Chronobiology International

The Journal of Biological and Medical Rhythm Research

ISSN: 0742-0528 (Print) 1525-6073 (Online) Journal homepage: http://www.tandfonline.com/loi/icbi20

Relationships among sleep timing, sleep duration

and glycemic control in Type 2 diabetes in
Thailand
Sirimon Reutrakul, Nantaporn Siwasaranond, Hataikarn Nimitphong, Sunee
Saetung, Naricha Chirakalwasan, Boonsong Ongphiphadhanakul, Ammarin
Thakkinstian, Megan M. Hood & Stephanie J. Crowley
To cite this article: Sirimon Reutrakul, Nantaporn Siwasaranond, Hataikarn Nimitphong, Sunee
Saetung, Naricha Chirakalwasan, Boonsong Ongphiphadhanakul, Ammarin Thakkinstian,
Megan M. Hood & Stephanie J. Crowley (2015): Relationships among sleep timing, sleep
duration and glycemic control in Type 2 diabetes in Thailand, Chronobiology International, DOI:
10.3109/07420528.2015.1105812
To link to this article: http://dx.doi.org/10.3109/07420528.2015.1105812

Published online: 23 Nov 2015.

Submit your article to this journal

Article views: 12

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=icbi20
Download by: [University of California, San Diego]

Date: 29 November 2015, At: 17:59

Chronobiology International, Early Online: 18, (2015)

! Taylor & Francis.
ISSN: 0742-0528 print / 1525-6073 online
DOI: 10.3109/07420528.2015.1105812

ORIGINAL ARTICLE

Relationships among sleep timing, sleep duration and glycemic

control in Type 2 diabetes in Thailand

Downloaded by [University of California, San Diego] at 17:59 29 November 2015

Sirimon Reutrakul1, Nantaporn Siwasaranond1, Hataikarn Nimitphong1, Sunee Saetung1, Naricha

Chirakalwasan2,3, Boonsong Ongphiphadhanakul1, Ammarin Thakkinstian4, Megan M. Hood5, and
Stephanie J. Crowley5
1

Section of Endocrinology, Department of Medicine, Faculty of Medicine Ramathibodi Hospital, Mahidol University,
Bangkok, Thailand, 2Division of Pulmonary and Critical Care Medicine, Department of Medicine, Faculty of Medicine,
Chulalongkorn University, Bangkok, Thailand, 3Excellence Center for Sleep Disorders, King Chulalongkorn Memorial
Hospital, Thai Red Cross Society, Bangkok, Thailand, 4Section for Clinical Epidemiology and Biostatistics, Faculty of
Medicine Ramathibodi Hospital, Mahidol University, Bangkok, Thailand, and 5Department of Behavioral Sciences, Rush
University Medical Center, Chicago, IL, USA

There is evidence that the sleep and circadian systems play a role in glucose metabolism. In addition to physiological
factors, sleep is also affected by behavioral, environmental, cultural and social factors. In this study, we examined
whether morning or evening preference, sleep timing and sleep duration are associated with glycemic control in
patients with type 2 diabetes residing in Thailand. Two hundred and ten type 2 diabetes patients who were not shift
workers completed an interview and questionnaires to collect information on diabetes history, habitual sleep duration
and sleep timing. Chronotype, an individuals tendency for being a morning or evening person, was assessed
using the Composite Score of Morningness (CSM), which reflects an individuals subjective preference for activities in
the morning or evening, as well as mid-sleep time on weekend nights (MSF), which reflects their actual sleep
behavior. Most recent hemoglobin A1c (HbA1c) values were retrieved from medical records. Evening preference (as
indicated by lower CSM), later bedtime on weekends, and shorter sleep duration correlated with higher HbA1c
(r 0.18, p 0.01; r 0.17, p 0.01 and r 0.17, p 0.01, respectively), while there was no association between
MSF or wake up time and glycemic control. In addition, later bedtime on weekends significantly correlated with
shorter sleep duration (r 0.34, p50.001). Hierarchical regression analyses adjusting for age, sex, body mass index,
insulin use and diabetes duration revealed that later bedtime on weekends was significantly associated with poorer
glycemic control (B 0.018, p 0.02), while CSM was not. Mediation analysis revealed that this association was fully
mediated by sleep duration. In summary, later bedtime on weekends was associated with shorter sleep duration and
poorer glycemic control in patients with type 2 diabetes. It is likely that patients with later weekend bedtimes curtail
their sleep by waking up earlier. Exploring the potential reasons for this phenomenon (e.g. cultural influences,
metropolitan lifestyle, environmental factors, family and social obligations) specific to a Thai population may help
identify behavioral modifications (i.e. earlier bedtime and/or sleep duration extension) that could possibly lead to
improved glycemic control in this population.
Keywords: Bedtime, chronotype, glycemic control, sleep duration, Type 2 diabetes

INTRODUCTION
and hormone secretions ([Arble et al., 2010; Aschoff
et al., 1975; Huang et al., 2011). Sleep is viewed as a state
of energy conservation and replenishment of energy
stores. This physiologic process is controlled in part by
the circadian timing system and in part by a homeostatic mechanism where the pressure for sleep increases
in proportion to the duration of prior wakefulness.

