AJH
2000;13:678 – 685
Home Blood Pressure Normalcy: The Didima
Study
George S. Stergiou, Georgia C. Thomopoulou, Irini I. Skeva, and Theodore D. Mountokalakis
To evaluate reference values of home blood
pressure (HBP) a cross-sectional community study
was conducted on 694 adult subjects (aged > 18
years) of the village Didima in southern Greece
(participation rate 76.4%). Clinic blood pressure
(CBP) was measured on two visits (triplicate
measurements, mercury sphygmomanometer) and
HBP on 3 workdays (duplicate morning and
evening measurements, oscillometric devices;
Omron HEM 705CP). After exclusion of 132
subjects (103 treated hypertensives and 29 with
incomplete data), 562 subjects were analyzed
(mean ⴞ SD aged 51.2 ⴞ 17.2 years, 42.7% men).
Average HBP (120.0 ⴞ 17.8/72.6 ⴞ 8.8 mm Hg,
systolic/diastolic) was strongly correlated (P <
.0001) with CBP (118.7 ⴞ 17.7/73.8 ⴞ 10.5 mm Hg).
Systolic CBP was 1.3 mm Hg lower than HBP (P <
.01, 95% confidence interval 0.4, 2.2), whereas
diastolic CBP was 1.2 mm Hg higher than HBP (P
< .0001, 95% confidence interval 0.6, 1.7). The
threshold of HBP normality determined using
three different approaches was 1) 139.7/83.0 mm Hg
(systolic/diastolic) using the distribution criterion
(95th percentile of the HBP distribution among 476
normotensive subjects); 2) 139.7/85.8 mm Hg using
the correspondence criterion (the percentiles of the
CBP distribution that correspond to CBP > 140/90
mm Hg were estimated, and the levels of BP that
correspond to these same percentiles on the HBP
distribution were calculated); and 3) 137.4/82.7 mm
Hg using the regression criterion (calculation of
the levels of HBP that correspond to CBP of 140/90
mm Hg using the regression equation between
HBP and CBP). Overall, the findings of the three
criteria suggest that average HBP < 137/82 mm Hg
might be considered as probably normal, > 140/86
mm Hg as probably abnormal, and within these
limits as borderline. Until mortality-based
prospective data are available, this approach might
be useful in the interpretation of HBP in clinical
practice. Am J Hypertens 2000;13:678 – 685 © 2000
American Journal of Hypertension, Ltd.
he use of self-monitoring of blood pressure
at home (HBP) is becoming increasingly
popular in clinical practice.1,2 Several medical organizations have encouraged the use of
HBP as a supplementary source of information to the
practicing physician.1,3,4
T
Very few cross-sectional population studies5–10 and
only one longitudinal study11 have attempted to define the normal range of HBP. Therefore, there is still
no agreement on what should be considered as the
upper limit of normal for HBP.2 Nevertheless, on the
basis of available evidence, the US Joint National
Received March 30, 1999. Accepted October 29, 1999.
From the Hypertension Center, Third University Department of
Medicine, Sotiria Hospital, Athens, Greece.
Address correspondence and reprint requests to George S. Ster-
giou, MD, Hypertension Centre, Third University Department of
Medicine, Sotiria Hospital, 152 Mesogion Ave., Athens 11527,
Greece.
© 2000 by the American Journal of Hypertension, Ltd.
Published by Elsevier Science, Inc.
KEY WORDS:
Home blood pressure, reference values,
hypertension, office blood pressure, blood pressure
measurement.
0895-7061/00/$20.00
PII S0895-7061(99)00266-6
AJH–JUNE 2000 –VOL. 13, NO. 6, PART 1
Committee on the Detection, Evaluation and Treatment of High Blood Pressure3 and an American Society of Hypertension Ad Hoc Panel, which presented
recommendations for the clinical use of HBP,2 suggested that HBP readings of 135/85 mm Hg or greater
should be considered elevated.
The Didima study is a cross-sectional study that
attempted to determine the threshold of normality for
HBP based on the distribution of HBP and the relation
of clinic blood pressure (CBP) with HBP in an unselected adult population.
SUBJECTS AND METHODS
Study Population This is a cross-sectional community study that was carried out in the village Didima
located in Argolida of Peloponesus in southern
Greece. All adults, 18 years of age or older, were
invited to participate. Subjects receiving antihypertensive drug treatment were excluded from the analysis.
