Angelos Vlhoyinnis,George Aphmis,Gregory C.Bogdnis,Giorgos K.Skks,Eleni Andreou ,Christoforos D.Ginnki,*

Department of Life and Health Sciences,University of Nicosia,Nicosia 1700,Cyprus

b School of Physical Education and Sport Science,National and Kapodistrian University of Athens,Athens 17237,Greece

c Department of Physical Education and Sport Science,University of Thessaly,Trikala 42100,Greece

d School of Sport and Health Sciences,Cardiff Metropolitan University,Cardiff CF5 2YB,UK

Abstract Purpose:This systematic review aimed to describe objective sleep parameters for athletes under different conditions and address potential sleep issues in this specif ic population.Methods:PubMed and Scopus were searched from inception to April 2019.Included studies measured sleep only via objective evaluation tools such as polysomnography or actigraphy.The modif ied version of the Newcastle－Ottawa Scale was used for the quality assessment of the studies.Results:Eighty-one studies were included,of which 56 were classif ied as medium quality,5 studies as low quality,and 20 studies as high quality.A total of 1830 athletes were monitored over 18,958 nights.Average values for sleep-related parameters were calculated for all athletes according to sex,age,athletic expertise level,training season,and type of sport.Athletes slept on average 7.2±1.1 h/night(mean±SD),with 86.3%±6.8%sleep eff iciency(SE).In all datasets,the athletes’mean total sleep time was ＜8 h.SE was low for young athletes(80.3%±8.8%).Reduced SE was attributed to high wake after sleep onset rather than sleep onset latency.During heavy training periods,sleep duration and SE were on average 36 min and 0.8%less compared to pre-season and 42 min and 3.0%less compared to in-season training periods,respectively.Conclusion:Athletes’sleep duration was found to be short with low SE,in comparison to the general consensus for non-athlete healthy adults.Notable sleep issues were revealed in young athletes.Sleep quality and architecture tend to change across different training periods.

Keywords:Actigraphy;Polysomnography;Sleep architecture;Sleep duration;Sleep quality

1.Introduction

Growing evidence suggests that more than half of active athletes report poor sleep during various occasions.1This fact bears high signif icance not only due to the detrimental effect of poor sleep on physical and mental performance,but also because of its impact on athletes’ health,as sleep is related to vital functions and may be associated with various pathological conditions and health problems.2－4

Sleep is a multidimensional construct that can be measured by both objective and subjective tools;thus,sleep evaluation methods may be crucial when attempting to draw safe conclusions about its impact on athletic performance.5,6Brief ly,sleep can be subjectively assessed using specif ic questionnaires and sleep diaries,whereas validated objective measurements of sleep include polysomnography(PSG)and actigraphic devices.7

An objective and detailed sleep measurement should assess the progress from wakefulness to sleep and its intermediate stages.Sleep evolves as follows:wakefulness,light sleep,deep sleep,and rapid eye movement (REM) sleep.These different states of brain activity compose the“sleep architecture”.According to the American Academy of Sleep Medicine(AASM),8there are 4 distinct sleep stages:three are non-REM(N1,N2,and N3);and the last is the REM sleep stage.Usually,the first 2 non-REM stages (N1 and N2) are denominated as“light sleep” and the third non-REM stage as “deep sleep” or slow wave sleep.

To measure these variations in brain function,sleep electroencephalography(EEG)or PSG is required.Specif ically,a typical PSG study includes central,frontal,and occipital EEG,recording of eye movements,chin,and muscle activity,electromyography,electrocardiogram,pulse oximetry,respiratory effort,nasal and oral airf low,and body position sensors.8PSG is considered to be the gold standard method for assessing sleep and sleep disorders despite its associated practical diff iculties.7Another critical measure for sleep is its duration,which is often referred as total sleep time(TST),while the percentage of duration of TST relative to the total time in bed is referred as sleep eff iciency (SE).An additional way to estimate TST and SE is via actigraphy,which translates activity/inactivity to wake/sleep cycles,9and has been adequately validated against PSG in athletes.10Thus,actigraphy and PSG effectively measure sleep-related parameters and they are objective tools of sleep assessment.In addition,subjective methods for assessing sleep duration and quality are very popular and include various self-reported sleep logs and questionnaires.However,their validity is unclear,since it has been shown that many athletes tend to overestimate their self-reported sleep.5,6Subjective sleep monitoring tools may be used as initial screening tools for sleep disorders in athletes,such as obstructive sleep apnea.11However,PSG examination should be performed to diagnose relevant disorders,12and it is proposed that it could be potentially used as an indicator of poor physical recovery.13

