Delaying Early Morning Training Time in Swimmers: Effects on Sleep, Recovery and Performance—A Randomized Controlled Trial

Abstract

Background

Swimmers often fail to achieve recommended sleep durations. Early morning training, common in swimming, is associated with shorter sleep. Sleep extension through later wake-up times appears to be an appropriate measure to counteract this. Therefore, this study investigated whether delaying early morning training has an effect on sleep, recovery and performance, and whether changes in sleep mediate eventual effects on recovery and performance.

Methods

Using cross-over design, 27 national-level swimmers (13 females, 14 males; age: 15.6 ± 1.0 years) completed two 8-day training camps in January and October 2023 in Saarbrücken, Germany. Morning sessions started at 07:00 AM (early) or 09:00 AM (late), with conditions randomized between camps. Sleep and recovery were assessed using actigraphy, sleep diaries, and the Short Recovery and Stress Scale. Performance was evaluated using 100 m main stroke and 800 m freestyle time-trials pre- and post-camp. Linear mixed-effects models and repeated measures analysis of variance were employed to assess the effect of training time on sleep, recovery and performance. Regression-based mediation analysis tested whether changes in sleep and recovery mediated the effects of training timepoint on recovery and performance.

Results

Sleep duration per night was significantly longer in the late compared to the early condition (1.0 ± 0.5 h; t = 14.0, p < .001). However, there was no significant effect on measures of recovery (p > .080), and 100 m and 800 m performance (p > .204). Mediation analysis showed no indirect effects of sleep duration on mental capacity or of mental capacity on performance (p > .05).

Conclusion

Delaying early morning training from 07:00 AM to 09:00 AM increased sleep duration by one hour on average, but this did not significantly affect recovery or performance in elite swimmers. Nevertheless, the study provides evidence that early morning training is associated with reduced sleep duration and highlights the persistent sleep challenges faced by adolescent swimmers.

Trial registration DRKS, DRKS00039267. Registered 17 February 2026—Retrospectively registered, https://www.drks.de/DRKS00039267.

Background

Sleep plays a crucial role in multiple restorative and developmental processes across the nervous, metabolic, musculoskeletal, and endocrine systems [1]. Accordingly, sleep loss can have wide-ranging consequences, impairing attention, working memory, decision-making, learning capacity, and mood regulation [2], while also increasing susceptibility to infections, inflammatory diseases [3], and reduced muscle protein synthesis [4].

A common form of sleep loss is sleep restriction, also referred to as partial sleep deprivation, which occurs when sleep duration is restricted below an individual’s sleep need, without complete absence of sleep [5]. Sleep restriction has been shown to negatively influence sporting performance, including anaerobic power, speed/power endurance, high-intensity interval exercise, maximal strength, strength-endurance, and skill execution [6]. Notably, when sleep is restricted at the end of the night (i.e., early wake-up times), referred to as "late sleep restriction", the performance impairments appear to be particularly pronounced [6–8]. Together, these findings suggest that sleep loss can be detrimental to performance, particularly when sleep is restricted in the morning, as with early morning training.

While the effects of sleep restriction on performance are well documented, evidence regarding its impact on recovery is limited. Rae et al. [9], observed that recovery from high-intensity interval training was impaired after one night of sleep restriction. These findings, together with the cognitive and physiological consequences, suggest a potential role of sleep for recovery. Accordingly, coaches and athletes recognize sleep as a key component of recovery [10–12].

Despite this recognition, previous research indicates that athletes often fail to obtain recommended sleep durations of 7–9 h for adults and 8–10 h for adolescents [13–15]. Sargent et al. [16] further reported that athletes not only sleep less than these recommendations but also less than they believe they need, with an average perceived sleep deficit of 96 min per night.

This shortfall appears to be even more pronounced in swimmers, with reported sleep durations ranging from 6.2 to 7.7 h [16–19], and an average perceived sleep deficit of 118 min [16]. Previous studies suggest that the prevalence of early morning training in swimming may contribute to this, with evidence showing that sleep on nights preceding sessions starting before 07:00 AM is reduced by 1–2 h compared to nights before later sessions or rest days [17, 18, 20]. Considering the impairments observed with (late) sleep restriction [6–9], early morning training may similarly compromise performance and recovery.

To mitigate these effects, several strategies have been proposed, such as napping, light therapy, cold water immersion and sleep extension [21]. Among these, sleep extension appears particularly promising, with a recent systematic review demonstrating consistent improvements in performance outcomes when sleep was prolonged by 46–113 min beyond a habitual duration of ~7 h, resulting in total sleep duration >8 h [21].

