Power Spectrum Of Heart Rate Variability Research Articles

Model based and experimental investigation of respiratory effect on the HRV power spectrum

The role of respiration in the genesis of heart rate variability (HRV) has been the subject matter of many experimental and modeling studies. It is widely accepted that the high frequency (HF) peak of a HRV power spectrum, which is centered at the average respiratory frequency, is caused by mechanisms activated by respiration. On the other hand, there is a debate on the possible role of respiration in the genesis of the low frequency (LF) peak which is usually centered around 0.1 Hz. In this study, a comprehensive cardiorespiratory interaction model is used to test various hypotheses regarding the role of respiration in the LF peak of HRV. In this model, chest and abdomen circumference signals and lung volume signal are used as respiratory inputs. Simulations are made for periodic, spontaneous and slightly irregular respiratory patterns, and it is observed that the more low frequency (LF) power there in the respiratory signals, the more LF power there in the model-predicted HRV. Experiments on nine volunteers are also performed for the same respiratory patterns and similar results are observed. Furthermore, the actual measured respiratory signals are input to the model and the model predicted and the actual HRVs are compared both in time domain and also with respect to their power spectra. It is concluded in general that respiration not only is the major contributor to the genesis of the HF peak in the HRV power spectrum, but also plays an important role in the genesis of its LF peak. Thus, the LF/HF ratio, which is used to assess sympathovagal balance, cannot be correctly utilized in the absence of simultaneous monitoring of respiration during an HRV test.

Estimation of HRV spectrogram using multiple window methods focussing on the high frequency power

In this paper, the results of different multiple window spectrum analysis methods are compared in the estimation of heart rate variability (HRV) power spectra, in the high frequency band (HF), around 0.25 Hz, related to respiratory sinus arrhythmia (RSA). The evaluation is performed by simulating different spectrum shapes and peak frequency locations and calculating the mean squared error of a frequency range close around the strongest spectral peak. The results show that it is preferable to use the Peak Matched Multiple Windows in most situations, but the Welch method and the Sinusoid Multiple Windows can be as reliable in certain aspects.

Cardiac vagal tone, exercise performance and the effect of respiratory training

Heart rate variability (HRV) at rest and heart rate recovery after exercise reflect cardiac vagal activity. The aim of this study was to determine whether increasing HRV during involuntary respiratory training induced by rebreathing air using a Hepburn heart and lung exerciser (HHALE) could, like exercise, improve vagal tone. Eighteen subjects (36-88 years) underwent a 6-week control period, then a 6-week training period with the HHALE following which half continued training for 6 weeks and half ceased training. Measurements were made of HRV, work at 60% predicted heart rate max for 15 min, heart rate recovery after exercise, resting blood pressure, heart rate, vital capacity and forced expiratory volume. After the first 6-week HHALE training, there was a significant increase of 13.2 +/- 5.7 nu in the high frequency peak of the power spectrum of HRV at rest, whereas, the low frequency peak decreased. Similarly, exercise performance showed a significant improvement of 0.031 +/- 0.012 J per heartbeat from a pre-training 0.128 +/- 0.022. Also, heart rate recovery after exercise significantly faster (drop in the first 20 s improving by 3.3 +/- 1.5 beats from a pre-training 12.9 +/- 1.6). The subgroup that continued training maintained or slightly improved these values. In those that ceased training the speed of heart rate recovery at the end of the exercise test returned to pre-trained levels, whereas, other responses were either maintained or decreased slightly. We conclude that training with the HHALE can, without additional exercise, increase cardiac vagal tone and exercise performance.

Optimal Electrical Stimulation Modality for Cortical Esophageal Evoked Potentials: Transmural or Intraesophageal?

Esophageal electrical stimulation using short and a relatively small number of (200 micros, 0.2 Hz, n = 25) electrical pulses generates a characteristic and well defined cortical evoked potential response (EP). There are two methods of stimulation: either through intraesophageal electrodes or with transmural electrodes. The objective of this paper is to compare EP response, sensations and heart rate variability power spectra elicited by both stimulation modalities in healthy volunteers. Our results suggest that transmural stimulation is more accurately perceived and at lower intensities, produces more reproducible peaks of higher amplitude than during intraesophageal stimulation. During either mode of esophageal stimulation, power within the high-frequency component of the heart rate variability power spectrum is enhanced.

