Severity of Dependence Scale: Establishing a cut-off point for cannabis dependence in the German adult population

Transcription

1 ORIGINALARBEIT DOI / Severity of Dependence Skala: Ermittlung eines Cut- Off-Wertes für Cannabisabhängigkeit in der deutschen Erwachsenenbevölkerung Susanne Steiner 1, 2, Sebastian E. Baumeister 1 & Ludwig Kraus 1 Severity of Dependence Scale: Establishing a cut-off point for cannabis dependence in the German adult population Key words Severity of Dependence Skala, Cannabis, Konstruktvalidität, Cut-Off-Wert Schlüsselwörter Severity of Dependence Scale, cannabis, construct validity, cut-off point Zusammenfassung Ziel: Ziel der Studie ist die Validierung der Severity of Dependence Skala (SDS) an einer Stichprobe von 18- bis 64-jährigen Cannabiskonsumenten in Deutschland. Durch den Vergleich der SDS Werte mit DSM-IV Cannabisdiagnosen wird ein Cut-Off-Wert zur optimalen Unterscheidung von abhängigen und nichtabhängigen Konsumenten berechnet. Methodik: Als Datengrundlage dient der Epidemiologische Suchtsurvey 2006 (ESA), die Analysestichprobe umfasst 456 Cannabiskonsumenten. Mittels Receiver Operating Characteristic (ROC) Analysen werden Sensitivität, Spezifität und optimale Cut- Off-Werte ermittelt. Ergebnisse: Die interne Konsistenz beträgt = 0,796, eine Hauptkomponentenanalyse (PCA) zeigt eine Ein-Faktor-Lösung und der SDS Gesamtwert korreliert signifikant mit DSM- IV Cannabisabhängigkeit, der Anzahl erfüllter DSM-IV Kriterien für Cannabisabhängigkeit, der Konsumhäufigkeit während der letzten 12 Monate und der durchschnittlichen Anzahl an Joints pro Tag. Ein optimaler Cut-Off-Wert von zwei bildet das beste Verhältnis von Sensitivität (93,6 %) und Spezifität (74,0 %). Für männliche Konsumenten wird ein Wert von vier gefunden. Schlussfolgerungen: Die deutsche Version des SDS stellt ein kurzes, reliables und valides Screeninginstrument für Cannabisabhängigkeit in der Allgemeinbevölkerung dar. Der SDS hat eine hohe Sensitivität, führt jedoch in epidemiologischen Untersuchungen zu einer Überschätzung der Prävalenz der Cannabisabhängigkeit. Abstract Aims: The study aimed to validate the Severity of Dependence Scale (SDS) in a sample of 18-to 64-year-old cannabis users in Germany. Data from current cannabis users were analysed to determine an optimal cut-off point that discriminates between dependent and non dependent users. SDS scores were compared against DSM- IV diagnoses of cannabis dependence as the gold standard. Method: Data came from the 2006 German Epidemiological Survey of Substance Abuse (ESA); the sample for analysis consisted of 456 cannabis users. Receiver Operating Characteristic (ROC) analysis was used to determine SDS sensitivity, specificity, and an optimal cut-off. Subgroup analyses by gender and age were conducted. Results: Internal consistency ( = 0.796) was high. A Principal Components Analysis resulted in a single factor solution, and the SDS total score correlated significantly with DSM-IV diagnosis of cannabis dependence, number of DSM-IV criteria, frequency of use within the past 12 months, and the mean number of joints smoked per day in the last 30 days. A cut-off point of two provided the best trade-off between sensitivity (93.6 %) and specificity (74.0 %). For male cannabis users, a cut-off of four was selected. Conclusions: The results indicate that the German brief version of the SDS is a reliable and valid instrument for screening for cannabis dependence in the general population. The SDS is highly sensitive, but it overestimates the prevalence of cannabis dependence in epidemiological studies. Introduction During the last two decades, almost all countries in the European Union have experienced a considerable increase in the use of cannabis (European Monitoring Centre for Drugs and Drug Addiction, 2004). In Germany, lifetime cannabis use increased from 14 % in 1990 to 34 % in 2006 (Kraus, Pfeiffer-Gerschel & Pabst, 2008). In addition, the number of treatment admissions for primary cannabis-related disorders increased tenfold between 1992 and 2005 (Simon & Kraus, 2007). Several instruments are available for estimating the prevalence of cannabis dependence in the general population. The Diagnostic and Statistical Manual of Mental Disorders (DSM-IV, American Psychiatric Association, 1994) is currently the»gold standard«for assessing cannabis dependence (Babor, 2006). The Composite International Diagnostic Interview (CIDI, World Health Organization, 1990) is a widely accepted and frequently used operationalization of the DSM-IV diagnostic criteria. However, the relevant CIDI section consists of 18 extensive questions concerning negative consequences of cannabis use. It is, therefore, too long and requires too much 1 IFT Institut für Therapieforschung, München 2 Institut und Poliklinik für Arbeits-, Sozialund Umweltmedizin der Ludwig-Maximilians-Universität München S57