While sleep extension appears to improve performance, its effects on post-exercise recovery remain to be investigated. Additionally, although early morning training has been linked to reduced sleep duration, existing studies are observational in nature, limiting conclusions about causality [17, 18, 20]. Consequently, it remains unknown whether delaying early morning training can extend sleep duration and, enhance recovery and performance. A training camp would provide a controlled setting to investigate this, and is particularly relevant as previous studies suggest that periods of high training load can affect sleep duration and sleep quality [22–25].

Therefore, the aim of this study was to investigate whether delaying early morning training during a training camp influences sleep, perceived recovery, and swimming performance. Beyond examining these outcomes in isolation, this study explored whether delaying morning training affects recovery through changes in sleep, and whether it affects swimming performance through changes in recovery.

Methods

Participants

Thirty-four tier 3 level swimmers (17 females, 17 males, age: 16 ± 1 years) participated in this study [26]. Recruitment was conducted through an invitation by the German national swimming federation to a training camp. An a priori sample size calculation for the Lamberts and Lambert Submaximal Cycling Test (LSCT; d = 0.2, α = 0.05, power = 0.80) indicated a required sample of 155. Since this was not practically feasible, a target of 30 was set, at which medium effects could be detected (d = 0.46). As the LSCT was not a primary outcome here, a simulation-based power analysis for the reported outcomes is provided (Sect. "Data Analysis"). Prior to the study, participants received detailed information about its objectives and procedures, and provided written informed consent. For participants under 18 years of age, written informed consent was also obtained from their parents or legal guardians.

Prior to the start of the study, participants were screened for sleep disorders using the Sleep Questionnaire B/Revised (SF-B/R; Görtelmeyer, 2011), and chronotype was assessed using the Morningness–Eveningness Questionnaire (MEQ; Horne & Östberg, 1976 [27]). Swimmers were excluded from participation in the case of an injury or medical condition that precludes physical exertion, or in the case of a clinically diagnosed sleep disorder. The study was conducted in accordance with the Declaration of Helsinki, with the exception that it was retrospectively registered at the clinical trial registry. The study was approved by the local Human Research Ethics Committee (Ärztekammer des Saarlandes, Saarbrücken, Germany, registration no.: 114/22).

Study Design

Two 8-day training camps were conducted in January 2023 and October 2023 at the Olympic training center in Saarbrücken, Germany. All training took place on-site, and participants resided in shared on-site accommodation, away from their habitual sleeping environment. Each camp included a total of 13 swimming sessions, consisting of 6 morning and 7 afternoon sessions. The first and final sessions took place in the afternoon of day 1 and day 7, respectively, with no training scheduled on day 8. Morning training sessions began either at 07:00 AM (early condition) or 09:00 AM (late condition). Participants were individually randomized to one of the two conditions, with allocation reversed in the second camp to allow for within-subject comparisons. Randomization was performed using a random number generator in Microsoft Excel without stratification. Both training camps were designed and organized by the same head coach, who kept the structure and content of the training sessions consistent across both conditions and camps. Swimming performance, recovery status and motivation were assessed on day 1 (pre-test) and day 8 (post-test). Daily measures of sleep, perceived recovery, and training load were collected during the two weeks preceding each camp as well as throughout the entire training camp.

Measurements

Performance

Swimming performance was assessed in a 50 m indoor pool through a 100 m time-trial in each participant’s main stroke and an 800 m freestyle time-trial. Participants were divided into series of 4–5 swimmers based on main stroke and personal best times. Time-trial performance was measured manually by experienced coaches using handheld stopwatches. Before the 100 m time-trial, participants completed a standardized warm-up consisting of 2 × 500 m and 1 × 200 m freestyle at 60%, 80% and 90% intensity, respectively, based on the Lamberts & Lambert Submaximal Cycling Test [28]. Time between the warm-up, the 100 m time-trial and the 800 m time-trial were 10–20 and 60–85 min, respectively, to ensure sufficient recovery. The start time of performance testing varied by condition: the early group began between 07:00 AM and 07:45 AM, while the late group started between 09:00 AM and 10:00 AM. Participants were instructed to replicate their nutritional intake during the 24 h preceding the post-test to match that of the 24 h prior to the pre-test, as recorded in a nutrition protocol. Breakfast prior to training and performance testing was provided on-site for all participants.

Sleep

Nocturnal sleep duration, sleep onset latency, sleep efficiency, and perceived sleep quality were assessed using actigraphy (BT-WGT3X ActiGraph, ActiGraph LLC, Pensacola, FL, USA) in combination with a sleep diary. Participants were instructed to wear the actigraph on their non-dominant wrist throughout the study period, except during water-based activities.

