The Effect of Halothane on the Amplitude and Frequency Characteristics of Heart Sounds in Children

Abstract

Although continuous auscultation has been used during surgery as a monitor of cardiac function for many years, the effect of anesthesics on heart sounds has never been quantified.We determined the root mean squared amplitude and frequency characteristics (peak frequency, spectral edge, and power ratios) of the first (S1) and second (S2) heart sounds in 19 healthy children during induction of anesthesia with halothane. In all patients, halothane decreased the amplitude of S1 (R² = 0.87 +/- 0.12) and S2 (R² = 0.66 +/- 0.33) and the high-frequency components (>80 Hz) of these sounds. These changes were clearly audible and preceded decreases in heart rate and blood pressure. The spectral edge decreased for S1 in 18 patients (R² = 0.73 +/- 0.24) and for S2 in 13 patients (R² = 0.58 +/- 0.25). Peak frequency did not change. The rapidity with which myocardial depression and its associated changes in heart sound characteristics occurred confirms that continuous auscultation of heart sounds is a useful clinical tool for hemodynamic monitoring of anesthetized infants and children. Implications: Heart sound characteristics can be used to monitor cardiac function during halothane anesthesia in children. The changes occur rapidly and precede noticeable changes in heart rate and blood pressure.

(Anesth Analg 1999;88:263-7)

Although continuous auscultation during anesthesia has been used for almost 90 yr ^[1] and anesthesiologists have stated in various texts that cardiac function can be assessed by using a precordial stethoscope ^[2-5], the amplitude and frequency characteristics of the heart sounds during anesthesia have never been studied. The amplitude of the first heart sound (S1) has been shown to be affected by cardiac contractility and plasma catecholamine levels in awake subjects ^[6], and the second heart sound (S2) has been correlated with blood pressure and inotropic state ^[7]. Because halothane causes dose-dependent myocardial depression ^[8], (¹) we hypothesized that inhaled halothane, in clinically relevant doses, would cause measurable and consistent changes in heart sound characteristics. We therefore recorded and analyzed the heart sounds of children undergoing anesthetic induction with inhaled halothane.

(¹) Diaz JH, Lockhart CH. Is halothane really safe in infancy? [abstract]. Anesthesiology 1979;51:S313.

Methods

After approval of our institution's committee on research involving human subjects and after informed parental consent was obtained, we studied the heart sounds of 19 ASA physical status I patients aged 6 mo to 12 yr presenting for urologic surgery. The patients were assigned to three age groups: I (6 mo to 2 yr), II (2-5 yr), and III (5-12 yr). Children with cardiac disease, active upper respiratory illness, or contraindication to inhaled induction with halothane were excluded.

No preoperative medications were given, and all patients had received nothing by mouth for at least 6 hr. In the operating room, a standard metallic 2-cm diameter precordial stethoscope was placed on the anterior chest wall, 1 cm medial to the left nipple. Blood pressure cuff, pulse oximeter, and electrocardiogram were applied; baseline measurements and sound recordings were made. Inhaled anesthesia was then achieved using a standard anesthesia circuit (North American Drager, Telford, PA). Nitrous oxide (70% in oxygen) was used initially to facilitate halothane delivery. The vaporizer setting was gradually increased (every three breaths) until the 5% setting was reached. The nitrous oxide was discontinued on loss of consciousness. The attending anesthesiologist (MAN) was instructed to perform the induction so as to accomplish endotracheal intubation without facilitation with muscle relaxants (relatively deep induction). Blood pressure was determined and recorded automatically each minute (vide infra). Pulse, oxygen saturation, end-tidal carbon dioxide, and inspired and expired anesthetic gas concentration were recorded at 15-s intervals using a custom computerized automated record keeping system ^[10]. Expired gas concentrations were measured by infrared spectroscopy (Capnomac Ultima; Datex, Helsinki, Finland), and the age-corrected minimum alveolar anesthetic concentration (MAC) equivalent was later calculated according to the findings of Gregory et al. ^[11] and Lerman et al. ^[12]. In each patient, spontaneous ventilation was maintained until the final 2 min of induction, at which time it was either assisted or controlled.

For the duration of the induction and before intubation, sound samples (15-s duration) were recorded each minute. This was accomplished with an omnidirectional condenser microphone with a flat frequency response between 20 and 20,000 Hz (GLM100; Crown International, Elkhart, IN) and a soundproof stethoscope adapter. The audio signals were amplified and routed to a personal computer (sampling rate 11,025 Hz, 8-bit resolution) and stereo headphones for continuous monitoring. The waveforms were processed with low-pass filtering with a cutoff at 150 Hz to minimize high-frequency noise. The resulting waveforms were later analyzed with mathematics software (Igor Pro; Wavemetrics, Inc., Lake Oswego, OR).

