Addictive Behaviors 39 (2014) 622–629

Contents lists available at ScienceDirect

Addictive Behaviors

A multiple motive/multi-dimensional approach to measure smokelesstobacco dependence

Nasir Mushtaq a,⁎, Laura A. Beebe b, Sara K. Vesely b, Barbara R. Neas b

a Department of Biostatistics and Epidemiology, University of Oklahoma Health Sciences Center, 4502 East 41st Street, SAC 1G06, Tulsa, OK 74135, USAb Department of Biostatistics and Epidemiology, University of Oklahoma Health Sciences Center, 801 NE 13th Street, CHB-309, Oklahoma City, OK 73104, USA

H I G H L I G H T S

• We used a multidimensional approach to measure dependence among ST users.• OSSTD identified seven latent constructs including 23 items to measure ST dependence.• OSSTD possesses better psychometric properties than FTND-ST.• OSSTD is an effective tool to measure ST dependence as a multidimensional construct.

⁎ Corresponding author. Tel.: +1 918 660 3680; fax: +E-mail addresses: Nasir-Mushtaq@ouhsc.edu (N. Mush

(L.A. Beebe), Sara-Vesely@ouhsc.edu (S.K. Vesely), Barbar

0306-4603/$ – see front matter © 2013 Elsevier Ltd. All rihttp://dx.doi.org/10.1016/j.addbeh.2013.11.016

a b s t r a c t

a r t i c l e i n f o

Keywords:

Smokeless tobaccoSmokeless tobacco dependenceDependence measureDependence scaleOklahoma Scale for Smokeless Tobacco DependenceOSSTD

Background:Unlike various research studies conducted to address dependence among smokers, only a few studieshave examined smokeless tobacco (ST) dependence. The Fagerström Tolerance Questionnaire (FTQ) andFagerström Test for Nicotine Dependence (FTND) based scales are the most widely used measures of nicotinedependence for both ST users and smokers. These scales were initially developed tomeasure physical dependenceand tolerance and not to assess other salient dimensions of dependence such as craving, compulsion, orwithdraw-al, as defined by DSM-IV and ICD-10. The aim of this study is to develop and validate a multidimensional scale

that has better content coverage, factor structure, and psychometric properties to measure dependence amongST users.Methods: 100 adult male smokeless tobacco users were recruited through email distribution lists and communityreferral. Participants completed three different nicotine dependence questionnaires and provided informationrelated to their tobacco use and demographic characteristics. They also provided a saliva sample for cotininemeasurement. In order to develop the new ST scale, subscales and items were selected based on correlation andfactor analysis of the modifiedWISDM-68. Reliability and validity of the new scale, Oklahoma Scale for SmokelessTobacco Dependence (OSSTD) were also assessed.Results: The new ST scale identified seven latent constructs including 23 items tomeasure ST dependence. Internalconsistency as measured by Cronbach's coefficient (α = 0.925) indicated better reliability of OSSTD than FTND-ST. Concurrent validity of OSSTD as evaluated by comparing it with dependence diagnosis and FTND-ST wasaffirmative. There was a significant correlation between the OSSTD total score and the cotinine levels and tobaccouse characteristics among study participants.Conclusion:OSSTDpossesses better psychometric properties andprovides an effective and efficient tool tomeasureST dependence as a multidimensional construct.

© 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Despite a decline in smoking prevalence, smokeless tobacco use isincreasing in the United States (SAMHSA, 2010). The Surgeon General'sreport of 1986 recognized that nicotine dependence associated withsmokeless tobacco use follows an addiction pattern similar to other

1 918 660 3671.taq), Laura-Beebe@ouhsc.edua-Neas@ouhsc.edu (B.R. Neas).

ghts reserved.

substance abuse (USDHHS, 1986). Clinical and laboratory studies haveidentified tolerance and withdrawal effects among ST users (Giovino,Henningfield, Tomar, Escobedo, & Slade, 1995; Hatsukami, Gust, &Keenan, 1987). For example in a study of baseball players, withdrawaleffects related to ST abstinence were observed in the form of cognitiveimpairment affecting players' performance during the game(Robertson et al., 1995). ST has unique pharmacokinetics; unlike ciga-rette smoking that delivers nicotine to arterial blood through thealveoli of the lungs, nicotine is absorbed through the oralmucosalmem-branes into the venous blood. Bioavailability of nicotine from ST differs

623N. Mushtaq et al. / Addictive Behaviors 39 (2014) 622–629

from that of cigarettes, as it depends on the nicotine concentration, pHlevel, and the tobacco cuttings of the ST product (Henningfield,Radzius, & Cone, 1995). ST users have higher concentrations of nicotine,as measured by cotinine levels, compared to cigarette smokers (Fant,Henningfield, Nelson, & Pickworth, 1999). Due to these differences inthe pharmacokinetics of the two nicotine delivery systems, behavioraland psychosocial factors explaining dependence among ST users maybe different from cigarette smokers. Relative to cigarette smoking,there are discrete sensory stimulations and cues associated withST use. A recent study of product-specific assessment of dependenceelaborated the role of nicotine and non-nicotine factors in repeateduse of tobacco products and noted product-specific behaviors andstimuli as a function of dependence (Fagerstrom & Eissenberg, 2012).These findings highlighted the importance of product-specificmeasuresof dependence for different tobacco products.

Unlike various research studies conducted to address dependenceamong smokers, only a few studies have been conducted for smokelesstobacco dependence (Boyle, Jensen, Hatsukami, & Severson, 1995;Ebbert, Patten, & Schroeder, 2006; Ferketich, Wee, Shultz, & Wewers,2007; Thomas et al., 2006). Most of these studies used the modifiedFagerström Tolerance Questionnaire (FTQ) or the Fagerström Test forNicotine Dependence (FTND) to measure ST dependence. The FTQdeveloped in 1989 was derived from the original Fagerström test,to measure nicotine dependence among smokers (Fagerstrom &Schneider, 1989). FTQ and FTND are the most widely used measuresin nicotine dependence assessment and their predictive validity hasbeen studied among cigarette smokers (Fagerstrom & Schneider,1989; Kozlowski, Porter, Orleans, Pope, & Heatherton, 1994; Pinto,Abrams, Monti, & Jacobus, 1987). The FTQ for smokers was notdeveloped according to standard psychometric methods. However,this questionnaire was intended to measure physical dependence andtolerance as the aim was to provide a tool for smoking cessation treat-ments to assess different levels of dependence of individual smokers.Thus the FTQand FTNDdonot assess other salient dimensions of depen-dence — craving or compulsion to smoke, as defined by DSM-IV andICD-10. FTND is also criticized for its multifactorial structure as it doesnot measure a single construct of physical dependence (Etter, Duc, &Perneger, 1999; Heatherton, Kozlowski, Frecker, Rickert, & Robinson,1989). The FTQ and FTND heavily rely on two constructs: heaviness oftobacco use and withdrawal. Due to inconsistent scores to the item re-sponses, questions related to the heaviness of use have more contribu-tions to the total score. Therefore, using cotinine levels, either salivaryor serum cotinine, to validate these dependence measures resulted incriterion contamination (DeVon et al., 2007; Piper et al., 2004). Validityof these scales against other criterion variables was not consistentlyestablished.

