Training Load and Injuries in Volleyball: An Approach Based on Different Methods of Calculating Acute to Chronic Workload Ratio

Abstract

Background:

Many questions persist regarding the relationship between training load and injuries in volleyball, as well as the best method for calculating acute:chronic workload ratio (ACWR). This study aimed to investigate the relationship between different metrics of training load and risk of injury in male professional volleyball players.

Hypothesis:

ACWR, as a training load measure, is useful for identifying injury risk in volleyball players, regardless of calculation method.

Study Design:

Longitudinal, prospective, and observational design conducted over 3 seasons of professional male volleyball.

Level of Evidence:

Level 3.

Methods:

The study included 43 male volleyball players. Internal training load was quantified using the Session Rating of Perceived Exertion. From daily training load values, absolute measures and relative measures were computed. For relative measures, 7 days were employed for acute training load, and 21 and 28 days for chronic training load. A distinction was made between coupled calculation and uncoupled calculation. Injuries were documented using the Injury Surveillance Form proposed by the International Volleyball Federation.

Results:

ACWR calculated in a coupled manner and by a rolling average demonstrated higher injury risks when analyzing the complete periods (odds ratio [OR] ACWR 7:28 = 2.040; ACWR 7:21 = 1.980) and competitive period (OR ACWR 7:28 = 2.044; ACWR 7:21 = 2.087). In contrast, during the preseason, the coupled exponential averages were more sensitive to the risk of injury (OR ACWR 7:28 = 4.370; ACWR 7.504).

Conclusion:

Both measures using rolling averages and those calculated from exponential averages can be employed to identify the risk of injuries in volleyball athletes.

Clinical Relevance:

The findings of this study can be useful to coaching staff, fitness trainers, and healthcare professionals involved in the challenge of reducing the risk of injury in volleyball athletes. The need for continuous monitoring and real-time adjustments of training load is emphasized.

Sports injuries manifest in a multifactorial nature involving both extrinsic and intrinsic factors. ¹² Within these determinants, evidence has pointed to training load as a significant risk factor.^{2,12,24,41,45}

Early studies on the relationship between training load and sports injuries aimed to establish a direct dose-response correlation between these variables, indicating that a higher magnitude of training load corresponded to an increased risk of injury.^1,18 It is now recognized that excessively reducing the training level may increase the risk of injuries due to inadequate athlete fitness. ¹⁶ Similarly, maintaining an appropriately elevated training load over a period can lead to positive physical adaptations, thereby reducing the risk of injury. ²⁷ Another aspect related to increased risks is the occurrence of abrupt load variations, referred to as spikes. ²⁵ A substantial portion of the current scientific literature on the subject suggests that variations in training load may be more closely associated with injuries than the absolute load.^2,41

The change in understanding about training load behavior, considering not only its absolute value but also its variations, revisits a concept introduced by Banister et al. ⁵ Based on this concept, a ratio between acute training load and chronic training load was proposed, termed acute:chronic workload ratio (ACWR). ²² It is calculated by dividing the training load an athlete has recently completed by the load they have accumulated over a longer period, with the commonly utilized time frames in studies being 7 days for acute load and 28 days for chronic load.^24,35 Some authors have suggested that ACWR could be a valuable measure to help practitioners progress training load while minimizing injury risk.^8,24,26

Although ACWR is an essential tool for training load management, its validity has been questioned.^29,36 In response to some of these initial concerns, a new approach was introduced called exponentially weighted moving averages (EWMA). This method uses weighted averages based on a calculation that assigns higher weights to the days closer to the analyzed event. ⁴⁶ This measure has been identified as more sensitive to sports injuries; however, further research on the subject is still needed.^24,35

A considerable portion of studies that aim to analyze the relationship between training load and injuries focuses on contact sports, such as rugby,^20,23 Australian football,^9,43 and soccer.^3,8 However, sports with lower physical contact demands, such as volleyball, may benefit more from an injury prevention program based on training monitoring. ¹⁷ Despite this, there is still a lack of research on the relationship between training load and injuries in volleyball. Some studies have correlated external training load variables (eg, training hours, jump frequency, number of sets played) with the occurrence of a specific injury (eg, jumper’s knee).^4,44 A study analyzed the risk of injury based on internal and external training load in volleyball, indicating that athletes who did not sustain injuries had lower ACWR values compared with other groups that experienced injuries. Similarly, the probability of injuries was higher in the presence of high ACWR. ⁴² A recent study examined different ACWRs to predict knee pain in elite volleyball athletes. They used 5 different ACWRs and indicated that these variables had only a small (or no) effect on knee pain scores. ¹¹ Three systematic reviews on the relationship between ACWR and sports injuries have been published recently,^2,24,35 yet only Griffin et al ⁴² cited a study on volleyball.

