assessment of blood pressure in patients with type 2 diabetes: comparison between home blood...
TRANSCRIPT
Assessment of blood pressure in patients with Type 2diabetes: comparison between home blood pressuremonitoring, clinic blood pressure measurement and24-h ambulatory blood pressure monitoring
M. G. Masding, J. R. Jones, E. Bartley and D. D. Sandeman
Abstract
Aims To compare a home blood pressure (BP) monitoring device and clinic
BP measurement with 24-h ambulatory BP monitoring in patients with Type 2
diabetes mellitus (DM).
Methods Fifty-®ve patients with type 2 DM had BP measured at three
consecutive visits to the DM clinic by nurses using a stethoscope and mercury
sphygmomanometer (CBP). Twenty-four-hour ambulatory BP was measured
using a Spacelabs 90207 automatic cuff-oscillometric device (ABPM). Subjects
were then instructed in how to use a Boots HEM 732B semiautomatic cuff-
oscillometric home BP monitoring device and measured BP at home on three
speci®ed occasions on each of 4 consecutive days at varying times (HBPM).
Results Correlations between HBPM and ABPM were r = 0.88, P < 0.001
for systolic BP and r = 0.76, P < 0.001 for diastolic BP, with correlations
between CBP and ABPM being systolic r = 0.59, P < 0.001, diastolic r = 0.47,
P < 0.001. HBPM agreed with ABPM more closely compared with CBP (CBP
+10.9/+3.8 (95% con®dence intervals (CI) 6.9, 14.8/1.6, 6.1) vs. HBPM +8.2/
+3.7 (95% CI 6.0, 10.3/2.0, 5.4)). The sensitivity, speci®city and positive
predictive value of HBPM in detecting hypertension were 100%, 79% and
90%, respectively, compared with CBP (85%, 46% and 58%, respectively).
Conclusions In patients with Type 2 DM, home BP monitoring is superior to
clinic BP measurement, when compared with 24-h ambulatory BP, and allows
better detection of hypertension. It would be a rational addition to the annual
review process.
Diabet. Med. 18, 431±437 (2001)
Keywords type 2 diabetes, blood pressure, blood pressure measurement
Abbreviations BP, blood pressure; DM, diabetes mellitus; HBPM, home blood
pressure monitoring; CBP, clinic blood pressure measurement; ABPM, 24-h
ambulatory blood pressure monitoring
Introduction
The pivotal role of hypertension in the development of
complications of diabetes was highlighted by the United
Kingdom Prospective Diabetes Study (UKPDS) [1].
Correspondence to: Dr Mike Masding, Research Registrar, The Endocrine
Unit, F Level, West Wing, Southampton General Hospital, Tremona Road,
Southampton SO16 6YD, UK. E-mail: [email protected]
Department of Diabetes and Endocrinology,
Southampton University Hospitals NHS Trust,
Southampton, UK
Accepted 3 February 2001
ã 2001 Diabetes UK. Diabetic Medicine, 18, 431±437 431
Knowledge of true blood pressure (BP) is therefore critical
to the care of patients with Type 2 diabetes (DM). In
clinical practice performing a resting BP during the routine
visit makes an estimate of true BP. Whilst BP measured by
this method is an important risk factor for cardiovascular
disease (CVD) in populations, its predictive value in
individual patients is poor [2]. The problem is greater for
those caring for patients with DM. The UKPDS has set
tighter goals for treatment (a target BP of 140/80 mmHg),
thus any `white-coat' effect will push patients with
normotension into the treatment group. The routine
practice of bringing patients back for repeat BP screening
and then a further clinic appointment for decision making
is inef®cient and there is need for an alternative strategy.
The current gold standard for measuring 24-h BP is
ambulatory monitoring (ABPM) over 24 h. This, however,
is expensive and cannot be offered to a large population
group such as patients with Type 2 DM. Previous studies
[3±6] in non-diabetic populations have been inconclusive
as to whether patient BP self-monitoring at home (HBPM)
is an effective strategy. In this study we have examined the
potential role of patient HBPM prior to the annual review
clinic, comparing it with ABPM and clinic BP (CBP)
assessment.
