Journal of Human Hypertension (2001) 15, 197–201
 2001 Nature Publishing Group All rights reserved 0950-9240/01 $15.00
www.nature.com/jhh
ORIGINAL ARTICLE
A clinical study of the Korotkoff phases of
blood pressure in children
J O’Sullivan1, J Allen2 and A Murray2
Departments of 1Paediatric Cardiology and 2Regional Medical Physics Department, Freeman Hospital,
Newcastle upon Tyne, UK
Background: Five Korotkoff phases are described in
adults, but there are no studies of the Korotkoff phase
distribution in children. This study determines the presence and length of Korotkoff phases in children, providing data on the repeatability of these measurements, the
relationship between the phases, and finally the
relationship between the phases and heart rate, blood
pressure and arm circumference.
Methods: Seventy, 11-year-old children were studied.
The Korotkoff sounds were recorded from the bell of a
stethoscope to a MiniDisc system and each sound
described twice on separate occasions as phase I, II, III
or IV, with phase V meaning disappearance of the
sound.
Results: Phases I, II, III, IV and V were present in 97%
(68/70), 61% (43/70), 51% (36/70), 88% (62/70) and 80%
(56/70) respectively. When the recordings were blindly
re-assessed there was no significant difference in the
phase distribution of the sounds. All five phases were
present in 40% (28/70). Phase III only occurred in the
presence of phase II (P ⬍ 0.0001). There was no significant relationship between the presence of the different
phases and heart rate or blood pressure. Arm circumference was significantly larger in children with phase V
present (P ⬍ 0.02).
Conclusions: The Korotkoff sounds and phase distribution present in normal children is described. Korotkoff sounds were consistently allocated to the various
Korotkoff phases. This study provides insights into the
problems of accurate diastolic blood pressure measurement. Phase V was more likely to be present with
increasing arm circumference, but the variation in the
occurrence of phases II and III remains unexplained.
Journal of Human Hypertension (2001) 15, 197–201
Keywords: Korotkoff sounds; diastolic blood pressure
Introduction
The measurement of blood pressure using the Korotkoff method is done by listening over the brachial
artery during deflation of an upper arm cuff. In 1905
Korotkoff1,2 described three different sounds that
could be distinguished by listening with a stethoscope placed over the brachial artery. The original
description of the sounds has been slightly modified
over the years and we now recognise four different
types of sound (and five phases).2
The pressure in the mercury sphygmomanometer
when the first sounds are heard represents systolic
blood pressure and there is continuing debate as to
whether disappearance (phase V) or muffling of
sounds (phase IV) best represents diastolic pressure
in children.3 Even though it is recognised that phase
V (disappearance of sounds) may be absent in children, the presence or otherwise of the other four
phases has not been described. This study describes
Correspondence: Dr John O’Sullivan, Department of Paediatric
Cardiology, Freeman Hospital, Newcastle upon Tyne, NE7 7DN,
UK. E-mail: J.J.Osullivan얀ncl.ac.uk
Received 22 May 2000; revised 5 August 2000; accepted 11
August 2000
the Korotkoff sounds present in normal children;
how consistently the sounds could be allocated to
the various phases on repeat assessment, the
relationship between phases, and how the Korotkoff
phases relate to heart rate, systolic and diastolic
blood pressure, and arm circumference.
Subjects and methods
Subjects
A local secondary school agreed to participate in the
study. Following a visit to the school, children took
a consent form home to their parent/guardian. The
study was carried out on a normal school day that
did not include a physical education class. Formal
approval was obtained from the local Ethics Committee.
Recording equipment
The Korotkoff sound measurement system is illustrated in Figure 1. The bell of a Littmann cardiology
grade stethoscope (No 2151, 3M) was attached to a
miniature high sensitivity ceramic microphone (BL1994, Knowles Electronics Co) using 60 cm of
Korotkoff sounds in children
J O’Sullivan et al
198
dren of this age the 5th centile for diastolic pressure
is over 40 mm Hg.4
Korotkoff sound analysis
Figure 1 The figure shows the arm cuff which is inflated automatically and which is deflated at an approximate constant rate.
The stethoscope is connected to a microphone, allowing the
investigator to listen through headphones to the sounds being
recorded onto MiniDisc.
stethoscope tubing. An amplifier was connected to
the microphone with the highest frequency range of
the amplifier chosen to provide realistic and faithful
audio-quality reproduction on listening. In addition,
all components of the measurement system were
chosen to extend down to 10 Hz to cover the audible
hearing range in humans. The signal was then
passed to a Sony MDS-JB930 MiniDisc digital
recorder which provided high quality sound reproductions with low noise. The operator listened
through a set of studio quality audio headphones
(Beyer, DT150) during recording and playback. To
obtain a consistent and controlled release of air from
the arm cuff an automatic blood pressure monitoring
device (A&D Blood Pressure Monitor, TM2421) was
modified to provide consistent cuff deflation from
130 down to 30 mm Hg at a mean deflation rate of
−3.3 mm Hg/s.
