A Replicate Crossover Trial on Nutritional Supplementation in Association Football (Soccer)

Background information \& study rationale: A number of investigations explored the effects nutritional supplements on proxy measures of sleep and recovery. Notably, different studies examined the properties of high glycemic index carbohydrates, tryptophan, tart cherries, glycine, magnesium, and L-theanine predominantly in isolation in sample of participants from general populations with and without sleep disorders. However, none of these studies involved elite athletes nor assessed the effects of and potential for personalizing nutritional supplementation according to evidence from appropriately designed experimental investigations involving the repeated administration of supplementation and placebo conditions adopting a replicate crossover trial design. With this in mind, and for the first time, the project seeks to quantify the magnitude of inter-individual variability in the effects of a novel nutritional blend supplementation in the form of a mixed juice (15000 mg cherry juice concentrate, 220 mg cherry extract, 250 mg cocoa extract, 200 mg of tryptophan, 100 mg of L-5-Hydroxytryptophan, 3000 mg of glycine, 300 mg of magnesium, 200 mg of theanine - https://sport.wetestyoutrust.com/supplement-search/science-sport/rest-juice) on proxy measures of recovery and sleep using a replicate crossover design. The project seeks to examine the consistency of these responses on repeated occasions. We hypothesize that there is 'true' inter-individual variability (i.e., exceeding random within-subject variability) in the effects of nutritional blend supplementation on actigraphy-based sleep outcome measures and proxies of recovery, and that these responses would be consistent on two occasions. Adequate sleep and recovery are pivotal to promote optimal preparation training and competition as well as foster high performance in elite sport (Fox et al., 2020). Importantly, continuous unperturbed sleep represents a physiological necessity to preserve normal cognitive and neurobehavioral functions (Maquet 2001; Van Someren et al. 2015; Logan and McClung 2019; Walsh et al. 2020). Nevertheless, youth athletes are vulnerable to sleep disorders that generally involve short durations inconsistent with health recommendations due to lifestyle factors, social commitments (e.g., school), coupled with the demands of training and competition (Fox et al. 2020). In this context, Muslim from general and athletic populations are particularly prone to issues of sleep disorders and schedule congestion throughout different seasons of the Gregorian year. Outcomes from a 14-day surveillance period from Aspire Academy youth football players (Lolli et al., 2021) were consistent with this line of evidence that highlighted children, adolescents, and young adults from Qatar do not meet the age-specific sleep recommendations (i.e., 8 hr to 9 hr of sleep per night). Furthermore, recent findings highlighted how the observance of the holy month of Ramadan exacerbated such issues in Aspire Academy student-athletes, with an average 4.2 hr (95% confidence interval, 3.6 to 4.7 hr) in the first block of night sleep immediately after Ramadan onset (Lolli et al., 2024a). From a clinical standpoint, sleep insufficiency issues can potentially disrupt a student-athlete's biological clock via desynchronisation of its key modulators, such as leptin, adiponectin and ghrelin, involved in appetite regulation and energy expenditure (Knutson et al., 2007). A recent randomized, controlled, crossover, 21-day inpatient study revealed experimental sleep restriction (i.e., 4 hr of night sleep) in free access to food conditions (i.e. ad libitum) caused increased energy intake with no changes in energy expenditure, leading to significant weight gain (Covassin et al., 2022). On the other hand, inconsistent eating habits might impair energy balance that, as a consequence, can offset and be detrimental to the youth athlete development process (Mountjoy et al., 2018). To address such issues, while also considering the number of narrative review and consensus statements, a number of studies in the literature examined various nutritional ingredients with proposed recovery and sleep enhancing qualities involving high glycemic index carbohydrates, tryptophan, tart cherries, glycine, magnesium, and L-theanine (Langan-Evans et al., 2023) yet without adopting appropriate research designs to explore the main effects and potential for personalization based on the assessment of interindividual differences in responses explored in elite athletes. Identification of meaningful inter-individual differences in response to nutritional interventions can be challenging. Notably, with conventional parallel-arm or crossover trials, it is not possible to determine whether any apparent difference between participants in the changes from pre- to post-intervention are due to genuine treatment response heterogeneity or are a consequence of random within-subject variability and/or measurement error (Senn 2016). Without repeated administration of trials, a typical crossover design does not allow formal estimation of variance attributable to participant-by-treatment interaction (Atkinson and Batterham, 2015; Senn 2019). Only with knowledge about this interaction can treatment response heterogeneity be quantified properly. As a form of n-of-1 trial, a replicate crossover trial constitutes a pragmatic research design for quantifying treatment response heterogeneity (Atkinson and Batterham, 2015; Senn et al., 2011; Senn 2016). This research design involves repeated administration of the intervention and control/placebo arms of a trial protocol in randomized order on at least two occasions. This design allows a principled estimation of various components of variation to examine whether genuine supplementation response heterogeneity can be distinguished from random within-subject variability for researchers to arrive at appropriate conclusions on the presence and consistency of inter-individual differences in the supplementation administration responses (Senn et al., 2011). Nevertheless, these aspects remain unexplored in the sports performance literature the modern Arab paediatric population from the State of Qatar. Notably, no study in the sports performance literature adopted a replicate crossover design to examine the possibility of personalized recommendations for dietary supplementation for optimal post-training and post-competition recovery management.