CBP Measurement
CBP was measured on two
morning visits, 4 to 14 days apart. Measurements were
taken by a well-trained physician (GCT), who fulfilled
the British Hypertension Society Protocol criteria for
observer agreement in BP measurement,12 using a
standard mercury sphygmomanometer (Baumomanometer, W.A. Baum Co., Inc., Copiague, NY) with
bladder size 12 ⫻ 35 cm. Triplicate BP measurements
were taken per visit on the subjects’ left arm in the
sitting position, after 5 min of rest and with 1 min
between measurements (Korotkoff phase V for diastolic BP). Participants with BP measured on a single
clinic visit were excluded from the analysis. Average
CBP of the three measurements of the second clinic
visit was used in the comparison of CBP with HBP.
HBP Measurement HBP was taken by the study
subjects themselves or by their relatives at home, on 3
working days between the first and the second clinic
visit. Duplicate BP measurements were taken in the
morning (06:30 to 10:00) and in the evening (17:00 to
23:00) of each day on the participants’ left arm in the
sitting position, after 5 min of sitting rest and with 1
min between measurements.
Measurements were taken using validated fully automated oscillometric devices Omron HEM 705 CP
(Omron Healthcare GmbH, Hamburg, Germany) that
comply with the standards of the Association for the
Advancement of Medical Instrumentation and the
British Hypertension Society Protocol for the evaluation of automated and semiautomated devices.13 Cuffs
with bladder size 14 ⫻ 28 cm were used and devices
were recalibrated every 3 months during the study, or
earlier when needed. Participants with less than six
valid HBP measurements were excluded from analysis. Average HBP of the eight measurements obtained
HOME BLOOD PRESSURE NORMALCY
679
on the second and the third HBP monitoring days was
used for the comparison of HBP with CBP.
Criteria for the Evaluation of the HBP Normalcy
Three different approaches were used to determine
the threshold of normality for HBP:
1. Distribution criterion. The 95th percentile of the
HBP distribution was used for the calculation of the
upper threshold of normality for systolic and diastolic
BP among 476 subjects with CBP ⬍ 140 mm Hg systolic and/or ⬍ 90 mm Hg diastolic.9
2. Correspondence criterion. The percentiles of
clinic systolic and diastolic BP distribution that correspond to CBP systolic ⱖ 140 mm Hg and diastolic ⱖ
90 mm Hg were estimated. The levels of systolic and
diastolic BP that correspond to these same percentiles
on the HBP distribution were calculated.6,7,10
3. Regression criterion. The regression equation,
which was estimated for the linear regression line
between CBP and HBP, was used for the calculation of
the levels of HBP that correspond to the CBP threshold
of 140/90 mm Hg.8
Statistical Analysis
Results are expressed as
means ⫾ SD. One-way analysis of variance (ANOVA)
and Student’s t tests were used to estimate differences
between mean values. Bonferroni’s correction for multiple comparisons were applied where appropriate.
Ninety-five percent confidence intervals (95% CI)
were given where relevant. Pearson correlations were
used to investigate the association of HBP and CBP.
The Bland-Altman approach14 was used for the investigation of the degree of similarity of HBP with CBP. A
probability value P ⬍ .05 was considered statistically
significant.
RESULTS
Among 908 adult subjects of the study area, 694
(76.4%) participated in the study. Twenty-nine subjects were rejected because of incomplete BP data and
103 treated hypertensives were excluded. Finally, a
total of 562 participants were included in the analysis.
Mean age was 51.2 ⫾ 17.2 (SD) years, 240 (42.7%) were
men and 322 (57.3%) women.
There was a progressive decline in systolic CBP in
the repeated readings of both the initial and the second clinic visit (Table 1). The same phenomenon was
observed for diastolic CBP in the initial visit only.
Average CBP of the second visit was lower than it was
on the initial visit (mean discrepancy 4.4 ⫾ 10.4 mm
Hg, P ⬍ .0001, 95% CI 3.6, 5.3 for systolic BP and
2.1 ⫾ 6.6 mm Hg, P ⬍ .0001, 95% CI 1.6, 2.7 for
diastolic). On the second visit, 86 subjects (15.3%) had
average CBP ⱖ140/90 mm Hg.
