Participants
An A-priori power analysis (G*power, version 3.1) was used to establish a minimum sample size (n = 30) for the present investigation. Sample size calculations were based on an effect size of 0.8 and a type I (α) error rate of 5%. A convenience sample of 36 female football players volunteered to take part; one participant withdrew following baseline data collection and was removed from the study and one participant withdrew in Week 3; meaning n = 34 completed the study. All participants gave informed consent prior to data collection. All participants (subject demographics detailed in Table 1) were part of the U21 or First Team squad at their football clubs in the United Kingdom, and had played regularly in the Women’s Super League (WSL) or the WSL Academy League in the previous season. Throughout the study, participants slept in their usual, home-based environment.
Table 1
Subject demographics
|
|
Mean
|
SD
|
Age (years)
|
20.3
|
1.4
|
Height (cm)
|
164.2
|
11
|
Mass (kg)
|
62.1
|
10.8
|
Weekly training hours (football)
|
10.4
|
4.1
|
Weekly training hours (gym based)
|
4.6
|
0.9
|
Across all participants, 15 reported regularly taking hormonal contraceptives (type unspecified), whilst 19 were classified as naturally menstruating women. Prior to the commencement of the study, all participants were informed of study requirements and gave informed consent. Participants were excluded if they reported a pre-existing sleep disorder, had a menstrual cycle outside the range of 21–35 days or did not give informed consent. Institutional ethical approval was issued (approval number 2023–12534, University of Brighton, UK) in accordance with the Declaration of Helsinki 1964 (revised 2013).
Experimental design
A randomised, controlled trial with repeated measures was used to assess whether sleep hygiene interventions could affect strength and power performance, and whether the method of SH delivery (individualised education vs. group) has any effect on sleep indices and performance. Given the potential for seasonal adjustment of sleep patterns (Allebrandt et al., 2014) and the potential variability of sleep patterns throughout a football season, it should be noted that data was collected during pre-season in July and August.
A random number generator (www.randomizer.org) was used to allocate participants into one of three groups: Control, Group SH, Individualised SH with n = 12 in each. A schematic of the study protocol is detailed below in Fig. 1.
Sleep monitoring – Week 1, Week 4 and Week 7
All participants completed the Athlete Sleep Behaviour Questionnaire (ASBQ) (Driller et al., 2018) to determine current sleep behaviours and sleep hygiene. The survey asked participants to rate on a Likert scale how frequently they engage in specific behaviours (never = 1, rarely = 2, sometimes = 3, frequently = 4, always = 5). Scores were summed to provide an ASBQ global score; higher scores were considered indicative of worse sleep habits and sleep hygiene. Participants also completed the reduced morningness:eveningness questionnaire (rMEQ, Adan and Almirall, 1991), with scores summed to determine chronotype classification as reported in Adan and Almirall (1991): definitely morning type (22–25), moderate morning type (18–21), neither type (12–17), moderate evening type (8–11), definitely evening type (4–7).
All participants were allocated an actigraph (GeneActiv Original, Activinsights, Cambridge UK) which they were instructed to wear continuously, only removing them for pre-season matches. The device contains a triaxial MEMS-accelerometer with a range of ± 8 g and a sensitivity of ≥ 0.004 g (te Lindert and Someren, 2013). It recorded both motion-related and gravitational acceleration and has a linear and equal sensitivity along the three axes. Devices were set with a sampling rate of 50Hz and participants were instructed to wear the device on whichever wrist they felt more comfortable with (Driller et al., 2017). Every week where sleep data was collected, each morning participants were asked to provide self-reported sleep quality (Likert scale response) and “lights out” time and wake up time via a Microsoft Forms questionnaire, sent by email to each participant. Participants were sent a daily reminder to complete this via a text message from coaches. If a missing data set was detected, participants were reminded again to submit their data at lunchtime and received an additional reminder from coaches to submit their data the following day – there was a 6% occurrence rate of this throughout the testing period. Actigraphy-derived sleep parameters are detailed in Table 2 below:
Table 2
Actigraphy-derived sleep parameters
Sleep variable
|
Units
|
Description
|
Latency
|
min
|
Number of minutes from time at lights out to sleep onset
|
Duration
|
hh:mm
|
Time at start of sleep interval to end of sleep interval, minus number of minutes awake (WASO)
|
Efficiency
|
%
|
Sleep duration divided by time in bed x 100
|
Strength and power assessment – Week 1, Week 4 and Week 7
In Week 1, 4 and 7 of the study, all participants completed testing of countermovement jump (CMJ), and isometric mid-thigh pull (IMTP). All athletes had prior experience of both tests as part of physical testing requirements from their club and had completed the test regularly throughout the previous season. Week 1 was considered as baseline data. Participants followed a standard 15-minute warm up following a RAMP protocol (Jeffreys, 2006) led by a strength and conditioning coach, after which warm up repetitions of each test were carried out (detailed below). Strength and power tests were conducted by the same tester throughout the study. Given the potential for circadian influence on performance (Drust et al., 2005), performance testing was carried out at the same time of the day throughout the study.
