Periodic cardiovascular and ventilatory activity during midazolam sedation

British Journal of Anaesthesia 1996; 76: 503–507

Periodic cardiovascular and ventilatory activity during midazolam sedation D. C. GALLETLY, T. B. WILLIAMS AND B. J. ROBINSON

Summary We have examined the effects of sedation with midazolam 0.1 mg kg91 and reversal with flumazenil 0.5 mg on beat-to-beat heart rate (HR) variability (HRV), systolic arterial pressure (SAP), finger photoplethysmograph amplitude (PLA) and impedence pneumography in eight volunteers. With the onset of sedation there was a small decrease in SAP and increase in HR (ns). Spectral analysis of the HR time series showed reductions in the proportion of power in the high (90.15 Hz) frequency “ventilatory” band consistent with midazolam causing vagolysis. During sedation, low frequency (:0.05 Hz) oscillations of PLA, HR, SAP and ventilation were observed. These were thought to be secondary to activity of coupled cardiorespiratory neurones within the brain stem and the ventilatory periodicity appeared similar to that observed during the early stages of sleep. The diminished high frequency and increased low frequency oscillations induced by midazolam sedation were reversed by administration of flumazenil. (Br. J. Anaesth. 1996; 76: 503–507) Key words Hypnotics benzodiazepine, midazolam. Antagonists benzodiazepine, flumazenil. Heart, heart rate. Arterial pressure, drug effects. Measurement techniques, plethysmography.

Analysis of heart rate variability (HRV) is a useful tool for examination of cardiac autonomic tone [1]. Because vagal and sympathetic components of the heart rate (HR) control mechanism are active over different frequency ranges, breakdown of a HR time series using Fourier methods of spectral analysis allows the shifting balance of autonomic control to be examined non-invasively. Similar analysis of systolic arterial pressure (SAP) variability allows extension of these observations to involvement of peripheral cardiovascular mechanisms [2]. Spectral methods of HR [3], SAP [2] and photoplethysmograph pulse amplitude [4] variability analysis have been used to examine the cardiovascular effects of i.v., volatile and inhalation anaesthesia. Using statistical or time domain methods, the benzodiazepines were one of the first groups of drugs to be studied with regard to HRV; they were observed to cause a decrease in respiratory sinus arrhythmia [5]. Midazolam has modest but clinically important cardiovascular effects and its ability to

obtund sympathetic and baroreflex function has been examined in detail [6]. In this study we have examined the effects of midazolam sedation on HRV, arterial pressure and plethysmograph amplitude.

Subjects and methods After obtaining Ethics Committee approval and informed consent, we studied eight healthy male volunteers, mean age 32 (26–38) yr, mean weight 74 (64–82) kg. Studies were conducted between 08:00 and 11 : 00 with subjects fasted and refraining from caffeine-containing beverages for at least 12 h. Each subject lay supine and breathed oxygen 5 litre min91 from a Hudson mask. A 22-gauge cannula was inserted into the right antecubital fossa. The following monitoring was commenced: HR (ECG lead CM5, Corometrics Neo-Trak 502), SAP (continuous non-invasive arterial tonometry, Nellcor N-CAT 500), ventilation (impedance pneumograph, Corometrics Neo-Trak 502) and infra-red finger plethysmography (Hewlett–Packard 78330A). The ECG signal was interfaced with a purpose-built R wave detector generating a 50-ms pulse synchronous with each R wave peak. This pulse triggered a Macintosh IIcx computer with 16-bit ADC board (National Instruments) to record the R–R interval, respiratory amplitude and to determine the trough (diastole) and peak (systole) of each subsequent photoplethysmograph and arterial pressure wave. For graphical display purposes, the output from the impedence pneumograph is shown as the unsigned value through which a five-point moving averager has been passed. In order to ensure that the ventilatory peak of HRV would not interfere with those frequencies less than 0.15 Hz, subjects were instructed to breathe in time with a 0.24-Hz microcomputer-generated auditory pulse throughout the study. After a 15-min stabilization period, the N-Cat 500 tonometer was calibrated and beat-to-beat recording of cardiorespiratory data commenced. After 5 min the tonometer was recalibrated and subjects received midazolam 0.1 mg kg91 i.v. Twelve to fifteen minutes later the tonometer was recalibrated and flumazenil

D. C. GALLETLY, FANZCA, FRCA, T. B. WILLIAMS, BM, CHB, BMEDSC, B. J. ROBINSON, MSC, PHD, Section of Anaesthesia, Wellington School of Medicine, PO Box 7343, Wellington South, New Zealand. Accepted for publication: November 8, 1995. Correspondence to D.C.G.