Autonomic nervous system activity in normotensive subjects with a family history of hypertension

This study was designed to address alterations in autonomic nervous system activity in normotensive subjects with a family history of hypertension. We compared the autonomic nervous system activity in 59 normotensives with a family history of hypertension and 46 normotensives with no family history of hypertension. Skin blood flow was measured using laser-Doppler method on the nailfold skin in the resting condition, during systemic cooling and during upright tilting. Finger blood pressure, pulse and ECG were monitored by a finapres device. Heart rate, systolic pressure and microvascular flow power spectral analyses were performed using fast Fourier transformation. Baroreflex sensitivity was estimated with the sequence method. Compared to the control group, normotensives with a family history of hypertension showed significantly higher systolic pressure, decreased proportion and area of the high-frequency band of the heart rate variability power spectrum and reduced baroreflex sensitivity in the resting condition as well as a decreased proportion and area of the high-frequency band of the heart rate variability power spectrum during systemic cooling. We also proved a different time course of baroreflex sensitivity during upright tilting in the two groups. In contrast, we did not find any significant differences in the parameters of systolic pressure and microvascular variability power spectra between the two groups. Our results indicate that even normotensives with a family history of hypertension exhibit an increased ratio of sympathetic to parasympathetic activity at the cardiac level; however, they do not show any alteration of the vascular sympathetic reactivity.

Interdependency between heart rate variability and sleep EEG: linear/non-linear?

Objective: To investigate whether the interdependency between heart rate variability (HRV) and sleep electroencephalogram (EEG) power spectra is linear or non-linear. Methods: Heart rate and sleep EEG signals were recorded in 8 healthy young men. Spectral analysis was applied to electrocardiogram and EEG sleep recordings. Synchronization likelihood was computed over the first 3 non-rapid eye movement–rapid eye movement sleep cycles between normalized high frequency of RR intervals (RRI) and all electroencephalographic frequency bands. Comparison to surrogate data of different types was used to attest statistical significance of the coupling between RRI and EEG power bands and its linear or non-linear character. Results: Synchronization likelihood values were statistically greater than univariate surrogate synchronization for all sleep bands both at the individual and the group levels. With reference to multivariate surrogates, synchronization values were statistically greater at the group level and, in a majority of cases, for individual comparison except for σ and β bands. Conclusions: While all electroencephalographic power bands are linked to normalized high frequency RRI band, this interdependency is non-linear for δ, θ and α bands. Significance: Non-linear description is required to capture the full interdependent dynamics of HRV and sleep EEG power bands.

Interdependency between heart rate variability and sleep EEG: linear/non-linear?

Objective : To investigate whether the interdependency between heart rate variability (HRV) and sleep electroencephalogram (EEG) power spectra is linear or non-linear. Methods : Heart rate and sleep EEG signals were recorded in 8 healthy young men. Spectral analysis was applied to electrocardiogram and EEG sleep recordings. Synchronization likelihood was computed over the first 3 non-rapid eye movement–rapid eye movement sleep cycles between normalized high frequency of RR intervals (RRI) and all electroencephalographic frequency bands. Comparison to surrogate data of different types was used to attest statistical significance of the coupling between RRI and EEG power bands and its linear or non-linear character. Results : Synchronization likelihood values were statistically greater than univariate surrogate synchronization for all sleep bands both at the individual and the group levels. With reference to multivariate surrogates, synchronization values were statistically greater at the group level and, in a majority of cases, for individual comparison except for σ and β bands. Conclusions : While all electroencephalographic power bands are linked to normalized high frequency RRI band, this interdependency is non-linear for δ , θ and α bands. Significance : Non-linear description is required to capture the full interdependent dynamics of HRV and sleep EEG power bands.

A mathematical analysis for the cardiovascular control adaptations in chronic renal failure.