Recent reviews that have systematically examined the prevalence of sleep disorders in athletes,as well as the effect of training or competition on sleep,have included studies that both objectively and subjectively measured sleep data.1,14These reviews showed that athletes may experience several sleep issues1,14and differentiations in sleep characteristics in relation to training,15,16competition,17or other related factors,such as chronotype16and sports discipline.18,19To address these issues regarding athletes’ sleep,the research community tested acute sleep hygiene strategies in order to improve athletes’ sleep.20Nevertheless,sleep guidelines have thus far not differentiated between athletes and non-athletes,21making it diff icult to design targeted sleep optimizing interventions.Thus,there is a need to gather more information about athletes’ sleep and to develop specif ic guidelines for this population.22

Based on our literature review,quantitative information about sleep quality,sleep quantity,and architecture in athletes,using only objectively measured sleep-related data,is poorly systematized.Thus,the current review aimed to extract from the literature typical sleep characteristics of athletes according to sex,age groups,level of competence,or type of sport and summarize these data.Specif ically,this review aims to:

· Provide data about objectively measured sleep quantity,quality,and architecture of athletes according to their (1)sex,(2) age,(3) athletic expertise,(4) training season,and(5) type of sport (primarily aerobic,anaerobic,or mixed sport,and individual sports vs.team sports).

·Address potential sleep issues and sleep fluctuations in athletes in relation to these categories.

2.Methods

2.1.Information sources and search strategy

This review was performed in accordance to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses(PRISMA) guidelines.23Two electronic databases (PubMed and Scopus) were systematically searched by two of the authors (AV and CDG) for studies published from inception to April 2,2019.For both databases,search terms included the following combination of keywords:(“athlete” OR “competitive” OR “sport” OR“elite”OR“player”)AND(“sleep”OR“electroencephalography”OR “actigraphy” OR “polysomnography”).Retrieved records from PubMed search were limited according to article type(clinical trial,controlled trial,randomized control trial,observational study,comparative studies,and case reports) and language(English).Retrieved records from the Scopus search were limited according to source type (journal),document type (article),and language (English).A supplementary search for relevant studies was conducted from the reference list of the screened manuscripts.

2.2.Eligibility criteria

The eligibility of retrieved records for inclusion in our review was determined by two of the authors (AV and CDG)according to the following criteria:studies were written in English and measured sleep-related parameters objectively via PSG,EEG,or actigraphy,or a combination of these 3 methods.No restrictions in study design were applied.Articles were excluded if (1) participants did not take part systematically in any sport or had a history of major diseases,(2)athletic expertise or sleep-related data were inadequately described,and (3)sleep-related data were assessed after travel or any other manipulated intervention (e.g.,dietary intervention,altitude,or light exposure);only the pre-intervention baseline sleep data from these studies were included,where available.

2.3.Study selection and data extraction

The process used for the literature research is depicted in Fig.1.After removal of duplicated records,titles (and,where required,the abstracts of the remaining studies)were evaluated to select potentially relevant studies for full-text article screening.Reasons for excluding a shortlisted article during full-text screening were recorded(Fig.1).

A standardized data extraction form was employed to obtain the following:(1) publication details—name of first author and year of publication;(2) participant characteristics—sample size and sex;(3) sports-related data—type of sport and training season;(4) sleep-related data—nights recorded,TST,SE,wake after sleep onset (WASO),sleep onset latency (SOL),REM onset latency,and sleep stages(REM,N1,N2,and N3).If data were graphically presented,Plot Digitizer (Version 2.6.8;SourceForge,San Diego,CA,USA) was used to extract data of interest.Where possible,data were extracted in as many sets as possible,separating datasets for age,sex,or other sports-related parameters.The mean and SD for all data were recorded.Where required,SD was calculated from reported conf idence intervals.The assumption that all groups of data were merely sub-samples of a single group was made.Thus,mean values and SDs were combined according to nights recorded.