504

British Journal of Anaesthesia

0.5 mg was administered i.v. Recordings were continued for a further 5-min period after administration of flumazenil. A 5-min epoch of data was obtained towards the end of the control, midazolam and post-flumazenil periods. Fast Fourier analysis on 256-s segments of HR, respiration, SAP and plethysmograph amplitude (PLA) were performed as described previously [7]. Spectral power (area under the power spectrum curve) was calculated for each variable, for each period, in each subject. This was done for total power and for the power in four frequency bands: very low frequency 0.01–0.02 Hz; low frequency (“vasomotor”) 0.02–0.08 Hz; mid frequency (“baroreflex”) 0.08–0.15 Hz; and high frequency (“ventilatory”) 0.15–0.45 Hz. The power in each band was then expressed as a percentage of the total for the corresponding subject and condition. In the time domain we calculated, for each subject and condition, the mean (SD) of each variable and for the R–R time series, the mean of the consecutive differences of the R–R interval time series. Changes in the time domain measures and the proportion of power in each spectral range were analysed using ANOVA followed by paired t tests (Statview). Total spectral power in each frequency range were compared using the non-parametric Mann–Whitney U test.

Results Table 1 shows the changes in time domain and spectral measures during the study period. The representative raw data are shown in figures 1 and 2. After midazolam, there was a trend towards

Figure 1 Effect of midazolam (M) and flumazenil (F) reversal of heart rate (HR), systolic arterial pressure (SAP), plethysmograph amplitude (PLA) and pneumogram amplitude (Pneumo.). PLA and pneumogram amplitude are shown on arbitrary scales.

increasing HR (P : 0.08), decreasing SAP (P : 0.08) and increasing PLA. These changes differed between subjects and a clear, sustained, increase in HR, reversed by flumazenil, is apparent in figures 1 and 2. Sustained increases in PLA are shown in figure 2, but these changes appeared not to be reversed by flumazenil. Pneumogram amplitude during midazolam sedation was similar to the control, although during sedation the frequency of ventilation was observed to

Table 1 Effect of midazolam sedation and flumazenil reversal on time and frequency domain measures (mean (SD)). Con Diff RR : mean of the consecutive differences, irrespective of sign, between R–R intervals within the times series. Total spectral power is given in arbitrary units Control R–R interval Mean SD (within subject) Mean Con Diff RR Total power % High % Mid % Low % Very low Pneumograph amplitude Total power Peak (Hz) SAP Mean SD (within subject) Total power % High % Mid % Low % Very low PLA Mean SD (within subject) Total power % High % Mid % Low % Very low

Midazolam

P

Flumazenil

P

1010 (53.4) 56.7 (8.3) 49.1 (10.5) 1.88 (0.54) 51.8 (6.0) 13.7 (1.9) 25.6 (4.7) 8.81 (2.0)

ns ns ns ns ns 0.05 ns ns

992 (50) 61.7 (37) 45.3 (5.6) 1.73 (0.11) 47.1 (6.3) 22.3 (3.6) 24.4 (11.2) 6.1 (1.6)

880 (38.9) 58.3 (7.8) 29.3 (6.36) 1.47 (0.41) 24.6 (6.5) 17.8 (3.0) 35.8 (7.4) 21.9 (4.1)

ns ns ns ns 0.03 ns ns 0.004

766 (260) 0.24 (0.0007)

410 (126) 0.26 (0.02)

ns ns

564 (246) 0.24 (0.004)

ns ns

132.7 (5.9) 4.36 (0.78) 11.1 (5.4) 15.3 (2.4) 11.9 (2.9) 47.4 (2.5) 25.3 (2.6)

