Addictive Behaviors, Vol. 15, pp. 257-264, 1990 0306-4603/90 $3.00 + .00 Printed in the USA. All rights reserved. Copyright © 1990 Pergamon Press pie

PSYCHOMETRIC PROPERTIES OF THE DRUG ABUSE SCREENING TEST IN A PSYCHIATRIC PATIENT POPULATION

DOUGLAS STALEY and NADY EL-GUEBALY st. Boniface General Hospital, Winnipeg, Canada

A b s t r a c t - - The Drug Abuse Screening Test (DAST) is a 28-item screening instrument developed by Skinner for clinical screening and treatment evaluation research in the substance abuse field. In a sample of 250 psychiatric patients drawn from four treatment programs, the DAST evidenced high internal consistency reliability and good item-total score correlations. A factor analysis of the DAST item correlation matrix revealed a predominantly unidimensional scale with the possibility of rotating five additional factors reflecting a continuum of drug abuse. The five factors were interpreted as (a) self-recognition of a drug problem, (b) serious social consequences of drug use, (c) help-seeking for drug abuse, (d) illegal drug-related activities, and (e) inability to control drug use. The diagnostic validity of the DAST in discriminating patients according to DSM-III Substance Abuse diagnostic criteria was high and a range of valid clinical DAST cutoff scores from 5/6 through I0/11 was identified. The DAST appears to be a valid measure of drug involvement and abuse in a psychiatric patient population, a finding of increasing clinical relevance.

The accurate measurement of drug use and abuse is an important issue in the substance abuse field. A variety of techniques has been utilized to measure substance abuse, including self-report screening questionnaires (Klitzner, Schwartz, Gruenewald, & Blasinsky, 1987), interviewer ratings (Cohen, Karras, & Hughes, 1977), severity indices (McLellan et al . , 1985) and structured clinical interviews and drug screening methods such as urinalysis (Roehrich & Gold, 1987).

Although studies suggest that self-report questionnaires may be reliable and valid measures of drug use (Smart & Jarvis, 198 l) there has been a lack of rigorous psychometric evaluation of such scales. The data obtained from non-standardized instruments may not accurately reflect the actual level of drug use among the questionnaire respondents (Hochhauser, 1979). A further problem arises due to the fact that most drug use scales have been developed in nonclinical settings often using student populations (e.g. , Klitzner et al. , 1987; Single, Kandel, & Johnson, 1975; Smart & Blair, 1978). Skinner (1982) cautions about the generalizabili ty of self-report measures of drug abuse to a clinical setting where respondents may be motivated to conceal or distort drug use behavior. There is a clear need for a standardized self-report instrument with known psychometric properties as evaluated in both clinical and nonclinical populations.

The 28-item Drug Abuse Screening Test (DAST) was designed and developed by Skinner (1982) to provide a convenient instrument for clinical screening and treatment evaluation research. The DAST is self-administered and consists of items which parallel the Michigan Alcoholism Screening Test (MAST), a widely used assessment instrument for alcohol abuse. The DAST items yield a quantitative index of the range of problems related to drug abuse with the total score ranging from 0 to 28. A case-finding cut-off score for the DAST of six or greater is suggested for clinical purposes.

An evaluation of the measurement properties of the DAST in a clinical sample of 256 clients who voluntarily sought help for drug and alcohol problems indicated a high degree of internal consistency reliability (coefficient alpha = .92) and generally substantial item-total

Requests for reprints should be sent to Douglas Staley, M.A., St. Boniface General Hospital, Department of Psychiatry. 409 Tache Avenue, Winnipeg, Manitoba R2H 2A6, Canada.

257

258 DOUGLAS STALEY and NADY EL-GUEBALY

scale correlations (Skinner, 1982). A factor analysis of the DAST item intercorrelations revealed a unidimensional scale suggesting that the DAST primarily measures a single dominant dimension of problems associated with drug abuse. The DAST clearly differenti- ated clients with primarily drug-related problems from clients with alcohol and mixed alcohol/drug problems, suggesting the DAST has discriminative validity. Reportedly there was also very little influence on the DAST score by the response style biases of denial and social desirability. The need for further validity work in other populations and settings was also emphasized.

