896
Original article
A population-basedstudlt of reducedsleep duration and
hypertension:the strongestassociationmay be in
premenopausal
women
bauerioStranges"'b,Joan M. Dornb'', FrancescoP. Cappucciod,
Richard P. Donahueb,Lisa B. Rafalson',KathleenM. Hoveyb,
MichelleA. Millero
Kandalad,
Ngianga-Bakwin
Jo L. Freudenheimb,
and MaurizioTrevisanb'r
Objectives Recentevidenceindicatesthat reduced sleep
durationmay be associatedwith an increasedrisk of
hypertensionwith possiblystronger effects among women
than men. We therefore examinedcross-sectionalsexspecificassociationsof sleep durationwith hypertensionin
a large population-basedsample from the Western New
York HealthStudy (1996-2001).
factors, and psychiatriccomorbidities,and is stronger
among premenopausalwomen. Prospectiveand
mechanisticevidenceis necessaryto support causality.
J Hypertens28:896-902 @ 2010 Wolters Kluwer Health I
LippincottWilliams& Wilkins.
Journalof Hypertension2010, 28:896-902
and
Methods Participantswere 3027 white men (43.50/o)
without prevalentcardiovascular disease
women (56.50/o)
(medianage 56 years).Hypertension
was definedas blood
pressureat least140 or at least90 mmHg or regularuse of
antihypertensivemedication.Multivariatelogistic
regressionanalyseswere performedto estimate odds
ratios (ORs) of hypertensioncomparing less than 6h of
sleep per night versus the referencecategory (X h) while
accountingfor a nurnber of potentialconfounders.
Keywords:blood pressure,epidemiology,hypertension,sex differences,
sleep duration
Abbreviations: BMl, body mass index; CES-D,Center for Epidemiologic
Studies DepressionScale;CI, confidenceinterval;CVD,cardiovascular
disease;HCFA,HealthcareFinancingAdministration;MONICA'Monitoring
trends and determinantson cardiovasculardiseases;NHANES'National
Healthand NutritionExaminationSurvey;OR, odds ratio;SDB' sleepdisorderedbreathing;SES,socioeconomicstatus;SF-36,Short Form'36;
SPSS, Statistical Package for Social Sciences; WNYHS'Western New York
Health Study
Results In multivariateanalyses,less than 6 h of sleep was
associatedwith a significantincreasedrisk of hypertension
comparedto sleepingat least 6 h per night,only among
women [OR: 1.66(1.09to 2.53)].No significantassociation
was found among men IOR: 0.93(0.62to 1.41X.
aHealthSciencesResearchlnstitute,University
of WarwickMedicalSchool,
Medicine,
of Socialand Preventive
Coventry,UnitedKingdom,bDepartment
cDepartment
of Exerciseand NutritionSciences,Schoolof PublicHealthand
HealthProfessions,State Universityof New York (SUNY)at Buffalo,Buffalo,New
of Wamick Medical
University
York,USA,dClinicalSciencesResearchInstitute,
eDepadment
of FamilyMedicine'Schoolof
School,Coventry,UnitedKingdom,
Medicineand BiomedicalSciences,State Universityof New York at Buffalo,
Buffalo,New York and rHealthSciencesSystem,NevadaSystemof Higher
Education,Las Vegas,Nevad4 USA
In subgroup analysesby menopausalstatus,the effect was
womenIOR: 3.25(1'37to
strongeramongpremenopausal
women [OR:1.49
7.76)lthan among postmenopausal
(0.92to 2.41)1.
to SaverioStranges,MD, PhD,AssociateClinicalProfessorof
Corresponde4ce
of
HealthSciencesResearchInstitute,University
Cardiovascular
Epidemiology,
Waruick MedicalSchool,MedicalSchoolBuilding,RoomA105, GibbetHill
Campus,CoveniryCV4 7AL, UK
Tel: + 44 2476151153; Iax:+ 44 2476528375i
e-mail:S.Stranges@wamick.ac.uk
Conclusion Reducedsleep duration,by increasingthe risk
of hypertension,may produce detrimentalcardiovascular
effects among women. The associationis independentof
socioeconomicstatus, traditionalcardiovascularrisk
Received4 August2009 Revised11 November2009
Accepted 22 November2009
lntroduction
disordersduring a mean follow-up of 8-10 years [7]' In
agreementr.viththis, in a cross-secdonalanalysisfrom the
SleepHeart Health Study on a largesampleof US adults of
both sexes, individuals sleeping less than 6h per night
reported a 66Vo higher prevalence of hypenension as
compared to those sleeping berween 7 and 8 h per night
[8]. These results were further supported by recent findings from the CARDIA study showing that reduced sleep
duration, measured by r,vristactigraphy,predicted higher
blood pressure levels and adverse changes in blood
Growing epidemiologicalevidence indicatesthat reduced
sleep duration and/or sleep deprivation may be associated
with a number of cardiovascular disease (CVD) and
cardiovascularrisk factorssuch as hypertensionU-11].
Specifically, in a prospective analysis of the first US
Narional Health and Nutrition Examination Survey
(NHANES-I), sleeping5 h or lessper night was associated
with a 60Vo increased risk of incident hypertension in
middle-aged US adults of both sexeswithout overt sleep
See editorialcommenton page 883
Sleepdurationandhypertension
risk Stranqes
ef al. gg7
pressureover 5 years among 578 African Americans and
whites aged33-45 yearsat baseline[9].