In an effort to address the shortcomings of the FTQ and DSM-IVbased smoking dependence measures, Piper et al. developed theWisconsin Inventory of Smoking Dependence Motives (WISDM-68)(Piper et al., 2004). This theory drivenmeasure of smoking dependenceidentifies 13motives of nicotine dependence among smokerswhich aremeasured by 68 items. Contrary to the other measures of tobaccodependence among smokers, WISDM-68 measures multiple motivesfor smoking that contribute to compulsive use and result in nicotinedependence. Preliminary studies focused on WISDM-68 have demon-strated excellent internal consistency for the overall scale (α = 0.96)although subscales had internal consistency ranging from 0.74 to 0.94.Other than some exceptions to the individual subscales, concurrentvalidity of this tool showed significant correlation with FTND, NicotineDependence Syndrome Scale (NDSS), smoking rate, alveolar carbonmonoxide, and DSM-IV based criteria of smoking dependence (Piperet al., 2004; Piper et al., 2008).

Regardless of the better psychometric properties and broader cover-age of dependence motives by WISDM-68, its length limits its use inresearch or clinical settings. More recent research on WISDM-68 hasexplored different possibilities to shorten the original scale, ranging

from simply combining highly correlated subscales to applying morerigorous strategies by integrating person centered analyses and variablecentered analysis to categorize subscales into broader constructs. Forexample, subscales such as Negative reinforcement and Positive rein-forcement were highly correlated; similarly Affiliative attachment andBehavioral choice–melioration subscales had stronger correlation(Smith et al., 2010). Latent class analysis and factor mixture modelsidentified two synthetic WISDM scales. The Primary Dependence Mo-tives (PDM) scale consisting of Automaticity, Loss of craving, Tolerance,and Craving subscales and the Secondary Dependence Motives (SDM)scale containing the remaining nine subscales (Piper et al., 2008).Based on these findings a shorter form of WISDM-68, Brief WISDMwhich included 11 subscales and 35 items was developed. It providedreliable and representative content coverage and demonstrated compa-rable psychometric properties (Smith et al., 2010). To our knowledge,neither the WISDM-68 nor Brief WISDM scales have ever been appliedto a ST using population.

A recent study of ST dependence measures compared modified FTQwith two newer scales, Glover-Nilsson Smokeless Tobacco BehavioralQuestionnaire (GN-STBQ), a variant of Glover-Nilsson Smoking Behav-ioral Questionnaire (GN-SBQ), and Severson Smokeless Tobacco Depen-dency Scale (SSTDS) which included FTQ items and items assessingbehavioral patterns of ST use and withdrawal symptoms. GN-STBQand SSTDS had diverse items which measured other dimensions of STdependence in combination with physical dependence. Both the scaleswere significantly associated with craving and withdrawal but did notpredict ST abstinence. These findings underscore the importance of amultidimensional scale to evaluate ST dependence (Ebbert, Severson,Danaher, Schroeder, & Glover, 2012). Similar to the previous ST depen-dence scales, this study employed a cigarette dependence measure todevelop a ST dependence scale, however, we adapted the WISDM-68,one of the most comprehensive and multiple motive measures ofsmoking dependence.

The goal of this study is to develop and validate a multidimensionalscale that has better content coverage, factor structure, and psychomet-ric properties to measure dependence among ST users, as comparedto previously used ST dependence measures. Specifically, the studyexamined a modified version of the WISDM-68 to assess dependenceamong ST users and retained sufficient items and subscales to measuremultiple motives of ST dependence.

2. Methods

2.1. Study population

Participants of this study were adult ST users living in Oklahomafrom May 2010 to December 2010. They were recruited through emaildistribution lists and community referrals. Eligibility criteria includedages 18 to 65 years, no current smoking, at least 1 year use of smokelesstobacco, and consuming at least one pouch or can of ST each week.Participantswith a history of other substance abuse or history of psychi-atric illness were not included. Following an initial telephone screeningto verify eligibility and obtain consent, study materials were mailed toparticipants. This study was approved by the University of OklahomaInstitutional Review Board (IRB#15079) and the Oklahoma StateUniversity IRB.

2.2. Data

140 adult ST users were screened for the eligibility criteria and 100of them completed three different nicotine dependence questionnairesand provided information related to their tobacco use and demographiccharacteristics. To assess saliva cotinine, saliva collection tubes werealso sent to the study participants. The saliva samples were returnedin the mail along with the surveys to the study center and werefrozen at −20 °C upon receipt. The samples were tested for salivary

624 N. Mushtaq et al. / Addictive Behaviors 39 (2014) 622–629

cotinine in duplicate using a highly-sensitive enzyme immunoassay bySalimetrics, LLC.

2.3. Measures

Participants completed a self-administered survey which includedthe following measures:

- ModifiedWISDM-68: Wisconsin Inventory of Smoking DependenceMotives Inventory was modified to be used for ST dependence.Participants responded to 68 questions using 7-point Likert scalesranging from “not true of me at all” to “extremely true of me”.

- Tobacco Dependence Screener (TDS): TDS is a screening question-naire which assesses DSM-IV and ICD10 criteria of nicotinedependence. It has good psychometric properties and has beenused in cigarette smoking studies to measure nicotine dependenceaccording to the psychiatric diagnostic criteria (Baker et al., 2007;Jones et al., 2009; Piper, McCarthy, et al., 2008). TDS was modifiedto be used for ST dependence as the references for smoking werechanged to smokeless tobacco use. This questionnaire includes 10items with dichotomous (yes/no) response. An individual was iden-tified asnicotinedependent if he/she answered yes tomore thanfiveof the questions (Kawakami, Takatsuka, Inaba, & Shimizu, 1999).

- Fagerström Test of Nicotine Dependence for Smokeless Tobacco(FTND-ST): The six-item FTND-ST has been commonly used intobacco dependence research. It provides a continuous measure ofnicotine dependence. A higher score indicates stronger dependence.

