Professional Documents
Culture Documents
Sleep Medicine: Luciane Gaspar Guedes, Gabriela de Azevedo Abreu, Katia Vergetti Bloch
Sleep Medicine: Luciane Gaspar Guedes, Gabriela de Azevedo Abreu, Katia Vergetti Bloch
Sleep Medicine
journal homepage: www.elsevier.com/locate/sleep
Original Article
a r t i c l e i n f o a b s t r a c t
Article history: Objective: At present, epidemiologic studies regarding the relationship between sleep duration and
Received 4 July 2017 glucose metabolism in adolescents are scarce. The objective was to investigate the association between
Received in revised form self-reported nocturnal sleep duration and glycosylated hemoglobin A in 12- to 17-year-old Brazilian
6 March 2018
adolescents.
Accepted 12 March 2018
Available online 6 April 2018
Patients/Methods: A school-based multicenter cross-sectional study was carried out in private and public
schools from 273 municipalities with more than 100,000 inhabitants. The final sample comprised 24,923
adolescents. A self-administered questionnaire was used. Blood tests included glucose, insulin, glyco-
Keywords:
Sleep length
sylated hemoglobin A and serum lipids. Age, sex, skin color, school type, body mass index (BMI), physical
Glucose metabolism activity, and Brazilian regions were studied as possible effect modifiers and/or confounders using linear
Adolescence regression.
Results: A significant positive association was found between more than 12 h of nocturnal sleep and
glycosylated hemoglobin A in two Brazilian regions: Southeast and South, even after adjustment for age,
sex, skin color, and BMI (coefficients of 0.142 and 0.339, respectively). No association was found with
nocturnal sleep duration <7 h.
Conclusion: Notably, a significant positive relationship was found between more than 12 h of nocturnal
sleep duration and glycosylated hemoglobin A in two Brazilian regions. The specific pubertal sleep
curtailment can be a compensatory mechanism for dealing with the insulin resistance during adoles-
cence. Those that escape from this regulatory strategy and sleep longer than the adequate duration, break
down this balance and tend to damage their glucose metabolism. To our knowledge, this is the first large
scale study, of the association between sleep duration and glucose metabolism in adolescents.
© 2018 Elsevier B.V. All rights reserved.
https://doi.org/10.1016/j.sleep.2018.03.013
1389-9457/© 2018 Elsevier B.V. All rights reserved.
L.G. Guedes et al. / Sleep Medicine 47 (2018) 60e65 61
utilization during sleep correspond to diminished brain metabolism medicines used in the previous day before the blood testing). There
related to the predominance of slow waves [13]. Like Javaheri et al. was a 9.9% loss of data, related to 2739 subjects with incoherent
[12], Koren et al., in a cross-sectional study with 62 obese adoles- answers for sleep hours.
cents, found the same quadratic relationship between sleep
duration and serial glucose in an oral glucose tolerance test, as well 2.2. Instruments for data collection
as with glycosylated hemoglobin A (HbA1c), regardless of obesity,
pubertal stage, sex, and sleep apnea measures. They also found The ERICA Questionnaire of the Adolescent (EQA) was self-
associations between decreased slow wave sleep and reduced in- administered using a personal digital assistant (PDA). The main
sulin secretion [14]. Likewise, the survey in the area of Iwaki-machi, areas covered were socioeconomic status, adolescent work, smok-
Japan, with subjects older than 20 years of age, using the definitions ing, alcohol consumption, physical activity assessment, health and
of pre-diabetes and diabetes mellitus (DM) of the American Dia- medical history, sleeping hours, eating behavior, oral health,
betes Association (ADA) as cut-off points, found the highest prev- common mental disorder, and reproductive health.
alence of high fasting glucose levels and of HbA1c in those with The ERICA questionnaire of the responsible parents/care giver
short or long sleep periods. Logistic regression showed an adjusted provided information on mothers' educational achievement, family
odds ratio of 4.96 (95% confidence interval (CI) ¼ 1.03e23.96) for history of cardiovascular and metabolic diseases, and circum-
the association between sleep of <6 h, and a high HbA1c level and stances related to the student health (birth weight, breastfeeding,
an adjusted odds ratio of 4.96 (95% CI ¼ 1.70e14.50) when the sleep diseases, and sleep duration). A printed form was sent to the
length was 9 h [15]. parents or caregivers through the students.