The sleep and circadian systems play a significant role

in energy metabolism. The circadian system, controlled
by the master circadian clock located in the suprachiasmatic nuclei of the hypothalamus, regulates daily
rhythms of sleep/wake and various metabolic outputs,
such as feeding behavior, peripheral tissue metabolism

Submitted September 9, 2015, Returned for revision September 28, 2015, Accepted October 6, 2015

Correspondence: Sirimon Reutrakul, Section of Endocrinology, Department of Medicine, Faculty of Medicine Ramathibodi
Hospital, 270 Rama VI Rd, Ratchathewi, Bangkok 10400, Thailand. Tel: +6692-265-8554. Fax: +662-201-1715. E-mail:
sreutrak10800@gmail.com

Downloaded by [University of California, San Diego] at 17:59 29 November 2015

S. Reutrakul et al.

Human behaviors and social factors may override these

physiological mechanisms, resulting in alterations of
sleep duration, quality or timing. These alterations of
sleep have been linked to abnormal energy metabolism,
including insulin resistance and type 2 diabetes
(Reutrakul & Van Cauter, 2014).
Circadian misalignment occurs when sleeping and
feeding are at inappropriate timing relative to the
endogenous circadian clock. Night shift work is an
example of severe circadian misalignment, as workers
are awake, active and eating during their circadian night
and trying to sleep and fast during their circadian day. In
controlled laboratory studies, experimentally induced
circadian misalignment in healthy human volunteers
resulted in impaired glucose tolerance (Buxton et al.,
2012; Leproult et al., 2014; Scheer et al., 2009).
Epidemiologic studies have also revealed that shift
work is associated with diabetes diagnosis (Gan et al.,
2015).
Many individuals in modern society experience a
form of mild circadian misalignment, especially during
the work week as they follow social rhythms imposed by
professional obligations, school schedules, family and
other commitments (Wittmann et al., 2006). Those with
more evening preference or later chronotype (later
bedtime and wake time), typically have a greater
degree of misalignment between social rhythms and
the circadian clock (Wittmann et al., 2006). Data are
emerging that this type of mild circadian misalignment
is associated with elevated cardiometabolic risk, including higher body mass index (BMI) (Arora & Taheri, 2015;
Culnan et al., 2013; Olds et al., 2011), meal skipping and
an unhealthy feeding pattern (Sato-Mito et al., 2011),
higher stress hormones and resting heart rate (Lucassen
et al., 2013) and metabolic syndrome (Yu et al., 2015). In
recent large population-based studies, those with evening preference had a significantly increased risk of
having a diabetes diagnosis (OR 1.72.5) (Merikanto
et al., 2013; Yu et al., 2015). In addition, in patients with
type 2 diabetes, later chronotype or more evening
preference has been found to be independently associated with poorer glycemic control (Iwasaki et al., 2013;
Osonoi et al., 2014; Reutrakul et al., 2013).
Factors that could affect chronotype include age
(becoming earlier chronotypes as individuals get older),
sex (males have later chronotypes for most of adulthood) (Roenneberg et al., 2007a), and geographical
location (Roenneberg et al., 2007b). In northern latitudes where day length varies significantly between
seasons, bedtime and wake time showed seasonal
variation while this was not observed in a location
close to the equator (Friborg et al., 2012). In addition,
studies suggest that individuals who reside in more
temperate climates are more morning-oriented than
those in less temperate climates, likely due to the
combination of sun exposure and warmer temperatures
(Horzum et al., 2015; Smith et al., 2002).

Insufficient sleep has been recognized as a risk factor

for diabetes (Cappuccio et al., 2010) and shown to be
associated with poorer glycemic control in patients with
type 2 diabetes (Kim et al., 2013; Ohkuma et al., 2013),
although not consistently (Knutson et al., 2011). We
have previously shown that in type 2 diabetes patients
residing in a northern latitude (Chicago, IL), sleep
duration had no relationship with glycemic control,
and sleep duration was similar across the distribution of
chronotypes (Reutrakul et al., 2013). Whether this
relationship is present in a more tropical climate and
whether it is associated with glycemic control in
diabetes patients is unknown. Therefore, the purpose
of this study is to examine the relationships among
chronotype, sleep duration and glycemic control in type
2 diabetes patients in Thailand.

MATERIALS AND METHODS

Participants
Adults with type 2 diabetes who were being followed in
an endocrinology clinic at Ramathibodi Hospital,
Mahidol University, Bangkok were invited to participate.
Exclusion criteria included shift work, pregnancy,
neurological or physical impairments that required
patients to depend on others for daily activity, such as
being confined to bed or needing assistance with
feeding. All participants gave written informed consent.
The protocol was approved by the Institutional Review
Board, Faculty of Medicine Ramathibodi Hospital,
Mahidol University, Bangkok, Thailand, which meet
the ethical standards of this journal outlined by
Portaluppi et al. (2008).
After obtaining informed consent, weight was measured. Age, height, current medications, and most recent
HbA1c values (within the prior 3 months) were extracted
from patient medical records. BMI was calculated using
the standard formula [weight (kg)/ height (meter)2].
Research personnel interviewed participants about their
work status [categorized into employed (working full
time or part time), or unemployed]. Depressive symptoms were assessed using the Thai version of The Center
for Epidemiologic Studies-Depression (CES-D) Scale
(Radloff, 1977; Trangkasombat et al., 1997). Additional
validated measures as described below were also collected (total assessment time ranged between 45 and
60 min).
Circadian and subjective sleep assessments
Morningnesseveningness preference was assessed
using the validated Thai version of the Composite
Score of Morningness (CSM) (Pornpitakpan, 1998;
Smith et al., 1989). The CSM consists of 13 questions
regarding the preferred time individuals would like to
wake up and go to bed, preferred time for physical and
mental activity and subjective alertness. The total score
ranges from 13 (extreme eveningness) to 55 (extreme
morningness).
Chronobiology International