Countermovement jump (CMJ)
The CMJ test was conducted prior to the isometric mid-thigh pull and was performed on VALD ForceDecks (Force Decks, VALD Performance, FD4000, Queensland, Australia) sampling at 1000 Hz. Participants were instructed to keep their hands on their hips to eliminate arm swing and perform a fast downward motion to around 90 knee flexion, followed by an immediate upward vertical jump as high as possible, all in one sequence (Slinde et al., 2008). Prior to the test attempts, participants performed 2 jumps at 75% maximal effort, each separated by 2 minutes; this was designed to act as an extended warm up, additional familiarisation, and reinforce test technique (Carroll et al., 2019). For the test attempts, participants were instructed to deliver a maximal attempt and performed the test 3 times, each attempt separated by 2 minutes. Jump height (cm) was calculated from impulse momentum (Frick, 1991; Linthorne, 2001) computed by the VALD ForceDecks software (VALD Performance, FD4000, Queensland, Australia). Software detected the initiation of movement as a 30 N deviation from the initial body weight calculation, eccentric to concentric phase moment as the lowest centre of mass displacement, and take-off as the moment the vertical forces fell 30 N below body mass (Merrigan et al., 2021). Perez-Castilla et al. (2021) stated the importance of defining and using a consistent threshold to identify take off and the importance of using a consistent threshold to enable comparisons between trials and testing sessions. The best of the 3 trials was used for analysis.
Isometric mid-thigh pull (IMTP)
Methodological guidelines from Comfort et al. (2019) were followed in the administration of this test, with testing carried out on VALD ForceDecks (VALD Performance, FD4000, Queensland, Australia) sampling at 1000 Hz. Participants were initially asked to self-select a start position that reflected the start of the second pull of a clean; this allows for athletes’ individual anthropometrics to be considered in the adoption of an optimal pulling position (Comfort et al., 2019). Knee and hip angles were then checked with a hand-held goniometer to ensure knee angles were within the range of 125–145 and hip angles were within the range of 140–150 (Comfort et al., 2019) and straps were used by all athletes to mitigate the risk of grip strength becoming a limiting factor (Comfort et al., 2019). Prior to testing, single reps were performed at 50% maximal effort for 5 seconds, and 75% maximal effort for 5 seconds, each separated by 60 seconds rest, with the purpose of serving as further warm up, additional familiarisation and reinforcing test technique (D’os Santos et al., 2017). For the beginning of the maximal attempts, the tester gave the athlete a countdown of 3,2,1 before the initiation of the test. Participants were instructed to “push their feet into the ground as hard and fast as possible”, maintaining the tension for a period of 5 seconds timed by the tester. This verbal cue has been previously shown to result in greater peak force than focusing on internal cues (Halperin et al., 2019). Each trial was separated by 2 minutes rest. The highest force generated was reported as the absolute peak force (PF) with relative PF then calculated by dividing this by the body mass of each participant (Haff et al., 2015). The best of the 3 trials was used for analysis.
Group sleep hygiene education – Week 2
A 40-minute group sleep hygiene education was delivered to both SH group and Individualised SH group in Week 2 of the study; the session was led by a strength and conditioning coach with specific expertise on athlete sleep. The session took place in a private room in the athletes’ training ground, with two technical coaches also present. The focus of the session was to provide athletes with general information regarding SH and provide practical tips on the following areas – maintaining a regular bedtime and wake time (Phillips et al., 2017), maintaining a cool and dark bedroom (Dautovich et al., 2022), avoidance of light-emitting screens before bed (Driller et al., 2019), and implementation of relaxation techniques before bed (McCloughan et al., 2014). The session was delivered in a way that focused on positive reinforcement and potential performance benefits, rather than negative impacts of bad habits. The session concluded with participants writing down 2–3 practical changes to their sleep habits which they would aim to implement following the session.
Individual sleep hygiene education – Week 5 and 6
Participants within the Individualised SH group were each given one one-on-one session per week, delivered via Microsoft Teams, where they were provided with individualised advice on their sleep hygiene, based on week 1 sleep data and self-reported perception of areas they needed to improve. Any areas reported above a “3 = sometimes” on the ASBQ was discussed as an area for improvement with each participant. Discussions aimed to establish and prioritise practical changes participants could implement daily and to overcome any concerns regarding changes. Participants were encouraged to ask questions and to focus on their own specific requirements, and each session concluded with the participant writing down 2–3 key areas of focus for their sleep habits which they would aim to implement. The initial individualised session for each participant lasted 30 minutes, with the second session lasting 20 minutes, to include a review of the success of previous action points, discussions of any concerns, and if necessary, amendments of any practical advice based on individual circumstances.
Statistical analysis
Descriptive statistics (mean ± SD) were calculated for all variables. Data was checked for normality using Shapiro-Wilk tests, and inspection of skewness-kurtosis. Within session test-retest reliability was assessed using two-way mixed intraclass correlation coefficients (ICC) for all performance outcome variables. ICC values were deemed as poor if ICC < 0.50; moderate 0.50–0.74; good if 0.75–0.90; and excellent if ICC > 0.90 (Koo and Li, 2016). Split-plot ANOVA were used to examine the effects of SH education on strength and power outcomes, by using a 3 (group: Individual SH, group SH, control) by 3 (time: week 1, week 4, week 7) design. Sphericity was verified by Mauchly’s test. For each variable, the main effects for group x week were examined, as well as the group x time interaction. Partial eta squared was reported to give an indication of effect size, with values of 0.01, 0.06 and 0.14 considered as small, medium and large effect sizes respectively (Girard et al., 2013). For chronotype, data was analysed from raw rMEQ scores rather than classifications. Statistical analyses were performed on SPSS (version 29.0, SPSS, Chicago, Illinois) and Microsoft Excel (Microsoft Office 365, Microsoft Corporation, USA).