Patients and methods
Ethical approval was obtained from the local ethics committee.
Consecutive patients with Type 2 DM (regardless of treatment)
aged < 75 years and attending the routine diabetes clinic were
invited to join the study. Type 2 DM was de®ned by the
requirement for treatment with either diet or oral hypoglycae-
mics for at least 1 year after diagnosis. Subjects were only
excluded if they were unable or unwilling to give consent to enter
the study. During the study period no changes were made to
their treatment for either diabetes or hypertension.
During the clinic visit, their BP was measured by the clinic
nurse, which is the usual practice at our clinic and also at the
General Practitioners who participate in the local shared care
scheme (nurses were instructed to measure BP to the nearest
2 mmHg using a mercury sphygmomanometer with an appro-
priately sized cuff and stethoscope, after the patient has been
seated for 5 min). The CBP measurement for each patient is the
mean of three measurements taken over 6 months at different
visits to the diabetes clinic at both the hospital and in general
practice. Whilst the nurses were briefed prior to the study on the
importance of hypertension and the need for accurate results,
they were not informed of the study itself.
After recruitment, subjects were ®tted with a previously
validated [7] Spacelabs 90207 automatic cuff-oscillometric
device to assess ABPM. The BP for each patient was measured at
half-hourly intervals during the daytime, and hourly during the
night. The mean BP reported is the average of all BP readings
throughout the 24 h, whilst the daytime and night-time values
are the means of readings between 06.00 h and 17.00, and 17.00
and 06.00 h, respectively.
Following the completion of 24-h ambulatory monitoring,
subjects were instructed how to use a Boots HEM 732B
semiautomatic cuff-oscillometric home BP monitoring device,
with an appropriately sized cuff. These devices are accurate to
6 4 mmHg (Boots PLC, personal correspondence). We valid-
ated the devices by measuring BP in ®ve healthy subjects using
the Boots HEM 732B device at the same time as the Spacelabs
90207 machine, to exclude systematic bias in the study. BP was
measured in the same and different arms on 10 different
occasions on each healthy subject to assess whether using the
right or left arm would affect the study. The bias for using the
HBPM and ABPM devices at the same time was 2.0/
+0.1 mmHg, whilst for different arms the bias was +1.6/
+2.7 mmHg, indicating there was an acceptable agreement
between the devices.
The subjects were instructed to check their BP at three speci®c
times on each of 4 consecutive days; these times were spread
throughout the day (08.00, 13.00, 17.00, 20.00, 23.00 h)
according to a prede®ned protocol. The mean BP, daytime and
night-time BP were calculated from these values (mean, all
values included: daytime, 08.00, 13.00, 17.00 h; night-time,
20.00, 23.00 h).
We prede®ned ranges for BP differences which would be
acceptable in clinical practice, at 610/5 mmHg. Using the
recently published British Hypertension Society guidelines for
hypertension management [8], we used the recommended audit
standard as a `cut-off' for hypertension (< 140/85 for CBP,
< 140/80 for HBPM and ABPM) and the sensitivity and
speci®city of CBP and mean HBPM values for detecting
hypertension were calculated. It has previously been estimated
that night-time BP should fall by 20% [9]. In order to assess
whether HBPM could be used to detect loss of night-time BP
variation, we compared the HBPM reading at 13.00 h with
HBPM at 23.00 h over the 3 days, looking for a fall of 20/
10 mmHg. Mean ABPM between 00.00 and 06.00 was then
assessed looking for a fall in BP of 20/10 mmHg, compared with
the mean day-time (09.00±17.00) ABPM. In both HBPM and
ABPM, if there was no such fall, the patients were classed as
`non-dippers'. Using measurement of ABPM as the gold stand-
ard for detecting `non-dippers', we calculated the sensitivity and
speci®city of comparing HBPM at 13.00 and 23.00 in detecting
these patients. The ABPM and HBPM daytime:night-time BP
ratios (ABPM: mean daytime BP/mean night-time BP; HBPM:
BP at 13.00/BP at 23.00) were also compared. HBPM measure-
ments on the days 1 and 4 of the study were compared to assess
whether there was a stress effect of measuring BP at home on the
®rst day.
