`
`NERVOUS AND
`MENTAL DISEASE
`
`VOL. 183, NO. 9
`
`September 1995
`SERIAL NO. 1354
`
`Alcohol, Cannabis, Nicotine, and Caffeine Use and
`Symptom Distress in Schizophrenia
`
`EDNA HAMERA, Ptt.D, 1 JOANNE KRAENZLE SCHNEIDER, PH.D.,2 AND STANLEY DEVINEY, PH.D.3
`
`The high prevalence of substance use, e.g., alcohol and illegal and nonprescribed drugs,
`in schizophrenia is widely recognized. One explanation for this high prevalence is that sub(cid:173)
`stance use may be a self-initiated method for managing symptoms. To test whether the intake
`of four substances-alcohol, cannabis, nicotine, and caffeine-would increase with increases
`in symptom distress, daily self-reports of symptom distress and substance intake over 12
`weeks were analyzed with pooled time series analyses. Compliance with neuroleptic medica(cid:173)
`tion was added to the analyses to control for any changes in prescribed medication compliance
`while using nonprescribed drugs or alcohol. Of the four substances studied, only nicotine
`was significantly related to symptom distress. Higher distress with prodromal symptoms was
`related to decreases in nicotine use. Analysis of caffeine did not meet the criteria for signifi(cid:173)
`cance but does provide direction for further research. Higher distress, with neurotic symp(cid:173)
`toms, was related to increases in caffeine use. Further research is needed to clarify the
`relationship between nicotine and symptoms.
`
`-J Nerv Ment Dis 183:559-565, 1995
`
`One explanation for the high prevalence of substance
`use in schizophrenia is that alcohol and drugs may be
`a self-initiated method for managing symptoms. Sub(cid:173)
`stance use in schizophrenia and its association with
`increases in psychosis, relapse, and rehospitalization
`is widely recognized (Carey et al., 1991; Cuffel, 1992;
`Kivlahan et al, 1991; Regier et al., 1990). If this high
`prevalence is an attempt to self-regulate symptoms, cli(cid:173)
`nicians need to be able to identify this behavior and
`offer alternative means of alleviating symptom distress.
`Studies supporting the use of alcohol and drugs to
`regulate internal experiences and symptoms are based
`on patient self-reports that relate symptom changes to
`substances used. Test and colleagues (1989) found sub(cid:173)
`jects reported more positive than negative changes in
`symptoms after recent substance use regardless of the
`substance used. Other investigators have focused on
`specific substances and their relationship with symp(cid:173)
`toms. Some subjects with schizophrenia reported that
`
`1 School of Nursing, University of Kansas Medical Center, 3901
`Rainbow Boulevard, Kansas City, Kansas 66160-7700. Send reprint
`requests to Dr. Hamera.
`2 Washington University, School of Medicine, St. Louis, Missouri.
`3 Department of Social Sciences, University of Maryland-Eastern
`Shore, Princess Anne, Maryland.
`This research was supported by grant R03 MH46650 from
`NIMH, ADAMHA.
`
`alcohol improved tension and depression. A smaller
`number reported that alcohol either relieved or
`worsened psychotic symptoms (Alpert and Silvers,
`1970; Bergman and Harris, 1985; Noordsy et al., 1991).
`Differential effects for alcohol, cannabis, and cocaine
`on symptoms has been reported by patients with
`schizophrenia (Dixon et al., 1991). Both alcohol and
`cannabis decreased anxiety, whereas cocaine in(cid:173)
`creased anxiety. Cocaine and cannabis increased en(cid:173)
`ergy more than alcohol, and cannabis increased suspi(cid:173)
`ciousness more than alcohol. Castaneda and colleagues
`(1991) also reported differential symptom effects for
`the kind of substance used. Cocaine users reported
`more symptoms worsened than improved, while those
`who used alcohol reported a balance between symptom
`improvement and worsening.