The purpose of the present study was threefold: (a) to evaluate some of the psychometric properties of the DAST scale and items in a sample of psychiatric inpatients and outpatients; (b) to conduct a principal components factor analysis of the DAST item correlation matrix in order to determine the internal structure of the instrument with a psychiatric patient population; and (c) to examine the test efficiency characteristics and the diagnostic validity of the DAST to accurately classify patients with a DMS-III diagnosis of Substance Abuse.

M E T H O D

Subjects The sample consisted of 250 consecutive admissions over an 8 month period to four

treatment programs in the department of Psychiatry of a teaching tertiary care hospital. There were 80 patients drawn from the Inpatient program, 50 patients from Day Hospital, 60 patients from Outpatient Anxiety Disorders and 60 from the Outpatient Substance Abuse programs. An additional 56 patients were not included in the study because they did not wish to participate or were considered ineligible due to lack of comprehension of English, severe organic brain syndrome or mental retardation. Subjects were approached within the first week of admission to a program and asked for their written consent to participate in the study.

The 250 patients in the sample were all 18 years of age or older with a mean age of 37.90 years (SD = 12.06). The sample group was 65% female and 35% male. With respect to marital status, 36% were single, 40% were married and 24% were either separated, divorced or widowed. Of the sample, 50% were employed part-time or full-time, 24% were unemployed and 26% were students, housekeepers or retired. With regard to education level, 27% had some high school, 24% had completed high school, 14% had vocational training diplomas and 28% had either some university education or had completed a university degree.

Instruments The DAST was administered as one of a number of instruments and questionnaires.

Subjects completed the DAST (Skinner, 1982), the 24-item Michigan Alcoholism Screening Test (MAST) (Seizer, 1971) and the Children of Alcoholics Screening Test (CAST) developed by Jones (1981). The MAST is a widely used screening instrument for alcohol abuse considered reliable and valid with clinical populations (Gibbs, 1983). The CAST is a 30-item instrument that measures feelings, attitudes, perceptions and experiences related to parental drinking behavior. Reliability and validity has been demonstrated for latency age and adult children of alcoholics (Pilat & Jones, 1985). In addition, subjects were given a sociodemographic questionnaire and a questionnaire regarding primary or secondary aspects of parental alcoholism. The DSM-III diagnosis of each subject was determined by the treating psychiatrist following a minimum of three admission days to a treatment program.

Psychometric properties of DAST 259

R E S U L T S

Correlations and diagnostic validity There was a low but significant correlation coefficient between the DAST and age (r =

- . 18, p < .01) and income (r = - . 15, p < .05), but none of the other sociodemographic variables showed a significant correlation. With respect to the other screening instruments, the DAST was positively correlated with the MAST (r = .45, p < .001) and the CAST (r = .31, p < .001).

To determine whether the DAST could discriminate between patients admitted to the Substance Abuse program and the other treatment programs, a one-way Analysis of Variance was computed. The results of this analysis indicated that the DAST scores were significantly different between the four programs (F(3, 245) = 28.91, p < .001). Multiple range tests showed that both the Substance Abuse and Inpatient programs differed significantly from the other two programs and from each other. The 60 patients in the Substance Abuse program all had a primary DSM-III diagnosis of Substance Abuse. The DAST clearly differentiated this group from the remaining sample of psychiatric patients. The Inpatient program also admitted substance abusers for a short period of stabilization.

The diagnostic validity of the DAST was assessed by dividing the total sample of 250 patients from all four treatment programs into two groups based on the presence or absence of a DSM-III Substance Abuse diagnosis. Subjects were then independently classified into test-positive and test-negative cases according to the total DAST score. Separate classifica-

Table 1..Validity analysis of the DAST cutoff thresholds

0/1 .98 .31 .32 .99 .47 1/2 .98 .53 .41 .99 .64 2/3 .98 .66 .49 .99 .74 3/4 .98 .75 .56 .99 .81 4/5 .96 .79 .60 .98 .83 5/6 .96 .81 .63 .98 .85 6/7 .94 .85 .68 .98 .87 7/8 .93 .88 .72 .97 .89 8/9 .91 .89 .74 .97 .89 9/10 .89 .90 .74 .96 .89