Sleep timing, sleep duration and glycemic control

Sleep duration

b (-0.016)

a (-0.355)

LnHbA1c

Weekend bedtime
c (0.009)

Downloaded by [University of California, San Diego] at 17:59 29 November 2015

FIGURE 1. In this mediation diagram, lnHbA1c is the outcome

variable, weekend bedtime is the causal variable, and self-reported
sleep duration is the mediator. Unstandardized beta coefficients
(B) are noted for each path.

Participants reported their usual bedtime, wake-up

time, sleep onset latency and actual sleep duration on
weekdays and weekends over the previous month. From
these, we calculated mid-sleep time separately for
weekdays and weekends as the midpoint between
sleep onset (bedtime plus sleep latency) and wake
time. The metric of chronotype, mid-sleep time on free
days (MSF), was derived from mid-sleep time on
weekend nights with further adjustment for the
weighted sleep duration (duration between sleep onset
and wake time) average of weekends and weekdays as
follows: MSF mid-sleep time on free days (weekend)
 0.5*(SDF  (5*SDW + 2*SDF)/7), where SDF is calculated sleep duration on weekend nights and SDw is
calculated sleep duration on weekday nights
(Roenneberg et al., 2004, 2007a). Social jetlag was
calculated based on the absolute difference between
mid-sleep time on weekdays and weekends
(Roenneberg et al., 2012), and categorized as 530 or
30 min.
Average self-reported sleep duration was calculated
as [(sleep duration on weekdays*5) + (actual sleep
duration on weekend*2)]/7. Sleep duration was derived
from the question During the past month, how many
hours of actual sleep did you get at night? Sleep debt
was calculated as the difference between self-reported
sleep duration and the desired sleep duration, from
which a weighted average was computed [(sleep debt on
weekdays*5) + (sleep debt on weekends*2)/7].
To assess sleep quality independently of sleep duration, we utilized the modified Pittsburgh Sleep Quality
Index (PSQI) score (Knutson et al., 2006). PSQI score
evaluates sleep duration and quality within the past
month, with a higher score indicating worse sleep
(Buysse et al., 1989), and has been validated in a Thai
population (Sitasuwan et al., 2014). The modified PSQI
score excludes the sleep duration component from the
PSQI to assess sleep quality only (Knutson et al., 2006).
Participants reported whether they had a diagnosis of
obstructive sleep apnea (OSA). Those without a previous
diagnosis were interviewed using the Berlin questionnaire to assess the risk of OSA, which categorizes
!

Taylor & Francis

respondents as high or low risk of having OSA (Netzer

et al., 1999). The questionnaire was previously validated
in a Thai population (Suksakorn et al., 2014).
Participants who had a diagnosis of or were at high
risk for OSA were grouped together as presence or high
risk of OSA (OSA/Risk).

Statistical analysis
Data are expressed as mean SD or median (interquartile range).HbA1c values were not normally distributed; therefore, the natural logarithm transformation
of HbA1c (lnHbA1c) was used for the analyses.
To determine the factors associated with glycemic
control, Pearson correlations were used to explore the
associations between the lnHbA1c and continuous
demographic, sleep and circadian variables, while
unpaired independent samples t-tests were used to
analyze differences in the lnHbA1c for dichotomous
variables.
To determine whether sleep timing, chronotype or
sleep duration was significantly associated with glycemic control beyond the role of other potential
contributing factors, we used hierarchical regression
modeling. Thus, age, sex (male reference), BMI, insulin
use (yes/no) and diabetes duration were entered in the
first step. Sleep timing, chronotype or sleep duration
was then entered in the second step and change in R2
was obtained to determine whether these variables
explained significant variance in HbA1c beyond the
role of the variables entered in step one.
Finally, mediation analysis for continuous data was
applied to construct mediation pathways, i.e. bedtime
on weekend ! sleep duration ! lnHbA1c, in which
bedtime was the independent variable, sleep duration
was the mediator and lnHbA1c was the outcome of
interest (Imai et al., 2010; MacKinnon et al., 2002, 2007).
The two equations were constructed as follows: the
sleep duration mediator was regressed on bedtime (path
a in Figure 1). Then, lnHbA1c was regressed on sleep
duration and weekend bedtime (path b in Figure 1). Age,
sex, BMI, diabetes duration and insulin use were also
included in both equations. A mediation effect was then
estimated using the product-of-coefficient method
(MacKinnon et al., 2002, 2007). A bootstrap analysis
with 1000 replications was then applied to estimate
average mediation effects without requiring the
assumption of normality (Preacher & Hayes, 2008). For
each bootstrap, the mediation effect was estimated,
averaged across 1000 replications, and its corresponding
95% confidence interval (CI) was determined using a
bias-corrected bootstrap technique. Analyses were performed using STATA 13.0 software. A p-value50.05 was
considered statistically significant.
RESULTS
Baseline demographic, glycemic, circadian and sleep
characteristics of the 210 participants are shown in

S. Reutrakul et al.

Table 1. Participants were 58.6 years, on average and

40% were male. The median HbA1c was 7.2 and 40% of
participants were using insulin. Average MSF was
2:16 A.M. and 26% of participants had social jetlag
30 min. Self-reported average sleep duration was
5.49 h and 39% were at risk of or had a diagnosis of OSA.