2 ORIGINALARBEIT S. Steiner et al. Severity of Dependence Skala time for routine use in health surveys. Only a few screening instruments exist that may be used instead. Although the Cannabis Abuse Screening Test (CAST; Legleye, Karila, Beck & Reynaud, 2007) is intended to measure the severity of cannabis abuse, and the Cannabis Use Disorders Identification Test (CUDIT; Adamson & Sellman, 2003) screens for cannabis-use disorders (abuse or dependence), other available screening questionnaires, such as the Marijuana Screening Inventory (Alexander, 2003) or the Substance Dependence Severity Scale (Miele et al., 2000a, 2000b), are too time-consuming to use in epidemiologic research. The Severity of Dependence Scale (SDS; Gossop et al., 1995) is a short and easily administered scale for measuring the degree of dependence on different types of drugs. It focuses on psychological components of dependence, whereas the DSM-IV covers a broader spectrum of dependence symptoms. As Edwards, Arif and Hadgson (1981) recognize in their seminal paper, the compulsive use of drugs is a central feature of an addictive behaviour. More recent definitions of addiction also emphasize the psychological components of dependence instead of tolerance and withdrawal (West & Gossop, 1994; Soellner, 2005). The SDS contains five items dealing with the individual s feeling of impaired control over his/her own drug use and with his/her concerns and anxieties about the use. Because of its simplicity, the scale seems to be useful as a screening instrument in both routine care (Swift, Copeland & Hall, 1998) and epidemiological studies (Gossop et al., 1995). Three validation studies on the SDS for cannabis use have been conducted, all of them with Australian samples. In the first study (Swift et al., 1998), 200 long-term cannabis users were recruited from newspaper advertisements or by word of mouth. The authors compared the SDS against DSM-III diagnoses and established a cut-off score of three. The second study (Martin, Copeland, Gates & Gilmour, 2006) assessed the practicability of the SDS in a convenience sample of young cannabis users between 14 and 18 years old and determined an optimal cut-off score of four. In their recent study, Hides, Dawe, Young and Kavanagh (2007) recruited cannabis users from acute psychiatric wards and examined the ability of the SDS to detect DSM-IV cannabis dependence in people with a psychosis. The authors reported a cut-off point of two that optimally discriminated between dependent and non-dependent users. Most research to date has used convenience samples or selective samples, such as psychiatric inpatients, to determine cut-off values. It is not clear whether these cut-off points can be generalized to the population of adult cannabis users. Against this background, the present study tested cannabis users selected from a generalpopulation sample in order to validate the SDS and determine cut-off points that might be generalized to the population of adult cannabis users. Method Sample The data for the study were taken from a nationwide, cross-sectional survey on the use of psychoactive substances (German Epidemiological Survey of Substance Abuse, ESA), which was conducted in Participants were selected using a two-stage probability design. In the first stage, 225 communities stratified by region and community size were selected proportional to the population size. In the second stage, 21,463 participants aged from 18 to 64 years and stratified by seven age groups were randomly selected from population registries. Younger age groups were oversampled. The net sample (after the exclusion of migrated or deceased persons) comprised 17,961 eligible participants, among whom 7,912 persons participated; this amounted to an adjusted response rate of 45 %. In an attempt to increase the response rate, a mixed mode design was used. People who did not complete the self-administered questionnaire after a second reminder were invited to answer the questions by telephone (Kraus & Baumeister, 2008). Inclusion criteria for the analytical sample were reported cannabis use during the last 12 months and a completed SDS and CIDI. The final sample for the analysis consisted of 456 current cannabis users. Within the total sample of 7,912 persons, two cases had missing information