Effect of training load on sleep parameters and biochemical fatigue markers in elite swimmers

INTRODUCTION

Sleep is considered a principal component of training and competition recovery, and essential for optimizing athletic performance [1]. Given the importance of sleep to athletic performance and recovery, it is surprising that poor sleep is still prevalent among elite athletes [2]. Many factors can affect the sleeping patterns (e.g., travel, unfamiliar environment, pre-competition anxiety, the use of electronic devices, altitude) of athletic populations [2–5], including intensified training periods [2, 3]. For instance, a 3-wk overload training program promoted overreaching in triathletes and this, in turn, promoted a decline in sleep quantity and quality [6]. Similarly, a short period of intensified training affected sleep efficiency among male cyclists [7]. Others have, however, found no association between TL variation and sleep [8, 9]. These discrepancies could be explained by the interdependence of sleep-exercise interactions, such that sleep might drive, and be driven by, shifts in TL.

Other difficulties arise in terms of identifying the sleep component that best reflects TL. Taylor et al., [10] examined sleep changes in female swimmers during three phases of a competitive season. Most indicators of sleep quantity (e.g., total sleep time) and quality (e.g., sleep latency) did not differ across the season, but time spent in deep sleep was higher during the onset and peak training (vs. tapering) period. A more recent study profiled competitive swimmers across four training phases [9]. They found only trivial to small differences in sleep quantity between phases, whilst other parameters (e.g., sleep latency) showed only small to moderate differences. Substantial between-individual variation in sleep responses also emerged. Hence, a broad assessment of sleep, ideally combining subjective and objective measures, is needed to explicate more complex sleep-TL interactions. Adding to these difficulties, it is not clear whether sleep disturbances are a consequence of physical or psychological overload, thereby contributing to an overreaching or overtraining state [11], or simply a manifestation of the aforementioned syndromes.

Training load is generally classified into two components; internal and external [12]. Internal TL (e.g., heart rate response to training), reflects the psychophysiological response to exercise, whilst external TL (e.g., total distance covered), refers to the physical work completed. Both external TL [9, 13] and internal TL measures have been used in sleep research on athletes [14, 15], but few have combined these approaches to capture the nuances of sports training [16]. Others expanded their assessment to include salivary testosterone (T) and cortisol (C) measures [17, 18], but linkages to sleep were equivocal due to individual and context specificity, with both measures more strongly affected by psychological stress rather than fatigue [17, 18]. As such, implementing a panel of blood biomarkers could help decode the impact of training-related fatigue mechanisms on sleep patterns or vice versa.

Though none of the currently available blood biomarkers are regarded as the definitive fatigue indicator, some are considered promising and used in science and practice to monitor fatigue and the recovery process (e.g., creatine kinase [CK], urea [URE], T, C) [19, 20]. Serum CK activity, for example, reflects metabolic and/or mechanical damage of the muscle fibres resulting from repeated and intense contractions [21]. Increased serum URE is a marker of enhanced protein breakdown (in the absence of excessive protein intake) and stimulated gluconeogenesis that results from increased training loads [19, 20]. Both CK and URE may also provide information pertaining to increased muscular and/or metabolic strain [19, 20]. Cortisol is an important signal for stress and metabolic regulation, sometimes promoting catabolic processes, and synchronization of peripheral clocks with the circadian cycle [22]. Testosterone often works antagonistically to cortisol, promoting protein synthesis and reducing protein breakdown; thus, increased C and/or decreased T levels might indicate high physiological strain [20].

The purpose of this study was to describe possible changes in actigraphy-based and subjective sleep parameters among professional swimmers subjected to different TLs, assessed using a combination of external and internal TL measures. Specifically, the swimmers were monitored across three training micro-cycles that differed in prescribed load: two preparatory periods (P1, P2) with a relatively high TL and a pre-competition tapering period (P3) with a relatively low TL. We investigated the interplay between TL and sleep, and also whether any adaptive shifts would manifest as biochemical (URE, CK, T, C) changes, both baseline activity and acute exercise-induced reactivity. We broadly hypothesized that a higher TL in P1 and P2 would lead to a deterioration in sleep quality and/or quantity, along with changes in biochemical fatigue markers (e.g., increase in baseline and exercise changes in CK and C, lowering of baseline T), with restoration of these outcomes in P3 following a lower TL.