Furthermore, the Whirehall II Study in the UK showed
potential sex differencesin the associadonofreduced sleep
duration with hypertension risk, with a significant effect
only among women [10]. In parricular, in cross-sectional
analyses, the odds ratio (OR) of hypertension among
women sleeping 5 h or lessper night was l.7Z [95% confidence incerval(CI) 1.07-2,.751
comparedwith the group
sleeping 7 h, whereasno associationwas dereccedin men
(OR 0.88,95VoCI0.63-1.23).In agreementwith this, a
cross-sectionalanalysisof the Heinz Nixdorf Recall Study
in Germany [11] showed a significantassociarionberween
reducedsleepduration(<5 h per nighr) and hypertension
only amongwomen (OR 1.24,95VoCI1.04-1.46).
controls:residentsof Erie and NiagaraCounties;age3579 years;no cancer.Porential participanrswere idencified
through rwo sources:Deparrment of Nlocor Vehicles of
New York State for participanrs aged 35-64 years. This
source was used because it is known rhat 95% of New
York residents in this age group have a driver license;
Healthcare Financing Administration (HCFA) lists for
participanrs aged 65-79 years. This source was used
because it includes virtually all individuals in the age
rangeof interesr.Between1996and 2001, atoalof 6837
porenrial participants were identified, conracted and
deemed eligible for the srudy. Of these, a total of 4065
agreed ro participate and .lvereexamined, for a participation rate of 59.5Vo.
All participanrscame to the Center for
Preventive Vledicine at the University at Buffalo for an
interview and physical examination that lasted approximately 2.5h. The study protocol was approved by the
University ac Buffalo Insritutional Review Board.
Altogether, thesefindings raiseconcernrhar reduced sleep
duration might produce detrimental cardiovasculareffects
particularly in women, as supporred by several independent studies evaiuaring sex-specific effects of sleep ' The exclusion criteria applied to rhe presenr analysis
were race other from r,vhite (//:381); a self-reported
duration on CVD-reiacedmorbidity and mortaliq [lZhistory of prevalent coronary heart disease, that is
161.These findings may have considerablepublic health
previous
myocardial infarction, coronary artery bypass
implications,given current declining trends in the average
graft
surgery,
angioplasry,or diagnosed angina pectoris
duration of sleep in the generalpopulation [17,18].
(I/:502); missing sleep data (,4/:155). The remaining
The mechanismsunderlying the sex-specificassociarion 3027 participants (56.5Vowomen), aged 35-79 years
of reduced sleep duration with hypertensionand other
(median 56 years),were included in this study.
cardiovascular outcomes are unclear. Hormonal influences and psychosocial distress may contribure to the
Sleep duration
observed associations,particularly during periods markSleep durarion in the past week was ascertainedwirh the
ing shifts in the reproductive stages of women, such as
Seven-Day PhysicalActivity Recall quesrionnairell3lby
menopause[19]. Nloreover,merhodologicalissuessuch
the question 'On average,how many hours did you sleep
as differential self-reporting of sleep habirs berween men
each night during the last five weekday nights (Sundayand women may also play a role [20]. However, the
Thursday)l' Responsecategorieswere collapsedin rwo
possibility that reduced sleep duration may represenra
main groups:lessthan 6 h and at leasr6 h. In our sample,
risk marker rarher than a casual risk factor for diseases there were few individuals who reported sleeping more
cannot be ruled out ar rhe presenrtime [21]
than 8 h per night (N:92 among women, Iy':81 among
men).
There fore, these participanrs were grouped
In the presenr analysis we sought to examine crosstogether
with the mid-range caregory (6-8 h).
sectional sex-specificassociationsof sleep duration with
hypertension in the Western New York Health Study
Covariates
(WNYHS), a iarge, well characterizedpopulation-based
The
following covariateswere examined: age, education
sample from the United States[21]. In chronic diseases,
(years), marital starus, and annual household income,
menopauserepresentsan imporrant effecr modifier of the
grouped as socio-demographics;
body massindex (BMII),
relation between exposure to endogenous factors and
waist circumference,pulse rate,physical activiry, drinking
disease risk 1221.Therefore, unlike previous srudies,
and smoking habits,and type 2 diabetes,grouped as CVD
we decided a priori to perform subgroup analyses by
risk factors; Short Form-36 (SF-36) physical and mental
menopausalstatus among women to further investigate
health
scores,and depressivesymproms,grouped as genpotential mechanismsfor the observedsex-specificeffect
eral
health/psychiatric
comorbidities.
of reduced sleep durarion on rhe risk of hypertension.
Methods
Studypopulation
Participants in this study were originally enrolled as
healthy control participanrs in the WNYHS, a series of
case-control studies to examine risk factors for chronic
disease,as describedin detail elsewhere[21]. The following eligibiliry criteria were used to identify porential
Specifically, marital starus was categorized as married/
unmarried. As a proxy measur€ of current or recent
socioeconomicstatus (SES), annual household income
was usedand divided in three categoriesin orderof decreasing income:above$70000,$30000-70 000,below930000.
For women, menopausalstatus was determined using
criteria taking into accountage,surgicaior naturalcessation
of menses,hormone use, and bilateral oopherecromy.
2010,Vol28 No5
of Hypertension
898 Journal
Health examination inciuded measurementsof anthropometric variables such as height, weight, and waist
circumference made by trained and certified intervielvers
on pafiicipants wearing light clothing and no shoes.BIVII
was calculated as weight in kilograms (kg)/heighc in
meters (m2). Waist circumference was determined r.vith
pardcipants standing erect with the abdomen relaxed,
arms at the side, and feet together. The tape was horizontally placed between the bottom of the rib cage and
the top of the iliac crest (hip bones) around the smallest
circumference between these two reference points. The
measurementwas taken at the €nd of a normalexpiration,
without the tape compressing the skin, to the nearest
0.1cm.
