Using Personality Assessment to Inform Comorbid Addiction Diagnosis

Sejmjl308

PersonalityandSubstanceUseDisordersAlcoholismVersusDrugUseDisorders.pdf

Home >Psychology homework help >Using Personality Assessment to Inform Comorbid Addiction Diagnosis

Journal of Consulting and Clinical Psychology 1999, Vol. 67, No. 3, 394-404

Personality and Substance Use Disorders: II. Alcoholism Versus Drug Use Disorders

Matt McGue University of Minnesota

Wendy Slutske University of Missouri

William G. lacono University of Minnesota

The relationship between personality and substance use disorders was investigated in a community-based sample of 638 individuals who were alcoholic and/or had a drug use disorder, and 1,530 individuals who did not have a substance use disorder. Personality was assessed by the Multidimensional Personality Questionnaire; substance use diagnoses were based on standard criteria as assessed by interview. Data were analyzed using a 3-factor (Gender X Alcoholism X Drug Use Disorder) multivariate analysis of variance. The significant alcoholism main effect was associated primarily with negative emotionality, whereas the significant drag use disorder main effect was associated primarily with constraint. No significant interactions with gender were observed. These findings suggest that the elevated levels of behavioral disinhibition observed with alcoholic individuals may be attributable to a subset of alcoholic individuals who also abuse drugs other than alcohol.

Several recent developments in psychology, as well as in the substance abuse research field, have helped spark a renewed in- terest in the role of personality in the etiology of alcoholism and other substance use disorders. First, behavioral genetic studies have consistently indicated that genetic factors contribute to risk of substance abuse (McGue, 1995; Pickens et al., 1991) and have thereby implicated the existence of inherited, individual-level risk factors for substance use disorders. Personality characteristics con- stitute an important instance of an individual-level risk factor that is both associated with substance abuse risk (e.g., Sher & Trull, 1994) and substantially heritable (e.g., Tellegen et al., 1988). Second, theoretical developments in the personality field have led to the advancement of hierarchical models of personality based on either a three- (Cloninger, 1987b; Eysenck, 1967; Tellegen, 1991) or a five-factor (Costa & McCrae, 1992) model. These models have facilitated research on personality not only by providing a theoretically coherent structure for the vast array of hypothesized personality traits but also by suggesting systematic approaches to the assessment of the major dimensions of personality. Finally, several prominent substance use researchers have proposed theo- retical models that accord personality factors a central role in the

Matt McGue and William G. lacono, Department of Psychology, Uni- versity of Minnesota; Wendy Slutske, Department of Psychology, Univer- sity of Missouri.

This research was supported in part by U.S. Public Health Service Grants AA00175, AA09367, and DA05147.

Correspondence concerning this article should be addressed to Matt McGue, Department of Psychology, University of Minnesota, 75 East River Road, Minneapolis, Minnesota 55455. Electronic mail may be sent to [email protected].

development of substance use disorders (Cloninger, 1987a; Tarter, Alterman, & Edwards, 1985).

The vast majority of research linking personality to substance use disorders has focused on alcoholism. This research (reviewed by Sher & Trull, 1994) is generally consistent in implicating two major dimensions of personality in the etiology of alcoholism: negative emotionality, or the tendency to experience negative mood states and psychological distress, and behavioral disinhibi- tion, or the inability or unwillingness to inhibit behavioral im- pulses. Compared with nonalcoholic individuals, alcoholic indi- viduals, on average, score higher on these two dimensions of personality, and there is some evidence, albeit limited, suggesting that these differences predate alcoholism onset (e.g., Caspi, Mof- fitt, Newman, & Silva, 1996; Cloninger, Sigvardsson, Reich, & Bohman, 1988; Kammeier, Hoffman, & Loper, 1973). Moreover, the association of these dimensions with alcoholism risk is robust, holding, for example, for both men and women (McGue, Slutske, Taylor, & lacono, 1997). Nonetheless, mean differences between representative samples of alcoholic and nonalcoholic individuals on indicators of negative emotionality and behavioral disinhibition appear only to be moderate in magnitude (McGue et al., 1997). If personality influences alcoholism risk it does so as one of a myriad of risk factors in a multifactorial system rather than as the effect of a unique configuration of personality characteristics, the posses- sion of which leads inevitably to an addiction (i.e., an "addictive personality"; Nathan, 1988).

The personality correlates of drug use disorders have been less widely researched than the personality correlates of alcoholism. Moreover, much of the research that does exist in this area suffers from serious methodological shortcomings, including the use of nonrepresentative, treatment-ascertained samples; poorly defined control groups; and psychometrically weak personality measures

394

PERSONALITY AND SUBSTANCE USE DISORDERS 395

(Sutker & Allain, 1988). Sensation seeking is the most widely studied personality correlate of drug use disorders, with the pre- dictable finding that indicators of this personality characteristic have been consistently associated with drug use disorder status (Kosten, Ball, & Rounsaville, 1994; Luthar, Anton, Merikangas, & Rounsaville, 1992; Sutker, Archer, & Allain, 1978; Vukov, Baba- Milkic, Lecic, Mijalkovic, & Marinkovic, 1995). The recent study by Krueger, Caspi, Moffitt, Silva, and McGee (1996) is the only

published investigation of the personality correlates of substance use disorders that used both a comprehensive assessment of per- sonality (the Multidimensional Personality Questionnaire [MPQ]),

and a community-based ascertainment scheme (a representative birth cohort from New Zealand). These investigators reported that a composite diagnosis of substance dependence was associated with elevated levels of negative emotionality and depressed levels

of constraint relative to non-substance-dependent controls. Unfor- tunately, Krueger et al. did not break down their findings by either gender or type of substance dependence, making it difficult to evaluate the generality of the correlations they reported. Although there have been few studies of the personality correlates of sub- stance use disorders, the personality correlates of drug use have been extensively investigated, especially in adolescent popula- tions. These studies have consistently reported that adolescent drug users score higher than adolescent non-drug users on indicators of behavioral disinhibition (e.g., Labouvie & McGee, 1986; Wills, Vaccaro, & McNamara, 1994).