Actigraphy data were processed using the GGIR package (version 3.1–5) [29], which has been previously validated against polysomnography [30]. The sleep diary was used to define the sleep period time window [31]. In cases where sleep diary data were unavailable, sleep timing was estimated using the HDCZA algorithm [32], implemented within GGIR.

Single nights were excluded from analysis if less than eight hours of valid wear data were recorded between noon and noon the following day, to account for potential non-wear periods before and after bedtime despite instructions to the contrary. Sleep periods containing more than 20% non-wear time were excluded, based on exploratory analyses showing increased variability in sleep measures beyond this threshold. Additionally, any night where the recorded wake-up time occurred after the scheduled morning training start was omitted.

At the start of the study, participants were instructed to abstain from napping throughout the training camp. However, on day 3, the coaching staff raised concerns about an expected high drop-out rate due to accumulated fatigue. Considering this and the common use of napping in training camps, daytime naps were permitted from day 3 onward in both camps and recorded using a nap diary.

Recovery and Training

Measures of recovery were obtained during the standardized warm-up prior to performance testing and through daily self-report. After each stage of the warm-up, athletes reported their rate of perceived exertion (RPE) and heart rate was assessed by palpation of the carotid artery for 10 s. Perceived recovery and stress were assessed each morning using the Short Recovery and Stress Scale (SRSS), an 8-item questionnaire [33]. Training load was assessed in the evenings, athletes reported training characteristics, including session duration and session rate of perceived exertion (sRPE). Session load was calculated as the product of sRPE and session duration, and a 7-day rolling average was computed to monitor training load over time. Data from days on which participants reported illness during the camp were excluded from analyses, as illness may affect both perceived recovery and stress. SRSS and RPE were collected via a custom, GDPR conforming smartphone application as previously published [34].

Data Analysis

Descriptive statistics were used to summarize participant characteristics, including sex, age, and chronotype distribution. Baseline sleep and chronic training load were compared between conditions (early, late) using repeated measures analysis of variance (ANOVA), to assess potential differences. Daily session load during the training camps was analyzed with a linear mixed-effects model including participant as a random intercept to assess whether training was comparable. To examine potential period effects, paired-sample t-tests were performed to compare the first and second camp in baseline average sleep, baseline chronic training load, average camp session load and pre-camp performance.

The effect of condition (early, late) on sleep, napping, recovery, performance, and motivation was first assessed in isolation. Linear mixed-effects models were used for sleep, nap duration, 24-h sleep duration and perceived recovery, and a Poisson mixed-effects model for nap count. Changes in heart rate and RPE during the standardized warm-up were analyzed using a two-way repeated measures ANOVA with intensity (60%, 80%, 90%), and condition (early, late) as within-participant factors. Performance and motivation were analyzed with two-way repeated measures ANOVA with timepoint (pre, post) and condition (early, late) as within-participant factors.

To assess whether the effect of delaying morning training on recovery and swimming performance operated through changes in sleep and subsequent recovery, a regression-based mediation analysis was conducted. Mediation analysis examines whether the effect of an independent variable (X) on a dependent variable (Y) occurs directly (X → Y) or indirectly through one or more intermediate variables (X  →  M  →  Y), referred to as mediators (M).

The hypothesized pathway is illustrated in a directed acyclic graph (DAG; Fig. 1), where arrows represent assumed causal directions. In this model, sleep, recovery, and performance are operationalized as sleep duration, mental capacity, and 100 m time, respectively.

Fig.1.

Directed acyclic graph representing the hypothesized mediation pathways linking morning training time (blue) to sleep, recovery and performance (green), directly or indirectly. Grey-shaded variables were considered conceptually but not included in the statistical mediation analysis. Arrows indicate the hypothesized direction of causal effects

Each pathway in the DAG was evaluated using linear models:

1. Sleep duration = a₁ x condition x sex,

2. Mental capacity = m₁ x sleep duration x sex + a₂ x condition x camp day,

3. Post 100 m time = m₂ x mental capacity + c x condition + pre 100 m time.

Random intercepts and slopes for participants were included in models 1 and 2. For model 3, only mental capacity measures obtained on the same day as the post-performance assessment were considered.

Exclusion of participants due to missing data was applied at two levels: general and outcome specific. First, participants who did not complete both training camps were excluded from all analyses. Second, within each outcome variable, participants with fewer than two observations were excluded from the corresponding analysis. As a result, missing data on one outcome did not preclude inclusion in analyses of other outcomes maximizing the use of available data.