Statistical analysis
CBP and HBPM values were compared with the `gold standard'
of ABPM using Bland and Altman's method [10], and using
correlation coef®cients obtained by the use of Statistical Package
for Social Sciences (SPSS) for Windows version 6.0. We took the
modulus of the difference between the CBP and ABPM values
and contrasted it with the corresponding differences between
HBPM and ABPM (for systolic and diastolic BP). Mean
differences and 95% CIs were calculated from a paired t-test
432 Blood pressure assessment in Type 2 diabetes · M. G. Masding et al.
ã 2001 Diabetes UK. Diabetic Medicine, 18, 431±437
procedure and P-values from a Wilcoxon signed ranks test from
SPSS.
Results
Patient characteristics
Fifty-eight patients were recruited into the studyÐthree
subjects dropped out (one because the home BP monitor
cuff was too narrow, one who refused to monitor BP at
home and one who could not operate the home BP
monitor). The characteristics of the 55 subjects included
in the analysis and their treatments for hypertension and
Type 2 diabetes are shown in Table 1.
Blood pressure readings
Mean (standard deviation (SD)) BP as assessed by ABPM
was 141/81 (19/11). The estimate of BP from CBP was 152/
85 (20/11) and from HBPM 149/84 (16/9). With respect to
systolic BP, the modulus of the differences between CBP
and ABPM were on average 4.5 mmHg (95% CI 1.4, 7.7)
greater than the modulus of the differences between HBPM
and ABPM (Wilcoxon signed rank test, P = 0.012). The
corresponding contrast for diastolic BP measurement
showed the differences between CBP and HBPM were
not so marked, the difference between CBP and ABPM
being on average 1.7 mmHg (95% CI 0.1, 3.4) greater than
the difference between HBPM and ABPM (Wilcoxon
signed rank test, P = 0.108). Correlations between HBPM
and ABPM were r = 0.88 (P < 0.001) for systolic BP and
r = 0.76 (P < 0.001) for diastolic BP, with correlations
between CBP and ABPM being systolic r = 0.59
(P < 0.001), diastolic r = 0.47 (P < 0.001).
However, Bland and Altman's method [10] makes a
more appropriate statistical assessment. The data were
examined for the bias, comparing the individual results for
each subject as obtained by the three methods. This
assumes the ABPM to be the true reading and analyses
the spread of individual differences between this result,
CBP and HBPM result (Figs 1 and 2). The mean differences
(bias) between CBP and ABPM were +10.9/+3.8, between
HBPM and ABPM +8.2/+3.7. The 95% Bland and Altman
range for the difference between CBP and ABPM in systolic
BP was ±18.6 to 40.4. Thus, 95% of the time, differences
between CBP and ABPM for systolic BP will lie between
18.6 mmHg below, and 40.4 mmHg above the true BP as
de®ned by ABPM. This range would not be clinically
acceptable. The 95% Bland and Altman ranges for the
difference between HBPM and ABPM are better
Difference from ABPM
Mean
(95% CI)
Bland and Altman precision and 95% range for differences
Upper estimated
value (95%CI)
Lower estimated
value (95% CI)
CBP±ABPM
Sys +10.9 ±18.6 (±25.5, ±11.7) +40.4 (33.5, 47.3)
(6.9, 14.8)
Dias +3.8 ±20.8 (±24.8, ±16.8) +13.1 (9.2, 17.1)
(±6.1, ±1.6)
HBPM±ABPM
Sys +8.2 ±7.8 (±11.6, ±4.0) +24.1 (20.4, 27.9)
(6.0, 10.3)
Dias +3.7 ±16.1 (±19.0, ±13.2) +8.7 (5.8, 11.6)
(±5.4, ±2.0)
Table 2 Differences for clinical blood
pressure measurement (CBP) and home blood
pressure monitoring (HBPM) when compared
with ambulatory blood pressure monitoring
(ABPM) (all values in mmHg)
Table 1 Subject characteristics; details of anti-hypertensive and