Downloaded by [University of California, San Diego] at 17:59 29 November 2015

Associations among HbA1c and demographics,

circadian and sleep parameters
There was no significant association between HbA1c
and age (r 0.097, p 0.16) or CES-D (r 0.02,
p 0.77), sex [median: male 7.4% (IQR 6.68.5) versus
female 7.1% (6.67.9), p 0.08], BMI [r 0.107, p 0.12],
TABLE 1. Baseline demographic, circadian and sleep characteristics (n 210).
Demographic and glycemic data
Age (year)
Male (n) (%)
BMI (kg/m2)
Employed (n) (%)
Insulin use (n) (%)
Diabetes duration (year)
HbA1c
CES-D
Circadian parameters
CSM
MSF*
Weekday bedtime*
Weekday wake time*
Weekend bedtime*
Weekend wake time*
Social jetlag 30 min (n) (%)
Sleep parameters
Average sleep duration (h)
Average sleep debt (h)
Modified PSQI
OSA/risk (n) (%)

58.6 11.0
84 (40.0)
28.4 4.8
106 (50.5)
84 (40.0)
10.0 (3.018.2)
7.2 (6.68.1)
11.7 6.3
44.15 5.46
2:16 1:22
22:02 1:58
5:34 1:17
22:13 1:23
6:05 1:26
56 (26)
5.49 1.52
1.41 1.53
5.85 2.97
83 (39.5%)

*Time presented as 24-h clock time. Data are presented as

mean SD, median (IQR) or n (%).
CES-D, Center for Epidemiologic Studies-Depression Scale; CSM,
Composite Morningness Scale (higher score indicates greater
morningness preference); HbA1c, Hemoglobin A1c; MSF,
mid-sleep time on weekends; PSQI, Pittsburgh Sleep Quality
Index.

employment status [median: working 7.3% (6.68.1)

versus not working 7.1% (6.68.1), p 0.78], having
social jetlag 30 min [30 min 7.2% (6.68.1) versus
530 min 7.2% (6.58.1), p 0.72], or having OSA/risk
[yes 7.2% (6.68.3) versus no 7.2% (6.68.0), p 0.35].
Those with longer diabetes durations had poorer glycemic control (r 0.212, p 0.002), as did those using
insulin [users 7.9% (7.09.1) versus non-users 6.9% (6.4
7.6), p50.001].
Associations among glycemic control and circadian
and sleep parameters are shown in Table 2. Eveningness
preference (lower CSM), later bedtime (weekday and
weekend) and shorter sleep duration were associated
with poorer glycemic control (higher HbA1c). Therefore,
these variables were the primary predictor variables we
tested in the hierarchical regression model.
To determine whether bedtime was independently
associated with HbA1c, a hierarchical multiple regression controlling for age, sex, BMI, diabetes duration and
insulin use was performed (Table 3). Because bedtime
on weekends likely better represents an individuals
sleep time preference than bedtime on weekdays,
weekend bedtime was used in this regression analysis.
Demographic variables, diabetes duration and insulin
use explained 16% of the variance in HbA1c. Bedtime on
weekends was added in the second step and was
significantly associated with HbA1c (unstandardized
coefficient, B 0.018, p 0.021). This model explained
an additional 2.1% of the variance in HbA1c
(DR2 0.021, p 0.021, total adjusted R2 0.18), which
indicated that bedtime on weekends contributed
slightly, but significantly to the variance of HbA1c
above and beyond demographic variables. When similar
analyses were performed using CSM in the second step,
this was not associated with HbA1c (B 0.003,
p 0.11).
To determine if sleep duration was also independently associated with glycemic control, similar hierarchical regression modeling was performed where
demographic variables were entered in the first step
and sleep duration was entered in the second step.
Shorter sleep duration was significantly associated with

TABLE 2. Correlations between glycemic, sleep and circadian variables (r correlation coefficient).

CSM
MSF
Bedtime weekday
Wake time weekday
Bedtime weekend
Wake time weekend
Sleep duration
Sleep debt
Modified PSQI

Ln HbA1c

CSM

MSF

Bedtime
weekday

Wake time
weekday

Bedtime
weekend

Wake time
weekend

Sleep
duration

Sleep
debt

0.177*
0.081
0.169*
0.022
0.169*
0.017
0.168*
0.120
0.097

0.544***
0.503***
0.407***
0.491***
0.446***
0.251***
0.271***
0.243***

0.619***
0.597***
0.763***
0.621***
0.235**
0.214**
0.135*

0.352***
0.884***
0.313***
0.388***
0.194**
0.017

0.305***
0.741***
0.085
0.102
0.094

0.382***
0.343***
0.211**
0.044

0.144*
0.048
0.048

0.552***
0.558***

0.429***

*p  0.05, **p50.01, ***p50.001.