The present study is the second in a series exploring the role of personality in the etiology of substance use disorders. In the first study (McGue et al., 1997), we investigated the association of alcoholism with personality and determined whether that associa- tion was moderated by gender or alcoholism subtype. In this second study, we contrasted the personality correlates of alcohol- ism with the personality correlates of drug use disorders. Although the use of a cross-sectional design did not allow us to determine whether the personality correlates we observed were causes or consequences of a substance use disorder, the present study did address several of the major limitations of research relating per- sonality and addictions (Sutker & Allain, 1988). In particular, the present study was based on a large sample of individuals with (n = 638) and without (n — 1,530) a substance use disorder. All individuals were ascertained using the same community-based procedure, and all individuals completed a comprehensive self- report personality inventory that is based on a theoretically mean- ingful three-factor model.

Method

Sample

The sample was composed of parent participants in the Minnesota Twin Family Study (MTFS). The MTFS is a prospective study of the develop- ment of substance use and related disorders in a large sample of adolescent twins and their parents (lacono, Lykken, & McGue, 1996; McGue, Lykken, & lacono, 1996). Twin families were ascertained from Minnesota birth records and located using various public databases, including telephone directories and driver's license registrations (Lykken, Bouchard, McGue, & Tellegen, 1990). For any given birth year, we were able to locate more than 90% of twin births. Families were excluded from participation if they lived farther than a day's drive from our laboratories in Minneapolis, Minnesota, or if either of the twins in the family had a physical or cognitive

handicap that would preclude their completing our day-long, in-person intake assessment. Among eligible families, 17% refused participation, with 78% of the nonrefusing families ultimately completing the intake assessment. Through telephone interviews, mail surveys, and access to information recorded on the birth records, we were able to determine the socioeconomic status (SES) and complete brief mental health assessments on 69% of the nonparticipating families. Analysis of these data indicated that participating families differed minimally from nonparticipating fami- lies, with the largest differences occurring on measures of SES in which parents in participating families averaged 0.6 years more education than parents in nonparticipating families. Consistent with Minnesota demo- graphics, over 98% of participating parents were Caucasian.

Substance use diagnoses were made from the clinical interviews of 2,698 parents (1,384 mothers, 1,314 fathers) and were based on Diagnostic and Statistical Manual of Mental Disorders (3rd ed., rev.; DSM-III-R; Amer- ican Psychiatric Association, 1987) lifetime criteria at two levels of cer- tainty: definite (met all criteria for dependence or abuse) and probable (all but one Criterion A dependence symptom were satisfied). Diagnoses were used to form four groups in a fully crossed 2 X 2 design (Alcohol Use Disorder x Drug Use Disorder). The first grouping factor was based on DSM-III-R diagnoses of alcohol dependence and distinguished individuals with a definite diagnosis of alcohol dependence (designated "alcohol yes") from those who did not meet criteria for alcohol dependence at either the definite or probable levels of certainty, (designated "alcohol no"). The second factor was based on DSM-III-R diagnoses of psychoactive sub- stance abuse or dependence for the following eight substances: amphet- amines, cannabis, cocaine, hallucinogens, inhalants, opiates, PCP-related substances, and sedatives. Individuals with a definite diagnosis of sub- stance abuse or dependence for at least one of the eight substances composed one group (designated "drug yes"), whereas those not diagnosed with a substance use disorder at either the definite (for dependence or abuse) or probable (for dependence) level for any of the substances formed the second group (designated "drug no"). Note that nicotine was not used as one of eight substances to determine group status and that individuals with a probable diagnosis of alcohol dependence, as well as individuals with no more than a probable dependence diagnosis for the eight sub- stances, were not classified into one of the four groups (n = 237, or 9%).

Measures

Demographics. Age, educational attainment in years, current marital status, and occupation were all assessed by self-report. On the basis of the occupation reports, we rated occupational status using the Hollingshead system (Hollingshead & Redlich, 1958), which ranges from 1 (profes- sional) to 7 (unskilled labor). Because full-time homemakers are not scaled in the Hollingshead system, we considered homemakers as "missing" in the analysis on occupational status presented here. Occupational status codes were not available for 123 participants (11%) in the female sample.

Clinical assessment. Participants were assessed in person by trained lay interviewers. Substance use disorders were assessed using an expanded version of the Substance Abuse Module (Robins, Babor, & Cottier, 1987), which was developed as a supplement to the World Health Organization's Composite International Diagnostic Interview (Robins et al., 1988). Major depression was assessed using the Structured Clinical Interview for DSM- III-R (Spitzer, Williams, & Gibbon, 1987), whereas conduct disorder and antisocial personality disorder (ASPD) were assessed using a clinical interview developed by MTFS staff. Prior to the assignment of individual diagnoses, a clinical case conference was completed in which the evidence for every symptom was reviewed (including, if necessary, the replaying of audiotapes from the interviews) by at least two advanced graduate students in clinical psychology. After consensus was reached, life-time diagnoses using DSM-III-R criteria were assigned by computer algorithm.

Because the DSM-III-R criteria are meant to be used to diagnose acute disorders, the lifetime substance use diagnoses were made at two levels of certainty to reduce the likelihood that affected individuals were classified

396 McGUE, SLUTSKE, AND IACONO

as "unaffected." A definite diagnosis was made if all necessary DSM-III-K criteria were satisfied; a probable diagnosis was made if all but one of the criteria were met. Because DSM-IIl-R substance abuse is defined by the presence of a single symptom, this approach was used for dependence only. As described previously, individuals with probable diagnoses of substance dependence were excluded from the sample. Findings for all of the relevant disorders presented here are based on lifetime definite diagnoses only. Because our diagnoses are lifetime rather than current, we provide descrip- tive data on two categorical indicators of the recency of substance use problems. (Our intake assessment procedures did not allow for the diag- nosis of current alcohol dependence and substance abuse and dependence.) Alcohol recency distinguished individuals who had at least one of the symptoms used to diagnose alcohol dependence in the past year from those who did not; drug recency distinguished individuals who had used any of the eight illicit substances in the past year from those who had not. Participants were also asked whether they had ever been treated for alcoholism.

Personality. Personality was assessed by self-report using the 198- item version of the MPQ (Tellegen, 1982; Tellegen & Waller, in press). The MPQ was developed using factor analysis and consists of 11 psycho- metrically sound primary scales and three higher order factor scales, the latter being linear-weighted composites of the primary scales. The MPQ is organized around three broad dimensions of personality: positive emotion- ality, or the tendency to be actively and pleasurably engaged with one's work and social environments; negative emotionality, or the tendency to experience psychological distress and negative mood states; and constraint, or the tendency to be cautious, to inhibit behavioral impulses, and to endorse conventional moral values (i.e., the complement of behavioral disinhibition). Four of the MPQ primary scales (Well-Being, Social Po- tency, Achievement, and Social Closeness) load principally on positive emotionality, three (Stress Reaction, Alienation, and Aggression) load principally on negative emotionality, and three (Control, Harm Avoidance, and Traditionalism) load principally on constraint. The eleventh primary scale, Absorption, does not load principally on any of the three higher order factors.