A model of baroreflex control of blood pressure (BP) is proposed in terms of a delay differential equation and this is used to predict the adaptation of short-term cardiovascular control in chronic renal failure (CRF) patients. Cardiac pump dynamics are explored by means of plots of blood flow vs. mean BP. The parameters of the model were determined from available data and from a sensitivity analysis. The model predicts stable and unstable equilibria close to the steady BP. It is shown that the unstable equilibrium point generates a quasiperiodic solution with two main harmonics for healthy subjects. We also show that the parameters for CRF patients predict solutions whose spectra exhibit a small high frequency component. This is due to the coalescence of the equilibrium points. The heart rate variability (HRV) time series and power spectra from healthy volunteers and CRF patients were compared with the model predictions. As an adequate measure of the sympathovagal balance we use the LF/HF index obtained from the power spectrum. The model allows the interpretation of the variability of the LF/HF index in terms of a specific set of cardiovascular parameters which are known to change from healthy to CRF patients. Comparisons of the changes in the LF/HF index predicted by the model are in agreement with actual observations for both the healthy and the CRF patients. These results show that the cardiac pump has a more restricted response in CRF patients. The model quantifies the cardiovascular adaptations to the CRF condition in terms of increased peripheral resistance and baroreflex delay and decreased arterial compliance, cardiac period, and stroke volume.

Autonomic Dysfunctions in Patients with Anxiety Throughout Therapy

Abstract Patterns of autonomic cardiovascular regulation were studied in patients with anxiety throughout the course of an intensive inpatient rehabilitation therapy at rest, and in response to an anxiety provocation, in order to measure reactivity changes. Patients had to meet the ICD-10 criteria: F40.0 (agoraphobia), F40.00 (agoraphobia without panic attacks), F40.01 (agoraphobia with panic attacks), or F41 (panic disorder). Thirty-eight patients (13 males, 25 females) were examined after recruitment (T1) and at the end of treatment (T2). Each laboratory session consisted of a resting baseline (5min) and two stress tests of 8min (presentation of word pairs with either neutral or with anxiety-related content). A continuous assessment of ECG, blood pressure, breathing pattern, and thoracic impedance allowed the calculation of heart rate variability (power spectra), baroreflex sensitivity, stroke volume, vascular compliance, and total peripheral resistance. In general, therapy had little impact on the parameters of autonomic cardiovascular regulation, neither on resting level nor on reactivity. Diminished baroreflex sensitivity values in our patients with anxiety were observed; in particular, 15% of the patients had values below 3ms/mmHg, which indicates an increased risk for cardiovascular mortality. Low baroreflex sensitivity in this group goes along with further changes within the cardiovascular system (higher heart rate, diminished heart rate variability, lower stroke volume, higher vascular compliance). This general effect describes altered sympathetic/parasympathetic balance and may be the consequence of illness-related changes in behavior. Our results suggest that a careful cardiovascular examination of patients with anxiety will detect patients with clinically significant cardiovascular alterations. This might lead to a more complex intervention in these patients including, for example, exercise training.

Heart rate variability and autonomic activity at rest and during exercise in various physiological conditions.

The rhythmic components of heart rate variability (HRV) can be separated and quantitatively assessed by means of power spectral analysis. The powers of high frequency (HF) and low frequency (LF) components of HRV have been shown to estimate cardiac vagal and sympathetic activities. The reliability of these spectral indices, as well as that of LF/HF ratio as a marker of autonomic interaction at rest and during exercise, is briefly reviewed. Modifications in autonomic activities induced by different physiological conditions, e.g. hypoxia exposure, training, and water immersion, have been found in HRV power spectra at rest. The changes in HF and LF powers and in LF/HF ratio observed during exercise have been shown not to reflect the decrease in vagal activity and the activation of sympathetic system occurring at increasing loads. HF peak was recognised in power spectra in the entire range of relative intensity, being responsible for the most part of HR variability at maximal load. LF power did not change during low intensity exercise and decreased to negligible values at medium-high intensity, where sympathetic activity was enhanced. There was no influence from factors such as fitness level, age, hypoxia, and blood distribution. In contrast, a dramatic effect of body position has been suggested by the observation that LF power increased at medium-high intensities when exercising in the supine position. The increased respiratory activity due to exercise would be responsible of HF modulation of HR via a direct mechanical effect. The changes in LF power observed at medium-high intensity might be the expression of the modifications in arterial pressure control mechanisms occurring with exercise. The finding of opposite trends for LF rhythm in supine and sitting exercises suggests that different readjustments might have occurred in relation to different muscular inputs in the two positions.

A study of the dynamic interactions between sleep EEG and heart rate variability in healthy young men.