`Although believed to be more benign, caffeine and
`nicotine are drugs that individuals with schizophrenia
`can readily use to modulate symptoms. Individuals with
`schizophrenia are more likely to smoke than either indi(cid:173)
`viduals with other psychiatric disorders or the general
`public (Hughes et al., 1986; Lohr and Flynn, 1992). Some
`smokers with schizophrenia have reported smoking in
`response to auditory hallucinations and medication
`side effects (Glynn and Sussman, 1990). In a cross(cid:173)
`sectional study of 78 outpatients with schizophrenia,
`
`559
`
`CFAD VI 1024-0001
`
`CFAD VI 1024-0001
`
`
`
`560
`
`HAMERA et al.
`
`Goff and colleagues (1992) found smokers had fewer
`extrapyramidal symptoms than nonsmokers when gen(cid:173)
`der, age, and caffeine consumption were controlled.
`This was despite the fact that smokers were prescribed
`twice the amount of neuroleptic medication. No differ(cid:173)
`ences were found between any other symptom and
`smoking status when using neuroleptic dose, caffeine,
`and age as covariates.
`Caffeine consumption by individuals with schizo(cid:173)
`phrenia is also high but tolerance appears to develop
`with chronic high intake. Koczapski and colleagues
`(1989) did not find changes in staff ratings on inpatients
`related to withdrawal or reintroduction of caffeine in
`inpatients. However, acute administration of caffeine
`has been found to produce significant changes in symp(cid:173)
`toms when preceded by a period of withdrawal from
`caffeine (Lucas et al., 1990). Psychotic and thought dis(cid:173)
`order symptoms worsened, while improvement was
`seen in mood, energy, and social involvement. These
`effects, however, appear to be short lived. The absorp(cid:173)
`tion of caffeine when taken concurrently with neuro(cid:173)
`leptic medication may lead to impairment in the absorp(cid:173)
`tion of both (Mikkelsen, 1978). Thus, individuals who
`consume large amounts of caffeine over long periods
`of time should experience fewer effects of both caffeine
`and neuroleptic medication.
`These studies are all based on subjects' recall of
`subjective experiences. Differences in findings may re(cid:173)
`flect the length of time between actual use and recall
`of use as well as whether subjects were asked to recall
`immediate versus delayed symptomatic effects of the
`substance used. A common problem in all the studies
`is that subjects are directed to link substance intake
`with symptoms. These post hoc attributions may ob(cid:173)
`scure the true natural relationship between symptoms
`and substance use, since they are susceptible to self(cid:173)
`perception biases. The purpose of the present time se(cid:173)
`ries study is to examine the daily relationship between
`symptom distress and substance intake. In addition,
`the role of compliance with neuroleptic medication in
`the relationship between symptoms and substance use
`is considered. The hypothesis tested in this study is
`that the intake of four substances-alcohol, cannabis,
`nicotine, and caffeine-will increase as symptom dis(cid:173)
`tress increases.
`
`Methods
`
`Subjects
`
`Subjects were recruited from community support
`programs (CSP) of two community mental health cen(cid:173)
`ters serving adjacent counties. Both CSPs used assert(cid:173)
`ive case management as their main modality of treat(cid:173)
`ment and offered an array of services, such as
`medication management and vocational counseling.
`
`The CSPs differed in the socioeconomic status of the
`counties they served. Initially, subjects (N = 14) were
`recruited from the first CSP, which serves a county
`with a per capita income of $23,346 and a 6% minority
`population. Three additional subjects were recruited
`from a second CSP serving a county with a per capita
`income of $12,752 and a 37% minority population.
`Clinical diagnosis of schizophrenia or schizoaffective
`disorder was confirmed by administering the Mood
`Syndromes and Psychotic and Associated Symptoms
`sections of the Structured Clinical Interview for DSM(cid:173)
`III-R (SCID; (Spitzer et al., 1989). Fifteen (88.2%) sub(cid:173)
`jects had a lifetime SCID diagnosis of schizophrenia.
`Two (11.8%) subjects had a lifetime SCID diagnosis of
`schizoaffective disorder.
`A criterion for participating in the study, the self(cid:173)
`report of alcohol and/or drugs at least weekly, was
`assessed by having subjects complete an investigator(cid:173)
`developed health behavior inventory. Of the subjects
`who met the diagnostic and substance use frequency
`criteria, 14 of 20 (70%) from the first CSP agreed to
`participate and three of four subjects (75%) from the
`second CSP agreed to participate. After procedures
`were explained fully, written consent was obtained.