Prior to their scheduled in-person assessment, parents were mailed the MPQ, along with several other self-report instruments, which they were asked to complete and either return by mail or bring with them to the assessment. If we failed to obtain a completed MPQ from a parent by the time of the in-person assessment, he or she was asked to complete the form at home and return it to us by mail. One additional telephone reminder was made if we still did not receive a completed MPQ. Of the 2,461 parents classified into one of the four substance use diagnosis groups, 236 (10%) did not return a completed MPQ and 57 (2%) returned a form but had either skipped too many items or did not pass the MPQ validity scale screens. We were more likely to receive a completed MPQ from women than from men (91% vs. 85%) and from those not having an alcohol or drug use diagnosis than from those having these diagnoses (90% vs. 83%).

After deleting those with missing or nonvalid MPQs, we found that the sample sizes for the four female groups were 1,000 (84% of the sample) for the alcohol no/drug no group, 66 (6%) for the alcohol yes/drug no group, 85 (7%) for the alcohol no/drug yes group, and 40 (3%) for the alcohol yes/drug yes group. The comparable figures for the four male groups were 530 (54%), 246 (25%), 68 (7%), and 133 (14%), respectively. The proportionate representation of gender in the four groups reflects the well-documented gender difference in prevalence for substance use disor- ders (e.g., Anthony & Helzer, 1991; Helzer, Bucholz, & Robins, 1992; Warner, Kessler, Hughes, Anthony, & Nelson, 1995), whereas the non- independence of the two classification variables reflects the substantial comorbidity that exists between alcoholism and other substance use dis- orders (Helzer & Pryzbeck, 1988).

Statistical Procedures

The major study questions were addressed with a three-factor multivar- iate analysis of variance (MANOVA). The three factors were gender (male vs. female), alcohol use disorder (no vs. yes), and drug use disorder (no vs. yes). Significant MANOVA effects were followed up with univariate analyses of variance (ANOVAs). Two separate MANOVAs were com- puted. The first had as dependent variables the 11 primary scales, whereas the second, because the higher order scales are linear composites of the primary scales, had as dependent variables the three higher order scales. In all analyses, we evaluated effects using a regression approach in order to control for the correlation between the two independent variables. That is, the alcohol effect was evaluated net of the drug effect, and, conversely, the drug effect was evaluated net of the alcohol effect.

Results

Demographic and Clinical Comparisons

Demographic and clinical comparisons are given in Table 1 for the four female groups and in Table 2 for the four male groups. We assessed group differences separately for men and women on the demographic and clinical indicators using either a one-factor (group membership) ANOVA for quantitative measures or a log- linear analysis for categorical variables. Because the drug use disorder classification was based on the substance use diagnoses, we compared rates of substance use disorders only for the alcohol no/drug yes and the alcohol yes/drug yes groups.

Among women, those with a substance use disorder were on average somewhat younger and less likely to be married than those not having a substance use diagnosis, and alcoholic women had on average about 1 year less of education than did nonalcoholic women. Among men, those with a substance use disorder were less likely to be married than those not having a substance use disorder, and alcoholic men had on average 1 year less of education and lower occupational status than did nonalcoholic men. As has been reported in other community-ascertained epidemiological samples (Helzer & Pryzbeck, 1988), the majority of alcoholic participants in both the male and female samples had not been treated for alcoholism. In the year previous to their assessment, a little less than 50% of the alcoholic men and women were positive for at least one alcohol-dependence symptom and approximately 20% to 25% of men and women with a drug diagnosis had used a sub- stance illicitly.

As expected, rates of conduct disorder, ASPD, and nicotine dependence were consistently higher among those having a sub- stance use disorder than among those not having a substance use disorder. Among men with a drug use disorder, alcoholism was significantly associated with higher rates of multiple-substance abuse or dependence. Nonalcoholic men with a drug use disorder met criteria for either abuse or dependence for an average of 1.3 (SD = 0.7) of the eight substances, with only 24% of these individuals being multiple-substance abusers (i.e., having a posi- tive diagnosis for more than one of the eight substances). In contrast, alcoholic men with a drug use disorder met criteria for abuse or dependence for an average of 2.1 (SD = 1.4) of the eight substances, and 54% of these individuals had a positive diagnosis for more than one of the substances. Among women with a drug use disorder, however, alcoholism was not associated with multiple-substance abuse; nonalcoholic women with a drug use disorder met criteria for abuse or dependence for an average of 1.4

PERSONALITY AND SUBSTANCE USE DISORDERS 397

Table 1 Demographic and Clinical Characteristics of the Female Substance Use Disorder Samples

Substance use disorder group

Alcohol no/ Characteristic drug no

Sample size Total sample size Personality sample sizeb

% Demographics

Age (years) M SD

Education (years) M SD

Occupation M SD

Married (%) Substance use diagnoses, treatment, and recency

Amphetamines (%) Cannabis (%) Cocaine (%) Hallucinogens (%) Inhalants (%) Opiates (%) PCP-related substances (%) Sedatives (%) No. of drug diagnoses

M SD

Alcohol treatment (%) Alcohol recency (%) Substance recency (%)

Other clinical diagnoses (%) Conduct disorder ASPD Major depression Nicotine dependence

1,092 1,000

42.1 5.1

13.9 1.9

3.7 1.6

0 0 0 0 0 0 0 0

0.0 0.0 0 8 1

1 0

20 24

Alcohol yes/ drug no

76 66 87

40.3 5.3

13.0 1.7

4.0 1.6

0 0 0 0 0 0 0 0

0.0 0.0

17 36 3

9 3

33 73

Alcohol no/ drug yes

94 85 90

39.5 4.6

14.1 2.1

3.3 1.3

34 73 11 11 2 4 1 5

1.4 0.9 2

14 17

5 0

35 48

Alcohol yes/ drug yes p'

50 40 80

37.8 4.1

13.2 1.8

3.9 1.7

33 83 20 13 0 8 3

1.7 1.1

20 55 20

23 3

45 70

.03

<.001

.003

<.001

ns ns ns ns ns ns ns ns

•C.001 <.001 <.001

<.001 <.001 <.001 <.001

Note. AH diagnoses are based on lifetime Diagnostic and Statistical Manual of Mental Disorders (3rd ed., rev.; DSM-III-R; American Psychiatric Association, 1987) criteria at a definite level of certainty. An individual was considered positive for each of the eight DSM-III—R substance use disorders if that individual met DSM—II1-R criteria for either abuse or dependence for that substance. An individual was considered to have an alcohol use disorder if he or she met DSM-III-R criteria for alcohol dependence (i.e., "alcohol yes"), and an individual was considered to have a drug use disorder (i.e., "drug yes") if he or she was positive for at least one of the eight substance use disorders. ASPD = antisocial personality disorder. 'p values are for the appropriate statistic for testing equality of group means or percentages. For the eight substance use diagnoses used in forming the drag use disorder grouping, the p value corresponds to the two-group comparison; in all other cases, it corresponds to the four-group comparison, ns = not significant at p < .05. b The personality sample size gives the size of the sample participating in this study. The remaining descriptive data in this table are based on this sample.