- Tobacco use characteristics: Participants also provided informationrelated to past cigarette smoking and current ST use. A smokinghistory questionnaire included items such as smoking status (daily,occasional, or non-smoker), age of initiation, whether smoked 100cigarettes in lifetime, number of cigarettes smoked per day, age ofquitting smoking, and quit method. The ST use questionnaireassessed characteristics such as age of initiation, age of regularuse, ST use status (daily or someday), quantity (can or pouch perweek), frequency (number of dips/chews per day), and type(e.g., chewing tobacco, moist snuff, snus) of ST use, lifetime quitattempts, and number of quit attempts during the last 12 months.

- A sociodemographic questionnaire acquired information such asage, gender, race, ethnicity, level of education, and income.

2.4. Statistical analyses

Exploratory analysis of the variables was performed to obtaindescriptive statistics. Diagnosis of nicotine dependence as measuredby TDS was used to categorize participants as dependent or non-dependent. Association of tobacco use characteristics with TDS baseddiagnosis was assessed by usingχ2 test of independence for categoricalvariables and t-tests for continuous variables.

Statistical analyses for the newST scalewere conducted in two steps.First we determined the number of factors to be retained in the ST-modified WISDM scale from the original 68 items. Then, we assessedthe psychometric properties of the new scale.

2.4.1. Development of the Oklahoma Scale for Smokeless TobaccoDependence (OSSTD)

To identify motives of ST dependence we analyzed the ST-modifiedWISDM-68 subscales as described by Piper et al. (2004). Correlationsbetween individual subscales and TDS-based diagnosis of nicotinedependence were calculated to identify subscales to be retained for ex-ploratory factor analysis (EFA). Subscales having statistically significantcorrelation with TDS-based dependence diagnosis were used for EFA.

2.4.1.1. Exploratory factor analysis. The second set of analyses was per-formed to determine the motives of ST dependence. Results of latentclass analyses and factor mixture models of WISDM had previously

identified two categories of the subscales in Brief WISDM (Piper, Bolt,et al., 2008). Primary Dependence Motives (PDM) comprises four sub-scales: Automaticity, Tolerance, Loss of control, and Craving; whereas,Secondary Dependence Motives (SDM) consists of the remaining ninesubscales. We examined the individual dimensionality of PDM andSDM scales of the OSSTD. Exploratory factor analysis was performedwith SAS v9.2 to examine loading pattern and cross loading ofitems. Two separate EFAs were computed to obtain factor loadings forsubscales contributing to PDM and SDM scales. We used principal com-ponentmethod followed by orthogonal rotation. Eigenvalue criteria of 1and scree plotswere used to select the number of factors. Factors havingless than three significant items were not selected.

2.4.1.2. Item selection analyses. Item selectionwas based on the results ofEFA and item-total correlation. Itemswere qualitatively analyzed for theclarity of relationship with the subscale and to discriminate items fromeach other. Following the approach used to develop the Brief WISDM,and given the importance of the PDM scale, four items were selectedfor the subscales contributing to PDM, and three items were selectedfor the subscales contributing to SDM scale. The items were selectedbased on the highest factor loading, item-total correlation, and thecontent coverage.

2.4.2. Psychometric properties of OSSTDThe psychometric properties of the new scale were evaluated by

examining its validity and reliability.

2.4.2.1. Reliability. To assess the reliability of OSSTD, internal consistencywas measured by Cronbach's coefficient alpha.

2.4.2.2. Validity. Criterion-based concurrent validity, construct validity,and convergent validity of the scale were assessed. Concurrent validityof OSSTD was validated against other measures of ST dependence, in-cluding FTND-ST and TDS. Similarly, salivary cotinine levels were usedas a primary criterion variable to assess concurrent validity. Correlationcoefficientswere calculated and regression analyseswere conducted forthe validation of the scale. Confirmatory factor analysis (CFA) was usedfor construct validity.

3. Results

Although recruitmentwasnot limited tomale ST users, all studypar-ticipants weremale. Study participants were also predominantlyWhite(93%)with amean age of 31.85 years (SD 12.08). About one-third (34%)of the respondents were college graduates and 65% had some college orhigh school education. 68% of the participantswere ex-smokers and 47%(n = 32) of ex-smokers used ST to quit smoking. Themajority of the STusers (91%) were daily users. A higher proportion of ST users (73%)reported ever quitting ST in the past as compared to participants whotried to quit during the past 12 months (34%). The mean age at regularuse of ST products was 18.7 years (standard deviation 7.6). Years of STuse was significantly higher for those who had TDS based diagnosis ofdependence (16.2 vs. 10.2 years) compared to those of non-dependentST users. The mean cotinine level was 480.30 ng/ml with a standarddeviation of 415.30. The median cotinine level was 350.53 ng/ml(min = 15.5, max = 1772.07). Sociodemographic and tobacco usecharacteristics of the study participants are reported elsewhere(Mushtaq, Beebe, & Vesely, 2011).

3.1. Evaluation of ST-modified WISDM-68

We evaluated the 13 subscales of ST-modified WISDM-68 as a firststep to identify motives associated with ST dependence. Correlationbetween TDS based diagnosis and WISDM-68 subscales revealed that11 subscales had statistically significant associationwith ST dependencediagnosis (r = 0.36 to 0.56). However two subscales related to SDM,

625N. Mushtaq et al. / Addictive Behaviors 39 (2014) 622–629

Social/environmental goads (r = −0.108, p = 0.286) and Taste/sensory processes (r = 0.084, p = 0.405) were not significantlycorrelated with TDS based diagnosis. Therefore, these two motives,which included 10 items, were excluded from subsequent analyses.

3.2. Exploratory factor analysis

Exploratory factor analyses were performed separately for PDMbased items (n = 18) and for the items (n = 40) constituting SDM.For the PDM subscale 18 items were analyzed and two factors wereyielded. These factors combined accounted for approximately 84% ofthe total variance. While evaluating the rotated factor pattern, wefollowed the criteria of a minimum factor loading of 40 for an item tobe retained (Hatcher, 1994). This underlying motive was labeled as“Loss of control or craving” because seven items loaded on the first fac-torwere originally classified into ‘Loss of control’ and ‘Craving’ subscalesof WISDM-68. Similarly seven items loaded on the second factor had afactor score of N0.4 and did not cross load on other factors. As theseitems belonged to two subscales, Tolerance and Automaticity ofWISDM-68, we subsequently labeled this factor as “Tolerance andautomaticity”.

We extracted a five-factor solution for the SDM related 40 items thataccounted for 84% of the total item variance. Each factor had a varyingnumber of significant items (4 to 9 items) that had a factor score ofN40 and did not cross load on other factors. The five factors extractedclosely corresponded to the five subscales identified in the BriefWISDM (Smith et al., 2010). These were strong factors with a meaning-ful number of items (ranging 4 to 9 items) based on the criteria of aminimum factor loading of 0.4 and non-cross loading items. The firstfactor, Affective enhancement, appeared to be a combination of theitems of WISDM-68 subscales Positive reinforcement and Negativereinforcement. Likewise the second factor, “Affiliative attachment”was represented by Melioration and Affiliative attachment subscales.