The Korotkoff phases were identified by the character of the sounds, as outlined by Geddes;2 phase I
sounds are loud with a clear-cut and snapping tone,
phase II sounds have a murmur-like quality, phase
III sounds are similar in character to first phase
sounds, and phase IV sounds have a dull or muffled
tone. The fifth phase represents the complete disappearance of the sounds. One of the investigators
(JO’S) allocated each of the sounds to one of these
phases and repeated it 2 months later (blinded) to
check consistency. Since the cuff deflated rapidly
below 30 mm Hg it was not possible to say if phase
V was absent, but for the purposes of analysis, if
sounds persisted to 30 mm Hg then phase V was
said to be absent. If the first sounds heard had a murmur-like quality then phase I was said to be absent.
The number of beats heard in phases I to IV was
counted and the length of each phase measured.
Statistical methods
The relationship between the blood pressure and
arm circumference was analysed using standard
regression plots. The differences in phase lengths
and also arm circumference for the phase groups
were assessed using the Mann-Whitney test. The
inter-relationship between phases II and III was
assessed using the Chi-squared test. In assessing
paired differences in repeat data the confidence
intervals of ±2 standard deviations about a mean difference level, as recommended by Bland and Altman, were calculated.5
Measurement procedure
Results
Children were studied in a quiet room in the school.
Each child sat quietly for at least 10 min before the
measurements. An appropriate sized pressure cuff
(bladder width at least 40% of the arm
circumference) was placed around the left arm and
was supported near to heart level using a cushion.
A Doppler ultrasound probe (Dopplex, Huntleigh
Healthcare) was used to locate the clearest pulsatile
signal at the antecubital fossa and the skin marked.
Blood pressure was first measured with a mercury
sphygmomanometer, using phase V (if present) for
diastolic. The Korotkoff sounds were then recorded
using the equipment described above. There were
two stages: a first one to preview the sounds and the
second to make a recording to MiniDisc. Subjects
were asked to keep still but relaxed during the
measurement. Each recording lasted for 1 min
allowing ample time for the deflation cycle. A
‘bleep’ was generated once the cuff pressure fell
below the start of rapid deflation at 30 mm Hg, as
this allowed the listener to ignore any sounds after
this point. This was considered valid since in chil-
Blood pressure and arm circumference
Seventy children (31 males), all aged 11 years, were
studied. Median height was 150 cm (Interquartile
range (IQR): 146–155 cm) and median arm circumference was 22 cm (IQR: 21–24 cm). The systolic
blood pressure ranged from 84 mm Hg to 130 mm Hg
and diastolic pressure from 50 mm Hg to 78 mm Hg.
These parameters are presented for boys and girls
separately in Table 1. There was a significant corre-
Journal of Human Hypertension
Table 1 The table details some of the measurements obtained in
boys and girls separately. The mean values (± standard deviation)
for heart rate (beats per minute), systolic blood pressure, diastolic
blood pressure (phase V), arm circumference (cm) and standing
height (cm) are provided
Male
Heart rate
Systolic BP
Diastolic BP
Arm Circumference
Height
80.2
99.9
61.7
22.8
150.8
(±11.4)
(±9.2)
(±9.2)
(±4.5)
(±27.4)
Female
83.8
99.7
63.1
23.3
150
(±12.3)
(±9.6)
(±9.5)
(±4.5)
(±8.8)
Korotkoff sounds in children
J O’Sullivan et al
lation between the systolic and diastolic blood
pressure (r = 0.32, P ⬍ 0.01). There was a significant
correlation between arm circumference and height
(r = 0.44; P ⬍ 0.001). There was also a significant
relationship between arm circumference and both
systolic and diastolic blood pressure (r = 0.44, P ⬍
0.001 and r = 0.29, P ⬍ 0.05), with larger arm circumferences associated with higher pressures.
Reproducibility of the Korotkoff sound analysis
When the same investigator (blinded) listened again
to the recordings, the phases identified in each child
were identical to those identified on the first except
for two children where phases IV was thought to be
present on one occasion but not on the other. The
number of beats allocated to each phase on both
occasions is shown in Figure 2. There was no significant difference with the repeats for any phase.