Fig.1.The Preferred Reporting Items for Systematic Reviews and Meta-Analyses(PRISMA)f low chart for study selection.

2.4.Evidencequality appraisal

All studies were assessed for their methodological quality using a version of the Newcastle－Ottawa Scale(NOS)for cross-sectional studies(Supplementary Table 1).This is an established tool,adapted by Gupta and colleagues,1and used in other relevant studies.14The NOS evaluates evidence through 8 items measuring selection,comparability and outcome criteria.These criteria were modif ied by 2 reviewers(AV and CDG)to represent adequately the aimsof thisreview.When a criterion was satisf ied,a scoreof 1(or in caseof Criterion 7,a score of 2)was awarded to the corresponding item.Otherwise,a score of 0 was awarded the corresponding item.The sum of the scoreswas used to categorize overall study evidence quality as low(＜5),moderate(5－7),or high(＞7).

2.5.Def initions of sleep-related parameters and terminology

The extracted sleep parameters are def ined in Table 1.These parameters were selected for the integrative view of sleep quantity (TST),quality (WASO,SE,and SOL),and architecture(REM onset latency,REM,N1,N2,and N3).

All TSTvaluesweretransformed into hoursfor purposesof uniformity.Data in regards to sleep stages were presented in min,as measured in the majority of the studies in literature.In case of merged N1 and N2 stages of sleep(as light sleep),the results are presented but not included in the calculation of the general or subgroup mean.In the case of sleep scoring with 5 stages(REM,N1,N2,N3,and N4),the N3 and N4 stages were combined and noted accordingly.8

2.6.Sample characteristics assessment and organization

Extracted datasets were categorized according to(1)sex—males and females;(2)age—available data were categorized into 3 groups according to the participants’age:(i)children and adolescents,6－17 years,(ii)young adults,18－24 years,and (iii)middle-aged adults,25－40 years;(3)sports parameters—datasets were classif ied according to(i)sports reliance on different typesof energy production pathway(primary aerobic,anaerobic,or mixed),(ii)individual or team sports,and(iii)the season of data collection (pre-season/tapering,in-season/competition phase,and heavy training phase),and(4)athletes expertise—all participants were evaluated according to the model for classifying the validity of expert samplesin sport psychology24to identify their level of competence(Supplementary Table 2).This tool evaluates athletic expertise by within-sports(level of performance,success,and experience)and between-sports(competitiveness of the sport in the country and in the world)criteria.Due to limited data regarding athletes’expertise and success,a modif ied version was used that had been previously employed in other studies.1,14The calculated score from the modif ied equation Eq.(1) was used to categorize athletes aselite(score of ≥8)or semi-elite(score of ＜8):1,14

Table 1Def inition of sleep-related terms used in data extraction.

where A=standard of performance,D=competitiveness in country,and E=global competitivenessof sport.

3.Results

3.1.Studies selection

During the literature research,3992 studies were identif ied(Fig.1).After duplicate removal,3683 records were screened by title and,where required,by abstract.A total of 159 full-text articles wereassessed for eligibility.Through full-text screening,96 articles were excluded and 18 additional records were identif ied,resulting in a total of 81 studies that were included in the qualitative synthesis.

3.2.Studies’characteristics

Included studies were published from 1978 to 2019,with 55.6%of the studies being published between 2016 and 2019.The sample included 1830 participants,with 230 females,1040 males,and 560 combined(without clear sex segregation),of which 177 were children and adolescents,1080 were young adults,and 410 were middle-aged adults;the age of 163 participants was not reported.The age of the participants was 22.2±5.5 years(mean ±SD).Overall,12 of the studies evaluated sleep-related parameters using PSG,4 studies used EEG,61 studies used actigraphy,and 4 studies performed PSG or EEGin conjunction with actigraphy,resulting in 206 datasets(PSG/EEG,n=37;actigraphy,n=169).A total of 18,958 recorded nights were extracted,of which 4.2% were obtained from PSG/EEG measurements and 95.8% from actigraphy.A total of 555 nights were recorded in children and adolescents,6298 in young adults,and 2330 in middle-aged adults;9775 nights of the 163 participants of undef ined age were excluded in this specif ic analysis.