Lifetime and current physical activity was aiso assessed.
For the lifetime assessment,participants were given a
table to recall how many hours per week they regularly
did strenuous leisur'e dme physical activity or exercise,
vigorous household chores or manual yard work during
each indicated time period, starting at age 10. The total
hours for lifetime strenuous physical activity were
obtained by summing up hours for each time period.
Average hours per week acrosslifetime were calculated
accordingly. Current physical acriviry in the past week
was ascertained with the Seven-Day Physical Activity
Recall questionnaire used in the Stanford Five-City
Projecr [23]. For drinking habits, current drinkers were
defined as participants who reported consuming at least
one alcoholic beverage in the 30 days prior to intervier,v;
former drinkers were defined as participants who
reported 12 or more drinks during their lifetime or in
any l-year period, but did not consume an alcoholic
beverage at least once in the past 30 days; lifetime
abstainers were defined as participants who reported
consumption of less that !2 drinks during their lifecime
or in any 1-yearperiod [21]. For smoking habits,participanrs reporting current use of cigarettes were classified
as current smokers; participants reporting not having
smoked at least 100 cigarettesin their lifetime were
classified as never smokers. Other participants r,vere
classifiedas former smokers.
Fasting glucose concentrations were determined by
glucose oxidase methods. Type 2 diabetes was defined
either as fastingglucoseat least 126m{dl (>7.0mmol/l)
or use of antidiabeticmedication.
General health status was assessedusing the physical and mental health component summaries of the
SF-36 health surveyquestionnaire[24]. These summary
measuresare standardizedas /-scoresand have higher
reliability than the individual scales.The presenceof
depressivesymptomswas assessedby using the Center
for Epidemiologic Studies Depression Scale (CES-D)
[25]; participantswere divided in two groups based on
the cut point for major depressive symptoms (score
>22).
Hypertension
Blood pressurewas measuredthree times in the sitting
position using a standard mercllry sphygmomanometer
by trained and cerdfied technicians, according to a stanwere askedto
dardizedprotocol[26]. Briefly, participancs
remain seated, legs uncrossed,feet flat on the floor, for
a period of 5 min and to refrain from ralking during
blood pressure measurement. The infladon pressure
.,vas 30 mmHg above the pulse obliteration pressure
and the cuff was released at a ftte of 2 mmHg/s. The
onset of the first phase(systolic)and fifth phase(diastolic)
Korotkoff sounds were recorded. The means of the
second and third measureswere used in the analyses.
Hypertension lvas defined as blood pressureat least 140/
90 mmHg or regular use of antihypertensive medication
[27]. Pulse rate from the radial artery at the wrist was also
recorded three times. The means of the secondand third
measureswere used in the analvses.
Statistical analysis
For continuous and categorical variables, respectively,
Kruskal-lVallis and 12 tests were used to determine the
sutistical significance of any difference in the distributions of selectedvariablesby sex and sleep duration
categories(<6h; >6h). Vlultivariate logistic regression
analyseswere performedto calculateORs and95% CIs of
hypertension comparing less than 6 h sieep versus the
reference category (>6 h). We used three progressive
levels of adjustment. Mlodel 1 was adjusted for age,
education,marital status,and householdincome (sociodemographics).iVlodel 2 was further adjusted for BIVII,
waist circumference, pulse rate, physical activiry, drinking status,smoking status,and type 2 diabetes (CVD risk
factors).Model 3 was further adjusted for SF-36 physical
and mental scores,and depressivesymptoms (general
health/psychiatriccomorbidities). All analyseswere stratified by sex (P<0.001 for sexx sleep duration interaction). To further investigate potential mechanismsfor
the observedsex-specificeffect ofreduced sleep duration
on hypertension risk, we decided a priori to perform
subgroup analysesby menopausalstatus among women,
despite lack of a statisticallysignificant interaction.There
was no sratistically significant interaction between age
(either as concinuousor categoricalvariable) and sleep
durationin either sex.Analyseswere conductedusing the
SPSS
StatisticalPackagefor SociaiSciences(SPSS-15.0;
Inc., Chicago,Illinois, USA).
