Alcohol and Alcoholism Advance Access originally published online on March 15, 2006
Alcohol and Alcoholism 2006 41(3):284-292; doi:10.1093/alcalc/agl012
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
DEVELOPMENT AND VALIDATION OF THE AMSTERDAM MOTIVES FOR DRINKING SCALE (AMDS): AN ATTEMPT TO DISTINGUISH RELIEF AND REWARD DRINKERS
1 Amsterdam Institute for Addiction Research, Amsterdam, The Netherlands, 2 Department of Psychiatry, Academic Medical Center and 3 Department of Clinical Psychology, University of Amsterdam, Amsterdam, The Netherlands and 4 Viersprong Institute for Studies on Personality Disorders (VISPD), Halsteren, The Netherlands
* Author to whom correspondence should be addressed at: Amsterdam Institute for Addiction Research (AIAR), Overschiestraat 65, 1062 XD, Amsterdam, The Netherlands. Tel.: +31 204087872; Fax: +31 20 4087862; E-mail: ooteman{at}aiar.nl
(Received 24 October 2005; first review notified 22 December 2005; in revised form 2 February 2006; accepted 3 February 2006)
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
Determination of alcoholic subtypes is a promising strategy for patient treatment matching with anti-craving interventions. The aim of this study is to develop and validate a questionnaire on drinking motives that can distinguish relief and reward drinkers. Methods: A 103-item self-report questionnaire was developed: the Amsterdam Motives for Drinking Scale (AMDS). The AMDS was tested in two samples of alcoholics (Sample A: n = 251; Sample B: n = 197). The psychometric properties were examined in Sample A and cross-validated in B. Results: The AMDS consists of a relief scale and a reward scale, both containing distinct subscales. All (sub)scales had good internal consistency. However, the relief and reward scales were highly correlated (Sample A: r = 0.84, P < 0.01; Sample B: r = 0.76, P < 0.01), and convergent and divergent validity was only partly confirmed. The lowest correlations were found between the relief subscale ‘stress and vulnerability’ and the reward subscale ‘stimulation seeking’ (Sample A: r = 0.33, P < 0.01; Sample B: r = 0.13, P > 0.01). Conclusions: The AMDS reliably measures two types of motives. However, the relief and reward scales are highly correlated. Lower correlations were found only at the subscale level. It is concluded that most treatment-seeking alcoholics have both relief and reward motives. More research is needed on the relationship between motives for drinking and other phenotypic, endophenotypic and genetic indicators of relief and reward drinking, and appropriate cut-off points. Only than we can draw firm conclusions regarding the potential of drinking motives for patient treatment matching.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
Alcohol use is widespread and carries a significant risk of adverse psychological, social and physical health consequences. Effective treatment of alcoholism is important to limit these negative consequences. However, not all patients benefit from the different treatment options available. A promising strategy to improve the effectiveness of existing treatments is more efficient patient treatment matching.
Successful patient treatment matching must be based on adequate understanding of individual differences between alcoholics and the relationship between these individual differences and the effectiveness of different treatment approaches. Our own research group recently reviewed the literature on phenotypic, endophenotypic and genetic matching indicators for the effectiveness of anti-craving medications in alcoholics (Ooteman et al., 2005
). We concluded that most studies have been conducted on phenotypic indicators (e.g. self-report questionnaires), producing inconsistent and rather disappointing results. Studies on genetic and endophenotypic indicators were scarce, but promising. Recently, Verheul et al. (1999) proposed a novel typology of drinkers based on two distinct reinforcing properties of alcohol: anxiety/stress-dampening properties (relief drinkers) and rewarding/stimulating properties (reward drinkers). Although Verheul et al. (1999) did not present tools for the identification of relief and reward drinkers, they did propose two hypotheses for more efficient patient treatment matching. Relief drinkers would benefit from the anti-craving compound acamprosate (a glutamate antagonist), most likely through a reduction of neuronal hyperexcitability, whereas reward drinkers would benefit from the anti-craving compound naltrexone (an opiate antagonist), most likely through blocking of opioid receptors.
Until now, attempts to distinguish relief and reward drinkers empirically have been scarce. Farber et al. (1980), who tried to distinguish relief and reward drinking by looking at drinkers' motives, found that 93% of alcoholics would be classified as relief drinkers using the Reasons for Drinking Questionnaire (RDQ). In addition, Kiefer et al. (2003) showed that most patients mention both ‘relief’ and ‘reward’ motives for their alcohol intake. These findings seem to indicate that no questionnaire is yet able to clearly distinguish between relief and reward drinkers. Equally, relief and reward drinking have never been defined precisely, and therefore these findings may also represent an insufficient conceptual distinction and/or insufficient measurement. The current study attempts adequately to distinguish the concepts of relief and reward drinking and subsequently to develop and test a new questionnaire.
