REF-1558 · Brandts et al., 2026 · Sports Medicine - Open · Paper · PubMed · Free full text

Does a later morning session pay?

Two hours later gave adolescent swimmers 1 h more night sleep, 0.8 h counting naps, but no significant difference in time-trial change over 8 days. Both arms did the same doubles.

Main caveat: Power unknown for performance and recovery, probably low. Sample sizes per analysis ranged from 14 to 25, power was estimated only for sleep, and many items…
moderate evidenceMeasured on highly trained athletes

Thirty-four swimmers started; 27 finished both 8-day camps, in January and October 2023. Each camp had 13 sessions, six in the morning and seven in the afternoon, the same structure under the same head coach, with morning sessions at 07:00 or 09:00 and the order reversed at the second camp.

Night sleep rose from 5.9 ± 0.6 to 6.9 ± 0.8 h (mean difference 1.0 ± 0.5 h, p < 0.001), wake time moved from 05:36 to 07:18 and bedtime from 22:54 to 23:18. The early arm took more naps, so the 24-h gain narrowed to 0.8 ± 0.6 h. Sleep efficiency fell slightly in the late condition (81.0 to 79.6%, p = 0.03).

Time-trial changes showed no significant difference: 100 m interaction p = .629, 800 m p = .884. Across both conditions 800 m improved (p = .014), within the 8.7-s error the authors cite for hand timing. Emotional balance declined over the early camp only; other recovery scores showed no significant condition effect. Motivation before the tests was higher in the late condition.

Mediation showed no indirect path through sleep to mental capacity, nor through mental capacity to 100 m time. Our reading: a clean demonstration that training hour moves sleep, and a weak null on whether that hour moves performance in eight loaded days.

Protocol

Design: randomised crossover by random-number generator, without stratification; 9 months between camps because of national-team calendars. Measures taken in the 2 weeks before each camp and through it. Ethics approval no. 114/22. Registered retrospectively in February 2026.
Camp: 8 days at an Olympic training centre, shared accommodation away from home; first session on the afternoon of day 1, last on the afternoon of day 7, tests on day 1 and day 8.
Naps: banned at first, then allowed from day 3 in both camps after coaches worried about drop-outs, and logged.
Performance: 100 m main stroke and 800 m freestyle in a 50-m pool, timed by hand. Test hour followed the condition (07:00-07:45 or 09:00-10:00).
Sleep: wrist actigraphy with a diary; nights with a wake time after the session start were excluded.
Recovery: Short Recovery and Stress Scale each morning; heart rate by carotid palpation and RPE in the warm-up. Illness days excluded.
Analysis: linear mixed-effects models and repeated-measures ANOVA; regression mediation with mental capacity chosen in advance. Samples across analyses 14 to 25. Power: simulation for sleep only (100%); target of 30 set by feasibility after a calculation for another test gave 155.
Funding and interests: Federal Institute of Sport Science (Germany) and German Swimming Federation; the authors declare no competing interests. Data and code are public.

