Accepted Manuscript

Impact of a Modest Delay in School Start Time in Hong Kong School Adolescents

Ngan Yin Chan, MPhil, Jihui Zhang, MD PhD, Mandy Wai Man Yu, MPH, Siu Ping
Lam, FHKAM (Psych), Shirley Xin Li, PhD, DClinPsy, Alice Pik Shan Kong, FRCP,
Albert Martin Li, FHKAM (Paeds), MD, Yun Kwok Wing, FRCPsych, Professor,
Director of Sleep Assessment Unit

PII: S1389-9457(16)30241-6
DOI: 10.1016/j.sleep.2016.09.018
Reference: SLEEP 3211

To appear in: Sleep Medicine

Received Date: 22 April 2016

Revised Date: 27 September 2016
Accepted Date: 29 September 2016

Please cite this article as: Chan NY, Zhang J, Yu MWM, Lam SP, Li SX, Kong APS, Li AM, Wing YK,
Impact of a Modest Delay in School Start Time in Hong Kong School Adolescents, Sleep Medicine
(2016), doi: 10.1016/j.sleep.2016.09.018.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT
Impact of a Modest Delay in School Start Time in Hong Kong School Adolescents

Ngan Yin Chana, MPhil; Jihui Zhanga, MD PhD; Mandy Wai Man Yua, MPH; Siu Ping Lama,

FHKAM (Psych), Shirley Xin Lib PhD, DClinPsy; Alice Pik Shan Kongc, FRCP; Albert

Martin Lid, FHKAM (Paeds), MD; Yun Kwok Winga, FRCPsych.

Affiliations:

PT
a
Department of Psychiatry, Faculty of Medicine, The Chinese University of Hong Kong,

RI
Shatin, Hong Kong SAR;
b
Department of Psychology, The University of Hong Kong, Pokfulam, Hong Kong SAR;

SC
c
Department of Medicine and Therapeutics, Faculty of Medicine, The Chinese University of

Hong, Shatin, Hong Kong SAR.;

d

U
Department of Pediatrics, Faculty of Medicine, The Chinese University of Hong Kong,
AN
Shatin, Hong Kong SAR.
M

Address correspondence to: Yun Kwok Wing, Professor, Director of Sleep Assessment
D

Unit, Department of Psychiatry, The Chinese University of Hong Kong, Shatin Hospital,
TE

Shatin, Hong Kong SAR; Tel: 852 26367748; Fax: 852 26475321; Email:

ykwing@cuhk.edu.hk
EP

Short title: Modest delay in school start time

C
AC

Word Counts: Text: 3725; Abstract: 255; Tables:3; Figures:1; Supplementary tables:3

1
 ACCEPTED MANUSCRIPT
ABSTRACT

Objectives: To examine the effect of a modest delay (15 minutes) in school start time (SST)

on adolescent sleep patterns, mood, and behaviors.

Methods: Two secondary schools in Hong Kong with a total of 1173 students (intervention: n

= 617; comparison school n = 556) completed both baseline and follow up questionnaires.

PT
School start time was delayed by 15 minutes from 7:45am to 8:00am in the intervention

RI
school. The comparison school maintained their regular SST at 7:55am. Students’ sleep-wake

patterns, daytime sleepiness, mental and behavioral aspects were assessed by validated

SC
questionnaires before and after the intervention.

Results: Students in intervention school significantly delayed their weekday wakeup time (P

U
< 0.001) and increased their total time in bed (P < 0.001) when compared to the comparison
AN
school. Both groups experienced a delay in their weekday bedtime. The students in the
M

intervention school showed improved mental health [General Health Questionnaire (GHQ)

score, P = .015], better prosocial behaviors (P = .009,), better peer relationship (P < 0.001,
D

more attentive (P < 0.001), less emotional problems (P = .002) and behavioral difficulties (P
TE

< 0.001) as measured by Strengths and Difficulties Questionnaire (SDQ).

Conclusions: A modest delay (15 minutes) of the school start time can increase adolescent
EP

sleep with corresponding improvement in mood and behaviors. Current findings have
C

significant implications for the education policy, suggesting that school administrators and
AC

policy makers should systematically consider delaying school start time to promote sleep and

health among school-aged adolescents.

Clinical Trial registration: ChiCTR-TRC-12002798. The trial protocol can be accessed at:

http://www.chictr.org/en/proj/show.aspx?proj=3955

Keywords: school start time; comparison study; sleep patterns; adolescents

2
 ACCEPTED MANUSCRIPT
1. Introduction

Sleep deprivation in adolescents is becoming a concerning health issue [1, 2] with numerous

studies consistently indicating that the suggested amount (8-10hrs)[3] of sleep is rarely

followed by adolescents[4-9]. Instead, they often compensate their accumulated sleep debt

PT
during weekends[5, 6, 9], resulting in a highly variable sleep-wake pattern. In addition, data

from Australia, Japan and Iceland collectively reported that children and adolescents are

RI
sleeping less than they used to be [2, 10-13]. However, a recent review including data from

SC
20 countries from 1905 to 2008 suggested that the secular change of sleep duration varied

across different regions, with some showing a decrease, and others showing an increase of

U
sleep duration[2]. Nonetheless, compared with those in Western countries, Asian adolescents
AN
are observed to obtain nearly an hour less daily sleep[2, 7, 8, 14]. This difference may be

attributed to the higher academic expectation and pressure in Asian culture[8, 15]. Chronic
M

sleep deprivation has been found to be related to a constellation of adverse consequences

D

including daytime sleepiness, mood disturbances, behavioral difficulties, accidents, poor

TE

academic performance, and elevated cardio-metabolic risk [6, 9, 16-22].