In general, the term ‘relief drinking’ is associated with negative reinforcement, which is the termination of an aversive stimulus immediately following a performance. The term ‘reward drinking’ is generally associated with positive reinforcement, which can be defined as the presentation of a reinforcing stimulus immediately following a performance. Following this definition, the stimulus can be a negative emotion (e.g. stress or anxiety) that is terminated by drinking alcohol (relief drinking) or a positive emotion (e.g. a kick or euphoria) following the drinking of alcohol (reward drinking) (Ferster et al., 1975). A more practical definition has been proposed by Cox and Klinger (1988), who suggested that individuals may drink in order to obtain a positive outcome (positive reinforcement) or to avoid a negative one (negative reinforcement). Both theory and empirical research suggest that the desire to regulate one's affective experience (emotion) is an important motive for drinking alcohol (Wills and Shiffman, 1985; Lang et al., 1999). According to Wills and Shiffman (1985), individuals use alcohol to reduce negative affects when they are anxious and over-aroused or to enhance positive effects when they are fatigued and under-aroused.
Based on these findings, Cooper et al. (1994, 1995) developed a motivational model of alcohol use that tried to distinguish coping and enhancement motives. However, coping and enhancement motives correlated quite highly in both the adolescent and adult non-alcoholic sample (respectively, r = 0.44, P < 0.001; r = 0.59, P < 0.001). In addition, in the adult sample, only 11% could be classified as pure coping drinkers and only 13% as pure enhancement drinkers (Cooper et al., 1995). A similar correlation between positive and negative reinforcement motives was found by Carey and Correia (1997) in a sample of non-alcoholic university undergraduates using the RDQ (r = 0.44, P < 0.001). Again, it seems that relief and reward drinking are overlapping concepts and no practical classification of relief versus reward drinkers can be made. Cooper's model integrated the valence (positive versus negative) and the intensity of arousal (low versus high) that drinkers seek. The idea that motives can vary in the valence as well as the intensity of arousal sought shows similarities with the circumplex model of emotion (Russell, 1980; Larsen and Diener, 1992) representing emotions in a circular order around an arousal axis and a valence axis (see Figure 1a). However, Cooper's model does not clearly define the nature of pre-existing emotional states for enhancement motives or post-existing emotional states for coping motives (Cooper et al., 1995). When looking at Cooper's questionnaire, all enhancement motives are positively formulated and therefore refer solely to the post-existing state (e.g. ‘I drink to get high’, ‘I drink because it is fun’) and most coping motives are negative formulated (e.g. ‘I drink to forget my worries’, ‘I drink to forget about problems’) and therefore are primarily referring to the pre-existing state.
|
In addition, Cooper's coping and enhancement scales are validated in a sample from the general population. Therefore certain motives that are specifically reported by alcoholics (often suffering from co-morbidity such as depression) are not present in Cooper's questionnaire and are difficult to fit into her model, e.g. ‘I drink to feel less passive’, ‘I drink to feel less empty’, ‘I drink to feel calmer’, ‘I drink to feel more at ease’, ‘I drink in order to be able to relax’, ‘I drink to fall asleep more easily’.
In order to further improve the measurement of relief and reward drinking, we developed a new two-dimensional model of relief and reward motives for alcoholics. This new model is based on Cooper's motivational model (Cooper et al., 1995), extended with motives integrating a wide array of (potentially pre-existing and post-existing) affective states as proposed by the circumplex model of emotion (Russell, 1980; Larsen and Diener, 1992; see Figure 1b). A diagonal reward axis is distinguished from a diagonal relief axis representing two distinct orthogonal dimensions. According to this model, reward motives aim at approaching or obtaining a positive reinforcing emotion with higher arousal than the current emotion of the patient (e.g. drinking for the kick), whereas relief motives aim at dampening or terminating a negative reinforcing emotion, resulting in an emotion with lower arousal than the current emotion of the patient (e.g. drinking to reduce stress).
The first aim of the current study is to develop a new questionnaire for drinking motives, the Amsterdam Motives for Drinking Scale (AMDS), which is based on the circumplex model of emotion (Russell, 1980; Larsen and Diener, 1992) and on Cooper's motivational model of alcohol use (Cooper et al., 1995). The second aim is to evaluate its psychometric properties. It is expected that this study will shed empirical light on the concepts of relief and reward motives in alcoholics, resulting in a measure that is able adequately to distinguish relief and reward drinkers and that could be used in patient treatment matching strategies.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
Questionnaire development
First, we conducted a review of the published literature and instruments on drinking motives, expectancies, drinking situations and anticipations of drinking outcome. The reason we also reviewed instruments on theoretical constructs other then motives (Wiers et al., 1997
) is because it has been questioned whether some of these constructs can be meaningfully distinguished (Goldman et al., 1999). A pool of 626 items was created (a complete list can be obtained from W.O.). If the original item represented a construct other than a drinking motive, the item was rewritten as a motive using the format ‘I drank alcohol in order to experience less/more [emotional state].’ For example, ‘After many drinks people become romantic’ is rewritten as ‘I drink to feel more romantic’. The reason we chose the motives format is that motives are generally seen as the most proximal determinant of consumption (Cooper et al., 1995). Both external (interpersonal) and internal (intrapersonal) motives were included.
