Population-Based Study of Sleep-Disordered

Breathing as a Risk Factor for Hypertension

Terry Young, PhD; Paul Peppard, MS; Mari Palta, PhD; K. Mae Hla, MD;
Laurel Finn, MS; Barbara Morgan, PhD; James Skatrud, MD

Background: Clinical observations have linked sleep-
disordered breathing, a condition of repeated apneas and
hypopneas during sleep, with hypertension but evi-
dence for an independent association has been lacking.
Understanding this relationship is important because the
prevalence of sleep-disordered breathing is high in adults.
Objective: To test the hypothesis that sleep-dis-
ordered breathing is related to elevated blood pressure
ing apnea-hypopnea index (P=.003 for systolic, P=.01
for diastolic, adjusted for confounding factors). The mag-
nitude of the linear association increased with decreas-
ing obesity. At a body mass index (weight in kilograms
divided by the square of the height in meters) of 30
kg/m2, an apnea-hypopnea index of 15 (vs 0) was asso-
ciated with blood pressure increases of 3.6 mm Hg for
systolic (95% confidence interval, 1.3-6.0) and 1.8
mm Hg for diastolic (95% confidence interval, 0.3-3.3).

independent of confounding factors.
Methods: The sample included 1060 employed women
and men aged 30 through 60 years who had completed
an overnight protocol as part of the Wisconsin Sleep Co-
hort Study. In-laboratory polysomnography was used to
determine sleep-disordered breathing status, quantified
as the number of apneas and hypopneas per hour of sleep
(apnea-hypopnea index). Blood pressure was measured
on the night polysomnography was performed.
The odds ratio for hypertension associated with an apnea-
hypopnea index of 15 (vs 0) was 1.8 (95% confidence
interval, 1.3-2.4).
Conclusions: There is a dose-response relationship be-
tween sleep-disordered breathing and blood pressure, in-
dependent of known confounding factors. If causal, the
high prevalence of sleep-disordered breathing could ac-
count for hypertension in a substantial number of adults
in the United States.

Results: Blood pressure increased linearly with increas- Arch Intern Med. 1997;157:1746-1752
_I_

From the Departments
of Preventive Medicine
(Drs Young and Palta and
Mr Peppard and Ms Finn),
Medicine (Drs Hla and
Skatrud), and Kinesiology
(Dr Morgan), University
of Wisconsin, Madison.
THE
repeatedepisodes of ap-
and hypopnea in sleep-
nea
disordered breathing are
known to cause transient
elevations in blood pres¬
sure (BP) during sleep, and it has been hy¬
pothesized that these episodes result in el¬
evated daytime BP as well.1 Because the
prevalence of untreated sleep-disordered
breathing, ranging from mild to severe, is
high in both women and men (9% and
24%, respectively), quantifying the role of
this condition in the development of hy¬
pertension is particularly important.2 Even
a modest role for sleep-disordered breath¬
ing in BP elevation would place a large
number of people at increased risk for car¬
diovascular morbidity and mortality.
A causal role of sleep-disordered
breathing in hypertension is supported by
strong biological plausibility and some ex¬
perimental evidence. The acute physiologi¬
cal consequences of apnea and hypopnea,
including hypoxemia, hypercapnia, arous-
als from sleep, and large negative intratho-
racic pressures, may affect BP regulation
through both neural and humoral mecha¬
nisms. Although a causal mechanism has
not been established, there is some evi¬
dence that patients with sleep-disordered
breathing have increased sympathetic nerve
activity,3 decreased baroreceptor sensitiv¬
ity,4 accentuated vascular responsive¬
ness,5 and abnormal salt and water metabo¬
lism,6 all of which could contribute to
hypertension. A few studies710 have shown
BP to decrease in patients with sleep-
disordered breathing after successful treat¬
ment by tracheostomy or by the most com¬
monly used therapy of nasal continuous
positive air pressure.
Widely cited clinical observa¬
tions1113 that 50% to 90% of patients with
sleep-disordered breathing have hyper¬
tension add support to the hypothesis.
However, inconsistent findings have re-

Downloaded From: http://archinte.jamanetwork.com/ by a University of California - San Diego User on 06/04/2015