A simulation-based power analysis was conducted to evaluate the ability to detect differences in sleep duration between the early and late condition, given the observed sample structure. The alpha level was set at 0.05, and 100 simulations were performed. Fixed-effect estimates and variance components were based on Ashby et al. (2024) [17]. This yielded an estimated power of 100% (95% CI 96–100%).

All analyses were conducted in R. (version 4.3.3). The corresponding code and data are available at https://github.com/MBrandts98/SleepTrain_analysis Results are reported as mean ± standard deviation or as unstandardized coefficients (b) with standard errors (SE), accompanied by relevant statistical estimates from the mixed-effects models or ANOVA, as appropriate.

Results

Participant Characteristics

Participant flow across the study is depicted in Fig. 2. Of the 34 athletes who took part in the first training camp, 27 completed both camps and were considered for the analyses (13 females, 14 males; age: 15.6 ± 1.0 years). No participants were excluded based on the SF-B/R. Chronotype distribution, based on self-report, was predominantly neutral (n = 21), with three participants classified as rather morning types and three as rather evening types. The specific analyses in which participants were included are illustrated in Fig. 3, resulting in samples ranging from 14 to 25 swimmers.

Fig.2.

Participant flow throughout the study

Fig.3.

Participant inclusion across analyses. Analyses are listed on the left y-axis, participant codes on the x-axis (top), and the number of participants included per analysis on right. Green dots indicate inclusion, red dots exclusion. DAG directed acyclic graph

Baseline Characteristics and Training Load

No statistically significant differences were found between the early and late condition in baseline sleep parameters (bedtime, waketime, and sleep duration) or chronic training load (p > 0.20). Additionally, session load during the training camps did not differ significantly between conditions (p = 0.47). Detailed values and test statistics comparing the early and late conditions are reported in Table 1. No significant condition x sex interaction effects or main effects of sex were observed (p > 0.15).

Table 1.

Summary of baseline characteristics, training load, sleep and recovery outcomes across the early and late training conditions

	Early	Late	Late vs. early a
	Mean ± SD	Mean ± SD	Mean ± SD	t-value	p-value
Baseline
Average bedtime (hh:mm)	23:00 ± 01:00	23:13 ± 00:49	00:13 ± 00:54	1.33	.199
Average waketime (hh:mm)	06:50 ± 01:12	06:59 ± 01:04	00:08 ± 01:11	0.68	.503
Average sleep duration (h)	6.6 ± 0.8	6.6 ± 0.7	− 0.0 ± 1.2	− 0.09	.923
7-day average session load b	435 ± 255	481 ± 272	46 ± 359	0.67	.506
Training
Average daily session load b	1005 ± 258	964 ± 296	− 40 ± 305	− 0.73	.465
Sleep
Average bedtime (hh:mm)	22:54 ± 00:28	23:18 ± 00:32	00:24 ± 00:22	6.25	< .001
Average waketime (hh:mm)	05:36 ± 00:09	07:18 ± 00:25	01:43 ± 00:28	40.59	< .001
Sleep onset latency (min)	25 ± 8	26 ± 13	1 ± 14	0.26	.794
Number of naps	2.3 ± 1.7	1.3 ± 1.1	− 1 ± 1.7	− 2.22	.023
24-h sleep duration (h)	6.5 ± 0.6	7.3 ± 0.5	0.8 ± 0.6	4.28	< .001
Average nap duration (min)	46 ± 37	42 ± 36	− 4 ± 57	− 0.65	.521
Recovery c
∆ HR @ 60% (bpm)	− 1 ± 16	− 8 ± 13	− 7 ± 21	− 1.87	.080
∆ HR @ 80% (bpm)	− 1 ± 15	− 7 ± 17	− 6 ± 26	− 1.32	.202
∆ HR @ 90% (bpm)	− 1 ± 16	− 4 ± 19	− 4 ± 24	− 0.64	.532
∆ RPE @ 60%	− 0.6 ± 5.1	0.0 ± 1.2	0.7 ± 5.2	0.54	.594
∆ RPE @ 80%	− 0.3 ± 4.9	0.7 ± 1.5	1.0 ± 4.7	0.86	.401
∆ RPE @ 90%	0.1 ± 3.1	1.1 ± 3.3	1.0 ± 3.7	0.99	.331
General recovery	2.6 ± 0.8	2.8 ± 0.6	0.2 ± 0.6	1.37	.184
Physical capacity	3.0 ± 0.8	3.1 ± 0.8	0.1 ± 0.7	0.62	.538
Mental capacity	3.2 ± 0.8	3.5 ± 0.9	0.3 ± 0.9	1.81	.084
Emotional balance	3.6 ± 1.0	3.8 ± 1.0	0.2 ± 0.9	1.16	.261
General stress	2.9 ± 1.3	3.0 ± 1.1	0.1 ± 1.6	0.44	.664
Muscular stress	3.0 ± 1.2	3.0 ± 1.0	0.1 ± 1.3	0.27	.793
Lack of activation	2.4 ± 1.4	2.0 ± 1.1	− 0.1 ± 1.6	− 1.05	.305
Emotional disbalance	1.8 ± 1.3	1.7 ± 1.1	− 0.4 ± 1.2	− 0.19	.855