Three representative audio samples were selected for each patient: baseline, mid-induction (approximately 1.5 MAC expired halothane), and end-induction (approximately 3.0 MAC expired halothane). Each sample consisted of the sound associated with at least three complete respiratory cycles. From each sample, three representative examples of the S1 and S2 were selected by visual inspection of the characteristic waveforms ^[13]. All calculations were performed using arbitrary linear digital quantizations. Root mean squared (RMS) amplitude was calculated and fast Fourier transform (FFT) was performed using a Hanning window. Peak frequency (frequency of maximal amplitude) was derived from the FFT. Power spectral density was calculated so that the 97% spectral edge and 20-Hz power ratios could be determined.

The relationship between the end-tidal halothane concentration and the S1 and S2 amplitude, peak frequency, spectral edge, and power ratios were analyzed using linear regression. Analysis of variance was used to determine differences in baseline heart sounds and regression characteristics among patients assigned to the three age groups. P values <0.05 were considered statistically significant; results are presented as mean +/- SD.

Results

There were 10 children in Group I, 4 in Group II, and 5 in Group III. The baseline heart sound characteristics and hemodynamic variables are shown in Table 1. Heart sounds were loudest in the older patients (Group III). The heart sounds of the youngest patients (Group I) showed a greater presence of high-frequency components than the other two groups (higher spectral edge). This resulted from contributions in the 60-80 and 80-100 Hz ranges, as illustrated by the power ratios shown in Figure 1 and Figure 2. Baseline peak frequencies were not statistically different among the age groups. In all patients, S2 exhibited higher frequencies than S1, as indicated by a higher spectral edge (P < 0.05). Peak frequency and RMS amplitude were not statistically different for S1 and S2 in any age group.

Table 1: Baseline Hemodynamic Variables and Heart Sound Characteristics

Figure 1: Baseline S1 frequency power ratios of the three age groups. Error bars indicate 95% confidence limits.

Figure 2: Baseline S2 frequency power ratios of the three age groups. Error bars indicate 95% confidence limits.

Halothane caused a dramatic dose-dependent decrease in S1 and S2 amplitude in all patients. As end-tidal halothane concentration increased from baseline to 3.0 MAC, the S1 amplitude decreased 66% +/- 15% (R² = 0.87 +/- 0.12) and S2 amplitude decreased 46% +/- 21% (R² = 0.66 +/- 0.33). These changes were clearly audible, occurred rapidly, and were followed by corresponding decreases in heart rate and blood pressure. The amplitude changes correlated more closely with hemodynamic variables than did frequency changes (Table 2). Characteristic heart sound waveforms at varying halothane end-tidal concentrations are shown in Figure 3.

Table 2: Matrix of the Correlation Coefficients Between Hemodynamic Variables and Heart Sound Characteristics

Figure 3: Representative heart sound waveforms for baseline, 1.5 minimum alveolar anesthetic concentration (MAC), and 3.0 MAC end-tidal halothane concentrations (Patient 1, Group I).

Halothane also caused a decrease in the spectral edge of S1 in 18 patients (R² = 0.73 +/- 0.24) and of S2 in 13 patients (R² = 0.58 +/- 0.25). This frequency "shift to the left" was particularly pronounced in the S1 of the Group I patients and is illustrated in Figure 4. There was no significant change in peak frequency associated with halothane in any age group.

Figure 4: Fast Fourier transform tracings for Patient 1 at baseline (A), 1.5 minimum alveolar anesthetic concentration (MAC) (B), and 3.0 MAC (C). Overall amplitude and high frequency components are damped by increasing halothane concentration, whereas peak frequency (the frequency of maximal amplitude) remains constant (25 Hz).

Discussion

Phonocardiography research has been focused on the determination of normal heart sound production, as well as abnormal sounds such as murmurs and clicks. Relatively little work has been done to investigate the sounds of the normal heart under abnormal conditions (stress testing, anesthesia). Luisada et al. ^[6] suggested that heart sounds should be studied during anesthesia because the changes that occur during stress testing are highly consistent and strongly suggestive of changes in myocardial function. They also stated that heart sound changes during stress may be more rapid and sensitive than changes in heart rate and blood pressure. Our investigation is a first attempt to characterize the heart sound changes that occur as a direct result of anesthetic induced myocardial depression. We chose to study children for three reasons. First, the thin chest wall provides for favorable sound transmission characteristics. Second, continuous auscultation is still used and advocated as a cardiac monitoring tool in children. Finally, particularly in younger children, decreased myocardial contractility caused by halothane can be significant ^[9].

