: risk of bias, inconsistency, indirectness, risk of publication bias, and imprecision. An effect direction plot was constructed to visualize data trends across the employed methods6.

The article selection process is illustrated in the figure. Initially, 1976 articles were identified. After removing duplicates and studies that did not meet the inclusion criteria, 172 articles remained. Of these, 22 full-text articles met all inclusion criteria and were included in the present study (Table 1).

When interpreting the findings, the limited amount of data available for certain markers was taken into account. It is important to note that no randomized clinical trials were included among the studies reviewed. Risk of bias assessment using the ROBINS-I tool revealed the following distribution: of the 22 included studies, 18 (81.8%) were characterized as having a moderate risk of bias, 3 (13.6%) as having a high risk of bias, and 1 (4.5%) as having a low risk of bias. Furthermore, 10 studies (45.5% of the total) showed a moderate risk of bias due to the influence of additional confounding factors.

This is attributable to the fact that the studies insufficiently accounted or adjusted for external influencing factors, such as energy deficit and nutritional disorders, psychosocial stress, sleep problems, and latent diseases. Additionally, a high risk with respect to this criterion was recorded in three studies (13.6%).

Therefore, the results should be interpreted with caution: the aforementioned external factors may influence several parameters — in particular, cortisol levels and athletes’ subjective perceptions.

Analysis of the dynamics of overtraining markers (Table 2) revealed a consistent direction of change for most markers upon the onset of overtraining: a decrease in maximal heart rate (HRmax) was observed, along with deterioration in emotional state and mood (assessed using the POMS questionnaire), and a reduction in muscle mass. Notably, the consistency of these findings was independent of the observation duration. Among the studies reviewed, three groups can be distinguished based on duration: short-term studies (1–10 days), which employed a comprehensive examination protocol; medium-term studies (3–14 weeks), involving comprehensive examinations with monitoring of overtraining marker dynamics; and long-term and retrospective studies (3 months–14 years), examining the evolving processes associated with a chronic condition.

These varying study durations enabled the diagnosis of specific conditions in each group: in the first group, functional overreaching was diagnosed; in the second group, non-functional overreaching was identified; in the third group, overtraining syndrome was confirmed.

Synthesis of the results and diagnostic value of the markers. Following the analysis of the selected articles, all overtraining markers were classified into five groups based on their physiological characteristics (Table 3). The consistency of 30 overtraining markers in athletes was evaluated using the vote-counting method. While this metric does not directly reflect the diagnostic performance of these markers, it may justify their inclusion in prospective studies designed to assess their diagnostic accuracy. High consistency was demonstrated by psychometric and subjective markers, as well as morphological, molecular genetic, and immunological markers. It should be noted that the high consistency (100%) observed for 11 out of 30 markers may be attributable to the fact that data for these markers were derived from a single study (n = 1). Combined with the low certainty of evidence as assessed by the GRADE approach, these findings necessitate confirmation in further studies. Divergent changes during overtraining were observed for the following markers: testosterone, cortisol, creatine kinase, lactate during exercise, anabolic index, and maximal oxygen uptake.

This systematic review demonstrates that assessing athletic overtraining requires a comprehensive set of methods and markers designed to evaluate the functional state of various physiological systems, as overtraining syndrome and the preceding states of overreaching affect endocrine, immune, metabolic, and psychophysiological processes.

Athletes in different sports exhibit variable adaptive responses, which can be attributed to the specific demands of each sport. In particular, the time course of overtraining development and its clinical manifestations may differ substantially [5]. Therefore, pooling athletes from different sports may oversimplify the assessment of overtraining.

Nevertheless, the current review intentionally included studies examining cyclic, strength, and team sports. This is because, despite the differences in the processes leading to overtraining, there are markers that remain informative regardless of the specific sport. For example, overtraining syndrome in weightlifters and distance runners shares common manifestations, including an inability to achieve supercompensation, mood disturbances, and depletion of adaptive reserves. Therefore, examining a broad range of studies enables the identification of the most robust markers with high result consistency, irrespective of sport type or load modality. For instance, decreased testosterone levels were observed in both strength and team sports, while mood deterioration, assessed via the POMS questionnaire, was reported in athletes participating in both strength and endurance disciplines. The effectiveness of this literature analysis approach is supported by the large-scale EROS study (Endocrine and Metabolic Responses on Overtraining Syndrome), which demonstrates the validity of pooling athletes from different sports. This is because similar physiological changes occur during overtraining development across athletic populations — for example, an increase in estradiol levels in males and a decrease in the neutrophil-to-lymphocyte ratio (NLR) [21].

