Indian Journal of Physiology and Pharmacology Original Article Study of Heart Rate Variability in Control and Hypertensive Subjects Manasi Sawant1, G. D. Jindal2*, Sandhya Agarwal3 and Alaka K. Deshpande4 Department of 1Biotechnology,
MGM Institute of Health Sciences,
Kamothe, Navi Mumbai, India.
*Corresponding author : |
Abstract
Autonomic nervous system comprising sympathetic and parasympathetic divisions has important role in regulating the cardiovascular system. Heart rate variability has been shown to give objective assessment of autonomic function. Peripheral Pulse Analyzer has been used to study heart rate variability in controls and hypertensive subjects for understanding manifestations of hypertension on autonomic activity. Subjects have been divided in two age groups; 18-30 years and 31-44 years and variability parameters have been compared with respect to gender stratification, age stratification and disease stratification. Statistical analysis has shown marked reduction in the coefficient of variation for variability parameter expressed as logarithm (to the base 10) in comparison to raw or average value of the parameter and has yielded higher discriminatory strength in various stratified groups. Excluding age and gender sensitive parameters, significant difference is observed in amplitude of low frequency component for lower age group male/female hypertensives and amplitude as well as area of low frequency component for higher age group female hypertensives. These observations are in agreement with similarly targeted previous studies. Higher age group male controls and hypertensives, however, could not be discriminated by variability study probably due to similar changes manifested by senility and hypertension. Thus amplitude and area of low frequency component in heart rate variability spectrum is identified as hypertension specific parameter. |
Introduction
Autonomic nervous system is a control system that acts unconsciously and regulates bodily function such as heart rate, blood pressure, respiratory rate, digestion, contraction of involuntary muscles, size of the pupil, secretion of glands etc. It comprises of sympathetic and parasympathetic divisions having opposite responses. Stimulation of sympathetic nervous system increases heart rate, constricts the blood vessels, decreases gastro-intestinal motility and so on, whereas that of parasympathetic induces opposite effect. The variation in sympathetic and parasympathetic activity can be studied through variability in physiological parameters (1). The most explored physiological variability is that of heart rate (HRV); other variabilities investigated are those of peripheral blood flow variability (PBFV), systolic and diastolic blood pressures (SBP, DBP) and peripheral pulse morphology index (MIV). Long term variability (usually 24 hours) is dominant with ultra low frequency (ULF) representing circadian, thermoregulatory rhythms etc. Short term variability gives 3 spectral components namely Very Low Frequency (VLF), Low Frequency (LF) and High Frequency (HF) peaks in frequency ranges 0.004-0.039, 0.04-0.15 and 0.15- 0.4 Hz respectively. Common consensus about the genesis of these frequency components is that HF is marker of vagal modulation and LF is marker of sympathetic modulation (2). Little is known about VLF; some researchers have associated this component with activity of renin-angiotensin (3). Manifestations of clinical conditions like myocardial infarction, hypertension, diabetes, renal failure, cardiac arrhythmia etc., have been reviewed by Acharya et al (4). They have also reviewed effect of smoking, alcohol, beta blockers, calcium channel blockers and sleep on HRV. In general, there is decrease in total power (TP) and HF power in HRV in majority of diseases. For instance HF power is seen to be reduced in myocardial infarction, hypertension and diabetes. In addition to these general manifestations, large numbers of studies have been conducted involving particular disease in search of specific manifestation. Akselrod et al (3) used nona-peptide converting enzyme inhibitor to block renin-angiotensin system in 4 un-anesthetized dogs and observed 2 to 4 fold increase in area under VLF in 3 out of 4 dogs. Baroreflex sensitivity assessment by bradycardia and tachycardia has shown inverse relation to blood pressure variability and direct relation with heart rate variability (5, 6). Baroreflex response sensitivity is evidenced by increase in HF power (7, 4). Power spectral density (PSD) analysis of heart rate and arterial pressure has shown increase in LF component in the presence of hypertension (8, 9, 10). HRV spectrum analysis at workplace study has revealed an increase in LF/(VLF+HF) during freeze frame and decrease during internal coherence (11). New onset hypertension in 931 male subjects has been observed to change HRV spectrum (LF) in contrast to that in 1111 female subjects (12). In normotensive, low HRV has been associated with higher risk of hypertension (13). Mean variability and HF component is observed to decrease significantly in hypertensives (14). Da Silva et al have observed that LF power decreases and LF/HF increases before treatment, which regress to normal after treatment. HF decreases after treatment (15). LF/HF ratio has significantly increased during mental stress in hypertensives in contrast to that in normal subjects (16). Lufti et al have shown that systolic, diastolic and mean blood pressure correlate positively with LF and LF/HF ratio; SBP correlates negatively with HF and DBP correlates negatively with HF and total power (17). Coexistence of hypertension with type- 2 diabetes has shown significant change in heart rate recovery (18); it has synergistic effect causing worsening of autonomic recovery. Similar work by Solanki et al on Gujarati population has shown decrease in HRV parameters, which is not affected by optimum pressure or glycemic control (19, 20). Half of the studies cited above have followed the
Task Force Recommendations for the study of heart
rate variability described by Camm et al (21). As
described above, LF power and LF/HF ratio in general
increase with increase in BP; VLF and HF power are
associated with renin-angiotensin system and baroreflex sensitivity respectively. Some of the above
studies have given conflicting views due to use of
absolute power of HRV components in place of
normalized power. Large values of standard deviation
of HRV parameters and non-segregation with respect
to age and sex in the above studies make it
necessary to have a fresh look on the manifestation
of hypertension on HRV spectrum. In present study
we aim to evaluate changes in spectrum of HRV in
controls and early untreated hypertensive subjects in segregated groups. Results of the study are briefly
described in this paper.
