0145-6008/92/1604-0721$3.00/0 ALCOHOLISM:CLINICAL AND EXPERIMENTAL RESEARCH

Vol. 16, No. 4 July/August 1992

Studies on a Wearable, Electronic, Transdermal Alcohol Sensor Robert M.Swift, Christopher S. Martin, Larry Swette, Anthony LaConti, and Nancy Kackley

The measurement of alcohol consumption over long time periods is important for monitoring treatment outcome and for research applications. Giner, Inc. has developed a wearable device that senses ethanol vapor at the surface of the skin, using an electrochemical cell that produces a continuouscurrent signal proportionalto ethanol concentration. A thermistor monitors continuous contact of the sensor with the skin, and a data-acquisition/logic circuit stores days of data recorded at 2- to 5-min intervals. Testing of this novel ethanol sensor/recorder was conducted on nonalcoholic human subjects consuming known quantities of ethanol and on intoxicated alcoholic subjects. The transdermal sensor signal closely follows the pattern of the blood alcohol concentration curve, although with a delay. This paper describes the concept of electrochemical ethanol measurement and presents some of the clinical data collected in support of the sensorlrecorder development. Key Words: Ethanol, Skin, Pharmacokinetics.

HE RELIABLE AND VALID MEASUREMENT of T alcohol consumption by passive, noninvasive means, over long time periods is important for monitoring indi-

ation to assess cumulative ethanol consumption over a 7to 10-day However, field trials of the sweat patch identified problems limiting its clinical utility. These include problems with ethanol storage and losses due to evaporation, back-diffusion and bacterial metabolism.20,21 Electrochemical detection of ethanol has been used for many years in sensor cells which oxidize ethanol and produce currents proportional to the ethanol concentration. Such cells are used in commercially available, portable breathalysers (e.g., Alco-Sensor I11 by Intoximeters, Inc., St. Louis, MO). The readings are well correlated with blood alcohol concentration (BAC).22 The adaptation of electrochemical technology for transdermal detection of ethanol seems an ideal approach for the continuous, noninvasive, long-term monitoring of ethanol use. Giner, Inc. has developed the Transdermal Alcohol Sensor/Recorder (TAS), a wearable, battery-operated device that continuously senses transdermal ethanol and stores an average value at 2- to 5-min intervals for up to 8 days. In the present paper, we describe the initial testing of the TAS on human subjects. The purpose of this study was to determine whether the prototype TAS will accurately track the pattern and the relative amplitude of the BAC over a range of ethanol doses. We tested: (1) nonalcoholic drinkers ingesting a moderate dose of ethanol under laboratory conditions, (2) intoxicated subjects presenting for inpatient detoxification (high ethanol dose), and ( 3 ) sober subjects not consuming ethanol.

viduals in research studies, monitoring those undergoing substance abuse treatment, and for forensic monitoring in special populations. Methods traditionally used for determining alcohol consumption include verbal report measures,1-6biochemical markers,’-’’ and direct measurement of urine, blood, saliva, breath or sweat alcohol levels. The ideal method for determining alcohol consumption would be inexpensive, highly specific and sensitive for ethanol, applicable across a range of users, usable over a range of time periods, and accepted by both the testers and the tested. One promising method of monitoring alcohol use has used the detection of ethanol transdermally. Measurable MATERIALS AND METHODS quantities of ingested ethanol are excreted through the human kin,'^,'^ by exocrine sweat glands,I4and by diffu- TransdermalAlcohol Sensor/Recorder sion across the skin.I5Transdermal ethanol levels generally The Giner TAS consists of two components: an electrochemical sensor that detects ethanol vapor and a data acquisition-recording device (Fig. reflect blood levels.l 6 The alcohol dosimeter or “sweat patch” is a wearable, 1). The patented sensor cell is a three-electrode, controlled potential noninvasive device which accumulates ethanol in perspir- device. The sensor is placed over the skin surface and continuously