diabetic treatments; details of proteinuria (all values as number
(percentage, unless otherwise stated))
Sex: Male 30 (54%)
Female 25 (46%)
Age (years)Ðmean (min±max) 58 (32±72)
Anti-hypertensive treatment:
None 25 (45%)
Single anti-hypertensive agent 17 (31%)
Combination of two agents 10 (18%)
Combination of three agents 3 (5%)
Type 2 diabetes treatment
Diet alone 6 (11%)
Oral hypoglycaemics alone 19 (35%)
Insulin alone 27 (49%)
Insulin + oral hypoglycaemics 3 (5%)
Proteinuria
Dipstick positive 13 (24%)
Original article 433
ã 2001 Diabetes UK. Diabetic Medicine, 18, 431±437
(±7.8, +24.1). The precision with which the lower and
upper limits of the 95% range are estimated are shown in
Table 2.
There was no difference in HBPM results between days 1
and 4.
Hypertension
Clinically it is important to distinguish patients who will
bene®t from intervention. We de®ned hypertension in our
patient group as per the 1999 British Hypertension
Guidelines [8]. HBPM monitoring showed 100% sensitiv-
ity and 79% speci®city in contrast to CBP (85% sensitivity
and 46% speci®city, i.e. by this measurement 54% of our
subjects had `white-coat hypertension' and would have
been unnecessarily treated). This results in a 90% positive
predictive value for HBPM, compared with 58% for CBP
measurement.
Dipper status
CBP is unable to assess dipper status. HBPM showed
good sensitivity for detecting non-dipper status by
comparing HBPM value at 13.00 and 23.00, but had
very poor speci®city: the sensitivity was 95%, speci®-
city 0% and positive predictive value 78%. The
daytime:night-time ratios for ABPM and HBPM were
signi®cantly different (ABPM mean (SD): systolic 1.03
(0.06), diastolic 1.07 (0.07); HBPM: systolic 1.00
(0.06), diastolic 1.02 (0.06), P < 0.001).
Figure 1 Comparisons between measurements of systolic BP. (a) Clinical blood pressure measurement (CBP) and ambulatory blood pressure
monitoring (ABPM). (b) Home blood pressure monitoring (HBPM) and ABPM.
434 Blood pressure assessment in Type 2 diabetes · M. G. Masding et al.
ã 2001 Diabetes UK. Diabetic Medicine, 18, 431±437
Discussion
Knowledge of the BP of patients with Type 2 DM is central
to their care; however, obtaining reliable estimates of BP
can be a frustrating experience for clinicians and patients
alike. White-coat hypertension is so prevalent, and the
goals of control post-UKPDS [1] so tight, that BP measured
at a hospital clinic visit rarely informs clinical management
but condemns the patient to multiple visits to the clinic or
the general practitioner. If the clinic monitors the ef®cacy
of treatment, the problem is compounded.
The aim of our study was to evaluate HBPM using a
standard cheap BP monitor available on the high street to
see if this is a practical alternative strategy. This could then
be incorporated into the annual review process.
Consecutive patients were recruited from the routine
diabetes clinic of a large city, so that the results of this
study could apply to most UK diabetic populations. HBPM
was explained to them by the research nurse, adding
approximately 5±10 min to the time of the clinic visit, a
period easily comparable to the time taken to obtain a
standard BP measurement. Our patients generally found
little dif®culty in using the BP monitor (only one out of 58
was unable to operate the monitor). If this became routine
management, it is likely that the time required educating
the patients would be reduced, and the accuracy of the
results may improve.