122.7 (4.8) 5.7 (0.63) 15.1 (5.3) 17.3 (3.2) 11.4 (3.3) 45 (4.9) 26.3 (6.4)

ns ns ns ns ns ns ns

124 (4.7) 4.22 (0.63) 9.3 (3.6) 26.9 (4.1) 8.86 (1.5) 44.6 (5.8) 19.6 (2.9)

ns ns ns 0.03 ns ns ns

6004 (1316) 471 (94.3) 115 (37) 28 (5.7) 10.7 (1.6) 40.0 (5.6) 21.3 (4)

7575 (606) 990 (247) 416 (171) 12.7 (4) 6.3 (2) 53.0 (2.48) 27.9 (4.8)

ns ns 0.03 0.05 ns 0.05 ns

8048 (495) 839 (223) 300 (146) 27.4 (6.9) 7.6 (2.2) 42.2 (6.82) 22.8 (7.2)

ns ns ns ns ns ns ns

Cardiovascular and ventilatory effects of midazolam

505

Discussion

Figure 2 Effect of midazolam (M) and flumazenil (F) reversal of heart rate (HR), systolic arterial pressure (SAP), plethysmograph amplitude (PLA) and pneumogram amplitude (Pneumo.). PLA and pneumogram amplitude are shown on arbitrary scales.

increase slightly from the control value of 0.24 Hz to 0.26 (0.02) Hz. Time domain measures (SD, mean consecutive differences) of HR and SAP variability did not differ significantly from control values. Spectral analysis revealed significant shifts in the HR and PLA power spectra. HR high frequency absolute power and the proportion of power in the high frequency band were reduced while the proportion of power in the band 0.01–0.02 Hz increased from 6 to 22 %. Examination of individual cardiorespiratory time series showed remarkable variability and complexity of effect. Consistent with the changes observed on spectral analysis, figures 1 and 2 show that with administration of midazolam, the appearance of very low and low frequency oscillations is evident in one or more of the cardiovascular measures (SAP, PLA, HR). These waves were most apparent in PLA and to the smallest extent in HR. The period of these oscillations was generally in the region of 40–75 s (0.013–0.025 Hz). The SAP waves appeared as transient elevations above a baseline level, rather than sinusoidal (figs 1, 2). The elevations were of two types: where SAP and PLA oscillations were out of phase, SAP elevations occurred during a reduction in plethysmograph amplitude; and where SAP and PLA oscillations were in phase, SAP pressure elevations were associated with an increase in plethysmograph amplitude (fig. 1). Low frequency SAP and PLA fluctuations were not confined solely to the period of midazolam sedation but were present in at least one subject before drug administration. Indeed in this subject administration of midazolam caused a transient slowing or cessation of these waves for several minutes. In three subjects oscillations in pneumogram amplitude were clearly observed in phase with the SAP waves; increased pneumogram amplitude occurred at the time of maximal increase in arterial pressure.