CSM, Composite Morningness Scale (higher score indicates greater morningness preference); HbA1c, Hemoglobin A1c; MSF, mid-sleep
time on weekends; PSQI, Pittsburgh Sleep Quality Index.
Chronobiology International

Sleep timing, sleep duration and glycemic control

Downloaded by [University of California, San Diego] at 17:59 29 November 2015

higher HbA1c (B 0.019, p 0.008), and sleep duration explained an additional 2.8% of the variance in
HbA1c (DR2 0.028, p 0.008, total adjusted R2 0.19).

duration. The outcome (i.e. lnHbA1c) model indicated

that sleep duration was negatively associated with
lnHbA1c (B 0.016, p 0.01) whereas bedtime on
weekends (B 0.010, p 0.132) was not directly associated with lnHbA1c. Therefore, sleep duration completely mediated the association between weekend
bedtime and lnHbA1c.
The mediation effect, estimated using a 1000-replication bootstrapping analysis suggested that for every
earlier hour of bedtime on weekend would be associated
with longer sleep duration, and a decrease in HbA1c
of 1% of its original value (B 0.006, p 0.026)
(Table 5).

Sleep duration as a mediator between bedtime on

weekends and glycemic control
Our findings suggested that those with later bedtimes
had shorter sleep durations, and both shorter sleep
duration and later bedtime on weekends were associated with poorer glycemic control. We therefore
explored further if the relationship between bedtime
and glycemic control was mediated by sleep duration in
a mediation analysis diagrammed in Figure 1. Results of
the mediation analysis are described in Table 4. After
adjusting for relevant covariates in the sleep duration
model, bedtime on weekends was negatively associated
with sleep duration (B 0.355, p50.001), i.e. later
bedtime on weekends was associated with shorter sleep

DISCUSSION
In this study conducted in Thailand, a country with a
tropical climate and close to the equator (latitude
13.8 N), we found that later bedtime, but not midsleep on weekends or wake time, was associated with
poorer glycemic control in type 2 diabetes patients. This
association was statistically significant although bedtime explained only a small amount of variance in
HbA1c. Those with later bedtimes reported significantly
shorter sleep durations and more sleep debt than those
with earlier bedtimes, suggesting that they curtailed
their sleep by waking up earlier than desired, even on
the weekends. In this population, shorter sleep duration
fully mediated the relationship between bedtime and
glycemic control.
These findings are different than those of the previous cohort of type 2 diabetes patients conducted in
Chicago (latitude 41.8 N) which revealed that sleep
timing measured by mid-sleep time on weekends (MSF),

TABLE 3. Hierarchical regression analysis with lnHbA1c as the

outcome (N 210).
Model 1
Variable
Age
Sex (reference: male)
BMI
Diabetes duration
Insulin use
Weekend bedtime
Adjusted R2
DR2

Model 2

p Value

p Value

0.002
0.019
0.002
0.002
0.127

0.070
0.411
0.419
0.236
50.001

0.002
0.022
0.002
0.001
0.125
0.018

0.084
0.335
0.463
0.274
50.001
0.021

0.163

0.181
0.021

0.021

B unstandardized coefficient.

TABLE 4. Mediation analysis of weekend bedtime, sleep duration and lnHbA1c.

Model
Sleep duration (path a)

lnHbA1c (path b)

Factors

SE

Weekend bedtime
Age
Male vs Female
BMI
Diabetes duration
Insulin use
Sleep duration
Weekend bedtime
Age
Male versus female
BMI
Diabetes duration
Insulin use

0.355
0.013
0.059
0.004
0.010
0.158
0.016
0.010
0.001
0.024
0.002
0.001
0.112

0.071
0.011
0.211
0.023
0.012
0.231
0.006
0.006
0.001
0.018
0.002
0.001
0.020

4.962
1.181
0.278
0.190
0.873
0.682
2.576
1.507
1.135
1.342
1.044
1.172
5.695

50.001
0.238
0.781
0.849
0.383
0.495
0.010
0.132
0.256
0.180
0.297
0.241
50.000

95% CI
0.495,
0.034,
0.354,
0.048,
0.034,
0.295,
0.028,
0.003,
0.003,
0.059,
0.002,
0.001,
0.074,

0.215
0.008
0.471
0.040
0.013
0.610
0.004
0.022
0.001
0.011
0.006
0.003
0.151

TABLE 5. Mediation effects of bedtime on weekend on lnHbA1c through sleep duration.

Effects

Pathway

SE

Bias

95% CI

Indirect
Direct

BT ! SD ! lnHbA1c (ab)
BT ! lnHbA1c (c)

0.006
0.009

0.003
0.006

2.22
1.62

0.026
0.106

0.000
0.000

0.001, 0.010
0.002, 0.022

BT, Weekend bedtime; SD, sleep duration

!