From the Brown UniversityCenterfor Alcohol and Addiction Studies, Providence, Rhode Island (R.M.S., C.S.M.), Roger Williams Medical Center, Providence, Rhode Island (R.M.S.), and Giner, Inc., Waltham Massachusetts (L.S.,A.L., N.K.). Received for publication August 12, 1991; accepted March 12, 1992 This research was supported by Alcohol, Drug Abuse, and Mental health Administration SBIR Grant 2R44AA07657. Reprint requests: Robert M. Swift. M.D., Ph.D., Department of Psychiatry, Roger Williams Medical Center, 825 Chalkstone Avenue, Providence, RI 02908. Copyright 0 I992 by The Research Society on Alcoholism. Alcohol C h Exp Res, Vol 16, No 4, 1992: pp 721-725

oxidizes ethanol. The oxidation current is used as a direct measure of the local ethanol vapor concentration. The sensor is highly responsive to ethanol and unresponsive to potential interferants such as oxygen and acetone. Two imbedded thermistors produce temperature signals, which are used to determine that the device was in continuous contact with the skin and to compensate for temperature changes. A schematic of the solid-polymerelectrochemical cell depicting the sensing (sens), reference (ref), and counter (ctr) electrodes is shown in Fig. 2. The sensor is attached by cable to the data acquisition-recording device. This battery-operateddevice samples the ethanol and temperature signals at intervals and stores days to weeks of data. The circuit is programmed, and data are down-loaded by personal computer with access through an RS232 port. 721

SWIFT ET AL.

722

Human Studies All human testing was approved by the Human Research Committee of Roger Williams General Hospital. Studies were in accordance with “Recommended Council Guidelines on Ethyl Alcohol Administration in Human Experimentati~n.”~~

Controlled Consumption Experiments

Fig. 1. Photograph of the Giner, Inc. Transdermal Ethanol Sensor/Recorder (TAS).

- R E T A I N E R RING

The TAS tracking of the BAC was assessed in 10 paid, volunteer, fasting, nonalcoholic subjects (eight males, two females). Mean age of the subjects was 27 years (range 2 1-40), and average alcohol consumption of the subjects was five drinks per week (range 3-9). After programming, one to three TAS devices were applied to unprepared skin in different locations, usually forearm or upper arm, and held in place with an elastic belt or tape. In the experimental session, subjects consumed 0.75 ml/kg absolute ethanol (as 100 proof vodka) in four parts cold orange juice over 20 min. Breath alcohol concentrations were assessed at numerous times with an Intoxometer 3000. Initial BAC readings were obtained after a mouth rinse with warm water and a 7-min waiting period; readings indicative of mouth alcohol were discarded. During the study period, subjects remained in the laboratory area and were allowed to sit or walk ad libitum. A light meal was served 2 to 3 hr after consumption. The TAS devices were worn until the BAC was 0 mg/dl, usually 6 to 8 hr after consumption. The data were down-loaded to computer for analysis.

Intoxicated Alcoholic Subjects

C T K . € L € C . BUSHIN G

Fig. 2. Schematic of component parts to the Giner, Inc. Transdermal Alcohol Sensor.

In Vitro Calibration TAS devices are calibrated in the laboratory, based on the projected transdermal response, with solutions of various concentrations of ethanol in water. The signal is set to the anticipated in vitro to transdermal ethanol concentration ratio and adjusted for variations in the thickness of the diffusion-limiting membrane. The relationship of current to ethanol concentration is linear over a concentration range of at least 10 to 400 mg/di ( r = 0.99). The TAS potentiometer is adjusted so that the device reads “500” to “800” for a solution of ethanol of 100 mg/dl, depending on membrane thickness.

Since we were unable to administer high doses of ethanol to subjects, we examined the TAS signal in intoxicated alcoholic individuals. Five intoxicated, but otherwise medically stable alcoholic subjects (three males, two females, age range 31-53) were recruited from patients admitted to a general hospital substance abuse unit. Subjects all had a history of alcohol dependence and had been drinking heavily for at least 2 weeks prior to admission. Blood alcohol levels were 150 mg/dl or greater in all subjects. After obtaining informed consent, one or two programmed TAS devices were attached to the subjects as described above. Ethanol levels were determined by breathalyzer hourly. Subjects wore the TAS until BACs were 0, usually 12 to 24 hr. Patients were ambulatory, ate, slept, and participated in individual and group therapy while wearing the TAS. Subjects were not paid for their participation.

