Article Text
PDF
Abstract
Background The purpose of this study was to evaluate the effects of training load (TL) and well-being on injury and illness risk in youth soccer players.
Methods Throughout a 20-week season, 75 female adolescent soccer players reported mood, fatigue, stress, soreness, sleep quality, sleep hours, TL, injuries and illnesses. Well-being measures were recorded from −3 (worst) to +3 (best). TL was expressed as daily, weekly and monthly, as well as an acute:chronic workload ratio (weekly divided by monthly). Variables were compared between days with and without an injury, and with or without an illness. Poisson regression models were developed to predict daily injuries and illnesses using well-being and TL (z-scores) as predictors.
Results 36 injuries and 52 illnesses were recorded. Days with an injury had lower (worse) daily mood (1.24±0.2 vs 1.16±0.1, p=0.012) and higher daily TL (517±138 vs 440±158, p=0.010). Average monthly TL was higher preceding days with an illness (12 442 ±409 vs 12 627 ±403, p=0.043), while no differences were found with respect to other measures of TL or well-being. Worse daily mood (p=0.011, OR=0.012), higher daily TL (p<0.001, OR=1.98), and higher prior day TL (p=0.040, OR=1.34) were independent predictors of injury, while weekly (p=0.005, OR=1.50) and monthly TL (p=0.007, OR=1.54) were predictors of illness.
Conclusions Lower mood and higher acute TL are associated with increased injury risk, while higher chronic TL increases the risk of illness. Monitoring well-being and TL may facilitate intervention to reduce in-season injury and illness.
- Athlete
- Well-being
- Illness
- Training load
- Injury
Statistics from Altmetric.com
Introduction
Although the benefits of exercise are considerable, sport participation is also associated with an increased risk of injury.1 ,2 In particular, injury rates for adolescent soccer athletes have been reported to be between 2 and 4 injuries per 1000 hours.3–6 In fact, soccer has been identified as the high school sport with the greatest risk of injury for female athletes,7 and self-reported injury rates among female high school soccer athletes have been documented as high as 15.3 per 1000 hours.8 This incidence of injuries represents a significant health burden for these athletes and a threat to athletic success.9 Consequently, the identification of risk factors in youth athletes that can be modified before or during the athletic season can potentially reduce in-season injury in athletes.
Although high internal and external training load (TL) have been found to be positively associated with injury,7–11 recent research in adult athletes suggests that higher chronic TL may actually be protective against injury by promoting increases in physical fitness, muscular strength, and biomechanical adaptations.12–15 The combination of short and long-term TL may provide the most useful measure, as increases in the acute:chronic workload ratio in adult athletes beyond a certain threshold appear to increase injury risk in a dose-dependent manner.9 ,12–14 ,16–19 Although these studies have used global positioning system (GPS) measures of total distance covered as a measure of external TL to predict injury, a similar relationship was identified between increased injury and high acute:chronic workload ratios using internal TL in elite cricket bowlers.20 Children may adapt differently to comparable workloads than adults; however, little research exists regarding the relationship between TL and injury in youth athletes,4 ,21 ,22 particularly among female youth athletes.
It has also been suggested that subjective well-being, including sleep, may be a predictor of injury in athletes. Milewski et al23 found that decreased sleep was an independent risk factor for injury in middle and high school athletes, with a significantly higher injury risk among those athletes who slept <8 hours per night, on average. Although a recent study of adult team sport athletes found that perceived recovery and stress were predictive of in-season injury,21 a study of elite soccer players failed to identify a relationship between injury and psychosocial stress.24 These studies evaluated stress monthly or every several weeks; however, emotional states may fluctuate much more frequently. As a result, the true influence of subjective well-being on injury in athletes remains unclear, particularly among youth athletes.
