Cocaine reduction in unmotivated crack users using carbamazepine versus placebo in a short-term, double-blind crossover design On the basis of cocaine-caused kindling in animals and the usefulness of carbamazepine in treating kindling-type seizures, carbamazepine has been tried in clinical settings with cocaine-dependent individuals. This report presents findings of a 20-day, double-blind, placebo-controlled crossover study in 32 nontreatment-motivated, paid, chronic crack cocaine users. Carbamazepine significantly lowered the mean number of positive urine specimens compared with placebo. Of clinical importance, serum carbamazepine levels of 4 &ml (17 ~ m o l / L )or more were associated with greater improvement. A consistent, clinically important trend linked therapeutic levels with improvement for all subjective and objective outcome variables. Comparison of daily acknowledged cocaine use or professed cocaine abstinence, with cocaine use indicated by daily urinalysis in these chronic cocaine users, has suggested the possibility of cocaine saturation as an important methodologic limitation inherent in outpatient studies of cocaine use THER1991; 50:81-95.) in humans. (CLINPHARMACOL
James A. Halikas, MD, Ross D. Crosby, PhD, Gregory A. Carlson, BA, Fred Crea, BA, Nina M. Graves, PharmD, and Larry D. Bowers, PhD Minneapolis, Minn. Repetitive cocaine use, particularly of the smokable cocaine products crack and freebase, is rapidly habituating and addi~ting,'.~associated with withdrawal symptoms and persistent, sometimes overwhelming, ~ r a v i n g . ~Craving, .~ defined as "an irresistible urge to use a substance that compels drug-seeking behavior"I0 can be generated spontaneously by internal mood states such as dysphoria or boredom or triggered by external prompts that remind the addict of some facet of cocaine use or effect. Examples of such external triggers include large amounts of cash, white powder,
From the Departments of Psychiatry and Laboratory Medicine, University of Minnesota School of Medicine, the Chemical Dependency Treatment Program, and the College of Pharmacy, University of Minnesota. Supported in part by a grant from the Chemical Dependency Program Division, Minnesota Department of Human Services (St. Paul, Minn. ) . Received for publication Feb. 15, 1991; accepted March 3, 1991. Reprint requests: James A. Halikas, MD, Department of Psychiatry, University of Minnesota, Box 393 UMHC, Minneapolis, MN 55455. 1311129228
the familiar smell, or the sound of a crackling noise. Craving is intense but f l ~ c t u a t e s . ~Craving ~ ~ ~ ~ ~per" sists intermittently for many month^.^ The frequency and duration of each episode of craving varies from addict to addict and situation to situation." In highrisk situations, craving may reappear indefinitely." This often-irresistible urge may have a different neurophysiologic mechanism than the immediate reinforcing properties of the drug that prompt repetition of early-use behavior." An important neuropharmacologic property of cocaine is "kindling." In animals cocaine sensitizes brain cells in the limbic system to fire at progressively lower drug concentrations with repeated intermittent co'~ of kindling have been caine e x p ~ s u r e . ' ~ -Stages identifiedI3 and described14 during the past 20 years, so the definition of kindling is no longer limited to overt seizure activity but now encompasses stereotypic movements and other changes in animal behavior. Recent work has suggested that one possible mechanism for the antiseizure efficacy of carbamazepine may be through its amelioration or elimination of kindling.I5 One of us (J.H.) hypothesized that this progressive
CLIN PHARMACOL THER JULY 1991
82 Halikas e t al. Table I. Drug and payment scheduling protocol Study day
Carbamazepine first 200 mg (n = 8) 400 mg (n = 8) Placebo first 200 mg (n = 8) 400 mg (n = 8) Dailypayment($)
100 200 200 400
200 400
200 400
200 200 400 400
200 200 400 400
200 400
200 400
0 0 5
0 0 5
0 0 5
0 0 5
0 0 5
0 0 5
100 200 200 400 5 10
0 0 5
0 0 5
sensitivity to cocaine at a cellular level may possibly be the physiologic explanation of the human cocainerelated behavior of craving." Carbamazepine was therefore suggested as an anticraving drug. The medication was tried in an open clinical mixed inpatientoutpatient setting with 35 The results indicated that carbamazepine reduced the acute withdrawal symptoms, minimized or eliminated the early and delayed craving urges, produced greater cocaine abstinence, and somewhat blocked cocaine's euphoric effects during subsequent cocaine experimentation,ll.16-18 A subsequent anecdotal report has also noted this blockage effect.I9 Experience with this clinical population indicated that the use of carbamazepine was well tolerated and safe. Treatment with carbamazepine was used by patients for approximately 3 to 12 