Mannheimer Craving Scale 1

The Mannheimer Craving Scale (MaCS): Psychometric properties in a non-clinical sample

and development of cut-off scores

[Die Mannheimer Craving Scale (MaCS): Psychometrische Eigenschaften in einer nicht-

klinischen Stichprobe und Entwicklung von Cut-Off Werten]

Adrian Meulea,*, Helmut Nakovicsb, & Andrea Küblera,c

a
Department of Psychology I, University of Würzburg, Marcusstr. 9-11, 97070 Würzburg,

Germany, adrian.meule@uni-wuerzburg.de, andrea.kuebler@uni-wuerzburg.de

b
Central Institute of Mental Health, Department of Addictive Behavior and Addiction

Medicine, J5, 68159 Mannheim, Germany, helmut.nakovics@zi-mannheim.de

c
Institute of Medical Psychology and Behavioural Neurobiology, University of Tübingen,

Gartenstr. 29, 72074 Tübingen, Germany

*Corresponding author:
Adrian Meule
Department of Psychology I, University of Würzburg
Marcusstr. 9-11
97070 Würzburg
Phone: +49 931 31 - 808 34
Fax: +49 931 31 - 8 24 24
E-mail address: adrian.meule@uni-wuerzburg.de
 Mannheimer Craving Scale 2

Abstract

Aims: The Mannheimer Craving Scale (MaCS) was developed by Nakovics and colleagues

(2009) for quantitative measurement of craving across different substances and to be suitable

for multiple substance use. The current study presents psychometric properties of the MaCS

as well as normative data and cut-off-scores. Method: An online survey was conducted in a

non-clinical sample (N = 616). Furthermore, a ROC-analysis including a sample of patients

(N = 264) with different substance use was calculated. Results: Factor structure, validity and

internal consistency could be replicated. As expected, retest-coefficients were higher for the

student sample than for patients. The MaCS discriminated between patients and control

participants with high sensitivity and specificity. We propose to interpret MaCS-scores

between 3 and 13 as risky and scores higher than 13 as pathological experiences of substance

craving. Conclusions: Psychometric properties of the MaCS could be replicated in a non-

clinical sample. High factorial stability and retest-reliability qualifies the MaCS for use in

longitudinal studies in which scores on the scale could be used to evaluate treatment success.

Keywords

Craving; substance abuse; OCDS; questionnaire; test validation

 Mannheimer Craving Scale 3

Abstract

Fragestellung: Die Mannheimer Craving Scale (MaCS) wurde von Nakovics und

Mitarbeitern (2009) zur Erfassung von Craving nach unterschiedlichen Suchtmitteln

entwickelt und eignet sich bei multiplem Substanzgebrauch. Die gegenwärtige Studie stellt

die psychometrischen Eigenschaften der MaCS sowie Norm- und Cut-off-Werte dar.

Methodik: Es wurde eine Onlinebefragung in einer nicht-klinischen Stichprobe

durchgeführt (N = 616). Darüber hinaus wurde eine ROC-Analyse unter Einbezug einer

Patientenstichprobe (N = 264) mit multiplem Substanzgebrauch berechnet. Ergebnisse:

Faktorstruktur, Validität und interne Konsistenz des Fragebogens konnten repliziert werden.

Die Retest-Reliabilität war erwartungsgemäß höher als in der Patientenstichprobe. Die

MaCS konnte mit hoher Sensitivität und Spezifität zwischen Patienten und gesunden

Probanden diskriminieren. MaCS-Werte zwischen 3 und 13 können als riskant interpretiert

werden, während Werte über 13 das Vorliegen pathologischen Substanzcravings anzeigen.

Schlussfolgerungen: Die psychometrischen Eigenschaften der MaCS konnten in einer

nicht-klinischen Stichprobe repliziert werden. Die MaCS eignet sich aufgrund hoher

faktorieller Stabilität und Retest-Reliabilität für den Gebrauch in Langzeitstudien zur

Evaluation von Therapieerfolg.

Schlüsselwörter

Craving; Substanzmissbrauch; OCDS; Fragebogen; Testvalidierung

 Mannheimer Craving Scale 4

Introduction

Craving is an urgent desire, longing or yearning for a substance, including food,

alcohol, tobacco, and other drugs [1]. Craving, and both its emotional and compulsive

aspects, is a substantial phenomenon in addiction. For example, criteria for the diagnosis of

substance dependence in the International Statistical Classification of Diseases and Related

Health Problems (ICD-10) include “a strong desire or sense of compulsion to take the

substance“ [2]. However, such cravings can even occur after successful treatment and are

implicated to contribute to relapse [3]. From a learning perspective, craving can be seen as a

conditioned reaction in response to internal or external cues which have been previously

associated with the consumption of the substance. Those cues can either cause withdrawal

symptoms, thereby triggering craving associated with relief of withdrawal symptoms, or

elicit craving, which is associated with positive, rewarding effects of drug use (reward

craving) [4]. Scores on measures of craving have been found to predict relapse, e.g. in

alcohol [5],[6] or tobacco addiction [7]. In smokers, craving not only peaks on the day of

relapse, it also increases steeply on days before relapse [7]. Therefore, assessment of

cravings during and after therapy may constitute an important aspect for evaluating treatment

success and risk of relapse.

Craving can be measured by various approaches. However, attempts to measure

craving objectively, e.g. based on physiological data, have been criticized to be unspecific

[8]. Until now, “subjective self-report seems the only viable assessment modality” [8] (p.