Previous studies found that there was a complex interactive web of biological, social, school,
EP

family and individual factors contributing to the epidemic of sleep deprivation[15, 23].

Among these factors, school start time (SST) has been suggested as one of the major
C

determinants[15, 23, 24]. In addition, coupling with increasing social and academic demands,
AC

the biologically driven delay in circadian rhythm that favors phase delay among pubertal

adolescents may contribute to a further delay of their bedtime [22, 25, 26]. This natural

tendency of delayed bedtime, however, is in direct conflict with the early SST, resulting in a

significant loss of sleep during school days[15, 27, 28]. A number of observational and cross-

sectional studies revealed that students with early SST reported increased daytime sleepiness,

poor school attendance, and a greater variability of sleep patterns compared to students with

3
 ACCEPTED MANUSCRIPT
later SST [28, 29]. With regard to this, few studies have attempted to delay SST to examine

its effect on adolescent’s sleep and well-being. These studies, in general, reported longer

sleep time, less daytime sleepiness, better behavior and school performance in students

following the intervention[30-33]. Nonetheless, these studies varied markedly in their study

design, sample size, measures and intervention duration, with a lack of control or comparison

PT
group being one of the major limiting factors [30, 32]. Some studies included small sample

RI
size and short intervention duration, for example, for only two weeks or even one day[34-36].

In addition, it is unclear whether later SST would affect adolescents’ sleep quality and

SC
lifestyle practice as these items were seldom measured[30]. On the other hand, the

implementation of delaying SST often encountered a series of logistic difficulties (for

U
example, adolescents’ after-school activities, transportation arrangements) which might be
AN
influenced by different cultures and education systems[24]. To our knowledge, only one
M

related study on school children has been conducted in Asia despite the emerging evidences

showing that Asian children and adolescents are among the most sleep deprived ones[2, 7,
D

37]. In light of a paucity of data and the methodological limitations of existing studies, the
TE

impact of delay SST on adolescent’s sleep patterns and outcomes merited further

investigation.
C EP

2. Materials and methods

AC

2.1. Study design

This study was part of our large-scale school-based sleep education programme among

school-aged children and adolescents in Hong Kong in 2012. The study involved 14

secondary schools (comparable to grade 7-12 in US school system)[7]. Despite intensive

negotiation with school administrators, only one school agreed to join this Delay SST project.

One of the secondary schools from the prior school education programme was chosen as

4
 ACCEPTED MANUSCRIPT
comparison school in this study. The comparison school did not receive any intervention

from the sleep education programme. The main reason for choosing this particular school was

that this school has a comparable SST (7:55am) as the intervention school (7:45am) and also

they had similar school time (intervention vs comparison school: 429 min vs 412 min). The

original daily school schedule for the intervention school was 7:45 am - 3:40 pm, while the

PT
comparison school was 7:55 am - 3:55 pm. The baseline and follow-up data for the

RI
comparison school were collected in February and May 2012, respectively, while the data for

the intervention school were collected in September 2012 and March 2013 respectively.

SC
Ethical approval was granted by the Ethics Committee of Institutional Ethics Review

Committee (Trial registration: ChiCTR-TRC-12002798).

U
AN
2.2. Participants
M

The study involved two secondary schools, at which a total of 1686 grade 7 to grade 11

adolescents were invited to participate in the study (Figure 1). Grade 12 students were
D

excluded as they had to sit for the public examinations.

TE

2.3. Procedure

Due to perceived logistic difficulties and stakeholders’ (mainly teachers and parents)
EP

concerns about the dismissal time, the school only agreed to a 15-minute delay. Parents and
C

eligible students signed informed consents and assents respectively. Upon completion of the
AC

baseline assessment, the school adjusted their start time from 7:45 am to 8:00 am and

maintained the same end time (3:40pm) by shortening the lunch period starting from October

2012. The re-evaluation took place five months later (March 2013). All intervention school

students were invited to prospectively record their sleep pattern in a sleep diary for a week at

baseline and follow up (n=560). Because of the limited resources, only a subgroup of

5
 ACCEPTED MANUSCRIPT
participants were invited to wear actigraphy (n=75) to correlate subjectively reported

(questionnaire and sleep diary) and objectively measured sleep variables.

2.4. Measurements

A set of validated questionnaires were used to assess adolescents’ sleep-wake pattern, mood

PT
and behaviors. Sleep-wake patterns, sleep-related symptoms (e.g. insomnia),

RI
sociodemographic data and lifestyle practice were measured by locally validated Hong Kong

Children Sleep Questionnaire (HKCSQ)[15, 38, 39]. Students were asked to indicate how

SC
frequently they smoked, drank alcohol, and consumed tea, coffee, energy drink and beverage.

It was defined as abnormal if adolescents rated these behaviors ≥3 times per week in past one

U
month. Daytime sleepiness was measured by Pediatric Daytime Sleepiness Scale (PDSS)[40],
AN
which consists of eight sleep-related behaviors rated on 5-point Likert scale. Strengths and
M

Difficulties Questionnaire (SDQ) [41] was used to measure adolescents’ behavioral

difficulties. It consists of five sub-scales that covers the following domains: conduct, peer
D

relationship, emotional problems, hyperactivity/attention and prosocial behavior. The total

TE

scores of these sub-scales except prosocial behavior yield a total score of overall behavioral

difficulties. In addition, 12-item General Health Questionnaire (GHQ)[42] was used to

EP

measure adolescents’ mental health.