months, after which it was spontaneously discontinued with no or minimal resurgence of craving. Clinically, cocaine-dependent patients appeared to need relatively low carbamazepine doses. Management, particularly during the induction phase, required clinical judgment and dosage modification based on side effects or breakthrough craving. The clinical results were encouraging enough to warrant the development of a double-blind research paradigm. Past outpatient treatment studies with chemically dependent individuals have used either patients motivated for treatment or patients under some coercion to maintain abstinence. This complicates studies of treatment efficacy, particularly studies of the efficacy of pharmacologic agents in discouraging drug use. Ideally, medication should demonstrate efficacy in the absence of psychologic assistance in a setting conducive to accurate self-report of drug-use information. Successful pharmacologic intervention should not require that a patient be motivated to give up cocaine use to be effective. Rather, the patient must only be willing to take the medication as prescribed. Theoretically, motivation should not be a prerequisite for anti-
0 0 5
100 0 200 0 200 400 10
0 0
0 0
0 0
0 0
200 400 10
200 400 10
200 400 10
200 400 20
craving medication efficacy any more than for antibiotic, anti-inflammatory, antidepressant, or antiarrhythmic efficacy.
METHODS Subjects In the study to be reported the variables of motivation and coercion were eliminated with a noncoercive, nontreatment situation. Unmotivated street crack addicts were found by word of mouth on the streets in the midst of the crack neighborhoods of Minneapolis who, for daily payment, would participate. The research site used was a nontreatment facility, a vacant laboratory area of a largely abandoned hospital building in the subjects' neighborhood. All subjects were poor adult men who indicated that they had no current wish to stop using crack cocaine. They were unmotivated for drug treatment. The informed consent form told them that this was a study of the efficacy of carbamazepine in reducing or eliminating their cocaine use. They were assured that all information obtained was strictly confidential, protected by federal guidelines, and could not be subpoenaed or in any other way used against them. Their sole apparent motivation for volunteering for this study was daily immediate cash payment for each visit. Design This was a 20-day, double-blind, placebocontrolled, crossover study of 32 chronic crack users. Each volunteer received carbamazepine for 10 days and placebo for 10 days in an AB or BA design (Table I). Subjects attended the study site daily, ingested their medication under observation, and participated in a research visit. Subjects received daily immediate cash payment at each visit, on a sliding scale to encourage study completion: $Stday through day 10, $10/day through day 15, and $20lday through day 20 (Table I). A missed day of attendance deferred the payment schedule by 1 day, thereby eliminating one
VOLUME 50 NUMBER 1
of the $20 days at the end of the study. A subject could collect a maximum of $235 during the entire 22day period. Half of the subjects received active medication, either 200 or 400 mg carbamazepine in one daily dose, during the first half of the study and placebo in the second half; the other half of the subjects reversed the order and received placebo during the first half of the study. Both order of medication and dose were assigned randomly. Because each subject received 9 consecutive days of full carbamazepine dose and 9 consecutive days of placebo in this AB or BA design, each served as his own control. Days with graduated or tapered doses (first and last days with each condition) were eliminated from statistical analysis to minimize crossover effects. Each subject received two tablets per day under direct visual observation regardless of the assigned condition. Daily supervised urine samples were obtained. Both active carbamazepine (200 mg tablet of Tegretol) and identical inert placebo were supplied by Ciba-Geigy Corp. (Summit, N. J.). Admission criteria to the study required selfreported current and past crack cocaine use at the level of at least 30 days of use in the previous 100 days. Craving, or compulsion to use cocaine, was not a requirement for admission. There were no restrictions regarding other drug use or alcohol use. There was no washout period. Subjects were at least 18 years of age and had at least an eighth-grade education. Subjects with clinically significant medical illnesses, seizure disorder, trauma-induced unconsciousness within the past year, or organic brain syndrome were excluded from the study, as were those with clinically significant abnormal laboratory findings at screening that could not be explained on the basis of cocaine use.