The initial HBP readings were consistently higher
than the repeated readings taken 1 min later in all
680
AJH–JUNE 2000 –VOL. 13, NO. 6, PART 1
STERGIOU ET AL
TABLE 1. CLINIC BLOOD PRESSURE OF THE INITIAL AND THE SECOND VISIT (MEAN ⴞ SD, mm Hg)
Visit
1st
2nd
Repeated Clinic Blood Pressure Readings
Blood
Pressure
1st
2nd
Systolic
Diastolic
Systolic
Diastolic
125.0 ⫾ 20.0
76.2 ⫾ 11.1
119.8 ⫾ 18.4
73.7 ⫾ 10.7
122.7 ⫾ 19.4*
75.8 ⫾ 11.2
118.4 ⫾ 18.1*
73.8 ⫾ 10.8
3rd
121.7 ⫾ 19.1‡*
75.6 ⫾ 11.1†
117.9 ⫾ 17.8*
73.8 ⫾ 10.8
Average Clinic
Blood Pressure
123.1 ⫾ 19.1
75.9 ⫾ 10.9
118.7 ⫾ 17.7§
73.8 ⫾ 10.5§
* P ⬍ .001, † P ⬍ .01 compared with the first blood pressure reading; ‡ P ⬍ .001 compared with the second blood pressure reading; § P ⬍ .001 compared
with the corresponding blood pressure of visit 1.
morning and evening measurements along the 3 HBP
monitoring days (range of discrepancy 3 to 3.5 mm Hg
for systolic and 1.5 to 2 mm Hg for diastolic, P ⬍
.0001 for all occasions, Figure 1). Average evening
systolic HBP were marginally higher (0.8 to 1.2 mm
Hg) than the corresponding morning values (P ⫽
.05). Ninety-five percent confidence intervals excluded any difference between morning and evening
systolic HBP greater than 2.1 mm Hg. No difference in
diastolic BP was observed between morning and
evening measurements. Average systolic HBP of the
initial monitoring day was higher than it was on the
second and third day, whereas there was no such a
difference between the second and third day (Figure
1). A marginal progressive decline in average diastolic
HBP was observed from the first to the third monitoring day. As indicated by 95% CI, this decline could not
exceed 1.4 mm Hg (Figure 1).
For the evaluation of the relation of HBP with CBP,
the average of all eight measurements obtained on the
second and the third HBP monitoring days (systolic
120.0 ⫾ 17.8 mm Hg, 95% CI 118.5, 121.5 and diastolic
72.6 ⫾ 8.8, 95% CI 71.9, 73.3) was compared with the
average CBP of the three measurements of visit 2
(systolic 118.7 ⫾ 17.7 mm Hg, 95% CI 117.2, 120.2 and
diastolic 73.8 ⫾ 10.5 mm Hg, 95% CI 72.9, 74.6). Average systolic CBP was lower than average HBP (mean
discrepancy 1.3 ⫾ 10.9 mm Hg, P ⬍ .01, 95% CI 0.4,
2.2), whereas diastolic CBP was higher than HBP
(mean discrepancy 1.2 ⫾ 7.1 mm Hg, P ⬍ .0001, 95%
CI 0.6, 1.7). The distributions of systolic and diastolic
HBP in comparison to CBP are presented in Figure 2.
HBP was strongly related to CBP with correlation
coefficient 0.81 and 0.75 for systolic and diastolic BP,
respectively (P ⬍ .0001 for both) (Figure 3). The
discrepancies for Bland–Altman technique between
HBP and CBP are given in Figure 4. The mean difference between the BP measures is presented with the
limits of agreement (⫾ 2 SD) within which 95% of the
differences are expected to lie.
FIGURE 1. Morning and afternoon home blood pressure (BP) readings. Numbers in boxes represent average home BP of each day.
Arrows indicate between days differences in average home BP. P ⬍ .0001 compared with the corresponding first BP reading taken 1 min
before.
AJH–JUNE 2000 –VOL. 13, NO. 6, PART 1
HOME BLOOD PRESSURE NORMALCY
681
FIGURE 2. Distribution of home and clinic blood pressures.
A progressive increase in both systolic CBP and
HBP was observed along the age bands, with the
oldest subjects having the highest levels (Table 2). For
diastolic BP, again there was a progressive increase up
to the third quartile of age, but a decline was observed
in the oldest subgroup of subjects for both CBP and
HBP.