 METHODS
DESIGN AND SAMPLE
A major aim of the ongoing Wisconsin Sleep Cohort Study
is to investigate prospectively cardiovascular morbidity as an
outcome of sleep-disordered breathing. For this investiga¬
tion we analyzed data on sleep-disordered breathing and BP
collected during the baseline overnight study. Details on the
study design and sample construction have been reported pre¬
viously2; a brief summary follows. All employees, aged 30
through 60 years, of 5 diverse state agencies in south central
Wisconsin were sent a questionnaire on their sleep habits and
other factors. Completed questionnaires were received from
79% of those contacted. A stratified probability sample for
cohort recruitment was drawn from the sampling frame of
survey respondents (n=4927). Potential participants were con¬
tacted by mail and subsequent telephone calls. Of those in¬
vited, 50% have successfully completed an overnight study
protocol; 42% have refused participation or failed to re¬
spond to recruitment as yet; and 8% are still considering par¬
ticipation. The most common reason for refusal was the in¬
convenience of being away from home overnight.
A high response rate was not expected because of the
participant burden, so the ability to assess response bias
was built into the study design by characterizing the sam¬
pling frame in detail. Data on 56 factors, including in¬
come, job description, sleep habits, medical history, and
lifestyle, were collected on the entire sampling frame. Com¬
parison of cohort participants and nonparticipants on these
factors showed no significant differences with respect to
age, gender, cardiovascular or pulmonary disease history,
marital status, sleep habits, or sleep complaints. Com¬
pared with nonrespondents, respondents had more edu¬
cation, had a slightly greater body mass index (BMI, a mea¬
sure of weight in kilograms divided by the square of the
height in meters), were more likely to report snoring, and
were less likely to report a history of hypertension. To spe¬
cifically determine whether bias could affect the results of
the present investigation, we used multiple regression to
examine the associations of self-reported snoring with self-
reported BP and hypertension in the cohort sample,
the sample of invitees, and the total sampling frame.
suited from recent clinic-based studies14'21designed to
specifically test thehypothesis thatsleep-disordered
breathing is related to increased BP afteradjustment for
obesity, age, and other potential confounding factors. Fur¬
thermore, with the exception of our study22 using am¬
bulatory BP monitoring in a subset of a population sample,
no population-based study to date has found a statisti¬
cally significant association between sleep-disordered
breathing and elevated BP.23'25 This lack of concurrence
has prompted skepticism regarding a meaningful link be¬
tween sleep-disordered breathing and hypertension, but
has also focused attention on méthodologie limitations
of these studies.1-26-27
The goal of the study reported herein was to
explicitly quantify the independent association of sleep-
disordered breathing with BP. The investigation was
conducted with a general population sample of 1069
middle-aged men and women enrolled in the Wiscon-
Downloaded From: http://archinte.jamanetwork.com/ by a University of California - San Diego User on 06/04/2015
The magnitude and significance of associations did not
materially differ across the 3 samples, indicating that
response bias should not lead to a spurious association
between sleep-disordered breathing and BP in this inves¬
tigation.
DATA COLLECTION
The study protocol was implemented during an overnight stay
in a sleep research laboratory designed specifically for the Wis¬
consin Sleep Cohort Study. The sleep study rooms are com¬
fortably furnished and have the appearance of typical bed¬
rooms. The protocol relevant to this investigation included
an interview, measures of body habitus, measures of seated
BP prior to bedtime, and overnight polysomnography.
The polysomnography consisted of a standard mon¬
tage of surface leads to record analog data on cardiopul-
monary function during sleep. Sleep state was monitored
by electroencephalography, electro-oculography, and sub-
mental electromyography. Airflow was detected at the na¬
res by end-tidal carbon dioxide detection (capnograph) and
at the mouth by thermistry (ProTec thermocouple, Pro-
Tec, Woodinville, Wash). Respiratory effort was mea¬
sured by chest and abdominal excursions using calibrated
inductance plethysmography (Respitrace Ambulatory Moni¬
toring, Ardsley, NY). Arterial oxygénation was recorded by
finger pulse oximetry (Omeda 3740, Englewood, Colo).
The polysomnography records were manually scored
using conventional criteria.30 Each 30-second epoch of the
recordings was scored for sleep stage (1-4) and rapid eye
movement, presence of an apnea (absence of airflow for s 10
seconds with either no respiratory effort or opposing chest
and abdomen excursions indicating attempt to breath against
a closed airway), and presence of a hypopnea (reduction
in respiratory effort with a s4% dip in oxygen satura¬
tion). A polysomnographic study of acceptable quality was
defined by adequate sleep and breathing signals through¬
out the night, at least 4 hours of objectively measured sleep,
and at least 1 period of rapid eye movement sleep.
The total number of scored apneas and hypopneas di¬
vided by the number of hours of sleep (apnea-hypopnea in¬
dex [AHI]) was determined for each participant as the
Continued on next page
sin Sleep Cohort Study, a longitudinal study of the
natural history of sleep-disordered breathing. The study
is unique in having the combined strengtbs of a large
population-based sample and of assessment of sleep-
disordered breathing by in-laboratory polysomnogra¬
phy, the standard of clinical practice for diagnosis.28-29
RESULTS
Systolic and diastolic BPs, unadjusted for any of the
covariates, increased with increasing severity of sleep-
disordered breathing (Table 1 ). The same trend was ap¬
parent for hypertension prevalence based on systolic BP
of 140 mm Hg or more, diastolic BP of 90 mm Hg or more,
or use of antihypertensive medication. Body habitus in¬
dicators, age, and the proportion of men increased with
increasing severity of sleep-disordered breathing, high-
 summary measure of sleep-disordered breathing. For de¬
scriptive analyses, AHI cutoff points at 5,15, and 30 were used.
Blood pressure was measured by conventional stan¬
dard mercury sphygmomanometer according to the Ameri¬
can Heart Association recommendations31 in the early
evening, following orientation and acclimation to the bed¬
room. Measurements were made using the left arm with
an appropriate-sized cuff, after participants had been seated
for 15 minutes. Three readings at 5-minute intervals of sys¬
tolic and diastolic (phase 5) BPs were recorded.
Body habitus measures, including height and weight
without shoes and waist, neck, and hip girths, were mea¬
sured using standard procedures.32 Body mass index was
calculated from the height and weight. Use of antihyper¬
tensive medication was determined by interview. Partici¬
pants who reported current use of a- and ß-blockers,