Procedures At the screening visit, after informed consent was obtained, the following information was collected: alcohol and drug use history; Addiction Severity In-
Cocaine reduction in crack users 83 dex''; medical and psychiatric history; physical examination; blood tests, including chemical profile, liver studies, white count and differential, thyroid profile, creatine phosphokinase, and, if indicated, creatine kinase MB; urinalysis and urine drug screen; and electrocardiogram (ECG). After all test results were returned, there was a baseline visit followed by 20 daily visits. At baseline and each subsequent daily visit the following data were collected: vital signs, ECG, urinalysis for cocaine alone, self-report cocaine craving scale," cocaine withdrawal scale, self-report of cocaine use and route, and self-report of other drug use in the past 24 hours. Monitoring of alcohol use during the study was based on self-report only. The blind medication was then taken under direct observation, and the subject was paid for the visit. Subjects were regularly encouraged by study staff to maintain abstinence from all drugs and alcohol.
Instruments used Carbamazepine serum levels were analyzed by HPLC." The standard curve between 0.5 and 20 pgiml showed a coefficient of variation of less than 5%. The method of urinalysis used was a semiquantitative fluorescence polarization immunoassay technique, with the Abbott TDx system (Abbott Diagnostics, Pasadena, Calif.). The TDx was calibrated in the range of 0.3 to 5 pglml to assess variations in very low levels of cocaine metabolite on a daily basis. Cocaine saturation and sequestration may complicate interpretation of urine test r e s ~ l t s , ~ ~ ~ ~ ~ the lthou~h presence of cocaine metabolite (benzoylecgonine) has usually been considered evidence of cocaine use within the previous 48 to 72 h o ~ r s . ~ ~ . ~ ~ Daily outcome measures Urinalysis. Positive cocaine presence was reported on the basis of presence of lowest measurable cocaine metabolite (0.30 pglml). The number of positive urine samples within each drug condition was calculated as a ratio of the observed number of positive urine samples per 9 days. Concentration of cocaine metabolite above the lowest amount of 0.30 pgiml was measured by the fluorescence polarization immunoassay technique for concentrations up to 5.0 pglml (the lower 0.1% to 1.O% of positive cocaineicocaine metabolite levels seen clinically in chronic users). Self-reported cocaine use. Days reporting cocaine use were calculated as a ratio of the number of days reporting use per 9 days within each drug condition; and, the reported amount of cocaine used. Self-reported ratings. Self-reported ratings were
84 Halihas et al. craving intensity,'' craving frequency, craving duration, withdrawal symptoms, withdrawal severity score, and global improvement. The Cocaine Withdrawal Scale is a 19-item scale of withdrawal symptoms (e.g., irritable, anxious, and depressed) gleaned from the cocaine treatment literature, modeled after the Minnesota Tobacco Withdrawal heck list.^^ Subjects respond on a four-point scale (0 = none, 1 = mild, 2 = moderate, and 3 = severe) based on symptoms at the time of completion. Total withdrawal score is computed by summing responses across items, with possible scores ranging from 0 to 57. Total positive symptom count is a count of non-zero items (i.e., responses of 1, 2, or 3), with possible scores ranging from 0 to 19. Clinical ratings. Clinical ratings were global improvement, side effects, and therapeutic effect of the medication. Clinical ratings of "therapeutic effects" are made on a four-point scale (0 = unchanged or worse, 1 = minimal, 2 = moderate, and 3 = marked) and are based exclusively on the perceived therapeutic effects of the medication in achieving reduction in craving, frequency or amount of cocaine use, and preoccupation with cocaine. The Patient Global Improvement Scale and Clinical Global Improvement Scale are subjective ratings of the patient's condition on a seven-point scale (1 = very much worse, 4 = no change, and 7 = very much improved), used often in outpatient drug studies. The Patient Global Improvement Scale and Clinical Global Improvement Scale are ratings of overall life conditions and may be based on subjective factors in addition to those relating directly to the therapeutic effects of the medication (e.g., employment, interpersonal relationships, and personal grooming). Vital signs. Vital signs included blood pressure and pulse, both sitting and standing, temperature, weight, and ECG. Blood samples. Laboratory tests performed on days 5, 10, 15, and 20 included complete blood count, chemistry profile, liver function tests, and serum carbamazepine level.