The HbA1c measurement has many advantages, such as its
stability, easy interpretation, fasting-independent blood test, and 2.3. Study variables
its direct relationship with cardiovascular consequences from
uncontrolled diabetes. The disadvantages are few, related to some Age (in years) was initially studied as a continuous variable in
specific groups: anemic patients, blood recipients, and those with the models. Age was also categorized as 17 years of age or younger
the human immunodeficiency virus [16,17]. than 17 years of age, as a proxy for insulin resistance: the oldest
Considering the incidence and the prevalence of type 2 diabetes subjects being considered the least-resistant. Other variables were
among children and adolescents around the world [18], including sex (female or male), skin color (white or non-white), and school
in Brazil [19,20], and the consequences for cardiovascular health, type (private or public).
this research aims to investigate the association between self- Sleep duration (h) was based on specific questions about sleep
reported nocturnal sleep duration and HbA1c in 12- to 17-year- listed in the EQA: “On a typical weekday, at what time do you
old Brazilian adolescents, from public and private schools. It es- usually sleep?”, “On a typical weekday, at what time do you usually
tablishes the hypothesis that nocturnal sleep length is associated wake up?” “On weekends, at what time do you usually sleep?” and
with the HbA1c levels, exploring the possibility of a quadratic “On weekends, at what time do you usually wake up?” For the
relationship between the exposure and the outcome variables, as evaluation of the hours/minutes of nighttime sleep by this method,
found in other studies [12,14,15]. the weighted average number of hours/minutes of sleep informed
on weekdays and on weekends was used. It was studied as a
2. Materials and methods continuous variable, but additionally it was divided into three
groups of sleep duration: <7 h, 7e12 h, and >12 h of nocturnal
This research is part of a multicenter school-based countrywide sleep, following the National Sleep Foundation's recommendations
cross-sectional study funded by the Brazilian Ministry of Health, [22], adapted for the age range of our study. Maximum limits of
the Study of Cardiovascular Risks in Adolescents (Portuguese ±4 h above or below those considered as “may be appropriate” for
acronym ‘ERICA’). the age range of this research were accepted, as recommended by
Brazil is a continental country, sub-divided into five geograph- the National Sleep Foundation [22].
ical macro-regions (North, Northeast, South, Southeast and Mid- Physical activity (min/week) was assessed through a question-
west), with several socioeconomic and cultural differences among naire included in the EQA, as previously validated [23], containing a
them. Therefore, the relationship between nocturnal sleep duration list of several activities considered moderate to vigorous. The
and HbA1c was assessed nationally and by its five macro-regions. subjects chose the activities exercised in the previous week, using
the PDA, describing the number of days and the daily duration
2.1. Study sample of each activity. This instrument allows the calculation of the
weekly physical activity time, in minutes. Those who performed
The ERICA sample was composed of adolescents aged 300 min/week or more were considered as active [24].
12e17 years, enrolled in the morning or afternoon shifts of private For BMI, the weight and height were measured using a digital
and public schools, from 273 municipalities with more than scale (model P150m, 200 kg of capacity and 50 g of precision), and a
100,000 inhabitants, on July 1, 2009. The overall sample size was portable stadiometer (with millimeter resolution and height up to
estimated in 75,060 subjects. In each of the 3753 selected classes, 213 cm), respectively. The adolescent nutritional status was
every student was invited to participate, but only the morning obtained by the index BMI for age and sex, expressed in z-scores,
students were included in the present research because of the need considering adapted World Health Organization reference values
for fasting blood dosages. The ERICA sample is considered a com- [25,26], and analyzed in three categories: malnourished/low weight,
plex sample, because it uses stratification, clustering, and unequal normal/overweight, obese.
probabilities in its selection stages. The study sampling in detail has All of the anthropometric procedures described above have
already been published [21]. been detailed elsewhere [26].
The data collection started in the first half of 2013 and finished Tanner's pubertal stage was determined by the EQA answers,
in November 2014. At the end of the collection 36,956 students had through self-evaluation boards. Despite its known relationship to
blood exams. HbA1c, glucose, and insulin were tested in 36,628 insulin resistance [27e29], pubertal stage was not used in the
subjects. Exclusion criteria were applied to those adolescents model because of the subjective aspect of the self-evaluation.