C

Total time in bed was calculated by the difference between wakeup time and bedtime for
AC

weekday and weekend respectively. For other sleep-related problems, it was considered as

abnormal if adolescents reported experiencing any of the sleep problems ≥ 3 times per week

for the past one month. Overall health was measured by one question “During recent one

month, how’s your health” in a 5-point Likert scale. Adolescents from intervention group

were also invited to complete a 7-day sleep diary and their official records were collected to

assess students’ school attendance. Anonymous feedback was collected from the teachers and

6
 ACCEPTED MANUSCRIPT
students of the intervention school. Teachers were asked to express their opinions regarding

the factors influencing school start time, perceived obstacles and decision making around the

implementation of delaying school start time as well as their perceived benefits and

difficulties during this intervention.

The primary outcome of the study was total time in bed as measured by the self-reported

PT
questionnaires and secondary outcomes included daytime sleepiness, mental health,

RI
behavioral problems, insomnia, daytime functioning and school attendances.

2.5. Statistical Analysis

SC
Descriptive data are presented as mean (standard deviation) for continuous variable and as

percentage for discrete variables. A 2 * 2 repeated analysis of variances (ANOVA) (by time;

U
pre vs post assessment; and by group: intervention vs comparison) was used to evaluate the
AN
effects of delayed SST on various sleep related outcome measures controlling for gender, age,
M

and family status. For discrete data, participants were divided into two groups based on their

baseline response: those without baseline sleep-related symptoms were used to explore
D

incidence rate and those with baseline symptoms were used to explore the persistence rate.
TE

Incidence was defined as new cases or disease over a specific period of time (from baseline to

follow-up in this study) while persistence was defined as cases with specific symptoms at
EP

both baseline and follow-up. Statistically significant level was set at p < 0.05. All analyses
C

were conducted using IMB SPSS Statistics 20.0 for Windows (SPSS Inc, Chicago, IL).
AC

3. RESULTS

3.1. Demographic data

A total of 1377 students (overall response rate: 81.7%; intervention: 729 out of 842 (86.6%);

comparison school: 648 out of 844 (76.8%)) returned parental consents, student assents and

baseline questionnaires. Among these students, 85% of them (n= 1173) completed follow up

assessment. Of those who completed both baseline and follow-up questionnaire, 617 (55.1%

7
 ACCEPTED MANUSCRIPT
boys) were from intervention school and 556 students (38.3% boys) were from comparison

school. Adolescents from intervention school (mean age: 14.8 ± 1.65) were slightly younger

than those from the comparison school (mean age: 15.1 ± 1.54; P < 0.001). Moreover, family

income is also lower in the intervention school (P < 0.001) (Table 1). These factors were

included as covariates in the analysis.

PT
3.2. Sleep-Wake Patterns

RI
Table 2 compares the sleep-wake patterns, daytime sleepiness, behavioral and psychological

functions of students for both intervention and comparison schools before and after

SC
intervention, respectively. There was a significant interaction effect of time (pre- and post-

assessment) on total time in bed during school days (P < .001). The adolescents in

U
intervention school slept longer after Delay SST (pre vs post: 7:24 ± 0:59 vs 7:28 ± 0:59)
AN
while adolescents in comparison school had reduced sleep at follow-up assessment (pre vs

post: 7:21 ± 0:56 vs 7:14 ± 0:59) (F (1, 1169) = 12.16, P < .001; partial η2 = 0.01). The
M

difference between two groups in school day total time in bed was mainly explained by later
D

wakeup time in intervention school (intervention: 6:38 ± 0:24 vs 6:47 ± 0:26 as the wakeup
TE

time of comparison school remained the same (comparison school: 6:41 ± 0:23 vs 6:41 ±

0:26; F (1, 1168) = 45.67; P < 0.001; partial η2 = 0.04). While for the school day bedtime, a
EP

significant main effect (pre-post assessment differences) indicated that both groups
C

experienced a slight delay in their bedtime but there was no interaction effect observed
AC

(intervention: 23:13 ± 0:56 vs 23:19 ± 0:57; comparison school: 23:20 ± 0:55 vs 23:27 ± 0:58;

F (1, 1168) = 0.26, P > 0.05). Significant interaction effects were also observed on weekend

total time in bed (intervention: 9:45 ± 1:32 vs 9:46 ± 1:33; comparison school: 9:59 ± 1:20 vs

9:44 ± 1:28; F(1, 1168) = 7.97; P = .005; partial η2 = 0.007) and wakeup time (intervention:

10:03 ± 1:43 vs 10:06 ± 1:42; comparison school: 10:11 ± 1:26 vs 10:01 ± 1:38; F(1, 1168) =

4.06; P = .04; partial η2 = 0.003) but no significant interaction effect was found in weekend

8
 ACCEPTED MANUSCRIPT
bedtime (intervention: 24:18 ± 1:22 vs 24:20 ± 1:22; comparison school: 24:11 ± 1:05 vs

24:17 ± 1:12; F(1, 1168) = 1.00, P > 0.05). However, weekend oversleep and bedtime delay

did not differ between two groups (all Ps > 0.05).

The correlations among questionnaire, prospective sleep diary and actigraphy on the baseline

sleep-wake patterns were moderate to high (all Ps<0.001, supplementary table 1).

