PHARNIACODYNANIICS AND
DRUG ACTION Dose-related effect of triazolam on postural sway Measurement of clinically relevant benzodiazepine drug effects at hypnotic and anxiolytic doses has been difficult because most measures are subjective, difficult to interpret, or relate to anesthetic doses. A potentially useful measure of drug effect is postural sway, which is a manifestation of the corrective mechanisms associated with the maintenance of upright posture. Postural sway was measured over 8 hours, with a biomechanics force platform, in six healthy male volunteers who received triazolam, 0.125, 0.250, or 0.375 mg, or placebo in a randomized double-blind study. Our results show a dose-dependent increase in postural sway measured as the elliptical area or the 95% confidence ellipse for the area covered by the subjects' sway. After triazolam, 0.250 and 0.375 mg, the area under the sway-time curve and peak effect increased significantly compared with placebo (p < 0.05). The number of losses of balance when subjects stood on one foot also showed a significant increase with increased dose (p < 0.05). The rate of loss of balance was positively correlated with the extent of postural sway (r = 0.802; p < 0.001). The extent of sway when the subjects were drug free predicted this increase in a subject's postural sway with triazolam. Thus at hypnotic or anxiolytic doses of triazolam, computer-assisted force platform measures of sway provide a clinically relevant measure of drug effect. Measurement of drug-induced postural sway may be useful in persons at risk for falls, such as the elderly. (CLIN PHARMACOL THER 1991;49:581-8.)
Deborah W. Robin, MD, Samer S. Hasan, BS, Michael J. Lichtenstein, MD, MSc,' Richard G. Shiavi, PhD, and Alastair J. J. Wood, MD Nashville, Tenn. Benzodiazepines are one of the most frequently prescribed classes of drugs. In spite of their widespread use, measurement of clinically relevant drug effects at hypnotic or anxiolytic doses has been difficult. Previously used measures of drug effect have included visual analog scales of sedation, psychometric testing, or spectral analysis of the electroencephalogram; how-
From the Division of Clinical Pharmacology and the Department of Biomedical Engineering. Vanderbilt University School of Medicine. Supported by U.S. Public Health Service grants RR 00095 and AG
01395. Received for publication Aug. 6. 1990; accepted Dec. 10, 1990. Reprint requests: Alastair J. J. Wood, MD, Division of Clinical Pharmacology. Department of Medicine, Vanderbilt University School of Medicine. Nashville. TN 37232. "Current address: GRECC, Audie L. Murphy VA Hospital, 7400 Merton Minter Blvd., San Antonio, TX 78284. 13/1/27315
ever, these measures can be difficult to interpret and their extrapolation and relevance to clinical effects are limited. Other measures of benzodiazepine effects have related to anesthetic doses. The development of appropriate measures of drug effect is extremely important to define sensitivity in different at-risk groups such as the elderly. A potentially useful measure of drug effect is postural sway." Sway is a manifestation of the corrective mechanisms associated with the maintenance of upright posture. Ability to balance may decrease as sway increases, and this can be quantified as a decrease in the ability to maintain the center of pressure within the base of support.7 Although everyone has some sway when standing still, postural sway increases with age,8-I4 and retrospective studies have shown that increased sway is associated with increased incidence of falls and CoIles' wrist fractUre.8.15-18 A prospective study of the physical factors
581
582
CL1N PHARMACOL THER MAY 1991
Robin et al.
associated with falls in a community-based elderly population followed for 1 year showed that increased sway was associated with a twofold increased risk of falls. 19 Another prospective cohort study associated increased velocity of sway with incident falls.2° Postural sway has previously been used as a measure of drug effect for such centrally acting drugs as alcohol, tetrahydrocannabinol, and barbiturates .21-24 A number of studies have measured postural sway after benzodiazepines with various techniques and have shown increased or altered patterns of sway. 1-6 Different benzodiazepines4 may have different effects on balance. Recent epidemiologic studies have suggested that use of benzodiazepines, especially those with long halflives, is associated with an increased risk of both falling and hip fractures.25-28 The purposes of this study were therefore to determine whether the short half-life benzodiazepine triazolam impairs balance and increases postural sway in a dose-dependent fashion and to develop a technique that will allow comparison of the effects of different benzodiazepines.