Statistical analysis The maximum number of missing observations on any dependent variable was four for any one subject. A within-subjects analysis of variance indicated no differences in the number of missing observations for carbamazepine and placebo conditions for each dependent variable. Missing observations were then replaced by computing the mean of the observations im-
CLIN PHARMACOL THER JULY 1991
mediately before and after the missing days in both drug conditions. The 2 days on which subjects received titrated doses (i.e., days 1 and 11 for the carbamazepine-first group and days 10 and 20 for the placebo-first group) were eliminated from statistical analysis to minimize crossover effects between drug conditions (Table I). The resultant design produced 9 days of observation within each drug condition (i.e., carbamazepine and placebo) for each subject; two between-subjects factors included drug order (i.e., carbamazepine first or placebo first) and dose (200 or 400 mg). A repeated-measures analysis of covariance was performed separately for each dependent variable (except nominal measures), with baseline scores serving as the covariate for between-subjects comparisons. Main-effect tests were computed for drug condition, drug order, and dose. A test of the drug condition-bydrug-order interaction was also computed to examine temporal changes from the first 9 days to the second 9 days of observation. An auxiliary analysis was performed to examine the impact of carbamazepine serum levels on outcome measures. Previous clinical experience with carbamazepine in patients using cocaine" and patients with epilepsy27,2Xhas suggested a minimum therapeutic serum level of approximately 4 pglml. Subjects were therefore divided into those who had attained a serum level of 4 pglml at some point in the study (n = 19) and those who had not (n = 13). A separate repeatedmeasures analysis of covariance was then performed for each outcome measure for the two serum levels with only the 9 carbamazepine observation days, with baseline scores serving as the covariate.
RESULTS Seven subjects were screened out because of existing medical conditions, illiteracy, or below-minimum cocaine use admission criteria. An additional five subjects failed to return after the screening visit. Nine subjects who were enrolled and dispensed medication failed to complete the study. Two subjects quit during the carbamazepine-first, one during the carbamazepine-second (average length, 9.7 days; SD, 4.6), and six during the placebo-first condition (average length, 5.0 days; SD, 2.8). There were no differences in the rate or severity of side effects found in either group of dropouts. Of the 16 total days of full medication for the three carbamazepine-condition dropouts, only 1 day of mild side effects was reported. The 32 subjects who completed the study were
VOLUME 50 NUMBER 1
Cocuine reduction in mack users 85
Table 11. Sociodemographic characteristics of the sample (n = 32) Marriage
Never married Divorced Separated Widowed Married Current common law relationships
n
Housing
n
19 Living alone in transient facilities 4 Rooming house 2 With sexual partner 2 2 3
12 CYear college degree 7 4
6 2
With relatives With friends On the street
Highest level of education
n 2
Some college High school completion or GED Some high school
James A. Halikas, MD, Ross D. Crosby, PhD, Gregory A. Carlson, BA, Fred Crea, BA, Nina M. Graves, PharmD, and Larry D. Bowers, PhD Minneapolis, Minn. Repetitive cocaine use, particularly of the smokable cocaine products crack and freebase, is rapidly habituating and addi~ting,'.~associated with withdrawal symptoms and persistent, sometimes overwhelming, ~ r a v i n g . ~Craving, .~ defined as "an irresistible urge to use a substance that compels drug-seeking behavior"I0 can be generated spontaneously by internal mood states such as dysphoria or boredom or triggered by external prompts that remind the addict of some facet of cocaine use or effect. Examples of such external triggers include large amounts of cash, white powder,
