1 Introduction MYOCARDIAL INFARCTION (MI) may cause imbalances in the autonomic nervous system (ANS) that persist for weeks to months following the event. Impaired parasympathetic responses to facial immersion, cold stimulus and isometric handgrip exercise tests have been demonstrated in patients studied 3-6 months after MI (RYANe t al., 1976). Reduced heart rate variations, suggestive of reduced parasympathetic activity, were found in recent MI (6-26 weeks post-MI) patients (KLEIGERe t al., 1987; BIGGER et al., 1988). Reduced baroreflexes were also demonstrated in 15-day post-MI patients, a response attributed primarily to reduced vagal activity (SCHWARTZe t al., 1988a; b; LA ROVERE et al., 1988). In patients studied within 2 weeks of their MI, using heart rate variability (HRV) analyses, hyperactive sympathetic and underactive parasympathetic responses have been reported (LOMBARDIet al., 1987). Autonomic imbalance in MI survivors has been implicated as a risk factor in sudden death (KLEIGERet al., 1987; SCHWARTZ et al., 1988b) and is therefore an important clinical parameter. The present study applies HRV analysis to patients during recovery from MI, in the periods 2-6 weeks and 1 year post-MI. The patients were studied in a relatively drug-free state to accurately assess their baseline levels of autonomic activity. The simple
analyses developed here, based on postural effects on HRV, have confirmed earlier studies that associated a depressed level of parasympathetic activity with the MI recovery period. A preliminary report has appeared (TANGELLAand CRAELIUS,1989). 2 M a t e r i a l s and m e t h o d s
2.1 Patients Three groups each of eight male subjects, were compared: Group A, consisting of patients who had MI within the previous 2-6 weeks of study; Group B, consisting of patients who had MI more than 1 year previously; and Group C, consisting of age-matched controls free of cardiac disease. Diagnosis of myocardial infarction was based on clinical electrocardiographic and enzymatic criteria (Table 1). Patients who had documented episodes of sustained arrhythmia, congestive heart failure, renal insufficiency, neuropathy or excessive premature beats were Table 1 Clinical characteristics
Group A (recent MI) 8 64 + 8 3.5 weeks
Group B (post MI) 8 63 + 5 25 months
Group C (controls) 8 68 _ 12 none
1
3
7
5
none none
First received l Oth January and in final form 18th July 1991
Number of subjects Age (mean + SD), years Post-MI period (mean) Site of MI Anterior Inferior
9 IFMBE: 1992
M I = m y o c a r d i a l infarction; S D = standard deviation
Correspondence should be addressed to Dr W. Craefius, Department of Biomedical Engineering, Rutgers University, Piscataway, NJ 08855, USA.
Medical & Biological Engineering & Computing
July 1992
385
excluded. N o n e of the patients was receiving any card• active drugs or angiotensin converting enzyme inhibitor drugs at the time of the study. 2.2 Protocol The procedure was identical to that of a previous study (BERHEIT et al., 1990). Subjects were asked to lie on a motorised tilt table and, after 5 m i n of stabilisation, the E C G was continuously recorded for at least 20 min in the supine position. The table was then tilted to 90 ~ and the E C G recorded for another 20min. During the recording, patients breathed normally and were relatively relaxed. Breathing rates were within normal range (11-17 breaths min-*). Blood pressures were measured using a conventional s p h y g m o m a n o m e t e r in both supine and tilt positions. All the subjects were carefully instructed about the study and all gave their written consent. The studies were performed between 9 and 11 am. The E C G signals were recorded onto F M tape (HewlettPackard 3968A) and played back through low-pass filters (cutoff frequency 500Hz) into a hardware-based Q R S recognition device that produced a fiducial marker for the fixed point in each cardiac cycle and eliminated nonsinus (i.e. ectopic) beats. Fiducial timing was insensitive to amplitude fluctuations in the E C G produced by breathing and was adjusted for optimal accuracy (CRAELIUS et al., 1988a). The TTL-level pulses representing R - R intervals were sampled at 2000Hz and successive intervals were stored in a microcomputer (80286). Approximately 600 continuous intervals were processed to derive a continuous function of instantaneous heart rate. R - R intervals that included an ectopic beat were recognised as artefact and were corrected by an interpolation algorithm. Records that contained excessive ectopy ( > 10 beats) were rejected. The continuous function representing the instantaneous heart rate (IHR) was derived from the R - R intervals using an algorithm developed by BERGER et al. (1986). I H R functions were determined for 600 beats, which usually represented 8-9 min of stationary data. Records were examined for the absence of irregularities such as large abrupt fluctuations, or large shifts in mean (i.e. stationarity) and, if acceptable, were spectrally analysed as described below. The power spectrum of I H R within the band 0-0-5 Hz was estimated using a fast Fourier transform based on the windowed periodogram method (PRESS et al., 1986). The Fourier method produced adequate H R V spectra based on our relatively large number of stationary samples with a high signal-to-noise ratio. Future analyses could be improved by the use of parametric modelling techniques, such as autoregression, that can provide an accurate H R V Table 2