METHODS Six healthy male volunteers ranging in age from 19 to 29 years (mean age, 23 ± 1.3 years I -±SEMD with a mean weight of 168.7 ± 9.8 pounds gave written, informed consent to participate in this study, which had been approved by the Vanderbilt University Committee for the Protection of Human Subjects. No subject had taken any medication or alcohol for at least 5 days before initiation of the study. All had normal histories, physical examinations, and routine laboratory test results. All subjects received triazolam, 0.125, 0.250, or 0.375 mg, or placebo on different days. Each drug or placebo administration was separated by at least 48 hours. The order of drug administration was determined randomly and was double blind. After an overnight fast, the subject received an oral drug dose between 8 and 9 AM with approximately 50 ml water. Postural sway was measured before and /4, /2, /4, - , 11/2, 2, 21/2, 3, 4, 5, 6, and 8 hours after the dose in all six subjects. At the same time points the subjects were asked to mark their degree of drowsiness on a 10 cm visual analog scale labeled as awake on the left and asleep on the right. Additional sway measurements were made on some days in all subjects before baseline and at 11/4, 13/4, and 21/4 hours. Postural sway was measured with a stationary stable level biomechanics force platform (model 0R6-3 force-torque dynamometer; Advanced Mechanical 1,
3,
1
Technology, Inc., Newton, Mass.) that consists of a flat surface beneath which lie four strain-gauge force transducers that produce a voltage proportional to the applied force. The mechanical deflection or strain in the transducer resulting from the applied force changes resistances in a bridge circuit.29 This unbalances the bridge circuit and produces an output voltage. The force platform used in these studies produced an output voltage proportional to each of the three orthogonal forces (sagittal, lateral, and vertical) and force moments. The data generated were acquired in real time with a PDP 11/23 computer (Digital Equipment Corp., Maynard, Mass.) sampling at 50 Hz. The data were filtered digitally to remove electronic noise and then used to determine the coordinates of the center of pressure, which is defined as the point of application of the applied load in the horizontal plane, which is the surface of the force platform. Movement of the center of pressure results from the corrective mechanisms associated with the maintenance of balance. The excursions of the center of pressure during a fixed time trial generate a sway tracing or stabilogram. The area covered by these excursions is a measure of the amount of postural sway; the greater the area the greater the sway and the imbalance. We have evaluated two measures, radial area and elliptical area, to quantify this area. The radial area (in square centimeters) was determined by calculating the radial distance to the center of pressure at each sampling interval from the geometric center of the stance. These radii were then averaged and the area of the circle with that average radius was determined. The elliptical area (in square centimeters) is the 95% confidence ellipse that was constructed with the geometric center of the stabilogram and the slope of the linear regression through the stabilogram to define the major axis of the ellipse. The perpendicular to the major axis at the center of the stabilogram defined the minor axis. For a 95% confidence ellipse the lengths of the axes were found by multiplying the standard deviation of the sway data along each axis by 1.96, and the area of the ellipse was determined. During the measurement of sway, subjects were asked to stand on the force platform with their feet comfortably apart. This foot position was traced on a piece of cardboard and maintained throughout the studies. Subjects wore the same low-heeled shoes throughout the study. The subjects were instructed to look straight ahead at a black target placed at eye level 15 feet away on a blank screen and to maintain balance. Measurements were made in three stances: dou-
VOLUME 49 NUMBER 5
Triazolam effect on postural sway 40
DOUBLE STANCE
:
583
EYES OPEN
30
.375 MG .250 MG .125 MG PLACEBO
10
0
200
100
300
400
500
TIME (min)
40
DOUBLE STANCE
:
40
EYES CLOSED
30
SINGLE STANCE
:
EYES OPEN
30 .375 MG .250 MG .125 MG PLACEBO
.375 MG .250 MG .125 MG PLACEBO 10
10
0
0
100
200
300
400
TIME (min)
500
100
200
300
400
500
TIME (min)
Fig. 1. Postural sway (mean ± SEM of six subjects) during varying doses of triazolam and placebo in three stances.