on 12-months cannabis use; 25 cases had missing values for at least one of the seven DSM- IV criteria for cannabis dependence; and 33 cases had incomplete information on the SDS. In the final sample of 456 current cannabis users, six cases had missing information about their age. They were excluded from the subgroup analysis according to age. Measures The original items on the SDS (Gossop et al., 1995) were adapted for cannabis and were translated into German and then independently backtranslated into English. The five items referring to cannabis use during the last 12 months were: (1) Did you think your use of cannabis was out of control?; (2) Did the prospect of missing a dose of cannabis make you anxious or worried?; (3) Did you worry about your use of cannabis?; (4) Did you wish you could stop the use of cannabis?; and (5) How difficult did you find it to stop or go without cannabis?. Each item is scored on a four-point scale: items 1 through 4 (0, never/almost never; 1, sometimes; 2, often; 3, always/nearly) and item 5 (0, not difficult; 1, quite difficult; 2, very difficult; 3, impossible). The total SDS score can range from 0 to 15. Diagnostic assessments were based on the paper-and-pencil version of the Munich Composite International Diagnostic Interview (M-CIDI; Wittchen et al., 1995). Past 12-months cannabis abuse and cannabis dependence were assessed according to the DSM-IV criteria. The M-CIDI is an updated version of the World Health Organization s CIDI version 1.2 (WHO-CIDI; World Health Organization 1990), which incorporates questions to cover DSM-IV diagnostic criteria (American S58

3 S. Steiner et al. Severity of Dependence Scale RESEARCH REPORT Psychiatric Association, 1994). The reliability and procedural validity of the M-CIDI have been established (Lachner et al., 1998). Frequency of cannabis use during the last 12 months was measured using a categorical response format (0, never; 1, once; 2, two to five times; 3, six to nine times; 4, ten to 19 times; 5, 20 to 59 times; 6, 60 to 99 times; 7, 100 to 199 time; 8, more than 200 times). Frequency of cannabis use during the last 30 days was assessed by a continuous variable which was categorized into four categories (0, none; 25 30, almost daily; 4 24, one to five times per week; 1 3, less often than weekly). The number of illegal drugs other than cannabis was calculated by counting the number of drugs (amphetamine, ecstasy, LSD, opiates, cocaine, crack or magic mushrooms) used more than five times during the last 12 months. Analysis The reliability of the SDS was assessed by Cronbach s coefficient denoting a steady rate of internal consistency. Three ways of analysing construct validity were used. First, Principal Components Analysis (PCA) was used to investigate the factor structure. Factors with eigenvalues greater than 1.00 were retained. Second, the total SDS score was correlated with related variables, such as DSM-IV cannabis dependence, the number of positive DSM-IV criteria, consumption frequency during the past 12 months, and the mean number of joints smoked per day during the last 30 days. Third, Receiver Operating Characteristic (ROC) analysis was conducted to compare the performance of the SDS against the diagnosis of DSM-IV cannabis dependence derived from the M-CIDI. The ROC curve describes the relationship between sensitivity and one minus specificity; it plots these two rates for a range of cut-off values. Sensitivity represents the proportion of true positives, i. e. the proportion of people being diagnosed as dependent by the SDS, to all people showing DSM-IV dependence. Specificity indicates the proportion of true negatives, i. e. people not being diagnosed as dependent either by the DSM-IV or the SDS. An important rate for describing the ability of a test to discriminate between cases and non-cases constitutes the area under the curve (AUC). The AUC can have a value between 0.5 (which describes a true case being diagnosed by chance) and 1.0 (which stands for perfect discrimination between cases and non-cases). In order to determine the best balance between sensitivity and specificity, the Youden index (Youden, 1950) was calculated for each score. Youden s index is the sum of sensitivity and specificity minus one; it gives equal weight to the two indicators. The cut-off with the highest Youden index indicates the value that best discriminates between the presence and absence of a DSM-IV diagnosis of cannabis dependence. Another major characteristic of a diagnostic test is its predictive value; it is the probability that the test (SDS) will give