Biochemical and endocrine markers. Cortisol level assessments yielded inconsistent results: in 4 out of 9 studies, an increase in the parameter was reported, while an equal number of studies documented a decrease. This bidirectional trend indicates low marker stability, making it inappropriate to use cortisol as a standalone indicator for overtraining diagnosis. The variability in cortisol levels is driven by multiple external and internal factors, including physical load, sleep quality, and stress exposure. The nature of the load itself is also important: inadequate control of its volume and intensity leads to a heterogeneous hormonal response, which complicates data interpretation [30]. According to the GRADE assessment, the certainty of evidence regarding cortisol as an overtraining marker is rated as very low.

The anabolic index (testosterone-to-cortisol ratio) is considered a meaningful criterion, with a data consistency level of 80%. However, when assessing its reliability, it is important to consider that salivary analysis is less informative than blood plasma testing [31]. Therefore, the GRADE certainty of this metric is rated as low.

In males with overtraining, a decreased testosterone-to-estradiol ratio is observed. This reflects structural maladaptation of the endocrine system and supports considering this parameter as a more reliable marker of overtraining.

Creatine kinase (CK) is a classical marker of muscle tissue damage; however, its application in overtraining evaluation must account for the specific sport discipline, as CK levels may vary across different athletic populations [9][20]. In speed-strength athletes, CK elevation can exceed 2600 U/L due to the intensity of protein metabolism [9]. In CrossFit, CK levels may vary within a wide range of 122–1650 U/L and do not necessarily differ from changes observed in non-overtrained athletes [20]. Due to its dependence on the type of muscle activity, low result reproducibility, and the fact that this marker reflects acute muscle damage rather than a chronic condition, the GRADE certainty of CK as an overtraining marker is very low.

Physiological markers. Analysis of HRV parameters demonstrated the highest consistency across two studies (Tables 2 and 3). These studies reported reduced HRV and a predominance of the sympathetic “stress response” [12][26]. Aerobic performance indicators showed lower consistency: in overtrained athletes, the anaerobic threshold occurred at a lower power output compared to non-overtrained athletes [29]. Athletes exhibiting signs of overtraining had difficulty reaching peak lactate levels during exhaustion tests, indicating a disturbance in anaerobic glycolysis as an energy supply mechanism. Regarding maximal oxygen uptake (VO2 max), this parameter was found to be reduced in only 1 out of 3 studies; no significant deviations were detected in the other two. The GRADE certainty of these results was assessed as very low.

HRmax is a primary indicator of exercise intensity; however, in overtrained athletes, this parameter was reduced during an exhaustion test [29]. This may be attributable to decreased sensitivity of cardiac β-adrenergic receptors to catecholamines — a protective mechanism that limits cardiac work [32] — or to a general suppression of the sympathetic response. In four studies (Table 2), an integrated marker combining two parameters — the HRmax/lactatemax index — was used to assess overtraining. A reduction in this index during submaximal exercise is associated with the overtrained state. This parameter is characterized by a moderate level of GRADE certainty.

Changes at the level of the central nervous system (CNS) can be used for early diagnosis of overtraining, preceding metabolic disturbances [9]. In five publications (Tables 2 and 3), speed, strength, and power tests demonstrated high consistency: overtrained participants exhibited reduced maximal squat strength, slower 5-km running speed, and lower average power output during a 1-hour race [14][15][23]. These alterations indicate impaired inhibition and excitation processes in the CNS, which supports the utility of such tests for field-based early detection of overtraining. However, speed-strength tests are characterized by a very low level of certainty. This low rating stems from the heterogeneity of test protocols across the studies: these included maximal squat strength; a maximal incremental treadmill running test; a standardized exhaustion treadmill test; speed-based squats and leg extensions; and a 5-km time-trial cycling test.