Although the relationship between immune function and exercise in adult athletes has received considerable attention in the past, this relationship in youth athletes is virtually unexplored. In adult athletes, it has been suggested that higher volumes of exercise and impaired well-being may be associated with an increased susceptibility to viral infections.25–27 Illness in adult athletes from multiple sports has been associated with increased weekly TLs in some studies,26 ,28–30 but not others.31 ,32 However, there is a paucity of research in this area on youth athletes. A single study of male soccer players aged 15–18 years found that higher training duration, as well as several components of psychosocial stress were predictive of in-season illness.24 These effects were evaluated separately; however, they did not account for the potentially meaningful interaction between TL and well-being which could confound the relationships with illness.
Although prior research has evaluated the predictive ability of data that is collected or aggregated on a weekly or monthly basis,9 the identification of injury or illness risk on a daily basis may allow for a more timely intervention to promote athlete's health. For example, while acute increases in TL can negatively impact well-being,33 short-term reductions in TL can restore these measures, potentially averting the increased injury and illness risk.12 ,34 In addition, higher levels of subjective well-being appear to predict greater tolerance of an increased workload later on the same day.35 Although the interaction between TL and well-being potentially confounds their relationships with injury and illness, we are aware of no research which has attempted to evaluate the independence of these relationships in youth athletes. Therefore, the purpose of this study is to determine whether TL and/or subjective well-being are independent predictors of injury and illness in youth athletes.
Methods
Study design
All procedures performed in this study were approved by the Institutional Review Board of the University of Wisconsin—Madison. Seventy-five female youth soccer athletes (15.5±1.6 years, 164.7±6.6 cm, 57.3±8.2 kg) provided self-reported TL, well-being, injury and illness data throughout a 20-week soccer season. Each morning during the study period prior to any soccer events, athletes were asked to provide a daily rating of fatigue, mood, soreness, stress, and sleep quality on a −3 (worst) to +3 (best) Likert scale with descriptive text prompts, as well as sleep volume in hours, using an online software program (fitfor90.com). Immediately following all physical activity during this period, participants provided the duration (minutes) and intensity (1–10) for all activity, which were multiplied to yield a session-rating of perceived exertion (sRPE) value as a measure of internal TL.30 ,36 Throughout the season, injuries and illnesses were self-reported by the participants through the use of the same computer software. Although this specific software has not been cited in the literature previously, injury self-reporting among adolescent female soccer players has been reported previously.8 ,37 In accordance with the consensus statement on soccer injury registration definition of time-loss injury,38 participants were asked to report any injury that occurred during a soccer training or game and resulted in the athlete being unable to continue to participate. They were asked to provide the date of the injury (if not reported on the same day it occurred), the body part involved, and the mechanism (if known). Acute injuries were defined as those with a sudden onset during an identifiable event, while overuse was defined as having a gradual onset and unrelated to a specific event. Both first-time and repeat injuries were included if they were felt to represent new injuries based on resolution of symptoms and return to full participation between injuries. Participants were similarly asked to report any illnesses that resulted in restriction from participation in soccer events, including the date of onset and predominant symptoms. Follow-up interviews were conducted by the primary investigator on phone or in person in instances in which the details of the injury or illness were unclear. At the time of any injury or illness resulting in lost time, participants were encouraged to report them by the coaching staff. Compliance with the completion of daily TL and well-being ratings was encouraged periodically throughout the study period by the coaching staff but this was not done in a prescribed or systematic fashion.
Statistical analysis
Data were initially evaluated for normality using descriptive statistics and histogram analysis. For each day of the season, TL was aggregated as a daily sum and average for the entire group. In addition, 7-day (weekly) and 28-day (monthly) rolling averages for the group were calculated for each day, as well as an acute:chronic workload ratio (weekly divided by monthly). Measures of subjective well-being were calculated as a daily group average. In order to determine the influence of TL on well-being, Spearman correlation coefficients were determined between average daily well-being measures and TL from the preceding day, week, and month.