Table 1Item-total correlations and factor scores of OSSTD items (retained after EFA).

Item no.a Subscales and items

Positive dependence motives Loss of control/craving1 Chew/dip controls me.

12 I'm really hooked on chew/dip.4 It's hard to ignore an urge to chew/dip

10 I frequently crave chew/dip.Tolerance/automaticity16 Other chewers/dippers would conside18 I chew/dip within the first 30 min of a13 I find myself reaching for chew/dip wi19 Sometimes I am not aware that I am c

Secondary dependent motives Affective enhancement2 Chewing/dipping improves my mood.

21 Chewing/dipping really helps me feel b22 Chewing/dipping makes me feel good.Affiliative attachment7 Chew/dip keeps me company, like a cl

15 I would feel alone without my chew/d3 Very few things give me pleasure each

Cognitive enhancement5 I chew/dip when I really need to conce9 Chewing/dipping helps me stay focuse

20 Chewing/dipping helps think better.Weight control6 I rely upon chewing/dipping to contro

11 Weight control is a major reason that I23 Chewing/dipping keeps me from overCue exposure8 There are particular sights and smells

14 I crave chew/dip at certain times of th17 Some things are very hard to do witho

Factor loading values are multiplied by 100 and rounded to the nearest integer. Item-total corra Item numbers from original WISDM-68.

The remaining three subscales were identified as Cognitive enhance-ment, Weight control, and Cue exposure.

3.3. Item selection

We selected four items for each of the two PDM related subscales.Items selected for Loss of craving & Tolerance subscales had factor load-ings ranging from 73 to 78 and significant item-total correlations(r = 0.617 to 0.752). For the Tolerance and automaticity subscale,two items with the highest factor scores were similarly worded; “Ifrequently chew/dip without thinking about it” and “I find myselfreaching for chew/dip without thinking about it”. Therefore, we select-ed the latter as it had a higher factor loading (72 vs. 76). The minimumfactor loading of an item in this scale was 64, and item-total correlationof the items in this subscale was from 0.418 to 0.616.

For the SDM related subscales, we selected three items for each ofthe five subscales. Affective enhancement had two items (factor scores79 and 73) from the Negative reinforcement subscale and one item(factor score 74) from the Positive reinforcement scale of WISDM-68.For the Affiliative attachment subscale, two items “Sometimes I feellike chew/dip are my best friends” and “Chew/dip keeps me companylike a close friend” were similarly worded. We selected the latter as ithad better factor loading (82 vs. 85). Two of the items in this subscalewere classified as Affiliative attachment in WISDM-68, however, thethird item, “Very few things give me pleasure each day like chewing/dipping” was a part of Melioration subscale. The remaining three SDMrelated subscales were also carefully examined to maintain the contentand construct of the subscale. Factor loadings and item-total correla-tions of the items are summarized in Table 1. Based on EFA two synthet-ic scales, PDM and SDM had two and five subscales respectively.

The OSSTD had seven subscales and 23 items. The total dependencescore for an individual was calculated as the sum of the means of theseven subscales. The mean score was 26.53 (SD = 8.44). The scoresfor the subscales were calculated by computing the mean of the items

Factor loading Item-total correlation

78 0.63073 0.64077 0.69675 0.760

r me a heavy chewer/dipper. 64 0.563wakening in the morning. 69 0.437thout thinking about it. 74 0.590hewing/dipping. 69 0.643

79 0.710etter if I've been feeling down. 73 0.717

74 0.661

ose friend. 85 0.639ip. 83 0.626day like chewing/dipping. 70 0.678

ntrate. 83 0.637d. 87 0.622

80 0.645

l my hunger and eating. 75 0.567chew/dip. 84 0.390eating. 81 0.573

that trigger strong urges to chew/dip. 70 0.513e day. 51 0.514ut chewing/dipping. 49 0.683

elations were significant with p b 0.0001.

Table 4Zero-order correlation between OSSTD and its subscales with other measures ofdependence (concurrent validity).

Measures TDS score FTND-ST Dependence diagnosis(TDS yes/no)

Table 2Descriptive statistics for OSSTD and its subscales.

Scale/subscale Mean (SD) Median (min–max)

OSSTD 26.53 (8.44) 27.12 (8.25–46.75)PDMa 4.04 (1.48) 3.87 (1.37–7.00)Loss of control/craving 4.30 (1.65) 4.5 (1.00–7.00)Tolerance/automaticity 3.77 (1.67) 3.50 (1.00–7.00)SDMb 3.69 (1.23) 3.70 (1.00–6.60)Affective enhancement 4.31 (1.70) 4.33 (1.00–7.00)Affiliative attachment 2.83 (1.73) 2.33 (1.00–7.00)Cognitive enhancement 4.46 (1.90) 4.67 (1.00–7.00)Weight control 2.60 (1.58) 2.33 (1.00–6.67)Cue exposure 4.27 (1.57) 4.33 (1.00–7.00)

Minimum and maximum possible scores of each subscale range from 1 to 7.a

PDM ¼ Loss of control cravingþ Tolerance automaticity2

:

b

SDM ¼ Affective enhancementþ Affiliative attachmentþ Cognitive enhancementþWeight controlþ Cure exposure5

:

626 N. Mushtaq et al. / Addictive Behaviors 39 (2014) 622–629

constituting the scale;whereas, PDMand SDMscoreswere themeans ofthe subscales representing the corresponding scale. Table 2 summarizesthe descriptive statistics of OSSTD.

3.4. Psychometric properties of the new scale

3.4.1. ReliabilityWe computed internal consistency for OSSTD and its subscales.

Cronbach's alpha coefficient for OSSTDwas 0.925. Although the internalconsistency of the PDM and SDM scales was lower than the overallOSSTD scale (α = 0.874 and 0.894), still it was higher than that ofFTND-ST (0.696) in our sample. Internal consistency for each of thesubscales was more than 0.86 except for the Cue exposure subscale(α = 0.662) and the Tolerance and automaticity subscale (α = 0.785).

All of the item-total correlations were significant and rangedbetween 0.517 and 0.787. Table 3 provides the correlation matrixshowing that the subscales were significantly intercorrelated, and thecorrelations ranged from 0.268 to 0.608.

3.4.2. ValidationConcurrent validity of the OSSTD was assessed by examining corre-

lations with dependence diagnosis (as assessed by TDS) and FTND-ST.There was a statistically significant correlation between OSSTD andTDS dependence diagnosis (r = 0.539, p b 0.0001) and FTND-ST(r = 0.515, p b 0.001). Similarly, all subscales had significant correla-tion with TDS dependence diagnosis, but for FTND-ST the Affectiveenhancement subscale did not have a significant correlation(r = 0.160, p = 0.1112). Table 4 shows correlations between subscalesand other ST dependence measures.