ble stance (two feet), eyes open; double stance, eyes closed; and single stance (one foot), eyes open. The measurement of sway in single stance is clinically relevant in that the majority of time spent in both walking and stair climbing is spent in single stance. In addition, maintenance of balance in single stance reflects the individual's ability to prevent a fall in an unstable stance. The arms were placed comfortably at their sides during the stances on two feet or out to the side to help maintain balance in the stance on one foot. Each stance measurement lasted for 10 seconds or until loss of balance and was repeated three times. A loss of balance could occur only during single leg stance and was defined as stepping on the force platform or around it, with the previously elevated foot, in an attempt to regain balance. Data analysis. The means of the elliptical areas (in square centimeters) for the three repetitions of each stance at each time point were determined for each subject and used in the subsequent analysis. Data are presented as means ± SEM; the statistical significance of differences was determined by a repeated-measures analysis of variance (ANOVA) followed by a
Wilcoxon signed rank test. The minimal level of significance accepted was p < 0.05.
RESULTS Comparisons were made between the elliptical and radial areas to determine which measure encompassed the greater proportion of the center of pressure data points. The elliptical area was found to contain a higher proportion of the data points than the radial area (85.7% ± 2.8% versus 56.1% ± 4.9%; n = 899; p < 0.001) and described the data more efficiently so that a greater percentage of center of pressure points per unit area was contained within the elliptical than the radial area (p < 0.001). The elliptical area measure, which is dependent on two parameters (major and minor axes), is a better descriptor of the data and was therefore used as the measure of postural sway throughout the reported studies. The effect on postural sway of the three doses of triazolam and placebo in the six subjects in three stances is shown in Fig. 1. As can be seen, the extent of sway increases with increasing drug dose and is greater during the less stable single stance. In addi-
584
CLAN PLIARMAC01. 'FLIER MAY 1991
Robin et al.
*
p < 0.05 compared to placebo
Lu
PLACEBO
.125 MG
.250 MG
.375 MG
TRIAZOLAM DOSE Fig. 2. Dose-response relationship relating dose of triazolam or placebo to area under the swaytime curve during single stance, eyes open.
tion, the time course of the response follows the expected time course of drug concentrations. The stability of the measurement is shown by the fact that on the different study days the pretreatment measurements (time 0) are similar and measurements during placebo administration are stable throughout the day. The area under the sway-time tracings for the first 240 minutes after drug, the time at which postural sway had returned to baseline, was determined by the trapezoidal rule and showed a significant dose-response relationship (Fig. 2) (p < 0.001, repeatedmeasures ANOVA). A significant dose-response relationship for triazolam was also demonstrable for the peak drug effects (p < 0.005, repeated-measures ANOVA). Similar dose-related changes were seen for both the area under the sway-time tracing and peak effect during double stance, eyes open, and double stance, eyes closed. While the subjects are standing on one foot with their eyes open (single stance, eyes open) they may become sufficiently imbalanced so that they have to put the other foot down (loss of balance). Triazolam increases the number of losses of balance in a dose-
dependent fashion (Fig. 3, A). The applicability of these measurements is also validated by the relationship between the number of losses of balance and the extent of postural sway (Fig. 3, B), so that the greater an individual's drug-induced instability measured by the area under the sway-time curve the greater the number of losses of balance. The extent of sway in subjects when they were drug free predicted the increase in sway after triazolam administration (Fig. 4), so there was an excellent correlation between the increase in sway produced by triazolam and the extent of sway when subjects were drug free (r = 0.917; p < 0.01). The data from the visual analog scale also show a dose-response relationship for triazolam (Fig. 5). However, the curves continue to decrease below baseline after 240 minutes, suggesting a withdrawal phenomena or that subjects rate themselves as more awake in the late afternoon than in the morning. The area under the visual analog-time tracings was determined by the trapezoidal rule and also showed a significant dose-response relationship (p < 0.05, repeated-measures ANOVA).