TAS Response in Sober Subjects To observe the TAS under conditions of no alcohol consumption, four healthy, nonalcoholic subjects wore the TAS device for periods of 48 to 96 hr. These subjects abstained from ethanol use, engaged in their usual activities and kept an activity log. Volatile blood-borne substances occur in hepatic and renal disease, which might produce false-positive responses. We therefore tested the TAS in sober subjects with jaundice ( n = biliary cirrhosis ( n = 1), alcoholic cirrhosis ( n = 2), and chronic renal failure ( n = 2), Subjects the TAS for at least hr.

Extended Wearing of the TAS

To explore the ability of the TAS to detect ethanol consumption over a longer period, four male, nonalcoholic subjects wore the devices for ’ periods of 5 to 7 days. During this period, subjects consumed (or remained abstinent) from alcohol according to their usual custom. SubBlood Alcohol Concentrations (BACs) jects identified a period of consumption by pressing an event marker on BACs were measured from breath samples using the Intoximeter 3000 the TAS device. (Intoximeters, Inc.) The Intoximeter was periodically calibrated with a standard ethanol solution. Deep-lung breath samples are highly correData Analysis lated with BACs as measured by gas chromatography (GC).” In our Stored data from each TAS device were downloaded to an IBMlaboratory, breathalyzer and BACs measured at the same time by GC, compatible computer and imported into a spreadsheet program (Lotus showed a blood/breath ratio of 0.994 ( r = 0.99).

723

WEARABLE, ELECTRONIC, TRANSDERMAL ALCOHOL SENSOR

123) for conversion to ethanol concentration and temperature and for graphing. Pharmacokinetic parameters including area under the curve (AUC) and curve slope were calculated with the PC version of MKh4ODEL.*' Means, t tests and Pearson's correlations for variables were calculated using Statistical Package for the Social Sciences (SPSS).

test). BAC and TAS curve peak amplitudes were correlated across subjects ( r = 0.6 1, p < 0.02). Fig. 4 shows a scatterplot of the BAC and TAS AUCs for the 10 subjects (n = 23 as some subjects wore multiple devices). The AUCs of the BAC and TAS curves are highly correlated ( r = 0.91, p < 0.001). Terminal slopes of the RESULTS BAC-time and AUC-time curves were poorly correlated A typical curve obtained for subjects ingesting ethanol and were significantly different by t test ( p < 0.03). To under controlled conditions is shown in Fig. 3. The units determine across-device, within-person correlations, we are TAS signal in pAmperes, BAC in mg/dl, and temper- examined the relationship between TAS signals and BAC ature in "C, all displayed on the Y-axis. The TAS signal is for two devices worn simultaneously. Time to TAS peaks of similar amplitude and time course to the BAC curve, for the two devices were highly correlated ( r = 0.86, p < although the TAS ethanol concentration versus time curve 0.001), as were TAS peak heights ( r = 0.71, p < 0.01) and lags behind the BAC. Mean time to peak for the 10 subjects TAS AUCs ( r = 0.94, p < 0.001). No significant correlawas 71 rf: 7 min (SEM) for the BAC curve and 107 k 12 tions were obtained for TAS parameters and height, min (SEM) for the TAS curve. The mean difference in onset weight, age, gender or quantity, and frequency of usual times was significantly different ( p < 0.02, by paired t alcohol consumption. To determine the ability of the TAS to quantitate difALCOHOL RECORD, TAS#34 ferent doses of ethanol within the same individual, two WT P subjects were tested on separate occasions over a several month period, receiving of 0.2 to 1.4 ml ethanol/kg body weight. The relationship between peak TAS signal and peak BAC is linear over the dosage range ( r = 0.97, p < 0.00 1) with a calculated threshold for TAS signal detection o \ of approximately 20 mg/dl. The AUCs for TAS and BAC curves show a high correlation across the different doses I (Y = 0.96, p < 0.001). In the intoxicated alcoholic subjects, the TAS curve was of similar amplitude and shape to the BAC curve, but lagged up to 120 min behind the BAC curve. Fig. 5 shows a typical curve from an intoxicated alcoholic subject. The 0 1 2 3 4 5 6 0 event marks (+) indicate TAS application and removal, Time (hrs) respectively. ! + fyents-0 SAC 1 In the sober subjects, no changes in TAS signal, which could be interpreted as a drinking event, were observed. Fig. 3. TAS output of an experimental session on a human subject consuming In the sober subjects with hepatic or renal disease, no 0.75 ml absolute ethanol per kg body weight. The units are TAS signal in pAmperes, 1