Table 3Zero-order correlation with p values of OSSTD subscales.

Subscale 1 2 3 4 5 6 7

1. Loss of control & craving 12. Tolerance &automaticity

0.593 1b .0001

3. Affective enhancement 0.589 0.310 1b .0001 0.0017

4. Affiliative attachment 0.593 0.533 0.509 1b .0001 b .0001 b .0001

5. Cognitive enhancement 0.323 0.319 0.606 0.268 10.0010 0.0012 b .0001 0.0070

6. Weight control 0.278 0.335 0.374 0.362 0.298 10.0051 0.0007 0.0001 0.0002 0.0026

7. Cue exposure 0.608 0.468 0.471 0.386 0.475 0.318 1b .0001 b .0001 b .0001 b .0001 b .0001 0.0013

Results of univariate logistic regression analysis showed that thesubscales were significantly associated with the dependence diagnosis(Table 5). Multiple regression analyses in which all subscales were en-tered as predictors, revealed that these subscales accounted for approx-imately 41% of the TDS score. Conversely, only 18% of the variation inTDS score was explained by FTND-ST.

3.4.3. CotinineBiological markers have consistently been used to evaluate the

validity of various tobacco dependence measures. We used salivarycotinine levels as the criterion variable to assess the validity of theOSSTD. Cotinine values tended to be skewed; therefore, square roottransformation was applied to the data. There was a significant correla-tion between the transformed cotinine variable and OSSTD (r = 0.267,p = 0.009). Although therewas no significant difference inmean cotin-ine levels for thosewho have TDS based dependence diagnosis (yes/no)and those who did not have dependence (p = 0.062), the TDS scorehad a significant correlation (r = 0.233, p = 0.023), and FTND-STwas more strongly associated (r = 0.613, p b 0.0001) with cotinineconcentration. Table 6 shows the results of simple linear regressionanalysis. Three subscales, Loss of control & craving, Tolerance& automa-ticity, and Affiliative attachment, had significant positive associationwith cotinine.

OSSTD 0.594 0.515 0.539b .0001 b .0001 b .0001

FTND-ST 0.430 1 0.401b .0001 b .0001

Loss of control & craving 0.585 0.530 0.531b .0001 b .0001 b .0001

Tolerance & automaticity 0.499 0.765 0.439b .0001 b .0001 b .0001

Affective enhancement 0.450 0.160 0.389b .0001 0.1112 b .0001

Affiliative attachment 0.389 0.394 0.352b .0001 b .0001 0.0003

Cognitive enhancement 0.320 0.204 0.3120.001 0.0422 0.0016

Weight control 0.292 0.201 0.2650.0032 0.0444 0.0076

Cue exposure 0.451 0.339 0.418b .0001 0.0006 b .0001

Table 7Correlation between OSSTD and its subscales with tobacco use characteristics.

Scale Correlation coefficient r (p value)

Cans/pouchesper week

Dips/chewper day

Years of ST use

OSSTD 0.294 (0.0031) 0.281 (0.0047) 0.297 (0.0027)Loss of control & craving 0.250 (0.0123) 0.381 (b .0001) 0.427 (b .0001)Tolerance & automaticity 0.541 (b .0001) 0.458 (b .0001) 0.447 (b .0001)Affective enhancement −0.012 (0.9053) 0.112 (0.2668) 0.089 (0.3796)Affiliative attachment 0.075 (0.4560) 0.145 (0.1508) 0.359 (0.0002)Cognitive enhancement 0.181 (0.0718) 0.043 (0.6689) −0.068 (0.5029)Weight control 0.186 (0.0636) 0.139 (0.1683) 0.124 (0.2179)Cue exposure 0.261 (0.0088) 0.146 (0.1468) 0.136 (0.1785)

Table 5Association between dependence diagnosis (based on TDS as yes/no) and OSSTDsubscales.

Subscales Odds ratios (95%CI)

Loss of control & craving 2.316 (1.627–3.297)Tolerance & automaticity 1.839 (1.375–2.465)Affective enhancement 1.675 (1.271–2.205)Affiliative attachment 1.593 (1.208–2.099)Cognitive enhancement 1.424 (1.131–1.792)Weight control 1.433 (1.090–1.884)Cue exposure 1.852 (1.357–2.526)OSSTD score 1.188 (1.106–1.277)

627N. Mushtaq et al. / Addictive Behaviors 39 (2014) 622–629

The OSSTD had significant correlation with all three components ofST use characteristics, frequency (Chews/dips per day), quantity(Cans/pouches per week), and duration (Years of ST use) (Table 7).The quantity of ST usewas significantly correlated with three subscales,Cue exposure (r = 0.261, p = 0.009), Tolerance & automaticity(r = 0.541, p b 0.0001), and Loss of control & craving (r = 0.250,p = 0.012). Both PDM based subscales had significant correlationwith the frequency of ST use. Duration of ST use was associated withsubscales Loss of control & craving (r = 0.427, p b 0.0001), Tolerance& automaticity (r = 0.447, p b 0.0001), and Affiliative attachment(r = 0.359, p = 0.0002).

Regression analyses were also conducted to identify the subscalesbest predicting themaximumvariance in these tobacco use characteris-tics. Stepwise selection procedure was used to acquire a parsimonioussolution. Four subscales, Loss of control & craving, Tolerance & automa-ticity, Affiliative attachment, and Cue exposure were the best predictorsof ST use frequency (r = 0.53, p b 0.0001). Tolerance & automaticityand Affiliative attachment accounted for 36% of the variance in STuse quantity (r = 0.596, p b 0.0001), and Loss of control & craving,Tolerance & automaticity, and Cue exposure explained 31% of thevariation in ST use duration (r = 0.555, p b 0.0001).

4. Discussion

This study describes the development and validation of a new scaletomeasure nicotinedependence among smokeless tobaccousers. A the-ory driven multidimensional approach was used to develop this scalewhich was analytically validated by following a data driven approach.OSSTD has improved reliability as compared to the previously studiedmeasures of ST dependence. Psychometric analysis of the OSSTDdemonstrated superior concurrent validity and construct validity.These analyses also established that the subscales of the OSSTD haveindependent and incremental associations with traditional dependenceconstructs, such as cotinine levels and tobacco use characteristics.

4.1. Structure and dimensions of OSSTD

The primary goal of this study was to develop a multidimensionalscale tomeasure ST dependence.We adapted theWISDM-68, amultiplemotive measure of smoking dependence, to identify seven factors

Table 6Linear regression results using new ST scale and subscales as predictor of cotinine levelsa.