`Four of the subjects did not complete the entire 84
`days of the study. One subject who had an unstable
`living situation dropped out after 21 days, one subject
`experiencing symptom exacerbation dropped out after
`28 days, one subject decided to drop out after 56 days,
`and one subject was hospitalized after completing 63
`days of the study. Data from all 17 subjects were in(cid:173)
`cluded in the analyses. Excluding the days after the
`four subjects dropped out, there was a total of 52 ( 4%)
`days with missing data.
`The size of the sample (N = 17) was adequate to test
`the hypothesis because the investigators pooled the
`individual time series (17 Ss X 84 days) (Kessler and
`Greenberg, 1981). Assuming a small effect size, 537
`data points (subjects X days) were needed to test the
`hypothesis (Cohen and Cohen, 1983). After adjustment
`for days lost (from subjects who dropped out before
`completing the study), power exceeded .90.
`
`Measures
`
`Three self-administered checklists were used in this
`study to measure symptom distress, substance use, and
`medication compliance. The Symptom Checklist is
`composed of 12 symptoms modified for self-administra(cid:173)
`tion from the expanded version of the Brief Psychiatric
`Rating Scale (Lukoff et al., 1986). Subjects rated the
`amount of distress with each symptom for the previous
`24 hours by checking one of five choices ranging from
`"not distressing at all" to "it was extremely distressing."
`A sixth choice, "did not have symptom," was also given
`so subjects would have a place to check when syrup-
`
`CFAD VI 1024-0002
`
`CFAD VI 1024-0002
`
`
`
`SUBSTANCE USE AND SYMITOM DISTRESS
`
`561
`
`toms were not present. This sixth choice was also
`scored as "not distressing at all." In addition to the 12
`symptoms, subjects also rated their distress on as many
`as three idiosyncratic prodromal symptoms. These are
`symptoms each subject identified as personal indica(cid:173)
`tors of relapse. Data supporting the validity and reliabil(cid:173)
`ity of the Symptom Checklist were assessed in a pilot
`study (N = 29). 4 Subjects' self-administered ratings of
`symptom distress correlated .81 with an interviewer(cid:173)
`administered Brief Psychiatric Rating Scale. The inter(cid:173)
`nal consistency of the Symptom Checklist using coeffi(cid:173)
`cient alpha was . 78.
`The Substance Use Checklist consisted of 33 sub(cid:173)
`stances, including caffeine, cigarettes, alcohol, illegal
`drugs, over-the-counter drugs, solvents, and inhalants.
`Subjects checked the substances they used and entered
`the amount of intake in the previous 24 hours.
`The validity of the subjects' self-report of substance
`use was assessed by urine drug screening. Urine speci(cid:173)
`mens were collected from each subject during two ran(cid:173)
`domly chosen weeks within the 12-week study. Urine
`was screened for amphetamines, cannabinoids (THC),
`alcohol, cocaine, barbiturates, opiates, and benzodiaze(cid:173)
`pines using an enzyme immunoassay (SYVA Emit"').
`Drug screening was performed by a laboratory certified
`by the National Institute for Drug Abuse. The urine
`screenings showed only one discrepancy between self(cid:173)
`report and use. One subject did not report using co(cid:173)
`caine that did show up in the urine screening. Analyses
`were computed both with and without this subject's
`data. The results were not significantly changed, so this
`subject's data were retained.
`The subjects' medications were listed on the Medica(cid:173)
`tion Checklist. Subjects were instructed to check one
`of four responses (took as prescribed, took some but
`not all prescribed, did not take at all, and took more
`than prescribed) to indicate how they took each of
`their medications in the previous 24 hours. For data
`analyses, neuroleptic medication was used as a covari(cid:173)
`ate for those subjects who took oral neuroleptic medi(cid:173)
`cations. To be added as a covariate, compliance had to
`be a dichotomous variable. Based on the understanding
`that some subjects were advised to adjust their daily
`dosages as needed, compliance was described as any
`of the following: "took as prescribed," "took some but
`not all," or "took more than prescribed." Noncompli(cid:173)
`ance was: "did not take at all."