Taylor & Francis

Downloaded by [University of California, San Diego] at 17:59 29 November 2015

S. Reutrakul et al.

and not sleep duration, was associated with poorer

glycemic control (Reutrakul et al., 2013). Part of the
reason for this difference could be that sleep duration in
Chicago was similar across chronotypes (MSF), suggesting that those who slept later also woke up later in that
location. This could be due to several reasons. First, the
findings of this study are consistent with previous
findings in younger populations that people residing in
a more tropical climate tended to have a greater
morning preference (Horzum et al., 2015; Smith et al.,
2002), and therefore likely an earlier wake time. MSF in
this study is 2:28 AM, compared with 3:29 AM in Chicago
despite participants being of similar age (58.6 versus
58.4 years). More consistent day length and sunrise time
throughout the year may play a role in this finding as
well. Chicagos earliest sunrise is 5:15 am in June and
latest is 7:15 am in January, while Bangkoks sunrises are
5:50 am and 6:45 am, respectively (www.timeanddate.
com, 2015). Chicagos longest and shortest day length is
9 h and 7 min and 15 h and 13 min, compared with 11 h
and 20 min and 12 h and 56 min in Bangkok (www.
timeanddate.com, 2015). In addition, higher and more
consistent temperature in Bangkok (year round average
high 8794 degrees F) could make it more uncomfortable to sleep in that climate. In addition to the
geographical location, differing social structures may
contribute to the need to wake up at different times.
Although both cities are metropolitan areas, Bangkok
tends to have especially long travel times to and from
work as a result of traffic, necessitating people to wake
up early to arrive at work on time. While we did not ask
the participants about their travel time to work, Bangkok
is well known as having some of the worst traffic
(Bangkok Traffic and Transportation Department, 2015).
Interestingly, employment was not related to glycemic
control in our cohort, and employment rates were not
different across quartiles of bedtime (result not shown),
suggesting that even those who were not working still
woke up relatively early. This could be due to the fact
that living in an extended family household is a norm in
Thailand. Therefore, individuals who are not working
may adjust their sleep-wake schedules to be similar to
the rest of the family for social reasons. Lastly, different
cultural expectations and religious practices may also
play a role in this finding. Late risers have been
traditionally viewed as being lazy in the Thai society.
In addition, most Thais are Buddhists who traditionally
wake up around 5 AM to offer food for the monks. While
this may not be routinely practiced in the modern
society,
it
is
still
practiced
by
25%
(Mahachulalongkornrajavidayala University, 2012).
Altogether, these factors could contribute to habitual
early wake times, which is associated with shorter sleep
duration and poorer glycemic control in those who
preferred later bedtimes in this cohort.
A causative role of partial sleep restriction in
promoting alterations in glucose metabolism was first
reported in 1999 (Spiegel et al., 1999). Intravenous

glucose tolerance testing following sleep restriction of

4 h per night for five nights in 11 healthy male participants resulted in a 24% decrease in insulin sensitivity as
well as a 30% decrease in the acute insulin response to
intravenous glucose (Leproult and Van Cauter, 2010;
Spiegel et al., 1999). This combination led to a more
than 40% decrease in glucose tolerance compared with
the fully rested condition. More recently, the mechanisms underlying abnormal glucose metabolisms as a
result of a combination of circadian misalignment and
sleep restriction were explored in an experimental study
of healthy volunteers (Buxton et al., 2012). Twenty-one
participants underwent a 28-h day forced desynchrony
protocol, in which they slept and consumed isocaloric
meals during a recurring cycle of 28-h days, with
concurrent sleep restriction (5.6 h/day) for 3 weeks to
explore the combined effects of sleep restriction and
circadian disruption as commonly experienced by shift
workers, followed by 9-day recovery period. Circadian
disruption combined with sleep restriction was associated with an 8% increase in fasting glucose levels and a
14% increase in postprandial glucose levels in response
to a standardized breakfast. There was an inadequate
pancreatic b-cell response because fasting and postprandial peak insulin levels were significantly reduced
(by 12 and 27%, respectively). In addition, the resting
metabolic rate decreased by 8%. While these studies
were not conducted in patients with type 2 diabetes,
these data could help explain the detrimental effects of
insufficient sleep and late bedtime on glycemic control
in our current study.
Our results highlight the interaction between sleep
timing, sleep duration and glycemic control in type 2
diabetes patients residing in a tropical area and have
potential clinical implications. While there are no
studies exploring the benefits of reducing mild circadian
misalignment on glucose metabolism, the effects of
sleep extension on glucose metabolism is starting to
emerge. A short-term home sleep extension for 6 weeks
in 16 non-obese healthy adults who were habitual short
sleepers found a significant positive correlation between
changes in total sleep time and indices of insulin
sensitivity at the end of the experiment (Leproult et al.,
2015). In another study, three nights of catch-up sleep
in 19 men with chronic sleep restriction during the work
days resulted in improved insulin sensitivity (Killick
et al., 2015). It is likely that a sleep extension in the Thai
population, if it were conducted, would have to involve
adjusting bedtime, rather than wake up time. This area
is the subject of future research.
Our study has strengths and limitations. To our
knowledge, this is the first study to explore the relationships among sleep timing, duration and glycemic control in a tropical climate. However, the cross-sectional
nature of the study does not provide causal evidence. All
the sleep variables were not objectively measured,
though they were based on standardized questionnaires.
Lastly, as this is a clinic-based sample, whether the
Chronobiology International

Sleep timing, sleep duration and glycemic control

Downloaded by [University of California, San Diego] at 17:59 29 November 2015

findings can be generalized to a larger population

remains to be investigated.
In summary, later bedtime on weekends is associated
with more evening preference, shorter sleep duration
and poorer glycemic control in type 2 diabetes patients.
The relationship between bedtime and glycemic control
was fully mediated by sleep duration. It is likely that
patients with later weekend bedtimes curtail their sleep
by waking up earlier than desired. Exploring the reasons
for this phenomenon, (i.e. culture, metropolitan lifestyle, environment, family and social obligations) may
lead to behavioral modifications (i.e. earlier bedtime
and/or sleep duration extension) and possibly improved
glycemic control.