Therefore, many questions persist regarding the relationship between training load and injuries in volleyball, as well as the best method for calculating ACWR. Points requiring clarification include the use of rolling averages or exponential averages, the period used for acute and chronic loads, coupled or uncoupled calculations, and the use of internal or external training load.^10,21 Thus, this study aimed to investigate the relationship between different metrics of training load and the risk of injuries in male professional volleyball players.

Methods

Subjects

The study included 43 male volleyball players (age, 23.0 ± 5.2 years; height, 192.9 cm; weight, 88.9 ± 12.2 kg) members of a team competing in national-level tournaments in Brazil over 3 seasons (2 seasons in the Brazilian “Superliga,” 1 season in the Brazilian National League B). Athletes who remained with the team for a period of ≤3 months of their respective season were excluded from the analysis. Among the 43 athletes, 14 were in the team in Season 1 (S1), 16 in Season 2 (S2), and 13 in Season 3 (S3); only 1 was present for all 3 years, while 4 participated in 2 analyzed seasons. The study was approved by the Human Research Ethics Committee of the Federal University of Juiz de Fora. After accepting the invitation, all athletes signed an Informed Consent Form, indicating their voluntary participation in the study.

Study Design

This study adopts a longitudinal, prospective, and observational design conducted over 3 seasons with a professional male volleyball team. Data regarding training load and injuries were collected daily. Within these 3 periods, the initial 13 weeks were designated as the preparatory phase. The competitive phase consisted of 20 weeks in S1, 22 weeks in S2, and 30 weeks in S3.

Training Load

Internal training load was quantified using the Session Rating of Perceived Exertion (RPE). ¹⁵ This involved multiplying the training duration (in minutes) by the score on the session-RPE scale ranging from 0 to 10. At 30 minutes after the end of each training session, athletes rated their perceived exertion on the scale by answering the question: “How was your workout?” When there were multiple training sessions in a day, the loads were summed to generate the daily load. Training load monitoring occurred in all training sessions and games, with rest days being designated as a daily training load of zero. The isolated session-RPE score and duration were used to measure training intensity and volume, respectively. ¹⁵

From the daily training load values, a series of absolute measures (duration, RPE, daily load) and relative measures (rolling average, EWMA, rolling average ACWR, and EWMA ACWR) were computed.

• Acute and chronic training load: used 7 days for acute load, and 21 or 28 days for chronic load. Thus, the calculation of ACWR was performed for both 7 to 21 days and 7 to 28 days.

• Rolling average ACWR: this calculation employs simple arithmetic means (acute:chronic), with each day being weighted equally in the calculation. In other words, the ACWR per rolling average assigns equal importance to all sessions. ¹⁹

• EWMA ACWR: applies exponentially decreasing weights to the loads performed each day, so that the loads performed on a previous day carry more weight compared with the 28th day, for example. The following formula is used for the calculation:

$${\mathrm{EWMA}}_{\mathrm{today}}={\mathrm{Load}}_{\mathrm{today}}\times {\lambda }_{\mathrm{a}}((1-{\lambda }_{\mathrm{a}})\times {\mathrm{EWMA}}_{\mathrm{yesterday}})$$

where λ_a is a value between 0 and 1 that represents the degree of decay, with higher values discounting older observations at a faster rate. The λ_a is given by

$${\lambda }_{\mathrm{a}}=2/\left(\mathrm{N}+1\right)$$

where N is the chosen time decay constant (7 days for acute load, and 21 or 28 days for chronic load). ⁴⁶

• Coupled and uncoupled approach: a further distinction was made between coupled ( acute load period used as the numerator in chronic load calculation) and uncoupled (acute load period excluded from chronic load calculation) calculation.^21,32

Injury

According to the most recent consensus statement of the International Olympic Committee, differences in injury definition stem from the specific sport or context for which statements were developed. ³¹ Therefore, we adopted the injury definition proposed by the International Volleyball Federation (FIVB): “any musculoskeletal complaint newly incurred due to competition and/or training during the tournament that received medical attention regardless of the consequences with respect to absence from competition or training.”^6,14 Medical staff, comprising 1 physician and 2 physiotherapists, were responsible for diagnosing and recording all injuries. Injuries were documented using the Injury Surveillance Form proposed by FIVB. ¹⁴