S172). One such approach that stresses obsessive and compulsive aspects of cravings is the

Obsessive-Compulsive Drinking Scale (OCDS) [9]. This 14-item-scale measures alcohol

cravings with regard to its cognitive and behavioral aspects. Obsessions refer to
 Mannheimer Craving Scale 5

preoccupation and thoughts about alcohol use while compulsions describe the inevitable

drinking behaviors [9].

Several studies found and suggested different factor solutions (two, three or four

factors; e.g. [6],[10]). In addition to obsessive thoughts, the OCDS seems to measure an

interference of craving with social and work functioning as well as alcohol consumption

quantity [6],[11],[12]. Here, interference was associated to appetitive reactions to alcohol

cues and drinking in positive situations while obsessive craving was associated to

withdrawal-like symptoms [13]. An examination of the different factor solutions under

consideration of the stability criterion showed however, that the two-factor solution (without

the quantity items) is stable and thus is recommended for calculation of OCDS-scores [14].

As the OCDS is restricted to alcohol craving, further versions have been provided

extending the scale to other substances, e.g. heroin [15], tobacco [16], and other drugs [17].

Other scales for the assessment of substance cravings have also been developed, e.g. the

Questionnaire on Smoking Urges [18], the Cocaine Craving Questionnaire [19], the Alcohol

Urge Questionnaire [20], the Marijuana Craving Questionnaire [21], the Tobacco Craving

Questionnaire [22], the Benzodiazepine Craving Questionnaire [23], the Alcohol Craving

Questionnaire [24], and others.

However, all of these scales are restricted to a specific substance. In clinical practice,

however, multiple substance use is common (e.g. [25]). Moreover, craving has been found to

be similar across substances. Activation of reward pathways and the insula was observed,

regardless of whether drug or food craving was experienced [4],[26],[27],[28],[29].

Recently, the Mannheimer Craving Scale (MaCS) was constructed to take these

findings into account [30]. This scale was derived from the German OCDS (OCDS-G; [31]),
 Mannheimer Craving Scale 6

but reformulated to be applicable to every addictive substance (e.g. “How strong is your urge

to take addictive substances?”). Another major difference between the two questionnaires is

that the MaCS does not include items that assess consumption quantity. Excluding such

items has previously been found to result in a stable and reliable factor structure [14].

Psychometric properties of the MaCS have only been tested in a sample of substance

dependence patients. Therefore, the present study provides a validation of the scale in a non-

clinical sample. Furthermore, validity of the MaCS was also tested by comparing our data to

the sample of patients investigated by Nakovics and colleagues [30]. Specifically, we

calculated specificity (SP) and sensitivity (SE) of the scale by including both samples.

Method

Procedure

The MaCS was part of an online study, of which the main purpose was the validation

of a food craving questionnaire [32]. Student councils of several German universities were

contacted via e-mail. Then, the internet address of the online survey was sent via the student

councils’ mailing lists. Five × 50.- Euro were raffled among participants who completed the

entire set of questions. Completion of the entire study lasted approximately 25 minutes.

Every question required a response in order to continue. Multiple entries were controlled for

by checking IP-addresses. Survey period lasted four weeks. Participants who entered their e-

mail-address and agreed to be contacted again were asked to fill out a retest three weeks after

the mid-point of the study period (i.e. one week after closure of the website). Participants

entered individual codes consisting of digits and letters to relate anonymously data from both

measurements.

Participants
 Mannheimer Craving Scale 7

The study-website was visited 1615 times. The entire survey was completed by N =

617 participants (38.2 %). Data from one participant were excluded from further analyses

because of implausible statements. The majority of participants were women (75.8 %, n =

467). Participants had a mean age of M = 24.5 years (SD ± 4.0). Almost all participants were

students (89.0 %) and had German citizenship (95.5 %). The percentage of participants who

reported to be smokers was 19.6% (n = 121). The retest was completed by n = 237

participants. However, data of only n = 197 participants could be used because individual

codes of n = 40 participants did not match to the ones specified in the primary survey.

The sample of patients (N = 292) was recruited from hospital wards specialized in

alcohol dependence (n = 105; 80.9 % male; age: M = 44.8 years, SD ± 9.4), drug dependence

(n = 110; 73.6 % male; age: M = 27.6 years, SD ± 5.7) or alcohol- and drug dependence (n =

77; 70.1 % male; age: M = 44.8 years, SD ± 9.4). MaCS-scores prior to the first treatment

session were available for n = 264 patients. Substance use in this sample included alcohol,

heroin, tobacco, benzodiazepines, and cannabis [30].

Instruments

The MaCS is an instrument to measure craving across multiple substances and

consists of 12 items. These include obsessive-cognitive, compulsive-motivational and

behavioral aspects of substance use. Participants are instructed that the questions refer to any

addictive substance (e.g. alcohol, drugs, or medication). All questions refer to cravings

experienced within the past seven days. Every item comprises five response categories

(coded 0-4). Here, interviewees ought to choose the most appropriate statement. All items

have the same polarity and higher scores indicate stronger intensity of craving. Four

additional items assess intensity and frequency of craving on visual analog scales (0-100)

and days of abstinence. A one-factor solution proved to be stable (as defined by factorial
 Mannheimer Craving Scale 8

invariance) and internal consistencies were α = .87 - .93 in a sample of patients with

different substance dependencies [30].

Statistical analyses

Relationships between MaCS-scores and sample characteristics were tested by

Spearman-correlations (age) or t-tests (gender, smoking). A principial component analysis

(PCA) was conducted to explore the factor structure of the MaCS. Using the Kaiser-criterion

[33] can lead to an overestimation of the number of factors [34], but a parallel analysis [35]

can result in an underestimation when the first eigenvalue is large [36]. Therefore, we

considered both methods for determination of the number of factors for extraction.