10/11 .82 .91 .75 .94 .88 11/12 .65 .91 .71 .89 .86 12/13 .63 .93 .76 .88 .86 13/14 .55 .95 .79 .86 .85 14/15 .45 .95 .76 .84 .83 15/16 .42 .96 .77 .83 .82 16/17 .38 .96 .78 .82 .82 17/18 .33 .97 .78 .81 .81 18/19 .29 .98 .84 .81 .81 19/20 .24 .98 .81 .80 .79 20/21 .20 .99 .85 .79 .78 21/22 . 16 .99 .90 .78 .78 22/23 .11 .99 .86 .77 .77 23/24 . 11 1.00 1.00 .77 .77 24/25 .05 1.00 1.00 .76 .76

DAST Positive Negative Cutoff Predictive Predictive Overall Score Sensitivity Specificity Value Value Accuracy

260 DOUGLAS STALEY and NADY EL-GUEBALY

Table 2. Point-biserial item-total correlations for the DAST

Item-Total Item-Total Item Correlation Item Correlation

1 .597 15 .754 2 .682 16 .319 3 .785 17 .653 4 .216 18 .603 5 .632 19 .524 6 .645 20 .293 7 .005 21 .553 8 .640 22 .473 9 .748 23 .791

I0 .710 24 .699 11 .745 25 .711 12 .638 26 .621 13 .527 27 .705 14 .716 28 .627

tions were completed using DAST cutoff thresholds ranging from 1 through 25. Test efficiency characteristics (sensitivity, specificity, positive predictive value, negative predic- tive value, and overall accuracy) were calculated using the DSM-III diagnosis as the criterion measure (Weinstein et al., 1980). Separate analyses were conducted for each of the DAST cutoff scores by cross-classifying cases by DAST score (positive or negative) and DSM-III Substance Abuse diagnosis (present or absent).

The results of the validity analysis for each of the DAST cutoff thresholds are presented in Table 1. The sensitivity of a test is equal to its true-positive rate, while specificity is the true-negative rate. Positive predictive value indicates the frequency of DSM-III Substance Abuse in those with positive test results, while negative predictive value is the frequency of non-Substance Abuse in those with negative test results. Overall accuracy is the proportion of correctly classified cases. The validity analysis indicated that the DAST attained a maximum accuracy of 89% and exhibited a high level of sensitivity, specificity and overall accuracy over a cutoff range from 5/6 to 10/11. Throughout this range overall accuracy was maintained at 85% or above, while sensitivity decreased from .96 to .82 and specificity increased from .81 to .91. This suggests a range of possible valid DAST cutoff score thresholds which may be selected by the clinician or researcher depending on the relative importance of either high sensitivity or high specificity. Using the lower DAST cutoff score of six or greater as an indication of clinical significance, the data shows that 75% of the Substance Abuse patients, 27.5% of the Inpatient sample, 18% of the Day Hospital sample and 13.3% of the Anxiety Disorders patients exceeded the defined cutoff value. The DAST appears to be a sensitive screening instrument for the abuse of drugs other than alcohol in a heterogeneous psychiatric patient population.

Item analysis The results of an item analysis of the DAST are summarized in Table 2. Item-total scale

correlation coefficients were computed with the point-biserial correlation and suggest that most items have a moderate to high item-total correlation. The pattern and magnitude of item-total scale correlations were similar to that obtained by Skinner (1982) in his clinical sample of 256 drug/alcohol abuse clients. Only five items showed relatively low item-total correlations: Item 4 ( "Can you get through the week without using drugs other than those required for medical reasons?"), Item 7 ( " D o you try to limit your drug use to certain

Psychometric properties of DAST 261

Table 3. Significant factor loadings for the DAST items on each factor following Varimax rotation

Factor I

.79 11, Do your friends or relatives know or suspect you abuse drugs?

.76 10, Does your spouse (or parents) ever complain about your involvement with drugs?

.72 3. Do you abuse more than one drug at a time?

.71 8. Have you had "blackouts" or "flashbacks" as a result of drug use?

.67 2, Have you abused prescription dugs?

.66 9. Do you ever feel bad about your drug abuse?