Statistical Analysis

The injury incidence was calculated as the number of injuries per 1000 hours of training. The odds ratios (OR) for injury associated with different training load metrics were estimated using logistic models with parameters calculated from the generalized estimating equation. For this estimation, the injury occurrence was calculated and presented using Z-score normalization for different categories: very low (≤-2.00 Z-score); low (-2.00 ≤ a < -1.00 Z-score); low to moderate (-1.00 < a ≤ 0 Z-score); moderate to high (0 ≤ a < 1.00 Z-score); high (1.00 ≤ a < 2.00 Z-score); very high (≥ 2.00 Z-score). For each of these, each session (training or game) for each athlete was used as the response variable, and each training load metric was used as the predictor variable for its respective model.

Similarly, the categorization of the 4 main ACWR measures was based on the Z-score, divided into: very low (≤ -2.00 Z-score); low (-2.00 ≤ a < -1.00 Z-score); low to moderate (-1.00 < a ≤ 0 Z-score); moderate to high (0 ≤ a < 1.00 Z-score); high (1.00 ≤ a < 2.00 Z-score); very high (≥ 2.00 Z-score). For each ACWR range, a model estimated the injury probability considering the number of training sessions, injuries per session and the total number of injuries. R Software Version 3.5.1 was used, and a significance level of 5% was adopted.

Results

A total of 13,283 hours (S1, 5428 hours; S2, 4295 hours; S3, 3560 hours) were recorded across 8021 sessions (S1, 3162; S2, 2980; S3, 1879 sessions) of individual training and games. During this period, 108 injuries occurred (S1, 48; S2, 29; S3, 31 injuries), resulting in a total incidence of 8.13 injuries per 1000 hours (S1, 8.84; S2, 6.75; S3, 8.70 injuries per 1000 hours).

An analysis of 17 different load measures was conducted, including 8 different ways to calculate ACWR. Substantial variability was observed across different periods of the season. Examining the data for the full season, it is evident that the risk of injury is relatively low for most measures, despite the model being significant for all of them (Figure 1a). The ACWR calculated in a coupled manner and by a rolling average (RA ACWR 7:21 C and RA ACWR 7:28 C) demonstrated higher injury risks when analyzing the complete periods. A similar trend was observed during the competitive period (Figure 1c). In contrast, during the preseason, the coupled exponential averages (EWMA ACWR C) were more sensitive to the risk of injury, representing a risk up to 7.5 times higher with an increase of 1 unit in the EWMA ACWR 7:21 C measure (Figure 1b). Only the EWMA ACWR 7:28 C metric during the competitive period did not show a significant value (P = 0.38). In general, the uncoupled ACWR metrics did not indicate significant injury risks.

Figure 1.

Injury risk related to different training load metrics in various periods of the season. (a) Full season. (b) Preseason. (c) Competitive period. ACWR, acute:chronic workload ratio; C, coupled; EWMA, exponentially weighted moving average; NC, noncoupled; OR, odds ratio; RA, rolling average; RPE, rating of perceived exertion; TL, training load; UC, uncoupled.

The absolute data of duration, session-RPE, daily load, as well as the cumulative measures of 7, 21, and 28 days using both rolling average and EWMA, showed injury risks very close to 1 in all periods. This indicates the low relevance of these measures in this analysis.

The main ACWR measures (MM 7:21, MM 7:28, EWMA 7:21, EWMA 7:28 - all coupled) were categorized into 6 ranges based on Z-score values, ranging from “very high” to “very low” (Table 1). The majority of training and game sessions fell within the “low to moderate” and “moderate to high” ACWR ranges. Consequently, most injuries occurred when athletes had ACWR values within these categories. The highest number of injuries per session was in the “moderate to high” ranges for ACWR calculated by the rolling average. For measures calculated by EWMA, the highest concentration of injuries was in the “low to moderate” range for 7:21 and “very high” for 7:28. The probability of injuries across various metrics was relatively low (1.5% to 4.5%). However, within the same measure, the probability of injury can be up to 3 times higher for athletes with “very high” ACWR compared with “very low” ACWR.

Table 1.