|
Materials and Methods
The study has been carried out at Bio-Medical Engineering Department of MGM College of Engineering and Technology, Kamothe, Navi Mumbai and has been approved by the human ethics committee of MGM Institute of Health Sciences (Deemed University) in its meeting held on 25th March 2015 item no. 2. The controls and hypertensive subjects, in the age group of 18 to 45 years, were derived from students and staff population of the institute conforming to inclusion and exclusion criteria. Smokers, tobacco chewers, those on regular medication and those suffered major sickness in the past were excluded from the study. Also subjects with history of cardiovascular (except hypertension) or autonomic sickness were excluded from the study. Initially subject’s information, past history and consent were taken. The subjects were then rested comfortably for 15 minutes in supine position. Continuing in supine position blood pressure is measured twice with digital BP monitor (Omron HEM- 7121) of which second reading has been considered for including the subject in the study. Blood pressure less than 130/80 were included in the control group. Those having >140/85, consistently on two different days, were taken in the hypertension group. These subjects have thus been freshly detected hypertensives, who have not been subjected to any treatment. Data has been stratified with respect to age, gender and hypertension in 8 groups (CML, CFL, CMH, CFH, HML, HFL, HMH and HFH). Demographic characteristics of these subjects are given in Table-I. Minimum sample size of 16 has been arrived at considering precision level to be 50% of the sample standard deviation. For assessment of heart rate variability, peripheral
blood flow (PBF) has been recorded in all the
subjects with the help of Peripheral Pulse Analyzer
(an R&D product of Bhabha Atomic Research Centre
(BARC), Mumbai), shown in Fig. 1. PBF recordings
have been made following the protocol and guidelines
recommended for HRV studies (22). Five consecutive recording of 275 seconds duration each (with an inter-recording gap of 2 minutes) have been done after 10
minutes of rest to the subject for the first reading.
Two hours gap was provided for recording specifically
after breakfast or lunch. With the subject in supine,
carrier electrodes C1 and C2 were applied on the
right upper extremity just below the elbow and palm
respectively, in form of loop around the limb. Sensing
electrodes S1 and S2 were placed 5 cm apart around
the wrist as shown in the figure. Subject’s information
was filled before acquisition. A Click on ACQUIRE
button (Fig. 2) in the application software starts data
acquisition, which stops after 275 seconds by default.
In case of motion artifact, it can be stopped in
between and restarted. Data gets saved with
subject’s information in a file. The acquired data is
then processed in another graphical user interface
(GUI) shown in Fig. 3. After loading pre-acquired data in the application, one can go to PROCESSING Panel (Fig. 3). Entire signal can be viewed in the top graph by panning with the help of arrows provided on right side. The systolic peak is marked by placing the cursor on the second peak and then clicking on LOCATE PEAKS. Vertical red lines appear on the peak positions as seen in the graph. The second graph, below the raw signal, displays time elapsed between two peaks (equivalent to approximate RR interval of ECG) as a function of time. By clicking on any irregularity in lower graph, one can see corresponding signal in the upper graph. Editing of peak can be done with the help of arrow buttons and insert/delete buttons provided in the bottom towards right. Edited data is then saved. From the application, one can move on to DISPLAY panel, which displays Heart Rate Variability (HRV) spectrum as shown in Fig. 4. Variability in time domain and frequency domain can be seen on left and right graph respectively. The computed variability parameters are displayed on right side of the panel. By clicking on SAVE EXCEL, data is transferred to designated excel sheet. A click on quit button exits the application. Data saved in Excel sheet is then processed for statistical analysis. To begin, five values of a particular HRV parameter (obtained from 5 recordings in a subject) are averaged. The average values, hereafter referred to as Average (Avg), for all the parameters in a particular subject are thus obtained. Mean and standard deviation (SD) values of these parameters are then obtained from this data in all the subjects in a particular group. The process is repeated for all the groups. Mean thus obtained has lowered value of SD in comparison to individual values as reported by Jindal et al (23). ANOVA and Tukey’s HSD (honestly significant difference) test were performed to find out the significant difference in the data stratified with respect to age, gender and disease. In view of the large scatter even in the averaged data, logarithm of every individual value to the base 10 was taken, hereafter referred to as Log. The whole statistical analysis performed on Avg data has been repeated on Log data. |
Results