PT
3.3. Mental health, daytime functioning, behaviors and caffeinated intake

RI
The students in intervention school had improved mental health (intervention: 1.81 ± 2.73 vs

1.69 ± 2.79, comparison school: 1.39 ± 2.35 vs 1.63 ± 2.64; F (1, 1150) = 2.89; P = .015;

SC
partial η2 = 0.005), better prosocial (intervention: 6.15 ± 2.11 vs 6.24 ± 2.17, comparison

school: 7.36 ± 1.97 vs 7.10 ± 2.03; F (1, 1160) = 6.90, P = .009; partial η2 = 0.006), emotion

U
(intervention: 3.19 ± 2.33 vs 3.14 ± 2.30, comparison school: 3.03 ± 2.25 vs 3.43 ± 2.33; F (1,
AN
1160) = 9.31; P = .002; partial η2 = 0.008), peer relationship (intervention: 3.29 ± 1.50 vs
M

3.15 ± 1.59, comparison school: 2.46 ± 1.53 vs 2.69 ± 1.53; F (1, 1159) = 14.87; P < 0.001;

partial η2 = 0.013), attention level (intervention: 3.98 ± 1.98 vs 3.85 ± 2.03, comparison
D

school: 3.57 ± 1.96 vs 3.86 ± 1.92; F (1, 1159) = 12.39; P < .001; partial η2 = 0.011) and
TE

overall behavioral score (intervention: 12.91 ± 5.39 vs 12.57 ± 5.36, comparison school:

11.22 ± 4.95 vs 12.29 ± 5.22; F (1, 1160) = 21.54; P < 0.001; partial η2 = 0.018). However,
EP

no significant difference in daytime sleepiness was found between students from the two
C

schools ((intervention: 16.25 ± 5.59 vs 16.38 ± 5.89, comparison school: 16.25 ± 5.59 vs
AC

16.27 ± 5.12; F (1, 1139) = 3.32; P > 0.05).

A significant lower incidence rate of difficulty maintaining sleep (1.0% vs 2.6%, P < 0.05)

and regular tea intake (40.4% vs 59.6%, P < 0.05) was found in the intervention school when

compared with comparison school (Table 3). However, other insomnia symptoms, overall

health and caffeinated intake were similar between the two schools.

9
 ACCEPTED MANUSCRIPT
There were no significant interaction effects on time spent on class, homework, tutorials,

electronics usage, and extracurricular activities at pre and post-assessment between two

schools (P >0.05).

3.4. School attendance in intervention school

There was an improvement of tardiness rate of school attendance in the follow-up month

PT
(0.24 days vs 0.09 days, Z = -3.00, P < .05), but not the absence rate (0.22 days vs 0.26 days,

RI
Z = -1.14, P > .05).

3.5. Teachers and students feedback

SC
Most teaching staff (n=60, 95.2%) appreciated the adjustment of the new SST. They reported

that their students improved concentration (74.1%), reduced dozing off (68.9%) and tardiness

U
rate (69.5%). In addition, 85% of teachers reported their own improvement of the teaching
AN
and daily schedule, while 63% of them reported improvement of their own sleep. Nonetheless,
M

some teachers worried about the repercussion of a later SST such as school dismissal time

and class duration (46.7%), extracurricular activities (36.7%), after-school tutorial (25%),
D

transportation arrangement (26.7%) and parental concern (8.3%). For those who provided
TE

written feedback, they stated that a 15 minutes delay can allow them to have more

preparation time and have a relaxed feeling in the morning. However, some teachers also
EP

mentioned that the implementation of delay intervention caused some additional workload.
C

Majority (75.2%) of students perceived Delay SST as beneficial with increased sleep (48.4%),
AC

improved attention (22.4%) and reduced tardiness (5.2%). A small group of students (4.6%)

reported some negative outcomes on transportation arrangement and shortening of lunch

period.

3.6. Subgroup analysis

As some students from the intervention school have participated in our sleep education

programme in previous academic year, we further investigated the possibility of any carrying

10
 ACCEPTED MANUSCRIPT
over effect of sleep education on their sleep patterns as some may argue that the improvement

observed in the current study might be due to sleep education. We evaluated the sleep-wake

pattern of this particular group of students who had enrolled in both sleep education and

delay school start time intervention. The data showed that adolescents consistently delayed

their weekday bedtime, possibly associated with advancing age (pre-post sleep education

PT
bedtime: 23:04 vs 23:14; pre-post delay intervention bedtime: 23:21 vs 23:27), while their

RI
weekday wakeup time remained almost the same before the implementation of Delay SST

intervention. (pre-post sleep education wakeup time: 6:41 vs 6:40;) Interestingly, the wakeup

SC
time only delayed after the Delay SST intervention(pre-post delay intervention wakeup time:

6:41 vs 6:49).

U
AN
3.7. Additional analysis
M

We ran additional analysis just to explore the effect of Delay SST on the intervention school

only (see the notes in Table 2). In the pre-post comparison, we used the 1-week prospective
D

sleep diary as the outcome measure. The adolescents had later school day bedtime (23:09 ±
TE

0:52 vs 23:15 ± 0:55, p =.003), later wakeup time (6:38 ± 0:29 vs 6:47 ± 0:28, P < 0.001) and

shorter sleep latency (P = .02) after Delay SST, resulting in a nearly significant improvement
EP

in weekday actual sleep duration (P = 0.07) (taking account of the sleep latency and WASO)
C

(supplementary table 2). The sleep pattern as reported in the sleep diary was consistent with
AC

the questionnaire findings. Moreover, more students reported having enough sleep (from

29.3% to 37.3%, P < 0.05), and lesser students reported prolonged sleep latency (≥30mins)

(11.3% to 8.1%, P< 0.05), and difficulty initiating sleep (≥3times/week) (6.3% to 3.5%, P<

0.05) (Supplementary Table 3).