3.3.Evidence quality appraisal

According to the NOS,evidence quality of 5 studies was classif ied as low,56 as medium,and 20 as high (Supplementary Table 1).The first NOS evaluation component,sample selection,had a median score of 2(0－4).The second component,comparability of studies,received a median score of 2(0－2),and the third component,outcome measurement,had a median score of 3(2－3).

3.4.Athletes expertise appraisal

Using the modif ied version of the model for classifying the validity of expert samples in sport psychology,the participants were categorized as elite (44 studies;females n=117,males n=716,combined males and females n=470)or as semi-elite(37 studies;females n=113,males n=324,combined males and females n=90) (Supplementary Table 2).Elite athletes were represented in 113 datasets of 13,262 nights recorded and semi-elite athletes in 93 datasets of 5696 nights recorded.

3.5.Overall sleep data(quality,quantity,and architecture)

Analytical sleep-related data obtained from PSG/EEG and actigraphy measures are presented in Tables 2 and 3,respectively.Datasets that included daytime naps,25or assessed sleep during ultra-endurance competition(duration of≥6 h),in single26or multiday races,27－29and individual26,28or team ultra-endurance events,27,29are presented separately and only their baseline data were used for the analysis.Overall results are showed in Table 3.Athletes slept approximately 7.2±1.1 h/night.SE was 86.3% ± 6.8%,with mean SOL time of 14.8 ± 17.0 min and WASO of 52.7 ± 32.0 min.According to PSG/EEG measures,REM stage occupied 114.1 ± 35.4 min of sleep,N1 stage occupied 47.9 ± 31.0 min,N2 occupied 240.5±53.6 min,and N3 occupied 91.5±35.8 min(Table 4).

3.6.Sleep quantity and quality according to sex

Discriminating datasets for sex showed that male athletes slept a total of 7.2 ± 1.1 h with 15.7 ± 17.0 min SOL,and female athletes slept 7.5 ± 1.2 h,with 9.7 ± 11.2 min SOL.SE was 85.2% ± 7.2% and 89.1% ± 4.8% for males and females,respectively.Sleep architecture data were available for male athletes only,as none of the included studies reported sleep architecture specif ic to female participants.Therefore,available data show that male athletes’sleep stages comprised REM:111.3 ± 26.0 min,N1:55.8 ± 29.0 min,N2:246.7 ±53.3 min,and N3:82.9±42.0 min(Table 4).

3.7.Sleep quantity and quality according to athletic expertise

Classif ication based on athletes’ level showed no differences in TST between elite and semi-elite participants.SOL was lower in semi-elite athletes than in elite athletes (9.5 ±10.5 min vs.18.4 ± 19.4 min).SE was 1.5% less in elite athletes compared with semi-elite athletes.It was found that elite athletes’ sleep was composed of less N1 and N2 sleep stages and more of N3 and REM sleep stages compared with semielite athletes’ sleep.In comparing SE between male and female elite athletes,it was found that SE was 4.6% less in males than in females.Sleep architecture data for elite athletes were very limited in the literature(N1 and N2 are presented as data from one study).30To the best of our knowledge,there were no discrete data according to sleep stages for male or female elite athletes and for semi-elite female athletes(Table 4).

3.8.Sleep quantity and quality according to age

As shown in Table 4,TST varied signif icantly throughout age groups.Children and adolescent athletes slept on average 60 min less than the young adults group,and 36 min less than middle-aged adults.No signif icant difference in SOL was found among age groups.SE was 5.6%and 6.6%less in children and adolescents,compared to young and middle-aged adults,respectively.The same trend was present for WASO.According to sleep architecture,data were lacking for middle-aged adults.REM sleep stage was 15% less in children/adolescents than in young adults,whilst there were no available data for middle-aged adults.Athletic expertise classif ication between age groups showed that SOL was longer in each elite age group,as compared to the semi-elite corresponding group.