Table II gives group mean and standard deviation (SD) values computed from Avg and Log data. As can be seen coefficient of variation has come down drastically for Log data. Histogram plot for a particular parameter in a particular group has been observed to be more symmetric for Log data than that of the Avg data. Similarly Table III gives ANOVA and Tukey’s HSD statistic values for Avg as well as Log data. Probability values are given for eight comparisons between various groups stratified with respect to gender, age and disease. For instance male and female controls are compared in the lower age groups (CML_CFL) and in the higher age group (CMH_CFH) for gender stratification. Similarly CML_CMH and CFL_CFH represent age stratification and CML_HML, CFL_HFL, CMH_HMH and CFH_HFH represent disease stratification. As can be seen from the table, ANOVA shows an overall significant difference in HRV parameters in the Avg as well as Log data derived from 8 different groups. Fine difference is revealed from Tukey’s HSD analysis, which shows honestly significant difference between any two groups. Significantly different values are highlighted as “bold”. Gender stratification shows no significant difference in Avg data from control subjects in respective age group, in contrast to that in Log data showing significant difference in 3 out of 16 comparisons. Similarly Age stratification shows significant difference in 3 and 8 comparisons (each out of 16) for Avg and Log data respectively. Thus it is seen that Log data is more sensitive in identifying significant difference in comparison to Avg data. For disease stratification, comparison is made between control and hypertensive groups for respective gender and age. For instance control-male-lower (CML) has been compared with hypertensive-male-lower (HML) and so on. Significant difference is noted in 8 and 16 comparisons (each out of 32) in Avg and Log data respectively, suggesting higher sensitivity of Log data. Excluding age and gender sensitive parameters, significant difference (highlighted by italics and bold) is observed in A_VLF for CFH_HFH; Amp_LF for CFL_HFL; A_HF for CFL_HFL in Avg data. Similarly significant difference (highlighted by italics and bold) is observed in Amp_LF for 3 comparisons (out of 4); and A_LF for CFH_HFH in Log data. |
Discussion
Gender stratification shows no significant difference in Avg data from control subjects in respective age group, in contrast, Log data shows significant difference in 3 out of 16 comparisons. Similarly Age stratification shows significant difference in 3 and 8 comparisons (each out of 16) for Avg and Log data respectively. Thus Log data appears more sensitive than Avg data. This is probably due to much lower coefficient of variation of the Log data (12.04% in contrast to 82.6% for TP). Over all 3 comparisons are significantly different in gender stratification, suggesting that gender stratification can be dispensed with little risk of error (9.4%). However age stratification need be maintained as the risk of error becomes (34.4%). Maintaining age and gender stratification, comparison of data in hypertensive subjects with that of respective control subjects shows significant difference in 8 and 16 comparisons (each out of 32) in Avg and Log data respectively, again suggesting higher sensitivity of Log data for the reason specified for control data. Therefore further discussion is limited to only Log data. Out of these comparisons, age and gender sensitive parameters need be eliminated, as these parameters (TP, RR_mean, Amp_VLF, A_VLF, Amp_HF and A_HF) are not changing due to hypertension per se. Thus significant difference observed in Amp_LF for lower age group male/female hypertensives and Amp_LF as well as A_LF for higher age group female hypertensives appears to be specific for hypertension. However, no parameter is specific for higher age group male hypertensives. It may probably be due to reason that senile changes are similar to that of hypertension in higher age group males. Since higher age group female hypertensives have two specific parameters, namely Amp_LF and A_LF, and male hypertensives in higher age group do not yield a specific parameter, it appears that gender stratification cannot be dispensed. In previous studies (8, 9, 10) increase in LF component and decrease in HF are associated in hypertension. In present study LF component is observed to be affected either in amplitude or in area. Thus this study is in agreement with previous observations. Thus this study indicates that increase in LF amplitude or/and area can be considered specific of hypertension and it can be used for early detection of hypertension. Conclusion The study demonstrates that Log data gives higher
discrimination in variability parameters in comparison
to raw or average data. Amplitude or power of LF component in HRV spectrum is observed to be
specifically related to untreated hypertension in young
males and young as well as old females and therefore
can be used to detect hypertension in these |
Aknowledgement
The authors are thankful to MGM Institute of Health
Sciences, Kamothe Navi Mumbai and MGM College
of Engineering and Technology, Kamothe, Navi
Mumbai, for their encouragement throughout the
work. Authors are grateful to Board of Research in
Nuclear Sciences (BRNS), Department of Atomic
Energy, Government of India, for providing us PPA
instrument under the project entitled “Development
of web-based pilot database and portal for data
analysis of autonomic nervous system”, which has
been used in this study for acquiring data in the
subjects. |
References
|