Additionally, we adapted the procedure used by Nakovics and colleagues [30] and

performed an orthogonal rotation (Varimax). This rotation method was chosen to compare

factor loadings of our data with factor loadings of the clinical sample. Overall solution

congruence and single-item congruence coefficients were calculated to evaluate similarity of

factor loadings between the two samples. To do so, the congruence coefficient, also referred

to as raw vector product coefficient, was computed with Orthosim-2 [37]. Stability of the

factor solutions was calculated by Pearson-correlations of factor loadings between the two

assessments (cf. [14],[30]).

To further corroborate factor solutions suggested by exploratory factor analysis, we

conducted confirmatory factor analysis (CFA) with STATISTICA 10.0 (StatSoft, Inc., Tulsa,

USA). Parameters were estimated with the asymptotically distribution free gramian (ADFG)

method because this method does not require normally distributed data. Model fit was

evaluated based on the root-mean-square error of approximation (RMSEA).

 Mannheimer Craving Scale 9

Item analysis was performed by calculating item difficulties (means), standard

deviations, and item-total-correlations (part-whole corrected). Internal consistency

(Cronbach’s α) and retest-coefficients (Spearman’s ρ) were computed to evaluate reliability

of the MaCS. Construct validity was tested by correlations (Spearman’s ρ) between total

MaCS-scores and the four additional items. Normative data were provided using percentile

ranks and stanine-scores. Furthermore, a cut-off-score was determined using receiver

operating characteristic (ROC) analysis by including total MaCS-scores of patients with

different substance dependencies prior to treatment (see description of this sample above or

[30]). Specifically, sensitivity and specificity of MaCS-scores were calculated for

discriminating between the samples (patient sample vs. student sample). In addition,

likelihood ratios, i.e. ratio between the probability of a positive or negative test result given

the presence of substance dependence and the probability of a positive or negative test result

given the absence of substance dependence, were determined (Positive likelihood ratio

(+LR) = true positive rate / false positive rate; negative likelihood ratio (-LR) = false

negative rate / true negative rate).

Results

MaCS-scores in relation to sample characteristics

MaCS-scores were independent of age (rρ = .07, ns) and did not differ between men

(M = 4.85, SD ± 5.62) and women (M = 4.04, SD ± 5.88, t(614) = 1.47, ns). Smokers (M =

9.98, SD ± 6.71) had a higher MaCS-score than non-smokers (M = 2.83, SD ± 4.61, t(148.82) =

11.10, p < .001).

Factor structure
 Mannheimer Craving Scale 10

The Kaiser-Meyer-Olkin-coefficient (KMO = .92) and the Bartlett-Test (χ²(66) =

4261.03, p < .001) indicated that data were appropriate for exploratory factor analysis. PCA

of the online-sample yielded two factors with an eigenvalue ≥ 1, namely 6.3 and 1.0 (Kaiser-

criterion; [33]). However, a parallel analysis [35] suggested a one-factor solution: only one

factor had an eigenvalue that exceeded eigenvalues of the simulated data. Therefore, factor

loadings of a one- and a two-factor solution are reported (Table 1). The one factor solution

explained 52.8% of the variance. Factor loadings of the single items ranged between .53 and

.84 (Table 1). Further, two factors were extracted and orthogonally rotated (Varimax). These

two factors explained 61.2% of the variance. Items 3, 7 and 8 had high factor loadings on the

second factor (Table 1). Those items assess the influence of obsessions and substance use on

occupational performance and social behavior. We therefore named this factor interference

whereas the subscale consisting of the remaining items was termed craving. An overall

congruence coefficient of .98 indicated identity between factor loadings of the patient and

non-clinical sample [37]. Single-item congruence coefficients ranged between .92 and 1.0

(Table 1).

In a subsequent CFA, we evaluated a one- and a two-factorial model as suggested by

exploratory factor analysis. The one-factorial model (χ²(54) = 155.31, p < .001, RMSEA =

.06) and the two-factorial model (χ²(53) = 150.30, p < .001, RMSEA = .06) had similar

adequate model fit.1

>Insert Table 1 here<

Stability

An identical factor analysis was conducted for the retest-data. Factor loadings of the

one-factor solution ranged between .51 and .84. For the rotated two-factor solution, factor
 Mannheimer Craving Scale 11

loadings of the first factor ranged between .10-.85 and of the second factor between 09-.91.

Correlating factor loadings between the first assessment and the retest resulted in high

stability coefficients for the one-factor- (r = .86, p < .001) and the two-factor solution (r1 =

.95, p < .001; r2 = .91, p < .001).

Item analysis

Item difficulties were high which can be seen in low item means (Table 1). Item-

total-correlations (part-whole corrected) were high (Table 1).

Reliability

Overall internal consistency was high (α = .89), but somewhat lower for the two

subscales (interference: α = .76, craving: α = .88). Retest-reliability for total MaCS-scores

was rρ = .76 (p < .001), while retest-coefficients for the four additional questions ranged

between rρ = .50 and rρ = .68 (all p’s < .001).

Validity

High positive correlations were identified between MaCS-scores and craving as

assessed with visual analog scales (average craving intensity: r = .70, maximum craving

intensity: r = .69, craving frequency: r = .66, all p’s < .001). Visual analog scales were also

highly intercorrelated (r = .73-.83, p < .001). There was a moderately negative correlation

between days of abstinence from a substance with the MaCS (r = -.27, p < .001). Visual

analog scales were only small or not correlated with days of abstinence (average craving

intensity: r = -.08, p < .05, maximum craving intensity: r = -.11, p < .01, craving frequency:

r = -.01, ns).