The thresholds of normality for HBP estimated using three different approaches are presented in Table
3. Using the correspondence criterion the CBP threshold of 140/90 mm Hg corresponded to the 87.4th and
FIGURE 3. Relation of home with clinic blood pressure.
93.5th percentile for systolic and diastolic BP, respectively. The regression equations for the calculation of
HBP from CBP were: HBP ⫽ 23.3 ⫹ 0.815 CBP (P ⬍
.0001) for systolic BP, and HBP ⫽ 26.7 ⫹ 0.622 CBP
(P ⬍ .0001) for diastolic BP (Figure 3, Table 3). The
upper 95% CI around the regression line calculated for
individuals9 were at 151.7/100.7 mm Hg (systolic/
diastolic). When subjects were divided in normotensives and hypertensives the slopes of the regression
lines between CBP/HBP was different in the two
groups for diastolic (P ⬍ .001) but not for systolic BP.
682
AJH–JUNE 2000 –VOL. 13, NO. 6, PART 1
STERGIOU ET AL
FIGURE 4. Discrepancies for Bland-Altman technique between clinic and home blood pressure (BP).
DISCUSSION
In clinical practice, for the application of HBP monitoring in decision making in hypertension, thresholds of
normality are urgently needed. Because only limited
data are available, the threshold of normality for HBP
remains largely unknown.2 The Didima study provides
the first epidemiologic data on the distribution of HBP in
Greece and is one of the few large studies with HBP
monitoring in unselected populations.5–11
In line with previous population studies that have
attempted to define normal values for HBP,6 –9 treated
hypertensives were excluded in this study to avoid the
confounding effect of treatment on BP. Nevertheless,
this represents a selection bias, because treated subjects might have altered the distributions of both CBP
and HBP, had they been included in the study in the
pretreatment state. A fully automated validated oscillometric device13 was preferred for the measurement
of HBP to eliminate the observer bias. We have recently shown that HBP monitoring using this device is
equally reliable and requires less training than with
calibrated aneroid devices.15
There are important differences between this and
previous cross-sectional population studies. In the
present study, BP of the second clinic visit, which gave
lower values than the initial visit, were compared with
HBP, whereas CBP of the initial visit were used in
other studies.6,8 Furthermore, in a previous study,
HBP of only one set of morning and evening selfmeasurements was used,8 whereas we discarded HBP
of the initial monitoring day because they gave the
highest systolic BP. As we have recently proposed, we
used the average BP of the second and third home
monitoring day as a reliable estimate of the level of
HBP.16 Other investigators have also preferred to exclude from the analysis HBP of the initial monitoring
day.17 In another previous study, single self-measurements were obtained,6 whereas several studies have
shown that repeated home measurements give lower
BP.7,10,16 Finally, in another population study, HBP
were taken by visiting nurses that probably represents
a different view of HBP.9
The levels of average HBP in this study were close
to those of previous population studies.5– 8,10,11 Nev-
TABLE 2. AVERAGE CLINIC AND HOME BLOOD PRESSURE IN QUARTILES OF AGE
(MEAN ⴞ SD, mm Hg)
Clinic Blood Pressure
Home Blood Pressure
Age
(yr)
No. of
Subjects
Systolic*
Diastolic*
Systolic*
Diastolic*
18–37
38–52
53–64
⬎64
143
145
131
143
109.1 ⫾ 14.4
115.7 ⫾ 14.4
124.1 ⫾ 17.0
126.4 ⫾ 19.1
69.0 ⫾ 11.5
76.9 ⫾ 10.2
77.1 ⫾ 8.9
72.3 ⫾ 9.1††
109.3 ⫾ 15.6
115.5 ⫾ 14.2
125.3 ⫾ 15.1
130.5 ⫾ 17.7†‡‡
67.0 ⫾ 7.9‡
73.3 ⫾ 8.7‡‡
76.4 ⫾ 7.1
73.9 ⫾ 8.5†‡
* P ⬍ .001 for changes in blood pressure along the age bands (ANOVA); † P ⬍ .05, †† P ⬍ .0001 compared with the age group of 53– 64 years; ‡ P ⬍
.01, ‡‡ P ⬍ .001 compared with the corresponding clinic blood pressure.