`
`Procedures
`
`Prior to the study, the potential subject pool was
`assessed at the first CSP for data collection by inter-
`
`4 Hamera E, Schneider JK, Potocky M, Casebeer MA. Validity of
`self-administered symptom scales in clients with schizophrenia and
`schizoaffective disorders. Manuscript submitted for publication.
`
`viewing case managers. Case managers identified 40
`individuals, or 23% of their caseload, with clinical diag(cid:173)
`noses of schizophrenia or schizoaffective disorders
`who were using drugs or alcohol. Of the 40 potential
`subjects, case managers believed 21 used drugs and
`alcohol at least once or twice a week, a criterion for
`participation in this study. The potential subject pool
`was not assessed at the second CSP because the pro(cid:173)
`gram was in the midst of relocating and enlarging.
`Subjects were recruited for the study by two meth(cid:173)
`ods. Case managers identified subjects thought to be
`eligible, i.e., those who had a clinical diagnosis of
`schizophrenia or schizoaffective disorder and were be(cid:173)
`lieved to be using either alcohol and/or drugs. Subjects
`also were recruited directly by posting sign-up sheets
`at the CSP. Individuals who indicated interest in partici(cid:173)
`pating were contacted by members of the research
`team to clarify eligibility.
`Subjects were given the checklists in notebooks with
`dividers separating each day of the week The order of
`the checklists was always consistent. Symptom distress
`was listed first, medication compliance was second,
`and substance use was third. Subjects were taught how
`to complete each of the checklists. A place to keep
`the notebook where they lived was identified and a
`consistent time each day to complete the checklists
`was specified. To help subjects remember to fill out
`the checklists daily, the time to complete them was
`linked to a habitual activity. Twice weekly, subjects met
`with a nurse who reviewed the completed checklists,
`clarified responses, and replenished the notebooks.
`Subjects completed the forms daily for 84 days. They
`were paid a total of $165 on a scheduled basis for
`their participation.
`
`Data Analysis
`
`Daily ratings for the 84 days of symptom distress and
`amount of alcohol and drug intake from the 17 subjects
`were entered into a pooled time series. Daily compli(cid:173)
`ance ratings of neuroleptic medication were used as
`a covariate. The pooled time series design, using the
`software selected for this study (Guass-TSCS 2.1, Ap(cid:173)
`tech Systems, Inc., Kent, Washington), analyzes data
`with unequal time series per subject. Normally, re(cid:173)
`peated-measures designs require that each subject con(cid:173)
`tribute data at each collection wave.
`The use of pooled time series data presents problems
`with nonindependence of data points over time (days
`in the present study). This nonindependence raises the
`possibility of correlated error. A Durwin-H was calcu(cid:173)
`lated for each equation to determine the presence of
`correlated error. Significant correlations were found
`in each equation. To adjust for this correlated error,
`generalized least squares regression (GLS), rather than
`
`CFAD VI 1024-0003
`
`CFAD VI 1024-0003
`
`
`
`562
`
`HAMERA et al.
`
`ordinary least squares regression, was used (Sayrs,
`1989).
`While the main focus of this study is on the structural
`parts of the equations that examine the relationship
`between symptoms and substance intake, the univari(cid:173)
`ate ARIMA models were computed separately on the
`substance intake and symptoms for each subject as
`well. First-order autocorrelations indicated that ratings
`for consecutive days were correlated.
`
`Results
`
`Demographic and Treatment Characteristics
`
`Thirteen (76.5%) of the 17 subjects were male; four
`(23.5%) were female. Age ranged from 21 to 54 years
`(mean age = 34.2 ± 8.5 years). Three subjects were
`black, one was a Pacific Islander, and the rest were
`white. One subject had less than a high school educa(cid:173)
`tion, seven had a high school education or equivalent,
`eight had some college, and one had a bachelor's de(cid:173)
`gree. Eight subjects were competitively employed, five
`were employed by the mental health center, and one
`was not employed. Most subjects lived independently
`(}11 = 12), three lived with relatives, and two had other
`living arrangements. The subjects' average monthly in(cid:173)
`come ranged from $379 to $2200 (mean = $597.3 ±
`425.8). All of the subjects received case management
`services.