Key Results

MeasureEarly (07:00)Late (09:00)Test
Night sleep (h)5.9 ± 0.66.9 ± 0.8difference 1.0 ± 0.5, p < 0.001
24-h sleep with naps (h)6.5 ± 0.67.3 ± 0.5difference 0.8 ± 0.6, p < .001
Wake time05:36 ± 00:0907:18 ± 00:25p < .001
Bedtime22:54 ± 00:2823:18 ± 00:32p < .001
Number of naps2.3 ± 1.71.3 ± 1.1p = .023
Sleep efficiency (%)81.0 ± 4.379.6 ± 4.4p = 0.03
Perceived sleep quality3.3 ± 1.03.3 ± 0.7p = 0.50
Daily session load (AU)1005 ± 258964 ± 296p = .465
100 m (s), pre → post66.1 ± 5.4 → 65.9 ± 5.466.9 ± 5.5 → 67.0 ± 6.0interaction p = .629; condition p = .291
800 m (s), pre → post607.6 ± 37.2 → 600.2 ± 36.3611.7 ± 37.7 → 605.2 ± 31.7interaction p = .884; time p = .014
Motivation, pre → post56.8 ± 18.6 → 49.3 ± 22.164.1 ± 12.8 → 63.6 ± 20.1condition p = .011; interaction p = .259
Mental capacity (0-6)3.2 ± 0.83.5 ± 0.9p = .084
Emotional balance across daysb = −0.1, p = 0.001b = −0.0, p = 0.75condition × day p = 0.04
Warm-up heart-rate change at 60% (bpm)−1 ± 16−8 ± 13p = .080
No clear difference, not "the same." No equivalence test was run. The authors themselves warn that secondary outcomes may be underpowered and their p values read with caution.
Sex. No condition × sex effect (p > 0.12). Across conditions boys slept less than girls (6.1 ± 0.9 against 6.7 ± 0.7 h, p = 0.01). The study was not built to compare sexes.
Period effect. Between camps, baseline sleep differed by 0.9 ± 0.8 h (p < 0.001) and pre-camp 800 m by 11.8 ± 17.1 s (p = 0.005); baseline did not differ by condition.
Below target in both arms. The 6.9 h in the late arm is under the 8-10 h the authors cite for adolescents.

Mechanisms

Measured: later session, later wake, slightly later bed, fewer naps, more total sleep, slightly lower efficiency.
Authors' interpretation: the extra hour was not enough to reach the sleep level at which sleep-extension studies show benefits; heavy load, chronic short sleep, naps and short duration may explain the null. They call the load explanation speculative.
Not measured: physiology beyond warm-up heart rate by palpation, sleep architecture, adaptation over weeks, training quality within sessions.

Practical Application

Does a later morning session pay? Two hours later gave adolescent swimmers 1 h more night sleep, 0.8 h counting naps, but no significant difference in time-trial change over 8 days. Both arms did the same doubles.
Good for: moving a morning session later when sleep is the concern. Sleep rose in a randomised comparison, with no clear difference in swimming. Steadier emotional balance is a side result (one interaction among many items, p = .04). Motivation was higher in the late condition, but already on day 1 before the first session (our reading), so the delay is not shown to have raised it.
Good for: resisting a claim that an early session has a proven performance cost. Eight days in adolescents did not show one, though this null is weak.
Not good for: saying sleep does not matter. Night sleep stayed under 7 h in both arms, the camps were short, and hand timing is coarse.
Not good for: whether to train twice a day. Both arms did.
Not good for: adult cyclists or runners on normal weeks, by direct evidence; a transfer of direction at most.
Not from this paper: how many hours an athlete needs. SYNTH-07 states the base holds no entry on this; PHYS-34 calls the need individual.

Limitations

Power unknown for performance and recovery, probably low. Sample sizes per analysis ranged from 14 to 25, power was estimated only for sleep, and many items were tested.
Nine months between camps in adolescents, with a measured period effect in sleep and 800 m.
Retrospective registration, so pre-specification cannot be checked from the registry.
Hand timing of time trials (error 8.7 s for 800 m as cited) and heart rate by palpation.
Test hour differed by condition, and the post-test on day 8 came the morning after the last session on the afternoon of day 7.
Naps were allowed mid-study and are not in the mediation; this limits the path from sleep to recovery.
Short. Eight days cannot show a long-term effect; the authors say so.
Questionnaire time. The time of day each person completed the recovery questionnaire was not recorded.
Not blinded, and it could not be.
How this entry was checked and changed
Why this level

controlled crossover with n ≥ 15, read from the full text (CC BY). It sits at the low edge of the band: registration came more than two years after the camps (our arithmetic), performance analyses had 20-21 swimmers (our reading of the degrees of freedom), power was calculated only for sleep, and there was a period effect between camps