Twenty-four-hour ABPM was used as a gold standard,
and for the purposes of the study mean BP by this method
was considered the true BP. HBPM was conducted on the
next 4 days. Biological variation plus reproducibility of the
method would result in differences between mean BP in the
same patient even if ABPM were repeated on a consecutive
day. Previous studies of the reproducibility of measure-
ments by the Spacelabs 90207 system have shown standard
deviations of the differences (SDD), a measure of reprodu-
Figure 2 Comparisons between measurements of diastolic BP. (a) Clinical blood pressure measurement (CBP) and ambulatory blood pressure
monitoring (ABPM). (b) Home blood pressure monitoring (HBPM) and ABPM.
Original article 435
ã 2001 Diabetes UK. Diabetic Medicine, 18, 431±437
cibility, of 6±8 mmHg for systolic BP and 4±6 mmHg for
diastolic BP [11,12]. The bias (i.e. mean of individual
differences) of HBPM compared with ABPM is very similar
to these ®gures, which suggests that the reproducibility
differences in ABPM could account for some of the
difference between HBPM and ABPM. However, the bias
of CBP compared with HBPM is well outside the
reproducibility measurements for ABPMÐthe difference
between CBP and ABPM could not be accounted for by
reproducibility differences in ABPM.
In our study the mean difference between HBPM and
ABPM was +8.2/+3.7 mmHgÐthis is more accurate than
CBP, which when compared with ABPM shows a mean
difference of +10.9/+3.8 mmHg. The bias for HBPM fell
within our predetermined limits of 6 10/5 mmHg, whilst
that of CBP did not. However, whilst the 95% Bland and
Altman range for HBPM with ABPM does not lie within
this prede®ned range, this could be partly explained by
day-to-day physiological BP variation and the reproduci-
bility of ABPM. The 95% Bland and Altman range with
ABPM is much tighter for HBPM monitoring compared
with CBP measurement, and the con®dence intervals
around these limits do not overlap for systolic BP
measurement, suggesting that there is a difference between
HBPM and CBP measurement, with HBPM being super-
ior. We also tested the modulus of the differences between
CBP and ABPM, and HBPM and ABPM, and the results of
these tests suggest that HBPM was signi®cantly closer to
ABPM than CBP, making HBPM a superior measuring
tool for BP.
Despite HBPM failing to satisfy our prede®ned limits of
agreement, its use would dramatically improve clinical
practice. HBPM missed no patients with hypertension
(100% sensitivity), compared with 15% missed by CBP
measurement (85% sensitivity). HBPM falsely identi®ed
21% of normotensive patients as hypertensive (79%
speci®city)Ðhowever, 54% of normotensive patients
were identi®ed as hypertensive by CBP measurement
(46% speci®city). Most patients would therefore be spared
regular clinic review if HBPM was adopted, a bene®t for
both the patient and the overworked clinic.
During this study we found a high percentage of patients
with `white-coat hypertension' (54% on CBP measure-
ment). This is high compared with non-diabetic popula-
tions, possibly due to the effect on medial sclerosis in
diabetic subjects, resulting in reduced arterial compliance,
thus amplifying the white-coat effect. Therapeutic deci-
sions in UKPDS and other BP-lowering trials were made
on the basis of clinic BP measurements, and as always
there is concern that we should not try to exclude patients
who would have been treated in such trials. The protocols
for BP measurement in these trials were strictly monitored
and adhered to, and we regret that the conditions of these
trials are not easily reproducible in the clinic setting. We
would suggest that the results of HBPM provide a better
assessment of true BP in aiding management of hyperten-
sion than CBP in a standard clinic.