the correct diagnosis. Whereas a positive predictive value (PPV) constitutes the probability of cannabis dependence in a person with a positive diagnosis based on the SDS, a negative predictive value (NPV) represents the probability of not being a case when the SDS is negative (Fletcher, Fletcher & Wagner, 1996). Furthermore, a subgroup analysis by gender and age group was conducted to explore the stability of the cut-off point. Finally, a cut-off for cannabis dependence syndrome that omitted tolerance and withdrawal as defined by the American Psychiatric Association (1994) was calculated. Data were weighted using probability and poststratification weights. Bivariable statistics and multivariable regression models were calculated using SVY procedures in the Stata software package (Stata Special Edition 9.2; Stata Corp., 2005). Taylor series linearization was used for statistical inference testing, including confidence intervals. Results Characteristics of cannabis users The mean age of the subsample of cannabis users was 28.7 years (sd = 8.5); 62.1 % were male. The group was primarily of German descent (94.9 %), with other ethnicities ranking far behind. The majority of the respondents were single (83.8 %); 8.3 % were married and cohabitating; 2.1 % were married but not cohabitating; 4.4 % were divorced; and 0.5 % were widowed. On average, respondents had tried cannabis for the first time at the age of 18.4 years (sd = 5.1; range = years). Frequency of use within the last 30 days varied widely. About 10.4 % of the sample reported almost daily use; 17.8 % used cannabis one to five times per week; 21.1 % used cannabis less often than weekly; and 50.7 % had not used cannabis within the last 30 days. Fortythree (9.7 %) of the cannabis users met DSM-IV criteria for cannabis dependence; 63.6 % of the sample reported no symptom of cannabis dependence; 26.7 % reported one or two symptoms; 8.5 %, three or four symptoms; and 1.2 %, five or more symptoms. About 15.7 % of the sample fulfilled the criteria for a DSM-IV diagnosis of cannabis abuse. The mean SDS Scale score was 1.6 (sd = 2.5, range = 1 15). Over half (54.8 %) of the participants had a score of zero on the SDS scale; 28.0 % scored one to three; 11.9 %, four to six; 3.5 %, seven to nine; and 1.6 % had a score of 10 or higher. Reliability and validity Within the subsample of cannabis users, the SDS was found to have high internal consistency (Cronbach s alpha = 0.796). An unrotated PCA resulted in a single-factor solution. This factor had an eigenvalue of 2.9, and it accounted for 57.6 % of the total variance in the scores. Factor loadings ranged from 0.54 to The factorability of the correlation matrix was supported by Barlett s test of sphericity (p <.001) and a Kaiser-Meyer-Olkin value of 0.8 (Tabachnick & Fidell, 2007). The total SDS score was significantly associated with related variables. The Spearman correlation between SDS scores and a DSM-IV diagnosis of cannabis dependence was 0.49 S59

4 ORIGINALARBEIT S. Steiner et al. Severity of Dependence Skala (p < 0.001); between SDS scores and the total number of DSM-IV cannabis dependence criteria, it was 0.59 (p < 0.001); between SDS scores and use frequency within the past 12 months, it was 0.45 (p < 0.001); and between SDS scores and mean number of joints smoked, it was 0.32 (p < 0.001). Receiver Operating Characteristics Analysis Figure 1 displays the ROC curve. The area under the curve (AUC) is (95 % CI: ), indicating a good ability of the SDS to discriminate between individuals with and without a diagnosis of cannabis dependence. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and the corresponding Youden index for relevant cut-off points are presented in Table 1. A maximum Youden index (0.686) was found at a cut-off point of two. This score provided the best trade-off between sensitivity (93.6 %; 95 % CI: ) and specificity (74.0 %; 95 % CI: ). Compared to the DSM-IV diagnoses, the SDS had a NPV of 99.0 % (95 % CI: ) and a PPV of 28.8 % (95 % CI: ). Table 2 shows the frequency distribution of diagnoses as dependent or not dependent according to both the DSM-IV and the SDS. In 75.4 % (n = 344) of the cases, the two instruments led to congruent results, i. e. correct positive (n = 39) and correct negative (n = 305) diagnoses. In 112 of the cases, diagnoses from the two instruments were incongruent. Among the incongruent cases, four participants were diagnosed as positive according to the DSM-IV but as negative on the basis of the SDS, i. e. the diagnoses were false negatives; 108 of the cannabis users were diagnosed as