Results
Selected characteristicsof participantsby categoriesof
sleep duration are reporred in Table 1 for women and
men separately. For both sexes, there were significant,
consistentassociations
for severalcorrelatesof reduced
sleep duration. For example, participantssleeping less
than 6 h per night were less educated, heavier, more
likely to be unmarried, noncurrent drinker and current
Sleep durationand hypertensionrisk Stranoesef a/. g99
Table1 Selectedcharacteristicsa
of participantsby categoriesof
sleep duration (N = 3027)
Table2 Odds ratios' (950/0Cl) of hypertensionby categoriesof
sleep durationin women (N:1710)
Sleep duration
Variable
>6h
Sleep duration
Pb
W o m e n( n : 17 1 0 )
Age (years)
(70)
Premenopausal
Education(years)
Unmanied(%)
Lowest income(o/o)
BMI (kg/m2)
Waist (cm)
Pulse rate (beats/mjn)
Physicalactivity(h/week)
Currentdrinker(o/o)
Currentsmoker(o/o)
Diabetes(o/o)
SF-36PhysicalScore
SF-36 MentalScore
Depressivesymptoms(o/o)
Hypertension(o/o)
M e n ( n : 1 3 17 )
Age (years)
Education(years)
Unmarried(o/o)
Lowest income(o/o)
BMI (kg/m2)
Waist (cm)
Pulserate (beats/minute)
Phydicalactivity(h/week)
Currentdrinker(o/o)
Currentsmoker(o/o)
Diabetes(07s1
SF-36PhysicalScore
SF-36MentalScore
Depressivesymptoms(o/o)
Hypertension(076)
>6h
Overall(N: 1710)
223
56.7
28.3
12.9
Jd.b
45.6
zv.o
90.6
7'1.5
R'
50.5
18.4
6.6
44.9
50.4
21.8
39.5
191
54.9
13.4
22.5
28.9
28.7
99.3
68.5
5.4
23.6
8.7
49.3
51.7
12.8
30.9
1487
cJ.b
34.4
I J.O
28.2
32.0
27.7
85.7
70.9
5.0
o+.J
14.7
6.8
49.3
52.6
10.1
26.8
223
0.178
0.068
<0.001
0.002
<0.001
<0.001
<0.001
0.380
<0.048
<0.00J
0.149
0.932
<0.001
0.001
<0.001
<0.001
1126
c/.o
11.1
15.0
23.7
24.1
98.1
68.5
76.6
14.1
9.8
cu.c
Ft o
4.8
36.7
0.003
0.001
0.009
0.138
0.132
0.314
0.984
o.726
0 . 19 4
0.001
o.642
0.071
<0.001
<0.001
0 . 13 4
Western New York Health Study, Buffalo,USA (j996-2001). "Data are
bP valuesfor comparison
expressedas the meanor as percenlages.
between
sleep durationcategoriesusing the X2 test for categoricalvariablesand ihe
Kruskal-Wallis
test for continuousvariables.
smoker, reporred lower SF-36 physical and menral scores
and more frequenr depressive symptoms rhan parricipants sleepingar leasr6 h per night.
The prevalence of diabetes did not significantly differ
berween sleep duration caregories.The prevalence of
hypertension was significantly higher in pardcipants
sleepingless rhan 6 h per night only among women.
TableZ displaysORs (95% CI) of prevalenrhyperrension
by categoriesof sleep duration in women, comparing less
than 6h sleep versus rhe reference caregory (>6h).
Reducedsleepduration(<6 h per nighr) was consistenrly
associatedwith a significanrhigher risk of hyperrensionas
compared to the group sleeping at least 6 h, in both
unadjusted [OR:1.78 (1.33 r.o 2.38)] and adjusted
[OR: I.66 (1.09to 2.53)l analyses.When analyseswere
stratified by menopausal srarus, the associarion of
reduced sleep duration wirh hypertension was srronger
and significantamongpremenopausalwomen [OR:3.25
(1.37 to 7.76))as compared ro posrmenopausalcounrerparts [OR:L49 (0.92 to 2.41)).
Unadjustedmodel
Modet1b (SES)
Model 2c (CVD risk factors)
Model 3d (fullyadjusted)
Premenopausal
women(N: 575)
I ln.Ai'
'.ta.i
m^Aal
Model 3b (fullyadjusted)
Postmenopausal
women(/V: 1135)
Unadjusted
model
Model 3b (fullyadjusted)
J ? q r / 1e a - o
l . , v
\ | . v v
ea\
Z l v v l
1 . 7 ef i . 2 7 - 2 . 4 8 1
1. 7 5 ( 1 . 2 0 - 2 . 5 5 )
1 . 6 6( 1 . 0 9 - 2 . 5 3 )
1447
1.00
't.00
1.00
1.00
63
2.46(1.25-4.84]|
3 . 2 5( 1 . 3 7 - 7 . 7 6 )
512
1.00
1.00
160
1. 5 8 ( 1 . 1 2 - 2 . 2 1 )
1 . 4 9( 0 . 9 2 - 2 . 4 1 \
975
1.00
1.00
WesternNew YorkHealthStudy,Buffalo,USA ( j 996-2001). Thefullyadjusted
modelincludesmenopausal
statusin the overallsampleof women. uOddsratios
of prevalenthypertension
comparingless than 6 h sleep versusthe reference
bModel
category(>6h).
1: age, education,maritalstaius,householdincome.
'Model
2: M1 +BMl, waistcircumference,
pulserate,physicalactivity,drinking
status,smokingstatus,type 2 diabetes.dModel3: M2+SF-36ohvsical/mental
scores,depressivesymptoms.
Table 3 displaysORs(95VoCI)of prevalenrhyperrension
by categoriesof sleep duration in men, comparing less
than 6h sleep versusthe referencecaregory(>6h). No
significant associarionstvere found in either unadjusted
or adjusted analysestOR:0.93 (0.62 to 1.41)1.We also
conducred sffarified analyses by age caregories (i.e.
middle-agedversuselderly) in men, bur results did not
yield any significant associarion(data not shown).