VOLUME 49 NUMBER 5
1.0
1.25
*
< 0.05 compared to placebo
Z
DRUG ACTION Dose-related effect of triazolam on postural sway Measurement of clinically relevant benzodiazepine drug effects at hypnotic and anxiolytic doses has been difficult because most measures are subjective, difficult to interpret, or relate to anesthetic doses. A potentially useful measure of drug effect is postural sway, which is a manifestation of the corrective mechanisms associated with the maintenance of upright posture. Postural sway was measured over 8 hours, with a biomechanics force platform, in six healthy male volunteers who received triazolam, 0.125, 0.250, or 0.375 mg, or placebo in a randomized double-blind study. Our results show a dose-dependent increase in postural sway measured as the elliptical area or the 95% confidence ellipse for the area covered by the subjects' sway. After triazolam, 0.250 and 0.375 mg, the area under the sway-time curve and peak effect increased significantly compared with placebo (p < 0.05). The number of losses of balance when subjects stood on one foot also showed a significant increase with increased dose (p < 0.05). The rate of loss of balance was positively correlated with the extent of postural sway (r = 0.802; p < 0.001). The extent of sway when the subjects were drug free predicted this increase in a subject's postural sway with triazolam. Thus at hypnotic or anxiolytic doses of triazolam, computer-assisted force platform measures of sway provide a clinically relevant measure of drug effect. Measurement of drug-induced postural sway may be useful in persons at risk for falls, such as the elderly. (CLIN PHARMACOL THER 1991;49:581-8.)
Deborah W. Robin, MD, Samer S. Hasan, BS, Michael J. Lichtenstein, MD, MSc,' Richard G. Shiavi, PhD, and Alastair J. J. Wood, MD Nashville, Tenn. Benzodiazepines are one of the most frequently prescribed classes of drugs. In spite of their widespread use, measurement of clinically relevant drug effects at hypnotic or anxiolytic doses has been difficult. Previously used measures of drug effect have included visual analog scales of sedation, psychometric testing, or spectral analysis of the electroencephalogram; how-
From the Division of Clinical Pharmacology and the Department of Biomedical Engineering. Vanderbilt University School of Medicine. Supported by U.S. Public Health Service grants RR 00095 and AG
01395. Received for publication Aug. 6. 1990; accepted Dec. 10, 1990. Reprint requests: Alastair J. J. Wood, MD, Division of Clinical Pharmacology. Department of Medicine, Vanderbilt University School of Medicine. Nashville. TN 37232. "Current address: GRECC, Audie L. Murphy VA Hospital, 7400 Merton Minter Blvd., San Antonio, TX 78284. 13/1/27315
ever, these measures can be difficult to interpret and their extrapolation and relevance to clinical effects are limited. Other measures of benzodiazepine effects have related to anesthetic doses. The development of appropriate measures of drug effect is extremely important to define sensitivity in different at-risk groups such as the elderly. A potentially useful measure of drug effect is postural sway." Sway is a manifestation of the corrective mechanisms associated with the maintenance of upright posture. Ability to balance may decrease as sway increases, and this can be quantified as a decrease in the ability to maintain the center of pressure within the base of support.7 Although everyone has some sway when standing still, postural sway increases with age,8-I4 and retrospective studies have shown that increased sway is associated with increased incidence of falls and CoIles' wrist fractUre.8.15-18 A prospective study of the physical factors
581
582
CL1N PHARMACOL THER MAY 1991
Robin et al.
associated with falls in a community-based elderly population followed for 1 year showed that increased sway was associated with a twofold increased risk of falls. 19 Another prospective cohort study associated increased velocity of sway with incident falls.2° Postural sway has previously been used as a measure of drug effect for such centrally acting drugs as alcohol, tetrahydrocannabinol, and barbiturates .21-24 A number of studies have measured postural sway after benzodiazepines with various techniques and have shown increased or altered patterns of sway. 1-6 Different benzodiazepines4 may have different effects on balance. Recent epidemiologic studies have suggested that use of benzodiazepines, especially those with long halflives, is associated with an increased risk of both falling and hip fractures.25-28 The purposes of this study were therefore to determine whether the short half-life benzodiazepine triazolam impairs balance and increases postural sway in a dose-dependent fashion and to develop a technique that will allow comparison of the effects of different benzodiazepines.