Second, all items were then allotted to one of the two dimensions of the model. The two-dimensional model of drinking motives consists of (i) relief motives, which are to eliminate or dampen a negative emotion with higher arousal than the post-existing emotional state, and (ii) reward motives, which are to achieve or enhance a positive emotion with higher arousal than the pre-existing emotional state (see Figure 1b). All items that did not fit the model or that contained no emotion in the motive were excluded. Furthermore, we excluded overlapping items, items exclusively aimed at youths or adolescents (e.g. conformity to peer pressure) and items with an unclear intensity of arousal or an unclear valence (e.g. ‘I drink because I like the feeling’, ‘I drink because I like the taste of alcohol’, ‘I drink to become a better driver’). By excluding these items we tried to focus on motives with a strong emotional component and a clearly defined valence and arousal intensity.
Third, the remaining 193 items were then allotted to distinct aspects of the relief and reward dimensions. The five aspects of relief motives were reduction of anxiety (anx), reduction of vulnerability and stress (vul), reduction of angry hostility (hos), reduction of autonomic hyper-arousal (aut) and reduction of social anxiety/shyness (shy). The four aspects of reward motives were achievement of stimulation (sti), achievement of hedonic positive affect (paf), achievement of energy (eny) and achievement of dominance (dom). Items that did not fit with the relief and reward aspects were excluded.
Finally, these 193 items were judged by the research group in terms of their clarity, lack of ambiguity, simplicity, formulation, generalization and overlap. Based on consensus, 11–13 items were selected for each motivational aspect, leaving 103 items for the study. Each motivational aspect included some negatively formulated items (e.g. ‘I drink to feel less bored’) and some positively formulated items (e.g. ‘I drink for the kick’). The rationale for the subscale breakdown into positive and negative wording is based partly on the theory of positive and negative reinforcement. Positive reinforcement is generally defined as achieving something pleasant (e.g. feeling more elated), whereas negative reinforcement is the mechanism of reducing something unpleasant (e.g. feeling less anxious). This definition uses a positive formulation for positive reinforcement and a negative formulation for negative reinforcement only. However, when looking at our circular model (Figure 1b), positive reinforcement motives can be positively and negatively formulated (e.g. ‘I drink to feel more elated’ and ‘I drink to feel less empty’), as can negative reinforcement motives (‘I drink to feel less anxious’ and ‘I drink to feel more relaxed’). We therefore wanted to formulate both positive and negative items for the relief and reward scale. For example, the motivational aspect stimulation seeking (‘sti’) consists of ‘sti-pos’ (positively formulated items) and ‘sti-neg’ (negatively formulated items). The order of the statements was determined randomly. Ten items were duplicated in order to test consistency, resulting in a total of 113 items in the original AMDS. Of the 113 items, 54 were selected, resulting in a short version for efficient cross-validation in a second sample (AMDS-short). The AMDS answer format is a 6-point scale measuring the frequency of each drinking motive in the past 12 months ranging from ‘never’ (0) to ‘very often’ (5).
Subjects
Two groups of alcohol-dependent patients were used for the validation of the AMDS. A large sample (Sample A) of treatment-seeking subjects from four different alcohol treatment centres in The Netherlands (Jellinek, Boumanhuis, Parnassia, Kuno van Dijk Stichting) was used for the investigation of the psychometric properties of the AMDS. Another sample (Sample B) was used for cross-validation of the AMDS. This latter sample contained treatment-seeking and non-treatment-seeking alcoholics who were recruited at the Jellinek addiction treatment clinic in Amsterdam and through advertisements in newspapers and radio interviews.
All subjects had a primary DSM-IV diagnosis of alcohol dependence. They were required not to have had heavy drinking days (
5 units) for 1 week to 6 months in order to ensure adequate recollection. Exclusion criteria were co-morbid cocaine or heroin dependence, cocaine or heroin use in the 30 days before intake, current psychosis, current use of anti-psychotic medication, current use of anti-craving compounds (naltrexone, acamprosate) and disulfiram, severe cognitive deficits and insufficient command of the Dutch language.
A total of 1019 subjects were screened: 495 subjects in Sample A and 524 subjects in Sample B. A total of 80 (Sample A) and 188 (Sample B) subjects refused to participate. Finally, 107 (Sample A) and 139 subjects (Sample B) had to be excluded; the most important reasons for exclusion were no primary diagnosis of alcohol dependence and 6 months of abstinence (Sample A), medical conditions and current use of anti-craving compounds (Sample B). This left 308 subjects in Sample A and 197 in Sample B. A total of 251 of 308 subjects (81.5%) in Sample A and 183 of 197 subjects (92.9%) in Sample B completed the study. A subgroup of 34 subjects in Sample A was asked to administer the AMDS twice with a 3 week interval in order to establish (test) retest reliability.