Discussion
Findings from the WNYHS corroborarethe norion that
reduced sleep duration may be associaredwith higher
risk of hypertension only among women. Consistenr
with previous unrelated reporrs [10,11] from different
countries (UK and Germany), we found a 66Vo higher
prevalence of hypertension among women sleeping less
than 6 h per night as compared ro rhe group sleeping at
least 6 h per night. This associationwas independent
of socioeconomic starus, traditional cardiovascularrisk
factors,and psychiatriccomorbidiries. As a novel finding,
Table
3 Oddsratiosa(950/o
Cl) of hypertension
by categories
of
sleepduration
in men(N:1317)
SIeep duration
Unadjustedmodel
Modet1b (sES)
Model 2' (CVD risk factors)
Model 3d (tutlyadjusted)
<6h
>6h
191
0.78(0.56-1.08)
0 " 9 2( 0 . 6 5 - 1 . 3 0 )
0.94(0.64-1.37)
0 . 9 3 ( 0 . 6 2 -1. 4 1 )
I 126
1.00
1.00
1.00
1.00
Western New York Health Study, Buffalo, USA (1996-2001). aOdds ratios of
prevalent hypertension comparing less than 6 h sleep versus the reference
category (>6h).'Model
1: age, education, marital status, household income.
"Model 2: M1 +BMl, waist circumference, pulse rate, physical activity, drinking
status, smoking status, type 2 diabetes. o Model g: M2 + SF_36 phvsical/mental
scores, depressive symptoms.
2010' Vol28 No5
of Hypertension
9OO Journal
in subgroup analysesby menopausalstatus, the effect of
reduce"dsleep durarion on hypertension was exacerbated
and more than doubled (OR 3'25 versus 1'49) among
premenopausalwomen as comparedto postmenopausal
.orrnr"rpurrr. No significant associations were found
among men.
tension, current evidence points to several potentiai
mechanisms.For example, acrtte sleep restriccionmay
induce an overactiviry of the sympathetic nervous system
leading to higher blood pressurein boch normotenslve
and hypertensive individu alsl40-421. Other contributing
mechanismsmay include ov€ractivityof the renin-angiosystem,renal impairment,endochelial
tensin-aldosterone
dysfunction, proinflammatory responses,metabolic and
On the contrary, interendocrine alterations 139,43,441.
vention studiesto improve duration and quality of sleep
have been effective in reducing both daytime and nighttime blood pressures[39,45].Horvever,it is possiblethat
reducedsleep duration may representa marker of health
status and quality of life rather than a causal factor for
hypertension and other health outcomes 121,46)'
Observationalepidemiologicalstudiessuggestthat both
'short' and 'long' sleep duration may be associatedwith
an increased ,Gk fo, health outcomes, such as total
mortality, cardiovascuiardisease,type 2 diabetes mellitus, obesiry, and poor general health status [1-16,21]'
These ,.rrdi"t have often shown a U-shaped relationship
between sleep duration and health t28'Z9l' However,
whereas findings on the associationsof long sleep
duration with health outcomes are not fully consisrent
Potential mechanisms that may explain sex-specific
and their underlying mechanisms are lictle understood
health
effectsofreduced sleepdurationon cardiovascular
reducedsleep
[21,30],there is increasingrecognitionthat
tlme;
present
the
at
unknown
are
risk
hypertension
and
duration (or chronic sleep restriction)may representan
Howspeculative'
is
highly
discussion
any
such
therefore
emerging risk factor or risk marker for increasedmorbidever, it is possible that hormonal influences may play a
iry and mortality [31]. Specifically,consistentepidemiorole,especiallyduring times suchaschemenopausewhen
logical evidence indicates that reduced sleep duration
women are more vulnerable due to major hormonai
may be associatedwith a number of cardiovascuiaroutfluctuations t19,22]. N{oreover,there are sex differences
comes l2-ll], and may increasemortality risk primarily
in stressand reaction to stress[47]' Recent findings from
through effects on CVD [1]' Funhermore, there is conthe Whitehall II Study also poinc to potential sex differthu, reduced sleep duration may produce detrimenences in the associationof sleep duration with inflam""r.r
tal cardiovasculareffects particularly among women, as
matory markers [48]. Furthermore, as shown in our
supported by severalunrelated studies examining sexdescriptive analyses,the disuibution of correlatesof
rpl"in" effects of sleep duration on cardiovascularendreduced sleep duration that have the potential to affect
points [10-16]. For example,findings from the Vlonitorrisk (e.g. BNII, waist, lowest SES' SF-36
ing trends and determinantson cardiovasculardiseases hyperrension
health, depressive symptoms) was
mental
and
pirysical
(MONICA) Augsburg Study on a large sample of German
sexes and may have parbetween
different
iomewhat
adultsshoweda prospectiveassociationof reducedsleep
associations'Finally,
observed
the
to
contributed
tially
duration (<5 h per night) with incident myocardial infracof sleep habits
self-reporting
differential
possibility
of
the
tion only among women (hazatd ratio 2'98,95% CI 1'48ruled out' as
be
cannot
women
and
men
between
6.03),as comparedwith the group sleeping 8h [16]; the
Sleep Heart
the
of
analysis
previous
a
in
suggested
(95%
CI
correspondinghazard rario among men was I'13
subjecdve
to
sex
Health Study on the relationship of
0.66-l.gZ). Moreover, two recentiy published studies,
measuresof sleePiness[20].
from the UK and Germany, consistently showed signifisleep
reduced
associationsbetween
cant cross-sectional
Altogether, most of the studies on the effect of reduced
duration (<5 h per night) and hypertension only among
sleepdurationon blood pressureievels and hypertension
women t10,11], though the effect size was somewhat
risk have found significant associationsonly in youngerl
different between the two samples (OR 1'72 versus
Three studieswere
middle-agedindividuals 17,9,49,501'
1.24, respecrively). The present study additionally
sleep duration and
between
associations
unable to find
suggests that hypertensive effects of reduced sleep
eiderly
the
in
17,57,521'The present
hypertension
durarion may be strongerin premenopausalwomen'
status could represent
menopausal
that
,*dy tngg"sts
age-related effect of
for
rhe
explanation
a plausible
Sleep-disorderedbreathing (SDB) has been linked to
risk at least in
hypertension
on
duration
sleep
reduced
elevated blood pressure and risk of hypertension in
women.