.63 12. Has drug abuse ever created problems between you and your spouse?

.61 23, Have you ever experienced withdrawal symptoms as a result of heavy drug intake?

.61 6, Do you abuse drugs on a continuous basis?

.60 24. Have you had medical problems as a result of your drug use?

.48 1. Have you used drugs other than those required for medical reasons?

.44 5. Are you always able to stop using drugs when you want to?

.42 25. Have you ever gone to anyone for help for a drug problem?

.41 15. Have you ever neglected your family or missed work because of your use of drugs?

Factor

.75 19.

.70 17.

.68 20.

.67 13.

.58 16.

.57 18.

.55 14.

.50 15.

.45 24.

.40 26.

1I

Have you ever been arrested because of unusual behavior while under the influence of drugs? Have you ever lost a job because of drug abuse? Have you ever been arrested for driving while under the influence of drugs? Has any family member ever sought help for problems related to your drug use? Have you ever been in trouble at work because of drug abuse? Have you ever gotten into fights when under the influence of drugs? Have you ever lost friends because of your use of drugs? Have you ever neglected your family or missed work because of your use of drugs? Have you had medical problems as a result of your drug use? Have you ever been in hospital for medical problems related to your drug use?

Factor

.82 27.

.78 28.

.65 25.

.55 23.

.54 26.

.48 15.

.43 9.

III

Have you ever been involved in a treatment programme specifically related to drug use? Have you ever been treated as an out-patient for problems related to drug abuse? Have you ever gone to anyone for help for a drug problem? Have you ever experienced withdrawal symptoms as a result of heavy drug intake? Have you ever been in hospital for medical problems related to your drug use? Have you ever neglected your family or missed work because of your use of drugs? Do you ever feel bad about your drug abuse?

Factor 1V

.76 22. Have you ever been arrested for possession of illegal drugs?

.70 21. Have you ever engaged in illegal activities in order to obtain drugs?

Factor V

.72. 7. Do you try to limit your drug use to certain situations?

.66 4. Can you get through the week without using drugs (other than those required for medical reasons)?

.52 5. Are you always able to stop using drugs when you want to?

s i t u a t i o n s ? " ) , I t e m 16 ( " H a v e y o u e v e r b e e n in t r o u b l e at w o r k b e c a u s e o f d r u g a b u s e ? " ) ,

I t e m 20 ( " H a v e y o u e v e r b e e n a r r e s t e d fo r d r i v i n g w h i l e u n d e r t h e i n f l u e n c e o f d r u g s ? " ) a n d

I t e m 22 ( " H a v e y o u e v e r b e e n a r r e s t e d fo r p o s s e s s i o n o f i l l ega l d r u g s ? " ) . It is o f i n t e r e s t to

n o t e t ha t t h e s e s a m e f i ve i t e m s w e r e a l so a m o n g t he l o w i t e m - t o t a l c o r r e l a t i o n i t e m s

i d e n t i f i e d b y S k i n n e r . B o t h i t e m s 4 a n d 7 a re r e l a t e d to a l o s s o f c o n t r o l i s s u e , d i s t i n g u i s h i n g

b e t w e e n r e g u l a r v e r s u s m o r e e p i s o d i c d r u g u s e . I t e m s 16, 20 a n d 22 i n d i c a t e s e r i o u s

262 DOUGLAS STALEY and NADY EL-GUEBALY

consequences due to drug abuse and typically manifest at a later stage of abuse. Therefore, the five low item-total correlation items are all indexing serious, long-term consequences of fairly sustained drug abuse and may be endorsed much less consistently than other DAST items, accounting for their low correlations.

The coefficient alpha, a measure of internal consistency reliability of the DAST items, was computed to be .94 which is consistent with the value of .92 obtained by Skinner (1982). This suggests uniform responding to all DAST items and a fairly homogeneous content among the items (Nunnally, 1978).

Factor analysis A principal components factor analysis was computed on the 28-item intercorrelation

matrix of the DAST using the SAS FACTOR program with Varimax rotation on a mainframe computer. The eigenvalues for the first five factors were 11.8, 2.1, 1.5, 1.3 and 1.2 and these five factors were retained for Varimax rotation and subsequent interpretation, The five factors accounted for 22.5%, 15.3%, 13.7%, 6.8% and 6.0%~ respectively, of the total variance. For purposes of interpretation only factor loadings greater than .40 were considered significant. Table 3 presents the significant factor loadings for items on each of the five factors.