BAC in mg/dl. and temperature in "C, all displayed on the Y-axis, Time in hours on the X-axis.

ALCOHOLIC SUBJECT 06, TAS #31

I ,

AUC BACs I AUC TAS 10 Ss 5 devices, multiple runs 50000 40000

30000 0 3

a

20000

An

10000

0

I

0

. .

L

y=2175.198 + 1.402 x

4

1u

(I

1L

14

I0

lime (hrs)

I

I

I

I

10000

20000

30000

40000

AUC BACs Fig. 4. Scatterplot of AUC of the TAS and BAC vs time curves for the 10 experimental subjects. Graph includes multiple sessions per subject and multiple TAS devices per experimental session.

I+

.

Events

BAC

Fig. 5. TAS output from an intoxicated alcoholic subject. Device was worn for the first 17 hr after an inpatient hospitalization. The units are TAS signal in pAmperes, BAC in mg/dl, and temperature in "C, all displayed on the Y-axis, Time in hours on the X-axis.

SWIFT ET AL.

724

increases in TAS signal from baseline were observed. BAC tested by breathalyzer was 0.000 to 0.00 1 in these subjects. Wearability Issues Wearing of the TAS was well tolerated by all sober and intoxicated subjects. No skin irritation or rashes developed in skin over the area of the sensor. Although subjects were requested not to move or remove the TAS devices, the devices were occasionally removed by the intoxicated alcoholic subjects. The site preferred by subjects was the inner surface of the lower arm or outer surface of the upper arm. In all cases, the thermistor temperature signals accurately tracked attachment and removal of the TAS. Anecdotally, subjects reported interest and willingness to wear the device. Devices were worn in a single location for periods of up to 7 days without discomfort. Fig. 6 shows a typical TAS curve from a nonalcoholic subject wearing the TAS for a 6-day period on the inner forearm. The device was worn continuously during all activities, including sleep, and vigorous exercise (jogging at 14 hr). During daily showering, the TAS was covered with plastic wrap. The subject consumed alcohol twice during the period (4 ounces of vodka at 38 hr and 6 ounces of wine each at 84 hr and 110 hr. Each drinking period is identified by the event marker.

DISCUSSION

The present study indicates that transdermal ethanol measurement with the Giner TAS determines the time course of alcohol consumption in a semi-quantitative fashion. The peak amplitudes of the TAS signal and the BAC curves are highly correlated. The AUCs for the TAS and BAC curves are highly correlated across subjects. The threshold sensitivity for the TAS for measuring ethanol is approximately a BAC of 20 mg/dl. In the controlled consumption experiment, the peak values of the TAS concentration-time curves are rightshifted by approximately 30 min as compared with the breathalyzer concentration time curves. In the intoxicated alcoholic subjects, the peak TAS signal lags up to 120 min ALCOHOL RECORD, TAS#34, [6Day]

loo,

i I

I

In

I

E 1 1 .