The cardiovascular effects of midazolam have been investigated most frequently in patients with coronary artery disease [6, 8–11] and to a lesser extent in healthy subjects [12]. Midazolam causes a decrease in SAP, a small increase in HR, a reduction in cardiac output and little change or a decrease in systemic vascular resistance; ventricular filling pressures are usually decreased or may be unchanged and a direct cardiac depressant effect has been proposed. These haemodynamic changes most likely involve a reduction in SAP secondary to decreased systemic vascular resistance, decreased venous return caused by venodilatation and direct myocardial depression reducing cardiac output [13]. Small doses of diazepam have little effect on baroreflex function in patients with heart disease [14] whereas both diazepam 0.4 mg kg91 and midazolam 0.3 mg kg91, when used for induction, cause significant but transient decreases in the pressor baroreflex slope which appears to be less than that induced by potent anaesthetic agents [6]. A compensatory, albeit blunted, baroreceptor-mediated increase in HR is therefore possible allowing haemodynamic status to be returned towards normal limits. The respiratory effects of benzodiazepines are similarly modest but clinically important. Although having minimal effects on respiratory mechanics [15] they diminish central respiratory drive and the ventilatory response to inhaled carbon dioxide [16], and may induce both central apnoea and upper airway obstructive apnoea [17]. The autonomic circulatory and respiratory effects of benzodiazepines are largely mediated within the brain stem by modulation of GABAergic tone [18, 19]. GABA is involved in the endogenous regulation of SAP [20] and regulation of vagal outflow to the heart and respiratory tract [21]. Benzodiazepines are believed to attenuate vagal tone through a central parasympatholytic effect [5, 22–24]. The reduction in respiratory sinus arrhythmia (which is vagally mediated) by diazepam is believed to be mediated through the same mechanism. A vagolytic effect for midazolam can also be inferred both from its known ability to increase HR and the observed reduction in both absolute and proportional high frequency “ventilatory” beat-to-beat HRV in this study. HR periodicities within this frequency range (90.15 Hz) are mediated predominantly through fluctuations in vagal tone. Increased HR in concert with reduction in ventilatory induced HRV waves is seen after injection of midazolam. An unexpected feature of midazolam sedation was the appearance of cardiovascular oscillations at frequencies :0.025 Hz. Although these oscillations were most apparent in PLA and SAP they were also present in HR and in the pneumogram amplitude. The genesis of low frequency periodicities in HR at less than the Mayer wave or baroreflex frequency of approximately 0.1–0.15 Hz is generally ascribed to vasomotor fluctuations in peripheral vascular tone and indeed we observed them to be most apparent in the digital photoplethysmogram. These low frequency oscillations are well described in normal

506 subjects [25, 26] and were clearly observed in one of our own subjects before sedation. We have observed [unpublished observations] similar fluctuations in unmedicated subjects who are warm and relaxed. Augmentation of these oscillations by ACE inhibitors suggests a modulatory role for the renin– angiotensin system [25] although some authors believe that they reflect an oscillation within a thermoregulatory reflex arc comprising peripheral thermoreceptors, afferent input, central integration and efferent sympathetic vascular response [27, 28]. It is of interest that some periodic SAP elevations were associated with a decrease in PLA while others were associated with an increase. The former probably reflect waves of sympathetic excitation which cause a secondary increase in arterial pressure while the latter may reflect a passive increase in PLA secondary to waves of SAP elevation, perhaps caused by constriction of other vasculature beds. Although these low frequency waves most likely represent sympathetic vasoconstriction, their appearance probably represents an overall decrease in peripheral cutaneous sympathetic tone from a relatively nonoscillatory tonic level at which vasoconstriction is more intense. In several traces there was a clear association between respiratory activity and very low frequency oscillations in SAP and PLA; elevations in arterial pressure and decreased PLA occurring at the same time as increased ventilatory amplitude. Periodic ventilatory activity is well described in healthy adults under normal and hypoxic conditions, and in the elderly and neonates. In neonates, Waggener and colleagues observed periodic ventilatory patterns with short (approx. 20 s) and long (approx. 60 s) cycles [29]. In adults, Lenfant observed oscillations in healthy awake adults with periods of 2–6 breaths, 25–50 breaths and 150–200 breaths [30]. Ventilatory periodicity has been noted during sleep and is particularly marked during the early stages of sleep or sleep onset. It is possible that the ventilatory abnormalities of sleep apnoea may be an exaggerated form of this periodicity [31]. These ventilatory fluctuations may represent the activity of a Servo control loop oscillation in a chemoreceptor/ ventilation arc, as postulated for the genesis of Cheyne–Stokes respiration. Indeed several theoretical and experimental studies having shown that prolonged circulatory times (e.g. heart failure) and enhanced chemoresponsiveness (e.g. altitude or hypoxic conditions) may lead to periodic patterns of breathing [32]. These mechanisms seem unlikely explanations for the observations in the present study however, as all subjects were healthy, the effects of midazolam on cardiac output are small and all subjects breathed oxygen which is a potent depressant of chemoreceptor activation. It seems probable that the slow periodic ventilatory effects were generated by a mechanism similar to that during light sleep and, as both cardiac and respiratory oscillations were coupled, they may have been generated within the interconnected respiratory and cardiovascular neurones of the brain stem [33]. Induction of these oscillations by midazolam and their reversal by flumazenil suggests they are