^aThe final column presents the mean difference (late–early), associated t-values, and p-values for paired comparisons

^bSession load was calculated as session RPE multiplied by session duration

^cMeasures of perceived recovery are provided on a scale from 0 to 6

^dSD standard deviation, HR heart rate, RPE rate of perceived exertion

Briefly, comparing the first and second camp, significant differences were observed in baseline sleep duration (mean difference = − 0.9 ± 0.8 h, p < 0.001), baseline wake time (mean difference = − 00:37 ± 01:00, p = 0.008), and pre-camp 800 m performance (mean difference = − 11.8 ± 17.1 s, p = 0.005). No significant differences were observed for baseline bedtime (mean difference = 00:09 ± 00:54, p = 0.422), baseline chronic training load (mean difference = 81 ± 353, p = 0.244), training-camp session load (mean difference = -70 ± 299, p = 0.237) and pre-camp 100 m performance (mean difference = − 1.5 ± 3.6 s, p = 0.084).

Sleep

The effect of the condition on sleep duration is presented in Fig. 4a and is reflective of path a₁ in Fig. 1. Sleep duration was significantly greater in the late condition compared to the early condition (early: 5.9 ± 0.6 h; late: 6.9 ± 0.8 h; mean difference: 1.0 ± 0.5 h; t = 14.0, p < 0.001). This increase was accompanied by a significant delay in both bedtime (p < 0.001), and wake-up time (p < 0.001), with the latter showing a larger shift (Table 1). Sleep efficiency was significantly lower in the late condition compared to the early condition (early: 81.0 ± 4.3%; late: 79.6 ± 4.4%; mean difference: –1.3 ± 3.7%; t = − 2.2, p = 0.03; Fig. 4b). No significant differences were observed in perceived sleep quality (early: 3.3 ± 1.0; late: 3.3 ± 0.7; mean difference = 0.0 ± 0.9; t = 0.7, p = 0.50; Fig. 4c) or in sleep onset latency (p = 0.53; Table 1).

Fig.4.

Mean values (bars) and individual data points (dots and lines) for average sleep duration (A), average sleep efficiency (B), and average perceived sleep quality (C) in the early (pink) and late (blue) condition. Asterisks indicate significant differences between conditions: *p < .05, ***p < .001

Considering sex differences, no condition × sex interaction effects were observed (p > 0.12). Across both conditions, men slept significantly less than women (men: 6.1 ± 0.9 h; women: 6.7 ± 0.7 h; p = 0.01) and went to bed significantly later (men: 23:24 ± 0:29; women: 22:48 ± 0:26; p < 0.001). No differences between men and women were observed for wake-up time, sleep efficiency, perceived sleep quality, or sleep onset latency (p > 0.18).

The effect of condition on napping behavior is presented in Table 1. There were significantly fewer naps in the late compared to the early condition, reflecting an estimated reduction of 58% (p = 0.03). Nevertheless, 24-h sleep duration was significantly greater in the late condition (p < 0.001), with men sleeping less than women (men: 6.6 ± 0.2 h, women: 7.0 ± 0.5; p = 0.04). No significant difference was observed in average nap duration between conditions (p = 0.53) and no significant main effects of sex were observed for napping frequency and duration (p > 0.30).

Recovery

Changes in heart rate and RPE during the standardized warm-up did not differ between conditions at any intensity (p > 0.08; Table 1). For perceived measures of recovery, there was a significant condition × day interaction on perceived emotional balance (t = 2.1, p = 0.04). Emotional balance declined significantly across camp days in the early condition (b = − 0.1, SE = 0.03, t = − 3.3, p = 0.001), but not in the late condition (b = − 0.0, SE = 0.03, t = − 0.3, p = 0.75). There were no significant condition x day effects for any other recovery-related items (p > 0.15), nor main effects of condition over the course of the training camp (p > 0.16; Table 1).