We investigated the interactions between heart rate variability and sleep electroencephalogram power spectra. Heart rate and sleep electroencephalogram signals were recorded in 8 healthy young men. Spectral analysis was applied to electrocardiogram and electroencephalogram recordings. Spectral components of RR intervals were studied across sleep stages. The cross-spectrum maximum was determined as well as coherencies, gains and phase shifts between normalized high frequency of RR intervals and all electroencephalographic frequency bands, calculated over the first 3 NREM-REM cycles. RR intervals increased from awake to NREM and decreased during REM. Normalized low frequency decreased from awake to NREM and increased during REM while normalized high frequency evolved conversely. Low to high frequency ratio developed in opposition to RR intervals. Coherencies between normalized high frequency and power spectra were high for all bands. The gain was highest for delta band. Phase shift between normalized high frequency and delta differed from zero and modifications in normalized high frequency preceded changes in delta by 41+/-14 degrees. Our study demonstrates that: (1) all electroencephalographic power bands are linked to normalized high frequency; (2) modifications in cardiac vagal activity show predominantly parallel changes and precede changes in delta band by a phase shift corresponding to a lead of 12+/-5 min.

AT1 antagonism by eprosartan lowers heart rate variability and baroreflex gain

Introduction: Blockade of the renin–angiotensin system (RAS) by ACE inhibitors has been demonstrated to reduce total mortality in cardiovascular diseases. This advantage was attributed in part to changes of autonomic cardiovascular control, exemplified by an increase of heart rate variability (HRV) and baroreflex gain (BRG). We sought to assess the effects of the angiotensin type 1 (AT1) receptor blocker eprosartan on HRV and BRG. Materials and methods: In a double-blind randomized cross-over design 25 young males took eprosartan (600 mg/day) and placebo each for a period of 7 days with a wash-out period of at least 4 weeks in between. At the end of the intake phases simultaneous recordings of arterial blood pressure (AP; Finapres) and electrocardiogram (ECG) were taken. Power spectra of HRV and arterial blood pressure variability (APV) were calculated by fast Fourier transform (FFT) and served to calculate BRG. Ang-II levels were measured by radioimmunoassay. Results: Eprosartan tended to lower mean AP, it slightly increased heart rate (HR) ( p<0.05), and markedly increased circulating Ang-II levels ( p<0.01). Eprosartan diminished the total power of HRV ( p<0.05) and the BRG ( p<0.01). The low/high frequency (LF/HF) ratio of HRV and the APV were not altered. Conclusions: AT1 antagonism by eprosartan lowers heart rate variability and baroreflex gain. We speculate that these findings are due to the marked increase in circulating angiotensin II (Ang II). Further studies are needed to clarify whether angiotensin type 1 (AT1) blockers with potential actions inside the blood–brain barrier (BBB) may have different effects on HRV and BRG.

Healthy high-hostiles: reduced parasympathetic activity and decreased sympathovagal flexibility during negative emotional processing

This study was designed to assess the physiological impact of processing neutral, positive, and negative emotional stimuli among a group of low- and high-hostile individuals. Advances were made by (1) measuring both reactivity and recovery to mood induction and (2) using Heart Rate Variability (HRV) to more specifically quantify sympathovagal (Low Frequency divided by High Frequency power, or LF/HF, within the HRV power spectrum) and parasympathetic arousal (Respiratory Sinus Arrhythmia; RSA) at the myocardium. In the present study, men and women were equally divided into low- (N=30) and high-hostile (N=30) groups based on their scores on the Cook–Medley Hostility Scale. Electrocardiogram data were collected before, during, and after being given the negative, positive, or neutral version of the Affective Auditory Verbal Learning Test (AAVL). Results indicate that high-hostiles had reduced parasympathetic activity relative to low-hostiles, as measured by RSA. Moreover, relative to low-hostiles, high-hostile participants evidenced reduced sympathovagal reactivity and recovery to the negative AAVL. Results are discussed in terms of their potential value in understanding risk factors for coronary heart disease.

Frequency characteristics of long-term heart rate variability during constant-routine protocol.