DECLARATION OF INTEREST
S.R. receives speaker honoraria from Sanofi Aventis and
Medtronic, and research grant from Merck. All other
authors have nothing to disclose.
This study was funded by a grant from Mahidol
University, Bangkok, Thailand.

REFERENCES
Arble DM, Ramsey KM, Bass J, Turek FW. (2010).
Circadian disruption and metabolic disease: Findings from
animal models. Best Pract Res Clin Endocrinol Metab. 24:
785800.
Arora T, Taheri S. (2015). Associations among late chronotype,
body mass index and dietary behaviors in young adolescents.
Int J Obes (Lond). 39:3944.
Aschoff J, Hoffmann K, Pohl H, Wever R. (1975). Re-entrainment of
circadian rhythms after phase-shifts of the Zeitgeber.
Chronobiologia. 2:2378.
Bangkok Traffic and Transportation Department. Bangkok Traffic
Statistics 2013. http://www.bangkok.go.th/traffic/ Updated
2015. [last accessed 16 Aug 2015].
Buxton OM, Cain SW, OConnor SP, et al. (2012). Adverse
metabolic consequences in humans of prolonged sleep restriction combined with circadian disruption. Sci Transl Med. 4:
129ra43.
Buysse DJ, Reynolds III CF, Monk TH, et al. (1989). The Pittsburgh
Sleep Quality Index: A new instrument for psychiatric practice
and research. Psychiatry Res. 28:193213.
Cappuccio FP, DElia L, Strazzullo P, Miller MA. (2010). Quantity
and quality of sleep and incidence of type 2 diabetes: A
systematic review and meta-analysis. Diabetes Care. 33:41420.
Culnan E, Kloss JD, Grandner M. (2013). A prospective study of
weight gain associated with chronotype among college freshmen. Chronobiol Int. 30:68290.
Friborg O, Bjorvatn B, Amponsah B, Pallesen S. (2012). Associations
between seasonal variations in day length (photoperiod), sleep
timing, sleep quality and mood: A comparison between Ghana
(5 degrees) and Norway (69 degrees). J Sleep Res. 21:17684.
Gan Y, Yang C, Tong X, et al. (2015). Shift work and diabetes
mellitus: A meta-analysis of observational studies. Occup
Environ Med. 72:728.
Horzum MB, Randler C, Masal E, et al. (2015). Morningness
eveningness and the environment hypothesisA cross-cultural
comparison of Turkish and German adolescents. Chronobiol
Int. 32:1421.
Huang W, Ramsey KM, Marcheva B, Bass J. (2011). Circadian
rhythms, sleep, and metabolism. J Clin Invest. 121:213341.
!

Taylor & Francis

Imai K, Keele L, Tingley D. (2010). A general approach to causal

mediation analysis. Psychol Methods. 15:30934.
Iwasaki M, Hirose T, Mita T, et al. (2013). Morningness
eveningness questionnaire score correlates with glycated
hemoglobin in middle-aged male workers with type 2 diabetes
mellitus. J Diabetes Investig. 4:37681.
Killick R, Hoyos CM, Melehan KL, et al. (2015). Metabolic and
hormonal effects of catch-up sleep in men with chronic,
repetitive, lifestyle-driven sleep restriction. Clin Endocrinol
(Oxf). 83:498507.
Kim BK, Kim BS, An SY, et al. (2013). Sleep duration and glycemic
control in patients with diabetes mellitus: Korea National
Health and Nutrition Examination Survey 20072010. J Korean
Med Sci. 28:13349.
Knutson KL, Ryden AM, Mander BA, Van CE. (2006). Role of sleep
duration and quality in the risk and severity of type 2 diabetes
mellitus. Arch Intern Med. 166:176874.
Knutson KL, Van Cauter E, Zee P, et al. (2011). Cross-sectional
associations between measures of sleep and markers of glucose
metabolism among subjects with and without diabetes: The
Coronary Artery Risk Development in Young Adults (CARDIA)
Sleep Study. Diabetes Care. 34:11716.
Leproult R, Deliens G, Gilson M, Peigneux P. (2015). Beneficial
impact of sleep extension on fasting insulin sensitivity in adults
with habitual sleep restriction. Sleep. 35:70715.
Leproult R, Holmback U, Van Cauter E. (2014). Circadian misalignment augments markers of insulin resistance and inflammation, independently of sleep loss. Diabetes 63:18609.
Leproult R, Van Cauter E. (2010). Role of sleep and sleep loss in
hormonal release and metabolism. Endocr Dev. 17:1121.
Lucassen EA, Zhao X, Rother KI, et al. (2013). Evening chronotype
is associated with changes in eating behavior, more sleep
apnea, and increased stress hormones in short sleeping obese
individuals. PLoS One. 8:e56519.
MacKinnon DP, Fairchild AJ, Fritz MS. (2007). Mediation analysis.
Annu Rev Psychol. 58:593614.
MacKinnon DP, Lockwood CM, Hoffman JM, et al. (2002).
A comparison of methods to test mediation and other
intervening variable effects. Psychol Methods. 7:83104.
Mahachulalongkornrajavidayala University. http://www.mcu.ac.
th/site/news_in.php?group_id1&NEWSID7599.
Last
updated 2012. [last accessed 20 Aug 2015].
Merikanto I, Lahti T, Puolijoki H, et al. (2013). Associations of
chronotype and sleep with cardiovascular diseases and type 2
diabetes. Chronobiol Int. 30:4707.
Netzer NC, Stoohs RA, Netzer CM, et al. (1999). Using the Berlin
Questionnaire to identify patients at risk for the sleep apnea
syndrome. Ann Intern Med. 131:48591.
Ohkuma T, Fujii H, Iwase M, et al. (2013). Impact of sleep duration
on obesity and the glycemic level in patients with type 2
diabetes: The Fukuoka diabetes registry. Diabetes Care. 36:
61117.
Olds TS, Maher CA, Matricciani L. (2011). Sleep duration or bedtime? Exploring the relationship between sleep habits and
weight status and activity patterns. Sleep. 34:12991307.
Osonoi Y, Mita T, Osonoi T, et al. (2014). Morningnesseveningness
questionnaire score and metabolic parameters in patients with
type 2 diabetes mellitus. Chronobiol Int. 31:101723.
Pornpitakpan C. (1998). Psychometric properties of the composite
scale of morningness. Personal Individ Differ 25:699709.
Portaluppi F, Touitou Y, Smolensky MH. (2008). Ethical and
methodological standards for laboratory and medical biological
rhythm research. Chronobiol Int. 25:9991016.
Preacher KJ, Hayes AF. (2008). Asymptotic and resampling
strategies for assessing and comparing indirect effects in
multiple mediator models. Behav Res Methods. 40:87991.
Radloff L. (1977). The CES-D Scale: A slef-report depression scale
for research in the general population. Appl Psychol Measur 1:
385401.