MATERIALS AND METHODS

Study design

A single group within-subject design was used to address the study aim. The participants were monitored for weekly internal and external TL, sleep patterns, and biochemical changes across a 1-week block during three training periods (P1, P2, P3), as they prepared for the Polish National Swimming Championships. Four weeks of training separated P1 (aerobic endurance capacity; session duration = 113 ± 13 min, session volume = 6.4 ± 1.4 km) and P2 (aerobic endurance capacity; session duration = 111 ± 19 min, session volume = 5.3 ± 1.4 km), with six weeks separating P2 and P3 (tapering; session duration = 66 ± 9 min, session volume = 3.2 ± 0.4 km). P3 ended some three days before the championship event, where success was crucial to obtain a qualifying time standard for other international events (e.g., 18^th FINA World Championships, European Junior Swimming Championships). A traditional linear periodization plan (low-intensity and high-volume training across the first periods of the macrocycle, with progressive increases in training intensity and concurrent decreases in training volumes of the consecutive periods) was developed by the head trainer to ensure peak performance. To achieve this goal, an incremental decrease in training volume (in km) was prescribed from P1 to P2 to P3. Training frequency did not change within each micro cycle. This entailed 6 training days a week, comprising of 6 morning (starting 6:00 am in P1 and P2, 6:45 am in P3) and 5 afternoon (starting 4:00–6:00 pm) sessions. Figure 1 shows an overview of the study design as well as the collected parameters.

FIG. 1

Schematic representation of the study design, including time points of training sessions, blood sampling, and the session rating of perceived exertion (sRPE) collection.

RESULTS

Athlete training intensity varied, based on five different “zones” of physical effort. The division of training intensity into five different zones was adopted arbitrarily by the coaching staff. The percentage of training volume spent in each zone is depicted in Figure 2.

FIG. 2

Percentage of training volume performed at a given intensity zone.

Biochemical parameters

Training period did not affect (p = 0.682) pre-session URE concentration (Figure 3A). Pre-session CK (p = 0.001), C (p = 0.010) and T (p = 0.031) did, however, vary with respect to training phase. CK levels declined from P1 to P3 (Figure 3B) with a large effect (d = -0.8, p = 0.001), with a similar difference from P2 to P3 (d = -0.6, p = 0.036). Plasma T concentration rose from P1 to P3 (Figure 3C), a moderate effect (d = 0.7, p = 0.008), but neither differed from P2. A similar pattern emerged for C (Figure 3D), with a higher concentration in P3 than P1 (d = 0.6, p = 0.024), whereas P2 results were similar (non-significant) to both P1 and P3.

FIG. 3

Pre-session urea, creatine kinase (CK), testosterone (T), and cortisol (C) concentration in each training period. Data are presented as estimated marginal means with a 95% CI. ASignificant from P1, BSignificant from P2, all p<0.05.

The within-session biochemical responses (%) were interpreted as being significant from baseline, if the 95% CI did not cross zero. We found no significant URE changes (Figure 4A), whereas positive CK changes were seen in each training period (Figure 4B). Conversely, a declining T (Figure 4C) and C (Figure 4D) response was observed in each period. When comparing these responses, a period effect on URE responsivity (p = 0.033) emerged, but the post-hoc comparisons did not reach statistical significance. The CK and C changes were unaffected by training phase (p ≥ 0.127). For T, the training effect was significant (p = 0.038) and follow-up tests identified a smaller (negative) change in P2 versus P1 (d = 0.6, p = 0.037).

FIG. 4

Acute within-session changes (%) in urea, creatine kinase (CK), testosterone (T), and cortisol (C) in each training period. The Monday and Friday sessions were pooled for the analysis. Data are presented as estimated marginal means with a 95% CI. *Significant from baseline, A Significant from P1, B Significant from P2, all p<0.05.

DISCUSSION

This study is the first to examine the interplay between TL, sleep and biochemical fatigue in elite swimmers across different training phases preceding a major competition. The swimmers experienced insufficient and fragmented sleep, regardless of the physical demands associated with different training periods. Only bedtime and get-up time differed (P3 > P1 and/or P2) in this study, but these differences resulted from the later start time of morning training across P3 (6:45) vs P1 (6:00) and P2 (6:00). Nevertheless, the magnitude of this difference was not large enough to be meaningful in real life; therefore, our hypothesis that higher levels of TL would affect sleep quantity and quality was not supported. Baseline plasma CK activity best reflected changes in TL, decreasing in a stepwise manner, although the anticipated rebound in P3 did not occur.