Factor I was defined by 14 items with factor loadings of .40 or greater. The highest loading items on this factor suggest that Factor I is concerned with indications of early psychosocial complications due to drug abuse and to the acknowledgment and self-recognition of a drug problem. Factor II, with 10 items showing significant factor loadings, measures more advanced, late onset social consequences of drug use, such as arrests, job loss and medical problems. The seven items with significant loadings on Factor III suggest a further stage of substance abuse related primarily to treatment or help-seeking as a consequence of drug abuse. Factor IV consisted of just two items with significant factor loadings, both concerned with illegal drug-related activities. The final factor, with three items showing significant loadings, can be interpreted as an inability to control drug use factor.

When interpreting this five factor solution it must be kept in mind that the pattern of eigenvalues indicates that the first unrotated factor accounts for 42.4% of the variance, while the second unrotated factor accounts for 7.5% of the variance. The size of the first unrotated factor suggests a fairly strong general factor underlying the DAST item structure, implying a predominantly unidimensional scale. Additional rotated factors, which distribute the variance, may then be derived and interpreted. The present multifactor solution suggests that the DAST may be measuring a continuum of problems associated with drug abuse, from the early stages of recognition of a drug problem to more advanced stages representing serious social and medical consequences of abuse.

D I S C U S S I O N

The Drug Abuse Screening Test appears to be a useful instrument for measuring drug involvement and problems associated with the abuse of drugs other than alcohol. The DAST evidenced high internal consistency reliability and good item-total scale correlations in a diverse psychiatric patient population. With the increasing prevalence of substance abuse recorded among younger general psychiatric patients (Galanter, Castaneda, & Ferman, 1988) and the substantive minority of substance abusers with concurrent psychopathology (Mirin, Weiss, & Michael, 1988), the DAST's validity in our sample is clinically significant.

The factor analysis of the DAST item intercorrelations suggests that the DAST items form a predominantly unidimensional scale, similar to the finding reported by Skinner (1982). This conclusion is supported by the fact that the first unrotated factor of the principal

Psychometric properties of DAST 263

components analysis accounted for 42.4% of the variance, in contrast to 7.5% of the variance accounted for by the second unrotated factor. However, following Varimax rotation, five additional factors were identified, suggesting a continuum of drug abuse. Factor I, consisting of half of the DAST items, was interpreted as self-recognition of a drug problem. Subsequent factors detailed more extensive involvement in substance abuse: Factor II (serious social consequences of drug use), Factor III (help-seeking for problems due to drug abuse), Factor IV (illegal drug-related activities) and Factor V (lack of ability to control drug use).

A five-factor solution of the DAST item structure has also been reported by Skinner and Goldberg (1986), who conducted a factor analysis of the DAST in a clinical sample of 105 narcotic users and identified five factors reflecting general drug dependence, social problems, medical problems, polydrug use and previous treatment. These five factors are similar in many respects to the factors described in the present investigation. Thus it appears that although the DAST is fundamentally a unidimensional measure of general drug abuse (Skinner, 1982), additional factors may be interpreted reflecting specific aspects and problems related to drug abuse.

The diagnostic validity analysis indicated that the DAST attained a maximum overall accuracy of 89% in classifying patients according to DSM-III Substance Abuse diagnosis. High sensitivity, high specificity and an overall accuracy rate above 85% were maintained between DAST cutoff thresholds of 5/6 through 10/11. The data suggests that this range of cutoff scores offers the clinician/researcher a choice of valid cutoff points depending upon the need for high test sensitivity or high test specificity. Using the lower cutoff threshold of 5/6, the DAST has maximum sensitivity and is very good at detecting substance abuse cases. At the higher cutoff score specificity is maximized and the DAST is effective at screening out non-substance abusers.