several epidemiological studies L32'361' The observational data have been corroborated by findings of
The cross-sectionalnature of the present scudy does
mechanisticstudiesemphasizingthe critical role of symnoc allow establishingcausalityand temporality of the
Dathetic overactivity in the cause of SDB-related hyperGiven the self-reportingof sleep
observedassociations.
le.rsio.r [37,38], although other mechanisms are likely to
habits, we cannot disregard the likelihood that health
be involved [39].
outcomes such as hypertension may influence reports of
habitual sleep habits [53] and impair sleep quality [54],
Wirh regard to the biological plausibility of the associand not vice versa.Nevertheless,self-reportassessments
ation between behavioral sleep deprivation and hyper-
Sleep duration and hypertension risk Stranges et a/. 901
of sleep have been shown to be valid measurescompared
to quantitative sleep assessmentswirh actigraphy or
polysomnography[55,56]. As a' further limication, we
were unable to control for the potential effect of SDB,
which is clearly linked to elevated blood pressureand
risk of hypertension 132-361.Moreover, the suboptimal
participation rate (59.5%) may leave the possibility for
selection bias and restrict the generalizability of chese
findings to the generalpopulation. However, the consistency in the magniude of the observed association
between reduced sleep duration and hypertension risk
with previous unrelated studies [10,11] strengthensthe
external validity and generalizability of these findings.
The strengthsof this study include the extensiveassessment of several covariatesknown to be related to either
sleep habits or hypertension risk, and the large community-wide sample.
Perspectives
The presencstudy corroboratessex-specificassociations
befi,veensleep duration and hypertension risk. Specifically, reduced sleep duration (<6 h per nighc)was associated with a higher risk of hypertension only among
womgn. As a novel observation, findings of the present
studj' showed that this association may be stronger in
premenopausaiwomen. These findings raise additional
concern that behavioral curtailment of sleep might produce detrimental effects on cardiovascularhealth particularly among \Momen. This may have important public
health implications given current declining trends in the
average duration of sleep in the general population
[17,18].Prospectiveand mechanisticevidenceis necessary to support causality.
10
11
t3
14
tc
to
Acknowledgements
The Western New York Health Study has been supported by grants from the US National Institute on
Alcohol Abuse and Alcoholism (grant 5 P50 AA09802
to NI.T.); the US Nacional Institutes of Health (grant
NIH R01 DK60587 to R.P.D.).
17
'19
20
F.P.C. holds the Cephalon Chair, an endowed post at
Warwick Vledical School, the result of a donation from
the company. The appointment to the Chair was made
entirely independently of the company and the postholder is free to devise his own program of research.
Cephalon do not have any sqke in IP associatedwith the
postholderand the Chair has complete academicindependence from the company"
tt
IJ
References
1
2
FerrieJ, ShipleyMJ,CappuccioFP, BrunnerE, MillerMA, KumariM,
lrlarmotMG. Sleepdurationand changein sleepduration:
associations
with mortalityin the Whitehallll cohort.Sleep2007',
30:1659-1666.
ShankarA, Koh WP, YuanJM, Lee HP, Yu MC. Sleepduration
and coronaryheartdiseasemortalityamongChineseadultsin
cohortstudy.Am J Epideniol20O8;
Singapore:
a population-based
1 6 8 : 31 6 7 - 13 7 3 .
25
26
EguchiK, PickeringTG, SchwartzJE,HoshideS, lshikawaJ, lshikawaS,
predictorof cardiovascular
el a/.Shortsleepdurationas an independent
eventsin Japanesepatientswith hypertension.Arch lntern Med 2Oa8;
168t2225-2231.
KingCR, KnutsonKL,RathouzPJ,SidneyS, Liu K, Lauderdale
DS. Short
sleepdurationand incidentcoronaryarterycalcification.
J Am Med Assoc
2008;300:2859-2866.
StrangesS, CappuccioFP,KandalaN-8, MillerMA,TaggartFM,KumariM,
versusprospective
et a/. Cross-sectional
associations
of sleepduration
with changesin relativeweightand bodyfat distribution:
the Whitehallll
-329.
Study.4m J Epideniol2008; 167:321
CappuccioFP,TaggartFM,KandalaNB, CurrieA, PeileE, StrangesS,
of shortsleepdurationandobesityin childrenand
MilierMA.Meta-analysis
adults.S/eep2008; 31:619-626.
SB, Boden-Albala
GangwischJE,Heymsfield
B, BuijsRM, KreierF,
fG, et al. Shortsleepdurationas a riskfactorfor hypertension:
Pickering
analysesof the first NationalHealthand NutritionExaminationSurvey.
2006; 47:833-839.
Hypertension
GottliebDJ,RedlineS, NietoFJ,BaldwinCM, NewmanAB, ResnickHE,
PunjabiNM. Associationof usualsleepdurationwith hypertension:
the
. / e e p2 0 0 6 ; 2 9 : 1 0 0 9 - 1 0 1 4 .
S l e e pH e a r tH e a l t hS t u d y S
KnutsonKL, Van CauterE, RathouzPi, Yan LL, HulleySB, Liu K,
DS. Association
beiweensleepand bloodpressurein midlife:
Lauderdale
the CARDIAsleepstudy.Arch lnternMed 2009;169:1055- 1061.