British Journal of Anaesthesia mediated at least in part through brain stem GABAergic transmission. It is of interest that cardiovascular oscillations at these very low frequencies were seen far more commonly than ventilatory oscillations. This may indicate that cardiovascular oscillations within the brain stem are the primary source of ventilatory periodicity. The periodic ventilatory activity may be of clinical relevance to the observation of respiratory instability during benzodiazepine sedation and may warrant further study. Administration of flumazenil to patients during benzodiazepine sedation results in few haemodynamic responses and minimal changes in SAP and HR [34]. Although these variables were changed little, administration of flumazenil was associated with return of periodic cardiorespiratory oscillations towards those before administration of midazolam. In summary, we observed that midazolam caused changes in HR and HRV consistent with vagolysis, in some subjects provoked periodic (40–75 s) oscillations in HR, PLA and SBP and that these oscillations were at times coupled to periodic ventilatory activity. The study highlights the value of detailed time series analysis in demonstrating the significant effects of a drug on cardiorespiratory control when standard mean haemodynamic measures are changed little.

References 1. van Ravenswaaij-Arts C, Kollee L, Hopman J, Stoelinga G, van Geijn H. Heart rate variability. Annals of Internal Medicine 1993; 118: 436–447. 2. Scheffer GJ, Ten Voorde BJ, Karemaker JM, Ros HH, De Lange JJ. Effects of thiopentone, etomidate and propofol on beat to beat cardiovascular signals in man. Anaesthesia 1993; 48: 849–855. 3. Muzi M, Ebert TJ. Quantification of heart rate variability with power spectral analysis. Current Opinion in Anesthesiology 1993; 6: 3–17. 4. Robinson B, Buyck HCE, Galletly DC. The effect of propofol anaesthesia on heart rate, arterial pressure and digital plethysmograph variability. British Journal of Anaesthesia 1994; 73: 167–173. 5. Adinoff B, Mefford I, Waxman R, Linnoila M. Vagal tone decreases following diazepam. Psychiatry Research 1992; 41: 89–97. 6. Marty J, Gauzit R, Lefevre P, Couderc E, Farinotti R, Henzel C, Desmonts JM. Effects of diazepam and midazolam on baroreflex control of heart rate and on sympathetic activity in humans. Anesthesia and Analgesia 1986; 65: 113–119. 7. Galletly DC, Tobin P, Robinson B, Corfiatis T. Effect of 30 % nitrous oxide inhalation on spectral components of heart rate variability in conscious subjects. Clinical Science 1993; 85: 389–392. 8. Samuelson PN, Reves JG, Kouchoukos NT, Smith LR, Dole KM. Hemodynamic responses to anesthetic induction with midazolam or diazepam in patients with ischemic heart disease. Anesthesia and Analgesia 1981; 60: 802–809. 9. Reves JG, Samuelson PN, Lewis S. Midazolam induction in patients with ischaemic heart disease; haemodynamic observations. Canadian Anaesthetists Society Journal 1979; 26: 402–409. 10. Schulte-Sasse U, Hess W, Tarnow J. Haemodynamic responses to induction of anaesthesia using midazolam in cardiac surgical patients. British Journal of Anaesthesia 1982; 54: 1053–1057. 11. Marty J, Nitenberg A, Blanchet F, Zouioueche S, Desmonts JM. Effects of midazolam on the coronary circulation in patients with coronary artery disease. Anesthesiology 1986; 64: 206–210. 12. Lebowitz P, Cote ME, Daniels AL, Ramsey FM, Martyn JA,

Cardiovascular and ventilatory effects of midazolam

13. 14. 15. 16. 17. 18.

19.

20.

21.

22.