analysis of relatively small samples. H R V powers were computed in units of [beats s - * ] 2 and, for comparison, are presented as dimensionless units (NU) normalised with respect to total power (with power < 0-02 Hz subtracted). H R V power within a low frequency band (LF = 0.040-12 Hz) determined during the upright position was used as a measure of sympathetic nervous activity (SYMP). H R V power within a high frequency band (HF = 0.180.28 Hz) determined during the supine position was used as a measure of parasympathetic activity (PARA). Autonomic balance was defined as SYMP/PARA. Statistical comparisons were made by the Student's t-test and differences with p < 0.05 were considered significant (Table 2). 3 Results
The mean sympathetic/parasympathetic (SYMP/PARA) ratio of G r o u p A, 17 + 17 NU, was significantly greater than the control G r o u p C value of 4 __+2 NU, as shown in Fig. 1. G r o u p B was not significantly different from either group. There were no significant differences among groups with
p
analyses developed here, based on postural effects on HRV, have confirmed earlier studies that associated a depressed level of parasympathetic activity with the MI recovery period. A preliminary report has appeared (TANGELLAand CRAELIUS,1989). 2 M a t e r i a l s and m e t h o d s
2.1 Patients Three groups each of eight male subjects, were compared: Group A, consisting of patients who had MI within the previous 2-6 weeks of study; Group B, consisting of patients who had MI more than 1 year previously; and Group C, consisting of age-matched controls free of cardiac disease. Diagnosis of myocardial infarction was based on clinical electrocardiographic and enzymatic criteria (Table 1). Patients who had documented episodes of sustained arrhythmia, congestive heart failure, renal insufficiency, neuropathy or excessive premature beats were Table 1 Clinical characteristics
Group A (recent MI) 8 64 + 8 3.5 weeks
Group B (post MI) 8 63 + 5 25 months
Group C (controls) 8 68 _ 12 none
1
3
7
5
none none
First received l Oth January and in final form 18th July 1991
Number of subjects Age (mean + SD), years Post-MI period (mean) Site of MI Anterior Inferior
9 IFMBE: 1992
M I = m y o c a r d i a l infarction; S D = standard deviation
Correspondence should be addressed to Dr W. Craefius, Department of Biomedical Engineering, Rutgers University, Piscataway, NJ 08855, USA.
Medical & Biological Engineering & Computing
July 1992
385
excluded. N o n e of the patients was receiving any card• active drugs or angiotensin converting enzyme inhibitor drugs at the time of the study. 2.2 Protocol The procedure was identical to that of a previous study (BERHEIT et al., 1990). Subjects were asked to lie on a motorised tilt table and, after 5 m i n of stabilisation, the E C G was continuously recorded for at least 20 min in the supine position. The table was then tilted to 90 ~ and the E C G recorded for another 20min. During the recording, patients breathed normally and were relatively relaxed. Breathing rates were within normal range (11-17 breaths min-*). Blood pressures were measured using a conventional s p h y g m o m a n o m e t e r in both supine and tilt positions. All the subjects were carefully instructed about the study and all gave their written consent. The studies were performed between 9 and 11 am. The E C G signals were recorded onto F M tape (HewlettPackard 3968A) and played back through low-pass filters (cutoff frequency 500Hz) into a hardware-based Q R S recognition device that produced a fiducial marker for the fixed point in each cardiac cycle and eliminated nonsinus (i.e. ectopic) beats. Fiducial timing was insensitive to amplitude fluctuations in the E C G produced by breathing and was adjusted for optimal accuracy (CRAELIUS et al., 1988a). The TTL-level pulses representing R - R intervals were sampled at 2000Hz and successive intervals were stored in a microcomputer (80286). Approximately 600 continuous intervals were processed to derive a continuous function of instantaneous heart rate. R - R intervals that included an ectopic beat were recognised as artefact and were corrected by an interpolation algorithm. Records that contained excessive ectopy ( > 10 beats) were rejected. The continuous function representing the instantaneous heart rate (IHR) was derived from the R - R intervals using an algorithm developed by BERGER et al. (1986). I H R functions were determined for 600 beats, which usually represented 8-9 min of stationary data. Records were examined for the absence of irregularities such as large abrupt fluctuations, or large shifts in mean (i.e. stationarity) and, if acceptable, were spectrally analysed as described below. The power spectrum of I H R within the band 0-0-5 Hz was estimated using a fast Fourier transform based on the windowed periodogram method (PRESS et al., 1986). The Fourier method produced adequate H R V spectra based on our relatively large number of stationary samples with a high signal-to-noise ratio. Future analyses could be improved by the use of parametric modelling techniques, such as autoregression, that can provide an accurate H R V Table 2
analysis of relatively small samples. H R V powers were computed in units of [beats s - * ] 2 and, for comparison, are presented as dimensionless units (NU) normalised with respect to total power (with power < 0-02 Hz subtracted). H R V power within a low frequency band (LF = 0.040-12 Hz) determined during the upright position was used as a measure of sympathetic nervous activity (SYMP). H R V power within a high frequency band (HF = 0.180.28 Hz) determined during the supine position was used as a measure of parasympathetic activity (PARA). Autonomic balance was defined as SYMP/PARA. Statistical comparisons were made by the Student's t-test and differences with p < 0.05 were considered significant (Table 2). 3 Results
The mean sympathetic/parasympathetic (SYMP/PARA) ratio of G r o u p A, 17 + 17 NU, was significantly greater than the control G r o u p C value of 4 __+2 NU, as shown in Fig. 1. G r o u p B was not significantly different from either group. There were no significant differences among groups with
p