(reported or treated diabetes or any other health problem or Therefore, age was studied as proxy for insulin resistance.
62 L.G. Guedes et al. / Sleep Medicine 47 (2018) 60e65
Table 1
Characteristics of the adolescents in the study samplea.
Skin color
White 39.3 37.2 41.5 21.6 19.6 23.7 27.7 23.7 32.2 33.6 30.5 36.8 40.3 36.8 44.0 67.3 63.5 71.0
Non-white 60.7 58.5 62.8 78.4 76.3 80.4 72.3 67.8 76.3 66.4 63.2 69.5 59.7 56.0 63.2 32.7 29.0 36.5
Weight
Very low weight 0.3 0.2 0.5 0.7 0.3 1.7 0.4 0.3 0.6 0.2 0.1 0.5 0.4 0.2 0.7 0.0 0.0 0.1
Low weight 2.0 1.7 2.4 2.3 1.8 2.9 2.3 1.8 3.0 2.2 1.7 2.8 2.0 1.5 2.7 1.2 0.7 1.8
Adequate weight 71.8 70.1 73.4 74.3 72.0 76.5 70.4 67.6 73.0 72.8 71.0 74.6 72.8 69.9 75.6 67.5 64.0 70.7
Overweight 17.0 15.8 18.2 16.1 14.5 17.9 17.3 15.6 19.1 16.7 15.0 18.6 16.3 14.4 18.5 20.0 17.8 22.4
Obese 8.9 8.1 9.7 6.5 5.6 7.6 9.6 8.1 11.4 8.0 6.9 9.3 8.5 7.3 9.8 11.4 9.0 14.2
School
Public 77.4 71.8 82.2 88.1 81.7 92.5 66.2 54.8 76.0 73.9 66.0 80.5 78.6 68.5 86.1 87.0 77.3 92.9
Private 22.6 17.8 28.2 11.9 7.5 18.3 33.8 24.0 45.2 26.1 19.5 34.0 21.4 13.9 31.5 13.0 7.1 22.7
Med 25p 75p Med 25p 75p Med 25p 75p Med 25p 75p Med 25p 75p Med 25p 75p
Physical activity 290.0 80.0 690.0 300.0 60.0 750.0 260.0 60.0 645.0 345.0 90.0 770.0 275.0 80.0 670.0 360.0 120.0 770.0
(min/week)
Sleep duration (h) 7.9 7.1 8.9 8.0 7.0 9.0 8.0 7.1 8.9 7.9 7.1 8.6 7.9 7.1 8.9 8.0 7.3 8.9
HbA1c (%) 5.4 5.2 5.6 5.4 5.2 5.6 5.4 5.2 5.6 5.4 5.2 5.6 5.4 5.2 5.6 5.3 5.2 5.6
HOMA-IR 1.7 1.2 2.5 1.5 1.0 2.2 1.6 1.1 2.4 1.5 1.0 2.2 1.8 1.2 2.5 2.0 1.4 2.8
Insulin (mU/L) 8.2 5.7 11.6 7.3 5.1 10.2 7.7 5.3 11.0 7.4 5.2 10.6 8.4 5.8 11.7 9.5 6.7 13.0
Glucose (mg/dL) 86.0 82.0 91.0 85.0 80.0 89.0 87.0 82.0 91.0 84.0 80.0 89.0 86.0 81.0 90.0 88.0 83.0 93.0
CI, confidence interval; HbA1c, glycosylated hemoglobin A; HOMA-IR, homeostatic model assessment of insulin resistance; Med, median; 25p, 25th percentile; 75p, 75th
percentile.
a
Brazilian Study of Cardiovascular Risks in Adolescents.
Table 2
Glycosylated hemoglobin A by sleep duration groups in Brazil and Brazilian macro-regions.