Study goals and objectives: To investigate the main effects and whether inter-individual variability exists in free-living objective sleep and subjective training load responses to novel nutritional blend supplementation in the form of a mixed juice (15000 mg cherry juice concentrate, 220 mg cherry extract, 250 mg cocoa extract, 200 mg of tryptophan, 100 mg of L-5-Hydroxytryptophan, 3000 mg of glycine, 300 mg of magnesium, 200 mg of theanine) adopting a replicate crossover design for exploring whether it may work in youth athletes and the potential for supplement administration personalization.

Study Design and Research Population: Considering the nature of a replicate crossover trial being an alternative form of n-of-1 trial in which each participant will represent a distinct trial and will act as his own control (Senn 2019), the present investigation will aim to assess the effects of nutritional supplementation embedded into routine service provision settings. Accordingly, the target sample will involve youth football players (target age range: 13 to 17 years) that will follow a period of preparation in the lead to the FIFA U-17 World Cup.

Description of subjects: Considering the nature of a replicate crossover trial being an alternative form of n-of-1 trial in which each participant will represent a distinct trial and will act as his own control (Senn 2019), the present investigation will aim to assess the effects of nutritional supplementation in routine service provision settings. Accordingly, the target sample will involve youth football players (target age range: 13 to 17 years) that will follow a period of preparation in the lead to the FIFA U-17 World Cup. In the absence of established target difference (Cook et al., 2018) and design-specific variance (Senn, 2019) values relevant to each primary outcome measure considered in this study, and in keeping with procedures adopted to sample size justification in previous research adopting a replicate crossover design (Gonzalez et al., 2024), a minimum sample size of 12 participants would translate to a one-tailed between-cycle correlation of 0.5, with a 90% confidence interval for this correlation ranging from 0.00 to 0.80, assuming a null and a negative correlation would suggest non-rejection of the null hypothesis (r ≤ 0).

Methodology: The study methods definition will follow the recent guidance (Chatters et al., 2024) and the Consolidated Standards of Reporting Trials (CONSORT) 2010 statement extended to randomized (replicate) crossover trials (Dwan et al., 2019). The relevant checklist (Dwan et al., 2019) for the design and conduct of the study will be completed accordingly. Using a replicate crossover design (Senn 2016), participants will complete two identical supplementation and two identical placebo l trials in a randomized order separated by at least four days. The randomization sequence for the four experimental trials will be conducted using the PLAN procedure available in SAS (Deng \& Graz, 2002). The supplement treatment will involve administering a nutritional blend consisting of a mixed juice containing 15000 mg cherry juice concentrate, 220 mg cherry extract, 250 mg cocoa extract, 200 mg of tryptophan, 100 mg of L-5-Hydroxytryptophan, 3000 mg of glycine, 300 mg of magnesium, 200 mg of Theanine sought to be manufactured by Science in Sport (https://sport.wetestyoutrust.com/supplement-search/science-sport/rest-juice) and within safety limits for administration in children and adolescents (Barrett et al., 2013; Bruni et al., 2004; Cynober et al., 2016; Elango, 2023; Fernstrom, 2012; Garlick, 2004; Institute of Medicine Standing Committee on the Scientific Evaluation of Dietary Reference, 1997; Lyon et al., 2011; Merry and Ristow, 2016; Rose et al., 1965). The placebo will involve a 40-g dual-source carbohydrate gel, having a 1:0.8 ratio of maltodextrin and fructose, as an alternative supplementation known for having trivial effects on sleep-related outcome measures (St-Onge et al., 2016). Both the supplement and the placebo will be administered in a random order across 2-night period for each respective main trial separated 5-day washout periods to mitigate carryover effects between trial periods and cycles consistent with the squad microcyle planning. The supplement and placebo administration will mimic a post-match recovery context, with both treatment conditions being administered within 1-hour post-training and coinciding with dinner time. Both the participant and the performance nutritionist administering the supplementation will be blinded to its administration and treatment episode allocation. The washout diet will involve an "eat as usual" protocol consistent with recommendations by Aspire Academy sports nutrition recommendations refraining from any caffeine and nicotine consumption. Participants will be required to undertake a food diary using the remote food photography method as previously utilised in adolescent male football players (Hannon et al., 2021). In brief, participants will be instructed on how to capture photographs at 45 degrees and 90 degrees angles before and after each eating and/or drinking occasion, which will be recorded and time stamped via WhatsApp (WhatsApp Inc., Mountain View, CA) as previously described (Hannon et al., 2021). These data will then be assessed by two accredited sports nutrition practitioners using dietary analysis software to establish daily energy and macronutrients values (Nutritics V5, Nutritics Ltd, Swords, Co. Dublin, Ireland). Parents of the participants will be fully informed regarding the aim of the study, however, the participants will be informed that the study's aim is to measure the effect of two supplemental treatments on physical performance outcomes. The intention of this deception blinding is to make participants unaware of the true aim of the study. A run-in period one week prior to formal trial initiation is considered to determine tolerability and assess potential compliance with study regimens (Shamseer et al., 2015), although no major risks are foreseen. Importantly, the adoption of replicate crossover trial design allows to maximize resources by minimizing participant recruitment requirements of parallel-arm trials requiring a larger pool of participants (Nikles et al., 2011). In a replicate crossover trial, each participant represents a distinct trial and will act as his own control (Senn 2019). Aggregation of each trial will be relevant to derive population- and individual-based estimates regarding supplementation effectiveness (Zucker et al., 1997).