(SD = 0.9) of the substances, whereas the corresponding mean for alcoholic women with a drug use disorder was 1.7 (SD = 1.1).

The Association of Personality With Alcohol and Drug Use Disorders

In the Gender X Alcohol X Drug MANOVA of the 11 MPQ primary scales, the gender main effect was statistically significant, F(ll,2150) = 52.1, p< .001, although none of the interaction terms involving gender were significant atp < .05. Both the alcohol, F(ll, 2150) = 4.60, p < .001, and drag, F(ll, 2150) = 9.01, p < .001,

main effects were statistically significant, as was the Alcohol X Drug interaction, F(l 1,2150) = 2.16,p < .05. The MANOVA for the three higher order MPQ scales paralleled the results obtained with the primary scales: The gender effect was significant, F(3, 2158) = 17.6, p < .001, but none of the interaction terms involving gender were significant. Both the alcohol, F(3, 2158) = 8.52, p < .001, and drug, F(3, 2158) = 28.7, p < .001, main effects, as well as the Alcohol X Drug interaction, F(3, 1903) = 3.80, p < .05, were statistically significant. Follow-up ANOVAs for the significant MANOVA ef-

fects are summarized in Table 3.

398 McGUE, SLUTSKE, AND IACONO

Table 2 Demographic and Clinical Characteristics of the Male Substance Use Disorder Samples

Substance use

Alcohol no/ Characteristic drug no

Sample size Total sample size Personality sample sizeb

% Demographics

Age (years) M SD

Education (years) M SD

Occupation M SD

Married (%) Substance use diagnoses, treatment, and recency

Amphetamines (%) Cannabis (%) Cocaine (%) Hallucinogens (%) Inhalants (%) Opiates (%) PCP-related substances (%) Sedatives (%) No. of drug diagnoses

M S£>

Alcohol treatment (%) Alcohol recency (%) Substance recency (%)

Other clinical diagnoses (%) Conduct disorder ASPD Major depression Nicotine dependence

604 530

44.9 5.7

14.6 2.4

3.4 1.8

0 0 0 0 0 0 0 0

0.0 0.0 1

11 1

8 0 9

Alcohol yes/ drug no

287 246 86

44.6 5.9

13.7 2.2

4.1 1.7

0 0 0 0 0 0 0 0

0.0 0.0

18 46

15 3

15 55

disorder group

Alcohol no/ drug yes

90 68 76

41.1 3.5

14.6 2.3

3.1 1.7

16 94

2 15 0 2 2 3

1.3 0.7 2

28 24

19 7

19 49

Alcohol yes/ drug yes p'

168 133 79

40.7 4.7

13.6 1.9

4.1 1.8

49 91 20 20 2

10 2

2.1 1.4

38 50 26

17 23 17 67

.002

<.001

<.001 ns

<.001 ns ns .03 ns .04

<.001

<.001 <.001 <-001

.005 <.001

Note. All diagnoses are based on lifetime Diagnostic and Statistical Manual of Mental Disorders (3rd ed., rev.; DSM—II1-R; American Psychiatric Association, 1987) criteria at a definite level of certainty. An individual was considered positive for each of the eight DSM-IIl-R substance use disorders if that individual met DSM-HI-R criteria for either abuse or dependence for that substance. An individual was considered to have an alcohol use disorder if he or she met DSM—IIl—R criteria for alcohol dependence (i.e., "alcohol yes"), and an individual was considered to have a drug use disorder (i.e., "drug yes") if he or she was positive for at least one of the eight substance use disorders. ASPD = antisocial personality disorder. a p values are for the appropriate statistic for testing equality of group means or percentages. For the eight substance use diagnoses used in forming the drug use disorder grouping, the p value corresponds to the two-group comparison; in all other cases, it corresponds to the four-group comparison, ns = not significant at p < .05. b The personality sample size gives the size of the sample participating in this study. The remaining descriptive data in this table are based on this sample.

MPQ scale means and standard deviations are given in Table 4 for the four female groups and in Table 5 for the four male groups. Because of the large size of our sample, modest effect sizes (ESs) could be statistically significant in the present study. Therefore, to facilitate interpretation and estimation of ES, we transformed scores separately for men and women so that each scale had a mean of 50.0 and a standard deviation of 10.0 in the alcohol no/drug no group (i.e., a T-score metric). Although this transfor- mation has the effect of removing the gender main effect, it allows us to focus on the ESs that are of principalinterest here (i.e., those associated with substance use disorders). Moreover, the transfor-

mation serves to highlight the replicability of findings across the male and female samples.

The pattern of significant ANOVA results summarized in Ta- ble 3 tends to be organized around the hierarchical structure of the MPQ. Alcoholism is primarily associated with elevated levels of negative emotionality. A significant alcohol effect was observed for the Negative Emotionality higher order scale, as well as for each of the three primary scales that load principally on this factor (i.e., Stress Reaction, Alienation, and Aggression). Despite this consistency of statistical results, the ESs associated with the alco- hol effect were only moderate in magnitude. For example, alco-

PERSONALITY AND SUBSTANCE USE DISORDERS 399

Table 3 Results of Follow-Up ANOVA Tests of Significant MANOVA Effects

MPQ scale

Primary scale Weil-Being Social Potency Achievement Social Closeness Stress Reaction Alienation Aggression Control Harm Avoidance Traditionalism Absorption"