4. Discussion

11
 ACCEPTED MANUSCRIPT
Our results demonstrated that adolescents earned more sleep after a modest delay of SST

from 7:45 to 8:00 am either comparing to a comparison school or self-comparison. The

increase in total time in bed was mainly due to later wakeup time. To best of our knowledge,

the delay of 15 minutes is the shortest delay intervention that has been reported in the existing

literature. Although it is reasonable to question that whether a few minute difference in sleep

PT
could have any clinical significance, our study demonstrated such sleep gain could readily

RI
benefit adolescents’ sleep, mental, behavioral and daytime functioning. In contrast to

Wahlstrom[33] and Owens[30] studies that reported either no change or some advancement

SC
in bedtime, students in our study reported a delay in their bedtime after the adjustment of SST.

There is a possibility that Hong Kong adolescents might perceive later SST as a permission to

U
stay up late for other activities such as homework or electronics usage. However, the similar
AN
delay in bedtime among the students from the comparison school argued that this delay might
M

be possibly related to the “natural circadian delay” associated with advancing age, or/and

academic quarter progress with increasing academic and other social demands[7, 15, 26, 43]
D
TE

Our finding also suggested that Delay SST could improve students’ behaviors, mood status

and school tardiness rate[30]. However, weekend oversleep and daytime sleepiness were not
EP

significantly improved. Although adolescents in intervention school reported longer sleep at

C

follow-up, the amount of extra sleep “earned” was not enough to improve their daytime
AC

sleepiness as the overall total time in bed (intervention: 7:28 ± 0:59 hr) was still far less than

the recommended amount of sleep (8-10 hours)[3]. In addition, we also assessed the impact

on adolescents’ sleep quality and the result indicated a lower incidence rate of insomnia

symptoms in the intervention school. These positive impacts on sleep quality in the

intervention school suggested that adolescents might have less sleep-related worries when

they perceive that they can wake up later in the morning. However, adolescents’ belief

12
 ACCEPTED MANUSCRIPT
towards sleep was not addressed in the current study, further study with the measurement of

adolescents’ sleep-related dysfunctional belief would be needed.

Increasing evidence supports the positive effects of later SST, albeit the ideal SST remains

unclear. Nonetheless, our study has provided further evidence that even a very modest delay

PT
(15 minutes) could lead to significant positive outcomes in an array of sleep, school,

RI
behavioral and emotional profiles. Recent study has reported a dose-response effect by

comparing schools with various delay schedules [37]. In fact, American Academy of

SC
Pediatrics recommended that SST should not be earlier than 8:30 am[23]. Nonetheless,

delaying SST is still perceived as a great challenge for schools and various stakeholders. In

U
our study, only two schools expressed their interest in this intervention, and one school
AN
withdrew due to the difficulty in arranging school bus system. Nonetheless, most teachers
M

and students in the intervention school appreciated the adjustment and perceived that the

number of positive outcomes related to Delay SST outweighed their worries. Intriguingly, our
D

previous attempts of a multi-modal and multi-level sleep education did not improve sleep
TE

duration of school adolescents[7, 43]. In this regard, Delay SST can be a highly cost-effective

measure for improving adolescent sleep and well-being[23]. In United States, one of the
EP

major concerns related to Delay SST is adolescent’s athletic programmes[24]. In contrast,

C

teachers and parents in Hong Kong worried about the potential negative impact of Delay SST
AC

on their adolescents’ academic performance including class duration, and after-school

tutorials. Thus, future studies must take into the accounts of potential cross-cultural

differences in Delay SST intervention.

Study strengths and limitations

13
 ACCEPTED MANUSCRIPT
The intervention school had a relatively adequate sample size with good retention rate and

our study incorporated a comparison school. In addition, the sleep-wake patterns of the

adolescents as reported by questionnaires were further verified by the prospective 1-week

sleep diary. In addition, the positive effect of delay SST was also evident even we just

examined the intervention school data only.

PT
There were several limitations. Firstly, the lack of a valid control school has limited our

RI
conclusion that the positive changes in sleep practice and behaviors are solely attributed to

later school start time alone. Although the pre-post comparison of intervention school alone

SC
revealed that adolescents have better sleep sufficiency and sleep quality, there remained a

possibility that the positive improvement might be due to other factors that have not been

U
measured in the study. However, the data collection period for intervention and comparison
AN
schools was about half year apart, at which there may be a possible seasonality effect on
M

adolescent sleep pattern. However, it seems that seasonal variation in sleep pattern was more

evident in adolescent’s bedtime as their wakeup time was mainly determined by SST[15].
D

Interestingly, adolescents’ schoolday wakeup time remained the same across two assessment
TE

periods in the comparison school, while their schoolday bedtime were delayed in parallel

with intervention school, suggesting that the effect of increasing age and circadian delay on
EP

bedtime might be more apparent than that of the seasonal variations. Secondly only two
C

schools were involved in the current study and the SST between intervention and comparison
AC

schools differed by 10 minutes, raising a concern that two schools may not be fully

comparable. Nonetheless, the similar data on students’ class duration time across the period

of measurement between these two schools supported their comparability. Thirdly, although

there were significant changes observed in the behavioral and mental health aspects, the

magnitude of changes were relatively small. Furthermore, a lack of detailed measurement on

the academic results limited our conclusion on the effect of Delay SST on adolescents’ school

14
 ACCEPTED MANUSCRIPT
performance. In order to address these limitations, further study should incorporate valid

control schools to better evaluate the delay intervention effects on adolescents’ sleep,

behavior and school performance.

5. Conclusion

PT
This study suggested that a modest delay of 15 minutes in school start schedule could result

RI
in a constellation of benefits across sleep, mood, behavior, and school attendance among

adolescents. In order to maximize the possible intervention effect, schools should consider a

SC
longer delay in their SST. However, delaying SST is an ongoing debate in different regions

and countries [23, 24, 31]and the perceived obstacles and logistical considerations in

U
changing SST have prevented school administrators and education authorities from taking
AN
appropriate and necessary action. More studies including risk-benefit analysis of Delay SST
M

across different regions and cultures are needed.