Procedure
Sample A was used for scale construction and estimation of convergent and divergent validity of the AMDS. Sample B was used for cross-validation. All subjects in Samples A and B self-administered a set of self-report questionnaires. The administration took 
1.5 h. All subjects received vouchers for 
12.50 after completion.
Self-report measures
All participants in Samples A and B completed the Alcohol Use Disorders Identification Test (AUDIT) (Babor et al., 1992). AUDIT scores can range from 0 to 40, and the generally accepted cut-off point for identifying a potential alcohol problem is 8. All participants in Sample B completed the short version (54 items) of the AMDS. All participants in Samples A and B also completed the following additional questionnaires in order to estimate the convergent and divergent validity of the AMDS: the Motives for Drinking Questionnaire (MDQ) (Cooper, 1994; translated in Dutch by W.O. and R.V.), the BIS/BAS-Scale (Carver and White, 1994), the Alcohol Effects Questionnaire (AEFQ) (Rohsenow, 1983), the NEO-PI-R Personality Inventory (Costa and McCrae, 1992), the Anxiety Sensitivity Index (ASI) (Reiss et al., 1986), the High Sensitive Person (HSP) Scale (Aron and Aron, 1997), the Boredom Proneness Scale (Farmer and Sundberg, 1986), the Jellinek Alcohol Craving Questionnaire (JACQ) ‘now’ and ‘past’ version (Ooteman et al., 2006), the Mood Adjective Checklist (MACL) (Mackay et al., 1978), the Hospital Anxiety and Depression Scale (HADS) (Zigmond and Snaith, 1983) and a series of self-constructed items to assess family history of alcoholism (FH), withdrawal symptoms and binge drinking. A positive family history was defined as at least one first-degree relative (parents or siblings) having (had) an alcohol problem.
Statistical analysis
A principal component analysis (PCA) with varimax rotation was performed to explore whether the underlying structure of the individual AMDS items resembles the theoretical two-dimensional motivational structure.
Next, the internal consistency of the theoretical scales was determined by computing Cronbach's alphas, and these values were compared with the internal consistency of the scales that resulted from the PCA. Interscale correlations, test–retest reliability, convergent validity and divergent validity were assessed using Pearson correlation coefficients. For family history and binge drinking effect sizes were presented as Cohen's d. Divergent validity in our definition means that an AMDS scale correlates lowly (r 
0.30) with variables reflecting a theoretically unrelated construct and convergent validity means that an AMDS scale correlates highly (r 0.60) with variables reflecting a theoretically related construct.
In order to distinguish between extreme groups of relief and reward drinkers, difference scores were calculated by subtracting the z transformed score on the reward scale from the z transformed score on the relief scale (z transformation adjusts for differences in mean and variance between the scales). The 10% of subjects with a difference score >0.77 were defined as relief drinkers, whereas the 10% with a difference score <–0.70 were defined as reward drinkers. A similar procedure was used for the two relief and reward scales with the lowest interscale correlation: the positively formulated ‘stimulation seeking’ subscale (sti-pos) and the negatively formulated ‘vulnerability’ subscale (vul-neg). The 10% of subjects with a score >1.50 were defined as relief drinkers, whereas the 10% with a score <–1.57 were defined as reward drinkers. Differences between relief and reward drinkers with respect to demographics and alcohol-related variables were tested using t-tests. Differences between reward and relief drinkers were tested with an 
of 0.01. This value takes into account both a Bonferroni correction for multiple testing and the relatively small sample sizes, which affect the power of the tests.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
Sample characteristics
Table 1 shows that most study participants were Dutch, male, of low educational level, unemployed and unmarried. In Sample A, most subjects were inpatient alcoholics (77.7%), whereas in Sample B 29.9% were inpatient, 41.1% were outpatient and 28.9% were currently not seeking treatment. Most subjects had moderate to severe levels of alcohol dependence. Sample A was less well educated and more severely addicted according to the AUDIT scores than Sample B. These differences seem to be related to the presence of non-treatment-seeking alcoholics in Sample B who were more highly educated (P < 0.01) and less severely addicted according to the AUDIT (P < 0.01) than the treatment-seeking alcoholics in Sample B.
|
Factor structure and internal consistency of the AMDS (sub)scales (content validity)
As a first step, the internal consistencies of the theoretical AMDS scales were calculated in Sample A. Internal consistencies of the relief and reward scales were high (
= 0.97 for both scales). Internal consistencies of the nine subscales were also good (ranging from = 0.90 to = 0.95 for the relief subscales and from = 0.87 to = 0.94 for the reward subscales).
