Stress and Heart Rate Variability
Prashant Suryanarayanan
1
and Dinesh Kant Kumar
1
School of Electrical and Computer Engineering,
RMIT University,
PO Box 2476V
Melbourne,
Australia - 3001
Abstract. Do people get accustomed to stress? Do people react differently to
a stressful environment when they have experienced the situation earlier? This
paper reports research to determine whether the above is a myth. It reports the
measurement of heart rate variability (HRV) of people when subjected to con-
trolled stress conditions and these conditions are repeated over different days.
The results indicate that while there is a large variability between different par-
ticipants, there is a consistent reduction in the heart rate variability from the first
experience of the stress condition to the subsequent applications of similar stress
conditions.
1 Introduction
Autonomic responses in psychologically challenging situation are a major area of psy-
chophysiological study and research. Psychophysiology is defined as the study of rela-
tions between psychological manipulations and resulting physiological responses, and
has developed over the years to promote better understanding between psychological
and physiological processes [1]. There are number of anecdotal evidences suggesting
the change in heart rate (HR) with stress of the person. Patients suffering from cardiac
conditions are often asked to reduce their anxiety and psychological stress. Changes
in cardiac activity had been observed by early scientists related to psychological phe-
nomena [1]. In the recent past, researchers have used heart rate and its variability as a
measure of mental load [2]. Variability of heart rate has also been used for ‘lie detector’
applications.
While the phenomena of change in heart rate with stress of people seems to be
widely accepted, Indian society has known the use of techniques such as yoga and med-
itation to control the cardiac activity for thousands of years. In our modern technology
driven society, biofeedback is being considered for helping individuals have some con-
trol on their cardiac activity. In each of these, the individual makes a conscious effort
and often undertakes training to develop a control over their heart rate.
Social health scientists believe that this stress causes increased cardiac stress and
results in long term damage to the body of the individuals living under these conditions
Suryanarayanan P. and Kant Kumar D. (2005).
Stress and Heart Rate Variability.
In Proceedings of the 1st International Workshop on Biosignal Processing and Classification, pages 68-77
DOI: 10.5220/0001194000680077
Copyright
c
SciTePress
[3], [4], [5]. But in our modern society, some of us are regularly exposed to stress and
anecdotal evidence suggests that we get accustomed to this stress. It is then arguable
that people who are used to living under stress may be developing immunity from the
stress at their physiological levels. It has been argued that stressful environment be-
comes less challenging to an individual who is used to working in such an environment
and such an individual will have little change in their heart rate when under stress.
Thus, it maybe suggested that the bodies of people who work in stressful and challeng-
ing professions would be able to cope with the stress easily with little changes to their
cardiac activity compared with other people. This paper reports experimental research
to test the above commonly held beliefs. The paper reports tests conducted on differ-
ent days to determine the correlation between changes to heart rate as the participants
became accustomed to the controlled stress conditions. For this purpose, controlled ex-
periments were conducted where electrocardiogram of the participants was recorded
and correlated with the controlled stress conditions.
1.1 Electrocardiogram
Electrocardiogram (ECG) is the electrical potentials of the heart and generally recorded
non-invasively from the surface of the body. The first known equipment to record ECG
was invented by Williem Einthoven. The origin of the electrical activity measured by
ECG is in the cardiac muscle fibres. ECG can be divided in phases of depolarization and
repolarization of different sections of the heart. The most significant section of ECG is
the QRS complex, the peak of which is ’R’ wave. QRS corresponds to the contraction
of the left ventricle.
ECG is a repetitive signal and corresponds with the beating of the heart. The rate
of repetition of ECG, measured using QRS peak, corresponds with the heart rate. The
HR of different people is not comparable due to differences such as body size, age
and genetics. Heart rate of an individual varies with number of factors including stress
and physical exercise. Due to these reasons, there is a very large inter and intra subject
variability and hence heart rate and ECG repetition are not useful for comparison. But
variation of heart rate is a more reliable measure of changes to the cardiac activities [6].
1.2 Heart Rate Variability
The heart rate variability (HRV) has been recognized as a non-invasive means to as-
sess the state of autonomic nervous system [7]. The autonomic nervous system (in-
voluntary nervous system) consists of two components; the Sympathetic and the Para-
sympathetic branch. Increased activity of sympathetic branch causes an increase in the
heart rate while increase in Para-sympathetic branch causes the slowing of heart rate.