Measures β SE t value p

Loss of control & craving 2.411 0.51480 4.68 b .0001Tolerance & automaticity 2.768 0.486 5.69 b .0001Affective enhancement 0.455 0.566 0.81 0.4197Affiliative attachment 1.412 0.533 2.65 0.0095Cognitive enhancement −0.608 0.499 −1.22 0.2255Weight control 0.527 0.607 0.87 0.3874Cue exposure 0.398 0.598 0.67 0.5070OSSTD score 0.280 0.108 2.58 0.0114

a Square root transformation of cotinine data.

which describe continual use of ST. It is an important finding that unlikedependence among smokers, Social and environmental goads and Tasteand sensory processes did not contribute to ST dependence. In theWISDM-68 study by Piper et al. the Social and environmental goadssubscale had the lowest number of significant correlations with theother subscales and was not significantly associated with FTND amongcigarette smokers (Piper et al., 2004).

The development of the OSSTD identified seven latent constructs in-cluding 23 items to measure ST dependence. Due to the distinct natureof primary dependence motives and secondary dependence motives,items were divided into two categories, PDM and SDM, for performingfactor analysis. We found Loss of control and Craving subscales to be asingle factor; similarly, Automaticity and Tolerance subscales were notseparable. Although SDM based subscales were closely correspondingto the subscales of WISDM-68, the items related to Negative reinforce-ment and Positive reinforcement subscales tap the same motive of Af-fective enhancement. These were combined, rather than assessing useto enhance pleasure differently from use to avoid negative internalstates. Similarly, items originally classified into two subscales, Affiliativeattachment and Behavioral choice–melioration subscales, explained asingle construct among ST users. These findings were consistent withthe results of Brief-WISDM for smokerswhich consolidatedNegative re-inforcement and Positive reinforcement subscales (Smith et al., 2010).However, unlike Brief-WISDM we did not completely exclude theMelioration subscale. One of the items in the Melioration subscale ofthe WISDM-68 was included in Affiliative attachment subscale of theOSSTD.

Nicotine dependence was initially predicted to be unidimensional;however later studies of smoking dependence measures, such asWISDM, revealed that multiple factors are required to explain thecovariation among the sets of indicators of the construct (Piper,McCarthy, et al., 2008; Piper et al., 2004).We examined the dimensionalstructure of ST dependence.We compared a single factormodel, assum-ing a unitary construct of ST dependence, with a two factor model,considering PDM and SDM as two constructs. The two factor model bet-ter fits the data as compared to the single factor solution. Based on theseresults and that of the previousWISDM based studies (Piper, Bolt, et al.,2008; Smith et al., 2010; Vajer, Urban, Tombor, Stauder, & Kalabay,2011), we independently analyzed PDM based items and SDM relatedOSSTD items. Our findings indicated that ST dependence as measuredby OSSTD is amultidimensional trait which is explained by distinct mo-tives. Our aim of performing CFA was not to identify the best structureor to revise the contents, but merely to validate the theory drivenmultiplemotive approach and the latent constructs as identified by EFA.

If multiple factors or subscales are intended to measure a commonoutcome then these factors should be associated with each other, butat the same time, the correlation between factors should not be toostrong to cost the distinction between these subscales. We observedmoderate correlations between the OSSTD subscales (r = 0.268 to0.606). There were a few redundancies in WISDM-68; therefore, wecarefully evaluated each item for its content and interpretation while

628 N. Mushtaq et al. / Addictive Behaviors 39 (2014) 622–629

selecting those items to be included in the OSSTD.We avoided selectingsimilarly worded items for the OSSTD. Exclusion of similar items from aquestionnaire potentially compromises its internal consistency. Howev-er, all the subscales of the OSSTD had acceptable internal consistency(α ≥ 0.86) with an exception of the Cue exposure subscale which hadinternal consistency of 0.66. Although these alpha coefficients of thesubscales were lower than theWISDM-68 subscales, they were compa-rable to the internal consistency of the Brief-WISDM subscales whichhad minimum internal consistency of 0.69 for the Cue exposuresubscale (Smith et al., 2010).

It is evident from the results that the OSSTD subscales have particu-lar contributions in measuring ST dependence. For instance, the diversepattern of relation between subscales and other dependence measuresand tobacco use indices highlights the existence of multidimensionalmotives of ST use and dependence (Piper, McCarthy, et al., 2008). Thecurrent study also found a varying degree of association between thesubscales and other criteria. All the ST use indicators were highlycorrelated with two subscales, Loss of control & craving and Tolerance& automaticity. The Cue exposure subscale predicted quantity of STuse, whereas the Affiliative attachment subscale was associated withduration of ST use.

4.2. Psychometric properties

Our study showed high reliability and construct validity ofOSSTD. Internal consistency as measured by Cronbach's coefficient(α = 0.925) indicated better reliability of OSSTD than FTND-ST.Although WISDM-68 had even higher internal consistency (α rangingfrom 0.97 to 0.99), a very high value of α suggests that some items aretesting that same question in a different diction. These results indicatethat the OSSTD has less redundancy than WISDM-68 (DeVon et al.,2007).

The concurrent validity of the OSSTD as evaluated by comparing itwith ST dependence diagnosis and FTND-ST was affirmative. Previousstudies of ST dependence scales did not find a significant associationbetween DSM-IV and FTQ-ST (Thomas et al., 2006). However, OSSTDhad a positive correlation with TDS based dependence diagnosiswhich was higher than the correlation between FTND-ST and depen-dence diagnosis. All the subscales showed positive correlation withdependence diagnosis and FTND-ST but the Affective enhancementsubscale was not associated with FTND-ST. These findings of significantcorrelation of subscales with criterion variables demonstrate conver-gent validity. While comparing with other criteria used in tobaccodependence studies, such as tobacco use characteristics, it is importantto note that OSSTD was consistently associated with the amount,frequency, and duration of ST use. When we explored this associationat the subscale level, the PDM based scales were better predictors ofthese variables. These results helped in establishing the validity of theOSSTD. Unlike FTND-ST and FTQ-ST which have items directly seekingthis information, the items on theOSSTDdo not query about the tobaccouse characteristics.

Cotininewasused as a criterion variable; previously conducted stud-ies of ST dependence have shown a modest correlation between cotin-ine levels and FTQ and FTND based dependence measures (Ferketichet al., 2007; Thomas et al., 2006). Our findings were consistent withthose studies, as OSSTD had a correlation of 0.27 with the cotininelevels. Relatively newer ST dependence scales, SSTDS and GN-STBQ,did not demonstrate a significant correlation with serum cotinine levels(Ebbert et al., 2012). Studies have demonstrated a relationship betweencotinine and the frequency and hours of ST use. Despite the fact that theOSSTD items do not query about the frequency or duration of ST use,finding a positive association with cotinine levels is promising. As ob-served with the tobacco use characteristics, PDM based subscales weresignificantly associated with cotinine levels, but SDM based subscale,Affiliative attachment, was also a significant predictor of cotinine levels.