>Insert Table 2 here<

 Mannheimer Craving Scale 12

Normative data and cut-offs

A Kolmogorov-Smirnov-test indicated that MaCS-scores were not normally

distributed (K-S-Z = .23, p < .001). Therefore, stanine-scores were derived from percentile

ranks to provide normative data (cf. [38], p. 264). Stanine-scores have a mean of M = 5 and a

standard deviation of SD ± 2. Hence, stanine-scores of 8 or higher indicate elevated test-

scores (≥ 90th percentile). This equals a MaCS-score of 13 or more (Table 2). MaCS-scores

of the non-clinical sample (M = 4.24, SD ± 5.83) were significantly lower than those of the

clinical sample (M = 15.03, SD ± 8.91, t(363.04) = -18.09, p < 001)2. ROC-analysis (Figure 1)

yielded highest sensitivity (80%) with highest specificity (75%) at a MaCS cut-off-score of

>6 (+LR = 3.23, -LR = 0.26; Table 3). In clinical practice, high sensitivity may be preferred

over high specificity to ensure that every vulnerable patient is detected. In this case, a

MaCS-score of >3 would be recommendable (SE = 89 %), at the cost of specificity (62 %;

+LR = 2.32, -LR = 0.17). Definite clinical relevant substance craving is present when a

person has a MaCS-score of >12 (SP = 90 %, SE = 57%, +LR = 5.47, -LR = 0.48). Area

under the curve was 86% (Figure 1).

>Insert Figure 1 here<

Discussion

Although there are different conceptualizations of craving [39], there is a consensus

that the term craving applies to a variety of substances [1]. Increasing evidence suggests that

the neurobiological basis of the craving experience may be similar across substances, e.g.

alcohol, tobacco, and cocaine [40]. The assessment of the subjective craving experience,

however, is usually constricted to a specific substance. The MaCS, which is the only

available measure to assess craving across substances, showed good psychometric properties
 Mannheimer Craving Scale 13

in patients with multiple substance use and, therefore, support the idea of measuring craving

independent of a specific substance.

Substance use is also prevalent in the general population. For instance, it has been

found that the 12-month-prevalence for alcohol use is 87% in Germany [41]. Furthermore,

some 60% of Germans have ever used tobacco [42]. Hence, control participants in our study

also experienced some substance craving, but scores were significantly lower than in

substance use patients. Accordingly, addicted patients could be discriminated from control

participants with high sensitivity and specificity. In addition, the present study demonstrated

that psychometric properties of the MaCS could be replicated in a non-clinical sample, but

stability and retest-reliability was higher compared to clinical samples [30].

Exploratory factor analysis suggested a one- or two-factor solution. Furthermore, a

subsequent CFA showed that both factor structures had similar and adequate model fit. A

one-dimensional factor structure was supported by parallel analysis and high internal

consistency and stability. Extraction of two factors resulted in similar factor loadings that

were reported in a sample of patients [30] and were found to measure either an impairment

of daily life (which was termed interference) or craving. Furthermore, these two subscales

also had satisfying internal consistency and stability was comparable to the one-factor

solution, but only in the non-clinical sample. It seems that the factor structure of the MaCS is

more differentiated in a non-clinical sample while a one-factor structure was found to be

stable in patients [30] and also in the non-clinical sample in this study. Therefore, we suggest

the use of total MaCS-scores, particularly when the MaCS is applied to patients or is

compared between clinical and non-clinical samples. Future research in healthy participants

might also consider the two subscales.

 Mannheimer Craving Scale 14

Retest-reliability of the MaCS was very high after three weeks (rρ = .76). Obviously,

retest-coefficients in the patient sample were found to be lower because substance cravings

decreased during treatment [30]. Notably, retest-coefficients of the MaCS were also higher

than for the visual analog scales. These findings support the use of the MaCS in longitudinal

studies because stable scores are produced in control participants but it is also sensitive to

treatment changes.

Validity of the MaCS could be seen in high correlations with average and maximum

craving intensity as well as craving frequency using visual analog scales. Days of abstinence

were inversely related to substance cravings. Moreover, smokers were found to have

significantly higher MaCS-scores than non-smokers. Therefore, the MaCS is a valid measure

of substance craving in a non-clinical sample. It should be noted though, that item

difficulties were very high, i.e. subjects chose the minimum item response category very

often. Hence, while the MaCS is able to discriminate between patients and control

participants, variability of MaCS-scores within control participants is low.

The MaCS has a high specificity at scores >12 which indicates intense substance

craving that exceeds a normal level. This was also corroborated using more than two

standard deviations from mean stanine score as criterion. When this score is used as a cut-

off, the probability of having a substance dependence is increased as indicated by a high

positive likelihood ratio. However, a high sensitivity may be preferred in clinical practice to

ensure that every vulnerable patient is detected. Accordingly, patients with a MaCS-score of

>3 should be further interviewed to decide if they need to be closely followed up after

therapy. Patients with different substance dependencies have been found to score between

14.7 and 17 on the MaCS at the beginning of treatment and between 14.1 and 10.1 two

weeks later [30]. Taken these findings together, we conclude that a MaCS-score ≤3 indicates
 Mannheimer Craving Scale 15

normal intensity of substance cravings. Accordingly, the probability of having a substance

dependence is decreased as indicated by a low negative likelihood ratio. A score between 3

and 13 may suggest an elevated, but not yet pathological level of substance cravings. A score

of 13 or higher indicates dangerous intensities of cravings. If a patient scores in this range

after therapy, risk of relapse may be significantly increased. Future longitudinal studies are

needed to validate the predictive value of these cut-off-scores. For the time being we

recommend that patients who score above 3 on the MaCS post therapy should be followed

up closely and receive specific counseling in what steps to take if craving becomes

irresistible. For patients with scores of 13 or above continued therapeutic intervention may

be recommendable.