AJH–JUNE 2000 –VOL. 13, NO. 6, PART 1
HOME BLOOD PRESSURE NORMALCY
TABLE 3. THRESHOLDS OF NORMALITY FOR
HOME BLOOD PRESSURE ESTIMATED USING
THREE DIFFERENT APPROACHES
(SYSTOLIC/DIASTOLIC, mm Hg)
All
Men
Women
Distribution criterion5,6,9* 139.7/83.0 139.7/83.8 139.6/82.1
Correspondence
criterion7*
139.7/85.8 140.8/86.0 138.2/85.1
Regression criterion8*
137.4/82.7 138.5/82.6 136.1/82.6
* Description of criteria in “subjects and methods: criteria for the estimation of home blood pressure normalcy.”
ertheless, in contrast to trends indicated in most of the
previous reports, there was a striking similarity between the distributions of CBP and HBP with only
small differences in their mean values. It should be
noted, however, that the wide distribution of data
points around the linear regression line (Figure 3) and
the wide limits of agreement between CBP and HBP
(Figure 4) suggest a significant lack of agreement.
Average systolic HBP was marginally higher (1.3 mm
Hg) than CBP, as was found in only one of the previous population studies,5 whereas an inverse relation
with differences ranging from 7 to 9 mm Hg was
reported in the other studies.5,7,8,10 Diastolic HBP was
consistently lower (3 to 7 mm Hg) than CBP in previous studies,5– 8,10 whereas little difference was found
in this study (1.2 mm Hg). The small CBP–HBP differences in this study might be attributed to the rejection of measurements of the initial clinic visit, which
are shown to give the highest and most unstable BP
(Table 1).16 It is important to note that in this study the
association of HBP with CBP (r ⫽ 0.81/0.75 for systolic/diastolic; Figure 3) was closer than it was in
previous reports (0.57/0.536; 0.60/0.505; 0.73/0.648;
0.76/0.7710; 0.76/0.75,7) probably because of the careful selection of both CBP and HBP. Finally, as it was
reported in previous studies, increasing age was associated with similar changes in CBP and HBP for both
systolic and diastolic BP (Table 2).6,7
In the absence of definitive studies that relate HBP
levels with risk, several mathematical approaches
have been used for the evaluation of the upper limit of
normality for HBP from cross-sectional studies in population samples.5–10 The methods used are either HBP
and CBP dependent (correspondence and regression
criteria) or solely HBP dependent (distributional criteria).
Distributional Criteria It is generally accepted that
for tightly regulated and normally distributed variables such as serum sodium, distributional criteria are
suitable for the estimation of the cutoff points. However, the widely used criterion of the mean ⫹ 2 SD5,6
often gives higher levels than the conventional values
683
used for CBP.8,9 In the present study, even when the
analysis was artificially restricted to “clinic normotensives,” mean HBP ⫹ 2 SD was at 144.1/85.8 mm Hg
(systolic/diastolic). Moreover, because the distribution curve for HBP is broad and skewed (Figure 2),
rather than using the mean ⫹ 2 SD method the 95th
percentile was preferred as this is a distribution-free
method.9 It should be mentioned, however, that the
use of any of the distributional criteria such as the
mean ⫹ 1 SD or ⫹ 2 SD and the 90th, 95th, or 99th
percentile might be regarded arbitrary.8
The use of distributional criteria has also been criticized for underestimating the prevalence of hypertension when applied to all untreated subjects of a population sample.6,18 This can be avoided by restricting
the analysis to clinic normotensives.9 However, because the clinical justification for performing out of
clinic BP monitoring is the belief that CBP is unreliable, the selection of subjects on the basis of normal
CBP might be regarded inappropriate.18
Regression Criteria The close association that was
found between CBP and HBP (Figure 3) suggests that
the regression equation may be used for the estimation
of the levels of HBP that correspond to the conventional 140/90 mm Hg threshold of normality for CBP.8
This procedure has the advantage that the threshold
for HBP is calculated by referring to the well-established and universally accepted threshold for CBP.8
Moreover, no BP limit is needed for the inclusion of
subjects in the reference population and both normotensives and untreated hypertensives can be included.18
The disadvantage is that this method is based on the
imperfect relationship between CBP and risk18 and on
the inconsistent CBP/HBP correlation (Figure 3).6 It is
clear that any mathematical attempt to predict HBP
levels from CBP will suffer from the substantial
amount of variability that characterized the relation of
the two BP measures (Figure 4). On the other hand,
because the CBP–HBP difference increases at higher
CBP,6 the slope of the CBP/HBP regression line was
not the same in the hypertensive group compared to
normotensives, suggesting that when these groups of
subjects are pooled the regression method for estimating the HBP normalcy is biased.9 Finally, because CBP
is subject to large variation, the slope of the CBP/HBP
association is probably underestimated (regression dilution bias).