CappuccioFP,StrangesS, KandalaN-8, MillerMA,TaggartFM,KumariM,
ef a/. Gender-specificassociationsof short sleep durationwith prevalent
the Whitehallll Study"Hypertension
and incidenthypertension:
2AO7;
5 0 : 6 9 3 -7 0 0 .
S, ErbelR, JockelKH, HeinzNixdorf
StangA, MoebusS, Mohlenkamp
Group.Gender-specific
associations
RecallStudylnvestigaiive
of short
hypertension.
Hypertension
sleepdurationwith prevalent
2008; 51:e15e 16 .
Ayas NT,White DP, MansonJE,StampferM.1,SpeizerFE,MalhoiraA, Hu
FB. A prospectivestudy of sleep durationand coronaryheartdiseasein
women.Arch lnternMed 2003; 163:205-209.
WK, MansonJE,StampferMJ,SpeizerFE,
AyasNT,White DP,Al-Delaimy
el a/.A prospective
studyof self-reported
sleepdurationand incident
diabetesin women.DrabetesCarc 2OO3-,26:380-384.
PatelSR, Ayas NT, MalhotraMR,White DP, Schernhammer
ES, Speizer
FE,et al A prospectivestudyof sleepdurationand mortalityriskin women.
Sleep20o4; 27:44o- 444.
PatelSR, MalhotraA, Whiie DP, Gottlieb DJ,Hu FB.Associationbetween
reducedsleep and weight gain in women.Am J Epideniol 20O6i
164:947-954.
A. Sleepduratron
Meisinger
C, HeierM, LowelH, SchneiderA, Dciring
and
and riskof myocardial
infarction
in middle-aged
menand
sleepcomplaints
The MONICA/KORAAugsburg
womenfromthe generalpopulation.
CohortStudy.Sleep2OO7i3O:1121-1127.
sleepdeprived.S/eep1995i
BonnetMH, ArandDL.We are chronically
1 8 : 9 0 8 - 9 .11.
KruegerPM, FriedmanEM.Sleepdurationin the UnitedStates:a crossstudy.lm J Epidemiol2009; 169:1052sectionalpopulation-based
10 6 3 .
ParryBL, Newton RP. Chronobiologicalbasis of female-specificmood
2OO1;25:S102-S108.
disorders.Neuropsychopharmacology
BaldwinCM, KapurVK, HolbergCJ, RosenC, NietoFJ,SleepHeart
betweengenderand measuresof
HealthStudyGroup.Associations
in the SleepHeartHeahhStudy.S/eep2004;
daytimesomnolence
27:305-31 .
M, MillerMA,ef a/.
StrangesS, DornJM,ShipleyMJ,KandalaNB,Trevisan
comparison
Correlatesof shortand long sleepduration:a cross-cultural
betweenthe UnitedKingdomandthe UnitedStates.lm J Epideniol2oOa',
16 8 : 13 5 3 - 13 6 4 .
its epidemiology
and
SowersMR, La PietraMT. Menopause:
potentialassociationwith chronic diseases.Epideniol Rev 1995;
17:287-3O2.
SallisJF,HaskellWL, Wood PD, FortmannSP,
RogersT, Blair SN, PaffenbargerRS jr. Physicalactivity
assessmentmethodologyin the Five-CiiyProject.Am J Epideniol 1985;
1 2 1 : 9 -1 1 0 6 .
BrazierJE,HarperR, JonesNM, O'CathainA, ThomasKJ,UshemoodT,
the SF-36healthsurueyquestionnaire:
new
WestlakeL. Validating
outcomemeasurefor primarycare.BMJ 1992;305:160-164.
RadloffLS.TheCES-Dscale:a self-reportdepressionscalefor researchin
Appl PsycholMeas 1977;1:385-401.
the generalpopulation.
for'INTERSALT',
an international
ElliottP, StamlerR.Manualof operations
studyon the relationof sodiumand potassiumto blood
cooperative
pressure.ControlClin lrmls 1988; 9 (Suppl):1S-117S.
902 Journalof Hypertension 2010' Vol 28 No 5
2 7 ChobanianAV, BakrisGL, BlackHR,CushmanWC, GreenLA,lzzoJL Jr,
on
Committee
JointNational
et al, NationalHeart,Lung,andBloodInstitute
andTreatment
of HighBloodPressure;
Detection,Evaluation,
Prevention,
NationalHigh Blood PressureEducationProgramCoordinating
The SeventhReportof the JointNationalCommitteeon
Committee.
andTreatment
of HighBloodPressure:
Detection,Evaluation,
Prevention,
the JNC 7 repon.JAMA 2003; 289:2560-2572.
2A KnutsonKL,TurekFW. The U-shapedassociationbetweensleep
and health:the 2 peaksdo not meanthe samething.S/eep2006;
31
eo
33
35
36
37
29:878-879.
whatthe U-shapedcurue
BliwiseDL,YoungTB. The parableof parabola:
can and cannottell us aboutsleep.S/eep2007; 30:1614- l 615'
for sleep
rationale
YoungstedtSD, KripkeDF. Longsleepand mortality:
restriction.SleepMed Rev 2OO4i8:159-174.
Al LawatiNM, Patel SR, Ayas NT. Epidemiology,risk factors,and
sleepapneaand shortsleepduration'Prog
of obstructive
consequences
CardiovascDrs 2009; 51:285-293.