Probability of injuries and injuries per session according to ACWR categories calculated in different ways

	Classification	Z-score	Limits	Injuries	Sessions	Injuries per session	Injury probability
RA ACWR 7:21 C	Very low	≤ –2	≤ 0.19	0	240	0.000	≤ 0.015
	Low	–2 < a ≤ –1	0.19 < a ≤ 0.56	2	640	0.003	0.015 < a ≤ 0.020
	Low to moderate	–1< a ≤ 0	0.56 < a ≤ 0.92	35	3249	0.010	0.020 < a ≤ 0.026
	Moderate to high	0 ≤ a < 1	0.92 ≤ a < 1.28	61	3131	0.019	0.026 ≤ a < 0.033
	High	1 ≤ a < 2	1.28 ≤ a < 1.65	7	504	0.013	0.033 ≤ a < 0.043
	Very high	≥ 2	≥ 1.65	3	257	0.011	≥ 0.043
RA ACWR 7:28 C	Very low	≤ –2	≤ 0.21	0	265	0	≤ 0.015
	Low	–2< a ≤ –1	0.21< a ≤ 0.59	3	638	0.016	0.015 < a ≤ 0.020
	Low to moderate	–1 < a ≤ 0	0.59 < a ≤ 0.96	31	3074	0.010	0.020 < a ≤ 0.026
	Moderate to high	0 ≤ a < 1	0.96 ≤ a < 1.34	62	3162	0.019	0.026 ≤ a < 0.034
	High	1 ≤ a < 2	1.34 ≤ a < 1.72	8	612	0.013	0.034 ≤ a < 0.045
	Very high	≥ 2	≥ 1.72	4	270	0.014	≥ 0.045
EWMA ACWR 7:21 C	Very low	≤ –2	≤ – 0.26	0	0	0	–
	Low	–2 < a ≤ –1	–0.26 < a ≤ 0.37	0	278	0	0 < a ≤ 0.025
	Low to moderate	–1 < a ≤ 0	0.37 < a ≤ 1.01	39	4430	0.088	0.025 < a ≤ 0.027
	Moderate to high	0 ≤ a < 1	1.01 ≤ a < 1.65	69	3167	0.022	0.027 ≤ a < 0.029
	High	1 ≤ a < 2	1.65 ≤ a < 2.29	0	57	0	0.029 ≤ a < 0.031
	Very high	≥ 2	≥ 2.29	0	89	0	≥ 0.031
EWMA ACWR 7:28 C	Very low	≤ –2	≤ 0.02	0	7	0	≤ 0.020
	Low	–2 < a ≤ –1	0.02 < a ≤ 0.50	0	595	0	0.020 < a ≤ 0.023
	Low to moderate	–1 < a ≤ 0	0.50 < a ≤ 0.97	27	3416	0.016	0.023 < a ≤ 0.026
	Moderate to high	0 ≤ a < 1	0.97 ≤ a < 1.45	77	3697	0.026	0.026 ≤ a < 0.030
	High	1 ≤ a < 2	1.45 ≤ a < 1.93	4	245	0.039	0.030 ≤ a < 0.035
	Very high	≥ 2	≥ 1.93	0	61	0	≥ 0.035

ACWR, acute:chronic workload ratio; C, coupled; EWMA, exponentially weighted moving averages; RA, rolling average.

Discussion

The current study analyzed injury risk related to the ACWR, calculated in various ways: rolling average and EWMA, coupled and uncoupled, chronic load over 21 or 28 days. This is the first study examining different ACWR calculation methods and injuries in male volleyball. Among the measures from 3 complete seasons, it is evident that coupled ACWRs calculated by rolling averages with 7:28 and 7:21 day periods were comparable and presented the highest injury risks. When considering only the competitive period, these measures were the most related to the risk of injury. Dalen-Lorentsen et al ¹⁰ conducted a similar analysis in a study with young football players. Although some ACWR measures showed a statistically significant association with injuries or health problems (with odds ratios reaching 11.94), when specifically analyzing new noncontact injuries, no associations were identified. ¹⁰ An earlier study that tracked volleyball athletes over a season reported an OR of 3.74 for ACWR calculated from the 7:28 rolling average in a coupled manner. This OR is higher than that found in the present study (1.14). ⁴² Meanwhile, another study associated different ACWR metrics (using the number of jumps) with knee pain scores in volleyball players, reporting positive associations between ACWR variants and knee pain scores, but the findings were not statistically significant. ¹¹

Regarding the preseason analysis, the coupled ACWRs calculated through EWMA (7:21 and 7:28) exhibited the highest OR values, suggesting that these measures may be more sensitive in identifying the risk of injury during this period. This behavior has been documented previously in other studies conducted in Australian football.^13,37 Of all the training load measures, whether in preseason or the competitive period, the coupled EWMA ACWR measures presented the highest injury risk values, reaching an OR of 7.5. In other words, an increase of 1 unit in this measure could increase the risk of injury by 7.5 times. Furthermore, it is noteworthy that, overall, the uncoupled ACWR metrics did not prove to be more sensitive to the risk of injury than their coupled counterparts.