Teplick RS, Davison JK. Comparative cardiovascular effects of midazolam and thiopental in healthy patients. Anesthesia and Analgesia 1982; 61: 771–775. Reves JG, Fragen RJ, Vinik HR, Greenblatt DJ. Midazolam: pharmacology and uses. Anesthesiology 1985; 62: 310–324. Markievicz W, Hunt S, Harrison DC, Alderman EL. Circulatory effects of diazepam in heart disease. Journal of Clinical Pharmacology 1976; 16: 637–644. Southorn P, Rehder K, Didier E. Midazolam sedation and respiratory mechanics in man. Anesthesiology 1981; 55: A367. Forster A, Gardaz J, Suter PM, Gemperle M. Respiratory depression by midazolam and diazepam. Anesthesiology 1980; 53: 494–497. Montravers P, Dureuil B, Desmonts JM. Effects of i.v. midazolam on upper airway resistance. British Journal of Anaesthesia 1992; 68: 27–31. Taveira Da Silva AM, Pineo S, Gillis RA. Gammaaminobutyric acid type A receptor blockade at the intermediate area of the ventral surface of the medulla counteracts the cardiorespiratory depression produced by intravenous midazolam. Pharmacology 1992; 45: 154–161. Gillis RA, Namath I, Easington C, Abrahams T, Guidotti A, Quest J, Hamosh P, Taveira Da Silva AM. Drug interaction with gamma amino butyric acid/benzodiazepine receptors at the ventral surface of the medulla results in pronounced changes in cardiorespiratory activity. Journal of Pharmacology and Experimental Therapeutics 1989; 248: 863–870. Suzuki M, Kuramochi T, Suga T. GABA receptor subtypes involved in the neuronal mechanisms of baroreceptor reflex in the nucleus tractus solitarii of rabbits. Journal of the Autonomic Nervous System 1993; 43: 27–35. Gillis RA, Quest JA, DiMicco JA. Central regulation of autonomic function by GABA receptors. In: Squires RF, ed. GABA and Benzodiazepine Receptors. Boca Raton, FL: CRC Press Inc, 1988: 141–147. Haxhiu MA, Van Luteren E, Cherniack NS, Deal EC. Benzodiazepines acting on the ventral surface of medulla cause airway dilation. American Journal of Physiology 1989; 257: R810–R815.

507 23. Hockman CH, Livingstone KE. Inhibition of reflex vagal bradycardia by diazepam. Neuropharmacology 1971; 10: 307–314. 24. DiMicco JA. Evidence for control of cardiac vagal tone by benzodiazepine receptors. Neuropharmacology 1987; 26: 553–559. 25. Akselrod S, Gordon D, Ubel F, Shannon D, Barger A, Cohen R. Power spectrum analysis of heart rate fluctuation: a quantitative probe of beat to beat cardiovascular control. Science 1981; 213: 220–222. 26. Miyakawa K, Koepchen HP, Polosa C eds. Mechanisms of Blood-Pressure Waves. Berlin: Springer-Verlag, 1984. 27. Bini G, Hagbarth K, Hynninen P, Wallin B. Thermoregulatory and rhythm generating mechanisms governing the sudomotor and vasoconstrictor outflow in human cutaneous nerves. Journal of Physiology (London) 1980; 306: 537–552. 28. Kitney R. An analysis of the nonlinear behaviour of the human thermal vasomotor control system. Journal of Theoretical Biology 1972; 52: 231–248. 29. Waggener T, Stark A, Cohlan B, Frantz I. Oscillatory breathing patterns leading to apnoeic spells in infants. Journal of Applied Physiology 1982; 52: 1288–1295. 30. Lenfant C. Time dependent variations of pulmonary gas exchange in normal man at rest. Journal of Applied Physiology 1967; 22: 675–684. 31. Pack A, Silage D, Millman R, Knight H, Shore ET, Chung D. Spectral analysis of ventilation in elderly subjects awake and asleep. Journal of Applied Physiology 1988; 64: 1257–1267. 32. Khoo M. Periodic breathing. In: Crystal R, West J, Barnes P, Cherniack N, Weibel E, eds. The Lung: Scientific Foundations. New York: Raven, 1990; 1419–1431. 33. Feldman JL, Ellenberger HH. Central co-ordination of respiratory and cardiovascular control in mammals. Annual Review of Physiology 1988; 50: 593–606. 34. Marty J, Joyon D. Haemodynamic responses following reversal of benzodiazepine-induced anaesthesia or sedation with flumazenil. European Journal of Anaesthesiology 1988: (Suppl. 2): 167–171.