`With the exception of one subject, all had one or
`more lifetime alcohol and/or drug abuse or dependency
`disorders. Fourteen subjects had a lifetime diagnosis
`of alcohol abuse or dependency. Twelve subjects had
`a lifetime diagnosis of cannabis abuse or dependency.
`A smaller number of subjects had lifetime diagnoses
`for stimulants, cocaine, opioid, hallucinogens, seda(cid:173)
`tives, and polydrug abuse or dependency.
`
`Symptom Distress
`
`Items from the Symptom Checklist that intercorre(cid:173)
`lated .6 or higher were parceled into subsets of items.
`This reduces multicollinearity, which is especially
`problematic in GLS. One item measuring confusion cor(cid:173)
`related highly with more than one symptom parcel and
`therefore was deleted from the analyses.
`Parcel I-Ideas of Reference included two items: a)
`feeling that something on the television or radio was
`about the subject or was sending special messages to
`the subject, and b) feeling that the subject caused spe(cid:173)
`cial or unusual things to happen. The daily mean for
`these two items was 2.96 ± 1.55 (possible and observed
`ranges= 2-10). Parcel II-Neurotic Symptoms included
`items that related to feeling depressed, guilty, nervous,
`restless, irritable, and isolating self. Daily mean score
`for these six items was 10.55 ± 4.77 (possible and ob(cid:173)
`served ranges = 6-30). Parcel III-Hallucinations was a
`
`single item dealing with hearing voices or sounds or
`seeing, smelling, or tasting things that others did not.
`Parcel III had a daily mean of 1.55 ± 1.00 (possible and
`observed ranges = 1-5). Parcel IV-Talk or Move Slow
`dealt with feeling that you were talking or moving
`slower than usual. This item had a daily mean of 1.51
`± .92 (possible and observed ranges= 1-5). The single
`item of Parcel V-Paranoid Symptoms dealt with feeling
`paranoid or suspicious and had a daily mean of 1.69 ±
`.98 (possible and observed ranges = 1-5). Only subjects'
`first idiosyncratic prodromal sign or symptom was used
`to form Parcel VI-Prodromal Symptom because some
`subjects did not identify more than one prodromal
`symptom. The mean rating for subjects' prodromal
`symptom was 1. 75 ± .98 (possible and observed
`ranges= 1-5). Two subjects reported psychotic symp(cid:173)
`toms as their first idiosyncratic prodromal sign, while
`the rest reported a variety of nonpsychotic signs and
`symptoms. Although the full range of symptom distress
`ratings was observed from the sample, the mean symp(cid:173)
`tom ratings were slightly positively skewed, indicating
`mild distress. Pooled time series analysis is robust
`enough to handle this restriction in variability.
`
`Type and Amount of Substance Use
`
`All subjects used caffeine, 16 (94.1 %) smoked ciga(cid:173)
`rettes, and 16 (94.1%) drank alcohol. Nine (52.9%) sub(cid:173)
`jects used cannabis, four (23.5%) subjects reported tak(cid:173)
`ing ephedrine, two (11.8%) reported taking caffeine
`pills, two (11.8%) subjects reported cocaine use, one
`(5.9%) used amphetamine, and one (5.9%) used benzodi(cid:173)
`azepines. Table 1 depicts the mean and range of per(cid:173)
`centage of days of substance use for subjects who used
`alcohol, cannabis, nicotine, and caffeine. It also in(cid:173)
`cludes the mean and range of daily use for those days
`of reported use. Subjects' alcohol entries were con(cid:173)
`verted to number of drinks based on alcoholic content.
`Correlations among the amount of alcohol, cannabis,
`caffeine, and nicotine over the 84 days of the study
`were computed using the Pearson product-moment
`correlation coefficient. Two pairs of substances were
`significantly correlated. Nicotine was moderately corre(cid:173)
`lated with caffeine (r = .30, p ~ .001) and alcohol was
`weakly correlated with cannabis (r = - .07, p ~ .01).