0

12

24

'\

I

. 36

1

80

48

72

.* 84

Ih

1

1

96

108

120

132

lime (hrs)

u' i

+ Events

Fig.6. TAS output from a subject wearing the TAS on the inner forearm continuously for a 6day period. The units are TAS signal in pAmperes. BAC in mg/ dl, and temperature in 'C. all displayed on the Y-axis, Time in hours on the X-axis.

behind the BAC, with a longer lag time observed for the time to zero for the TAS curve. The delay in the TAS peak, a delayed time to zero and a threshold sensitivity for the presence of ethanol suggest the existence of a distinct pharmacokinetic compartment for cutaneous ethanol, also proposed by Brown.26Urinary alcohol levels show a similar lag when compared with blood.27 No false positive TAS signals occurred in sober subjects, including those with liver or renal disease. No spurious signals are observed during daily activity, sleep, or vigorous exercise. The signal to noise ratio of the TAS device is extremely favorable. There is occasional noise in some devices, usually due to defects in the interface cable. However, transient noise is easily distinguished from a TAS ethanol event curve. Future versions of the device will integrate sensor and recorder in a single housing to reduce noise. At this time, the absolute values for the BAC can be approximated, but not directly derived, from the transderma1 ethanol signal. It should be noted that the TAS signal measures ethanol flux rather than concentration. This flux is related to concentration but is also affected by sensor geometry, type, and thickness of diffusion-limiting membrane, rates of excretion and/or diffusion through the skin and upon evaporation. Since the TAS totally consumes ethanol during the analysis, ethanol vapor at the electrode is not in equilibrium with vapor at the skin surface. A breathalyzer sampling breath alcohol vapor, similarly requires calibration according to the blood/breath partition ratio.28 In monitoring alcohol consumption over time, subject compliance must be high, and the method must be tamper-proof. Individuals whose drug or alcohol use is monitored for clinical or forensic purposes may go to great lengths to falsify or invalidate the results.29The temperature signal may be used to check compliance; however, temperature is influenced by a number of factors, including ambient temperature. Additional measures are under consideration to ensure that the device is properly placed and tamper-resistant and to detect tampering. These include a galvanic skin response detector incorporated into the sensor and adhesives which secure attachment of the TAS to skin. However, no method of alcohol assessment may be absolutely tamperproof. Further research is necessary to determine optimal conditions for use of the transdermal technology. This preliminary work indicates that transdermal ethanol detection shows promise for the prospective assessment of alcohol consumption on a continuous, realtime basis. The projected cost of the TAS device is approximately $800. Additional testing must be performed to optimize performance and to test reliability and specificity across individuals over a range of operating conditions and to compare the use of this methodology with other methods of assessing alcohol consumption. Potential applications of the TAS include treatment outcome re-

WEARABLE, ELECTRONIC, TRANSDERMAL ALCOHOL SENSOR

search studies, monitoring clinical treatment compliance, and monitoring individuals who are mandated to remain abstinent from alcohol. REFERENCES 1. Orrego HL, Blendis LM, Blake JE, et al: Reliability of assessment of alcohol intake based on personal interviews in a liver clinic. Lancet 2:1354-1356, 1977 2. Maisto SA, Sobell LC, Sobell MC: Comparison of alcoholics’ selfreports of drinking behavior with reports of collateral informants. J Consult Clin Psycho1 47:106-112, 1979. 3. Sobell MB, Maisto SA, Sobell LC, et al: Developing a prototype for evaluating alcohol treatment effectiveness,in Sobell LC, Sobell MB, Ward E (eds): Evaluating Alcohol and Drug Abuse Treatment Effectiveness. New York, Pergamon Press, 1980 4. Babor TF, Stephens RS, Marlatt GA: Verbal report methods in clinical research on alcoholism: response bias and its minimization. J Stud Alcohol 48(5):410-424, 1987 5. O’Farrell TJ, Maisto SA: The utility of self-report and biological measures of alcohol consumption in alcoholism treatment outcome studies. Adv Behav Res Ther 9:91-125, 1987 6. Fuller RK, Lee KK, Gordis E Validity of self-report in alcoholism research: Results of a Veterans Administration cooperative study. Alcohol Clin Exp Res 12:201-205, 1988 7. Ryback RS, Eckhardt MJ, Felsher B, Rawlings RR: Biochemical and hematologic correlates of alcoholism and liver disease. JAMA 24812261-2265, 1982 8. Cushman P, Jacobson G, Barboriak JJ, Anderson AJ: Biochemical markers for alcoholism: Sensitivity problems. Alcohol Clin Exp Res 8:253-257, 1984 9. Takase S, Takada A, Tsutsumi M, Matsuda Y: Biochemical markers of chronic alcohol. Alcohol 2:405-410, 1985 10, Schellenberg F, Benard JY, LeGoff AM, et al: Evaluation of carbohydrate-deficient transfemn compared with Tf index and other markers of alcohol abuse. Alcohol Clin Exp Res 13(5):605-610, 1989 11. Allen JP: In Allen JP, Litten R (eds): Assessing Alcohol Consumption: Psychosocial and Biochemical Methods. Clifton, NJ, Humana Press (in Press), 1992 12. Nyman E, Palmlov A: The elimination of ethyl alcohol in sweat. Scand Arch Physiol74:155-159, 1936