When the individual is under stress, the sympathetic activity is more profound than
the Para-sympathetic activity causing an imbalance in the autonomous nervous system
(ANS) and hence the HR of the person increases [8]. HRV analysis is a measure of the
variability in heart rate; specifically, variability in intervals between successive R waves
- ”RR intervals”. This is also called as Interbeat Interval (IBI) and is defined as time in
milliseconds between two normal R to R waves of an ECG. The variations of the heart
rate are affected by a diverse number of factors such as the metabolic activity (related
69
to physical activity) and emotional activity. The change in heart rate affects the overall
well-being of the person including their physical abilities and mental and emotional
capabilities. Drastic change in HR is known to affect the individual’s capability to take
decisions, solve problems, and the overall perception of the events and environment
[9]. The relationship between mental stress and HRV is a very active area of research
for human-machine interface. Number of studies have been reported that have analyzed
HRV with mental stress to identify the impact of various factors on our wellbeing such
as during physical work [10], with controlled breathing [11], with and without vocaliza-
tion [4], [12], effect of individual differences on stress responsivity [13], with electrical
stimuli [1]. The studies suggest that HRV is an effective, objective and non-intrusive
measure for mental stress evaluation.
This paper reports the study conducted to determine whether repeated exposure to
stress reduces the impact of emotional and psychological stress on the HRV of the
individuals. For this study, mental tasks have been used as the laboratory emotional
and psychological stressors. Based on other reported studies, tasks requiring mental
arithmetic (MA) have been used (e.g. [10]; [11]; [12]). The mental tasks that were
applied as stimuli ranged from simple to difficult mathematical operations, and memory
tests (e.g. English sentence reordering etc.).
2 Methodologies
The aim of this research was to identify the changes in HR due to emotional and psy-
chological stress. Towards this aim, controlled experiments were conducted where the
HR was recorded when the participant were made to undertake the mental tasks, with
all other conditions kept unchanged. These experiments were repeated for three days.
Participants performed five minute simple mental tasks on three consecutive days.
For each day, participants were asked to solve a series of mental tasks that are detailed
below.
2.1 Mental Tasks
The participants were given a number of mental tasks on each day, design of these tasks
adopted from [1]. These tasks consisted of three set of tasks; word memory, numeric
and tasks requiring planning. The set of these tasks were repeated three times (one par-
ticipant did this four times). Between each stimulus, the participants were encouraged
to relax for about one minute. In accordance with [12], the participants were asked to
vocalize their responses to the mental tasks. The mental tasks stimulus to which the
participants were subjected is described below:
2.2 Day 1
The first mental task that was given to the participants was a ’memory task’. The par-
ticipants were given eight randomly selected commonly used mid-size English words
and were asked to read these aloud (excluding 3 punctuation marks). Ten seconds after
this was completed, they were asked to repeat the eight words in the same order.
70
The second mental task for the day was a quantitative problem involving one step,
single digit multiplication. This required the participant to calculate the hourly speed of
the car, given the distance the car travelled for 15 minutes.
The third task required the participant to unscramble a jumbled word. The word was
selected randomly, it was five or six characters long, and of common usage.
2.3 Day 2
The tasks on the second day were similar to that of the first day, but were in a different
order and of greater complexity. The first task was a numeric task, similar to the second
task of day one, where the participant was given a simple single digit multiplication
task.
The second task for the second day was also a numeric task where the participant
was asked to divide a randomly generated ve digit number by ’7’ to the second decimal
places orally. The third task was a two digit oral multiplication. The fourth task was a
memory test, similar to task one on day one.
2.4 Day 3
The tasks and the order of the tasks used on the earlier two days were randomly selected
for the third day. The actual words and numbers were changed for each experiment.
3 Participants
Sixteen (male and female) University students volunteered to participate in the study.
All participants were in good health with no history of cardiovascular disease and were
not under any medications that may affect their cardiovascular functions. All partici-
pants were advised to abstain from caffeine, alcohol and nicotine for at least 2 hours
before testing.
4 Apparatus
Electrocardiogram (ECG) were measured using a BIOPAC ECG100C amplifier. The
ECG was measured using Ag-AgCl disposable sensors with a high conductivity wet
gel. AcqKnowledge (Biopac) software was used to record the ECG. The sampling rate
was 200 samples/sec. The analysis was done offline using MATLAB.
5 Procedure
The experiments were conducted in accordance with the University Human Ethics
Committee approval. The participants were informed that they were free to stop the ex-
periment when they so wished without giving any reason. The participants were made
aware of the details of the experimental procedure, the overall aim of the project and
71
were made familiar with the laboratory and the apparatus. The participants were pro-
vided with the plain language statement of the experiments. They were included in the
experiments after they signed the consent form. The room was air-conditioned with
temperature maintained between 20
o
C and 22
o
C throughout all the experiments.