Higher order scale PE NE CN

Principal factor loading

PE PE PE PE NE NE NE CN CN CN

— —

ANOVA effect p value

Gender

ns M > F*** M > F * p > M*** F > M***

ns M > F***

ns F > M***

ns p>M***

ns ns

F > M***

Alcohol (A)

ns ns ns ns

Yes > No*** Yes > No*** Yes > No*** No > Yes***

ns ns ns

ns Yes > No***

Drug (D)

ns Yes > No** No > Yes**

ns ns

Yes > No* Yes > No* No > Yes*** No > Yes*** No > Yes***

ns ns

No > Yes***

A X D

<.01 ns ns

<.01 <.05

ns ns ns ns

<.01 ns

ns ns

<.01

Note. In the overall MANOVA, the two-way interactions between gender and alcohol and between gender and drug, as well as the three-way interaction, were not statistically significant; therefore, follow-up ANOVAs for these effects are not reported. Principal factor loading gives the higher order factors on which each of the primary MPQ scales principally loads. Significance for the three main effects are summarized by indicating which group had the larger mean. "Yes" indicates that the participant has a relevant substance use disorder; "No" indicates that the participant does not have a relevant substance use disorder. ANOVA = analysis of variance; MANOVA = multivariate analysis of variance; MPQ = Multidimensional Personality Questionnaire; PE = Positive Emotionality; NE = Negative Emotionality; CN = Constraint; M = male participants; F = female participants. a Does not load principally on any one of the three higher order factors. *p<.05. * * / > < . 0 1 . * * * / > < . 001.

holic participants scored from 0.3 to 0.5 SDs higher than nonal- coholic participants on the Negative Emotionality higher order scale. A significant alcohol effect was also observed for the Con- trol scale, with the mean for alcoholic participants moderately lower than the mean for nonalcoholic participants (standardized ES of approximately 0.3 SDs).

The significant drug main effect was associated primarily with the MPQ higher order dimension of constraint; a signifi- cant drug main effect was observed for the Constraint scale, as well as for each of the three primary scales that load principally on this variable (Control, Harm Avoidance, and Traditional- ism). The ESs associated with these scales tended to be mod- erate to large in magnitude. For example, as compared with those without a drug use disorder, those with a drug use disorder averaged about 0.7 to 0.8 SDs lower on the Constraint higher order scale. A significant drug main effect was observed also for the Social Potency, Aggression, and Alienation scales; the mean for those with a drug use disorder was somewhat higher on the first two scales and somewhat lower on the last scale than the mean for those not having a drug use disorder (ES of less than 0.3 SDs in all cases).

Our evaluation of the significant alcohol and drug main effects needs to be conditioned by the existence of a significant Alcohol X Drug interaction, both in the overall MANOVA and in several of the follow-up ANOVAs. The significant ANOVA interaction for the Social Closeness, Traditionalism, and Constraint scales were all disordinal and had a similar pattern: The mean difference between nonalcoholic and alcoholic participants with a drug use

disorder was opposite in sign from the difference between nonal- coholic and alcoholic participants with a drug use disorder. For example, on the Constraint scale, alcoholic participants without a drug use disorder scored lower than their nonalcoholic counter- parts; however, alcoholic participants with a drug use disorder scored higher than their nonalcoholic counterparts. That is, the most "deviant" group on the Constraint scale was not the comorbid alcohol yes/drug yes group but rather the alcohol no/drug yes group.

Because the substance use diagnoses were based on lifetime criteria, our substance use disorder groups include a mixture of current and remitted alcoholic and drug-abusing individuals. We investigated whether the recency of individuals' alcohol or drug problems moderated the effects reported here by determin- ing whether the MPQ scales were associated with the recency of alcoholism symptom expression and illicit drug use. In a two- factor (Gender X Alcoholism Symptom Recency) MANOVA using alcoholics only, the recency main effect and the Re- cency X Gender interaction effect were not significant in the analysis of both the 11 primary, F(ll, 470) = 1.13, p = .33, and 3 higher order, F(3, 478) = 0.35, p = .79, MPQ scales. In a similar analysis of drug use recency among those with a drug abuse or dependence diagnosis, there were again no significant interaction effects, and the drug use recency effect was not significant in the analysis of the 3 higher order MPQ scales, F(3, 320) = 2.09, p = .10. The drug use recency main effect was, however, significant in the analysis of the 11 primary MPQ scales, F(ll, 312) = 2.42, p < .01. Follow-up ANOVAs

400 McGUE, SLUTSKE, AND IACONO

Table 4 MPQ Scale Means and Standard Deviations for the Four Female Substance Use Disorder Groups

MPQ scale

Primary scale Well-Being

M SD

Social Potency M SD

Achievement M SD

Social Closeness M SD

Stress Reaction M SD

Alienation M SD

Aggression M SD

Control M SD

Harm Avoidance M SD

Traditionalism M SD

Absorption M SD

Higher order scale Positive Emotionality

M SD

Negative Emotionality M SD

Constraint M SD

Discriminant functions First discriminant

M SD

Second discriminant M SD

Alcohol no/ drug no

(n = 1,000)

50.0 10.0

Substance use

Alcohol yes/ drug no (n = 66)

45.6 10.6

50.2 9.9

52.1 10.9

46.1 10.0

54.9 11.3

54.6 11.5

52.5 11.6

46.2 11.0

50.7 8.5

48.6 9.3

52.5 10.7

48.8 8.7

55.8 11.0

48.9 10.3

46.3 10.7

54.1 11.5

disorder group

Alcohol no/ drug yes (n = 85)

49.7 9.8

53.2 11.0

48.6 10.0

49.0 10.2

50.0 10.1

47.6 9.4

50.2 9.8

47.1 10.8

46.5 10.2

42.2 9.7

52.7 10.1

50.0 9.4

48.8 8.4

42.7 11.2

43.0 10.7

44.1 9.8

Alcohol yes/ drug yes (« = 40)

50.1 10.3

52.1 10.6

48.6 11.3

50.9 9.6

51.5 10.5

51.6 9.4

51.5 12.3

44.4 9.8

46.6 11.6

46.4 8.5

52.3 10.4

50.5 10.6

51.4 9.4

44.1 9.7

44.2 10.5

49.9 8.8

Note. Scale scores were linearly transformed to have a mean of 50.0 and a standard deviation of 10.0 in the alcohol no/drug no group. MPQ = Multidimensional Personality Questionnaire.

indicated that the overall significant effect could be attributed to recent drug users scoring higher on the Absorption scale (about 0.5 SDs) and somewhat lower on the Stress Reaction and Traditionalism scales (about 0.3 SDs) than those who had not

used illicit substances in the past year. Of the 3 scales that did show a significant drug use recency effect, only the Tradition- alism scale also significantly differentiated those with a diag- nosis of drug abuse or dependence from those not having a