Downloaded by [University of California, San Diego] at 17:59 29 November 2015

S. Reutrakul et al.

Reutrakul S, Hood MM, Crowley SJ, et al. (2013). Chronotype is

independently associated with glycemic control in type 2
diabetes. Diabetes Care. 36:25239.
Reutrakul S, Van Cauter E. (2014). Interactions between
sleep, circadian function, and glucose metabolism:
Implications for risk and severity of diabetes. Ann N Y Acad
Sci. 1311:15173.
Roenneberg T, Allebrandt KV, Merrow M, Vetter C. (2012). Social
jetlag and obesity. Curr Biol. 22:93943.
Roenneberg T, Kuehnle T, Juda M, et al. (2007a). Epidemiology of
the human circadian clock. Sleep Med Rev. 11:42938.
Roenneberg T, Kuehnle T, Pramstaller PP, et al. (2004). A marker
for the end of adolescence. Curr Biol. 14:R10389.
Roenneberg T, Kumar CJ, Merrow M. (2007b). The human
circadian clock entrains to sun time. Curr Biol. 17:R445.
Sato-Mito N, Sasaki S, Murakami K, et al. (2011). The midpoint of
sleep is associated with dietary intake and dietary behavior
among young Japanese women. Sleep Med. 12:28994.
Scheer FA, Hilton MF, Mantzoros CS, Shea SA. (2009). Adverse
metabolic and cardiovascular consequences of circadian misalignment. Proc Natl Acad Sci USA. 106:44538.
Sitasuwan T, Bussaratid S, Ruttanaumpawan P, Chotinaiwattarakul W.
(2014). Reliability and validity of the Thai version of the Pittsburgh
Sleep Quality Index. J Med Assoc Thai. 97(Suppl 3):S5767.

Smith CS, Folkard S, Schmieder RA, et al. (2002). Investigation of

moring-evening orientation in six countries using the preference scale. Personal Indiv Differ 32:94968.
Smith CS, Reilly C, Midkiff K. (1989). Evaluation of three circadian
rhythm questionnaires with suggestions for an improved
measure of morningness. J Appl Psychol. 74:72838.
Spiegel K, Leproult R, Van Cauter E. (1999). Impact of sleep debt on
metabolic and endocrine function. Lancet 354:14359.
Suksakorn S, Rattanaumpawan P, Banhiran W, et al. (2014).
Reliability and validity of a Thai version of the Berlin questionnaire in patients with sleep disordered breathing. J Med Assoc
Thai. 97(Suppl 3):S4656.
Trangkasombat U, Larpboonsarp V, Havanond P. (1997). CES-D as
a screen for depression in adolescents. J Psychiatric Assoc Thai.
42:213.
Wittmann M, Dinich J, Merrow M, Roenneberg T. (2006). Social
jetlag: Misalignment of biological and social time. Chronobiol
Int. 23:497509.
www.timeanddate.comSunrise, sunset and daylength. Last
updated 2015. www.timeanddate.com. Time and Date AS
[accessed 10 Jul 2015].
Yu JH, Yun CH, Ahn JH, et al. (2015). Evening chronotype is
associated with metabolic disorders and body composition in
middle-aged adults. J Clin Endocrinol Metab. 100:1494502.

Chronobiology International