Well-being and average TL were compared between days with and without an injury during the season. Whereas injuries were entirely acquired during soccer events, symptoms of an acute illness may start anytime during the day reported. As such, well-being and TL measures from the day prior to the onset of symptoms were compared between days with or without a reported illness. Comparisons were made using Wilcoxon rank-sum tests for means and χ2 or Fisher's exact tests for frequencies, as appropriate, with adjustment for multiple pairwise comparisons as previously described by Holm.10 Effect sizes for pairwise comparisons were calculated using Cohen's d. TL variables were then converted to z-scores for inclusion in subsequent injury and illness prediction models. Since the number of injuries was not sufficient to support a large number of predictors in a single regression model, univariable Poisson regression models were developed to predict in-season injury and illness using total TL and well-being variables as predictors. In order to account for the influences of the TL from the same day as well as the effect of TL on well-being, daily total TL, prior day total TL, and mood were included in a multivariable Poisson regression model to predict in-season injury. Since only two highly related variables were found to be significant predictors of in-season illness, no additional multivariable analysis was conducted. Significance level was determined a priori at the 0.05 level, effect sizes were defined as small (0.2), medium (0.5), large (0.8), and very large (>1.0), and all tests were two-tailed. All statistical analyses were performed in R (R: A language and environment for statistical computing [program]. Vienna, Austria: R Foundation for Statistical Computing, 2015).
Results
Data were initially evaluated descriptively to determine missing data in TL and well-being variables. Compliance was found to be 84.7% overall and no obvious differences in missing data were found with respect to date, age group, or specific individual participants. As a result, missing values were ignored and all available data were included for analysis. Thirty-six injuries were identified on 29 days in 28 individual athletes, while 54 illnesses were recorded on 41 days in 33 individual athletes. All recorded injuries were acute, and the locations and types of injuries are shown in table 1.
In-season injury characteristics among adolescent female soccer players
Illnesses consisted of 36 (67%) respiratory infections, 10 (19%) gastrointestinal infections, 4 (7%) isolated fevers, 2 (4%) acute otitis media, 1 (3%) episode of conjunctivitis and 1 (3%) episode of body aches. The distribution of injuries and illnesses throughout the season are shown in figures 1 and 2, respectively. The average weekly TL during the intervention period was 3167±324 arbitrary units (AUs), corresponding to an average individual weekly volume of 6.7±1.9 hours and an average reported intensity of 6.3±0.5 out of 10. Total daily, weekly, and monthly TL, and acute:chronic workload ratio for the entire group are shown in figure 3. Significant inverse correlations were identified between multiple elements of subjective well-being and TL from the previous day, but not the previous week or month (table 2).
Spearman correlation coefficients between average daily well-being and training load (TL) from the prior day, week, and month in adolescent female soccer players
Number of weekly injuries during a 20-week season in adolescent female soccer athletes.
Number of weekly illnesses during a 20-week season in adolescent female soccer athletes.
Internal training load (TL) during a 20-week season in adolescent female soccer athletes. Top graph shows the daily (shaded grey area), weekly (7-day rolling average; dashed black line), and monthly (28-day rolling average; solid black line) TL. Bottom graph shows the acute:chronic TL as the ratio of weekly to monthly each day through the season.
Compared with days without a reported injury, days in which an injury had occurred were found to have significantly lower daily mood, as well as significantly higher TL from the same day (table 3). No significant differences were identified in other measures of well-being or with respect to TL from the prior day, week, month, or acute:chronic workload ratio.
Differences in well-being and training load on days with and without an injury in adolescent female soccer players
Daily mood, daily TL, prior day TL, weekly TL, and acute:chronic workload ratio were all found to be significant predictors of injury in the univariable Poisson regression analysis, and after inclusion in the multivariable model, mood, daily TL, and prior day TL all remained independent predictors of injury (table 4).
Univariable and multivariable Poisson regression analysis to identify predictors of in-season injury in adolescent female soccer players
Higher chronic (monthly) TL was found to precede days with a reported illness, while no differences were found with respect to well-being, daily TL or weekly TL, or acute:chronic workload ratio (table 5).