We expected to find low correlations between relief and reward scales. However, the correlation for the relief and reward scales was 0.84 (P < 0.01), whereas the correlations between the relief and reward subscales ranged from 0.46 to 0.83 (P < 0.01) (Table 2). The lowest correlations were found between the negatively formulated relief and the positively formulated reward subscales, ranging from r = 0.33 (P < 0.01) for the correlation between the negatively formulated ‘vulnerability’ subscale (vul-neg) and the positively formulated ‘stimulation seeking’ subscale (sti-pos) to r = 0.78 (P < 0.01) for the correlation between the positively formulated ‘dominance’ subscale (dom-pos) and the negatively formulated ‘shyness’ subscale (shy-neg) (Table 3). Given the generally high interscale correlations between reward and relief motives, we questioned whether the AMDS really measures two dimensions of drinking motives. In order to explore this question, a PCA was performed. Most of the variance (37.0%) was explained by a factor closely resembling the relief scale and a much smaller portion of the variance (6.9%) was explained by a factor resembling the reward scale. The internal consistency of the explorative scales was high and very similar to those of the theoretical scales ( = 0.97 for both relief and reward). Since the content and the internal consistency of the explorative AMDS scales were very similar to those of the theoretical scales, it was decided to use the theoretical scales for further analysis.
|
|
As a final step, a short version of the AMDS was constructed that was used in Sample B. In Sample A, this 54-item version of the AMDS correlated 0.99 (P < 0.01) with the 113-item version. When the theoretical scales of AMDS-short were cross-validated in Sample B, interscale correlations and internal consistencies were in the same range as for the original long version of the AMDS in Sample A. Again, relief and reward correlated highly (r = 0.76, P < 0.01), with the lowest correlations between the positively formulated ‘simulation seeking’ subscale (sti-pos) and the negatively formulated subscales ‘anxiety’ (anx-neg), ‘vulnerability’ (vul-neg) and ‘angry hostility’ (hos-neg) (r = 0.18, P > 0.01; r = 0.13, P > 0.01; and r = 0.12, P > 0.01, respectively).
Patient characteristics and alcohol-related variables of reward and relief drinkers
Based on scores on the relief and reward scales, subjects in Sample A were divided into relief and reward drinkers. When the difference scores on the relief and reward scales were used, 20 patients (8%) were defined as relief drinkers and 20 patients (8%) were defined as reward drinkers. The remaining 211 subjects (84%) were defined as combination drinkers and were left out of further analyses. We used these criteria because currently we do not have an external criterion such as clinically relevant cut-off scores. Relief and reward drinkers were not significantly different with respect to most of the patient characteristics. However, a significant effect was found for education (P < 0.01), with lower education in the relief group.
When the definition of relief and reward was based on difference scores of the negatively formulated subscale ‘vulnerability’ (vul-neg) and positively formulated subscale ‘stimulation seeking’ (sti-pos), 24 subjects (10%) were defined as relief drinkers and 24 subjects (10%) were defined as reward drinkers. A significant effect was found for gender and AUDIT scores (P < 0.01), with more females and higher AUDIT scores in the relief group.
Also in Sample B, 17 relief drinkers (9%) and 17 reward drinkers (9%) were not significantly different with respect to any variables, except that a significant effect was found for education and the subject's setting (P < 0.01), with lower education and more inpatients in the relief group. When the definition of relief and reward was based on the negatively formulated ‘vulnerability’ and positively formulated ‘stimulation seeking’ subscales, a significant difference was found between the 18 relief drinkers (9%) and the 17 reward drinkers (9%) in setting: relief drinkers were more likely than reward drinkers to be inpatients (P < 0.01). A trend was found for treatment-seeking (P < 0.05). Relief drinkers were more likely to be in treatment.
Test–retest reliability (n = 34)
For Sample A, the test–retest reliability of the AMDS was assessed in 34 subjects. For most subscales pre-test scores were not significantly different from the scores on the retest 3 weeks later. Relief and reward scores at pre-test were highly correlated with relief and reward scores 3 weeks later (r = 0.93 and 0.94, respectively; P < 0.01). The test–retest results for the subscales were very similar: ranging from r = 0.86 to r = 0.90 (P < 0.01) for the relief subscales and from r = 0.84 to r = 0.95 for the reward subscales.
Convergent and divergent validity of the AMDS
As Table 4 shows there is not much evidence for convergent validity of the relief and reward scales. Most correlations with instruments and variables measuring constructs theoretically related to the constructs of the relief and reward scale were below threshold, with the exception of the correlations for the relief scale with the coping scale of the MDQ (r = 0.67; P < 0.01) and withdrawal symptoms (r = 0.60; P < 0.01) and correlations for the reward scale with the enhancement scale of the MDQ (r = 0.63; P < 0.01) and the global positive scale of the AEFQ (r = 0.64; P < 0.01). There is evidence for moderate to good divergent validity of the relief scale. Correlations <0.30 were found with the BAS scale (r = 0.10; P > 0.01), the extraversion scale of the NEO-PI-R (r = –0.15; P > 0.01), the arousal scale of the MACL (r = –0.23; P < 0.01) and the social pleasure scale of the AEFQ (r = 0.28; P < 0.01). However, the divergent validity of the reward scale was low, except for a below-threshold correlation with the relaxation scale of the AEFQ (r = 0.25; P < 0.01).
|
As expected, the correlation pattern with all variables was very similar for the AMDS relief and reward scales, probably reflecting the high correlation between the AMDS relief and reward scales (r = 0.84)—a situation that was also observed for the AMDS subscales, although to a slightly lesser extent. Very similar patterns of correlations were found in Sample B.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
The aim of the current study was to develop and validate a self-report questionnaire measuring two theoretically independent motives for drinking: relief and reward.