The largest difference was in phase IV with a difference of less than one beat. The discrepancy in phase
IV sounds (muffling) was due to their low amplitude
and these sounds faded rather than abruptly ceased
making it difficult to decide whether the sound had
disappeared or was still faintly present.
199
Pattern of Korotkoff phases present
The patterns of Korotkoff phases present is
presented in Table 2. In virtually all children phase
I was present (68/70, 97%) and phase IV (muffling)
was identified in 62/70 (88%). Sounds persisted to
the end of the controlled deflation (30 mm Hg) in
14 children suggesting that under these conditions
phase V does not occur in approximately 20% of
children. Phase II and phase III sounds were present
in 43/70 (61%) and 36/70 (51%) respectively. Less
than half of the children (28/70) had all five phases.
Phase III only occurred with phase II, with phase II
being present without phase III in only seven children. This was highly significant (P ⬍ 0.0001). The
proportions of the phases are shown in Figure 3.
There was no significant difference in the proportion of each phase in the repeat assessment.
The length of the phases is shown in Figure 4.
Phase I was significantly longer in those children
where phases II and III were absent (8.2 s [IQR 6.4 –
10.9] compared to 3.2 s [IQR 2.4 –4.6], P ⬍ 0.00001),
and similarly for the number of beats (10.5 beats
[IQR 9–15] compared to 4.0 beats [IQR 4.0–6.0], P
⬍ 0.00001). Phase IV was significantly longer when
phase V was absent (13.7 s [IQR 12.0–15.4] compared to 3.7 s [IQR 1.7–8.7], P ⬍ 0.00001), and similarly for the number of beats (17.0 beats [IQR 14.8–
22.3] compared to 7.3 beats [IQR 2.3–11.8], P ⬍
0.00001).
Relationship between Korotkoff phases and other
parameters
There was no significant relationship between the
relative duration of the various phases and heart
rate. Also, the presence or absence of the various
phases had no significant relationship with heart
rate, sex, systolic or diastolic blood pressure. We
did, however, find that arm circumference (Figure
5) was significantly larger in those with phase V
(23.0 cm, [IQR 22.0–25.0]) versus those with no
phase V (21.5 cm, [21.0–22.0], P ⬍ 0.02).
Discussion
Figure 2 (a) The number of beats allocated to the various phases
when the recordings were assessed on the first (1) and second (2)
occasions (median ± Inter-quartile range (IQR). (b) The difference
in the number of beats allocated to the different phases when
the recordings were assessed on the second occasion. The mean
difference ±2 standard deviations are shown in the figure.
The only previous studies on the clinical description of all Korotkoff phases in the literature were
undertaken in adults over 80 years ago. Swan6
described the Korotkoff phases in 200 adults. All
phases were present in about 40% and phase IV was
detected in less than half of the subjects. Goodman
and Howell7,8 measured the proportion of the pulse
pressure that the phases occupied and reported that
Journal of Human Hypertension
Korotkoff sounds in children
J O’Sullivan et al
200
Table 2 The table shows the different patterns of phases present in the 70 children. The area is shaded if that phase is present. No. =
number of children with the different patterns of phase distribution
I
II
III
IV
V
No. (=70)
28
15
7
6
5
3
2
1
1
1
1
68
43
36
Figure 3 The duration of the Korotkoff sounds for each individual was measured and the proportion occupied by each phase
calculated. These proportions are presented in the figure (median
± Inter-quartile range (IQR)) for each phase and for the two
occasions when the sounds were assessed. The numbers 1 and 2
on the figure refer to the first and second assessments
(respectively) of the recordings.
Figure 4 The length of each phase was measured, in seconds, and
is shown in the figure as the median duration, ± Inter-quartile
range.
Journal of Human Hypertension
62
56
Figure 5 Children were divided into those with (yes) and those
without (no) phase V. The difference in arm circumference
between these two groups is presented with each individual
value shown as an open circle and the group median (± Interquartile range) as the closed circles.
phase II was the longest phase. In these studies the
Korotkoff sounds were analysed during the actual
blood pressure measurement which has large potential errors as it is very difficult to achieve consistent
rate of manual cuff deflation and the investigator
had only one opportunity to allocate the sounds to
the various phases.