METHODS Six healthy male volunteers ranging in age from 19 to 29 years (mean age, 23 ± 1.3 years I -±SEMD with a mean weight of 168.7 ± 9.8 pounds gave written, informed consent to participate in this study, which had been approved by the Vanderbilt University Committee for the Protection of Human Subjects. No subject had taken any medication or alcohol for at least 5 days before initiation of the study. All had normal histories, physical examinations, and routine laboratory test results. All subjects received triazolam, 0.125, 0.250, or 0.375 mg, or placebo on different days. Each drug or placebo administration was separated by at least 48 hours. The order of drug administration was determined randomly and was double blind. After an overnight fast, the subject received an oral drug dose between 8 and 9 AM with approximately 50 ml water. Postural sway was measured before and /4, /2, /4, - , 11/2, 2, 21/2, 3, 4, 5, 6, and 8 hours after the dose in all six subjects. At the same time points the subjects were asked to mark their degree of drowsiness on a 10 cm visual analog scale labeled as awake on the left and asleep on the right. Additional sway measurements were made on some days in all subjects before baseline and at 11/4, 13/4, and 21/4 hours. Postural sway was measured with a stationary stable level biomechanics force platform (model 0R6-3 force-torque dynamometer; Advanced Mechanical 1,
3,
1
Technology, Inc., Newton, Mass.) that consists of a flat surface beneath which lie four strain-gauge force transducers that produce a voltage proportional to the applied force. The mechanical deflection or strain in the transducer resulting from the applied force changes resistances in a bridge circuit.29 This unbalances the bridge circuit and produces an output voltage. The force platform used in these studies produced an output voltage proportional to each of the three orthogonal forces (sagittal, lateral, and vertical) and force moments. The data generated were acquired in real time with a PDP 11/23 computer (Digital Equipment Corp., Maynard, Mass.) sampling at 50 Hz. The data were filtered digitally to remove electronic noise and then used to determine the coordinates of the center of pressure, which is defined as the point of application of the applied load in the horizontal plane, which is the surface of the force platform. Movement of the center of pressure results from the corrective mechanisms associated with the maintenance of balance. The excursions of the center of pressure during a fixed time trial generate a sway tracing or stabilogram. The area covered by these excursions is a measure of the amount of postural sway; the greater the area the greater the sway and the imbalance. We have evaluated two measures, radial area and elliptical area, to quantify this area. The radial area (in square centimeters) was determined by calculating the radial distance to the center of pressure at each sampling interval from the geometric center of the stance. These radii were then averaged and the area of the circle with that average radius was determined. The elliptical area (in square centimeters) is the 95% confidence ellipse that was constructed with the geometric center of the stabilogram and the slope of the linear regression through the stabilogram to define the major axis of the ellipse. The perpendicular to the major axis at the center of the stabilogram defined the minor axis. For a 95% confidence ellipse the lengths of the axes were found by multiplying the standard deviation of the sway data along each axis by 1.96, and the area of the ellipse was determined. During the measurement of sway, subjects were asked to stand on the force platform with their feet comfortably apart. This foot position was traced on a piece of cardboard and maintained throughout the studies. Subjects wore the same low-heeled shoes throughout the study. The subjects were instructed to look straight ahead at a black target placed at eye level 15 feet away on a blank screen and to maintain balance. Measurements were made in three stances: dou-
VOLUME 49 NUMBER 5
Triazolam effect on postural sway 40
DOUBLE STANCE
:
583
EYES OPEN
30
.375 MG .250 MG .125 MG PLACEBO
10
0
200
100
300
400
500
TIME (min)
40
DOUBLE STANCE
:
40
EYES CLOSED
30
SINGLE STANCE
:
EYES OPEN
30 .375 MG .250 MG .125 MG PLACEBO
.375 MG .250 MG .125 MG PLACEBO 10
10
0
0
100
200
300
400
TIME (min)
500
100
200
300
400
500
TIME (min)
Fig. 1. Postural sway (mean ± SEM of six subjects) during varying doses of triazolam and placebo in three stances.