Med 25p 75p Med 25p 75p Med 25p 75p Med 25p 75p Med 25p 75p Med 25p 75p
<7 h 5.4 5.1 5.6 5.5 5.2 5.6 5.4 5.2 5.6 5.4 5.2 5.6 5.4 5.2 5.6 5.3 5.2 5.5
7e12 h 5.4 5.2 5.6 5.4 5.2 5.6 5.4 5.2 5.6 5.4 5.2 5.6 5.4 5.1 5.6 5.3 5.1 5.6
>12 h 5.6 5.4 5.8 5.4 5.2 5.6 5.3 5.2 5.6 5.4 5.2 5.6 5.6 5.4 5.7 5.8 5.8 5.8
Table 3
Linear regression between self-reported nocturnal sleep duration and glycosylated hemoglobin A by Brazilian macro-regionsa.
Region Sleep duration (h)b Unadjusted coefficient 95% CI p Adjusted coefficientc 95% CI p
(<17 years; 17 years), sex, skin color (white; non-white) and BMI. resistance [11,12,35,39,40,44,45]. Despite its wide utilization,
There was no statistically significant association between sleep HOMA-IR [11,12,39,40,45] has been questioned for representing
<7 h and the outcome studied. only the insulin hepatic action [46].
HbA1c was chosen as the glucose metabolism marker in our
4. Discussion work; a few studies have used it for the specific purpose of studying
the association between self-reported sleep duration and glucose
Sleep influences adolescent health in several ways. It has metabolism [14,15,47]. Berentzen et al. did not find any association
recently been the central focus of many studies [10e12,14,31e44]. between those aspects, although the sample comprised 1481
However, epidemiologic studies about the relationship between adolescents of a restricted age range, from 11 to 12 years, at the
sleep duration and glucose metabolism are scarce. Such studies beginning of the puberty [47]. A quadratic relationship was found
point out the link between sleep length and increased insulin in the other two researches [14,15]; however, Nakajima et al.
64 L.G. Guedes et al. / Sleep Medicine 47 (2018) 60e65
studied 1062 adult subjects [15], and, Koren and colleagues studied mere act of going to bed as the sleep onset, and overestimating it
62 obese adolescents, aged 8e17 years [14]. [10,59]. Some available more detailed questionnaires could be used,
Our results did not show the U-shaped curve expected in the although they are inappropriate as our questionnaire is already
relationship studied. A statistically significant positive association extensive. Sleep diaries are also not practical for our population
was found only for sleep durations longer than 12 h and HbA1c, in research with adolescents. Despite probably being more accurate,
the South and Southeast Brazilian regions, even after adjustment those strategies are still subjective with respect to sleep duration
for potential confounding factors. measurement. Unfortunately, the objective techniques, such as the
Considering our findings, it is meaningful that the National polysomnography and the actigraphy, are also unfeasible for
Sleep Foundation does not recommend more than 12 h of sleep for population-based studies. Another limitation is the lack of evalu-
children and adolescents from 6 to 13 years, and, not more than ation of the sleep quality influence on our findings. Again, as
11 h for those from 14 to 17 years of age [22]. These recommen- explained previously, this assessment was not operationally
dations supported by studies such as Javaheri and colleagues [12], possible. However, as reported by Koren et al. in their review, the
Koren et al. [14], and Duggan et al. [48], pointing out not only the relationship between obstructive sleep apnea and insulin resis-
importance of sleep deprivation but also the possible undesirable tance in children is less clear, without conclusive results [10]. The
consequences of excessive hours of sleep. third limitation refers to our cross-sectional study design. It does
Still, sleep is a process composed of sequential, alternate and cy- not allow us to establish directly the temporality between cause
clic phases that occur in the brain: N1, N2 and N3 stages of non-rapid and effect. Hence, clinical trials and prospective longitudinal
eye movement (non-REM) sleep and REM sleep [49]. Some studies studies should be carried out.
[14,35,44,45,50] show negative association between N1 sleep stage Nevertheless, our research presents the strength of being a
percentage and insulin secretion and/or insulin sensitivity, and pos- population-based study, with a sample as large as 24,923 adoles-
itive association between N3 sleep stage percentage and the same cents, from all the Brazilian regions, with national and regional
conditions [13], independent on the total sleep duration. representation.