3.9.Sleep quantity and quality in relation to training season

For both sleep quality and sleep quantity,moderate fluctuations were present among training seasons.TST was less during heavy training phases than in pre-season/tapering phases or competition phases (6.7 h vs.7.3 h vs.7.4 h,respectively).SE was found to be less during heavy training phases and greater during in-season/competition phases (84.5% vs.87.5%).According to sleep architecture,even if data are elusive,N3 sleep during the pre-season/tapering phase was reported to be longer than during the in-season/competition and heavy training phases (130.0 min vs.97.0 min vs.74.6 min,respectively) (Table 4).

3.10.Sleep quantity and quality in relation to sports type

The majority of the included articles studied participants in purely aerobic or mixed sports,whereas only 2 studies9,31included participants in sports that require mostly power/explosiveness.According to this categorization,SOL was longer in power/explosiveness sports compared to other sports,with a large SD (24.2 ± 31.3 min).According to sleep architecture,athletes in anaerobic sports had slightly less REM sleep compared with athletes in aerobic or mixed sports(103.0 min vs.111.8 min vs.107.9 min,respectively).N3“deep sleep”was longer in athletes in mixed sports than in athletes in aerobic sports,and over 2-fold longer than athletes in power/explosiveness sports (139.5 min vs.78.2 min vs.61.0 min,respectively).Comparing individual and team sports,no great differences were present,except for N3 sleep,which was longer in athletes participating in team sports than for athletes participating in individual sports (Table 4).

3.11.Sleep during ultra-endurance competitions

The majority of the studies used actigraphic devices to evaluate sleep after ultra-endurance competitions.During multiday ultra-endurance competitions,mean TST ranged from 7.0± 0.9 h to 7.7± 2.0 h,with SE ranging from 83.0 ±6.0 h to 85.1%±5.9%.In case of extreme ultra-endurance team-sport races such as roparun,athletes obtained a mean of 2.2±0.3 h of TST with 83.0% ± 6.0% SE.According to sleep architecture,no longitudinal EEG data were available for multi-day ultra-endurance events.Nevertheless,it was shown that immediately after an ultra-triathlon race,REM sleep decreased,wakefulness increased,and slow wave sleep,corresponding to N3 sleep stage,was not different compared to a no-exercise scenario.

4.Discussion

The current review aimed to summarize objectively measured sleep data from a wide range of athletic populations in order to (1) develop an integrative perspective of athletes’sleep,(2)identify potential sleep issues or related risk factors,and (3) reveal possible gaps in current knowledge.Since norms and recommendations for athletes’ sleep have not been established yet,this review points toward a general depiction of athletes’ sleep.The current review reveals that sleep quantity of athletes is reduced and potentially insuff icient according to the general consensus of the AASM for non-athlete healthy adults.32Thus,overall sleep quality appears inadequate,which highlights the need for sleep-optimizing interventions,especially for children and adolescent athletes,for which specif ic guidelines are not available.Furthermore,periods of heavy training load were identif ied as being more sensitive to potential alterations in sleep quality,while sleep architecture may be modif ied during pre-season.

4.1.Sleep quantity

Evidence suggests that sleep duration in athletes is limited to 7.2 h/night,across all investigated categories.All studies reported TST mean values shorter than 8 h/night,which is considered to be at the lowest values of the cut-off point of 7 h according to the guidelines of AASM for healthy adults.32Notably,the shortest mean TST was 6.3 h,which was reported in children and adolescents,33in contrast with the current AASM guidelines for the pediatric population,which suggest 8－12 h/night.21In early adulthood,average sleep duration was marginally within the suggested ＞7 h of sleep per night.32In contrast,in middle-aged athletes,TST was below this recommended cut-off.However,these suggested limits for a healthy population may underestimate sleep needs in athletes,since the recovery demand in sports is increased.34This fact is supported by studies that found improvements in sports performance when sleep was extended up to approximately 2 h.3,4

The prevalence of insuff icient sleep duration in elite athletes is well documented in the literature.35In most cases athletes appear to obtain less than 8 h of sleep per night.36This issue is also highlighted in young athletes.33,37Student-athletes were found to sleep less that their non-athletic counterpart population.38Similarly,adolescent athletes had less sleep than senior athletes.39However,these studies included mainly athletes of Asian origin;hence,these results could have been inf luenced by the habits of this specif ic ethnic group.