The DAST showed a small significant negative correlation with both age and income. This finding is not surprising in light of previous research with substance abuse populations (Nicholi, 1983). The DAST was significantly related to both the MAST, a measure of alcohol abuse, and the CAST, an index of parental drinking behavior. A positive relationship between drug abuse, alcohol abuse and parental drinking problems is to be expected (Black, Bucky, & Wilder-Padilla, 1986).

In summary, the DAST exhibited valid psychometric properties in a heterogeneous psychiatric patient population. With the growing recognition of the "dual diagnoses" patient with both mental illness and substance abuse, such an instrument is clinically increasingly required. The fact that the items parallel those on the MAST supports the use of the DAST as a complementary measure to the MAST for quantifying levels of drug abuse. Further research to delineate the range of applicability of this promising instrument is recommended.

R E F E R E N C E S

Black, C., Bucky, S.F., & Wilder-Padilla, S. (1986). The interpersonal and emotional consequences of being an adult child of an alcoholic. International Journal of the Addictions, 21, 213-231.

Cohen, M., Karras, A., & Hughes, R. (1977). The usefulness and reliability of a drug severity scale. International Journal of the Addictions, 12, 417-422.

Galanter, M., Castaneda, R., & Ferman, J. (1988). Substance abuse among general psychiatric patients: Place of presentation, diagnosis and treatment. American Journal of Drug and Alcohol Abuse, 14, 211-235.

Gibbs, L.E. (1983). Validity and reliability of the Michigan Alcoholism Screening Test: A review. Drug and Alcohol Dependance, 12, 279-285.

Hochhauser, M. (1979). Bias in drug abuse survey research. International Journal of the Addictions, 14, 675~587. Jones, J.W. (1981). The Children of Alcoholics Screening Test (CAST). Chicago: Family Recovery Press. Klitzner, M., Schwartz, R.H., Gruenewald, P., & Blasinsky, M. (1987). Screening for risk factors for adolescent

alcohol and drug use. American Journal of Diseases of Children, 141, 45-49.

264 DOUGLAS STALEY and NADY EL-GUEBALY

McClellan, A.T., Luborsky, L., Cacciola, J., Griffith, J., Evans, F., Barr, H.L., & O'Brien, C.P. (1985). New data from the Addiction Severity Index: Reliability and validity in three centers. Journal of Nervous and Mental Diseases, 173, 412-423.

Mirin, S.M., Weiss, R~D., & Michael. J. (1988). Psychopathology in substance abusers: Diagnosis and treatment. American Journal of Drug and Alcohol Abuse, 14, 139-157.

Nicholi, A.M. (1983). The nontherapeutic use of psychoactive drugs: A modem epidemic. The New England Journal of Medicine, 308, 925-933.

Nunnally, J.C. (1978). Psychometric theory. New York: McGraw-Hill. Pilat, J.M., & Jones, J.W. (1985). Identification of children of alcoholics: Two empirical studies. Alcohol Health

and Research World, 9, 27-36. Roehrich, H., & Gold, M.S. (1987). Diagnosis of substance abuse in an adolescent psychiatric population.

International Journal of Psychiatry in Medicine, 16, 137-143. Seizer, M.L. (1971). The Michigan Alcoholism Screening Test: The quest for a new diagnostic instrument.

American Journal of Psychiatry, 127, 1653-1658. Single, E., Kandel, D., & Johnson, B.D. (1975). The reliability and validity of drug use responses in a large scale

longitudinal survey. Journal of Drug Issues, Fall, 426-443. Skinner, H.A. (1982). The Drug Abuse Screening Test. Addictive Behaviors, 7, 363-371. Skinner, H.A., & Goldberg, A.E. (1986). Evidence for a drug dependence syndrome among narcotic users. British

Journal of Addiction, 81,479-484. Smart, R.G., & Blair, N.L. (1978). Test-retest reliability and validity information for a high school drug use

questionnaire. Drug and Alcohol Dependence. 3, 265-271. Smart, R.G., & Jarvis, G.K. (1981). Do self-report studies of drug use really give dependable results? Canadian

Journal of Criminology, 23, 83-92. Weinstein, M.C., Fineberg, H.V., Elstein, A.S., Frazier, H.S., Neuhauser, D., Neutra, R.R., & McNeil, B.

(1980). Clinical decision analysis. Toronto: W. B. Saunders Company.