Psychometric and subjective indicators. The Profile of Mood States (POMS) questionnaire remains the most frequently used tool for overtraining assessment, demonstrating the highest consistency across all studies (Tables 2 and 3). It revealed negative mood profile changes — increased depression and fatigue, and reduced vigor — indicating an overtrained state. Notably, 3 out of 6 studies belonged to a single research project. An increase in the rating of perceived exertion (RPE) during exercise is considered the gold standard for detecting changes in subjective sensations associated with overtraining [32]. In the current review, RPE was employed in only one study, where it demonstrated the highest consistency [30]. Another tool for subjective assessment of the overtrained state is the Recovery-Stress Questionnaire for Athletes (RESTQ-Sport), which showed the highest consistency across three studies (Tables 2 and 3). In athletes exhibiting signs of overtraining, an increase in stress scale scores and a decrease in recovery scale scores were observed. The French Society of Sport Medicine questionnaire (QSFMS) was included in three publications (Tables 2 and 3) and demonstrated the highest consistency. Subjective scales and questionnaires frequently correlate with objective measures, which is supported by a moderate level of GRADE certainty in the current study. Therefore, their use — similar to speed-strength performance tests — enables field-based diagnosis and early detection of impairments.

Morphological markers. In three studies, despite a high training volume, a reduction in muscle mass and an increase in fat mass were observed (Tables 2 and 3). The EROS studies demonstrated that these morphological changes are accompanied by a decrease in basal metabolic rate (BMR) and slower fat oxidation [18]. Unlike biochemical markers (e.g., cortisol or lactate), morphological parameters are considered more precise indicators of overtraining because they reflect long-term metabolic alterations [18]. However, since the analyzed morphological data originated from a single research project, their GRADE certainty was rated as low.

Molecular genetic and immunological markers. Extracellular DNA markers address the issue of marker non-specificity with respect to the sport type, as they enable assessment of the overtrained state at the molecular level. The main limitation of their application lies in the cost of the equipment required for the analysis [15]. NLR demonstrated the highest consistency across two studies (Tables 2 and 3), conducted within a single project but involving different subject samples [21]. The utility of this marker lies in its ability to reflect adaptive changes in the immune system. Under normal conditions, training loads in athletes stimulate an increase in blood lymphocyte counts compared to non-athletic individuals. However, with the development of overtraining, the athlete-specific immune profile is disrupted. Despite the high consistency of these markers, they received a lower GRADE rating due to the risk of bias arising from confounding factors.

The study was not prospectively registered in international registries; however, the PRISMA guidelines were strictly followed during its conduct. The majority of the studies included in the analysis involved male athletes, and in several publications, information on the participants’ sex was not reported — this limits the generalizability of the findings to the female population. Methods for assessing the metabolic index varied across studies: some authors relied on blood plasma analysis, whereas others used saliva testing, and this discrepancy may have influenced the detection of effects. Study protocols also differed substantially in duration — ranging from 1 day to 14 years — which complicates the differentiation between functional and non-functional overreaching and overtraining syndrome.

When interpreting the results (Table 3), it should be noted that conclusions for a number of markers are based on a limited number of studies (36.7%). In such cases, this reflects not so much high consistency as the scarcity of available evidence. Particular caution is warranted when applying these markers, and further validation is required before they can be reliably used in practice. It is also important to note that several markers demonstrating high consistency originated from a single project, the EROS study [17][20][21], which was included in the analysis due to the diversity of markers it encompassed. The observed high consistency of these markers may be attributable to methodological homogeneity across the studies. These parameters include: muscle mass, BMR, NLR, testosterone and cortisol concentrations, and the POMS questionnaire. Clinical and methodological heterogeneity among the included studies may have introduced additional heterogeneity into the results and affected their consistency.