Limitations

In the current study, we directly compared data from a non-clinical sample with a

sample of patients. Firstly, it has to be noted that the methodology used in these two studies

differed which could have influenced self-ratings. We conducted an online survey whereas

Nakovics and colleagues [30] handed out paper-and-pencil versions of the MaCS. However,

there is evidence that questionnaire scores, e.g. in alcohol measures, assessed with internet-

based surveys do not differ from traditional assessment methods [43]. Also, other

psychological constructs can be assessed via the Internet as reliable and valid as with paper-

and-pencil administration [44-46]. Web-based surveys are particularly suitable in college

populations and advantageous for the assessment of sensitive questions, e.g., about substance

use [47]. Furthermore, we found similar factor loadings, indicated by high congruence

coefficients, as in the study of Nakovics and colleagues [30] which does support

comparability of the studies. Secondly, we did not assess actual substance use but smoking.

Although smokers had higher MaCS-scores, we cannot infer how other substance use is
 Mannheimer Craving Scale 16

related to the MaCS. Due to this missing information, we were also not able to identify

individuals with substance use disorders in the student sample. However, we may speculate

that truly substance dependent subjects (except smokers) are less likely to participate in such

a survey due to lower overall functioning and low prevalence among students. Although

substance use patients had higher MaCS-scores than smokers, discrimination between those

groups by means of MaCS-scores was reduced. There is clearly a need for future studies

investigating groups of, e.g., smokers or social drinkers who do not meet criteria for

substance use disorder and evaluate if it is feasible to discriminate those groups from

substance use patients using the MaCS. Thirdly, sample characteristics differed between

studies and we could not trace if there was a selection bias. Specifically, the majority of our

sample consisted of young, collegiate women. The patient sample comprised more men at a

higher age. However, possible gender differences can be ruled out because MaCS-scores did

not differ between men and women in our sample. Nonetheless, an underestimation of

MaCS-scores, compared to a representative sample that would include people with higher

age and lower social status, is quite possible. Therefore, as already denoted above, a cut-off

of 3 may result in false positive classification. Instead, a MaCS-score of 13 could be used as

critical value to evaluate clinical relevance of substance cravings.

 Mannheimer Craving Scale 17

Footnote

1
CFA was also run separately for males and females. In each group, model fit was similar for

the one-factorial model (females: χ²(54) = 136.28, p < .001, RMSEA = .06; males: χ²(54) =

125.11, p < .001, RMSEA = .08) and the two-factorial model (females: χ²(53) = 131.93, p <

.001, RMSEA = .06; males: χ²(53) = 114.36, p < .001, RMSEA = .08). Note that model fit in

males was still in an acceptable range considering the smaller sample size ([38], p. 425).

2
As smokers had higher MaCS-scores than non-smokers, we also tested if MaCS-scores

would differentiate between smokers in the student sample and substance use patients.

Patients had higher MaCS-scores than smokers (t(383) = 5.55, p < .001). Yet, discrimination

between groups was reduced compared with using the total sample, but was still higher than

what would be expected by chance (Area under curve = 66%, p < .001).

Acknowledgements

Funding for this study was provided by a grant of the research training group 1253/2

which is supported by the DFG by federal and Länder funds. DFG had no role in the study

design, collection, analysis or interpretation of the data, writing the manuscript, or the

decision to submit the paper for publication. The authors thank A. Lutz for aiding in data

collection. K. Roeser and B. Schwerdtle are also gratefully acknowledged for helping in data

analysis and -presentation.

 Mannheimer Craving Scale 18

References

[1] Hormes JM, Rozin P. Does "craving" carve nature at the joints? Absence of a synonym

for craving in many languages. Addict Behav. 2010;35:459-63.

[2] World Health Organization. International statistical classification of diseases and related

health problems - 10th revision. Geneva: World Health Organization; 2007.

[3] Sinha R, Li CSR. Imaging stress- and cue-induced drug and alcohol craving: association

with relapse and clinical implications. Drug Alcohol Rev. 2007;26:25-31.

[4] Heinz A, Beck A, Grusser SM, Grace AA, Wrase J. Identifying the neural circuitry of

alcohol craving and relapse vulnerability. Addict Biol. 2009;14:108-18.

[5] Bottlender M, Soyka M. Impact of craving on alcohol relapse during, and 12 months

following, outpatient treatment. Alcohol Alcoholism. 2004;39:357-61.

[6] Roberts JS, Anton RF, Latham PK, Moak DH. Factor structure and predictive validity of

the Obsessive Compulsive Drinking Scale. Alcohol Clin Exp Res. 1999;23:1484-91.

[7] Allen SS, Bade T, Hatsukami D, Center B. Craving, withdrawal, and smoking urges on

days immediately prior to smoking relapse. Nicotine Tob Res. 2008;10:35-45.

[8] Shiffman S. Comments on craving. Addiction. 2000;95:S171-5.

[9] Anton RF, Moak DH, Latham P. The Obsessive-Compulsive Drinking Scale - a self-

rated instrument for the quantification of thoughts about alcohol and drinking

behavior. Alcohol Clin Exp Res. 1995;19:92-9.