`
`Medication Use
`
`All subjects were on a neuroleptic medication. Five
`(29.4%) received an injectable neuroleptic only. Nine
`(52.9%) took an oral neuroleptic medication only. Three
`(17.6%) of the subjects received both injections and oral
`neuroleptic medications. A review of clinical records
`showed that the subjects received scheduled injections
`during the study.
`
`CFAD VI 1024-0004
`
`CFAD VI 1024-0004
`
`
`
`SUBSTANCE USE AND SYMPTOM DISTRESS
`
`563
`
`TABLE 1
`Substances Used, Number and Range of Days Used, and Mean and Range of Daily Use
`No. of
`Subjects
`Using
`Substance
`
`For Subjects Who Used
`Range of
`daily use
`
`Mean daily use
`
`For Subjects Who Used
`Mean days
`Range of days
`used(%)
`used(%)
`
`Substance
`
`Alcohol
`Cannabis
`Nicotine
`Caffeine
`
`16 (94.1)
`9 (52.9)
`16 (94.1)
`17 (100)
`
`43.7
`25.9
`96.2
`91.1
`
`4.8--100
`1.2-100
`85.7-100
`61.9-100
`
`6.16 drinks"
`4.70 joints
`31.32 cigarettes
`9.44 cups
`
`1-37
`1-9
`1-98
`1-65
`
`" One drink equals 1 oz. hard liquor or 1 can/bottle of beer or 5 oz. of wine.
`
`Relationship of Symptom Distress
`and Substance Use
`
`The relationships between the six symptom parcels
`and the amount of alcohol, cannabis, nicotine, and caf(cid:173)
`feine were analyzed by pooled time series analyses.
`Analyses were performed with and without the covari(cid:173)
`ate of compliance with oral neuroleptic medication.
`Nonstandardized parameter estimates were reported
`rather than standardized parameter estimates because
`the covariate neuroleptic compliance was a dichoto(cid:173)
`mous variable. Nonstandardized parameter estimates
`are also easier to interpret when using GLS and compar(cid:173)
`ing parameter estimates from analyses of different sam(cid:173)
`ples (Pedhazur, 1982). Although not completely inde(cid:173)
`pendent, the 12 subjects (768 cases) taking oral
`neuroleptic medications were analyzed as a group to
`examine the effect of medication compliance on each
`of the four substances in addition to analyses of the
`total 17 subjects (1189 cases). Four sets (with alcohol,
`caffeine, nicotine, and cannabis as dependent vari(cid:173)
`ables) of two regression equations (with and without
`the covariate neuroleptic compliance) were run for a
`total of eight analyses. The six symptom parcels were
`entered in each equation as predictor variables. Be(cid:173)
`cause this number of analyses increases family-wise
`error, an alpha level of .006 was used to identify signifi(cid:173)
`cant findings (Keppel, 1991).
`The results of the pooled time series on the same(cid:173)
`day ratings of symptoms and reports of substance use
`are shown in Table 2. Significant parameter estimates
`were obtained with nicotine as the dependent variable,
`but not when alcohol and cannabis were used as depen(cid:173)
`dent variables. The analyses with caffeine as the depen(cid:173)
`dent variable yielded parameter estimates that were
`very near the accepted alpha level of .006.
`For Symptom Parcel VI-Prodromal Symptom, param(cid:173)
`eter estimates were significant when the dependent
`variable was nicotine, measured by the number of ciga(cid:173)
`rettes smoked. An inverse relationship was found with
`nicotine. The nicotine intake decreased with increased
`distress with prodromal symptoms (Symptom Parcel
`VI), both with and without the covariate of neurolep(cid:173)
`tic compliance.
`
`Although not meeting the criterion for significance
`(.006), the relationship between caffeine and neurotic
`symptoms (Symptom Parcel II) reached an alpha level
`of .007. Caffeine intake increased as distress with neu(cid:173)
`rotic symptoms increased. This was true with and with(cid:173)
`out the neuroleptic compliance covariate.