PERSONALITY AND SUBSTANCE USE DISORDERS 401

Table 5 MPQ Scale Means and Standard Deviations for the Four Male Substance Use Disorder Groups

MPQ scale

Primary scale Weil-Being

M SD

Social Potency M SD

Achievement M SD

Social Closeness M SD

Stress Reaction M SD

Alienation M SD

Aggression M SD

Control M SD

Harm Avoidance M SD

Traditionalism M SD

Absorption M SD

Higher order scale Positive Emotionality

M SD

Negative Emotionality M SD

Constraint M SD

Discriminant functions First discriminant

M SD

Second discriminant M SD

Alcohol no/ drug no

(n = 530)

50.0 10.0

Substance use

Alcohol yes/ drug no

(n = 246)

47.9 9.9

48.8 9.9

48.7 11.4

48.2 10.2

53.7 10.2

53.2 10.8

54.3 11.2

45.6 9.6

48.7 10.1

48.9 9.2

50.3 10.5

47.9 9.8

53.8 10.8

46.9 9.2

45.7 9.6

54.0 9.6

disorder group

Alcohol no/ drug yes (n = 68)

48.0 8.4

52.0 9.2

47.2 11.4

46.7 9.1

50.7 9.9

49.8 8.8

54.4 11.4

45.5 10.1

43.7 8.0

43.4 9.5

51.7 11.5

48.3 8.5

51.0 9.3

41.0 9.3

41.1 9.0

47.0 8.3

Alcohol yes/ drug yes

(n = 133)

48.7 9.8

50.1 8.7

49.4 11.0

47.1 10.6

52.1 8.9

52.5 10.1

55.8 10.0

44.1 9.0

45.4 9.6

44.2 10.4

51.6 12.7

48.7 9.9

53.0 9.5

42.3 9.7

41.3 9.8

51.4 9.5

Note. Scale scores were linearly transformed to have a mean of 50.0 and a standard deviation of 10.0 in the alcohol no/drug no group. MPQ = Multidimensional Personality Questionnaire.

diagnosis (see Table 3). Thus, our analysis of recency effects suggests that the significant alcohol and drug effects we ob- served in the overall sample were not an artifact of our use of lifetime diagnoses. That is, the personality differences we have reported appear to characterize individuals with both remitted and continuing substance use problems.

Discriminant Function Analysis

To provide a more rigorous test of the proposition that alcohol- ism is primarily associated with negative emotionality, whereas drug use disorders are primarily associated with constraint, we performed a linear discriminant function analysis with the four

402 McGUE, SLUTSKE, AND IACONO

groups as the independent variable and the 11 MPQ scales as the dependent variables. Because our MANOVA analyses provided no evidence that the grouping variable interacted with gender, we pooled the male and female samples in this analysis after we had adjusted the personality scores for the main effect of gender. Of the three possible linear discriminant functions, the first, ;̂ (33, N = 2,168) = 277.5, p < .001, and the second, ̂ (20, N = 2,168) = 92.1, p < .001, but not the third, ^(9, N = 2,168) = 9.71, p = .37, were statistically significant. The first discriminant function appeared to predominantly reflect constraint, as it correlated most highly with the Constraint higher order scale (r = .90) and with the Traditionalism (r = .71), Control (r = .68), Harm Avoidance (r = .51), and Aggression (r — -.51) primary scales. The first discrimi- nant function also appeared to primarily differentiate those with and without a drug use disorder (means and standard deviations are reported in Table 4 for women and in Table 5 for men). The second discriminant function was most strongly aligned with negative emotionality, as it correlated most highly with the Negative Emo- tionality higher order scale (r = .63) and the Alienation (r = .64), Stress Reaction (r = .54), Traditionalism (r = .42), and Aggres- sion (r = .41) primary scales. Group differences on the second discriminant function appeared to primarily differentiate alcoholic individuals from nonalcoholic individuals (see Tables 4 and 5).

Discussion

We found that if drug use disorders are statistically controlled, personality differences between alcoholic and nonalcoholic indi- viduals arise primarily on measures of negative emotionality, whereas if one controls for alcoholism, personality differences between individuals with and without a drug use disorder arise primarily on measures of constraint. This pattern of results was observed in both the male and female samples and was suggested in the MANOVA analyses and confirmed in the follow-up linear discriminant analyses.

Our finding that alcoholism appears to be specifically associated with measures of negative emotionality may seem to run counter to the consistent observation that alcoholism is associated with both negative emotionality and behavioral disinhibition (McGue et al., 1997; Sher & Trull, 1994). However, previous research investi- gating the personality correlates of alcoholism has failed to control for comorbid drug use disorders. The present findings suggest that this may be a consequential omission. In particular, the relatively high level of behavioral disinhibition that has been associated with alcoholism may characterize only that subset of alcoholic individ- uals who also abuse drugs other than alcohol. In the present study, non-drug-abusing alcoholic individuals differed significantly from non-drug-abusing nonalcoholic individuals by scoring higher on measures of negative emotionality, not by scoring lower on mea- sures of constraint.

Our finding that drug use disorders are specifically associated with measures of constraint is consistent with other research indi- cating that the correlation of illicit drug use with measures of negative emotionality is substantially decreased once alcohol and cigarette consumption are controlled for statistically (Sieber, 1981). That constraint is the principal personality correlate of drug abuse is not altogether surprising, given the nature of the under- lying personality construct. Individuals who are low on constraint are characterized by the unwillingness or inability to inhibit be-

havioral impulses, a lack of caution, and a failure to endorse conventional moral standards. Drug use and abuse are strongly discouraged by conventional cultural standards and are considered risky activities because of their illicit nature. Individuals who are low on constraint may be at increased risk for drug abuse because they are less likely to accept—and less fearful of the consequences of failing to follow—cultural norms governing drug use.

Interestingly, the most deviant individuals on the constraint dimension were not those with a drug use disorder comorbid with alcoholism but rather those with a drug use disorder who were nonalcoholic. Given that alcohol use is usually a gateway to illicit drug use, this observation perhaps reflects that being a drug- abusing nonalcoholic individual is socially more deviant than being a drug-abusing alcoholic individual. This interpretation is not, however, supported by the clinical and demographic data reported in Tables 1 and 2, in which the alcoholic drug abusers are clearly the most deviant group. It is also interesting to note that the alcohol no/drug yes group had the highest mean level of education and highest mean occupational status of the four groups we stud- ied. Perhaps personality, specifically constraint, is a more signif- icant risk factor for drug use disorders among those with high rather than low levels of education. Although this is an interesting possibility, the modest ES associated with the significant Alco- hol X Drug interaction suggests that caution is needed until this interaction is replicated.