false positives; i. e. they had a positive SDS diagnosis, but they were not dependent according to the DSM-IV. The characteristics of the participants who were incorrectly identified as dependent (false positives) and those correctly identified as not dependent (correct negatives) are represented in Table 3. Whereas 26.3 % of the false positives fulfilled two criteria for DSM-IV cannabis dependence, only 1.7 % of the true negatives did so. A diagnosis of cannabis abuse among the false positives was almost five times as likely as among the true negatives. In addition, the false positives had a significantly higher frequency of cannabis use during the last 30 days. Subgroup analysis Cut-off points for age and gender Figure 1: Receiver Operating Characteristic (ROC) Curve Sensitivity Area under ROC curve = subgroups were also calculated. Whereas female cannabis users, young adult (aged 18 to 24 years) cannabis users, and adult (aged 25 to 64 years) cannabis users all had a cut-off point of two (the same as the entire sample), males had a cut-off point of four. Because the SDS is intended to assess the psychological component of cannabis dependence, the criteria tolerance and withdrawal were omitted from the DSM-IV definition of cannabis dependence, and a new cut-off point was determined. This change in the defini Specificity Table 1: Sensitivity, specificity, positive predictive value, negative predictive value and Youden index for the Severity of Dependence Scale (SDS) for various cut-off points SENS SPEC PPV NPV Youden SDS SDS SDS SDS SDS SDS SENS = sensitivity, SPEC = specificity; PPV = positive predictive value, NPV = negative predictive value; Youden = Youden Index (SENS+SPEC-1) Table 2: SDS (cut-off of 2) and DSM-IV diagnosis of cannabis dependence (n = 456) DSM-IV diagnosis SDS diagnosis dependent not dependent dependent not dependent S60

5 S. Steiner et al. Severity of Dependence Scale RESEARCH REPORT Table 3: Characteristics of cases incorrectly diagnosed as cannabis dependent by the SDS (false positives) and cases correctly diagnosed as not cannabis dependent by SDS and DSM-IV (correct negatives) DSM-IV cannabis DSM-IV Frequency of cannabis use Number of other illegal dependence cannabis (last 30 days) drugs used more than (no. of symptoms) abuse 5 times (last 12 months) % % % % False yes 42.7 none positives less than weekly (n=108) times per week almost daily and more 2.6 Correct yes 8.5* none negatives * less than weekly (n=305) 2 1.7* 1 5 times per week 12.9* almost daily 3.5* 3 and more 0.7 * p < 0.05, derived from a binary logistic regression with»false positives«vs.»correct negatives«as the dependent variable and age and gender as additional independent variables Table 4: Cut-off points, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and Youden index for the SDS for different subgroups and different DSM-IV definitions of cannabis dependence DSM-IV Subgroups n Cut-off SENS SPEC PPV NPV Youden Definition Cannabis dependence a male female years years Cannabis dependence b total sample a DSM-IV definition b DSM-IV definition without withdrawal and tolerance SENS = sensitivity, SPEC = specificity, PPV = positive predictive value, NPV = negative predictive value, Youden = Youden Index (SENS+SPEC-1) tion, however, had no significant impact on the results. The analysis yielded the same cut-off point of two as for the original definition (Table 4). Discussion The results of the present study demonstrate that the German version of the Severity of Dependence Scale has good diagnostic utility for cannabis dependence. We found both good internal consistency and adequate concurrent validity. The PCA showed that the SDS is a unidimensional scale, with all of the items loading on a single factor. Higher SDS scores were associated with more DSM-IV dependence symptoms, a higher rate of DSM-IV diagnoses of cannabis dependence, more frequent cannabis use, and a greater number of joints used per day. The ROC indicated that the ability of the SDS to identify cannabis users as dependent or not was good. The analysis suggested that a cut-off of two optimally discriminates between cases and non-cases of cannabis dependence in the general adult population. Compared to previous validation studies, the determined cut-off of two appears quite liberal. Swift et al. (1998) reported a cut-off of three for optimal discrimination in a convenience sample of 200 long-term cannabis users, and Martin et al. (2006) established a cut-off of four in a sample of young cannabis users. Hides et al. s (2007) validation study assessed a sample of cannabis users