Our method allowed consistent cuff deflation and
recording of the sounds for later analysis. Using this
approach each sound was described as phase I, II, III
or IV and the allocation of the sounds was consistent
when the recordings were analysed again. A number
of interesting observations could be made. Less than
half the children had all 5 phases and a number of
different patterns were observed. Phase II was
present in about two-thirds and was strongly associated with phase III, which itself was present in
about 50% of children. The length of the individual
phases also showed much variation and when some
phases were absent the remaining phase(s) were
longer. We found that phase V was more likely to
Korotkoff sounds in children
J O’Sullivan et al
be present in children with larger arm circumference. Arm circumference is positively associated
with increasing height and weight and therefore
larger arm circumference may be the main explanation for the fact that phase V is virtually always
present in normal adults. However other factors
such as pulse pressure and arterial wall compliance
may also be important. As there is no consensus on
how the Korotkoff phases are produced,9,10 one can
only speculate as to the significance of the observed
variation in the presence and length of the phases in
children with similar blood pressure. The apparent
differences between our findings and those reported
in historical adult studies6,7 might suggest that
arterial wall compliance is an important factor, but
methodological differences between the studies
make comparisons difficult.
Our study also provides some insights into the
problems of accurate diastolic blood pressure
measurement in children. It was clear that phase IV
(muffling) sounds are of low amplitude and often
difficult to hear. Under the optimal listening conditions of our study phase IV could not be detected
in 12% of children. The difficulty with the reliable
detection of phase IV is highlighted by the widely
different results of two large studies. A Finnish
study found that phase IV sounds were not detected
in 3% of over 1000 12-year-old children11 and an
American study on children aged 10–15 years found
no phase IV in over 50%.12 Using phase V for diastolic blood pressure measurement is also prone to
error as the phase IV sounds generally fade rather
than abruptly cease. The small difference in the
length of phase IV when the recordings were reanalysed suggests that under standard clinical conditions the reproducibility would be much poorer.
This was the case in a study on between- and
within-visit variance of blood pressure, which concluded that diastolic blood pressure (using phase IV
or V) was a ‘strikingly imprecise measurement’ for
children aged 8–12 years.13
Our study describes in detail the Korotkoff phases
detectable in normal children under optimal listening conditions. This is the first study to describe all
the Korotkoff phases in children and it provides
insight into the problems of accurate clinical
measurement of diastolic blood pressure in this age
group. It is noteworthy that almost 100 years since
their initial description we have made little progress
in understanding the clinical significance of the Korotkoff phases of blood pressure.
201
Acknowledgements
We are grateful to the parents, teachers and children
of Heaton Manor school who supported the study,
and to Mrs Toni Allen who helped to organise the
study and was funded by the Children’s Heart Unit
Fund (CHUF). The equipment for this study was
purchased by a grant from the Freeman Hospital
Special Trustees.
References
1 Korotkoff NS. On the subject of methods of determining blood pressure [in Russian]. Bull Imperial Military
Med Academy, St. Petersburg 1905: 11: 365.
2 Geddes LA, Hoff HE, Badger AS. Introduction of the
auscultatory method of measuring blood pressure –
including a translation of Korotkoff’s original paper.
Cardiovasc Res Center Bull 1966; 5: 57–74.
3 Gillman MW, Cook NR. Blood pressure measurement
in childhood epidemiological studies. Circulation
1995; 92: 1049–1057.
4 de Man SA et al. Blood pressure in childhood: pooled
findings of six European studies. J Hypertens 1991; 9:
109–114.
5 Bland JM, Altman DG. Statistical methods for
assessing agreement between two methods of clinical
measurement. Lancet 1986; i: 307–310.
6 Swan JM. The auscultatory method of blood pressure
determination: a clinical study. Int Clin 1914; 4:
130–190.
7 Goodman EH, Howell AA. Further clinical studies in
the auscultatory method of determining blood pressure. Am J Med Sci 1911; 142: 334 –353.
8 Goodman EH, Howell AA. Clinical studies in the auscultatory method of determining blood pressure. University Penn Med Bull 1910; 23: 469– 475.
9 Drzewiecki GM, Melbin J, Noordergraaf A. The Korotkoff sound. Annals of Biomed Engineering 1989; 17:
325–359.
10 McCutcheon EP, Rushmer RF. Korotkoff sounds. An
experimental critique. Circ Res 1967; 20: 149–161.
11 Uhari M, Nuutinen M, Turtinen J, Pokka T. Pulse
sounds and measurement of diastolic blood pressure
in children. Lancet 1991; 338: 159–161.
12 Sinalko AR, Gomez-Marin O, Prineas RJ. Diastolic
fourth and fifth phase blood pressure in 10–15 year old
children. Am J Epidemiol 1990; 132: 647–655.
13 Rosner B et al. Reproducibility and predictive values
of routine blood pressure measurements in children.
Am J Epidemiol 1987; 126: 1115–1125.
Journal of Human Hypertension