Multidimensional assessment of nicotine dependence in adolescents

Drug and Alcohol Dependence 77 (2005) 235–242

Multidimensional assessment of nicotine dependence in adolescents Duncan B. Clarka,∗ , D. Scott Wooda , Christopher S. Martina , Jack R. Corneliusa , Kevin G. Lynchb , Saul Shiffmanc a

Department of Psychiatry, Western Psychiatric Institute and Clinic, 3811 O’Hara Street, Pittsburgh, PA 15213, USA b University of Pennsylvania, 3900 Chestnut Street, Philadelphia, PA 19104-6178, USA c Department of Psychology, University of Pittsburgh, 130 N. Bellefield Avenue, Pittsburgh, PA 15213, USA Received 5 February 2004; received in revised form 11 August 2004; accepted 13 August 2004

Abstract Despite the critical importance of adolescent smoking, the assessment of nicotine dependence during this developmental period has been the subject of relatively little research. In this study, 301 adolescents (ages 12 through 18 years) reporting daily smoking were recruited for a project on alcohol use disorders (AUDs). The sample included 140 females and 161 males, 251 subjects from clinical and 50 from community sources, and 176 subjects with AUDs at the baseline assessment. Subjects were evaluated with the Nicotine Dependence Syndrome Scale (NDSS), the Fagerstrom Test for Nicotine Dependence (FTND) and a determination of average number of cigarettes per day (cigarettes/day). A varimax factor analysis of 27 NDSS items revealed four factors: (1) Drive/Tolerance (13 items; Cronbach α = 0.91); (2) Continuity (five items; Cronbach α = 0.67); (3) Priority (three items; Cronbach α = 0.64); (4) Stereotypy (five items; Cronbach α = 0.66). The NDSS total score, refined by the removal of four items, was also examined (23 items; Cronbach α = 0.90). Predicting cigarettes/day at follow-up, initial smoking rate was the best predictor, with the FTND and NDSS Total score showing significant and similar predictive validity. The NDSS Total showed incremental validity in the prediction of smoking progression in a model including demographic characteristics, initial smoking rate and FTND. The findings suggest that the NDSS has acceptable psychometric properties when applied to adolescents, complementing smoking rate and FTND in a multidimensional smoking assessment. © 2004 Elsevier Ireland Ltd. All rights reserved. Keywords: Adolescents; Nicotine dependence; Assessment

1. Introduction Adolescence is a critical developmental period in understanding the progression from the initiation of cigarette smoking to nicotine dependence. While most adolescents experiment with cigarette smoking, a minority become regular smokers (Clark and Cornelius, 2004; Shadel et al., 2000). Of regular smokers in adolescence, however, most continue as regular smokers into adulthood and many develop the attendant health consequences (Chassin et al., 1996). Comprehensive research on the development of nicotine dependence in adolescence requires adequate measurement methods (Clark and Winters, 2002). The methodology for assessing nicotine ∗

Corresponding author. Tel.: +1 412 246 5186; fax: +1 412 246 6550. E-mail address: [email protected] (D.B. Clark).

0376-8716/$ – see front matter © 2004 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.drugalcdep.2004.08.019

dependence in adolescents lags behind methods for adults, with no clear consensus on the optimal measure (Colby et al., 2000a, 2000b). One of the most commonly used measures for assessing nicotine dependence is the Fagerstrom Tolerance Questionnaire (FTQ: Fagerstrom, 1978), modified as the Fagerstrom Test for Nicotine Dependence (FTND: Heatherton et al., 1991). The FTQ is a well-established and brief selfreport measure with demonstrated validity (Fagerstrom and Schneider, 1989). While the unmodified FTQ has been used in adolescent populations, psychometric problems including low internal consistency (Colby et al., 2000a) have led to several attempts to improve the measure. The FTND, which eliminates items including “usual brand of cigarettes smoked,” has been shown to be an improvement over the FTQ in both internal consistency and validity (Pomerleau et al.,