`
`Discussion
`
`This study provides a rigorous test of the symptom
`self-regulation explanation for the high prevalence of
`substance use in individuals with schizophrenia. The
`type of analysis performed allowed us to examine the
`relationship between changes in symptom distress and
`changes in substance intake. Analysis and use of a
`pooled time series design make it difficult to compare
`the results with previous studies that have examined
`subjects' post hoc attributions about the relationship
`of symptoms to substance intake. Instead of asking
`subjects to link their symptoms with substance intake,
`in the present study we examined the natural relation(cid:173)
`ship between daily ratings of symptom distress and
`daily substance intake over an 84-day period.
`The absence of significant findings with alcohol and
`cannabis in the present study extends the findings of
`previous researchers (Alpert and Silvers, 1970; Berg(cid:173)
`man and Harris, 1985; Dixon et al., 1991; Noordsy et al. ,
`1991) who reported that some, but not all, subjects
`make post hoc associations linking alcohol and canna(cid:173)
`bis use with a decrease in symptoms. The present
`findings cast doubt on the empirical bases of self-re(cid:173)
`ports that alcohol and cannabis are used to regulate
`symptom distress. However, because investigators se(cid:173)
`lected a population who were not too impaired to par(cid:173)
`ticipate, the findings can be generalized only to individ(cid:173)
`uals with schizophrenia who are similar to the
`sample studied.
`The nicotine findings are new. Nicotine has not been
`studied as a method of self-regulating symptoms to the
`extent that alcohol and drugs have been. The significant
`inverse relationship between prodromal symptom dis(cid:173)
`tress and nicotine is contrary to the hypothesized direc(cid:173)
`tion that increases in symptom distress lead to in(cid:173)
`creases in substance use . This observed inverse
`
`CFAD VI 1024-0005
`
`CFAD VI 1024-0005
`
`
`
`564
`
`HAMERA et al.
`
`TABLE 2
`Unstandardized Parameter Estimates from Generalized Least Squares Regression of Effects of Symptom Distress
`on Use of Alcohol, Cannabis, Caffeine, and Nicotine
`Dependent Variables
`
`Alcohol
`
`Cannabis
`
`Caffeine
`
`Nicotine
`
`WIO
`neuroleptic
`covariate
`(1190 cases)
`
`WI
`neuroleptic
`covariate
`(768 cases)
`
`WIO
`neuroleptic
`covariate
`(1 190 cases)
`
`WI
`neuroleptic
`covariate
`(768 cases)
`
`WIO
`neuroleptic
`covariate
`(1189 cases)
`
`WI
`neuroleptic
`covariate
`(768 cases)
`
`WIO
`neuroleptic
`covariate
`(1191 cases)
`
`WI
`neuroleptic
`covariate
`(768 cases)
`
`Independent
`Variables
`
`2.23
`.10
`-.04
`- .12
`.00
`
`Constant
`Ideas of Reference
`Neurotic Sx
`Hallucinations
`Talk and
`Move Slow
`Paranoid Sx
`Prndromal Sx
`N euroleptic
`Compliance
`* p :s; .007; ** p :s; .006; *** p :s; .001.
`
`.17
`.20
`
`2.65
`.03
`- .05
`-.01
`.06
`
`.17
`.20
`- .14
`
`.54
`-.02
`.00
`.02
`.06
`
`-.01
`.00
`
`.89
`- .06
`.01
`.05
`.03
`
`- .03
`-.02
`- .12
`
`6.37
`.22
`.15*
`- .10
`- .32
`
`.00
`.11
`
`5.93
`-.05
`.13*
`- .01
`-.11
`
`.10
`.01
`.51
`
`27.76***
`.18
`.13
`.25
`- .74
`
`.90
`-1.12***
`
`29.74***
`.19
`.05
`.72
`- .54
`
`.43
`-1.15**
`-1.96
`
`relationship suggests either that subjects smoke more
`cigarettes when prodromal symptoms are less dis(cid:173)
`tressing or that they smoke fewer cigarettes when pro(cid:173)
`dromal symptoms are more distressing. This latter in(cid:173)
`terpretation of the relationship between smoking and
`symptoms suggests that smoking less may be a method
`of preventing relapse. Since cigarette smoking has been
`found to increase metabolism of neuroleptic medica(cid:173)
`tions, thereby decreasing blood levels, a decrease in
`smoking would make more medication available for
`use (Jarvik and Schneider, 1992).