While looking at these results and comparing the subscales as pri-mary dependence and secondary motives, one can argue the impor-tance of PDM based subscales and overlook the SDM related subscales.But we have observed unique and particular contributions of these sub-scales in addition to their incremental utility as the results of multipleregression analysis of subscales emphasized the role of one or more ofthese subscales in jointly predicting other criteria. These findings alsoendorse that ST dependence has two broader motives (PDM andSDM), and these motives are explained by multiple factors. Anotherreason for including these subscales was because our aimwas to identi-fymotives of ST dependencewhile preserving the theoretically rigorouscontent of WISDM-68.

In addition to following a comprehensive approach to developing aST dependence measure, we incorporated qualitative and quantitativemethods into the process resulting in better reliability and validity.Another strength of our study was the use of a sample of ST usersfrom the community, recruited without regard to their intentions toquit. Past research focused on tobacco dependence measures hasinvolved study subjects from treatment programs or clinical trials oftobacco cessation (Ebbert et al., 2006; Ferketich et al., 2007; Thomaset al., 2006). Such study participants are different from other tobaccousers as they are highly motivated to quit and are already in a processof behavioral change. This study provides insight into the motivesbehind ST dependence among ST users in the general population.

4.3. Limitations

There are a few limitations of this study. Although the sample size ofthe study was comparable to most of the previously conducted STdependence studies, it was smaller than the WISDM-68 based studiesof smoking dependence (Piper et al., 2004). Relatively smaller samplesize did not compromise the power of baseline analyses, such as reliabil-ity, correlation, simple linear regression, and other univariate analyses.But the results based on higher order analysis, especially factor analysis,should be interpretedwith caution because a sample with 100 observa-tions provides moderate power. Due to a smaller sample size multivar-iate analysis was not performed; specifically, study did not evaluate thepossible confounding of previous smoking behaviors. All the study par-ticipants were male, and they were predominantlyWhite non-Hispanicdaily ST users, similar to past studies of ST dependence. Due to thesesimilarities, the results of our study can be compared to the previousresearch, but these issues limit the generalizability of our findings.

Despite conducting a number of validation tests to determine thepsychometric properties of the new ST scale, we did not evaluate thepredictive validity of the scale. Predictive validity assessing cessationand withdrawal while abstinent is an important component for thevalidation of a dependence measure. Another reason for retaining amaximum number of motives (subscales) as mentioned above was toexamine the contributions of these subscales in the predictive validityof the new ST scale.

5. Conclusion

In conclusion, motives of tobacco dependence differ betweensmokers and ST users. This study is a first step towards measuring STdependence with a multiple motive approach. OSSTD, with betterpsychometric properties than previously used scales, provides a moreeffective and efficient tool to measure ST dependence as a multidimen-sional construct. Future research should be focused on establishing thepredictive validity of the OSSTD. This could include long-term predic-tion of abstinence, relapse, and severity of the withdrawal symptoms.

Role of funding sourcesFunding for this study was provided by the Oklahoma Tobacco Research Center

(OTRC). OTRC had no role in the study design, collection, analysis or interpretation ofthe data, writing the manuscript, or the decision to submit the paper for publication.

629N. Mushtaq et al. / Addictive Behaviors 39 (2014) 622–629

ContributorsNM: Conception and design, literature review, analysis, interpretation of results, and

drafting of the article.LAB: Conception and design, interpretation of results, and critical revision of the

manuscript for important intellectual content.SKV: Statistical analysis, interpretation of results, and critical revision of the

manuscript.BRN: Statistical analysis and revision of the manuscript.All authors contributed to and have approved the final manuscript.

Conflict of interestAll authors declare that they have no conflicts of interest.

AcknowledgmentsThe authors would like to thank Dr. Raymond Boyle for his helpful feedback on the

initial draft of the paper.This study was funded by the Oklahoma Tobacco Research Center.

References

Baker, T. B., Piper,M. E., McCarthy, D. E., Bolt, D.M., Smith, S. S., Kim, S. Y., et al. (2007). Timeto first cigarette in themorning as an index of ability to quit smoking: Implications fornicotine dependence. [Research Support, N.I.H., Extramural]. Nicotine & TobaccoResearch, 9(Suppl. 4), S555–S570, http://dx.doi.org/10.1080/14622200701673480.

Boyle, R. G., Jensen, J., Hatsukami, D. K., & Severson, H. H. (1995). Measuring dependence insmokeless tobacco users. Addictive Behaviors, 20(4), 443–450, http://dx.doi.org/10.1016/0306-4603(95)00013-3.

DeVon, H. A., Block, M. E., Moyle-Wright, P., Ernst, D.M., Hayden, S. J., Lazzara, D. J., et al.(2007). A psychometric toolbox for testing validity and reliability. Journal of NursingScholarship, 39(2), 155–164, http://dx.doi.org/10.1111/j.1547-5069.2007.00161.x.

Ebbert, J. O., Patten, C. A., & Schroeder, D. R. (2006). The Fagerstrom Test for NicotineDependence-Smokeless Tobacco (FTND-ST). Addictive Behaviors, 31(9), 1716–1721,http://dx.doi.org/10.1016/j.addbeh.2005.12.015.

Ebbert, J. O., Severson, H. H., Danaher, B. G., Schroeder, D. R., & Glover, E. D. (2012). A com-parison of three smokeless tobacco dependence measures. Addictive Behaviors,37(11), 1271–1277, http://dx.doi.org/10.1016/j.addbeh.2012.06.011.

Etter, J. F., Duc, T. V., & Perneger, T. V. (1999). Validity of the Fagerstrom Test For NicotineDependence and of the Heaviness of Smoking Index among relatively light smokers.Addiction, 94(2), 269–281, http://dx.doi.org/10.1046/j.1360-0443.1999.94226910.x.

Fagerstrom, K., & Eissenberg, T. (2012). Dependence on tobacco and nicotine products: Acase for product-specific assessment. [Research Support, N.I.H., Extramural]. Nicotine& Tobacco Research, 14(11), 1382–1390, http://dx.doi.org/10.1093/ntr/nts007.

Fagerstrom, K. O., & Schneider, N. G. (1989). Measuring nicotine dependence: A review ofthe Fagerstrom Tolerance Questionnaire. Journal of Behavioral Medicine, 12(2),159–182, http://dx.doi.org/10.1007/BF00846549.