236

D.B. Clark et al. / Drug and Alcohol Dependence 77 (2005) 235–242

1994; Radzius et al., 2001). Nevertheless, the FTND remains a sub-optimal measure (Payne et al., 1994). Most critically, the FTND items are based on a narrow, unidimensional conceptualization of nicotine dependence (Colby et al., 2000a; Shiffman et al., 2004). Shiffman et al. (2004) have developed a multi-dimensional measure of nicotine dependence, the Nicotine Dependence Syndrome Scale (NDSS). Based on Edwards’ conceptualization of dependence (Edwards, 1986), an initial pool of questionnaire items was generated. In an initial sample of 317 adult smokers, psychometric analyses revealed five factors. The five NDSS factors are: (1) Drive, including compulsion to smoke, craving and withdrawal; (2) Tolerance, indicating reduced sensitivity to nicotine effects; (3) Continuity, indicating smoking regularity; (4) Priority, reflecting preference for smoking over other activities; and (5) Stereotypy, indicating a rigid pattern of cigarette use. These factors have been confirmed and refined in other adult samples (Shiffman et al., 2004). 1.1. Study aims The purpose of the current study was to examine the psychometric properties of the NDSS applied to an adolescent sample. The study examined NDSS factor structure, scale reliability, construct validity, predictive validity, and incremental validity in an adolescent sample.

2. Methods 2.1. Participants The sample was 301 adolescents (mean age 17.0 ± 1.3; range 12 through 18 years; 140 females, 161 males) with daily smoking recruited from clinical (n = 251) and community (n = 50) sources. The ethnic groups represented were European American (n = 259; 86%), African American (n = 29, 10%; and others (n = 13, 4%). By Hollingshead two-factor scale, socio-economic status (SES) groups were as follows: Class I: n = 21, 7%; Class II: n = 117, 39%; Class III: n = 85, 28%; Class IV: n = 44, 15%; Class V: n = 33, 11% (n.b., SES was missing in one subject). The sample was recruited to serve several projects on adolescent alcohol use disorders (AUDs). Recruitment was conducted to include adolescents with AUDs from treatment programs, adolescents with other diagnoses from clinical programs, and reference adolescents from community sources. Clinical sources included licensed drug and alcohol treatment sites and hospital-based psychiatric treatment facilities. Subjects from community sources were recruited by two methods: (1) In an investigator-initiated method recruiting a representative sample (i.e., Community A), adolescents were identified from the local geographic area through sampling frames purchased from two survey firms (n = 36); (2) Subjects were also recruited by advertisement (n = 14).

Cigarette smoking rates for the sample had a mean ± standard deviation of 16.9 ± 9.4 cigarettes per day at the initial assessment, 15.9 ± 9.1 at the 1-year follow-up point, and 15.4 ± 8.5 at the 3-year follow-up point. A more detailed description of the rationale, design, and recruitment for the larger study may be found in prior publications (Clark et al., 2003). The assessment battery including the NDSS was added to the assessment protocol for all visits in the course of an on-going longitudinal study. Follow-up (FU) assessments occurred at 1st, 3rd and 5th years after the baseline assessment. For these analyses, the first administration of the NDSS was utilized as the initial assessment, and subsequent assessments used for follow-up data. Subjects included here reported daily smoking at the time of the initial assessment. Daily smoking is considerably more likely than is less frequent smoking to be characterized by nicotine dependence (Mayhew et al., 2000). 2.2. Measures 2.2.1. Nicotine dependence syndrome scale The NDSS version administered in this study included 24 items described in the original reports (Shiffman et al., 2004) and six additional exploratory items. The items were presented as a questionnaire with the following instructions: “Using the scale below, circle the corresponding number that indicates how well each of the following statements describes you.” Each item is answered on a five-point Likert-type scale with the following anchors: 1. Not at all true; 2. Somewhat true; 3. Moderately true; 4. Very True; 5. Extremely true. The NDSS items are provided in Table 1. By the Flesch–Kincaid method in SPSS, reading level for the NDSS was determined to be 7.6. Compared with the previously reported version (Shiffman et al., 2004), two items referencing quit attempts and an item requesting an estimate of the amount of money the individual would pay for a cigarette were not included. Added items in this version are 9, 13, 17, 20, 25, and 27 (see Table 1). Consequently, 27 items were examined here. Since nicotine dependence is generally considered to be a phenomenon occurring among daily smokers, the scale was considered to primarily apply to such individuals. Therefore, the NDSS was administered only to those individuals indicating daily smoking on a screening question. At the initial administration, the NDSS was missing items in 6 of 307 cases. The analyses were therefore based on 301 cases. 2.2.2. Fagerstrom test for nicotine dependence The FTND administered here included six items (Heatherton et al., 1991). The item topics and response options were as follows: (1) time from waking up to first cigarette (four responses ranging from after 61 min: zero points to within 5 min: three points); (2) difficult to refrain from smoking where forbidden (Yes: one point/No: zero points); (3) cigarette hate to give up most (first one in morning: one point/all others: zero points); (4) cigarettes per day (from 10 or less: zero points to 31 or more: three points); (5) smoke more frequently during first hours after waking

D.B. Clark et al. / Drug and Alcohol Dependence 77 (2005) 235–242

237

Table 1 NDSS factor scores in 301 adolescents NDSS item questions

Item

Mean ± S.D.