D
TE
C EP
AC

15
 ACCEPTED MANUSCRIPT
Acknowledgement Sections

The authors thank all the principals and teachers of the participating secondary schools for

approving and accommodating the implementation of the program. We also thank all the

parents, teachers, and students who participated in the study.

PT
Funding source: This project was supported by Public Policy Research of University Grants

RI
Committee (Reference number: CUHK4012-PPR-11), Hong Kong SAR, China. The funding

body has no role in conception, design, conduction, interpretation and analysis of the study or

SC
in the approval of the publication.

U
Financial Disclosures: Dr. Wing has received sponsorship from Lundbeck Export A/S,
AN
Servier Hong Kong Ltd, Pfizer company Ltd, and Celki Medical Company. Dr. Kong has
M

received honorarium for consultancy or giving lectures from Astra Zeneca, Pfizer, Sanofi,

Novo-nordisk, Eli-Lilly, Merck Serono and Nestle.

D
TE

Conflicts of interest: none

C EP
AC

16
 ACCEPTED MANUSCRIPT

Table 1. Demographic data between participants of two schools

Intervention Comparison school p-value
No. of students 617 556 NA
School start time 7:45am 7:55am NA
Age 14.8 (1.65) 15.1 (1.54) 0.018*
Gender (male, %) 55.1 38.3 0.000**
Family income 24.5 52.5 0.000**
(≥20,000)

PT
Note: ** p <0.001, * p<0.05
Family income dichotomized at HKD$20,000 as it is the median monthly domestic household income

RI
in Hong Kong according to the data from the Census and Statistics Department, HKSAR

U SC
AN
M
D
TE
C EP
AC

17
 ACCEPTED MANUSCRIPT

Table 2. Sleep-wake patterns, daytime sleepiness, mental and behavioral aspects between two schools
before and after intervention
Delay school start Comparison school p-value Partial
eta2

PT
N=617 N=556 (interaction)
Baseline Follow up Baseline Follow-up
Weekday (n=1173)

RI
Total time in bed 7:24 (0:59) 7:28 (0:59) 7:21 (0:56) 7:14 (0:59) 0.001** 0.01
a
Bedtime 23:13(0:56) 23:19 (0:57) 23:20 (0:55) 23:27 (0:58) 0.61 0.00
Wakeup time 6:38 (0:24) 6:47(0:26) a 6:41 (0:23) 6:41 (0:26) 0.000** 0.04

SC
Weekend
Total time in bed 9:45 (1:32) 9:46 (1:33) 9:59 (1:20) 9:44 (1:28) 0.005** 0.007
Bedtime 24:18 (1:22) 24:20 (1:22) 24:11 (1:05) 24:17 (1:12) 0.23 0.001

U
Wakeup time 10:03 (1:43) 10:06 (1:42) 10:11 (1:26) 10:01 (1:38) 0.044** 0.003
AN
Weekend oversleep 2:20 (1:38) 2:18 (1:38) 2:38 (1:32) 2:29 (1:39) 0.34 0.001
Weekend Bedtime 1:04 (1:04) 1:00 (1:07) 0:50 (0:52) 0:50 (0:56) 0.37 0.001
M

delay

PDSS 16.25 (5.59) 16.38 (5.89) 16.25 (5.59) 16.27 (5.12) 0.055 0.003
D

GHQ 1.81 (2.73) 1.69 (2.79) 1.39 (2.35) 1.63 (2.64) 0.015* 0.005

SDQ prosocial 6.15 (2.11) 6.24 (2.17) 7.36 (1.97) 7.10 (2.03) 0.009* 0.006
TE

behavior
SDQ emotional 3.19 (2.33) 3.14 (2.30) 3.03 (2.25) 3.43 (2.33) 0.002* 0.008
symptoms
EP

SDQ 3.98 (1.98) 3.85 (2.03) 3.57 (1.96) 3.89 (1.92) 0.000** 0.011
hyperactive/inattention
SDQ conduct 2.45 (1.60) 2.42 (1.58) 2.15 (1.43) 2.29 (1.54) 0.14 0.002
problems
C

SDQ peer relationship 3.29(1.50) 3.15(1.59) a 2.46(1.53) 2.69(1.53) 0.000** 0.013

AC

problems
SDQ total difficulties 12.91(5.39) 12.57(5.36) 11.22(4.95) 12.29(5.22) 0.000** 0.018

Note: Data are expressed as mean (SD). Time variables were presented in the format of hh:mm.
** Significant interaction effect between two groups across time (P < 001).
* Significant interaction effect between two groups across time (P <0.05)
Paired sample t-test was performed on intervention group for self-comparison.
a
Significant differences between baseline and follow up for the intervention school only by using paired sample t-test (P < 0.05)
PDSS = Pediatric Daytime Sleepiness Scale; GHQ = General Health Questionnaire; SDQ = Strengths and Difficulties Questionnaire
Adjusted for age, gender and family income in the model, family income is dicto

18
Table 3. Incidence and Persistence rate ACCEPTED
of insomnia MANUSCRIPT
symptoms, daytime functioning, overall health and
lifestyle practice of participants between two schools.
Incidence Persistence
Intervention Compari p- Intervention comparison p-value
son value school
school
DIS 2.3 1.9 0.59 22.2 35.3 0.31
DMS 1.0 2.6 0.04* 26.3 20.0 0.71
EMA 1.7 0.9 0.24 29.0 25.0 0.82
Any insomnia 3.7 4.0 0.78 29.6 41.9 0.22