The findings concerning the difference in injury risk based on the ACWR measures used confirm that the choice of the calculation method for this ratio can significantly impact the study results.^10,40 Consequently, any form of comparison between studies using different metrics is considered limited and not recommended. Although it is an essential tool for monitoring training load, the validity of the ACWR has been questioned.^29,36 Some studies suggest that one solution could be the use of absolute measures of training load.^30,33 However, the results of the current study demonstrated that absolute training load data (duration, RPE, daily load, rolling averages or exponential averages of 7, 21, or 28 days) did not show relevant injury risks, with OR values remaining very close to 1.

To identify a safe training load for athletes, various cut-off points within these measures have been speculated. One study referred to the ACWR range between 0.8 and 1.3 as the “sweet spot” and an ACWR above 1.5 as a “danger zone.” ¹⁹ Since then, many practitioners have used this reference as a “magic number.” However, this research was conducted with rugby, cricket, and Australian football athletes, which does not allow these values to be extrapolated as a reference for sports with different demands. Furthermore, the authors themselves emphasized that a single metric of training load would be incapable of predicting injuries due to their multifactorial etiology. ¹⁶ In this context, the present study categorizes different ACWR metrics into 6 ranges, from “very low” to “very high,” indicating an increase in the injury probability with higher ACWR values. However, despite these results being useful as a reference for the sport, it should not be used as a “cut-off point” for training load in volleyball. Therefore, ACWR should not be analyzed in a cause-effect relationship with injuries but as a reference value in a multifactorial training load management approach.

Another point highlighted is that the ACWR is calculated based on different measures of external and internal load. The present study used the session-RPE to calculate the internal training load. In volleyball, jump count remains the most used measure of external load; however, there is a question as to whether this is the best measure, given that there are other important demands in volleyball beyond jumping.^38,39 Therefore, these findings are not comparable with others that calculated ACWR using different measures.

The creation of categories based on ACWR ranges calculated from the session-RPE has also been explored in other sports.^13,25,34 However, this is the first study to investigate these subdivisions in volleyball. The probability of injury values is relatively low, reaching a maximum of 4.5%. This finding aligns with the multifactorial nature of the etiology of sports injuries, where training load alone is unable to predict this event in isolation. It is essential to perceive training as one among various determinant factors related to sports injuries.^7,28 This understanding justifies the need for a multidisciplinary approach, incorporating the greatest number of risk factors. Nevertheless, there are factors beyond the control of professionals and athletes, which makes the prevention of sports injuries a significant challenge.

Although our findings provide important new information on the relationship between training load, recovery, and injury in elite volleyball, there are limitations. First, our findings are specific to elite male volleyball players, and should not be generalized to adolescent or female players. A major limitation of this study is the absence of other objective tools for measuring training load, such as the measurement of external load using data from GPS and accelerometers. These tools can provide a range of data on distance, speed, acceleration, and jumps. Moreover, this study controls only for training load, overlooking various other factors influencing injury occurrence, such as age and previous injuries. Finally, our data collection was limited to a single team, which restricts the generalizability of the findings.

The findings of this study can be useful and applicable to members of coaching staff, fitness trainers, and healthcare professionals involved in the challenge of reducing the risk of injury in volleyball athletes. It emphasizes the need for continuous monitoring and real-time adjustments of training load. In this way, professionals can provide athletes with training that enhances performance without increasing the risk of injury. It is important to highlight that the analysis of different ACWR metrics can be useful for professionals to choose those that best fit the specific demands of volleyball.

Conclusion

In summary, regarding the different calculations of ACWR, both measures using rolling averages and those calculated from exponential averages can be employed to identify the risk of injuries in volleyball athletes. Only during the preseason period were the EWMA measures more sensitive to the risk of injury. Generally, uncoupled measures do not yield relevant results, and neither do absolute load measures such as session-RPE, duration, daily load, or chronic load in isolation. Finally, categorizing ACWR values suggests what might be considered low or high values for a group of male volleyball players. However, it is not recommended that these values be used as a threshold or absolute rule to reduce the risk of injuries in the sport. The application of this study should be interpreted as another step toward a deeper understanding of the relationships between training load and injuries in volleyball, prompting further research to discern other aspects of this complex relationship.