NDS factor

After not smoking for a while, I need to smoke in order to keep myself from experiencing any discomfort Whenever I go without a smoke for a few hours, I experience craving After not smoking for a while, I need to smoke to relieve feelings of restlessness and irritability Where regulations require that I go outdoors to smoke, it’s worth it to be able to smoke a cigarette, even in cold or rainy weather I rarely go for very long without smoking Compared to when I first started smoking, I need to smoke a lot more now in order to get what I really want out of it Compared to when I first started smoking I can smoke much, much more now before I start to feel nauseated or ill I can function much better in the morning after I’ve had a cigarette Sometimes even when I’m telling myself I’m not going to have a cigarette, I find myself smoking anyway I smoke consistently and regularly throughout the day When I’m really craving a cigarette, it feels like I’m in the grip of some unknown force that I cannot control If I wake up during the night, I feel I need a cigarette Not even a torrential rainstorm could stop me—if I were out of cigarettes, I would be immediately on my way to the store to get some more Sometimes, without realizing it, I go for several hours without smoking. (r) The number of cigarettes I smoke per day is often influenced by other factors—how I’m feeling, what I’m doing, etc. (r) My smoking pattern is very irregular throughout the day. It is not unusual for me to smoke many cigarettes in an hour, then not have another one until hours later. (r) I smoke at different rates in different situations. (r) It’s hard to estimate how many cigarettes I smoke per day because the number often changes. (r) I feel a sense of control over my smoking. I can “take it or leave it” at any time. (r) My cigarette smoking is fairly regular throughout the day I smoke about the same amount on weekends as on weekdays. I smoke just about the same number of cigarettes from day to day Since the time when I became a regular smoker, the amount I smoke has either stayed the same or has decreased somewhat My smoking is not much affected by other things. I smoke about the same amount whether I’m relaxed or working, happy or sad, alone or with others, etc. Sometimes, I decline offers to visit with my non-smoking friends because I know I’ll feel uncomfortable if I smoke I tend to avoid restaurants that don’t allow smoking, even if I would otherwise enjoy the food Even if traveling a long distance, I’d rather not travel by airplane because I wouldn’t be allowed to smoke