Differences in well-being and training load immediately preceding days with and without an illness in adolescent female soccer players*
None of the well-being variables were found to be significant predictors of illness, while higher preceding weekly and monthly TL were significant predictors of in-season illness (table 6).
Univariable Poisson regression analysis to identify predictors of in-season illness in adolescent female soccer players
Discussion
TL, well-being and injury
The primary finding of this study was that acute TL and mood are independent predictors of injury in youth female soccer players. While the increased injuries associated with higher daily TL are consistent with prior studies,39 ,40 we also found that higher TL from the previous day exerted an independent effect on the risk of injury. Even though several measures of well-being were found to be negatively impacted by higher acute TL, the effect of the prior day's TL on injury persisted after accounting for the effect of mood. Previous research has suggested that an increase in z-score of 1 for acute TL represents the difference between a moderate–low and moderate–high training workload.14 Using the results of the multivariable model, an increase of this amount in same day TL or prior day TL would be accompanied by an increased injury risk of 98% and 38%, respectively. In addition, these effects are independent, suggesting that the physiological effects of high TL carry over and translate into increased injury risk the following day, and high TL on consecutive days may result in a cumulative increased risk of injury.
We also present the novel finding that decreased daily mood was an important predictor of in-season injury, even after controlling for the effects of acute TL. Although prior studies have found that measures of well-being can predict injury risk in adults, far less information is available in children.23 ,40 A prior study found that retrospective accounts of average sleep were predictive of injury in adolescent athletes,23 but we did not find any relationship between injury and sleep quality or quantity from the preceding night. Two prior studies of adolescent athletes found that overreached players demonstrated impairments in several measures of psychosocial well-being compared with controls up to 2 months prior to the diagnosis of overreaching.41 ,42 In a similar study of adolescent male soccer players, Brink et al24 failed to demonstrate the ability of these measures to predict injury. In these studies, stress and recovery were measured every 3 weeks,41 monthly,24 or only at the time of diagnosis of overreaching.42 Mood states may fluctuate much more rapidly than this, and our data demonstrate the clear and immediate influence of acute TL on several measures of well-being. It is possible that acute changes in mood can lead to poor decisions during training and competition, potentially making athletes vulnerable to injury. Daily measurement of well-being and TL therefore may allow for a more precise measurement of this relationship during the season. While this was not an interventional study, it is feasible that monitoring of daily well-being and recent prior TLs may allow for immediate modification to upcoming TL to reduce the risk of injury.
The influence of chronic TL on injury was less clear. Although we did not identify any relationship between injury and monthly TL, acute:chronic workload ratio was a significant predictor of injury. This is similar to the relationship that has been consistently identified in adult athletes in multiple sports,13 ,14 ,43 ,44 and previously suggested as a useful metric to determine readiness to return to play.45 Although these studies suggested a protective effect of higher chronic TL, we did not find a positive or negative influence of monthly TL on injury. Similarly, although these studies suggested that acute:chronic workload ratio provided a better indicator of injury risk than acute TL alone, we were not able to differentiate the relative predictive ability of these two measures in the current study. It should be noted that the acute:chronic workload ratios identified in this study were generated from measures of internal rather than external workload, and varied far less than in these prior studies.
TL, well-being, and illness
In this study we also found that higher chronic TL was predictive of illness in youth soccer players. Specifically, an increase in weekly and monthly TL was associated with a significant increase in illness, while no relationship was identified with daily TL or acute:chronic workload ratio. As stated above, it has previously been suggested that a z-score difference of 1 represents a meaningful difference in workload for both acute and chronic TLs.14 Based on the results of the current study an increase of this amount in weekly or monthly TL would be accompanied by an increased illness risk of 50% and 54%, respectively. This is similar to studies in adult athletes which identified increased illness risk during times of increased weekly TL,26 ,28 ,29 ,31 and a recent systematic review which found a similar relationship in the majority of included studies of adult athletes.9 Our findings are also consistent with the only prior study of TL and illness in youth athletes of which we are aware found that increased training duration was associated with illness risk.24 We did not, however, find a significant relationship between illness and daily TL, weekly TL, or acute:chronic workload ratio and we are not aware of any prior studies which have examined these relationships. This suggests that illness risk is likely a function of accumulated effects of higher TLs over the preceding weeks, without a protective effect of higher chronic TL.