Several factors,such as training or competition,have been shown to consistently correlate with athletes’sleep quantity.14An athlete’s training schedule impairs sleep duration because it reduces the time available for sleep.40,41For young athletes,their daily schedule,including school attendance,may be a key factor in reducing sleep quantity.Furthermore,reduced sleep duration is often observed during training camps,42,43where not only training schedules,but also training volume and intensity are increased.It has been previously shown that sleep duration is impaired in endurance athletes who experienced overreaching due to high training volume28,44and that high-intensity training requires suff icient sleep time since recovery demands are elevated.45For example,during multiple-day ultra-endurance competitions,such as roparun(a nonstop relay run between Rotterdam and Paris),athletes sleep is considerably reduced to 2.2 h.29

Because sleep is a key restorative factor for daytimeinduced fatigue,insuff icient sleep duration may be a risk factor for several hormonal imbalances that suppress muscle growth46and may result in unfavorable body composition.47Inadequate sleep quantity in athletes extends its consequences to sports performance.In contrast,benef its derived from adequate sleep duration translated into higher team rankings in a netball tournament.48Overall,variation in sleep quantity affects mostly psychomotor vigilance.49Because adolescent athletes tend to sleep more over the weekends to counterbalance their weekday sleep debt,their psychomotor vigilance time was found to be shorter on Monday than on Tuesday or Friday.50Additionally,in a study by Choi and colleagues,33longer sleep duration correlated with improved shuttle bouncing performance in junior badminton players.

4.2.Sleep quality

A key finding of the present review is that reduced sleep quality is reported among athletes.In the reviewed athletic sample,children and adolescents,independently of athletic expertise,exhibited a notably low SE (80.3%).However,SE alone cannot recognize short episodes of wakefulness;therefore,examination of SOL and WASO is important.In the present review,no signif icant differences were reported for SOL,but WASO was longer among children and adolescents (74.0 min).This fact may be indicative of long or frequent wake bouts during nocturnal sleep.Furthermore,low SE and elevated WASO was more noticeable in pre-season training compared with in-season and off-season(Table 4).

Overall,there seems to be an interrelation among several sleep quality parameters such as SOL,WASO,and SE.9As shown in elite young athletes/rowers,falling asleep quickly(reduced SOL) was related to more restful sleep and fewer awakenings.37However,overall SE was shown to be lower in elite athletes compared to the age-and sex-matched non-athletic population.9

As with sleep quantity,sleep quality is affected by training and competition.51A short period of intensif ied training results in progressive decrements in sleep quality among cyclists and triathletes.44,52Extending this period—for example,during training camps42or during a cycling grand tour28—may also impair SE.Overall,Wall and colleagues53suggested that SE may be an objective predictor of overreaching in swimmers.

In regards to other health issues,low SE values are indicative of the development or establishment of health problems such as increases in systolic blood pressure2or lower cognitive function.54Also,Choi and colleagues33showed that higher SE was related with better shuttle bouncing performance in young badminton players.

4.3.Sleep architecture

The data we reviewed on sleep architecture showed that athletes of both sexes spend approximately 114.1 min of their sleep time in the REM stage,47.9 min in N1 stage,240.5 min in N2 stage,and 91.5 min in N3 stage.Converting these sleep stages from minutes to percentages of TST,the following distribution is obtained:23.1% REM,9.7% N1,48.7% N2,and 18.5% N3 sleep stage.In comparison,a typical non-athletic adult spends approximately 25% of sleep time in REM stage,5%in N1 stage,50%in N2 stage,and 20%in N3 stage.8Early studies showed that athletes tended to spend more time in non-REM sleep and less time in REM sleep compared to non-athletic controls.55However,ultra-endurance competition changes athletes’ sleep architecture by increasing REM sleep stage,while N3 sleep stage is not affected.26

A factor that was identif ied as being associated with alterations in sleep architecture in athletes was pre-season training period.It was observed that during the pre-season period the amount of sleep in the N3 sleep stage is extended.However,these data were calculated from a limited number of studies,and generalized assumptions should be made with caution.