At the start of each session, the participants washed their hands using warm wa-
ter and soap free cleanser; they sat down in a comfortable chair. Disposable Ag/AgCl
electrodes were attached in the 3-lead configuration. The location of the electrodes was
marked to ensure repeatability over the different experiments. The apparatus was cali-
brated to ensure reproducibility.
The participant was asked to relax for ve minutes after the electrodes were attached
to them. At this end of this period, the participant continued to relax for another five
minutes, and during this time their ECG was recorded. This ECG corresponded to the
relaxed state and HR during this period was taken as the base level for all other com-
parisons. At the completion of this period, the participant was given a series of mental
tasks while their ECG was being continuously recorded. The instants when the stimu-
lus (mental task) was administered and when the participant completed the task were
recorded.
The experiments were repeated on three different days. The experiments were con-
ducted at the same time each day to reduce variability due to the time of the day.
6 Data Analysis
The first step of the analysis involved identifying the QRS complex. This involved a
series of band-pass filters, differentiators and squaring the differentiated signal, a tech-
nique described by Pan and Tompkins [14], [15]. This algorithm is based on the slope,
amplitude and width of the ECG. An auto-threshold selection was done using two clus-
ter unsupervised learning using nearest neighbour criteria described in [16].
Heart rate (HR) was computed by measuring the R-R intervals from the ECG signals
using the QRS detected above. Heart rate variability (HRV)was subsequently calculated
using the software package Advanced HRV Analysis [17].
For the purpose of studying the inter-day variation, data was normalized by consid-
ering the first day as the reference point.
7 Statistical Analysis
After extracting the HR from the ECG recordings, statistical analysis was performed
to identify the variability and the significance due to the stimulus. The analysis was
performed using the Data Analysis tool-pack add-in provided with Microsoft Excel. A
Student t-test was performed on the data, which gives the probability that the difference
between the two means is caused by chance. It is customary to say that if this probability
is less than 0.05, then the difference is ’significant’ and it is not caused by chance. The
t-test analysis is appropriate whenever one needs to compare the means of two groups.
The t-test is a ratio, with the numerator being difference between the two means or
averages and denominator being measure of the variability or dispersion of the scores
as expressed below:
72
t value =
Dif ference Between the Group M ean
V ariability of the Groups
To find the numerator, the difference between the means of the groups is calculated.
To find the denominator, which is also called as the standard error of the difference, the
variance of each group is calculated and is divided by number of people in the group.
By taking the square root of the sum of both the variances we get the standard error.
The variance is the average of the square of the distance of each data point from the
mean. It is also known as mean squared deviation.
The t-value will be positive if the first mean is larger than the second and negative
if it is smaller. Once you compute the t-value you have to look it up in a table of sig-
nificance to test whether the ratio is large enough to say that the difference between the
groups is not likely to have been a chance. To test the significance, you need to set a
risk level called the alpha level. Alpha level is the risk of rejecting the Null Hypothesis
when in fact it is true. In other words, stating a difference exists where actually there is
none. Alpha risk is stated in terms of probability (such as 0.05 or 5%). The p-value is
the probability of finding a difference between the two group means and as explained
above if this p-value is less than 0.05, then the difference is significant’ and it is not
caused by chance.The alpha level was always set to 0.05 for all the t-test performed. A
Paired t-test was performed on the mean HR for all the participants paired for:
The same question position on different days.
Between different question on the same day.
A Paired t-test is normally used to compare means on the same or related subject
over time or in differing circumstances and it does not assume that the variance of both
populations are equal. A two sample t-test with unequal variance was also conducted on
the heart rate for each of the participants to determine the overall HRV for an individual
on different days.
8 Results
There was a variation of the total number of questions that were answered by the dif-
ferent participants on different days. The minimum questions answered were nine, and
hence the first nine questions for each experiment were considered for the paired t-test
analysis. The two samples t-test with unequal variance to identify changes between the
different days was considered for all the questions answered by the participants on each
of the days. Table 1 lists the mean heart rate of all the participants for all the three days.
The mean heart rate here implies the average of the participants’ heart rate during all the
mental tasks conducted on that day. The tables (2, 3, 4) list the p-values for two-tailed
distribution analysis of the data for the following
Table 2 for identifying changes due to different questions on each of the three days.