from acute psychiatric wards and also found a cut-off of two. Along with earlier studies, the current results indicate that a standard diagnostic criterion for the SDS does not exist, and cut-off points need to be defined independently for different samples. Depending on the context in which the SDS is applied, one needs to consider whether to maximize the proportion of correct positives or of true negatives (Rey, Morris-Yates, & Stanislaw, 1992). Consistent with previous studies recommendations (Kaye & Darke, 2002; Topp & Mattick, 1997; Swift et al., 1998), we suggest that users of the SDS refer to our Table 1 to determine whether a different cut-off point might be more favourable. Whereas sensitivity and specificity should theoretically be stable characteristics of an instrument, regardless of the population being tested, predictive values are highly dependent on the prevalence of a disease found in the population tested (Rothman, 2002). The rarer the disease, the more likely it is that a negative test result will indicate no disease (high NPV), and the less likely it is that a positive test result will indicate that the disease is present (low PPV). Even if the sensitivity and the specificity of a test are high, the PPV will be low if the prevalence of the disease is low (Altman & Bland, 1994). The prevalence of cannabis dependence in our sample of cannabis users was low; hence, the positive predictive value was low (28.8 %), and the negative predictive value was high (99.0 %). S61

6 ORIGINALARBEIT S. Steiner et al. Severity of Dependence Skala Ideally, diagnostic tools should have both high sensitivity and high specificity. In epidemiologic research in which screening for cannabis dependence has implications for future preventive measures, an instrument with high sensitivity is preferable. High sensitivity of a screening instrument assures that vulnerable people just below the threshold of dependence will be detected. The results of a study conducted in outpatient treatment centres (Simon & Kraus, 2007) underscores the fact that people with cannabis-related problems who do not meet criteria for cannabis-related disorders might still require treatment or other support. Clients primary cannabis diagnoses provided by their therapists were re-evaluated using computerized clinical interviews (the CIDI). In about 25 % of the cases, patients did not meet the criteria for a cannabis-related disorder. The present results indicate that some cannabis users who do not meet diagnostic criteria for a cannabis disorder still have behavioural problems related to their cannabis use. Whereas 43 cases of cannabis dependence were diagnosed according to DSM-IV criteria, 147 were identified as dependent (including 104 false positives) using the SDS. Compared to the users who were correctly diagnosed as not cannabis dependent, those incorrectly identified as dependent with the SDS (false positives) had significantly more DSM-IV cannabis-dependence symptoms and more often diagnoses of DSM-IV cannabis abuse. Furthermore, they had used cannabis more frequently during the last 30 days and had used more other illegal drugs during the last 12 months than had the true negatives. Our results show that the SDS is highly sensitive. However, because of the low prevalence of cannabis dependence, the positive predictive value of the SDS is rather low. As a consequence, in epidemiological studies the SDS will overestimate the prevalence of cannabis dependence. In routine care settings, the SDS might be used to screen clients prior to using time-consuming diagnostic instruments such as the CIDI. The high NPV assures that 99 % of the cases identified as not dependent with the SDS will not show DSM-IV cannabis dependence. It is important to keep in mind that PPV and NPV strongly depend on the prevalence of a disease; they cannot necessarily be generalized to different populations or settings (Shapiro, 1999). However, PPVs and NPVs are more useful than sensitivity and specificity when the effectiveness of a screening instrument in the general population needs to be assessed (Shenassa, 2002). The prevalence rate of a disease in the target population can also indirectly influence the generalizability of a screening. Screening instruments developed in populations with low prevalence might be of limited value in populations in which a wider range of symptoms is manifest. Generalizability can be improved by taking relevant covariates (e.g. gender) into account when estimating the parameters of the test. As Greiner and Gardner (2000) recommended, stratum-specific estimates