From the Departments of Psychiatry and Laboratory Medicine, University of Minnesota School of Medicine, the Chemical Dependency Treatment Program, and the College of Pharmacy, University of Minnesota. Supported in part by a grant from the Chemical Dependency Program Division, Minnesota Department of Human Services (St. Paul, Minn. ) . Received for publication Feb. 15, 1991; accepted March 3, 1991. Reprint requests: James A. Halikas, MD, Department of Psychiatry, University of Minnesota, Box 393 UMHC, Minneapolis, MN 55455. 1311129228
the familiar smell, or the sound of a crackling noise. Craving is intense but f l ~ c t u a t e s . ~Craving ~ ~ ~ ~ ~per" sists intermittently for many month^.^ The frequency and duration of each episode of craving varies from addict to addict and situation to situation." In highrisk situations, craving may reappear indefinitely." This often-irresistible urge may have a different neurophysiologic mechanism than the immediate reinforcing properties of the drug that prompt repetition of early-use behavior." An important neuropharmacologic property of cocaine is "kindling." In animals cocaine sensitizes brain cells in the limbic system to fire at progressively lower drug concentrations with repeated intermittent co'~ of kindling have been caine e x p ~ s u r e . ' ~ -Stages identifiedI3 and described14 during the past 20 years, so the definition of kindling is no longer limited to overt seizure activity but now encompasses stereotypic movements and other changes in animal behavior. Recent work has suggested that one possible mechanism for the antiseizure efficacy of carbamazepine may be through its amelioration or elimination of kindling.I5 One of us (J.H.) hypothesized that this progressive
CLIN PHARMACOL THER JULY 1991
82 Halikas e t al. Table I. Drug and payment scheduling protocol Study day
Carbamazepine first 200 mg (n = 8) 400 mg (n = 8) Placebo first 200 mg (n = 8) 400 mg (n = 8) Dailypayment($)
100 200 200 400
200 400
200 400
200 200 400 400
200 200 400 400
200 400
200 400
0 0 5
0 0 5
0 0 5
0 0 5
0 0 5
0 0 5
100 200 200 400 5 10
0 0 5
0 0 5
sensitivity to cocaine at a cellular level may possibly be the physiologic explanation of the human cocainerelated behavior of craving." Carbamazepine was therefore suggested as an anticraving drug. The medication was tried in an open clinical mixed inpatientoutpatient setting with 35 The results indicated that carbamazepine reduced the acute withdrawal symptoms, minimized or eliminated the early and delayed craving urges, produced greater cocaine abstinence, and somewhat blocked cocaine's euphoric effects during subsequent cocaine experimentation,ll.16-18 A subsequent anecdotal report has also noted this blockage effect.I9 Experience with this clinical population indicated that the use of carbamazepine was well tolerated and safe. Treatment with carbamazepine was used by patients for approximately 3 to 12 months, after which it was spontaneously discontinued with no or minimal resurgence of craving. Clinically, cocaine-dependent patients appeared to need relatively low carbamazepine doses. Management, particularly during the induction phase, required clinical judgment and dosage modification based on side effects or breakthrough craving. The clinical results were encouraging enough to warrant the development of a double-blind research paradigm. Past outpatient treatment studies with chemically dependent individuals have used either patients motivated for treatment or patients under some coercion to maintain abstinence. This complicates studies of treatment efficacy, particularly studies of the efficacy of pharmacologic agents in discouraging drug use. Ideally, medication should demonstrate efficacy in the absence of psychologic assistance in a setting conducive to accurate self-report of drug-use information. Successful pharmacologic intervention should not require that a patient be motivated to give up cocaine use to be effective. Rather, the patient must only be willing to take the medication as prescribed. Theoretically, motivation should not be a prerequisite for anti-
0 0 5
100 0 200 0 200 400 10
0 0
0 0
0 0
0 0
200 400 10
200 400 10
200 400 10
200 400 20
craving medication efficacy any more than for antibiotic, anti-inflammatory, antidepressant, or antiarrhythmic efficacy.