The most widely accepted theory for the genesis of heart sounds is the "cardiohemic model," which states that the sounds are produced by the vibration of the entire heart and its contents ^[13]. This vibration is triggered by valve closure (the mitral valve for S1, the aortic and pulmonic valves for S2). The amplitude of these sounds depends on the force with which the valves close, which, in turn, depends on the pressure gradient across the valve at the time of closure. Also of significance is the volume of the heart and its contents, which determines the resonance of the cardiohemic system. The properties of the chest wall are crucial in the transmission of sound from inside the thorax to the surface of the chest; we have noted in our previous studies comparing esophageal and precordial breath sounds that the chest wall is a low-pass filter ^[14]. The frequencies present in heart sounds are probably determined by the volume of the vibrating mass (smaller volume has a higher resonance frequency) and the tension generated in the walls of the heart and great vessels. This explains the fact that S2 is normally of higher frequency than S1 (the aorta is of lower volume than the heart) and that younger children exhibited higher heart sound frequencies than older children.

In adult patients undergoing stress testing, heart sound amplitude has been shown to be directly related to myocardial contractility ^[6]. Halothane caused a decrease in the amplitude of S1 and S2 in our study. This is illustrated in Figure 3, which shows a progressive decrease in size of the amplitude waveforms for Patient 1 at baseline and 1.5 MAC and 3.0 MAC end-tidal halothane concentrations. This phenomenon is easily explained by the known effects of halothane on myocardial contractility. The decreased rate of pressure increase in the left ventricle during systole likely results in decreased force of closure of the mitral valve and, thus, a decreased amplitude of S1. Two mechanisms would explain the decrease in S2 amplitude associated with halothane. First, myocardial depression results in an increase in left ventricular pressure during diastole, decreasing the driving pressure for aortic valve closure. This has been noted in awake adult patients with mild congestive heart failure ^[7]. Halothane also decreases systemic blood pressure (primarily via its negative inotropic effects), further decreasing the force of valve closure. The loss of high-frequency components of S1 and S2 probably results from a decrease in wall tension in the heart and great vessels. The use of frequency components as an indicator of wall tension and intravascular pressure has been described by other investigators ^[15] but has not to our knowledge, been reported to characterize the inotropic state of the heart.

Various mathematical methods have been used to describe heart sounds, including RMS amplitude, FFT, and wavelet transform ^[16]. Although there is no general agreement on the best method of analysis, the time domain (RMS amplitude) and the frequency domain (FFT) methods have both proven useful ^[13]. We used FFT in the calculation of peak frequency and power spectral density (the square of the FFT). The latter allowed us to determine the spectral edge and power ratios. Although spectral edge and power ratios have not been previously used in the analysis of heart sounds, we find them descriptive and have demonstrated their value in the quantification of breath sounds ^[14]. Because peak frequency is a descriptor of only a single point, it was therefore not a useful factor in describing heart sound changes resulting from variations in myocardial contractility (Figure 4). In this investigation, hemodynamic variables (heart rate and blood pressure) correlated more closely with amplitude than with frequency (Table 2). This may be explained by the fact that amplitude is primarily determined by one factor-force of valve closure-whereas frequency depends on the force of closure, heart volume, and the resonance frequencies of the heart and great vessels. Thus, differences in heart size and intravascular volume status could explain the greater variability (and, thus, weaker statistical correlation) in frequency characteristics than amplitude.

Although the use of continuous auscultation has recently declined ^[17], it is still advocated by many anesthesiologists, particularly pediatric specialists. Today, digital technology allows the rapid and inexpensive analysis of sound data, which will probably lead to increased understanding of the genesis of heart sounds and, ultimately, automated computerized auscultation. Further studies during anesthesia are required to determine the exact relationship between cardiac events and the sounds produced (correlation with echocardiography), the relationship between intracardiac pressures and heart sound characteristics, and the effect of chest wall characteristics on sound transmission.

In this investigation, the changes in heart sounds were clearly audible and preceded changes in heart rate and blood pressure. However, without the use of continuous invasive monitoring, our model did not allow precise verification of this timing. The rapidity with which myocardial depression and its associated changes in heart sound characteristics occurred confirms that continuous auscultation of heart sounds is an extremely useful clinical tool for hemodynamic monitoring of anesthetized infants and children.