16

2.8 ± 1.2

D/T

24

3.1 ± 1.3

26

Drive/Tolerance

Continuity

Sterotypy

Priority

0.76

−0.05

−0.03

−0.19

D/T

0.76

0.09

−0.20

0.01

3.1 ± 1.2

D/T

0.76

−0.02

−0.14

−0.07

12

3.1 ± 1.2

D/T

0.72

0.02

−0.02

−0.07

13 10

2.7 ± 1.2 2.8 ± 1.3

D/T D/T

0.74 0.65

0.16 −0.16

−0.15 0.11

−0.12 −0.08

11

3.2 ± 1.4

D/T

0.65

−0.20

−0.13

−0.01

15

2.9 ± 1.3

D/T

0.67

0.05

−0.04

−0.25

19

3.0 ± 1.3

D/T

0.60

−0.11

−0.10

−0.17

8 21

2.8 ± 1.3 2.4 ± 1.2

D/T D/T

0.63 0.58

0.22 −0.13

−0.31 −0.04

−0.10 −0.31

14 4

2.3 ± 1.3 2.4 ± 1.4

D/T D/T

0.58 0.58

0.10 0.15

−0.05 −0.05

−0.41 −0.43

5

3.2 ± 1.3

C

−0.43

−0.56

−0.03

−0.06

20

2.8 ± 1.2

C

0.33

−0.71

0.13

0.15

1

3.5 ± 1.3

C

−0.21

−0.53

0.11

−0.05

9 17

2.9 ± 1.1 3.2 ± 1.3

C C

0.26 0.25

−0.56 −0.50

0.09 0.32

0.20 −0.10

18

3.8 ± 1.2

–

−0.42

−0.32

−0.13

−0.14

25 27

2.9 ± 1.1 2.7 ± 1.3

S S

0.46 0.17

0.19 0.16

−0.64 −0.53

−0.06 0.08

22

2.6 ± 1.2

S

0.28

0.35

−0.58

−0.09

23

2.3 ± 1.2

S

−0.20

−0.14

−0.44

0.03

2

2.5 ± 10.1

S

−0.01

0.10

−0.45

−0.30

6

1.7 ± 1.0

P

0.03

−0.11

−0.13

−0.75

7

1.9 ± 1.1

P

0.33

0.08

−0.09

−0.71

3

2.1 ± 1.3

P

0.43

0.15

−0.04

−0.51

Note: NDSS scale abbreviations: D/T = Drive/Tolerance; C = Continuity; S = Stereotypy; P = Priority. The above table presents the NDSS item, whether the item is reverse scored (r), the scale to which the item was assigned, and factor scores for each item.

238

D.B. Clark et al. / Drug and Alcohol Dependence 77 (2005) 235–242

(Yes: one point/No: zero points); (6) smoke if you are ill (Yes: one point/No: zero points). The FTND items and scoring were described in detail in the original report (Heatherton et al., 1991). In a recent analysis, Chabrol et al. (2003) found that deleting item three resulted in improved psychometric properties. Therefore, item three was not used and the FTND score was based on the five remaining items. With this refinement, FTND has improved but still problematic internal consistency (Cronbach α = 0.63), somewhat lower here than in a previously reported sample (i.e., Cronbach α = 0.86 in Chabrol et al., 2003). The FTND was missing in seven cases. 2.2.3. Cigarettes per day Subjects estimated the average number of cigarettes per day by phases using a method derived from the Skinner Lifetime Alcohol Use Interview (see Clark et al., 2001). At the baseline assessment, 12 month time periods are defined by birthday, and the time from the last birthday to the assessment day. At the follow-up assessments, time periods were from the day of the last assessment to the birthday, and from the last birthday to the day of the follow-up assessment. For these analyses, the most recent time period dating back from the day of the assessment was utilized. This measure was missing in one case. 2.3. Statistical analyses A confirmatory factor analysis was conducted to determine whether the factor structure for this sample fit that predicted by observations with adult samples. The confirmatory factor analysis was conducted with Mplus. Using a maximum likelihood estimator, the expected factor structure was specified by 23 items comprising the five-factor structure indicated in a previously reported adult sample (Shiffman et al., 2004). In the event that the proposed factor structure did not fit these data, an exploratory factor analysis was planned. The planned factor analysis used Mplus exploratory factor analysis with varimax rotation with the default extraction of un-weighted least squares. Several criteria were considered in determining the optimal number of factors: (1) eigenvalue >1.0; (2) scree plot indicating discontinuity between adjacent factor structures; (3) root mean square residual of <0.05; and (4) minimum of 5% variance explained by each factor. After constructing scales based on these properties, NDSS scales were compared with prior analyses. To the extent feasible, similar scales were labeled to represent constructs determined in prior analyses (i.e., Shiffman et al., 2004). NDSS scales were constructed by defining scale constructs according to the factor analysis dimensions, and assigning items to scales according to their highest factor loading. Items with factor loadings of less than 0.4 were excluded. NDSS scales were calculated by averaging the item scores. Internal consistency was examined for each scale using Cronbach’s alpha. In addition, the NDSS scales were examined for internal consistency for use in a total score. Items notably diminishing scale or total score reliability were deleted. To determine concur-

rent and discriminative validity, the NDSS scale scores and total were correlated with FTND, cigarettes per day, and each other. Predictive and incremental validity were determined by examining the initial assessment of cigarettes/day, FTND and NDSS as predictors of cigarettes/day at the subsequent follow-up assessment. Regression models were constructed with initial assessment of cigarettes per day, FTND, and NDSS Scales as independent and 1-year outcomes on cigarettes per day as dependent variables. A series of regression analyses were conducted to examine the incremental validity of the NDSS. With cigarettes per day at the follow-up assessment as the dependent variable, a multivariate model was also constructed with the following independent variables from the initial assessment entered in a stepwise fashion: Step 1: age, sex, ethnic group, SES, recruitment source; Step 2: cigarettes per day; Step 3: FTND and; Step 4: NDSS. Throughout the analyses, a significance threshold of p < 0.05 was used.