METHODS Subjects In the study to be reported the variables of motivation and coercion were eliminated with a noncoercive, nontreatment situation. Unmotivated street crack addicts were found by word of mouth on the streets in the midst of the crack neighborhoods of Minneapolis who, for daily payment, would participate. The research site used was a nontreatment facility, a vacant laboratory area of a largely abandoned hospital building in the subjects' neighborhood. All subjects were poor adult men who indicated that they had no current wish to stop using crack cocaine. They were unmotivated for drug treatment. The informed consent form told them that this was a study of the efficacy of carbamazepine in reducing or eliminating their cocaine use. They were assured that all information obtained was strictly confidential, protected by federal guidelines, and could not be subpoenaed or in any other way used against them. Their sole apparent motivation for volunteering for this study was daily immediate cash payment for each visit. Design This was a 20-day, double-blind, placebocontrolled, crossover study of 32 chronic crack users. Each volunteer received carbamazepine for 10 days and placebo for 10 days in an AB or BA design (Table I). Subjects attended the study site daily, ingested their medication under observation, and participated in a research visit. Subjects received daily immediate cash payment at each visit, on a sliding scale to encourage study completion: $Stday through day 10, $10/day through day 15, and $20lday through day 20 (Table I). A missed day of attendance deferred the payment schedule by 1 day, thereby eliminating one
VOLUME 50 NUMBER 1
of the $20 days at the end of the study. A subject could collect a maximum of $235 during the entire 22day period. Half of the subjects received active medication, either 200 or 400 mg carbamazepine in one daily dose, during the first half of the study and placebo in the second half; the other half of the subjects reversed the order and received placebo during the first half of the study. Both order of medication and dose were assigned randomly. Because each subject received 9 consecutive days of full carbamazepine dose and 9 consecutive days of placebo in this AB or BA design, each served as his own control. Days with graduated or tapered doses (first and last days with each condition) were eliminated from statistical analysis to minimize crossover effects. Each subject received two tablets per day under direct visual observation regardless of the assigned condition. Daily supervised urine samples were obtained. Both active carbamazepine (200 mg tablet of Tegretol) and identical inert placebo were supplied by Ciba-Geigy Corp. (Summit, N. J.). Admission criteria to the study required selfreported current and past crack cocaine use at the level of at least 30 days of use in the previous 100 days. Craving, or compulsion to use cocaine, was not a requirement for admission. There were no restrictions regarding other drug use or alcohol use. There was no washout period. Subjects were at least 18 years of age and had at least an eighth-grade education. Subjects with clinically significant medical illnesses, seizure disorder, trauma-induced unconsciousness within the past year, or organic brain syndrome were excluded from the study, as were those with clinically significant abnormal laboratory findings at screening that could not be explained on the basis of cocaine use.
Procedures At the screening visit, after informed consent was obtained, the following information was collected: alcohol and drug use history; Addiction Severity In-
Cocaine reduction in crack users 83 dex''; medical and psychiatric history; physical examination; blood tests, including chemical profile, liver studies, white count and differential, thyroid profile, creatine phosphokinase, and, if indicated, creatine kinase MB; urinalysis and urine drug screen; and electrocardiogram (ECG). After all test results were returned, there was a baseline visit followed by 20 daily visits. At baseline and each subsequent daily visit the following data were collected: vital signs, ECG, urinalysis for cocaine alone, self-report cocaine craving scale," cocaine withdrawal scale, self-report of cocaine use and route, and self-report of other drug use in the past 24 hours. Monitoring of alcohol use during the study was based on self-report only. The blind medication was then taken under direct observation, and the subject was paid for the visit. Subjects were regularly encouraged by study staff to maintain abstinence from all drugs and alcohol.
Instruments used Carbamazepine serum levels were analyzed by HPLC." The standard curve between 0.5 and 20 pgiml showed a coefficient of variation of less than 5%. The method of urinalysis used was a semiquantitative fluorescence polarization immunoassay technique, with the Abbott TDx system (Abbott Diagnostics, Pasadena, Calif.). The TDx was calibrated in the range of 0.3 to 5 pglml to assess variations in very low levels of cocaine metabolite on a daily basis. Cocaine saturation and sequestration may complicate interpretation of urine test r e s ~ l t s , ~ ~ ~ ~ ~ the lthou~h presence of cocaine metabolite (benzoylecgonine) has usually been considered evidence of cocaine use within the previous 48 to 72 h o ~ r s . ~ ~ . ~ ~ Daily outcome measures Urinalysis. Positive cocaine presence was reported on the basis of presence of lowest measurable cocaine metabolite (0.30 pglml). The number of positive urine samples within each drug condition was calculated as a ratio of the observed number of positive urine samples per 9 days. Concentration of cocaine metabolite above the lowest amount of 0.30 pgiml was measured by the fluorescence polarization immunoassay technique for concentrations up to 5.0 pglml (the lower 0.1% to 1.O% of positive cocaineicocaine metabolite levels seen clinically in chronic users). Self-reported cocaine use. Days reporting cocaine use were calculated as a ratio of the number of days reporting use per 9 days within each drug condition; and, the reported amount of cocaine used. Self-reported ratings. Self-reported ratings were