Our findings provide partial support for current typological models of alcoholism. In particular, both Cloninger (1987a) and Babor et al. (1992) have hypothesized the existence of two types of alcoholic individuals. The first is characterized by a relatively late age of alcoholism onset, a low familial loading, and absence of other antisociality, whereas the second is characterized by a rela- tively early age of alcoholism onset, a high familial loading, and the existence of other antisocial behavior. Of particular relevance to the present study is the hypothesis that personality factors are differentially associated with the two forms of alcoholism: The first type of alcoholism is thought to be specifically associated with relatively high levels of negative emotionality, whereas the second type is hypothesized to be associated with relatively high levels of behavioral disinhibition. Our findings are consistent with the proposition that constraint and negative emotionality mark distinct pathways to alcoholism but are inconsistent with the proposition that these two pathways are primarily distinguished clinically by different levels of general antisocial behavior. Rather, our data suggest that comorbid drug abuse (admittedly, a specific instance of antisocial behavior) may be the critical clinical factor differentiating the two types of alcoholism, a possibility certainly deserving further exploration.

Although the present study was designed to address several of the major methodological shortcomings of previous research in- vestigating the relationship between personality and substance use disorders, it is not without its own shortcomings, which deserve mention here. First, although nearly 90% of all MTFS parents completed the MPQ, response rates were lowest among alcoholic individuals and among those having a drug use disorder. Our results may have been different had we been able to achieve a higher response rate with these latter two groups. Second, although the MTFS community-based ascertainment procedure is likely to produce a more representative sample than the treatment-based sampling methods that have been used in most of the research on

PERSONALITY AND SUBSTANCE USE DISORDERS 403

personality and drug use disorders, it is important to recognize that the present sample may not accurately represent the general- population distribution of alcoholic individuals and those with a drug use disorder. In particular, all participants in the present sample were parents. As becoming a parent likely involves some selection for mental health (e.g., Slater, Hare, & Price, 1971), the present findings may underestimate the association of personality with substance use disorders. Third, the present study was cross- sectional rather than longitudinal. Consequently, we were unable to determine whether the personality differences we observed predated the onset—or developed as a consequence—of a sub- stance use disorder. Moreover, it should be emphasized that the significant ESs we observed tended to be moderate in magnitude, indicating that at best personality is one of a multitude of factors influencing the risk of having a substance use disorder. Finally, although the MPQ is an omnibus personality inventory, there may be personality characteristics that are associated with substance use disorders but that are not assessed by the MPQ.

References

American Psychiatric Association. (1987). Diagnostic and statistical man- ual of mental disorders (3rd ed., rev.). Washington, DC: Author.

Anthony, J. C, & Helzer, J. E. (1991). Syndromes of drug abuse and dependence. In L. N. Robins & D. A. Regier (Eds.), Psychiatric disor- ders in America: The Epidemiologic Catchment Area Study (pp. 116- 154). New York: Free Press.

Babor, T. R, Hofraann, M., DelBoca, R K., Hesselbrock, V., Meyer, R. E., Dolinsky, Z. S., & Rounsaville, B. (1992). Types of alcoholics: I. Evidence for an empirically derived typology based on indicators of vulnerability and severity. Archives of General Psychiatry, 49, 599- 608.

Caspi, A., Moffitt, T. E., Newman, D. L., & Silva, P. A. (1996). Behavioral observations at age 3 predict adult psychiatric disorders: Longitudinal evidence from a birth cohort. Archives of General Psychiatry, 53, 1033-1039.

Cloninger, C. R. (1987a). Neurogenetic adaptive mechanisms in alcohol- ism. Science, 236, 410-416.

Cloninger, C. R. (1987b). A systematic method for clinical description and classification of personality variants. Archives of General Psychia- try, 44, 573-588.

Cloninger, C. R., Sigvardsson, S., Reich, T., & Bohman, M. (1988). Childhood personality predicts alcohol abuse in young adults. Alcohol- ism: Clinical and Experimental Research, 12, 494-505.

Costa, P. T., Jr., & McCrae, R. R. (1992). Four ways five factors are basic. Personality and Individual Differences, 13, 653-665.

Eysenck, H. J. (1967). The biological basis of personality. Springfield, IL: Charles C Thomas.

Helzer, J., & Pryzbeck, T. (1988). The co-occurrence of alcoholism and other psychiatric disorders in the general population and its impact on treatment. Journal of Studies on Alcohol, 49, 219-224.

Helzer, J. E., Bucholz, K., & Robins, L. N. (1992). Five communities in the United States: Results from the Epidemiologic Catchment Area Survey. In J. E. Helzer & G. J. Canino (Eds.), Alcoholism in North America, Europe, and Asia (pp. 71-95). New York: Oxford University Press.

Hollingshead, A., & Redlich, F. C. (1958). Social class and mental disease. New York: Wiley.

lacono, W. G., Lykken, D. T., & McGue, M. (1996). Psychophysiological prediction of substance abuse. In H. W. Gordon & M. D. Glantz (Eds.), Individual differences in the biobehavioral etiology of drug abuse (pp. 129-149). Rockville, MD: National Institute on Drug Abuse.

Kammeier, M. L., Hoffman, H., & Loper, R. G. (1973). Personality

characteristics of alcoholics as college freshman and at time of treat- ment. Quarterly Journal of Studies on Alcohol, 34, 390-399.

Kosten, T. A., Ball, S. A., & Rounsaville, B. J. (1994). A sibling study of sensation seeking and opiate addiction. The Journal of Nervous and Mental Disease, 182, 284-289.

Krueger, R. F., Caspi, A., Moffitt, T. E., Silva, P. A., & McGee, R. (1996). Personality traits are differentially linked to mental disorders: A multitrait-multidiagnosis study of an adolescent birth cohort. Journal of Abnormal Psychology, 105, 299-312.

Labouvie, E. W., & McGee, C. R. (1986). Relation of personality to alcohol and drug use in adolescence. Journal of Consulting and Clinical Psychology, 54, 289-293.

Luthar, S. S., Anton, S. R, Merikangas, K. R., & Rounsaville, B. J. (1992). Vulnerability to drug abuse among opioid addicts' siblings: Individual, familial, and peer influences. Comprehensive Psychiatry, 33, 1-8.