3. Results 3.1. Factor structure A confirmatory factor analysis was conducted using the 23 items and five-factor structure indicated in an adult sample (Shiffman et al., 2004). The proposed factor structure did not sufficiently fit these data, as indicated by a significant lack of model fit (χ2 = 379.4, d.f. = 42, p < 0.001). Therefore, other factor structures were explored. An exploratory factor analysis was conducted with unweighted least squares and varimax rotation. The 27 NDSS items available were included. Five factors had eigenvalues greater than 1.0 (i.e., 8.4, 3.3, 2.0, 1.6, and 1.1). This pattern, inspected with a scree plot, suggested a disjunction between the eigenvalues for factor four (i.e., 1.6) and the remaining factors. Four factors accounted for 57% of the variance (i.e., 31, 12, 8, and 6%) while the fifth factor accounted for only 4% of the variance. The root mean square residuals (RMSR) for these solutions were as follows: one factor: RMSR = 0.12; two factors: RMSR = 0.08; three factors: RMSR = 0.06; four factors: RMSR = 0.04; five factors: RMSR = 0.04. By the criteria indicated in Methods (Section 2), these results consistently selected four factors as accounting for the factor structure. Note that the results did not support the separation of Drive and Tolerance factors reported for adult samples. High internal consistency for items identified as a Drive/Tolerance factor (see below) further supported a single-factor structure for these items. The constructs represented by these factors included: (1) Drive/Tolerance (13 items: loadings 0.58–0.76); (2) Continuity (five items, loadings 0.50–0.56); (3) Stereotypy (five items, loadings 0.44–0.64); and (4) Priority (three items, loadings 0.51–0.75). The NDSS factor structure with item loadings is shown in Table 1. Note that all item factor load-

D.B. Clark et al. / Drug and Alcohol Dependence 77 (2005) 235–242

239

Table 2 NDSS scale internal consistency and correlations NDSS scales Drive/Tolerance

Continuity

Stereotypy

Priority

Total

Items (#) Cronbach’s α Item average (mean ± S.D.) Item sum (mean ± S.D.)

13 0.90 2.8 ± 0.9 36.5 ± 11.3

5 0.67 3.1 ± 0.8 14.5 ± 4.0

5 0.66 2.7 ± 0.9 12.9 ± 3.8

3 0.63 1.9 ± 0.9 5.7 ± 2.6

23 0.90 2.8 ± 0.7 63.3 ± 15.7

Correlations FTND Cigarettes/day

0.56 0.41

0.25 0.11

0.35 0.23

0.38 0.30

0.60 0.42

–

−0.01 –

0.42 0.32 –

0.49 0.06 0.28 –

0.95 0.19 0.60 0.60

Drive/Tolerance Continuity Stereotypy Priority

Note: all correlations are significant at p < 0.01 except those where r < 0.15 above. FTND: Fagerstrom test for nicotine dependence.

ings were over 0.4 for their primary factor. Note that five items showed factor loadings over 0.4 on secondary dimensions. 3.2. NDSS scales 3.2.1. Scale construction, reliability and stability Scales were constructed by assigning each item to the dimension with the highest loading according to the exploratory factor analysis, and averaging item scores. The mean and standard deviation for each item are provided in Table 1. Each of the resulting four scales included the number of items noted in the factor structure above. We elected to retain items with complex loadings. While eliminating such items would result in greater distinctions among the subscales, Reise et al., 2000 caution against aiming for mathematically-based simple structure solutions since psychosocial research often involves items that appropriately have complex factor loadings. Inclusion of complex items may improve the clinical validity and utility of the subscales. Descriptive statistics for NDSS scales are presented in Table 2. Internal consistency by Cronbach alpha was 0.90 for Drive/Tolerance, and below 0.7 for the other NDSS scales. To facilitate generalizability, item average scale scores rather than factor scores were used in subsequent analyses. Consequently, the range for each scale was one to five. Since smokers with AUDs, compared with smokers without AUDs, may have higher levels of nicotine dependence (Novy et al., 2001; Cornelius et al., 2001), the relationship among AUD and NDSS scores were examined. Controlling for cigarettes per day, demographic variables and recruitment source, there were no significant differences noted between adolescents with AUDs and adolescents without AUDs on NDSS total score (F = 2.8, p = 0.10) or subscales (Drive/Tolerance: F = 3.32, p = 0.07; Continuity: F = 0.00, p = 0.98; Priority: F = 0.04, p = 0.83; Stereotypy: F = 0.76, p = 0.38). The internal consistencies of NDSS items were also examined for use in a total score. Four NDSS items had item-total correlations of <0.1 and diminished alpha (i.e., items 9, 17, 20, and 23). With these items deleted, the NDSS Total score had excellent internal consistency.