Description of the drug/device/vaccine/dietary intake that is being tested or in social sciences for example providing training or information to groups of individuals: The supplement treatment will involve administering a nutritional blend of natural ingredients consisting of a mixed juice containing 15000 mg cherry juice concentrate, 220 mg cherry extract, 250 mg cocoa extract, 200 mg of tryptophan, 100 mg of L-5-Hydroxytryptophan, 3000 mg of glycine, 300 mg of magnesium, 200 mg of theanine sought to be manufactured by Science in Sport (https://sport.wetestyoutrust.com/supplement-search/science-sport/rest-juice) and within safety limits for administration in children and adolescents (Barrett et al., 2013; Bruni et al., 2004; Cynober et al., 2016; Elango, 2023; Fernstrom, 2012; Garlick, 2004; Lyon et al., 2011; Merry and Ristow, 2016; Rose et al., 1965). The placebo will involve a 40-g dual-source carbohydrate gel, having a 1:0.8 ratio of maltodextrin and fructose, as an alternative supplementation known for having trivial effects on sleep-related outcome measures (St-Onge et al., 2016). Supplementation may elicit potential gastrointestinal discomfort and presents minimal risks that, if occur, will involve the student-athlete withdrawing from the study. A run-in period one week prior to formal trial initiationmay be considered to determine tolerability and assess potential compliance with study regimens (Shamseer et al., 2015), although no major risks are foreseen. Nevertheless, the length of our protocol washout periods is deemed adequate to mitigate any unforeseeable issue due to potential side effects. Both active supplementation (https://sport.wetestyoutrust.com/supplement-search/science-sport/rest-juice) and placebo (https://sport.wetestyoutrust.com/supplement-search/science-sport/beta-fuel-dual-source-energy-gel-beta-fuel-maltodextrin-fructose) are Informed Sport Tested since commercially available. Informed Sport is the world's leading testing and certification program for brands producing sports and nutritional supplements. Designed for elite sport, it protects athletes from inadvertent doping caused by supplements contaminated with banned substances.

Procedures: Time asleep, defined as 'the actual time that a person spends asleep excluding non- sleep-related activities in bed', will represent the primary outcome measure in this investigation (Reed and Sacco 2016). Secondary outcome measures were ratings of perceived exertion (RPE; au) on the CR-100 scale and a proxy measure of high-intensity activity determined as high-speed running distance in meters (Borg 1998; Di Salvo et al. 2009). Time in bed and wake-after-sleep-onset represented other secondary outcome measures (Reed and Sacco 2016). Study participants will wear an ActiGraph GT9X Link (ActiGraph, Pensacola, FL, USA) activity monitor set for date of birth, stature, body mass, and non-dominant wrist. Devices will track all activities during the surveillance period except involvement in training and official matches only. The ActiGraph GT9X Link monitors (35mm×35mm×100mm, and a weight of 14 g) will be configured as per the manufacturer recommendations, with the Cole-Kripke sleep scoring algorithm used to analyze the data and the Tudor-Locke algorithm (Tudor-Locke et al. 2014) set as default for automated sleep period detection (Cole et al. 1992; Quante et al. 2018). Student-athletes will rate the global intensity (RPE) of all sessions using level-anchored semi-ratio Borg CR-100® scale (Borg 1998). Running distances will be monitored during all training sessions with 10-Hz global positioning system (GPS; Fitogether). In line with procedures outlined by Short et al., 2017, participants will be provided with sleep diary for completion during the selected observational time. The purpose of using a sleep diary intends to gather the following information: time someone went to bed; time someone thought fell asleep; time someone woke up (Short et al., 2017). A run-in period one week prior to formal trial initiation will be considered to determine tolerability and assess potential compliance with study regimens (Shamseer et al., 2015), although no major risks are foreseen. The randomization sequence for the four experimental trials will be conducted using the PLAN procedure available in SAS (Deng \& Graz, 2002). Both the participant and the performance nutritionist administering the supplementation will be blinded to its administration and treatment episode allocation.