[10] Schippers GM, DeJong CAJ, Lehert P, Potgieter A, Deckers F, Casselman J, et al. The

Obsessive Compulsive Drinking Scale: Translation into Dutch and possible

modifications. Eur Addict Res. 1997;3:116-22.

[11] Bohn MJ, Barton BA, Barron KE. Psychometric properties and validity of the

Obsessive-Compulsive Drinking Scale. Alcohol Clin Exp Res. 1996;20:817-23.

 Mannheimer Craving Scale 19

[12] Kranzler HR, Mulgrew CL, Modesto-Lowe V, Burleson JA. Validity of the obsessive

compulsive drinking scale (OCDS): Does craving predict drinking behavior? Alcohol

Clin Exp Res. 1999;23:108-14.

[13] Heinz A, Lober S, Georgi A, Wrase J, Hermann D, Rey ER, et al. Reward craving and

withdrawal relief craving: Assessment of different motivational pathways to alcohol

intake. Alcohol Alcoholism. 2003;38:35-9.

[14] Nakovics H, Diehl A, Croissant B, Mann K. Modifications of the Obsessive

Compulsive Drinking Scale (OCDS-G) for use in longitudinal studies. Addict Behav.

2008;33:1276-81.

[15] Franken IHA, Hendriks VM, van den Brink W. Initial validation of two opiate craving

questionnaires - The Obsessive Compulsive Drug Use Scale and the Desires for Drug

Questionnaire. Addict Behav. 2002;27:675-85.

[16] Hitsman B, Shen BJ, Cohen RA, Morissette SB, Drobes DJ, Spring B, et al. Measuring

smoking-related preoccupation and compulsive drive: Evaluation of the obsessive

compulsive smoking scale. Psychopharmacology. 2010;211:377-87.

[17] Morgan TJ, Morgenstern J, Blanchard KA, Labouvie E, Bux DA. Development of the

OCDS-revised: A measure of alcohol and drug urges with outpatient substance abuse

clients. Psychol Addict Behav. 2004;18:316-21.

[18] Tiffany ST, Drobes DJ. The development and initial validation of a questionnaire on

smoking urges. Brit J Addict. 1991;86:1467-76.

[19] Tiffany ST, Singleton E, Haertzen CA, Henningfield JE. The development of a cocaine

craving questionnaire. Drug Alcohol Depen. 1993;34:19-28.

[20] Bohn MJ, Krahn DD, Staehler BA. Development and initial validation of a measure of

drinking urges in abstinent alcoholics. Alcohol Clin Exp Res. 1995;19:600-6.

 Mannheimer Craving Scale 20

[21] Heishman SJ, Singleton EG, Liguori A. Marijuana Craving Questionnaire:

Development and initial validation of a self-report instrument. Addiction.

2001;96:1023-34.

[22] Heishman SJ, Singleton EG, Moolchan ET. Tobacco Craving Questionnaire: Reliability

and validity of a new multifactorial instrument. Nicotine Tob Res. 2003;5:645-54.

[23] Mol AJJ, Voshaar RCO, Gorgels WJMJ, Breteler MHM, van Balkom AJLM, van de

Lisdonk EH, et al. Development and psychometric evaluation of the Benzodiazepine

Craving Questionnaire. Addiction. 2003;98:1143-52.

[24] Raabe A, Grusser SM, Wessa M, Podschus J, Flor H. The assessment of craving:

Psychometric properties, factor structure and a revised version of the Alcohol

Craving Questionnaire (ACQ). Addiction. 2005;100:227-34.

[25] Diehl A, Scherbaum N. [Nicotine dependence as comorbidity of alcohol dependence -

Epidemiology, etiology and therapy]. Fortschr Neurol Psyc. 2008;76:14-20.

[26] Filbey FM, Schacht JP, Myers US, Chavez RS, Hutchison, KE. Marijuana craving in

the brain. PNAS. 2009;106:13016-21.

[27] Garavan H. Insula and drug cravings. Brain Struct Funct. 2010;214:593-601.

[28] Naqvi NH, Bechara A. The hidden island of addiction: The insula. Trends Neurosci.

2009;32:56-67.

[29] Pelchat ML, Johnson A, Chan R, Valdez J, Ragland JD. Images of desire: Food-craving

activation during fMRI. NeuroImage. 2004;23:1486–93.

[30] Nakovics H, Diehl A, Geiselhart H, Mann K. [Development and validation of an overall

instrument to measure craving across multiple substances: The Mannheimer Craving

Scale (MaCS)]. Psychiat Prax. 2009;36:72-8.

[31] Mann K, Ackermann K. [The OCDS-G: Psychometric characteristics of the German

version of the Obsessive Compulsive Drinking Scale]. Sucht. 2000;46:90-100.

 Mannheimer Craving Scale 21

[32] Meule A, Lutz A, Vögele C, Kübler A. Food cravings discriminate differentially

between successful and unsuccessful dieters and non-dieters. Validation of the Food

Craving Questionnaires in German. Appetite. 2012;58:88-97.

[33] Kaiser HF. The application of electronic computers to factor analysis. Educ Psychol

Meas. 1960;20:141-51.

[34] Hayton JC, Allen DG, Scarpello V. Factor retention decisions in exploratory factor

analysis: A tutorial on parallel analysis. Organ Res Meth. 2004;7:191-205.

[35] Horn JL. A rationale and test for the number of factors in factor analysis.

Psychometrika. 1965;30:179-85.

[36] Turner NE. The effect of common variance and structure pattern on random data

eigenvalues: Implications for the accuracy of parallel analysis. Educ Psychol Meas.