We did not find that measures of well-being were predictive of illness. Although prior research has identified decreased sleep as a consequence of overtraining and a possible risk factor for illness in athletes,12 we failed to identify a similar relationship. In addition, recent studies of collegiate football players46 and adolescent male soccer players24 found that illness risk was associated with increases in psychosocial stress. These prior studies have evaluated the effects of well-being over longer periods; however, it is possible that the increased risk of illness is due to cumulative effects over time which we did not evaluate in the present study. This remains an important area of inquiry, as risk-factor identification and interventions to reduce the burden of infectious disease have important implications for overall health and athletic success.
This study has several limitations. Although self-reported injuries resulting in time loss have been used in a prior similar research,8 it is possible that not all injuries were reported during the study period. In addition, while we achieved relatively high compliance, we did have missing data. While this was felt to be random within the data set, we did run a separate analysis after imputation of the missing data that yielded virtually identical results. As a result, we felt it was reasonable to present the data without imputation of missing values. It should be recognised that additional risk factors for injury and/or illness such as anatomic differences, body composition, fitness level, or growth rate were not accounted for in our analysis. In addition, a larger sample size in future studies will allow for the evaluation of additional risk factors and their interactions, as well as separate prediction models for contact and non-contact injuries. Finally, this study was conducted among a group of elite, adolescent female soccer players and may not be generalisable to other populations.
Conclusion
This study demonstrates that acute TL and mood are independent predictors of injury, while high chronic TL is associated with an increased risk of illness in adolescent youth soccer players. Whereas the avoidance of chronically high TL may minimise in-season illness risk, the use of daily ratings of TL and well-being may allow for the identification of increased risk of injury on an individual or group level. For example, tracking of daily TL can help avoid consecutive days of high internal TL which can impair well-being but also increase risk of injury. In addition, proactive monitoring of measures of well-being such as mood may allow better identification of individuals or groups at risk for injury that might otherwise have remained unknown. While the majority of prior research has evaluated the predictive ability of data that is collected or aggregated on a weekly or monthly basis,9 these results suggest that information provided by athletes on a daily basis may be useful to identify risk. Finally, the inclusion of additional risk factors in larger evaluations across multiple sports remains an important area of future research to allow for the identification of independent, modifiable, sport-specific and gender-specific risk factors that can be used to guide intervention to improve athlete health.
What are the findings?
Decreased mood and higher acute, but not chronic, training load are independent predictors of in-season injury.
Higher weekly and monthly training load are associated with increased in-season illness.
Acute training load has an immediate, negative impact on several measures of well-being.
How might it impact clinical practice in the future?
Monitoring of daily subjective ratings of well-being and training load may allow for intervention to promote athlete health.
References
Footnotes
Twitter Follow M Alison Brooks @DrABrooksUWisc
Contributors AW conceptualised the study design; participated in the data collection, data analysis and interpretation; drafted the manuscript and approved the final manuscript as submitted. SB contributed to the study design; participated in the data collection, data analysis and interpretation; reviewed the manuscript and approved the final manuscript as submitted. AB contributed to the data analysis and interpretation; reviewed the manuscript and approved the final manuscript as submitted. WD contributed to the study design; analysed and interpreted the data; reviewed the manuscript and approved the final manuscript as submitted.
Competing interests None declared.
Ethics approval Institutional Review Board of the University of Wisconsin—Madison.
Provenance and peer review Not commissioned; externally peer reviewed.