Findings based on training volume and sleep architecture show that after endurance races,REM proportion is reduced and N3 stage sleep may increase.26,56Driver and colleagues26found that the REM stage was a more sensitive indicator of exercise-induced stress than N3,since REM changed only after 15.0 km and 42.2 km runs.Sapiro and colleagues56showed that after a 92-km road race,REM proportion was decreased and N3 stage sleep was increased.During pre-season and peak training period,N3 sleep has also been found to increase.57Recent data show that N3 stage sleep increases in response to higher physiological restorative demand in athletes(as indicated by higher resting heart rate and lower heart rate variability).58Trinder et al.31found that,according to the type of sport,N3 was higher in athletes practicing an aerobic sport compared to an anaerobic sport.

4.4.Sleep optimization interventions

Several interventions have been implemented in order to improve athletes’ sleep,such as bright-light exposure,59cold-water immersion,60napping,61nutritional interventions,3and sleep hygiene education.62Some interventions may improve sleep in the short term,but these improvements are not sustained in the long term.62Some factors that were thought to negatively affect sleep,such as the use of electronic devices in the evening,should perhaps be reconsidered before relevant interventions are established.63Thus,awareness should be raised when investigating easily applicable and targeted sleep optimization interventions.

Supplementing nocturnal sleep with daily naps has shown mixed results.In adolescent athletes,daily naps showed varying effects on sports-specif ic performance.For example,20-m run performance improved but no effect was observed in shooting performance.64Furthermore,after travel,napping showed no signif icant benef it in Wingate performance.61On the other hand,Blanchf ield and colleagues65reported benef its in endurance performance after nap supplementation of nocturnal sleep among athletes with a TST of ＜7 h.Similarly,Waterhouse and colleagues66showed that post-lunch napping improved alertness and sprint performance in sleep-deprived subjects.

More promising results were found in studies that tried to prolong nocturnal sleep duration.In basketball players,extension of TST correlated with improvements in athletic performance.4In a recent study,extension of sleep duration induced by a meal containing high glycemic index carbohydrates was related to faster reaction time the following morning.3Notably,sleep quantity and quality were modif ied by administering a post-exercise meal containing high glycemic index carbohydrates,thus underlining the need for easily applicable sleepoptimizing interventions.For example,tart cherry juice was found to be an effective nutrition intervention for improving sleep duration and quality in healthy individuals.67Nutritional interventions should be used with caution,however,as they may result in improved athletic performance but still have a negative effect on sleep.For example,a study by Miller and colleagues68showed that caffeine supplementation lead to improvements in endurance sports performance but impaired sleep duration in athletes.

4.5.Other factors that may affect sleep in athletes

Sleep is a multidimensional construct that alters throughout the human life span according to an individual’s habits,such as electronic device use.69Due to the nature of both their training and competition,athletes are exposed to several additional risk factors that may alter sleep patterns and induce sleep deprivation.For instance,environmental factors that inf luence sleep include traveling across different time zones,70training in heat,60conditions that lead to hypoxia,71,72and psychological issues derived from competition stress.73

Practicing a sport since adolescence seems to alter sleep quantity.It has been shown that sleep onset and wake time differ between athletes and non-athletes,with young athletes going to sleep earlier and waking earlier compared to controls who are training recreationally or are sedentary.74Because training schedules demand that student athletes wake up early in the morning,electronic device usage may contribute considerably to the reduced sleep of adolescent athletes.For example,the use of smartphones after lights out may impair athletes’sleep,69and unrestricted Internet access appears to be related to fewer hours of sleep among young athletes.75On the other hand,some recent studies have shown that electronic device use does not necessarily affect sleep quantity and quality in young adults.6,63