of sensitivity and specificity should be optional test parameters. Thus, in the current study specific cutoff points for subgroups were calculated. We found a different cut-off point for male cannabis users than for female users. Although the SDS does not use the DSM-IV physical symptoms tolerance and withdrawal, our analysis that excluded these criteria yielded a similar cut-off point. This study has some limitations. First, although the sample size (456 adult cannabis users) was relatively large, only 9.7 % (n = 43) of them met DSM-IV criteria for cannabis dependence. Second, the illegal nature of cannabis use is often associated with inaccurate or incorrect self-reports of use (Behrendt, Bühringer, Heinemann & Rommelspacher, 2006). Respondents tendency to give socially desirable answers is often assumed to be the motivation for deliberately underreporting or denying their use of a substance (Johnson & Fendrich, 2005). The reported prevalence rates might, therefore, be underestimated. Third, the eligibility of DSM-IV diagnoses as the»gold standard«for measuring cannabis dependence needs to be assessed. It has been noted that the application of the same threshold criterion of three positive symptoms as for opiate and alcohol dependence is questionable (Soellner, 2005; Budney, 2006), that diagnosis of cannabis dependence should focus on psychological dependence only (Soellner, 2005), and that incorporating a dimensional approach could enhance the utility of the DSM-IV (Helzer, Bucholz & Gossop, 2007). Although these authors argued for a reformulation of the DSM- IV cannabis dependence diagnosis, current evidence supports the present DSM-IV definition (Babor, 2006). In conclusion, the present study validated the utility of the German version of the Severity of Dependence Scale as a diagnostic measure of cannabis dependence, and an appropriate cut-off point was calculated. The SDS is brief, and it seems to be a suitable and reliable instrument that can be used for detecting cannabis dependence in the general population of adult cannabis users. Acknowledgements Funding of the Epidemiological Survey on Substance Abuse was provided by the German Federal Ministry of Health (BMG) (Grant No /32). The funding does not implicate any conditions. Declaration of possible conflicts of interest There were no conflicts of interest related to the preparation of this paper. References Adamson, S. J. & Sellman, J. D. (2003). A prototype screening instrument for cannabis use disorder: the Cannabis Use Disorders Identification Test (CUDIT) in an alcohol-dependent clinical sample. Drug and Alcohol Review, 22 (3), Alexander, D. E. (2003). A marijuana screening inventory (experimental version): description and preliminary psychometric properties. The American Journal of Drug and Alcohol Abuse, 29 (3), S62

7 S. Steiner et al. Severity of Dependence Scale RESEARCH REPORT Altman, D.G. & Bland, M. (1994). Diagnostic tests: predictive values. British Medical Journal, 309, 102. American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders (4 th ed.). Washington DC: American Psychiatric Association. Babor, T. (2006). The diagnosis of cannabis dependence. In R. Roffman & G. A. Marlatt (Hrsg.), Cannabis dependence (International Research Monographs in the Addictions, S ). Cambridge: Cambridge University Press. Behrendt, K., Bühringer, G., Heinemann, A. & Rommelspacher, H. (2006). Drogenabhängigkeit [Drug dependence] (Suchtmedizinische Reihe, Band 4). Hamm: Deutsche Hauptstelle für Suchtfragen e.v. Budney, A. J. (2006). Are specific dependence criteria necessary for different substances: how can research on cannabis inform this issue? Addiction, 101 (Suppl 1), Edwards, G., Arif, A. & Hadgson, R. (1981). Nomenclature and classification of drugand alcohol-related problems: a WHO memorandum. Bulletin of the World Health Organisation, 59 (2), European Monitoring Centre for Drugs and Drug Addiction (2004). An overview of cannabis potency in Europe (Insights No 6). Lisbon: European Monitoring Centre for Drugs and Drug Addiction. Fletcher, R. H., Fletcher, S. W. & Wagner, E. H. (1996). Clinical epidemiology. Baltimore: Lippincott, Williams and Wilkins. Gossop, M., Darke, S., Griffiths, P., Hando, J., Powis, B., Hall, W. & Strang, J. (1995). The Severity of Dependence Scale (SDS): psychometric properties of the SDS in English and Australian samples of heroin, cocaine and amphetamine users. Addiction, 90 (5), Greiner, M. & Gardner, I. A. (2000). Epidemiologic issues in the validation of veterinary diagnostic tests. Preventive Veterinary Medicine, 45 (1 