Data from two administrations from 84 subjects with no change in cigarettes per day were compared. The test–retest interval was 1.8 ± 1.1 years. For NDSS total score, the two administrations were significantly correlated (r = 0.44, n = 80, p < 0.001; four cases with missing data) and were not significantly different (paired t = 0.9, d.f. = 79, p = 0.4). For a comparison, the FTND total score was similarly examined. The two FTND administrations were also significantly correlated (r = 0.59, n = 82, p < 0.001; two cases with missing data) and not significantly different (t = 0.1.0, d.f. = 81, p = 0.3). 3.2.2. Construct validity The correlations among NDSS scales and other nicotine dependence measures are presented in Table 2. FTND represents a general nicotine dependence measure, and significant correlations between NDSS and FTND therefore confirm that NDSS provides an assessment of the nicotine dependence construct. Since the NDSS scales represent nicotine dependence dimensions, significant correlations with FTND as well as cigarettes/day were expected. The NDSS scales and total score were all significantly correlated with FTND score and the estimate of cigarettes per day (i.e., smoking rate). Among NDSS scales, Drive/Tolerance was most correlated with FTND, whereas Stereotypy and Continuity were least correlated with FTND. NDSS Continuity showed nonsignificant correlations with cigarettes per day and NDSS Drive/Tolerance. Not represented in the table, FTND includes a smoking rate item, and therefore, a significant correlation with smoking rate would be expected and was seen here (r = 0.45, n = 293, p < 0.001). 3.3. Predictive and incremental validity Among the 301 subjects with complete NDSS data, 248 (82%) had a subsequent assessment. The follow-up (FU) assessment occurred after a period of 1.9 ± 1.2 years. Individual variables at the initial assessment were examined as predictors of cigarettes/day at follow-up. The initial assessment of cigarettes/day was significantly predictive of cigarettes per

240

D.B. Clark et al. / Drug and Alcohol Dependence 77 (2005) 235–242

Table 3 NDSS incremental validity regression models for predicting follow-up cigarettes per day: incremental validity Individual variable

Step 1 Sex Age SES Ethnic group

Overall model

β

t

p

0.2 0.06 −0.08 −0.1

3.6 1.0 −1.3 −1.9

<0.001 0.3 0.2 0.06

4.7

4,238

F change

d.f.

R2

R2 change

0.001

0.07

–

p

Step 2. Cigarettes per day

0.5

8.2

<0.001

66.8

1,237

<0.001

0.28

0.20

Step 3. FTND

0.2

3.6

0.001

11.5

1,236

0.001

0.31

0.03

Step 4. NDSS total

0.2

2.1

0.04

0.01

1,235

0.04

0.32

0.01

Note: At each step, variables in prior step are retained. The model change statistics represent incremental validity over the model with all prior variables. Standardized beta coefficients are reported.

day at FU (R2 = 0.23, F = 75.1, d.f. = 1246, p < 0.001), as were FTND (R2 = 0.18, F = 54.4, d.f. = 1241, p < 0.001) and the NDSS Total score (R2 = 0.15, F = 41.6, d.f. = 1246, p < 0.001). NDSS subscales as a group significantly predicted cigarettes/day at follow-up (R2 = 0.15, F = 10.7, d.f. = 1243, p < 0.001). In addition, each individual NDSS subscale significantly predicted cigarettes/day at follow-up (i.e., Drive/Tolerance: F = 36.8, p < 0.001; Continuity: F = 4.6, p = 0.03; Priority: F = 14.7, p < 0.001; Stereotypy: 4.4, p = 0.03). A regression model was constructed to determine the incremental validity of the NDSS. Given the superior psychometric characteristics of the NDSS total score and the lack of any observed advantage to the NDSS subscale scores in predictive validity, the NDSS total score was used. The results are presented in Table 3. After considering the variance accounted for by demographic characteristics, recruitment source, the initial assessment of cigarettes per day and the FTND, the NDSS Total added significantly to the prediction model.