Recently,Fowler and colleagues70showed that professional rugby players who travelled from Australia to the UK needed on average of 5 days to restore their sleep-wake patterns.Waterhouse and colleagues76showed that appropriate itinerary arrangements might reduce some of the negative effects of jet lag.In contrast,one study showed that interstate travel did not affect sleep patterns in athletes on the night before a game.77

Pedlar and colleagues71found that sleep duration was not impaired when training was performed at high altitude and that sleeping in a normobaric hypoxic tent led to impaired duration of stage N1 and N2 sleep but led to no signif icant changes in deep sleep or REM sleep.Sleeping at high altitude showed that REM sleep initially falls acutely but afterwards rises gradually from Day 1 to Day 15.78,79In contrast,deep sleep was shown to decline gradually during exposure to altitude.79Similarly,sleep disturbances as a response to high altitude have also been observed in female middle distance runners and young soccer players of various levels.78,80,81Ambient temperature is an additional environmental parameter that may impair sleep.As shown by Skein and colleagues,605 consecutive days of training in the heat can impair sleep quality.

Another factor that has been investigated for its effect on athletes’ sleep is chronotype.It is proposed that chronotype not only affects ratings of perceived exertion and fatigue scores and performance,16but it also affects sleep quality following exercise training.82In regard to chronobiology and circadian rhythm,it was recently shown that rest－activity circadian rhythms vary among different sport disciplines,18and this has an impact on sleep onset and wake times.18

4.6.Methodological issues and future research directions

The current study aimed to analyze athletes’sleep intra-variability using only objectively measured sleep-related data while excluding extreme environmental factors that could affect sleep.Objectively measured data was used because it is well documented that subjective measures,such as sleep logs,often overestimate athlete sleep duration.5,6Likewise,commercially available wearable devices also show poor accuracy and reliability for sleep tracking.83,84In athletic populations,wearable devices may overestimate TST,providing questionable results.10Nevertheless,wearables are practical,user-friendly,non-invasive devices that,after adequate validation,could be useful for coaches and sports scientists to use in longitudinally monitoring athletes’sleep.

Analyzing only objectively measured data revealed gaps in current methodologies and existing literature.When comparing data across studies,differences in the way each sleep variable was calculated may signif icantly affect the outcomes.For example,in some studies,27,81SE was artif icially higher because its calculation did not include the SOL measure.Regarding sleep-related measures,data were scarce for certain categories of participants,and especially for sleep architecture measures.Sleep stages of female athletes,at both the elite and semi-elite levels,were not presented separately.Sleep architecture data were also understudied for elite male athletes and middle-aged adult athletes.No data were retrieved in the literature search for masters athletes or athletes in late adulthood.

Thus,it is diff icult to measure sleep architecture in elite athletes across all ranges of the human lifespan.Therefore,additional studies are needed to investigate sleep stages and their relations to various physiological measures in athletes.Future studies should also focus on sleep evaluation of female athletes.An important outcome of this review is that athletes,especially children and adolescents,present overall poor sleep quality and quantity.Thus,easily applicable interventions should be investigated to optimize sleep and sleep-related parameters in junior athletes.

5.Conclusion

The current systematic review showed that athletes slept on average 7.2 h/night with 86.3%SE.Notably,athletes in junior categories demonstrated low SE (80.3%),with no signif icant increases in SOL,but with long WASO.Several exercise-and sports-related parameters that can potentially inf luence sleep quantity and architecture were analyzed,such as athletic expertise,training season,and sports type.Data were lacking for certain age-and sex-specif ic categories,such as females and masters athletes.Therefore,more studies should be conducted to evaluate sleep architecture among these categories of athletes.Further research is needed to establish sleep recommendations for athletes and promote sleep-optimizing interventions,especially for children and adolescent athletes.

Authors’contributions

AV performed the literature review and data extraction and wrote the first draft of the manuscript;CDG performed the literature review and data extraction and wrote and revised the first draft of the manuscript;GA,GCB,GKS,and EA revised the manuscript.All authors have read and approved the final version of the manuscript,and agree with the order of presentation of the authors.

Competing interests

The authors declare that they have no competing interests.

Supplementary materials

Supplementary material associated with this article can be found in the online version at doi:10.1016/j.jshs.2020.03.006.