PT
Morning headache 3.9 2.0 0.06 30.4 33.3 0.85
Tired during day 15.4 11.0 0.56 57.4 50.8 0.27
Sleep latency (≥30 1.8 1.7 0.82 21.4 25.0 0.85

RI
mins)
Overall health 7.9 6.9 0.62 33.3 34.6 0.92
Tea 40.4 59.6 0.04* 29.4 70.6 0.58

SC
Coffee 62.1 37.9 0.28 70.6 29.4 0.90
Energy drink 42.1 57.9 0.40 100 0 0.18
Alcohol 100 0 0.10 0 0 NA
Cigarette 0 0 NA 50.0 50 0.39

U
Beverage 44.1 55.9 0.21 52.0 48.0 0.16
AN
Note: data is presented as %; ≥3times/week defined as abnormal; DIS= Difficulty Initiating Sleep; DMS=Difficulty
Maintaining Sleep; EMA=Early Morning Awakening
*
p<0.05
M
D
TE
C EP
AC

19
 ACCEPTED MANUSCRIPT
References

[1] Owens J, Adolescent Sleep Working Group, Committee on Adolescents. Insufficient sleep in
adolescents and young adults: an update on causes and consequences. Pediatrics 2014;134:e921-32.
[2] Matricciani L, Olds T, Petkov J. In search of lost sleep: secular trends in the sleep time of school-
aged children and adolescents. Sleep Med Rev 2012;16:203-11.
[3] Hirshkowitz M, Whiton K, Albert SM, Alessi C, Bruni, DonCarlo L, et al. National Sleep
Foundation’s sleep time duration recommendations: methodology and results summary. J National
Sleep Foundation 2015;1:40-3.

PT
[4] Gibson ES, Powles AC, Thabane L, O'Brien S, Molnar DS, Trajanovic N, et al. "Sleepiness" is serious
in adolescence: two surveys of 3235 Canadian students. BMC Public Health 2006;6:116.
[5] Liu X, Zhao Z, Jia C, Buysse DJ. Sleep patterns and problems among chinese adolescents.

RI
Pediatrics 2008;121:1165-73.
[6] Kong AP, Wing YK, Choi KC, Li AM, Ko GT, Ma RC, et al. Associations of sleep duration with obesity
and serum lipid profile in children and adolescents. Sleep Med 2011;12:659-65.

SC
[7] Wing YK, Chan NY, Yu MWM, Lam SP, Zhang J, Li SX, et al. A school-based sleep education
program for adolescents: a cluster randomized trial. Pediatrics 2015;135:e635-43.
[8] Yang CK, Kim JK, Patel SR, Lee JH. Age-related changes in sleep/wake patterns among Korean
teenagers. Pediatrics 2005;115:250-6.

U
[9] Wing YK, Li SX, Li AM, Zhang J, Kong AP. The effect of weekend and holiday sleep compensation
on childhood overweight and obesity. Pediatrics 2009;124:e994-e1000.
AN
[10] Dollman J, Ridley K, Olds T, Lowe E. Trends in the duration of school-day sleep among 10- to 15-
year-old South Australians between 1985 and 2004. Acta Paediatr 2007;96:1011-4.
[11] Iglowstein I, Jenni OG, Molinari L, Largo RH. Sleep duration from infancy to adolescence:
M

reference values and generational trends. Pediatrics 2003;111:302-7.

[12] Thorleifsdottir B, Bjornsson JK, Benediktsdottir B, Gislason T, Kristbjarnarson H. Sleep and sleep
habits from childhood to young adulthood over a 10-year period. J Psychosom Res 2002;53:529-37.
D

[13] Matricciani LA, Olds TS, Blunden S, Rigney G, Williams MT. Never enough sleep: a brief history of
sleep recommendations for children. Pediatrics 2012;129:548-56.
[14] Gradisar M, Gardner G, Dohnt H. Recent worldwide sleep patterns and problems during
TE

adolescence: a review and meta-analysis of age, region, and sleep. Sleep Med 2011;12:110-8.
[15] Zhang JH, Li AM, Fok TF, Wing YK. Roles of Parental Sleep/Wake Patterns, Socioeconomic Status,
and Daytime Activities in the Sleep/Wake Patterns of Children. J Pediatrics 2010;156:606-U105.
EP

[16] Fallone G, Owens JA, Deane J. Sleepiness in children and adolescents: clinical implications. Sleep
Med Rev 2002;6:287-306.
[17] Wolfson AR, Carskadon MA. Understanding adolescents' sleep patterns and school performance:
a critical appraisal. Sleep Med Rev 2003;7:491-506.
C

[18] Reid GJ, Hong RY, Wade TJ. The relation between common sleep problems and emotional and
behavioral problems among 2- and 3-year-olds in the context of known risk factors for
AC

psychopathology. J Sleep Res 2009;18:49-59.

[19] Hidalgo MP, Caumo W. Sleep disturbances associated with minor psychiatric disorders in
medical students. Neurol Sci 2002;23:35-9.
[20] Spiegel K, Leproult R, Van Cauter E. Impact of sleep debt on metabolic and endocrine function.
Lancet 1999;354:1435-9.
[21] Au CT, Ho CK, Wing YK, Lam HS, Li AM. Acute and chronic effects of sleep duration on blood
pressure. Pediatrics 2014;133:e64-72.
[22] Carskadon MA, Vieira C, Acebo C. Association between puberty and delayed phase preference.
Sleep 1993;16:258-62.
[23] American Academy of Pediatrics, Adolescent Sleep Working Ggroup, Committee on
Adolescence, Council of Sleep Health. School start times for adolescents. Pediatrics 2014;134:642-9.