Performance

As shown in Table 2, no significant effects were observed for 100 m performance (p > 0.29). For the 800 m, there was no significant interaction effect or main effect of condition (p > 0.20), but a significant main effect of timepoint indicated overall improvement from pre- to post-training camp (p = 0.01). Regarding motivation, no significant interaction effect or main effect of timepoint was found (p > 0.10); however, a significant main effect of condition indicated greater motivation in the late compared to the early condition (p = 0.01).

Table 2.

Performance and motivation values across conditions (early and late) and timepoints (pre and post)

		Pre mean ± SD	Post mean ± SD	Post–Pre mean ± SD	Intervention	Timepoint	Intervention x timepoint
100 m (s)	Early	66.1 ± 5.4	65.9 ± 5.4	− 0.2 ± 1.9	F1,19 = 1.18 p = .291 pη2 = .06	F1,19 = 0.03 p = .858 pη2 < .01	F1,19 = 0.24 p = .629 pη2 = .01
	Late	66.9 ± 5.5	67.0 ± 6.0	0.1 ± 1.8
800 m (s)	Early	607.6 ± 37.2	600.2 ± 36.3	− 7.4 ± 20.1	F1,20 = 1.72 p = .204 pη2 = .08	F1,20 = 7.32 p = .014 pη2 = .27	F1,20 = 0.02 p = .884 pη2 < .01
	Late	611.7 ± 37.7	605.2 ± 31.7	− 6.6 ± 13.5
Motivation	Early	56.8 ± 18.6	49.3 ± 22.1	− 7.5 ± 18.7	F1,23 = 7.58 p = .011 pη2 = .25	F1,23 = 2.87 p = .104 pη2 = .11	F1,23 = 1.34 p = .259 pη2 = .06
	Late	64.1 ± 12.8	63.6 ± 20.1	− 0.5 ± 18.6

^aANOVA results for the intervention × timepoint interaction and main effects of intervention and timepoint are included

^bSD standard deviation, pη² partial eta squared

Mediation Analysis

The effect of the condition on mental capacity, accounting for sleep duration, is illustrated in Fig. 5. No significant association was found between sleep duration and mental capacity (b = − 0.0 SE = 0.1, t = − 0.28, p = 0.79; path m₁), nor was there a significant direct effect of the condition on mental capacity after adjusting for sleep duration (b = 0.7, SE = 0.4, t = 1.81, p = 0.07; path a₂), although the result approached significance. No significant condition × day effect (p = 0.21) was observed.

Fig.5.

Relationship between sleep duration and mental capacity in the early (top) and late (bottom) condition. Each panel displays a 2D binned heatmap showing the distribution of observations across sleep duration and mental capacity levels. Color intensity reflects the number of observations per bin, with brighter colors indicating higher densities

The relationship between 100 m performance after the camp, perceived mental capacity on that day, and the condition is depicted in Fig. 6. There was no significant relationship between mental capacity and 100 m performance (b = − 0.2, SE = 0.3, t = − 0.5, p = 0.60; path m₂), nor was there a significant direct effect of condition on 100 m performance (b = 0.6, SE = 0.8, t = 0.8, p = 0.46; path c).

Fig.6.

Relationship between pre- and post-100 m performance in the early (left) and late (right) condition. Each point represents an individual observation. The diagonal line indicates equal performance pre- and post-camp, with the shaded area representing a ± 2% range around this line. Points below the line indicate improved performance; points above indicate decreased performance. Color intensity corresponds to mental capacity levels, with brighter colors representing higher mental capacity

Discussion

The main findings of the current study were: (1) sleep duration increased by one hour on average when training commenced at 09:00 AM compared to 07:00 AM; (2) and the decline of emotional balance was mitigated when training time was delayed; (3) however, there were no further significant effects on recovery or swimming performance; (4) sleep duration did not explain the relationship between delayed morning training and perceived mental capacity; (5) nor did perceived mental capacity explain the relationship between delayed morning training and 100 m performance.

Sleep

Previous studies have shown that later training sessions are associated with longer sleep duration compared to earlier ones [17, 18, 20]; however, these studies used observational designs with naturally varying training times. The present study, employing an experimental design, supports these findings and extends them by providing causal evidence that delaying early morning training leads to increased sleep duration. This was the case for both men and women, although men slept significantly less than women regardless of condition, alongside later bedtimes. These findings are in line with previous research showing that males develop later bedtimes than females, with the greatest sex differences during adolescence [35, 36]. Overall, the one-hour increase observed here was insufficient for most swimmers to achieve the recommended >8 h of sleep [15, 37]. Considering that wake times in the late condition were earlier than free-day wake times reported in comparable adolescent swimmer samples [17, 38], a further delay in training start time might have resulted in additional sleep extension. Sleep efficiency was also below the recommended 85% [39], in both conditions, and decreased slightly when training was delayed, which may be related to longer sleep duration. Despite this, perceived sleep quality remained unchanged, consistent with the findings of Ashby et al. [17].