The impact of training load on sleep patterns in swimmers is equivocal [9, 10, 14, 15, 31]. As expected, all TL indices decreased incrementally over time with respect to P1, but with little corresponding change in athlete’s sleep quantity or quality. Only bedtime and get-up time differed in this work (P3 > P1 and/or P2), likely reflecting training schedule rather than reduced TL and shorter workouts prescribed during athlete tapering for the championship event. However, the results are not practically significant. Firstly, even though in P3 participants got out of bed later than in P1 and/or P2 they went to bed later as well. Secondly, changes in bedtime and get-up time did not translate to an increase in total sleep duration. A similar pattern was seen in swimmers during a 14–day altitude training camp at 1500 m [4]. In the monitored athletes, gradually increasing training loads (sRPE) did not jeopardize their sleeping patterns. Our findings also concur with with Walsh et al., [9] who reasoned that a threshold effect might be operating. That is, training differences in sleep parameters are more likely to occur when a much higher TL, and one that induces excessive overload, is prescribed. Still, there is no consensus as to whether impaired sleep is a cause or a consequence of heavy training loads [11, 32]. Sleep disturbances are common in sport, particularly among athletes experiencing training maladaptation that elicit overreaching or overtraining [6, 31]. Recently, Vachon et al. [32] reported that poor sleep quality prior to intensive training predisposed athletes to the development of functional overreaching. Throughout this study, all swimmers showed systematic improvements in performance, which indicates the prescribed TLs were not excessive, but entirely appropriate to promote positive adaptations. Moreover, based on the assessed fatigue biomarkers, it might be assumed that our subjects were not in an overreached or overtrained state.

Our major finding is that, irrespective of the training period and TLs, the swimmers slept less (~6.3 h) than the recommended 8–10 h or 7–9 h of sleep per night for healthy adolescents and adults, respectively [33, 34]. Supporting our results, many studies have reported disruptions in the sleep patterns of swimmers depending on their training schedule [9, 35–38]. For instance, the amount of sleep achieved by Australian swimmers during nights preceding training days (5.4 h) was substantially lower than prior to rest days (7.1 h) [35]. This reduction in sleep was linked to early morning training sessions (i.e., 6:00 am) and a failure to compensate with an earlier bedtime. Others have attributed low sleep quantity (i.e., 6.6 h) among swimmers to early-morning starts (i.e., 6:30 am) [36]. Correspondingly, a longer total sleep time (7.7–7.9 h) was found among swimmers who begun their first sessions at a later time of day (7:30–8:30 am) [9]. However, Vitale et. al. [15], who monitored sleep in swimmers 4 nights before and 14 nights during a 14-day training camp, frequently observed reduced sleep times (< 7 h) during this study, even though the athletes started training at 8:30 am. Thus, our observation of insufficient sleep quantity most likely stems from the early-morning workouts, scheduled around 6:00–6:45 am, which is typical of elite swimmers. The justification for this is to fit around schooling or work activities, and to provide maximum time for recovery between the sessions, since there is no performance rationale for early-morning starts. Interestingly, morning exercise might cause phase advances regardless of the chronotype [39]. People with late chronotypes, as often seen in young adults, are particularly susceptible to circadian disruption caused by misalignment of the internal rhythms with early-morning duties [39]. Further research is needed to confirm if swimmers with late chronotypes may benefit from phase advances induced by exercise in the morning. Although, participating in this study, being elites with several years of training experience, were training routinely in the early morning, so their sleep patterns were likely adjusted to the early morning start. Nonetheless, the negative impact of early-morning training sessions on sleep quantity is well recognized, as is the consequences of sleep restriction (e.g., fatigue, excessive daytime sleepiness, decreased exercise and cognitive performance, decreased mood) [2, 34].

Regardless of TL, the subjects did not meet recommendations for WASO (i.e., ≥ 41 min) [40], indicating restless or fragmented sleep. In the current study, swimmers spent less time awake after sleep was initiated (WASO: 41–45 min) than adolescent Icelandic swimmers (WASO : 64.8 ± 24.8 min) [38], but more than adult open-water swimmers (WASO: 22–27 min) monitored across altitude camp [15] or Dutch elite athletes (WASO: 33.3 ± 17.0 min) [8] from various sports disciplines during seven consecutive days of regular training. Impaired sleep continuity has been attributed to increased TLs [41], muscle soreness and general fatigue in athletes [14]. These data indicate that the swimmers obtained an insufficient amount of fragmented sleep, indicating an urgent need for sleep optimizing strategies focused on sleep extension and promoting continuous sleep. Hence, key swimming stakeholders should consider different strategies to maximize sleep, especially at the individual level. For instance, when feasible, delaying morning training sessions to allow longer sleep duration should be considered. Sleep hygiene education may encourage earlier bedtimes and reduce causes of excessive nighttime awakenings. Moreover, napping should be encouraged, particularly if extending nocturnal sleep is not possible.