In asking the patient to monitor their BP until they went
to bed, we hoped that we could identify those patients who
lose nocturnal BP variation on ABPM (`non-dippers').
ABPM requires BP measurement throughout the night,
which is inconvenient for the patient. However, whilst
HBPM was highly sensitive, it was not speci®c. Thus we
cannot recommend that HBPM be used for this purpose.
In conclusion, this study has shown that home BP
monitoring using a cheap semiautomatic cuff-oscillo-
metric BP monitoring device is superior to clinic BP
measurement, and allows far superior detection of hyper-
tension. Clinic BP measurement revealed `white-coat
hypertension' in over half of the sampled patients, which
could lead to widespread over-treatment of hypertension.
It would seem a reasonable strategy to incorporate Home
Blood Pressure monitoring in the preinvestigation of
patients prior to an annual review along with an estimate
of glycaemic control.
Acknowledgements
We would like to thank most sincerely the Non-Invasive
Cardiology Department at the Royal South Hants
Hospital, Southampton, for their invaluable help. We
would also like to thank Sr. Vivien Foster for her help.
References
1 United Kingdom Prospective Diabetes Study Group. Tight blood
pressure control and risk of microvascular and macrovascular
complications in type 2 diabetes: UKPDS 38. Br Med J 1998; 317:
703±713.
2 Stanton A, O'Brien E. Noninvasive 24-hour ambulatory blood
pressure monitoring: current status. Postgrad Med J 1993; 69:
255±267.
3 Kjeldsen SE, Moan A, Petrin J, Weber AB, Zwei¯er AJ, Julius S.
Evaluation of self-measured home vs. clinic intra-arterial blood
pressure. Blood-Press 1993; 2: 28±34.
4 Stergiou GS, Malakos JS, Voutsa AV, Achimastos AD,
Mountokalakis TD. Home monitoring of blood pressure: limited
value in general practice. J-Hum-Hypertens 1996; 10: 219±223.
5 Stergiou GS, Voutsa AV, Achimastos AD, Mountokalakis TD.
Home self-monitoring of blood pressure: is fully automated
oscillometric technique as good as conventional stethoscopic
technique? Am-J-Hypertens 1997; 10: 428±433.
6 Brueren MM, de Schouten HJA, Leeuw PW, van Montfrans GA, van
Ree JW. A series of self-measurements by the patient is a reliable
alternative to ambulatory blood pressure measurement. Br J Gen
Prac 1998; 48: 1585±1589.
7 O'Brien E, Mee F, Atkins N, O'Malley K. Accuracy of the Spacelabs
90207 determined by the British Hypertension Society Protocol. J
Hypertens 1991; 9: S25±S31.
8 Ramsay LE, Williams B, Johnston GD, MacGregor GA, Poston L,
Potter JF et al. British Hypertension Society guidelines for hyperten-
sion management: summary. Br Med J 1999; 319: 630±635.
9 Williams B. The pathogenesis of hypertension in diabetes mellitus.
436 Blood pressure assessment in Type 2 diabetes · M. G. Masding et al.
ã 2001 Diabetes UK. Diabetic Medicine, 18, 431±437
In: Diabetes and HypertensionÐa Fatal Attraction Explained.
Maidstone: Publishing Initiatives Books, 1997: 26±27.
10 Bland JM, Altman DG. Statistical methods for assessing agreement
between two methods of clinical measurement. Lancet 1986; 327:
307±311.
11 Fotherby MD, Potter JF. Reproducibility of ambulatory and clinic
blood pressure measurements in elderly hypertensive subjects. J
Hypertens 1993; 11: 573±579.
12 Engfeldt P, Danielsson B, Nyman K, Aberg K, Aberg H. 24-hour
ambulatory blood pressure monitoring in elderly normotensive
individuals and its reproducibility after one year. J Hum Hypertens
1994; 8: 545±550.
Original article 437
ã 2001 Diabetes UK. Diabetic Medicine, 18, 431±437