A common strategy among athletes to compensate for nocturnal sleep loss is daytime napping, which has a known potential to enhance physical performance and reduce perceived fatigue [40–42], with some evidence suggesting these benefits are more pronounced following sleep restriction [41]. The current findings suggest that swimmers indeed attempted to compensate for the reduced nocturnal sleep, reflected by 58% more naps in the early than in the late condition; however, this only slightly attenuated the difference in sleep duration between conditions.

Together these findings suggest that while delaying training start to 09:00 extends sleep duration, it may be insufficient to achieve the recommended sleep quantity and quality for adolescent swimmers—at least within the limited time frame of a training camp.

Recovery

Although sleep duration increased by one hour when training was delayed, the impact of morning training delay on recovery was limited. While the decline in emotional balance across the training camp was mitigated, there was no further effect of delaying early morning training on other recovery measures. This was accompanied by the absence of a significant relationship between sleep duration and perceived mental capacity. These findings are in line with previous research, albeit in a very different population of adult ultra-runners, where Roberts et al. [43] reported no significant relationship between sleep duration and mental capacity the following day. On the other hand, Roberts et al.[43] did find a small yet significant relationship between sleep duration and overall recovery. To avoid fishing and inflating the type I error, the present study focused on mental capacity a priori, given the influence of sleep on attention and mood [2]. However, including an additional SRSS items in the analysis, such as emotional balance, may have yielded different results. These results suggest that sleep duration might not be as important for perceived recovery as presumed. However, it should be considered that the present study covered only a narrow range of sleep durations over a limited timespan. It therefore remains unclear whether a relationship between sleep duration and recovery would emerge across a broader range of sleep durations, at different parts of that range, or following longer-term adaptation. Future research is needed to examine the relationship between sleep and recovery across a broader spectrum of sleep durations and over longer period of time.

Performance

The current study is the first to assess the effect of early morning training delay on swimming performance. The findings suggest that the change in performance across an 8-day training camp is not altered by delaying early morning training from 07:00 AM to 09:00 AM. In contrast, a meta-analysis by Cunha et al.[21], indicated that sleep extension of 46–113 min can improve sports performance. Additionally, improvements in motor and cognitive performance have been reported after 4–7 days of sleep extension in non-athlete populations [45–47]. This discrepancy may relate to differences in performance type, as Fullagar et al. [48] suggested that cognitively demanding tasks may be more susceptible to performance decrements upon sleep loss. Additionally, in the studies included by Cunha et al.[21], sleep extension typically resulted in total sleep durations exceeding 8 h, whereas in the present study mean sleep duration reached only 6.9 h. Moreover, unlike the studies summarized by Cunha et al., the present investigation was conducted during a period of high training load, which may have masked any potential performance benefits of sleep extension, though this remains speculative. Notably, despite the intensive camp, 100 m performance was maintained and 800 m performance even significantly improved. However, the observed improvement was within the standard error of measurement of 8.7 s previously reported for hand-timed 800 m swimming performance and should be interpreted accordingly [49]. Nonetheless, daily training load during the camp doubled compared to baseline, indicating a substantial increase in training load without any observed performance decrement. These findings suggest that swimming performance may be relatively robust to both high training load and extended sleep duration, in the short term.

In contrast to performance, motivation prior to performance testing was generally greater in the late than in the early condition. This is in line with previous research showing greater motivation following ~1.4 h sleep extension over four days [47]. Although there is also evidence for a time-of-day effect on motivation [50], the variation in test time in the current study (~ 2 h) was relatively small. Greater motivation in the late condition may have implications for training adherence and quality, and might thereby influence performance over the longer term.