125

13. Pawan GLS, Grice K Distribution of alcohol in urine and sweat after drinking. Lancet 2:1016, 1968 14. Brusilow SW, Gordis EH: The permeability of the sweat gland to non-electrolytes. Am J Dis Child 112:328-333, 1966 15. Scheuplein RJ, Blank IH: Permeability of the skin. Physiol Reviews 5 1(4):702-747, 197 1 16. Brown DJ: A method for determining the excretion of volatile substances through skin. Methods Find Exp Clin Pharmacol 7(5):269274, 1985 17. Phillips M. An improved adhesive patch for long-term collection of sweat. Biomater Med Devices Artif Organs 8( I): 13-2 I , I980 18. Phillips M, McAloon MH: A sweat-patch test for alcohol consumption: Evaluation in continuous and episodic drinkers. Alcohol Clin Exp Res 4(4):391-395, 1980 19. Phillips M: Sweat patch test for alcohol consumption: Rapid assay with an electrochemical detector. Alcohol Clin Exp Res 6(4):532-534, 1982 20. Phillips ELR, Little RE, Hillman RS, et a]: A field test of the sweat patch. Alcohol Clin Exp Res 8(2):233-237, 1984 2 1. Parmentier AH, Liepman MR, Nirenberg T Reasons for failure of the alcohol sweat patch. Alcohol Clin Exp Res 15:376 (abstract), 1991 (abstr) 22. Gibb KA, Yee AS, Johnston CC: Accuracy and usefulness of a breath alcohol analyzer. Ann Emerg Med 13(7):516-520, 1984 23. Dubowski K:Recent developments in breath alcohol analysis, in Israelstam S and Lambert S (eds): Proceedings of the 6th International Conference on Alcohol, Drugs and Traffic Safety. Toronto, Canada, ARF of Toronto, 1976 24. NIAAA: “Recommended Council Guidelines on Ethyl Alcohol Administration in Human Experimentation.” National Advisory Council on Alcohol Abuse and Alcoholism, 1989 25. Holford N: MKMODEL, Version 4. Cambridge UK, Biosoft, 1990 26. Brown DJ: Pharmacokinetics of alcohol excretion in human perspiration. Meth Find Exp Clin Pharmacol 7( 10):539-544, 1985 27. Batt RD: Adsorption, distribution and elimination of alcohol. In Crow KE, Batt RD (eds): Human Metabolism of Alcohol, vol. I. Boca Raton, FL, CRC Press, 1989 28. Dubowski K Human pharmacokinetics of ethanol 1. Peak blood alcohol concentrations and elimination in male and female subjects. Alcohol Technical Reports 555-63, 1976 29. Swift RM, Camara P, Griffiths W: Use of toxicological analysis in the identification of drugs of abuse, in Stoudemire A, Fogel B (eds): Medical Psychiatric Practice, vol. 1. Washington, APA Press, 199 1

Studies on a wearable, electronic, transdermal alcohol sensor.

The measurement of alcohol consumption over long time periods is important for monitoring treatment outcome and for research applications. Giner, Inc...
566KB Sizes 0 Downloads 0 Views