calcium channel blockers, diuretics, or angiotensin-
converting enzyme inhibitors for treatment of hypertension
were coded as positive for medication use. Information on
medical history, smoking, alcohol use, education, age, and
other sociodemographic factors were also obtained as part
of the interview.
DATA ANALYSIS
Data were analyzed with SAS statistical software33 and
SUDAAN34 software modules for descriptive statistics, con¬
tingency tables, multiple linear regression, and logistic re¬
gression. To account for the stratified sampling of the sleep
cohort, all analyses were weighted to give unbiased esti¬
mates. The SUDAAN software was used to compute ap¬
propriate SEs for the weighted analyses.
The association between sleep-disordered breathing
and BP adjusted for confounding factors was quantified by
multiple regression techniques. Apnea-hypopnea index, the
primary independent variable, was used in all models as a
continuous variable.
Multiple linear regression was used to estimate the
change in BP associated with increases in the AHI. Al¬
though linear regression has the advantage of modeling BP
as a continuous variable, the ability to estimate the true re¬
lationship is hampered by the influence of antihyperten¬
sive medication on BP. To account for this, we fit models
lighting the need to account for these factors that are also
correlates of hypertension.
Multiple linear regression modeling, adjusting for
age, sex, and several indicators of body habitus (BMI,
waist-hip ratio, neck girth, and skinfolds thickness),
showed that sleep-disordered breathing, measured by AHI,
was significantly related to systolic and diastolic BPs. The
body habitus variables were all significant in the regres¬
sion models and reduced the coefficient for AHI to the
same degree. Use of combinations of the body habitus
variables produced the same results. Body mass index has
been most widely used, so this measure of body habitus
was adopted. There was no interaction of AHI and age
or sex, indicating that the relationship between AHI and
BP did not vary by age or sex. There was, however, a sig¬
nificant negative interaction of AHI and BMI, indicating
the association of AHI and BP decreases with increasing
BMI. Although the decrease is small, to correctly esti-
Downloaded From: http://archinte.jamanetwork.com/ by a University of California - San Diego User on 06/04/2015
with an interaction term, allowing associations between
sleep-disordered breathing and BP to be estimated for those
using and not using antihypertensive medication. Since, as
expected, there was no linear association between AHI and
BP for those receiving medication, only the regression re¬
sults for those not using medication are presented. (The
results were the same as those obtained from performing
regression on the sample after medication users are ex¬
cluded.) This strategy is commonly used in linear regres¬
sion analyses of BP as the outcome, but it has been shown
that this leads to underestimation of associations.33 Use of
multiple logistic regression modeling with hypertension as
the outcome variable avoids the analytic problem due to
data with BPs influenced by medication. This approach al¬
lows both people with BPs above specified cutoff points and
people with treated hypertension, regardless of BP level,
to be coded as having hypertension. For the logistic mod¬
eling, 3 hypertension variables were created: (1) systolic
BP of 140 mm Hg or more or antihypertensive medication
use, (2) diastolic BP of 90 mm Hg or more or antihyper¬
tensive medication use, and (3) systolic BP of 140 mm Hg
or more or diastolic BP of 90 mm Hg or more or antihy¬
pertensive medication use.
Age, gender, body habitus (height, weight, BMI, skin-
folds, and waist, hip, and neck circumferences), smoking
(current, former, or never status and amount in pack-
years) alcohol use (usual weekly consumption and amount
,
consumed 24 hours before the sleep study), education, and
physical activity were investigated as confounding fac¬
tors. Confounding was assessed by the standard method
of comparing the association of interest before and after ad¬
dition of each potential confounding factor.36 When ad¬
justment changed the regression coefficient for AHI by 15%
or more, the covariate was retained in the final model. In
addition, interactions between the covariates and AHI with
respect to BP were tested for statistical significance. The
statistical significance of linear regression coefficients was
assessed by í tests and that of logistic regression coeffi¬
cients was assessed by Wald x2 tests. Two-tailed P values
of less than .05 were considered to indicate statistical sig¬
nificance. Standard regression diagnostics were per¬
formed to assess model fit and adequacy of compliance with
the modeling assumptions.
mate the association of AHI and BP, it is necessary to
specify a BMI level. Final models of systolic and dias¬
tolic BPs, centered at a BMI of 30 kg/m2 for individuals
not using antihypertensive medication, are given in
Table 2. Under the conditions of the model, the ß co¬
efficient for AHI indicates an increase of 0.24 and 0.12
mm Hg in systolic and diastolic BPs, respectively, for each
additional apnea or hypopnea per hour of sleep. The model
predicts, for example, that BPs will be 3.6 mm Hg (sys¬
tolic) and 1.2 mm Hg (diastolic) higher for mild sleep-
disordered breathing (AHI, 15) vs no sleep-disordered
breathing (AHI, 0). To illustrate the decrease in this
effect with increasing obesity, predicted BP increases
associated with mild to moderate sleep-disordered breath¬
ing over 3 BMI levels are given in Figure I.
The final multiple logistic regression model of sleep-
disordered breathing and hypertension, with terms for
AHI, sex, age, BMI, and an interaction term for BMI and
 Table 1. Sample Characteristics According to AHI Categories: Sleep Cohort Study*
AHI Stratum
<2 <2 to <5 =s5to<15 =15to<30 >30
Characteristics (n=642) (n=191) (n=141) (n=58) (n=37)
Mean (SD), AHI 0.5(0.6) 3.3(0.8) 8.6(2.7) 21.0(4.0) 52.9(17.7)1
Mean (SD) age, y 44.0(7.3) 47.1(7.9) 47.2(7.7) 47.9(8.1) 46.9(8.3)
Male, No. (%) 315(49.1) 130(68.1) 95(67.4) 45(77.6) 32(86.5)
Current smokers, No. (%) 128(20.1) 41(21.6) 27(19.3) 9(15.5) 8(22.2)
Mean (SD)
Alcoholic drinks per wk 3.9(5.8) 5.1(9.7) 4.4(7.5) 5.2(7.1) 3.2(5.7)
Body mass index 27.7(5.6) 30.5(6.1) 31.7(6.2) 34.1(6.9) 38.1(8.7)
Waist girth/hip girth 0.87(0.09) 0.92(0.08) 0.92(0.10) 0.95(0.06) 0.99(0.06)
Neck girth, cm 36.8(4.0) 39.5(4.1) 39.8(3.8) 41.7(3.5) 43.9(3.3)
Systolic BP, mm Hg 123.0(13.7) 127.3(13.7) 129.6(14.7) 133.1(13.5) 136.6(15.8)
Diastolic BP, mm Hg 80.8(9.4) 83.2(8.8) 84.7(9.1) 87.0(10.5) 88.2(11.8)
Antihypertensive medication, No. (%) 53(8.3) 26(13.7) 23(16.4) 9(15.5) 16(44.4)
Systolic BP ==140 mm Hg, No. (%) 72(11.2) 34(17.8) 31(22.0) 13(22.4) 16(43.2)
Diastolic BP ==90 mm Hg, No. (%) 100(15.6) 39(20.4) 37(26.2) 21(36.2) 15(40.5)
Systolic BP&140 mm Hg or receiving BP 105(16.4) 50(26.2) 47(33.3) 18(31.0) 25(67.6)
medication, No. (%)
Diastolic BP >90 mm Hg or receiving BP 130(20.0) 55(28.8) 51(36.2) 26(44.8) 25(67.6)
medication, No. (%)
Systolic BP ==140 mm Hg or diastolic BP >90 151(23.5) 66(34.6) 59(41.8) 28(48.3) 30(81.1)
mm Hg or receiving BP medication. No. (%)