Fant, R. V., Henningfield, J. E., Nelson, R. A., & Pickworth, W. B. (1999). Pharmacokineticsand pharmacodynamics of moist snuff in humans. Tobacco Control, 8(4), 387–392,http://dx.doi.org/10.1136/tc.8.4.387.

Ferketich, A. K., Wee, A. G., Shultz, J., & Wewers, M. E. (2007). A measure of nicotinedependence for smokeless tobacco users. Addictive Behaviors, 32(9), 1970–1975,http://dx.doi.org/10.1016/j.addbeh.2007.01.005.

Giovino, G. A., Henningfield, J. E., Tomar, S. L., Escobedo, L. G., & Slade, J. (1995). Epidemi-ology of tobacco use and dependence. Epidemiologic Reviews, 17(1), 48–65.

Hatcher, L. (1994). A step-by-step approach to using the SAS system for factor analysis andstructural equation modeling (1 ed.)Cary, NC, USA: SAS Publishing.

Hatsukami, D. K., Gust, S. W., & Keenan, R. M. (1987). Physiologic and subjective changesfrom smokeless tobacco withdrawal. Clinical Pharmacology and Therapeutics, 41(1),103–107, http://dx.doi.org/10.1038/clpt.1987.17.

Heatherton, T. F., Kozlowski, L. T., Frecker, R. C., Rickert, W., & Robinson, J. (1989). Measur-ing the heaviness of smoking: Using self-reported time to the first cigarette of the dayand number of cigarettes smoked per day. British Journal of Addiction, 84(7), 791–799,http://dx.doi.org/10.1111/j.1360-0443.1989.tb03059.x.

Henningfield, J. E., Radzius, A., & Cone, E. J. (1995). Estimation of available nicotine contentof six smokeless tobacco products. Tobacco Control, 4, 57–61, http://dx.doi.org/10.1136/tc.4.1.57.

Jones, H. E., Heil, S. H., O'Grady, K. E., Martin, P. R., Kaltenbach, K., Coyle,M. G., et al. (2009).Smoking in pregnant women screened for an opioid agonist medication studycompared to related pregnant and non-pregnant patient samples. [ComparativeStudy Randomized Controlled Trial Research Support, N.I.H., Extramural].American Journal of Drug and Alcohol Abuse, 35(5), 375–380, http://dx.doi.org/10.1080/00952990903125235.

Kawakami, N., Takatsuka, N., Inaba, S., & Shimizu, H. (1999). Development of a screeningquestionnaire for tobacco/nicotine dependence according to ICD-10, DSM-III-R,and DSM-IV. Addictive Behaviors, 24(2), 155–166, http://dx.doi.org/10.1016/S0306-4603(98)00127-0.

Kozlowski, L. T., Porter, C. Q., Orleans, C. T., Pope, M.A., & Heatherton, T. (1994). Predictingsmoking cessation with self-reported measures of nicotine dependence: FTQ, FTND,and HSI. Drug and Alcohol Dependence, 34(3), 211–216, http://dx.doi.org/10.1016/0376-8716(94)90158-9.

Mushtaq, N., Beebe, L. A., & Vesely, S. K. (2011). Determinants of salivary cotinine concentra-tions among smokeless tobacco users. Nicotine & Tobacco Research, http://dx.doi.org/10.1093/ntr/ntr279.

Pinto, R. P., Abrams, D. B.,Monti, P.M., & Jacobus, S. I. (1987). Nicotine dependence and like-lihood of quitting smoking. Addictive Behaviors, 12(4), 371–374, http://dx.doi.org/10.1016/0306-4603(87)90052-9.

Piper, M. E., Bolt, D.M., Kim, S. Y., Japuntich, S. J., Smith, S. S., Niederdeppe, J., et al. (2008).Refining the tobacco dependence phenotype using the Wisconsin Inventory ofSmoking Dependence Motives. Journal of Abnormal Psychology, 117(4), 747–761,http://dx.doi.org/10.1037/a0013298.

Piper, M. E., McCarthy, D. E., Bolt, D.M., Smith, S. S., Lerman, C., Benowitz, N., et al. (2008).Assessing dimensions of nicotine dependence: An evaluation of the Nicotine Depen-dence Syndrome Scale (NDSS) and the Wisconsin Inventory of Smoking DependenceMotives (WISDM). Nicotine & Tobacco Research, 10(6), 1009–1020, http://dx.doi.org/10.1080/14622200802097563.

Piper, M. E., Piasecki, T. M., Federman, E. B., Bolt, D.M., Smith, S. S., Fiore, M. C., et al.(2004). A multiple motives approach to tobacco dependence: The Wisconsin Inven-tory of Smoking Dependence Motives (WISDM-68). Journal of Consulting andClinical Psychology, 72(2), 139–154, http://dx.doi.org/10.1037/0022-006X.72.2.139.

Robertson, P. B., DeRouen, T. A., Ernster, V., Grady, D., Greene, J., Mancl, L., et al. (1995).Smokeless tobacco use: How it affects the performance of major league baseballplayers. Journal of the American Dental Association, 126(8), 1115–1121 (Discussion1121–1114).

SAMHSA (2010). Results from the 2009 National Survey on Drug Use and Health:Detailed tables. Rockville, MD. (Retrieved from). http://www.oas.samhsa.gov/NSDUH/2k8NSDUH/2k8results.cfm#Ch4

Smith, S. S., Piper,M. E., Bolt, D.M., Fiore,M. C.,Wetter, D.W., Cinciripini, P.M., et al. (2010).Development of the Brief Wisconsin Inventory of Smoking Dependence Motives.Nicotine & Tobacco Research, 12(5), 489–499, http://dx.doi.org/10.1093/ntr/ntq032.

Thomas, J. L., Ebbert, J. O., Patten, C. A., Dale, L. C., Bronars, C. A., & Schroeder, D. R. (2006).Measuring nicotine dependence among smokeless tobacco users. Addictive Behaviors,31(9), 1511–1521, http://dx.doi.org/10.1016/j.addbeh.2005.11.005.

USDHHS (1986). The health consequences of using smokeless tobacco. A report of the advi-sory committee to the Surgeon General. (NIH Publication No. 86-2874) (Bethesda,MD: Retrieved from http://profiles.nlm.nih.gov/ps/access/NNBBFC.pdf)

Vajer, P., Urban, R., Tombor, I., Stauder, A., & Kalabay, L. (2011). Psychometric propertiesand construct validity of the Brief Wisconsin Inventory of Smoking Dependence Mo-tives in an internet-based sample of treatment-seeking Hungarian smokers. Nicotine& Tobacco Research, 13(4), 273–281, http://dx.doi.org/10.1093/ntr/ntq254.