Lykken, D. T., Bouchard, T. J., McGue, M., & Tellegen, A. (1990). The Minnesota Twin Family Registry: Some initial findings. Acta Gemel- logicae et Medicae, 39, 35-70.

McGue, M. (1995). Mediators and moderators of alcoholism inheritance. In J. R. Turner, I. R. Cardon, & J. K. Hewitt (Eds.), Behavior genetic approaches in behavioral medicine (pp. 17-44). New York: Plenum Press.

McGue, M., Lykken, D. T., & lacono, W. G. (1996). Genotype- environment correlations and interactions in the etiology of substance abuse and related behaviors. In H. W. Gordon & M. D. Glantz (Eds.), Individual differences in the biobehavioral etiology of drug abuse (pp. 49-72). Rockville, MD: National Institute on Drug Abuse.

McGue, M., Slutske, W., Taylor, J., & lacono, W. G. (1997). Personality and substance use disorders: I. Effects of gender and alcoholism subtype. Alcoholism: Clinical and Experimental Research, 21, 513-520.

Nathan, P. (1988). The addictive personality is the behavior of the addict. Journal of Consulting and Clinical Psychology, 56, 183-188.

Pickens, R. W., Svikis, D. S., McGue, M., Lykken, D. T., Heston, L. L., & Clayton, P. J. (1991). Heterogeneity in the inheritance of alcoholism: A study of male and female twins. Archives of General Psychiatry, 48, 19-28.

Robins, L. N., Babor, T. F., & Cottier, L. B. (1987). Composite Interna- tional Diagnostic Interview: Expanded Substance Abuse Module. St. Louis, MO: Washington University, Department of Psychiatry.

Robins, L. N., Wing, J., Wittchen, H. U., Helzer, J. E., Babor, T. F., Burke, J., Fanner, A., Jablenski, A., Pickens, R., Regier, D. A., Sartorius, N., & Towle, L. H. (1988). The Composite International Diagnostic Interview: An epidemiologic instrument suitable for use in conjunction with dif- ferent diagnostic systems and in different cultures. Archives of General Psychiatry, 45, 1068-1077.

Sher, K. J., & Trull, T. J. (1994). Personality and disinhibitory psychopa- thology: Alcoholism and antisocial personality disorder. Journal of Abnormal Psychology, 103, 92-102.

Sieber, M. F. (1981). Personality scores and licit and illicit substance use. Personality and Individual Differences, 2, 235—241.

Slater, E., Hare, E. H., & Price, J. (1971). Marriage and fertility of psychotic patients compared with national data. Social Biology, 18, S60-S73.

Spitzer, R. L., Williams, J. B. W., & Gibbon, M. (1987). Structured Clinical Interview for DSM-III-R. New York: New York State Psychi- atric Institute, Biometrics Research Department.

Sutker, P. B., & Allain, A. N. (1988). Issues in personality conceptualiza- tions of addictive behaviors. Journal of Consulting and Clinical Psy- chology, 56, 172-182.

Sutker, P. B., Archer, R. P., & Allain, A. N. (1978). Drug abuse patterns, personality characteristics, and relationships with sex, race, and sensa- tion seeking. Journal of Consulting and Clinical Psychology, 46, 1374- 1378.

Tarter, R. E., Alterman, A. I., & Edwards, K. L. (1985). Vulnerability to

404 McGUE, SLUTSKE, AND IACONO

alcoholism: A behavior—genetic perspective. Journal of Studies on Al- cohol, 46, 329-356.

Tellegen, A. (1982). Brief manual of the Multidimensional Personality Questionnaire. Unpublished manuscript, University of Minnesota.

Tellegen, A. (1991). Personality traits: Issues of definition, evidence and assessment. In D. Ciccheti & W. Grove (Eds.), Thinking clearly about psychology: Essays in honor of Paul Everett Meehl (Vol. 2, pp. 10-35), Minneapolis: University of Minnesota Press.

Tellegen, A., Lykken, D. T., Bouchard, T. J., Jr., Wilcox, K. J., Segal, N. L., & Rich, S. (1988). Personality similarity in twins reared apart and together. Journal of Personality and Social Psychology, 54, 1031-1039.

Tellegen, A., & Waller, N. G. (in press). Exploring personality through test construction: Development of the Multidimensional Personality Ques- tionnaire. In S. R. Briggs & J. M. Cheek (Eds.), Personality measures: Development and evaluation (Vol. 1). Greenwich, CT: JAI Press.

Vukov, M., Baba-Milkic, N., Lecic, D., Mijalkovic, S., & Marinkovic, J. (1995). Personality dimensions of opiate addicts. Acta Psychiatrica Scandinavica, 91, 103-107.

Warner, L. A., Kessler, R. C., Hughes, M., Anthony, J. C., & Nelson, C. B. (1995). Prevalence and correlates of drug use and dependence in the United States: Results from the National Comorbidity Survey. Archives of General Psychiatry, 52, 219-229.

Wills, T. A., Vaccaro, D., & McNamara, G. (1994). Novelty seeking, risk taking, and related constructs as predictors of adolescent substance use: An application of Cloninger's theory. Journal of Substance Abuse, 6, 1-20.

Received November 4, 1996 Revision received November 13, 1998

Accepted November 30, 1998

ORDER FORM Start my 1999 subscription to Journal of Consulting and Clinical Psychology! ISSN: 0022-ooex

$92.00, APA Member/Affiliate $ 183.00, Individual Non-Member $379.00, Institution TOTAL AMOUNT ENCLOSED $

Subscription orders must be prepaid. (Subscriptions are on a calendar basis only.) Allow 4-6 weeks for delivery of the first issue. Call for international subscription rates.

SEND THIS ORDER FORM TO: American Psychological Association Subscriptions 750 First Street, NE Washington, DC 20002-4242

ASSOCIATION

Or call (800) 374-2721, fax (202) 336-5568. TDD/TTY (202)336-6123. Email: [email protected]

Send me a Free Sample Issue Q

Q Check Enclosed (make payable to APA)

Charge my: Q VISAQMasterCard Q American Express Cardholder Name Card No. Exp. date

Signature (Required for Charge) Credit Card Billing Address City State Zip. Daytime Phone SHIP TO: Name ArMrpss

Citv_ State Zip APA Customer #

GAD99

PLEASE DO NOT REMOVE - A PHOTOCOPY MAY BE USED