*
N= 1069. AHI Indicates apnea-hypopnea index; body mass index, a measure of weight in kilograms divided by the square of the height in meters; and BP, blood
pressure.
tRange for AHI, 30.5 to 97.5.

Table 2. Multiple Linear Regression Models for Systolic and Diastolic BPs: Sleep Cohort Study*
Systolic BP Model Diastolic BP Model
r 1 r 1
Independent Variable SE(ß) SE(ß)
AHI, events per h .24 0.08 .003 .12 0.05 .01
Age, 1 y .34 0.07 <.001 .13 0.05 .004
Sex, Male 5.50 0.95 <.001 3.77 0.66 <.001
BMIf .71 0.09 <,001 .39 0.08 <.001
AHIxBMIf -.014 0.006 .03 -.010 0.004 .01

*
N=936. The formulas to calculate the millimeters of mercury increase in systolic and diastolic blood pressure (BP) associated with other apnea-hypopnea
Index (AHI)-body mass index (BMI, a measure of weight in kilograms divided by the square of the height in meters) combinations can be calculated as millimeters
of mercury=[0.24+(BMI- 30)x (-0.014)]xAHI. The increase in millimeters of mercury in diastolic BP associated with any AHI-BMI combination can be calculated
as millimeters of mercury=[0.12+(BMI-30)x (-0.01)]xAHI.
\Body mass Index is centered at 30 kg/m2.