It has been suggested that rather than using 95% CI
around the regression line,8 which actually predict
populations means, 95% CI for individuals should be
used for the estimation of the upper threshold for
HBP.9 However, the HBP threshold provided by the
upper 95% CI calculated for individuals was higher
684
AJH–JUNE 2000 –VOL. 13, NO. 6, PART 1
STERGIOU ET AL
TABLE 4. CRITERIA USED TO DETERMINE THE UPPER THRESHOLD OF NORMALITY FOR HBP IN
CROSS-SECTIONAL POPULATION STUDIES
Participants
Study Name
Criteria for Home BP Normalcy
Number
Untreated
Hypertensives
Excluded
Mean ⴞ
2 SD
95th
Percentile
Correspondence
Criterion
Regression
Criterion
HBP Threshold
(systolic/diastolic)
(mm Hg)
608
729
390
503
1438
871
562
No
Yes
No
No
No
No
No*
⫹
—
—
—
—
⫹
—
—
⫹
—
—
—
⫹
⫹
—
—
⫹
⫹
—
⫹
⫹
—
—
—
—
⫹
—
⫹
139/89
136/86
147/86
133/86
132/81
137/86
140/86
5
Tecumseh
Belgian9
French Society7
Dubendorf10
PAMELA8
Osahama6
Didima
* Untreated hypertensives were excluded when the 95th percentile criterion was applied.
than the conventional CBP level of 140/90 mm Hg
(systolic/diastolic).
Correspondence Criteria The rational for applying
the correspondence criteria is that the estimated HBP
normalcy is established from the prognostic value of
CBP measurement using the same subjects at the same
level of distribution.7,10 Moreover preselection of subjects by BP level is avoided.7 The disadvantages are
again attributable to the imperfect relationship between CBP and risk18 and the inconsistent CBP/HBP
correlation6 (Figures 3 and 4). Finally, the correspondence criterion assumes that the same percentage of
subjects is classified as hypertensives either using CBP
or HBP, ignoring thereby the existence of the white
coat effect in the subjects with high CBP.
Because all the methods that have been proposed
for the evaluation of the HBP normalcy have important limitations, each of the published studies has used
a different approach (Table 4). Therefore, we preferred
to combine results obtained using all three criteria and
to propose limits of uncertainty, which reflect the discrepancy between these criteria (Table 5). Home BP
above these limits might be regarded as probably
abnormal and below these limits probably normal.
Previous reports, where different criteria were used,
TABLE 5. SUGGESTED LIMITS OF NORMALITY
FOR HOME BLOOD PRESSURE (mm Hg)
Blood
Pressure
Systolic
Diastolic
Probably
Normal Borderline
All
Men
Women
All
Men
Women
⬍137
⬍138
⬍136
⬍82
⬍82
⬍82
137–140
138–141
136–140
82–86
82–86
82–85
Probably
Abnormal
⬎140
⬎141
⬎140
⬎86
⬎86
⬎85
estimated the threshold for systolic HBP between 132
and 147 mm Hg and for diastolic between 81 and 89
mm Hg (Table 4).5–11 Men had consistently higher
HBP than women by 2 to 11/1 to 7 mm Hg (systolic/
diastolic.5–7 These data, together with the present findings support the 135/85 mm Hg HBP threshold that
has been recently proposed by the US Joint National
Committee on the Detection, Evaluation and Treatment of High Blood Pressure3 and an American Society of Hypertension Ad Hoc Panel.2
It should be remembered that the prognostic relevance of reference values provided by cross-sectional
studies is unknown. Prospective population studies
that investigate the association of HBP directly with
long-term outcome to define the normal range of HBP
are still awaited. The only study that proposed reference values for HBP based on prognostic criteria in a
rural Japanese community estimated the threshold of
HBP normality at 137/84 mm Hg.11 Interestingly, this
was very close to the threshold previously estimated
by the same investigators using distributional criteria6
and lies within the limits of uncertainty proposed in
the present study. Therefore, until more mortalitybased prospective data are available, reference values
obtained from cross-sectional population studies
might be useful in the application of HBP as a supplementary source of information in clinical practice.