20
 ACCEPTED MANUSCRIPT
[24] Owens J, Drobnich D, Baylor A, Lewin D. School Start Time Change: An In-Depth Examination of
School Districts in the United States. Mind Brain Educ 2014;8:182-213.
[25] Carskadon MA. Patterns of sleep and sleepiness in adolescents. Pediatrician 1990;17:5-12.
[26] Roenneberg T, Kuehnle T, Pramstaller PP, Ricken J, Havel M, Guth A, et al. A marker for the end
of adolescence. Curr Biol 2004;14:R1038-9.
[27] Dexter D, Bijwadia J, Schilling D, Applebaugh G. Sleep, sleepiness and school start times: a
preliminary study. WMJ 2003;102:44-6.
[28] Short MA, Gradisar M, Lack LC, Wright HR, Dewald JF, Wolfson AR, et al. A Cross-Cultural
Comparison of Sleep Duration Between US and Australian Adolescents: The Effect of School Start

PT
Time, Parent-Set Bedtimes, and Extracurricular Load. Health Educ Behav 2013;40:323-30.
[29] Wolfson AR, Spaulding NL, Dandrow C, Baroni EM. Middle school start times: the importance of
a good night's sleep for young adolescents. Behav Sleep Med 2007;5:194-209.
[30] Owens JA, Belon K, Moss P. Impact of delaying school start time on adolescent sleep, mood, and

RI
behavior. Arch Pediatr Adolesc Med 2010;164:608-14.
[31] Minges KE, Redeker NS. Delayed school start times and adolescent sleep: A systematic review of
the experimental evidence. Sleep Med Rev 2016;28:86-95.

SC
[32] Danner F, Phillips B. Adolescent sleep, school start times, and teen motor vehicle crashes. J Clin
Sleep Med 2008;4:533-5.
[33] Wahlstrom K. Changing times: findings from the first longitudinal study of later high school start

U
time. NASSP Bulletin 2002;86:3-21.
[34] Lima PF, Medeiros AL, Araujo JF. Sleep-wake pattern of medical students: early versus late class
AN
starting time. Braz J Med Biol Res 2002;35:1373-7.
[35] Lufi D, Tzischinsky O, Hadar S. Delaying school starting time by one hour: some effects on
attention levels in adolescents. J Clin Sleep Med 2011;7:137-43.
[36] Vedaa Ø, Saxvig, I. W., Whilhelmsen-Langeland, A., Bjorvatn, B., & Pallesen, S. School start time,
M

sleepiness and functioning in Norwegian adolescents. Scand J Educ Res 2012;56:55-67.

[37] Li S, Arguelles L, Jiang F, Chen W, Jin X, Yan C, et al. Sleep, school performance, and a school-
based intervention among school-aged children: a sleep series study in China. PLoS One
D

2013;8:e67928.
[38] Li AM, Cheung A, Chan D, Wong E, Ho C, Lau J, et al. Validation of a questionnaire instrument for
TE

prediction of obstructive sleep apnea in Hong Kong Chinese children. Pediatr Pulmonol
2006;41:1153-60.
[39] Zhang J, Li AM, Kong AP, Lai KY, Tang NL, Wing YK. A community-based study of insomnia in
Hong Kong Chinese children: Prevalence, risk factors and familial aggregation. Sleep Med
EP

2009;10:1040-6.
[40] Drake C, Nickel C, Burduvali E, Roth T, Jefferson C, Pietro B. The pediatric daytime sleepiness
scale (PDSS): sleep habits and school outcomes in middle-school children. Sleep 2003;26:455-8.
[41] Lai KY, Luk ES, Leung PW, Wong AS, Law L, Ho K. Validation of the Chinese version of the
C

strengths and difficulties questionnaire in Hong Kong. Soc Psychiatry Psychiatr Epidemiol
AC

2010;45:1179-86.
[42] Chan DW, Chan TS. Reliability, validity and the structure of the General Health Questionnaire in
a Chinese context. Psychol Med 1983;13:363-71.
[43] Chan NY, Lam SP, Zhang J, Yu MWM, Li SX, Li AM, Wing YK. Sleep Education in Hong Kong. Sleep
Biol Rhythm 2016;14:21-5.

21
 ACCEPTED MANUSCRIPT
Figure1. Flow chart of schools and subjects recruitment process

Delay School start time

(2 schools, student n = 1686)

PT
Enrollment

RI
SC
Intervention school Control school
1 school, students (n = 842) 1 school, student (n=844)

U
AN
Baseline
M

Lost baseline data Lost baseline data

Students (n=113, 13.4%) Students (n=196, 23.2%)
Did not return baseline questionnaire Did not return baseline questionnaire
D
TE

Follow up
Lost to follow up Lost to follow up
Students (n=110, 17.8%) Students (n=92, 14.2%)
EP

Did not return follow up questionnaire Did not return follow up questionnaire
C
AC
Highlights
ACCEPTED MANUSCRIPT
a. Sleep deprivation in adolescents is becoming a concerning health issue globally.
b. Emerging data suggest the beneficial effect of delaying school start time.
c. There has been limited study in Asia despite evidence that Asian children and adolescents are
among the most sleep deprived ones.
d. A modest delay of 15 minutes in the school start schedule could result in a constellation of
benefits across sleep, mood, behavior, and school attendance among adolescents. The delay of 15
minutes is the shortest delay intervention that has been reported in the existing literature.
e. School administrators and policy makers should systematically consider delaying school start
time to promote sleep and health among school-aged adolescents.

PT
.

RI
U SC
AN
M
D
TE
C EP
AC