2), Hides, L., Dawe, S., Young, R. M. & Kavanagh, D. J. (2007). The reliability and validity of the Severity of Dependence Scale for detecting cannabis dependence in psychosis. Addiction, 102 (1), Helzer, J.E., Bucholz, K.K. & Gossop, M. (2007). A dimensional option for the diagnosis of substance dependence in DSM-V. International Journal of Methods in Psychiatric Research, 16 (Suppl 1), S Johnson, T. & Fendrich, M. (2005). Modeling sources of self-report bias in a survey of drug use epidemiology. Annals of Epidemiology, 15 (5), Kaye, S. & Darke, S. (2002). Determining a diagnostic cut-off on the Severity of Dependence Scale (SDS) for cocaine dependence. Addiction, 97 (6), Kraus, L., Pfeiffer-Gerschel, T. & Pabst, A. (2008). Cannabis und andere illegale Drogen: Prävalent, Konsummuster und Trends. Ergebnisse des Epidemiologischen Suchtsurveys 2006 [Cannabis and other illegal drugs: prevalence, consumption patterns and trends. Results of the 2006 Epidemiological Survey of Substance Abuse.] SUCHT, 54 (Sonderheft 1), S16 S25. Kraus, L. & Baumeister, S. E. (2008). Studiendesign und Methodik des Epidemiologischen Suchtsurveys 2006 [Study design and methodology of the 2006 Epidemiologic Survey of Substance Abuse]. SUCHT, 54 (Sonderheft 1), S6 S15. Legleye, S., Karila, L., Beck, F. & Reynaud, M. (2007). Validation of the CAST, a general population Cannabis Abuse Screening Test. Journal of Substance Use, 12 (4), Lachner, G., Wittchen, H. U., Perkonigg, A., Holly, A., Schuster, P., Wunderlich, U., Turk, D., Garczynski, E. & Pfister, H. (1998). Structure, content and reliability of the Munich-Composite International Diagnostic Interview (M-CIDI) substance use sections. European Addiction Research, 4 (1 2), Martin, G., Copeland, J., Gates, P. & Gilmour, S. (2006). The Severity of Dependence Scale (SDS) in an adolescent population of cannabis users: reliability, validity and diagnostic cut-off. Drug and Alcohol Dependence, 83 (1), Miele, G. M., Carpenter, K. M., Smith, C. M., Trautman, K. D., Blaine, J. & Hasin, D. S. (2000a). Substance Dependence Severity Scale (SDSS): reliability and validity of a clinician-administered interview for DSM-IV substance use disorders. Drug and Alcohol Dependence, 59 (1), Miele, G. M., Carpenter, K. M., Smith, C. M., Dietz, T. K., Blaine, J. & Hasin, D. S. (2000b). Concurrent and predictive validity of the Substance Dependence Severity Scale (SDSS). Drug and Alcohol Dependence, 59 (1), Rey, J. M., Morris-Yates, A. & Stanislaw, H. (1992). Measuring the accuracy of diagnostic tests using receiver operating characteristics (ROC) analysis. International Journal of Methods in Psychatric Research, 2, Rothman, K. J. (2002). Epidemiology: an introduction. Oxford: Oxford University Press. Shapiro, D. E. (1999). The interpretation of diagnostic tests. Statistical Methods in Medical Research, 8 (2), Shenassa, E. D. (2002). Delivering the goods: the importance of screening accuracy for effective community intervention and prevention. Journal of Community Psychology, 30 (2), Simon, R. & Kraus, L. (2007). Validität von cannabisbezogenen Diagnosen in Suchtberatungsstellen [Validity of cannabisrelated diagnoses in addiction counselling centres] SUCHT, 53 (5), Soellner, R. (2005). Dependence on cannabis an ever lasting issue. Substance Use & Misuse, 40 (6), Stata Corp. (2005). Survey data reference manual, release 9. College Station, TX: Stata Press. Swift, W., Copeland, J. & Hall, W. (1998). Choosing a diagnostic cut-off for cannabis dependence. Addiction, 93 (11), Tabachnick, B. G. & Fidell, L. S. (2007). Using multivariate statistics (5 th ). Boston: Allyn and Bacon. Topp, L. & Mattick, R. P. (1997). Choosing a cut-off on the Severity of Dependence Scale (SDS) for amphetamine users. Addiction, 92 (7), West, R. & Gossop, M. (1994). Overview: a comparison of withdrawal symptoms from different drug classes. Addiction, 89 (11), Wittchen, H.-U., Beloch, E., Garczynski, E., Holly, A., Lachner, G., Perkonigg, A., Pfütze, E.-M., Schuster, P., Vodermaier, A., Vossen, A., Wunderlich, U. & Zieglgänsberger, S. (1995). Münchener Composite International Diagnostic Interview (M-CIDI), Paper-pencil 2.2, 2/95. München: Max-Planck-Institut für Psychiatrie, Klinisches Institut. World Health Organization (1990). Composite International Diagnostic Interview (CIDI). Geneva: World Health Organization. Youden, W. J. (1950). Index for rating diagnostic tests. Cancer, 3 (1), Korrespondenzadresse PD Dr. Ludwig Kraus IFT Institut für Therapieforschung Leiter des Fachbereichs Epidemiologische Forschung Parzivalstr München Tel Fax kraus@ift.de Eingereicht: Angenommen: S63