1998;58:541-68.

[37] Barrett PT. Orthosim-2. [Computer software]. 2006; available from

http://www.pbarrett.net/orthosim2.htm

[38] Bühner M. Einführung in die Test- und Fragebogenkonstruktion. 3rd ed. München:

Pearson Studium; 2011.

[39] Sayette MA, Shiffman S, Tiffany ST, Niaura RS, Martin CS, Shadel WG. The

measurement of drug craving. Addiction. 2000;95:S189-210.

[40] Kühn S, Gallinat J. Common biology of craving across legal and illegal drugs - a

quantitative meta-analysis of cue-reactivity brain response. Eur J Neurosci.

2011;33:1318-26.

[41] Kraus L, Bloomfield K, Augustin R, Reese A. Prevalence of alcohol use and the

association between onset of use and alcohol-related problems in a general

population sample in Germany. Addiction. 2000;95:1389-401.

 Mannheimer Craving Scale 22

[42] John U, Meyer C, Hapke U, Rumpf HR, Schumann A. Nicotine dependence, quit

attempts, and quitting among smokers in a regional population sample from a country

with a high prevalence of tobacco smoking. Prev Med. 2004;38:350-8.

[43] Miller ET, Neal DJ, Roberts LJ, Baer JS, Cressler SO, Metrik J, et al. Test-retest

reliability of alcohol measures: Is there a difference between Internet-based

assessment and traditional methods? Psychol Addict Behav. 2002;16:56-63.

[44] Campos JADB, Zucoloto ML, Bonafe FSS, Jordani PC, Maroco J. Reliability and

validity of self-reported burnout in college students: A cross randomized comparison

of paper-and-pencil vs. online administration. Comput Hum Behav. 2011;27:1875-

83.

[45] Naus MJ, Philipp LM, Samsi M. From paper to pixels: A comparison of paper and

computer formats in psychological assessment. Comput Hum Behav. 2009;25:1-7.

[46] Vallejo MA, Mananes G, Comeche MI, Diaz MI. Comparison between administration

via Internet and paper-and-pencil administration of two clinical instruments: SCL-90-

R and GHQ-128. J Behav Ther Exp Psy. 2008;39:201-8.

[47] Kays K, Gathercoal K, Buhrow W. Does survey format influence self-disclosure on

sensitive question items? Comput Hum Behav. 2012;28:251-6.

 Mannheimer Craving Scale 23

Table 1

Item statistics and factor loadings of the MaCS

Item M SD ritc Factor loadings Factor loadings after rotation Congruence

(one-factor solution) Factor 1 Factor 2
1. How much of your time when you are not taking addictive .44 .67 .77 .84 .80 .31 .99
substances is occupied by ideas, thoughts, impulses, or images related
to it? [Wenn Sie keine Suchtmittel nehmen, wie viel Ihrer Zeit wird
dann von Vorstellungen, Gedanken, Impulsen oder Bildern
beansprucht, die etwas damit zu tun haben?]
2. How frequently do these thoughts occur or ideas occur? [Wie häufig .43 .66 .75 .82 .81 .27 1.0
treten diese Gedanken oder Vorstellungen auf?]
3. How much do these ideas, thoughts, impulses, or images interfere .18 .47 .63 .71 .41 .65 .97
with your social or work functioning? Is there anything you do not or
cannot do because of them? [Wie stark werden Ihre berufliche
Tätigkeit oder Ihr soziales Verhalten von diesen Vorstellungen,
Gedanken, Impulsen oder Bildern beeinflusst? Gibt es etwas, was sie
deswegen nicht tun oder nicht können?]
4. How much distress or disturbance do these ideas, thoughts, .29 .58 .72 .79 .68 .41 1.0
impulses, or images cause you when you are not taking an addictive
 Mannheimer Craving Scale 24

substance? [Wenn Sie kein Suchtmittel nehmen, wie sehr leiden Sie
dann unter den Vorstellungen, Gedanken, Impulsen oder Bildern, die
damit zu tun haben bzw. wie sehr werden Sie dadurch gestört?]
5. How much of an effort do you make to resist these thoughts or try .40 1.0 .48 .53 .52 .18 .95
to disregard or turn your attention away from these thoughts when you
are not taking an addictive substance? [Wenn Sie kein Suchtmittel
nehmen, wie sehr bemühen Sie sich dann, diesen Gedanken zu
widerstehen, Sie nicht zu beachten oder Ihre Aufmerksamkeit auf
etwas Anderes zu lenken?]
6. How successful are you in stopping or diverting these thoughts .35 .67 .70 .77 .69 .37 1.0
when you are not taking an addictive substance? [Wenn Sie kein
Suchtmittel nehmen, wie erfolgreich können Sie dann diese Gedanken
beenden oder sie zerstreuen?]
7. How much does taking addictive substances interfere with your .12 .36 .55 .63 .18 .84 .99
work functioning? Is there anything you do not or cannot do because
of it? [Wie stark wird Ihre berufliche Tätigkeit durch die Einnahme
von Suchtmitteln beeinflusst? Gibt es etwas, was Sie deswegen nicht
tun oder nicht können?]
8. How much does taking addictive substances interfere with your .23 .53 .56 .65 .23 .81 1.0
social functioning? Is there anything you do not or cannot do because
 Mannheimer Craving Scale 25