Limitations

The present study has some limitations that should be considered when interpreting the results. First, the two camps were separated by 9 months, allowing for period effects (systematic differences in the participants’ baseline state due to factors such as performance development or maturation). In fact, these period effects were observed for baseline sleep measures and 800 m performance. Although baseline values did not differ between conditions, these effects should be considered when interpreting the results. Unfortunately, a shorter time interval was not feasible, as both camps had to fall within high-training-load periods, occur in seasonally comparable periods, and align with the schedules of two national-level youth teams from different nations. Second, missing values reduced the sample size. While the study was adequately powered to detect changes in our primary outcome, sleep duration, this cannot be assumed for the secondary outcomes, for which sample sizes but also other determinants of statistical power could vary. Therefore, the respective p-values should be interpreted with explicit caution. Relatedly, napping was not included in the mediation analysis because it would have reduced the sample size and added complexity to the model. Considering that napping is known to enhance physical performance and reduce perceived fatigue [40–42], this may have influenced the results. Third, this study did not record the time of day at which the SRSS questionnaire was completed. Since time of day can influence perceived mental and physical fatigue independently of preceding sleep duration [44], this may have affected the SRSS results. However, participants were provided with a link to the questionnaire in the morning and instructed to complete it at that time. Nevertheless, future research should consider the timing of questionnaire completion. Relatedly, test times differed between conditions, which could be considered as a potential confounder of the camp-induced performance change. However, given the small discrepancy in test times, the inconsistent evidence for diurnal variation in endurance performance in the literature [51], and, in line with this, the absence of a main effect of condition on absolute performance, this is considered unlikely. Moreover, a review by Chtourou & Souissi [52], suggested that training adaptations are most apparent at the time of day at which training took place, supporting the present approach of aligning test times with training times. Additionally, the use of 10-s carotid palpation for heart rate and handheld stopwatches for swimming performance are less precise than their electronic alternatives, which should be considered when interpreting these outcomes. However, electronic timing was not available, measuring heart rate through chest-straps is impractical in men due to slippage during swimming, and neither outcome was a primary measure of the study. More broadly, the applied setting meant that conditions could not be fully standardized, potentially increasing variability and influencing the results. Nevertheless, this real-world context is also a strength of the study, enhancing its ecological validity and practical relevance. Finally, most swimmers in the current study exhibited a neutral chronotype; this limited variation precluded a meaningful analysis with chronotype as a moderator and limits the generalizability of the findings to early-birds and late-owls. It would be interesting to explore whether responses to delayed early morning training differ between chronotypes, specifically whether late-owls benefit more from delayed morning training time.

Conclusion

The present study assessed the effect of delaying early morning training on sleep, perceived recovery, and swimming performance in high-performing adolescent swimmers. The results demonstrate that delaying morning training start from 07:00 AM to 09:00 AM increases sleep duration by one hour, but this was insufficient for swimmers to reach the recommended 8–10 h of sleep. Additionally, delaying early morning training did not significantly affect perceived recovery or swimming performance in response to the training load imposed during the camp. These findings may be explained by chronic sleep deprivation, the high training load, naps during the training camp. Moreover, the relatively short study period may have limited the opportunity for changes in performance or recovery to manifest. Nevertheless, the study provides evidence that early morning training reduces sleep duration and highlights the persistent sleep challenges faced by national-level adolescent swimmers. Coaches should plan early morning sessions carefully and, when possible, delay morning training start times to support healthy sleep in their athletes. Future research is warranted to explore strategies that enable swimmers to meet sleep recommendations and to examine the effects on recovery and performance over a longer period.

Abbreviations

ANOVA

Analysis of variance

DAG

Directed acyclic graph

RPE

Rate of perceived exertion

SE

Standard error

sRPE

Session rate of perceived exertion

SRSS

Short Recovery and Stress Scale

Author Contributions

SF and TM conceptualized the study design. MB and SF further developed the study protocol and collected the data. MB performed the data analysis with support from AH. MB drafted the manuscript. All authors critically revised the manuscript, approved the final version, and agree to be accountable for all aspects of the work.

Funding

Open Access funding enabled and organized by Projekt DEAL. This project was funded by the Federal Institute of Sport Science (Bundesinstitut für Sportwissenschaft, BISp), Germany, and financially supported by the German Swimming Federation (Deutscher Schwimm-Verband, DSV).

Data Availability

The datasets generated and/or analyzed during the current study are available in the SleepTrain_analysis repository on GitHub, https://github.com/MBrandts98/SleepTrain_analysis.

Code Availability

All analysis code is openly available on GitHub: https://github.com/MBrandts98/SleepTrain_analysis.

Declarations

Ethics Approval and Consent to Participate

The study was approved by the Human Research Ethics Committee of the Ärztekammer des Saarlandes, Saarbrücken, Germany (approval no.: 114/22), and conducted in accordance with the Declaration of Helsinki, with the exception that it was retrospectively registered. Written informed consent was obtained from all participants and, for participants under 18 years of age, from their parents or legal guardians.

Consent for Publication

Consent for publication was obtained from all participants and/or their legal guardians.

Competing Interests

The authors declare that they have no competing interests.