AHI, indicated that each additional apnea or hypopnea
per hour of sleep increased the risk of having hyperten¬
sion by approximately 4% (ß coefficient, .037). The odds
ratios and 95% confidence intervals for hypertension as¬
sociated with AHI levels of 5, 15, and 30 predicted by
the models (centered at a BMI of 30 mm Hg) are given
in Table 3; the slight decrease in effect with increasing
BMI is illustrated in Figure 2.
COMMENT
This investigation, based the largest community
on
sample that has been studied with in-laboratory poly¬
somnography, has shown explicitly that occult,
untreated sleep-disordered breathing is significantly
associated with elevated systolic and diastolic BPs and
with hypertension, independent of age, sex, and body
habitus (BMÎ, waist-hip ratio, neck girth, and skinfold
thickness). We consistently saw a dose-response trend:
BP increased with increasing sleep-disordered breathing
severity. Of particular public health relevance, even
mild sleep-disordered breathing (eg, AHI, 5-15) appears
to carry a slightly increased risk for elevated BP.
The use of probability sampling from a well-
defined base of employed men and women increases con¬
fidence in the validity of the study findings, as well as
the generalizability of our results to middle-aged adults.
Furthermore, the large sample size allowed precise esti¬
mates of associations and enhanced our ability to con¬
trol simultaneously for numerous potential confound¬
ing factors.
An additional important strength of our study lies
in the quality of the measurements of sleep-disordered
breathing and BP. Use of attended, laboratory-based poly¬
somnography, the diagnostic standard for assessment of
sleep-disordered breathing,28-29 ensured high-quality data

Downloaded From: http://archinte.jamanetwork.com/ by a University of California - San Diego User on 06/04/2015

 and accurate assessment of the AHI. Furthermore,
use of standardpolysomnography means that our find¬
ings are expressed with the same metrics used to de¬
scribe the spectrum of sleep-disordered breathing in sleep
medicine. A shortcoming of our analysis is the lack of
exploration of other physiological parameters of sleep-
disordered breathing. The AHI, as a measure of the fre¬
quency of apneas and hypopneas, is the most com¬
monly used measure of sleep-disordered breathing
occurrence and severity, but it is possible that variables
based on the underlying physiological effects, such as level
of oxygen desaturation and cortical arousals, better de¬
scribe severity in relation to BP. The BP measures, con¬
ducted by trained technicians, were taken under similar
conditions for all subjects and the average BPs for the to¬
tal sample are comparable with those for the adult US
population.37
This study was designed to analyze the effect of hy¬
pothesized and established factors that would be ca¬
pable of confounding the findings. Extensive investiga-
our tion with statistical modeling indicated that there was an
independent association between sleep-disordered breath¬
ing and BP. However, it is possible that an unknown risk
factor for both sleep-disordered breathing and hyperten¬
sion exists that may have biased the findings. Further¬
more, although we used variables based on commonly
used measures of potential confounding factors, it is not
possible to know with certainty that these factors were
fully accounted for. Obesity, because of its strong asso¬
ciation with both sleep-disordered breathing and BP, is
of special concern. It is unlikely that variables derived
from the simple parameters of weight, height, circum¬
ferences, and skinfolds truly capture the aspect of obe¬
sity that contributes to both sleep-disordered breathing
and elevated BP. The consistent results from using sev¬
eral body habitus variables singly and in combination,
however, is reassuring.
The major limitation of our study is the cross-
sectional nature of the data. The temporal direction of
the association cannot be discerned because the onset
dates of sleep-disordered breathing and hypertension are
not known. Research has been focused mostly on the
physiological sequence of events following apneas and
hypopneas, with sleep-disordered breathing hypoth¬
esized as the causal factor. However, the reverse path¬
way cannot be dismissed and, although little relevant re¬
search has been conducted, there is also evidence in
support of a causal role of hypertension in breathing in¬
stability during sleep. Treatment of hypertension with
ß-blockers or angiotensin-converting enzyme inhibi¬
tors has been shown to reduce the frequency of the ap¬
neas and hypopneas in patients with sleep-disordered

breathing.38 Some studies suggest that hypertension aug¬

Figure 1. Predicted increase in systolic blood pressure (BP) associated with
sleep-disordered breathing at 3 body mass index (BMI, a measure of weight in
kilograms divided by the square of the height in meters) levels. Increases are
based on linear regression after exclusion from the sample of individuals
receiving antihypertensive medication andaré adjusted forage, sex, and BMI.
ments peripheral chemoreceptor responsiveness,39 an ef¬
fect that has been shown to cause breathing instability
during sleep at high altitude.40 Evidence that one causal
direction exists, however, does not disallow the other and
the possibility of synergistic effects cannot be dis¬
missed.
Our findings in support of an independent associa¬
tion between sleep-disordered breathing and elevated BP
are in agreement with those of only a few clinic-based

Table 3. Odds Ratios for Sleep-Disordered Breathing and Hypertension: Sleep Cohort Study*
Systolic BP ==140 mm Hg
Systolic BP a140 mm Hg Diastolic BP -90 mm Hg orDiastolic BP 90 mm Hg
or the Use of or the Use of or the Use of
Antihypertensive Medication Antihypertensive Medication Antihypertensive Medication
r ~l 1 1
Independent Variable OR 95% CI OR 95% CI OR 95% CI
AHI
5vs0 1.21 1.10-1.34 1.18 1.07-1.30 1.21 1.09-1.34
15 vsO 1.78 1.32-2.38 1.64 1.22-2.21 1.75 1.28-2.40
30vs0 3.15 1.75-5.67 2.68 1.48-4.86 3.07 1.65-1.74