84 Halihas et al. craving intensity,'' craving frequency, craving duration, withdrawal symptoms, withdrawal severity score, and global improvement. The Cocaine Withdrawal Scale is a 19-item scale of withdrawal symptoms (e.g., irritable, anxious, and depressed) gleaned from the cocaine treatment literature, modeled after the Minnesota Tobacco Withdrawal heck list.^^ Subjects respond on a four-point scale (0 = none, 1 = mild, 2 = moderate, and 3 = severe) based on symptoms at the time of completion. Total withdrawal score is computed by summing responses across items, with possible scores ranging from 0 to 57. Total positive symptom count is a count of non-zero items (i.e., responses of 1, 2, or 3), with possible scores ranging from 0 to 19. Clinical ratings. Clinical ratings were global improvement, side effects, and therapeutic effect of the medication. Clinical ratings of "therapeutic effects" are made on a four-point scale (0 = unchanged or worse, 1 = minimal, 2 = moderate, and 3 = marked) and are based exclusively on the perceived therapeutic effects of the medication in achieving reduction in craving, frequency or amount of cocaine use, and preoccupation with cocaine. The Patient Global Improvement Scale and Clinical Global Improvement Scale are subjective ratings of the patient's condition on a seven-point scale (1 = very much worse, 4 = no change, and 7 = very much improved), used often in outpatient drug studies. The Patient Global Improvement Scale and Clinical Global Improvement Scale are ratings of overall life conditions and may be based on subjective factors in addition to those relating directly to the therapeutic effects of the medication (e.g., employment, interpersonal relationships, and personal grooming). Vital signs. Vital signs included blood pressure and pulse, both sitting and standing, temperature, weight, and ECG. Blood samples. Laboratory tests performed on days 5, 10, 15, and 20 included complete blood count, chemistry profile, liver function tests, and serum carbamazepine level.
Statistical analysis The maximum number of missing observations on any dependent variable was four for any one subject. A within-subjects analysis of variance indicated no differences in the number of missing observations for carbamazepine and placebo conditions for each dependent variable. Missing observations were then replaced by computing the mean of the observations im-
CLIN PHARMACOL THER JULY 1991
mediately before and after the missing days in both drug conditions. The 2 days on which subjects received titrated doses (i.e., days 1 and 11 for the carbamazepine-first group and days 10 and 20 for the placebo-first group) were eliminated from statistical analysis to minimize crossover effects between drug conditions (Table I). The resultant design produced 9 days of observation within each drug condition (i.e., carbamazepine and placebo) for each subject; two between-subjects factors included drug order (i.e., carbamazepine first or placebo first) and dose (200 or 400 mg). A repeated-measures analysis of covariance was performed separately for each dependent variable (except nominal measures), with baseline scores serving as the covariate for between-subjects comparisons. Main-effect tests were computed for drug condition, drug order, and dose. A test of the drug condition-bydrug-order interaction was also computed to examine temporal changes from the first 9 days to the second 9 days of observation. An auxiliary analysis was performed to examine the impact of carbamazepine serum levels on outcome measures. Previous clinical experience with carbamazepine in patients using cocaine" and patients with epilepsy27,2Xhas suggested a minimum therapeutic serum level of approximately 4 pglml. Subjects were therefore divided into those who had attained a serum level of 4 pglml at some point in the study (n = 19) and those who had not (n = 13). A separate repeatedmeasures analysis of covariance was then performed for each outcome measure for the two serum levels with only the 9 carbamazepine observation days, with baseline scores serving as the covariate.
RESULTS Seven subjects were screened out because of existing medical conditions, illiteracy, or below-minimum cocaine use admission criteria. An additional five subjects failed to return after the screening visit. Nine subjects who were enrolled and dispensed medication failed to complete the study. Two subjects quit during the carbamazepine-first, one during the carbamazepine-second (average length, 9.7 days; SD, 4.6), and six during the placebo-first condition (average length, 5.0 days; SD, 2.8). There were no differences in the rate or severity of side effects found in either group of dropouts. Of the 16 total days of full medication for the three carbamazepine-condition dropouts, only 1 day of mild side effects was reported. The 32 subjects who completed the study were
VOLUME 50 NUMBER 1
Cocuine reduction in mack users 85
Table 11. Sociodemographic characteristics of the sample (n = 32) Marriage
Never married Divorced Separated Widowed Married Current common law relationships
n
Housing
n
19 Living alone in transient facilities 4 Rooming house 2 With sexual partner 2 2 3
12 CYear college degree 7 4
6 2
With relatives With friends On the street
Highest level of education
n 2
Some college High school completion or GED Some high school