According to the GRADE assessment, none of the studies were classified as high certainty (Table 3). The certainty level for most markers was rated as moderate (40%), low (30%), or very low (30%). This is attributable both to the impossibility of conducting randomized trials due to ethical constraints and to other methodological limitations, including small sample sizes, uncontrolled external factors, and non-standardized sample collection procedures. However, the high consistency across studies for several markers (Table 3) supports their potential use as diagnostic tools for the overtraining state — when applied in combination with other methods. Nevertheless, the low certainty of evidence according to the GRADE assessment indicates the need for further validation of the proposed overtraining markers in large prospective studies.

Based on the conducted review and the observed consistency of studies regarding the applied methods, we propose a two-tier approach to diagnosing overtraining in athletes:

Both coaches and other professionals involved in diagnosing an athlete’s overtraining state should account for external factors that may substantially influence test results, including sleep, stress, illness, and the nature of training sessions performed on the day prior to assessment.

This systematic review classified potential markers for assessing athletic overtraining into five categories: biochemical and endocrine, physiological, psychometric and subjective, morphological, and molecular-genetic and immunological markers. The highest consistency was shown by the psychometric and subjective markers (POMS: 6/6 studies; RESTQ-Sport: 3/3; QSFMS: 3/3). The remaining markers demonstrated lower consistency (cortisol: 4/9 studies; testosterone: 5/8; VO2 max: 1/3; anabolic index: 4/5; lactate during exercise: 3/4; CK: 3/4) or exhibited high consistency but were supported by a limited number of studies (n = 1) or derived from analyses of studies conducted within a single research project. According to the GRADE assessment, 18 out of 30 markers were assigned a low or very low level of certainty.

The obtained data indicate the potential value of the identified markers for diagnosing the overtraining state; however, their application is advisable only within a comprehensive approach that involves the simultaneous assessment of markers from different functional systems. Furthermore, further prospective studies are required to standardize diagnostic protocols and evaluate the diagnostic accuracy of overtraining markers, including athletes of different sexes and training levels. Achieving an objective evaluation of overtraining status, however, requires not only monitoring marker levels but also accounting for external factors that may influence the athlete’s condition, such as sleep, stress, psychosocial factors, and caloric intake restrictions.

Authors’ contributions. All authors confirm that their contributions meet the ICMJE criteria for authorship. The primary contributions are distributed as follows: Fanis A. Mavliev — conceptualization and planning, literature search, data extraction and systematization, manuscript preparation; Adelia Sh. Abdrakhmanova — literature search, data extraction and systematization, manuscript drafting; Vladislav R. Karfik — data extraction and systematization, manuscript review and editing; Timur V. Sabirov — participation in the discussion of results, manuscript review and editing; Firuza R. Zotova — conceptualization, methodology, manuscript preparation, manuscript review and editing.

1. Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. PLoS Medicine. 2021;18(3):e1003583. https://doi.org/10.1371/journal.pmed.1003583 2. Methley AM, Campbell S, Chew-Graham C, McNally R, Cheraghi-Sohi S. PICO, PICOS and SPIDER: a comparison study of specificity and sensitivity in three search tools for qualitative systematic reviews. BMC Health Services Research. 2014;14:579. https://doi.org/10.1186/s12913-014-0579-0 3. Campbell M, Mckenzie JE, Sowden A, Katkireddi SV, Brennan SE, Ellis S, et al. Synthesis without meta-analysis (SWiM) in systematic reviews: reporting guideline. BMJ. 2020;368:l6890. https://doi.org/10.1136/bmj.l6890 4. Sterne JAC, Hernán MA, Reeves BC, Savović J, Berkman ND, Viswanathan M, et al. ROBINS-I: A tool for assessing risk of bias in non-randomised studies of interventions. BMJ. 2016;355:i4919. https://doi.org/10.1136/bmj.i4919 5. Schünemann HJ, Brennan S, Akl EA, Mustafa RA, Langendam M, Dahm P. The development methods of official GRADE articles and requirements for claiming the use of GRADE–A statement by the GRADE guidance group. Journal of Clinical Epidemiology. 2023;159:79–84. https://doi.org/10.1016/j.jclinepi.2023.05.010 6. Thompson HJ, Thomas S. The effect direction plot: visual display of non-standardised effects across multiple outcome domains. Research Synthesis Methods. 2013;4:95–101. https://doi.org/10.1002/jrsm.1060