*N=1069. The formulas to calculate odds ratios (ORs) for each definition of hypertension associated with other apnea-hypopnea index (AHI)-body mass index
(BMI, a measure of weight in kilograms divided by the>square of the height In meters) combinations are as follows: for the logistic regression models, the OR for
hypertension defined as systolic blood pressure (BP) 140 mm Hg or the use of antihypertensive medication associated with any AHI-BMI combination can be
calculated as ORHm=el003S*<mi-3O,x<-0JX"98>lxAHI. The OR for hypertension defined as diastolic BP >90 mm Hg or the use of antihypertensive medication associated
with any AHI-BMI combination can be calculated as ORHm=eiao33,tBMI~m'<<~omi>,xA"1. The OR for hypertension defined as systolic BP > 140 mm Hg or diastolic BP
>90 mm Hg or the use of antihypertensive medication associated with any AHI-BMI combination can be calculated as ORHm=eIO037*<BMI-3O>xl-OO022>,xAHI. HTN
indicates hypertension. Body mass index centered at 30 kg/nf.

Downloaded From: http://archinte.jamanetwork.com/ by a University of California - San Diego User on 06/04/2015

 Figure 2. Predicted increase in diastolic blood pressure (BP) associated with
sleep-disordered breathing at 3 body mass index (BMI, a measure of weight in
kilograms divided by the square of the height in meters) levels. Increases are
based on linear regression after exclusion from the sample of individuals
receiving antihypertensive medication and are adjusted for age, sex, and BMI.
studies.16"18 Nearly all the other studies141519"21-23"25 show
a strong association between sleep-disordered breath¬
ing and elevated BP that becomes weaker and statisti¬
cally insignificant when BMI, age, and sex are taken into
consideration. Although these findings are generally taken
as evidence against an association of sleep-disordered
breathing with BP, méthodologie limitations may have
made the studies incapable of detecting a true but small
association.
Our analysis indicates that the association of sleep-
disordered breathing with BP is not large. To detect the
effect with statistical significance requires a sample size
that provides adequate study power, accurate measure¬
ments, and proper multivariable modeling. Many of the
clinic studies1419"21 with negative findings had small sample
sizes or used control groups of symptomatic patients who
could be considered to have mild or preclinical sleep-
disordered breathing. In view of our finding that even
mild sleep-disordered breathing is related to elevated BP,
use of these controls would underestimate the associa¬
tion. Another concern is random error in the measure¬
ment of sleep-disordered breathing, which would bias the
effect estimate toward the null. This problem may ac¬
count for insignificant findings of the previous population-
based studies, none of which used polysomnography to
measure sleep-disordered breathing. Surrogate mea¬
sures with poor or unknown validity were used in all these
studies.23"25 Finally, body habitus was controlled for in
all studies, but an interaction term has never been con¬
sidered previously. Our finding that the association be¬
tween sleep-disordered breathing and BP diminishes with
increasing BMI levels may explain negative findings from
studies with a high proportion of extremely obese pa¬
tients. In support of this explanation, positive findings
were reported from investigations in Sweden16 and Aus¬
tralia17 in which the average BMI of the patients studied
was less than 30 kg/m2.
In summary, the current controversy of whether an
association between sleep-disordered breathing and el¬
evated BP truly exists should be lessened considerably
Downloaded From: http://archinte.jamanetwork.com/ by a University of California - San Diego User on 06/04/2015
Figure 3. Odds ratios and upper 95% confidence intervals for
sleep-disordered breathing and hypertension (defined as systolic blood
pressure «s 140 mm Hg or diastolic blood pressure > 90 mm Hg or receiving
antihypertensive medication) for 3 body mass index (BMI, a measure of
weight in kilograms divided by the square of the height in meters) levels.
by our findings that demonstrate a statistically signifi¬
cant relationship, independent of known confounding fac¬
tors. Our findings are of both clinical and public health
significance.
A patient with even mild unrecognized sleep-
disordered breathing is more likely to have an elevated
BP than an otherwise similar patient without sleep ap-
nea. Recognition of sleep-disordered breathing in such
patients may be helpful in designing treatment strate¬