Table 3 for identifying changes due to different days.
Table 4 for identifying changes due to each of the questions on different days.
73
Table 1. Mean Heart Rate of the all the participants during the mental tasks.
Participant Day 1 HR Day 2 HR Day 3 HR Day 2/Day 1 Day 3/Day 1
1 100.11 99.96 86.21 0.998 0.861
2 83.43 77.96 75.57 0.934 0.906
3 88.39 81.22 80.27 0.919 0.908
4 75.37 71.34 67.26 0.947 0.892
5 87.92 88.97 84.30 1.012 0.959
6 86.71 78.13 72.59 0.901 0.837
7 86.48 73.55 73.78 0.850 0.853
8 79.98 86.47 70.57 1.081 0.882
9 83.04 83.74 82.29 1.008 0.991
10 89.07 93.23 80.77 1.047 0.907
11 98.43 87.09 81.07 0.885 0.824
12 75.79 68.59 69.08 0.905 0.911
13 82.10 82.25 85.36 1.002 1.040
14 96.52 86.24 95.97 0.894 0.994
15 79.08 75.18 68.89 0.951 0.871
16 78.72 71.49 74.97 0.908 0.952
Table 2. p-values of Paired t-test Between Questions on Same Day for all the participants
Question Day 1 Day 2 Day 3
Q1/Q2 0.0685 0.8166 0.6055
Q2/Q3 0.6044 0.9455 0.0376
Q3/Q4 0.6876 0.2782 0.2611
Q4/Q5 0.4037 0.0996 0.0489
Q5/Q6 0.5541 0.6431 0.1137
Q6/Q7 0.5429 0.9761 0.6027
Q7/Q8 0.5813 0.8892 0.0990
Q8/Q9 0.5937 0.7124 0.4410
Table 3. p-values of Paired t-test Between the Mean Heart Rate for all the participants
T-Test Day 1/Day 2 0.013578121
T-Test Day 2/Day 3 0.243691025
T-Test Day 1/Day 3 0.000128381
74
Table 4. p-values of Paired t-test Between the Same Task Order on Different Days for all partic-
ipants
Question Day 1 / Day 2 Day 2 / Day 3 Day 1 / Day 3
Q1 0.357610 0.460882 0.035439
Q2 0.055732 0.422769 0.000077
Q3 0.076067 0.228351 0.000106
Q4 0.200562 0.085746 0.000898
Q5 0.018164 0.270949 0.000823
Q6 0.031875 0.596949 0.003855
Q7 0.024736 0.712155 0.013709
Q8 0.022571 0.533416 0.001042
Q9 0.025960 0.728800 0.038138
9 Observations
From Table 1, it is clear that there is a consistent reduction of Heart rate from Day 1 to
Day 2 to Day 3 for a significant number of participants. From Table 2, it can be seen
that there is no statistically significant change between the HR between two questions
on the same day, although there is drop in the HR of the participant over the experiment
period.
From Table 3, it is observed that there is a significant reduction in HRV from day
one to day two. The change from day two to the third day is much smaller and not
significant. This is also confirmed by the two sample T-test of unequal variance. It is
also observed (not tabulated) that the average heart rate change between questions on
the same day for each participant considering all participants was found to reduce from
±3.69 on Day 1 to ±2.60 on Day 3. The two sample T-test assuming unequal variance
on the heart rate was also performed for all the questions between the following day
pairs (Day 1/Day 2, Day 2/Day 3, Day 1/Day 3), and it showed a weak relationship of
HR with all subjects considered.
From Table 4, it is observed that between different days, for the same order of the
task, there is small but relatively higher relationship between the HRV compared to the
comparison between different questions on the same day.
10 Conclusion
From the experimental results, it has been observed that the mental tasks do cause a
variation in the heart rate of the participants. It is also observed that this variation is the
greatest on the first day. On the subsequent days, even when the tasks are made more
difficult, the variation is smaller. It is also observed that the changes to the variation of
the heart rate is from the first day to the second, but from the second to the third the
change is relatively small. It is also observed that the intra-day variation of HR for is
maximum for the first day, and minimum for the third day, from ±3.69 to ±2.60. It is
also observed that the HR for different participants was significantly different. For the
same participant, the HR for the three days was comparable, even though a reduction
from the first day was observed.
75
Based on the above, the authors conclude that:
Absolute values of HR of different people should not be compared.
The HRV reduces when the individual has experienced the experimental stress prior
to the experiment.