The effects of such behavioral factors as physical activity, food intake, and circadian rhythm on long-term heart rate variability (HRV) in humans remain poorly understood. We therefore studied their effects on HRV using a constant-routine protocol that included simultaneous core body temperature (CBT) correction. Seven healthy subjects completed the constant-routine and daily-routine protocols, during which HRV and CBT were continuously monitored. During the constant routine, subjects were kept awake for 27 h in a semirecumbent posture with minimal physical activity; small isocaloric meals were provided every 2 h. During the daily routine, subjects carried on their lives normally. Data were analyzed using generic spectral analysis based on a fast Fourier transform; coarse-graining spectral analysis was also used to eliminate periodicity due to the regular meals for raw HRV and for the CBT-corrected HRV without circadian and/or low-frequency ultradian components. The results showed that 1) the power spectra of HRV in the constant routine and daily routine had similar power-law scalings at frequencies above approximately 10(-3.5) Hz, while 2) below that crossover frequency, HRV was smaller in the constant routine than in the daily routine, with the difference becoming significant (P < 0.05) at <10(-4) Hz, 3) coarse-graining spectral analysis eliminated diet-induced peaks in generic spectral analysis-based HRV spectra during the constant routine and emphasized the crossover at approximately 10(-3.5) Hz, and 4) CBT correction did not alter the results. Below a frequency of approximately 10(-3.5) Hz (a period >1 h), HRV is strongly influenced by behavioral factors; above that crossover frequency, HRV is behavior independent, possibly reflecting an intrinsic regulatory system.

Age-related changes in sleep-wake rhythm in dog

To investigate a sleep-wake rhythm in aged dogs, a radio-telemetry monitoring was carried out for 24 h. Electrodes and telemetry device were surgically implanted in four aged dogs (16–18 years old) and four young dogs (3–4 years old). Electroencephalogram (EEG), electromyogram (EMG) and electrocardiogram (ECG) were recorded simultaneously as parameters to determine vigilance states and an autonomic nervous function. Wakefulness, slow wave sleep (SWS) and paradoxical sleep (PS) were identified according to the EEG and EMG pattern. We also examined whether absolute powers and the low frequency-to-high frequency ratio (LF/HF) derived from the heart rate variability power spectrum could detect shifts in autonomic balance correlated with aging. The aged dogs showed a marked reduction of PS and a fragmentation of wakefulness in the daytime and a sleep disruption in the night. The pattern of 24 h sleep and waking was dramatically altered in the aged dog. It was characterized by an increase in the total amount of time spent in SWS during the daytime followed by an increasing of time spent in wakefulness during the night. Furthermore, LF/HF ratio showed a very low amplitude of variance throughout the day in the aged dog. These results suggest that the aged dog is a useful model to investigate sleep disorders in human such as daytime drowsiness, difficulties in sleep maintenance. The abnormality in sleep-wake cycle might be reflected by the altered autonomic balance in the aged dogs.

Aerobic training and cardiovascular responses at rest and during exercise in older men and women.

The effects of an intense 8-wk aerobic training program on cardiovascular responses at rest and during exercise, including heart rate variability (HRV) as an expression of autonomic modulation, were evaluated in subjects over 70 yr (mean: 73.9 +/- 3.5 yr). Before and after training in 7 men and 8 women: a) heart rate (HR), blood pressures (BPs), pulse pressure (PP), and oxygen uptake were measured at rest, during, and after exhausting incremental exercise; b) HRV power spectra were calculated at rest in supine and sitting, and during and after two submaximal constant loads (5 min). Power in low-frequency (LF, 0.04-0.15 Hz) and high-frequency (HF, >0.15 Hz) bands were expressed as a percent of total power minus power < 0.04 Hz. After training: a) at rest HR and HRV parameters (in both body positions) were unchanged, whereas BPs decreased; b) peak cycle resistance and oxygen consumption increased by 25% and 18%, respectively, but no change in maximal HR and BPs were found; c) during submaximal loads HR was unchanged at the same metabolic demand, whereas SBP and DBP were lower than before at low loads whereas PP was unchanged. LF power decreased and HF increased at oxygen uptakes above about 0.7 L.min-1 similarly before and after training; and d) recovery of all parameters was similar to pretraining and complete after 10 min The increase in exercise capacity without changes in cardiovascular parameters suggests that 8 wk of aerobic training augmented peripheral gas exchange but not delivery to muscle. The lack of effect on HRV indicates that the improvements in aerobic power and cardiac autonomic modulation, at least in subjects over 70 yr, are dissociated. Moreover, the metabolic demand seems to be the main factor for the changes in HRV power spectra that occur during exercise.