ble stance (two feet), eyes open; double stance, eyes closed; and single stance (one foot), eyes open. The measurement of sway in single stance is clinically relevant in that the majority of time spent in both walking and stair climbing is spent in single stance. In addition, maintenance of balance in single stance reflects the individual's ability to prevent a fall in an unstable stance. The arms were placed comfortably at their sides during the stances on two feet or out to the side to help maintain balance in the stance on one foot. Each stance measurement lasted for 10 seconds or until loss of balance and was repeated three times. A loss of balance could occur only during single leg stance and was defined as stepping on the force platform or around it, with the previously elevated foot, in an attempt to regain balance. Data analysis. The means of the elliptical areas (in square centimeters) for the three repetitions of each stance at each time point were determined for each subject and used in the subsequent analysis. Data are presented as means ± SEM; the statistical significance of differences was determined by a repeated-measures analysis of variance (ANOVA) followed by a
Wilcoxon signed rank test. The minimal level of significance accepted was p < 0.05.
RESULTS Comparisons were made between the elliptical and radial areas to determine which measure encompassed the greater proportion of the center of pressure data points. The elliptical area was found to contain a higher proportion of the data points than the radial area (85.7% ± 2.8% versus 56.1% ± 4.9%; n = 899; p < 0.001) and described the data more efficiently so that a greater percentage of center of pressure points per unit area was contained within the elliptical than the radial area (p < 0.001). The elliptical area measure, which is dependent on two parameters (major and minor axes), is a better descriptor of the data and was therefore used as the measure of postural sway throughout the reported studies. The effect on postural sway of the three doses of triazolam and placebo in the six subjects in three stances is shown in Fig. 1. As can be seen, the extent of sway increases with increasing drug dose and is greater during the less stable single stance. In addi-
584
CLAN PLIARMAC01. 'FLIER MAY 1991
Robin et al.
*
p < 0.05 compared to placebo
Lu
PLACEBO
.125 MG
.250 MG
.375 MG
TRIAZOLAM DOSE Fig. 2. Dose-response relationship relating dose of triazolam or placebo to area under the swaytime curve during single stance, eyes open.
tion, the time course of the response follows the expected time course of drug concentrations. The stability of the measurement is shown by the fact that on the different study days the pretreatment measurements (time 0) are similar and measurements during placebo administration are stable throughout the day. The area under the sway-time tracings for the first 240 minutes after drug, the time at which postural sway had returned to baseline, was determined by the trapezoidal rule and showed a significant dose-response relationship (Fig. 2) (p < 0.001, repeatedmeasures ANOVA). A significant dose-response relationship for triazolam was also demonstrable for the peak drug effects (p < 0.005, repeated-measures ANOVA). Similar dose-related changes were seen for both the area under the sway-time tracing and peak effect during double stance, eyes open, and double stance, eyes closed. While the subjects are standing on one foot with their eyes open (single stance, eyes open) they may become sufficiently imbalanced so that they have to put the other foot down (loss of balance). Triazolam increases the number of losses of balance in a dose-
dependent fashion (Fig. 3, A). The applicability of these measurements is also validated by the relationship between the number of losses of balance and the extent of postural sway (Fig. 3, B), so that the greater an individual's drug-induced instability measured by the area under the sway-time curve the greater the number of losses of balance. The extent of sway in subjects when they were drug free predicted the increase in sway after triazolam administration (Fig. 4), so there was an excellent correlation between the increase in sway produced by triazolam and the extent of sway when subjects were drug free (r = 0.917; p < 0.01). The data from the visual analog scale also show a dose-response relationship for triazolam (Fig. 5). However, the curves continue to decrease below baseline after 240 minutes, suggesting a withdrawal phenomena or that subjects rate themselves as more awake in the late afternoon than in the morning. The area under the visual analog-time tracings was determined by the trapezoidal rule and also showed a significant dose-response relationship (p < 0.05, repeated-measures ANOVA).
VOLUME 49 NUMBER 5
1.0
1.25
*
< 0.05 compared to placebo
Z