It is suggested that HRV should not be used to measure the emotional impact of
a stimulus on a person as it is possible that they may have got accustomed to the
stress and the experiments.
It is not possible for the authors to confirm whether people would get accustomed
to overall stressful situations, but it can be stated that when a new stress condition
is reapplied, the impact is less than the first application.
References
1. Andreassi, J.L.: Psychophysiology: Human Behaviour & Physiological Response. Fourth
edn. Lawrence Erlbaum Associates, London (2000)
2. Myrtek, M., Weber, D., Brugner, G., Muller, W.: Occupational stress and strain of female
students: results of physiological, behavioral, and psychological monitoring. Biological Psy-
chology 42 (1996) 379–391
3. Krantz, D.S., B.Manuck, S.: Acute psychophysiologic reactivity and risk of cardiovascular
disease: A review and methodologic critique. Psychological Bulletin 96 (1984) 435–464
4. Sloan, R.P., Korten, J.B., Myers, M.M.: Components of heart rate reactivity during mental
arithmetic with and without speaking. Physiology & Behavior 50 (1991) 1039–1045
5. Johnston, D.: The measurement of heart rate, posture, and motion in the study of psycholog-
ical stress. In: IEE Colloquium on Data Logging of Physiological Signals. (1995) 7/1–7/4
6. Tompkins, W.J.: Electrocardiography. In Tompkins, W.J., ed.: Biomedical Digital Signal
Processing, C-Language Examples and Laboratory Experiments for the IBM PC. Prentice
Hall, New Jersey (1993)
7. Gohara, T., Mizuta, H., Takeuchi, I., Tsuda, O., Yana, K., Yanai, T., Yamamoto, Y., Kishi, N.:
Heart rate variability change induced by the mental stress: the effect of accumulated fatigue.
In: Biomedical Engineering Conference, 1996., Proceedings of the 1996 Fifteenth Southern.
(1996) 367–369
8. Sims, J., Vashishtha, D., Rani, P., Brackin, R., Sarkar, N.: Stress detection for implicit
human-robot co-operation. In: World Automation Congress, 2002. Proceedings of the 5th
Biannual. Volume 14. (2002) 567–572
9. Seong, H., Lee, J., Shin, T., Kim, W., Yoon, Y.: The analysis of mental stress using time-
frequency distribution of heart rate variability signal. In: Engineering in Medicine and Bi-
ology Society, 2004. EMBC 2004. Conference Proceedings. 26th Annual International Con-
ference of the. Volume 1. (2004) 283–285 Vol.1
10. Pagani, M., Lucini, D., Rimoldi, O., Furlan, R., Piazza, S., Binacardi, L.: Effects of physical
and mental excercise on heart rate variability. In Malik, M., Camm, A.J., eds.: Heart rate
variability. 1 edn. Futura Pub. Co, Armonk, NY (1995)
11. Bernardi, L., Wdowczyk-Szulc, J., Valenti, C., Castoldi, S., Passino, C., Spadacini, G.,
Sleight, P.: Effects of controlled breathing, mental activity and mental stress with or without
verbalization on heart rate variability. Journal of the American College of Cardiology 35
(2000) 1462–1469
12. Tomaka, J., Blascovich, J., Swart, L.: Effects of vocalization on cardiovascular and electro-
dermal responses during mental arithmetic. International Journal of Psychophysiology 18
(1994) 23–33
76
13. Walsh, J.J., Wilding, J.M., Eysenck, M.W.: Stress responsivity: The role of individual differ-
ences. Personality and Individual Differences 16 (1994) 385–394
14. Rangayyan, R.M.: Event detection. In: Biomedical Signal Analysis: A Case-Study Ap-
proach. IEEE Press Series in Biomedical Engineering. IJohn Wiley & Sons (2002) 187–190
15. Afonso, V.X.: Ecg qrs detection. In Tompkins, W.J., ed.: Biomedical Digital Signal Process-
ing, C-Language Examples and Laboratory Experiments for the IBM PC. Prentice Hall, New
Jersey (1993) 236–264
16. Jacobson, A.: Auto-threshold peak detection in physiological signals. In: Engineering in
Medicine and Biology Society, 2001. Proceedings of the 23rd Annual International Confer-
ence of the IEEE. Volume 3. (2001) 2194–2195 vol.3
17. Niskanen, J.P., Tarvainen, M.P., Ranta-aho, P.O., Karjalainen, P.A.: Software for advanced
hrv analysis. Computer Methods and Programs in Biomedicine 76 (2004) 73–81
77