First, the explorative factor analysis showed a factor solution that was similar to our theoretical model. We therefore decided to continue our analyses with the scales that were based on our theoretical model.
Second, the AMDS showed good test–retest reliability, high internal consistencies of all (sub)scales and high interscale correlations between relief and reward motives (r = 0.84). The high correlation does not support the idea that reward and relief motives are independent factors. Only at a narrowly defined subscale level were lower correlations found.
In Sample A the negatively formulated relief aspect ‘vulnerability’ correlated lowly with the positively formulated reward aspect ‘stimulation seeking’ (r = 0.33). Apparently, these subscales showed the best discrimination. After inspection of the model (Figure 1b), it can be concluded that the negatively formulated ‘vulnerability’ and the positively formulated ‘stimulation seeking’ subscales, as well as Cooper's scales, represent only the upper half of our model. This result is in line with the definition of positive and negative reinforcement, using a positive formulation for reward and a negative one for relief. However, when all negatively formulated items were correlated with all positively formulated items of the AMDS, we did not find high discrimination, which may indicate that the content rather than the wording of the subscales is of importance. In Sample B, the correlation between the subscales ‘vulnerability’ and ‘stimulation seeking’ was even lower (r = 0.13, P > 0.01), which may be explained by the fact that in Sample B a more heterogeneous sample with less severely dependent non-treatment-seeking alcoholics was studied. It is suggested that in alcoholics at the start of their pathological drinking career (first admissions, non-treatment-seeking alcoholics) it is easier to distinguish reward and relief motives, whereas chronically addicted inpatients with a long history of treatment may have developed secondary relief drinking owing to the adverse effects of their addiction (e.g. withdrawal symptoms) or may have developed a behavioural drinking pattern that is driven by automatic processes rather than by rational motives. This post hoc explanation is supported by the observed trend that treatment-seeking alcoholics showed a tendency to endorse more relief motives than non-treatment-seeking alcoholics.
Third, relief and reward drinkers were not significantly different with respect to most of the demographic or alcohol-related variables. There may be a relationship with education and setting. In our study, relief drinkers were more likely to be less well educated and inpatients. When the definition of relief and reward motives was based on the subscales ‘vulnerability’ and ‘stimulation seeking’ only, relief drinkers were also more likely to be treatment-seeking females with more severe alcohol problems. These findings are partially in line with the definition of a relief drinker by Verheul et al. (1999).
Fourth, with respect to the convergent and divergent validity of the AMDS, it can be concluded that convergent and divergent validity for the relief and reward scales are only partly confirmed. The AMDS was strongly correlated with the MDQ, which is probably due to a large overlap in content and formulation. However, it should be noted that the MDQ does not include all emotions of the circumplex model of emotion (Figure 1a). Therefore, we expected slightly lower correlations between the relief and reward scales of the AMDS and the MDQ.
The correlation pattern of the relief and reward scales of the AMDS with the other variables was quite similar. These findings probably reflect the high correlation between the relief and reward scales of the AMDS (r = 0.84). Both relief and reward motives correlated only moderately with craving now (JACQ-now; r = 0.32 and 0.33, respectively) and rather highly with craving in the past (JACQ-past; r = 0.58 and 0.48, respectively), suggesting that the concepts of craving in the past and drinking motives are related phenomena. Therefore, the reported types of motive may be an indication for the pathways underlying a patient's craving. However, craving may not necessarily be accompanied by a motive or vice versa. More research is needed to unravel the relationship between motives and craving.
Fifth, and most important, we found that most patients proved to be so-called drinkers with combined motives, whereas the extreme groups of relief and reward drinkers were very small (16–20%). Apparently, most treatment-seeking alcoholics report both relief and reward motives. This is in line with earlier studies (Farber et al., 1980; Kiefer et al., 2003). The most likely explanation may be that the subjects had distinct relief or reward motives in the past but have developed both motives for drinking as a result of their dependence. However, the existing questionnaires, including the AMDS, may not be able to differentiate between persons with a (predominantly) positive reinforcement history and persons with a (predominantly) negative reinforcement history. We therefore suggest investigating the effect of the course of an alcohol problem on the development of reward and relief motives, say by developing a ‘past’ version of the AMDS. Another option would be to investigate the influence of drinking career on relief and reward drinking by looking at genotypic and endophenotypic indicators of relief and reward drinking in addition to phenotypic indicators, as genotypic and endophenotypic indicators are less influenced by environmental factors.
An alternative explanation may be that positive and negative reinforcement mechanisms are two distinct mechanisms that have always co-existed to a similar extent in most drinkers (activated simultaneously or depending on the situation). If future research indicates this, the distinct subtypes ‘relief drinker’ and ‘reward drinker’ do not exist, according to our model.