Procedures for analysis and Interpretation of Data: The statistical analysis framework involves a four-step approach in line with existing research (Senn et al., 2011; Senn 2016; Goltz et al., 2018; Goltz et al., 2019; Shen et al., 2024) and more recent advances (Senn 2024) for appropriate examination of continuous data from a replicate crossover experiment. Response pairs will be formulated by calculating the placebo-adjusted treatment effect for the first supplementation and placebo pair in each participants' sequence (response 1; supplementation 1 minus placebo 1) along with the second supplementation and placebo pair (response 2; supplementation 2 minus placebo 2). The first and second replicates will be calculated as the difference between the supplementation and placebo trial pre-to-post change scores in two comparisons:

1. The immediate impact of supplementation minus placebo on the first administration night;
2. The immediate impact of supplementation minus placebo for the outcome average over the two administration nights as a summary measure (Senn, 2000; Yang et al., 2025).

First, estimation of Pearson's product moment correlation coefficients between the two replicates of the supplementation and placebo differences for each sleep outcome (Senn 2016) is relevant to explore the consistency of the supplementation effect between the replicate trials, with thresholds of 0.1, 0.3 and 0.5 indicating small, moderate and large coefficients, respectively. Second, a naïve estimate (estimate 1) of the true (placebo-adjusted) individual differences SD will be determined by calculating the difference in SDs of the pre-to-post change between the supplementation and placebo trials (Atkinson and Batterham, 2015; Atkinson et al., 2019). These calculations utilise the appropriate equation for pooling SDs across the two replicates of the supplementation and placebo trials. A positive SDIR indicates greater heterogeneity in the supplementation response compared to any random within-subject variability. Third, separate within-participant linear mixed modelling will conducted in SAS OnDemand for Academics for each outcome to derive the participant-by-trial interaction which models trial, period (trial sequence) and the period-by-trial interaction as fixed effects, whereas participant and the participant-by-trial interaction will be modelled as random effects. Estimate 2 of the true individual differences SD will be derived from the participant-by-trial interaction term. Standard residual diagnostics will be performed to assess the adequacy and stability of the modelled covariance parameter estimates and a sensitivity analysis will be conducted excluding outliers that were > 3 times higher or lower than the sample SD. Fourth, a sample estimate of within-subjects variance will be calculated and converted it to a standard error using appropriate degrees of freedom given the completed cycles to derive per participant replicate-averaged treatment effects (Senn 2024). A random-effects meta-analysis with Hartung-Knapp adjustment (IntHout et al., 2014) will summarise individual-participant replicate-averaged treatment effects and respective sampling errors (Senn 2024) conducted using the metagen() function. The restricted maximum-likelihood estimation method will determine the tau-statistic (τ) value describing the between-participant replicate-averaged treatment effect response variability across the distribution of true supplementation-related effects (Langan et al., 2019; Veroniki et al., 2016) The uncertainty surrounding the point τ-statistic estimate was described using 95% CI derived using the generalised Q-statistic method (Viechtbauer 2007) Weighted raw replicate-averaged treatment effects will be reported as descriptive statistics alongside the respective 95% prediction interval illustrating the range for the distribution of true mean differences expected for 95% of similar trials (Borenstein 2024; IntHout et al., 2016) Meta-analyses will be conducted in R (version 3.6.3, R Foundation for Statistical Computing). Mean differences and correlation coefficients will be reported with their corresponding 95% confidence intervals (CI). Accordingly, the phenomenon of statistical shrinkage of treatment effects toward the sample-level mean by deriving best linear unbiased predictor, or shrunk, estimates using the blup() function following random effects meta-analysis modelling using the metafor package (Senn, 2024). Visual inspection of unadjusted (or naïve) estimates (x-axis) against shrunk estimates (y-axis) informed judgements relevant to interpreting the clinical meaningfulness of the estimated response heterogeneity. A minimal clinical important difference (MCID) of ± 30 minutes will inform interpretations for time asleep and total time in bed variables. These values are in line with proposed thresholds defining the acceptable difference in sleep outcomes between actigraphy and polysomnography devices (de Zambotti et al., 2019; Youngstedt 2003) and are broadly consistent with the expected night-to-night variability in these outcomes (Lolli et al., 2024b).