`With the rapid effects of nicotine (Henningfield et al.,
`1993) and only one data point within a 24-hour period,
`the present study may not have allowed an adequate
`assessment of the relationship between smoking and
`symptoms. An experimental design examining short(cid:173)
`term changes in symptom distress, both in the presence
`and in the absence of nicotine, is needed.
`It is unclear why prodromal symptoms were the only
`significant symptom group related to substance intake.
`Since these symptoms were idiosyncratic-some psy(cid:173)
`chotic, others nonpsychotic-they differed widely in
`severity. Their only commonality is the association sub(cid:173)
`jects made with relapse. Since relapse is a disturbing
`experience, prodromal symptoms may be more dis(cid:173)
`tressing than other symptoms. The mean rating for pro(cid:173)
`dromal symptoms was slightly higher than for the other
`symptoms measured supporting this interpretation.
`Symptoms not included on the symptom checklist may
`also relate to substance intake. Deficit or negative
`symptoms and medication side effects are more diffi(cid:173)
`cult to measure by self-report.
`Although the relationship between caffeine and neu(cid:173)
`rotic symptoms did not meet the criterion for signifi(cid:173)
`cance, it provides a useful direction for further re(cid:173)
`search. When distress with neurotic symptoms such
`
`as depression and tension increased, caffeine intake
`increased. This suggests that subjects drink more cof(cid:173)
`fee when feeling tense and depressed. Tension and
`mood changes are symptoms of caffeinism. Individuals
`with schizophrenia may be caught in a cycle driven by
`the reinforcing effects of caffeine on symptom distress
`while attempting to avoid the withdrawal effects
`(Greden and Walters, 1992).
`The findings from this study are limited by sample
`selection and characteristics. To be in the study, sub(cid:173)
`jects had to self-report drug or alcohol use at least
`weekly. Subjects who denied substance use were not
`included, and the average daily consumption of alcohol
`and cannabis suggest that heavy users were not in(cid:173)
`cluded. In addition, these subjects may not represent
`substance-using individuals with schizophrenia from
`poorer inner-city areas, since the majority came from
`a wealthier suburban setting.
`The second issue limiting the findings pertains to
`self-report of substance intake. The issues involve reac(cid:173)
`tivity of self-monitoring and validity of self-report.
`Urine screening included in this study provides valida(cid:173)
`tion for the type of substance used but not the amount.
`Presently there is no way to obtain continuous mea(cid:173)
`sures of quantity over time other than self-report. There
`has been research on an older nonpsychiatric popula(cid:173)
`tion's daily self-reported quantities, which were compa(cid:173)
`rable both to recall at 1-week intervals and to collateral
`daily reports of intake (Samo et al., 1989). This finding
`supports the accuracy of self-report.
`Reactivity involves changes in drinking behavior due
`to self-monitoring. Several methods were used to de(cid:173)
`crease reactivity in this study. Rating multiple catego(cid:173)
`ries, e.g., symptoms, compliance, and numerous sub(cid:173)
`stances, tends to decrease reactivity (Hayes and Carver,
`1977, 1980). Another corrective precaution was the se-
`
`CFAD VI 1024-0006
`
`CFAD VI 1024-0006
`
`
`
`SUBSTANCE USE AND SYMPTOM DISTRESS
`
`565
`
`lection and training for data collection. The nurses col(cid:173)
`lecting data were not connected with the community
`support programs, and were trained to be nonjudg(cid:173)
`mental to increase openness. However, it is possible
`that recording errors occurred on days when larger
`amounts were used due to diminished recall.
`
`Conclusions
`
`In summary, the findings from this time series study
`show an inverse relationship between prodromal symp(cid:173)
`toms and nicotine, but not in the hypothesized direc(cid:173)
`tion. The most logical interpretation of this relationship
`is that individuals with schizophrenia smoke fewer cig(cid:173)
`arettes when prodromal symptoms are more dis(cid:173)
`tressing. Further research needs to examine subjects'
`symptom experien