Some hypotheses are suggested to explain our findings. First, the
specific pubertal sleep curtailment, in fact, is a compensatory 5. Conclusions
mechanism of the human body for facing the insulin resistance that
occurs during this time. Some studies corroborate this hypothesis A positive association was found between self-reported
[51,52]. Crowley et al. showed decreased salivary melatonin during nocturnal sleep length of >12 h in adolescents and HbA1c from
puberty in 69 subjects aged 9.6e17.8 years [51]. Maslowsky and two Brazilian regions. Sleep should be included in the adolescent
Ozer, in a longitudinal research with 15,701 adolescents, verified a health agenda and discussed at different levels of national health
progressive decrease in the sleep length across puberty, with initial policies. It should be also systematically assessed by health
recovery after 18 years, reaching the pre-pubertal levels at around professionals during adolescents' routine consultations.
23 years [52]. Those that escape from this functional regulatory
strategy and sleep longer than what is adequate, break down this Acknowledgements
balance and tend to increase the HbA1c. This is probably a result of
an increased percentage of a more superficial sleep, with smaller The authors would like to thank the adolescents who partici-
amounts of slowewave activity, both negatively related to insulin pated in this research, and CAPES (Coordenaça ~o de Aperfeiçoa-
secretion and insulin sensitivity, with glucose metabolism impair- mento de Pessoal de Nível Superior) for L.G.G.’s scholarship
ment, as previously described [14,35,44,45,50]. (BEX 3610/15-2). This article is part of the first author's doctorate
The second hypothesis concerns Brazilian regional diversity, thesis. The ERICA Study was funded by the Ministry of Health of
regarding the different adolescent lifestyles. Other ERICA researches Brazil (MS), Public Call e MCT/FINEP/MS/SCTIE/DECIT e CT-Saúde
have shown that adolescents from the South (one of the most and FNS e Síndrome Metabo lica e 01/2008 (Grant No. FINEP:
developed regions in Brazil) have greater alcohol consumption [53], 01090421, CNPq: 565037/2010-2 and 405009/2012-7) (the sponsor
with a relative frequency of 27.5%, preference for beverages with had no involvement in the study design; collection, analysis, and
higher alcohol levels, and higher smoking prevalence compared to interpretation of data; writing of the report; or the decision to
the other regions [54]. In addition, they have more screen time submit the paper for publication).
(use of computer, TV, and electronic games more than 2 h a day by
77.9% of those teenagers) [55], which could be related to lower sleep
Conflicts of interest
duration and quality [56]. Differences related to eating habits among
the adolescents were also found in all regions [57,58]. Considering
The authors have no conflicts of interest to declare regarding
our findings, the most relevant results of the last two researches
this study.
were the highest prevalence of soft drinks intake [58] and the lowest
The ICMJE Uniform Disclosure Form for Potential Conflicts of
water consumption [57] among those from the South. Macronu-
Interest associated with this article can be viewed by clicking on the
trient percentages were not different among the regions [58]. All
following link: https://doi.org/10.1016/j.sleep.2018.03.013.
of the mentioned factors might have contributed for the macro-
regions-related differences found in Table 3.
Third, the lack of a negative association between <7 h of References
nocturnal sleep and HbA1c levels might have been caused by [1] Mansur AP, Favarato D. Mortality due to cardiovascular diseases in Brazil and
compensating naps during the day. Unfortunately, information in the metropolitan region of Sa ~o Paulo: a 2011 update. Arq Bras Cardiol
about this sleep aspect was not collected. 2012;99(2):755e61.
[2] Soares GP, Klein CH, Silva NA, et al. Evolution of cardiovascular diseases
Finally, residual confounding also might have occurred because
mortality in the counties of the State of Rio de Janeiro from 1979 to 2010. Arq
of the broad categories of the sleep groups in our research. Bras Cardiol 2015;104(5):356e65.
They were adapted and based on the National Sleep Foundation [3] Berenson GS, Srinivasan SR, Bao W, et al. Association between multiple
recommendations [22]. cardiovascular risk factors and atherosclerosis in children and young adults.
The Bogalusa Heart Study. N Engl J Med 1998;338(23):1650e6.
Our study has the subjective sleep measure as its main limita- [4] Raghuveer G. Lifetime cardiovascular risk of childhood obesity. Am J Clin Nutr
tion, related to the memory and also to the interpretation of the 2010;91(5). 1514Se9S.
L.G. Guedes et al. / Sleep Medicine 47 (2018) 60e65 65