How to avoid misinterpretation of heart rate variability power spectra?

Spectral analysis of R-R Interval time series is increasingly used to determine periodic components of heart rate variability (HRV). Particular diagnostic relevance is assigned to a low-frequency (LF) component, associated with blood pressure regulation, and a high-frequency (HF) component, also referred to as respiratory sinus arrhythmia (RSA) in the HRV power spectra. Frequency ranges for parametrisation of power spectra have been defined for either component in numerous publications. Results obtained from examinations with standardised psychic load in which ECG and respiratory signal are continuously recorded and adequately processed have shown that the true individual frequency range of the HF component can be reliably determined only by means of characteristics of respiration (respiratory rate (RR), range and median value of RR, tidal depth). Respiratory rhythms are interindividually extremely differentiated and of individual-specific nature. In many cases LF and HF components may be totally superimposed on each other and, consequently, cannot be diagnostically evaluated.

Heart rate variability and arterial blood pressure variability show different characteristic changes during hemorrhage in isoflurane-anesthetized, mechanically ventilated dogs.

We assessed the changes in heart rate variability (HRV) and blood pressure variability (BPV) as indices of autonomic nervous system and volume status during hemorrhage in isoflurane-anesthetized, mechanically ventilated dogs. Nine dogs were used. They were sequentially subjected to withdrawal of 30% estimated blood volume and graded isoflurane inhalation of 1% and 2% followed by discontinuation of isoflurane and retransfusion. The power spectra of HRV and BPV were computed using the fast Fourier transformation, and were quantified by determining the areas of the spectrum in two component widths: low-frequency component (LF) (0.04-0.15 Hz) and high-frequency component (HF) (0.15-0.4 Hz). During hemorrhage and isoflurane anesthesia, both HRV-LF and HRV-HF were decreased and plateaued at the smaller concentration of isoflurane, whereas BPV-LF decreased concentration-dependently. BPV-HF showed a completely different response and increased significantly during 2% isoflurane. We speculate that HRV and BPV-LF would be affected by the autonomic nervous activity, whereas BPV-HF would depend on relative/absolute change in circulating blood volume. Power spectra of heart rate variability (HRV) and blood pressure variability (BPV) were computed using the fast Fourier transformation. The HRV and BPV showed their differential characteristics during hemorrhage, isoflurane anesthesia, and retransfusion, and would help to assess changes in autonomic nervous system and preload under mechanical ventilation.

Innocuous mechanical stimulation of the neck and alterations in heart-rate variability in healthy young adults

The present study examined the effects of cervical spinal manipulation, a widely applied form of physical therapy, which involves innocuous mechanical stimulation, on heart rate and heart-rate variability, in a cohort of healthy young adults. Using a cross-over treatment design, with a one-week washout period and, in contrast to a sham procedure, the authentic manipulation produced significant alterations in both heart rate and measures of heart-rate variability calculated from power spectrum analysis. In particular, there was an increase in the ratio of low-frequency (LF)-to-high-frequency (HF) components of the power spectrum of heart-rate variability, which may reflect a shift in balance between sympathetic and parasympathetic output to the heart.

Power spectrum of heart rate variability in exercising humans: The effect of exercise intensity

Power spectral analysis of heart rate variability (HRV) has been estimated by means of a Fourier transform method at rest in seven healthy men and three women, during a 30‐minutes steady state cycle exercise test at 40%, 50%, 60%, 70% and 80% respectively, of maximal heart rate reserve (MHRR) defined as (HRmax ‐ HRrest). Total power (PT) of HRV was divided into three separate components (PHf, Plf, and Pvlf). All four indices for HRV decreased significantly with increasing exercise intensity. The fractional Pvlf (%VLF), VLF/LF ratio, and VLF/HF ratio increased significantly with increasing exercise intensity. On the contrary, %LF and %HF decreased significantly with increasing exercise intensity. The following conclusions were made: (1) the contribution of parasympathetic activity to the power spectrum of HRV decreases with exercise intensity; (2) the change in %VLF, VLF/LF ratio, and VLF/HF ratio respectively, can be used as indices of change in sympathovagal balance during exercise; (3) the change in the LF/HF ratio cannot be used during exercise as an index of change in sympathovagal balance, since a change in vagal activity effects predominantly both numerator and denominator of the LF/HF ratio.