To our knowledge, no previous research has investigated the impact of training on sleep in swimmers combining workload with blood fatigue biomarkers. Baseline plasma CK activity best reflected changes in TL, but with no corresponding change in athlete’s sleep pattern. Considering the entire preparation period, the most visible shifts in the biochemical parameters were identified for biochemical parameters prior to training sessions. This included a decline in CK levels, coupled with a gradual rise in T and C concentrations, as the athletes transitioned from high-volume training to low-volume, pre-event tapering. There are conflicting reports concerning fatigue biomarkers and TL in swimmers. In swimmers starting from second week of preseason training, CK levels have been shown to decrease across subsequent weeks of training [42] and tapering period [43–45]. Similar to our findings, T level decreased when swimming training intensified and increased during periods of decreased TL (i.e., post –taper) [44]. Even though the C increase across the taper period was unexpected, it was in line with previous findings as well [43]. On the contrary, some [46] have demonstrated weekly changes in URE but not CK, T or C levels in adolescent swimmers and triathletes, whilst others found no CK and endocrinological differences to periods of intense swimming training and taper [47].

Serum CK level is considered an indirect marker of muscle tissue damage [21]. Although swimming is a non-contact sport, where concentric muscle actions predominate over eccentric actions, increased TL and highly repetitive motions may enhance metabolic and/or mechanical disruption of muscle cells, leading to a rise in baseline serum CK changes [21], as we demonstrated herein. Also noteworthy is that the CK response to acute exercise was consistently elevated (by 15–20%) above baseline. Two possibilities could explain this finding: (1) exercise intensity was sufficiently high to induce changes in skeletal muscle membrane permeability, resulting in increased CK release independent of mechanically-induced damage; (2) swimming is a repetitive movement that places high demands on joint articulations and thus, leading to muscular strain that also promotes greater CK efflux [48].

The hormones T and C contribute to athletic performance and training adaption through several mechanisms [49]. One example being the anabolic to catabolic balance in muscle tissue regeneration and TL tolerance [50]. The observed rise in pre-training T level, when TL was the lowest, could represent greater pituitary activation following the more intense training period, leading to a downstream increase in T when tapering [50]. The corresponding increase in C might reflect a need to partition metabolic resources in preparation for the impending National Swimming Championships, where success was pivotal to obtain a qualifying time standard for 18th FINA World Championships and/or European Junior Swimming Championships [51]. In addition, it cannot be ruled out that higher C level might reflect an emotional response (eg. activation of the sympathetic nervous system and hypothalamic-pituitary-adrenal axis) in anticipation to a crucial sporting event [51]. We also observed a decline (by -8 to -40%) in T and C reactivity to exercise that arguably reflects normal circadian variation in steroid hormones [52], but no patterns emerged with regards to TL. Serum urea concentration (baseline and reactive) did not differentiate between high and low TLs, perhaps due to individual variability and other metabolic factors (e.g., availability of carbohydrates).

Finally, a number of important study limitations need to be considered. Although the biochemical parameters are recognized as surrogate markers of TL, their use as definitive biomarkers of fatigue should be done so with caution, namely due to high inter-subject variability and individual response patterns [46]. Furthermore, changes of plasma volume, that could affect post-training biochemistry, were not assessed. Another limitation is the lack of a chronotype assessment, which would allow for a better understanding of the impact of TL on sleep, since submaximal RPE and fatigue score are likely to be influenced by chronotype [53]. In addition, the sample size was small, due to recruitment restrictions when undertaking research on elite athletes, but this was offset somewhat by the repeated measures design and intensive monitoring procedures. Inferences of any TL effect are also limited by the number of assessment occasions, especially when the swimmers were already experiencing a high volume of training at study inception. Additionally, data were pooled across both sexes to increase our sample size. Recruiting a larger number of male and female swimmers would allow a more detailed examination of gender-related differences in the studied variables. Future studies should look to increase sample size, but where restrictions on recruitment are a limiting factor, it is advisable to adopt a repeated measures design to increase study power and subsequent inferences. Moreover, the phase assessment of TL did not include non-specific training (i.e., gym sessions), although the “dry-land” training presumably had no significant effect on the monitored parameters.

CONCLUSIONS

In conclusion, the profiling of elite swimmers revealed insufficient and fragmented sleep, regardless of the physical demands associated with different training periods. In fact, neither sleep quality nor quantity were affected by different magnitudes of TL. Even though swimming is a sport with a predominant share of concentric muscle actions, baseline plasma (pre-training) CK activity best reflected the physiological response to TL among the monitored biochemical fatigue markers.