of it? [Wie stark wird Ihr soziales Verhalten durch die Einnahme von
Suchtmitteln beeinflusst? Gibt es etwas, was Sie deswegen nicht tun
oder nicht können?]
9. If you were prevented from taking an addictive substance when you .36 .71 .59 .67 .51 .45 .95
desired one, how anxious or upset would you become? [Wenn Sie ein
Suchtmittel nehmen möchten, aber daran gehindert wären, wie
ängstlich oder ungehalten würden Sie dann werden?]
10. How much of an effort do you make to resist consumption of .71 1.3 .55 .60 .69 .06 .98
addictive substances? [Wie sehr bemühen Sie sich der Einnahme von
Suchtmittel zu widerstehen?]
11. How strong is the drive to consume addictive substances? [Wie .40 .61 .77 .83 .74 .38 .92
stark ist Ihr Drang, Suchtmittel zu nehmen?]
12. How much control do you have over consumption of addictive .34 .62 .73 .79 .74 .33 .96
substances? [Wie viel Kontrolle haben Sie über Ihren
Suchtmittelkonsum?]
A13. On average, how strong has been your craving for an addictive 17.6 24.3 - - - - -
substance during the past seven days? [Wie stark war während der
letzten sieben Tage Ihr Verlangen nach Suchtmittel im Durchschnitt?]
A14. Please try to remember the moment within the past seven days 27.1 31.2 - - - - -
when your craving for an addictive substance was strongest. How
 Mannheimer Craving Scale 26

strong was this craving? [Denken Sie bitte einmal an den Moment
innerhalb der letzten sieben Tage zurück, als das Verlangen nach
Suchtmittel am stärksten war. Wie stark war dieses Verlangen?]
A15. How often during the past seven days have you been craving an 18.7 25.2 - - - - -
addictive substance? [Wie häufig hatten Sie während der letzten
sieben Tage Verlangen nach einem Suchtmittel?]
A16. When was the last time you consumed an addictive substance 265.9 1394.6 - - - - -
[Wann haben Sie zuletzt ein Suchtmittel genommen?]
Notes. Items A13-A16 are additional items and are not included in the total MaCS-score. M = mean (item difficulty); SD = standard deviation;
ritc = item-total-correlation (part-whole corrected). Congruence refers to similarity of factor loadings to the patient sample as indexed by
congruence coefficients.
 Mannheimer Craving Scale 27

Table 2

Normative data of MaCS-scores

Stanine-score MaCS-score Percentile rank Relative frequency (%)

1 0 0-4 4
2 0 >4-11 7
3 0 >11-23 12
4 0 >23-40 17
5 1-3 >40-60 20
6 3-7 >60-77 17
7 8-12 >77-89 12
8 13-18 >89-96 7
9 >18 >96-100 4
 Mannheimer Craving Scale 28

Table 3

Sensitivity, specificity, and likelihood ratios of MaCS-scores

MaCS-score Sensitivity Specificity +LR -LR

(95% CI) (95% CI) (95% CI) (95% CI)

≥0 100.00 0.00 1.00 -

(98.6 - 100.0) (0.0 - 0.6) (-) (-)

>0 96.97 40.42 1.63 0.08

(94.1 - 98.7) (36.5 - 44.4) (1.5 - 1.8) (0.04 - 0.1)

>1 95.45 49.84 1.90 0.09

(92.2 - 97.6) (45.8 - 53.9) (1.8 - 2.1) (0.05 - 0.2)

>2 92.42 56.17 2.11 0.13

(88.5 - 95.3) (52.1 - 60.1) (2.0 - 2.3) (0.09 - 0.2)

>3 89.39 61.53 2.32 0.17

(85.0 - 92.8) (57.6 - 65.4) (2.2 - 2.5) (0.1 - 0.2)

>4 87.88 67.53 2.71 0.18

(83.3 - 91.6) (63.7 - 71.2) (2.5 - 2.9) (0.1 - 0.3)

>5 84.47 70.94 2.91 0.22

(79.5 - 88.6) (67.2 - 74.5) (2.7 - 3.1) (0.2 - 0.3)

>6* 80.30 75.16 3.23 0.26

(75.0 - 84.9) (71.6 - 78.5) (3.0 - 3.5) (0.2 - 0.3)

>7 77.27 77.27 3.40 0.29

(71.7 - 82.2) (73.8 - 80.5) (3.1 - 3.7) (0.2 - 0.4)

>8 73.86 80.36 3.76 0.33

(68.1 - 79.1) (77.0 - 83.4) (3.5 - 4.1) (0.3 - 0.4)

 Mannheimer Craving Scale 29

>9 69.32 82.47 3.95 0.37

(63.4 - 74.8) (79.2 - 85.4) (3.6 - 4.3) (0.3 - 0.5)

>10 62.12 85.23 4.21 0.44

(56.0 - 68.0) (82.2 - 87.9) (3.8 - 4.6) (0.3 - 0.6)

>11 59.47 88.47 5.16 0.46

(53.3 - 65.4) (85.7 - 90.9) (4.7 - 5.7) (0.4 - 0.6)

>12 56.82 89.61 5.47 0.48

(50.6 - 62.9) (86.9 - 91.9) (4.9 - 6.1) (0.4 - 0.6)

>13 51.89 90.75 5.61 0.53

(45.7 - 58.1) (88.2 - 92.9) (5.0 - 6.3) (0.4 - 0.7)

Notes. Asterisk indicates the criterion with the highest Youden-index (= sensitivity +
specificity – 1). CI = confidence interval; +LR = positive likelihood ratio; -LR = negative
likelihood ratio.
 Mannheimer Craving Scale 30

Figure caption

Figure 1. ROC-analysis for determining specificity and sensitivity of the MaCS. High values
indicate good discrimination between substance abuse patients and controls.