gies for both the elevated BP and the sleep-disordered
breathing. Of particular clinical importance, sleep-
disordered breathing is a stronger risk factor for el¬
evated BP in the nonobese compared with the obese
patient. Although obesity is a strong risk factor for sleep-
disordered breathing, there is a significant prevalence of
unrecognized sleep-disordered breathing in the non-
obese patient. Currently, suspicion for sleep-disordered
breathing is focused on obese snorers; nonobese pa¬
tients who do not present with sleep complaints are un¬
likely to be questioned on their snoring or sleeping his¬
tory. Our findings suggest that during the routine medical
examination, questions relevant to sleep-disordered
breathing should not be restricted to obese patients.
A prevalent risk factor associated with even a small
increase in BP is of public health significance. Although
the causal direction of the sleep-disordered breathing and
elevated BP association has not been established, it is pos¬
sible using our estimates to appreciate what the popula¬
tion impact would be if sleep-disordered breathing does
directly contribute to hypertension, defined as systolic
BP of 140 mm Hg or more, diastolic BP of 90 mm Hg or
more, or using antihypertensive medication. For this, we
used standard formula for attributable risk,41 in conjunc¬
tion with (1) 1990 US Census population estimates, (2)
National Health and Nutrition Examination Survey hy¬
pertension prevalence estimates,37 (3) the prevalence of
sleep-disordered breathing in middle-aged women (5%
for an AHI of 5-15 and 4% for an AHI of > 15) and men
(15% for an AHI of 5-15 and 9% for an AHI of >15),2
and (4) the relative risk estimates for hypertension as-
 sociated with AHI and BMI midpoints of the 2 AHI cat¬
egories (1.3 for the 5-15 category for both men and
women, and 2.0 and 2.1 for the >15 category for men
and women, respectively). Under the assumption that
there is a causal association between sleep-disordered
breathing and hypertension, we estimate from these data
that among adults in their most productive years, sleep-
disordered breathing would contribute to hypertension
in approximately 400 000 women and 2 million men.
Acceptedfor publication January 16, 1997.
This study was supported by grants P01HL42242 and
RR03186 from the National Institutes of Health, Bethesda,
Md.
Polysomnographic evaluation was conducted by the
Sleep Laboratory Core of the University of Wisconsin Spe¬
cialized Center of Research in Cardiopulmonary Disorders
of Sleep. We are gratefulfor the technical expertise of Steven
Weber, PhD, Tony Jacques, Linda Evans, Deborah Brown,
Leah Steinberg, MS, Andrea Damer, Kathy Kenneway, Hyun
Kim, MS, and Kathy Fluff.
Reprints: Terry Young, PhD, Department of Preven¬
Medicine, University of Wisconsin, 504 N Walnut St,
tive
Madison, Wl 53705.
REFERENCES
1. Fletcher EC. The relationship between systemic hypertension and obstructive sleep
apnea: facts and theory. Am J Med. 1995;98:118-128.
2. Young T, Palta M, Dempsey J, Skatrud J, Weber S, Badr S. The occurrence of
sleep-disordered breathing among middle-aged adults. N Engl J Med. 1993;328:
1230-1235.
3. Carlson JT, Hedner J, Sellgren J, Elam M, Wallin BG. Augmented resting sym-
pathetic activity in awake patients with obstructive sleep apnea. Chest. 1993;
103:1763-1768.
4. Carlson JT, Hedner J, Elam M, Sellgren J, Wallin BG. Depressed baroreflex sen-
sitivity in patients with obstructive sleep apnea. Am J Respir Crit Care Med. 1996;
154:1490-1496.
5. Hedner JA, Wilcox I, Laks L, Grunstein RR, Sullivan CE. A specific and potent
effect of hypoxia in patients with sleep apnea. Am Rev Respir Dis. 1992;146:
1240-1245.
6. Rodenstein DO, D'Odemont JP, Pieters T, Aubert-Tulkens G. Diurnal and noc-
turnal diuresis and natriuresis in obstructive sleep apnea. Am Rev Respir Dis.
1992;145:1367-1371.
7. Guilleminault C, Simmons FB, Motta H, et al. Obstructive sleep apnea syndrome
and tracheostomy: long-term follow-up experience. Arch Intern Med. 1981;141:
985-988.
8. Fletcher EC, Miller J, Schaff JW, Flectcher JG. Urinary catecholamines before
and after tracheostomy in patients with obstructive sleep apnea and hyperten-
sion. Sleep. 1987;10:35-44.
9. Sullivan CE, Issa FG, Ellis E, Bye T, Vela Bueno A. Treatment of cardiorespiratory
disturbances during sleep. Interdiscipl Topics Gerontol. 1987;22:47-67.
10. Wilcox I, Grunstein RR, Hedner JA, et al. Effect of nasal continuous positive air-
way pressure during sleep on 24-hour blood pressure in obstructive sleep ap-
nea. Sleep. 1993;16:539-544.