YoungT, PeppardP, PaltaM, Hla KM, FinnL, MorganB, SkatrudJ'
study of sleep-disorderedbreathingas a risk factor for
Population-based
hvoertension.Atch Intern Med 1997; 157:17 46' 1752.
N:ietoFJ,YoungTB, LindBK, ShaharE' SametJM' RedlineS, et a/'
sleepapnea,and hypertension
breathing,
of sleep-disordered
Association
sludy'JAMA 2000; 283:1829-1836'
in a largecommunity-based
PeppardPE, Young T, PaltaM, Skakud J. Prospectivestudyof the
breathingand hypertension'N Engl
associationbetweensleep-disordered
J Med 2OOO:342,137A- 1344.
sleepapnea-hypopnea
DuranJ, EsnaolaS, RubioR, lztuetaA. Obstructive
sampleof subjectsaged
andrelatedclinicalfeaturesin a population-based
30 to 70 yr.Am J RespirCrit CareMed 2OO1;163:685-689.
HaasDC, FosterGL, NietoJ, RedlineS, ResnickHE, RobbinsJA,et al'
breathingand
betweensleep-disordered
associations
Age-dependent
Circulation2005; 111:614-621 .
hvDertens;on.
NarkiewiczK, SomersVK.Thesympatheticneruoussystemand obstructive
J Hypertens1997; 15:1613for hypertension.
sleepapnea:implicaiions
1616.
FletcherEC. Sympatheticoveractivityin the etiologyof hypertensionof
obstructivesleepapnea.S/eep2003; 26:15- l I'
39 Wolk R,SomersVK. Sleepandthe metabolicsyndrome.ExpPhysiol2007;
92:67-74.
sleepon
40 TochikuboO, lkedaA, MiyajimaE, lshiiM. Effectsof insufficient
recorder.
bloodpressuremonitoredby a new multibiomedical
1996; 27:l 318-1 324'
Hypertension
41 LusardiP, MugelliniA, Preti P, Zoppi A, DerosaG, FogariR' Effecisof a
in
bloodpressuremonitoring
restrictedsleepregimenon ambulatory
subjects'Am J Hypertens1996; 9:503-505'
normotensive
38
LusardiP, Zoppi A, Preti P, Pesce RM, PiazzaE' FogariR Effectsof
patients:a 24-hstudy'
insufficient
sleepon bloodpressurein hypedensive
Am J Hypertens1999; 12:63-68.
te
sleep'obesityand cardiovascular
MillerMA, CappuccioFP.Inflammation,
2007i 5:95-102.
disease.Curr Vasc Pharmacol
44 ScheerFA, HiltonMF, MantzorosCS, Shea SA' Adversemetabolicand
cardiovascularconsequencesof circadianmisalignment'Proc NatlAcad
Sci U S,4 2009i106:4453-4458.
ScheerFA,Van MontgransGA, van SomerenEl, MairuhuG' Bui.isRM'
Dailynight-timemelatoninreducesbloodpressurein malepaiientswith
2OO4i431192-197'
Hypertension
essentialhypertension.
and in both directions'
Zee PC, TurekFW. Sleepand health:everywhere
Arch lnternMed 2006; 166:1686- 1688.
4 7 RediehsM, ReisJ, CreasonN. Sleepin old age:focuson gender
S/eep1990: 13:41O- 424.
differences.
M, KumariM, BrunnerEJ,LoweGD, ef ai'
4 8 MillerMA,KandalaN-B,Kivimaki
beiweensleep
relationships
in the cross-sectional
Genderdifferences
Whitehallll Study.S/eep2009;
durationand markersof inflammation:
3 2 : 8 5 7- 8 6 4 .
WellsJCK,HallalPC, ReichertFF,MenezesAMB,AraujoCLP' Victoria
viewingin relationto obesityandblood
CG. Sleeppatternsandtelevision
birthcohort'lnt I Obes
pressure:evidencefrom an adolescentBrazilian
2008;32:1042-1049.
S, StorferisserA, RosenCL,RedlineS. Sleepqualityandelevated
5 0 Javaheri
Circulation2008; 118:1034-1 040'
bloodpressurein adolescents.
cl
van den Berg JF,TulenJHM,NevenAK, HofmanA, MiedemaHME,
are not
WittemanJCM,TiemeierH. Sleepdurationand hypertension
2007; 50:585-589'
associatedin the elderly.Hypertension
with
MF, RochaFL.Usualsleepdurationis not associated
5 2 Lima-Costa
elderly:the BambuiHealthAgingStudy(BHAS).
in Brazilian
hypertension
S l e e pM e d 2 0 0 8 ;9 : 8 0 6 - 8 0 7 .
5 3 LauderdaleDS, KnutsonKL, Yan LL, Liu K, RathouzPJ.Self-reportedand
measuredsleep duration:how similarare they? Epidemiology2ooa;
19 : 8 3 8 - 8 4 5 .
5 4 SahlinC, FranklinKA, StenlundH, LindbergE' Sleepin women:normal
valuesfor sleep stages and positionand the effect of age, obesity,sleep
SleepMed 2009;10:1025apnea,smoking,alcoholand hypertension.
1030.
and
5 5 LockleySW, SkeneDJ,ArendtJ. Comparisonbetweensubjective
actigraphicmeasurementof sleep and sleep rhythmsJ S/eepRes 1999;
8:175-183.
in flightcrew:
c o SignalTL, GaleJ, GanderPH' Sleepmeasurement
Avrat
comparingactigraphicand subjectiveestimatesto polysomnography.
SpaceEnvironMed 2OO5;75:1058-1 063'