An import factor with respect to the validity of the AMDS and other drinking motives questionnaires is that there may be individual differences in the extent to which patients are consciously aware of their implicit underlying mechanisms resulting in (excessive) drinking. The degree of awareness or detection of motivational mechanisms can therefore influence the way the AMDS is filled out. Results from a previous study by our research group may illustrate this important point. It was shown that the explicit (aware) phenotype ‘craving’ and the related but implicit (not necessarily aware) endophenotype ‘physiological cue reactivity’ correlated poorly because of individual differences in detection levels of physiological cue reactions (Ooteman et al., 2006). Motives are by definition subjective experiences and the end product of various factors. Knowing your motives implies a sufficient level of awareness and explicit information processing. In fact, sufficient awareness of these mechanisms is a prerequisite for valid administration of the AMDS, and therefore a prerequisite for valid distinction of relief and reward drinkers (e.g. for patient treatment matching). In addition, we do not know whether patients always report what they experience. However, the mechanisms of positive and negative reinforcement do not assume sufficient awareness or honest reporting. Therefore, more research is warranted on the relationship between the AMDS and indicators of relief and reward drinking that require lower levels of awareness and that are less affected by report bias, such as (i) other (more objective) phenotypic indicators (age of onset); (ii) endophenotypic indicators (physiological cue reactivity) and (iii) genetic indicators (mu-opioid receptor polymorphism) of relief and reward drinking. This may shed light on the external validity of the AMDS. However, we do not think that lack of awareness is a very likely explanation for the high correlation between the relief and reward motives of the AMDS. If our patients had had low level of awareness, we would have expected low to moderate correlations between the (sub)scales owing to a lack of variance.
A final limitation of this study is that our definition of relief and reward drinking was based on difference scores on the relief and reward scales, and not on clinically relevant cut-off scores. Therefore our percentages may not be accurate. More research is needed on appropriate cut-off scores.
In conclusion, it seems that the AMDS is a reliable questionnaire for measuring different types of drinking motives, but that it is not able to distinguish between relief and reward drinkers in a population of mainly treatment-seeking alcoholics. Apparently, the broad concepts of relief and reward motives are closely related in moderate to severe alcoholics. This is disappointing because it seems that currently the AMDS cannot be used in the process of patient treatment matching. More research is needed regarding the external validity of the AMDS, i.e. correlations with phenotypic, endophenotypic and genetic indicators of relief and reward drinking. Research is also warranted regarding the use of appropriate clinical cut-off points and the discriminative properties of the AMDS (sub)scales. Only then will we be able to draw firm conclusions regarding the potential of using drinking motives for patient treatment matching. However, even if we are able to distinguish relief and reward motives in the future, this does not automatically imply predictive value in patient treatment matching. And the opposite applies as well: even if we still find overlap between relief and reward motives and the AMDS cannot be used for matching strategies based on our model, this would not per se speak against a use for matching strategies based on other models. Until now, these issues have remained empirical questions. Future research will have to show whether categorization of extreme groups of relief and reward drinkers will contribute to better patient treatment matching and whether a combination of interventions (e.g. polypharmacy) is the future strategy for drinkers with both relief and reward motives.
|
APPENDIX |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
AMDS-short (54-item version, translated from Dutch)
This questionnaire aims at getting an impression of your reasons for drinking alcohol during the past 12 months, or in other words what your motives were to drink. The questionnaire contains a few statements (motives) that ask about reasons for drinking alcohol. Please indicate how often your experience matched each of the following statements by placing a circle around one of the numbers. There are no wrong or right answers. For example, the number ‘5’ indicates that your experience matched the statement often; the number ‘0’ indicates never. We are interested in your own reasons for drinking alcohol in the ‘past 12 months’. If you have quit drinking, please fill out the questionnaire for the time when you were still drinking. Please read carefully, but do not think too long about your answer. Usually your first impression is the best one. Please complete every item.
In the past 12 months I drank alcohol ...
|
|
ACKNOWLEDGEMENTS |
|---|
We thank Simone Höhner, Premal Koning, Ron Baas, Sigrid Wittenberg, Dorine Eeken, Kai Hulstijn, Weena Chen, Gjalt de Vries and Hans Kronemeijer for their assistance. Source of support: NWO/ZON-MW 985-10-013.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
Aron, E. N. and Aron, A. (1997) Sensory-processing sensitivity and its relation to introversion and emotionality. Journal of Personality and Social Psychology 73, 345–368.[CrossRef][ISI][Medline]
Babor, T. F., De la Fuente, J. R., Saunders, J. et al. (1992) The Alcohol Use Disorders Identification Test: Guidelines for Use in Primary Health Care. WHO Publication No. 92.4, World Health Organization, Geneva, Switzerland.