11. Guilleminault C, Tilkian A, Dement WC. The sleep apnea syndrome. Ann Rev Med.
1976;27:465-484.
12. Williams A, Houston D, Finberg S, Lam C, Kinney J, Santiage S. Sleep apnea sy-
drome and essential hypertension. Am J Cardiol. 1985;55:1019-1022.
13. Burack B, Pollack C, Borowiecke B, Weitzman E. The hypersomnia-sleep apnea syn-
drome: a reversible major cardiovascular hazard. Circulation. 1977;56:111-117.
Downloaded From: http://archinte.jamanetwork.com/ by a University of California - San Diego User on 06/04/2015
14. Hoffstein V, Mateika J. Evening-to-morning blood pressure variations in snor-
ing patients with and without sleep apnea. Chest. 1992;101:379-384.
15. Millman RP, Redline S, Carlisle C, Assaf A, Levinson P. Daytime hypertension in
obstructive sleep apnea. Chest. 1991;99:861-866.
16. Carlson J, Hedner J, Ejnell H, Peterson L. High prevalence of hypertension in
sleep apnea patients independent of obesity. Am J Respir Crit Care Med. 1994;
150:72-77.
17. Grunstein R, Wilcox I, Yang T, Gould Y, Hedner J. Snoring and sleep apnoea in
men: association with central obesity and hypertension. Int J Obes. 1993;17:
533-540.
18. Strohl K, Novak R, Singer W, et al. Insulin levels, blood pressure and sleep ap-
nea. Sleep. 1994;17:614-618.
19. Rauscher H, Popp W, Zwick H. Systemic hypertension in snorers with and with-
out sleep apnea. Chest. 1992;102:367-371.
20. Mendelson W. The relationship of sleepiness and blood pressure to respiratory
variables in obstructive sleep apnea. Chest. 1995;108:966-972.
21. Escourrou P, Jirani A, Nedelcoux H, Duroux P, Gaultier C. Systemic hyperten-
sion in sleep apnea syndrome. Chest. 1990;98:1362-1365.
22. Hla K, Young T, Bidwell T, Palta M, Skatrud J, Dempsey J. Sleep apnea and
hypertension: a population-based study. Ann Intern Med. 1994;120:
382-388.
23. Stradling JR, Crosby JH. Relation between systematic hypertension and sleep
hypoxaemia or snoring: analysis in 748 men drawn from general practice. BMJ.
1990;300:75-78.
24. Olson L, King M, Hensley M, Saunders N. A community study of snoring and
sleep disordered breathing: health outcomes. Am J Respir Crit Care Med. 1995;
152:717-720.
25. Jennum P, Sjol A. Snoring, sleep apnoea and cardiovascular risk factors: The
MONICA II Study. Int J Epidemiol. 1993;22:439-444.
26. Bresnitz E, Golberg R, Kosinski R. Epidemiology of sleep apnea. Epidemiol Rev.
1994;16:210-227.
27. Levinson P, Millman R. Causes and consequences of blood pressure alterations
in obstructive sleep apnea. Arch Intern Med. 1991;151:455-462.
28. Block A, Cohn M, Conway W, et al. Indications and standards for cardiopulmo-
nary sleep studies. Sleep. 1985;8:371-379.
29. American Thoracic Society Consensus Conference on Indications and Stan-
dards for Cardiopulmonary Sleep Studies. Indications and standards for cardio-
pulmonary sleep studies. Am Rev Respir Dis. 1989;139:559-568.
30. Rechtschaffen A, Kales AA, eds. A Manual of Standardized Terminology, Tech-
niques and Scoring System for Sleep Stages of Human Subjects. Washington,
DC: US Government Printing Office; 1968. NIH publication 204.
31. American Society of Hypertension. Recommendations for routine blood pres-
sure measurement by indirect cuff sphygmomanometry. Am J Hypertens. 1992;
5:207-209.
32. Lohman T, Roche A, Martorell R. Anthropometric Standardization Reference
Manual. Champaign, III: Human Kinetics; 1988:1-55.
33. SAS Institute Inc. SAS User's Guide: Statistics, Version 6.08. Cary, NC: SAS In-
stitute Inc; 1990.
34. SAS Institute Inc. SUDAAN User's Manual, Release 6.0. Cary, NC: SAS Institute
Inc; 1992.
35. Heitjan DF, Landis JR. Assessing secular trends in blood pressure: a multiple
imputation approach. J Am Stat Assoc. 1994;89:750-759.
36. Rothman KJ. Modern Epidemiology. Boston, Mass: Little Brown & Co Inc; 1986:
125-126.
37. Burt V, Cutler J, Higgins M, et al. Trends in the prevalence, awareness, treat-
ment and control of hypertension in the adult US population. Hypertension. 1995;
26:60-69.
38. Mayer J, Weichler U, Herres-Mayer B, Schneider H, Marx U, Peter JH. Influence
of metoprolol and cilazapril on blood pressure and on sleep apnea activity. J Car-
diovasc Pharmacol. 1990;16:952-961.
39. Trzebski A, Tafil M, Zoltowski M, Przybyiski J. Increased sensitivity of the arte-
rial chemoreceptor reflex in young men with mild hypertension. Cardiovasc Res.
1982;16:163.
40. Pack AI, Gottschalk A. Mechanisms of ventilatory periodicities. Ann Biomed Eng.
1993;21:537-544.
41. Kleinbaum DG, Kupper L, Morgenstern H. Epidemiologic Research: Principles
and Quantitative Methods. Belmont, Calif: Wadsworth; 1982.