ACKNOWLEDGMENTS
We thank Omron Healthcare, Germany GmbH for providing 15 Omron HEM 705CP devices and G. Leoussis SA,
Athens, Greece, for technical support and regular calibration
of these devices throughout the study.
REFERENCES
1.
World Hypertension League: Self measurement of
blood pressure: a statement by the World Hypertension
League. J Hypertens 1988;6:257–261.
2.
Pickering T, for an American Society of Hypertension
AJH–JUNE 2000 –VOL. 13, NO. 6, PART 1
4.
685
11.
Joint National Committee on Detection, Evaluation and
Treatment of High Blood Pressure: The sixth report of
the Joint National Committee on the detection, evaluation and treatment of high blood pressure (JNC VI).
Arch Intern Med 1997;24:2413–2446.
Tsuji I, Imai Y, Nagai K, et al: Proposal of reference
values for home blood pressure measurement: prognostic criteria based on a prospective observation of the
general population in Ohasama, Japan. Am J Hypertens 1997;10:409 – 418.
12.
WHO/ISH: Guidelines for the management of mild
hypertension: memorandum from a World Health Organisation/International Society of Hypertension
meeting. J Hypertens 1993;11:905–918.
O’Brien E, Petrie J, Littler W, et al: The British Hypertension Society Protocol for the evaluation of blood
pressure measuring devices. J Hypertens 1993;11(suppl
2):S43–S63.
13.
O’Brien E, Mee F, Atkins N, Thomas M: Evaluation of
three devices for self-measurement of blood pressure
according to the revised British Hypertension Society
Protocol: the Omron HEM-705CP, Philips HP5332 and
Nissei DS-175. Blood Pressure Monitoring 1996;1:55–
61.
14.
Bland JM, Altman DG: Statistical methods for assessing
agreement between two methods of clinical measurement. Lancet 1986;i:307–310.
15.
Stergiou GS, Voutsa AV, Achimastos AD, Mountokalakis TD: Self home-monitoring of blood pressure:
is fully automated oscillometric technique as good as
conventional stethoscopic technique? Am J Hypertens
1997;10:428 – 433.
Ad Hoc Panel: Recommendations for the use of home
(self) and ambulatory blood pressure monitoring. Am J
Hypertens 1995;9:1–11.
3.
HOME BLOOD PRESSURE NORMALCY
5.
Mejia AD, Julius S, Jones KA, et al: The Tecumseh blood
pressure study. Arch Intern Med 1990;150:1209 –1213.
6.
Imai Y, Satoh H, Nagai K, et al: Characteristics of a
community based distribution of home blood pressure
in Osahama in northern Japan. J Hypertens 1993;11:
1441–1449.
7.
De Gaudemaris R, Chau NP, Mallion JM, for the
Groupe de la Mesure, French Society of Hypertension:
Home blood pressure: variability, comparison with office reading and proposal for reference values. J Hypertens 1994;12:831– 838.
8.
Mancia G, Sega R, Bravi C, et al: Ambulatory blood
pressure normality: results from the PAMELA study.
J Hypertens 1995;13:1377–1390.
16.
9.
Staessen JA, Bieniaszewski L, O’Brien ET, et al: An
epidemiological approach to ambulatory blood pressure monitoring: the Belgian Population study. Blood
Press Monit 1996;1:13–26.
Stergiou GS, Skeva II, Zourbaki AS, Mountokalakis TD:
Self monitoring of blood pressure at home: how many
measurements are needed? J Hypertens 1998;16:725–
731.
17.
Chatellier G, Dutrey-Dupagne C, Vaur L, et al: Home
self blood pressure measurement in general practice:
the SMART study. Am J Hypertens 1996;9:644 – 652.
18.
Pickering TG: The ninth Sir George Pickering memorial
lecture: ambulatory monitoring and the definition of
hypertension. J Hypertens 1992;10:401– 409.
10.
Weisser B, Grune S, Burger R, et al: The Dubendorf
Study: a population-based investigation on normal values of blood pressure self-measurement. J Hum Hypertens 1994;8:227–231.