Carey, K. B. and Correia, C. J. (1997) Drinking motives predict alcohol-related problems in college students. Journal of Studies on Alcohol 58, 100–105.[Medline]
Carver, C. S. and White, T. L. (1994) Behavioural inhibition, behavioral activation, and affective responses to impending reward and punishment: the BIS/BAS scales. Journal of Personality and Social Psychology 67, 319–333.[CrossRef]
Cooper, M. L. (1994) Motivations for alcohol use among adolescents: development and validation of a four-factor model. Psychological Assessment 6, 117–128.[CrossRef]
Cooper, M. L., Frone, M. R., Russell, M. et al. (1995) Drinking to regulate positive and negative emotions: a motivational model of alcohol use. Journal of Personality and Social Psychology 69, 990–1005.[CrossRef][ISI][Medline]
Costa, P. T. and McCrae, R. R. (1992) Revised NEO Personality Inventory (NEO-PI-R) and the Five Factor Inventory (NEO-FFI): Professional Manual. Psychological Assessment Resources Inc., Odessa, Florida.
Cox, W. M. and Klinger, E. (1988) A motivational model of alcohol use. Journal of Abnormal Psychology 97, 168–180.[CrossRef][ISI][Medline]
Farber, P. D., Khavari, K. A. and Douglass, F. M. (1980) A factor analytic study of reasons for drinking: empirical validation of positive and negative reinforcement dimensions. Journal of Consulting and Clinical Psychology 48, 780–781.[CrossRef][ISI][Medline]
Farmer, R. and Sundberg, N. D. (1986) Boredom proneness—the development and correlates of a new scale. Journal of Personality Assessment 50, 4–17.[CrossRef][ISI][Medline]
Ferster, C. B., Culbertson, S., and Perrott Boren, M. C. (1975) Positive and negative reinforcers and punishment. In Behavior Principles, pp. 151–165. Prentice-Hall, Inc., Englewood Cliffs, NJ.
Goldman, M. S., Del Boca, F. K. and Darkes, J. (1999) Alcohol expectancy theory: the application of cognitive neuroscience. In Psychological Theories of Drinking and Alcoholism, Blane, H. T. and Leonard, K. E. eds, pp. 203–246. Guilford Press, New York.
Kiefer, F., Jahn, H., Tarnaske, T. et al. (2003) Comparing and combining naltrexone and acamprosate in relapse prevention of alcoholism: a double-blind, placebo-controlled study. Archives of General Psychiatry 60, 92–99.
Lang, A. R., Patrick, C. J. and Stritzke, W. G. K. (1999) Alcohol and emotional response: a multidimensional multilevel analysis. In Psychological Theories of Drinking and Alcoholism, Blane, H. T. and Leonard, K. E. eds, pp. 328–371. Guilford Press, New York.
Larsen, R. J. and Diener, E. (1992) Promises and problems with the circumplex model of emotion. In Emotion, Clark, M. S. ed, pp. 25–59. Sage Publications, Inc., Thousand Oaks, CA.
Mackay, C., Cox, T., Burrows, G., and Lazzerini, T. (1978) An inventory for the measurement of self-reported stress and arousal. British Journal of Social and Clinical Psychology 17, 283–284.
Ooteman, W., Verheul, R., Naassila, M. et al. (2005) Patient-treatment matching with anti-craving medications in alcohol-dependent patients: a review on phenotypic, endophenotypic and genetic indicators. Journal of Substance Use, 10, 75–96.
Ooteman, W., Koeter, M. W., Verheul, R. et al. (2006) Measuring craving: an attempt to connect subjective craving with cue reactivity. Alcoholism Clinical and Experimental Research 30, 57–69.[Medline]
Reiss, S., Peterson, R. A., Gursky, D. M. et al. (1986) Anxiety sensitivity, anxiety frequency and the prediction of fearfulness. Behaviour Research and Therapy 24, 1–8.[CrossRef][ISI][Medline]
Rohsenow, D. J. (1983) Drinking habits and expectancies about alcohol's effects for self versus others. Journal of Consulting and Clinical Psychology 51, 752–756.[CrossRef][ISI][Medline]
Russell, J. A. (1980) A circumplex model of affect. Journal of Personality and Social Psychology 39, 1161–1178.[CrossRef]
Verheul, R., Van den Brink, W. and Geerlings, P. (1999) A three-pathway psychobiological model of craving for alcohol. Alcohol and Alcoholism 34, 197–222.
Wiers, R. W., Hoogeveen, K. J., Sergeant, J. A. et al. (1997) High- and low-dose alcohol-related expectancies and the differential associations with drinking in male and female adolescents and young adults. Addiction 92, 871–888.[Medline]
Wills, T. A. and Shiffman, S. (1985) Coping and substance use: a conceptual framework. In Coping and Substance Use, Shiffman, S. and Wills, T. A. eds, pp. 3–24. Academic Press, Orlando, FL.
Zigmond, A. S. and Snaith, R. P. (1983) The hospital anxiety and depression scale. Acta Psychiatrica Scandinavia 67, 361–370.[ISI][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||