4. Discussion These results support the reliability and validity of the NDSS for use in adolescents. The NDSS was developed to assess multiple dimensions derived from a contemporary conceptualization of nicotine dependence constructs. The factor analysis presented here indicates a four-factor structure was optimal for these data. The Drive and Tolerance dimensions observed to be distinct in adult samples were here found to form one dimension. With this exception, the factor structure was similar to that reported in adult samples. With nearly half of the NDSS items, the combined Drive/Tolerance scale showed a high degree of internal consistency and, among the NDSS scales, the highest correlation with both FTND and cigarettes/day. The content of these items included smoking to avoid withdrawal symptoms, craving, morning smoking, and increasing smoking to achieve similar effects. The results

suggest that these items reflect a dimension most akin to a general dependence construct as embodied in existing scales such as the FTND. In this regard, the construct validity of the Drive/Tolerance scale was supported. This merger of the Drive and Tolerance factors in this adolescent sample was unexpected. It could reflect the developmental trajectory of smoking and nicotine dependence. During early phases of smoking, craving/withdrawal may develop in tandem with increased smoking (tolerance), and thus be highly correlated. Such tight coupling is in fact predicted by theories that conceptualize withdrawal as the converse of tolerance – i.e., as revealing the neuroadaptation that underlies tolerance (Jaffe, 1989). Later, as smokers mature and progress, tolerance may asymptote – as indicated by the fact that smoking rates do not continue to rise over the entire lifetime – even as craving and withdrawal continue to increase, thus decoupling the two constructs and allowing the factors to separate in adult smokers. It should be noted that we did not test for a difference in NDSS factor structure between adults and adolescents; the different structure observed here could be due to other characteristics of our sample, and other differences between our sample and the samples in Shiffman et al. (2004). The remaining three NDSS scales represented Continuity, Priority and Stereotypy. These NDSS scales were also significantly correlated with FTND. In the case of Priority, item content overlap with FTND can be noted (i.e., both refer to situations where smoking is prohibited). The correlation between the NDSS Continuity Scale and FTND was lower; this may represent a dimension that was not tapped by the FTND. For these four NDSS scales, there may well be constructive uses not reflected in the rather narrow focus of this study. We tested the ability of the NDSS scales and total to predict subsequent smoking progression. The use of the four NDSS scales did not show an advantage over use of the NDSS as a single total score. Nevertheless, the predictive validity of the NDSS was similar to that of the FTND. We found that the NDSS total score significantly predicted subsequent smoking rate, even after demographic characteristics, recruitment source, baseline smoking rate and FTND were accounted

D.B. Clark et al. / Drug and Alcohol Dependence 77 (2005) 235–242

for. This suggests that early development of dependence may be an important driver of progression in smoking. This is consistent with theory, which would predict that development of dependence would put adolescents on a trajectory towards progression in smoking, while those who do not develop dependence would be less likely to progress. While the increment in predictive validity was modest, the NDSS significantly contributed to this prediction even after these other variables were accounted for suggesting the scale provided complementary information about nicotine dependence. In an adolescent application, the NDSS thus showed strengths and weaknesses. Used as a unidimensional scale, the NDSS total when compared with the FTND showed superior internal consistency, similar construct and predictive validity, and incremental validity after including the FTND in regression models predicting smoking progression. The NDSS total also showed a pattern of correlations supporting its construct validity. These strengths are notable in the context of the application of a scale designed for adults and applied to adolescents. The four NDSS scales showed some weaknesses. Three of the four NDSS scales had coefficient alphas of less than 0.70, indicating weak internal consistency (Cicchetti, 1994). Whether the instrument is best conceptualized and utilized as one or multiple scales remains a matter of interpreting the results of this and other analyses. The extent to which specific scales validly reflect labeled constructs, particularly for adolescents, will need to be considered in future studies. In addition, the test–retest reliability for the FTND was higher than that for the NDSS, suggesting the NDSS may be a somewhat less stable measure. The limitations of this study include a relatively small sample, the lack of a single systematic sampling method, and the absence of a confirmatory laboratory measure of nicotine use. The sample was relatively small for the purposes of establishing a stable factor structure, and the sample size diminished at follow-up. The generalizability of these results was further hampered by the recruitment of the majority of subjects from clinical programs. The clinical sample was collected for the most part to represent adolescents with AUDs. While selected as daily smokers, this sample had relatively high smoking rates for an adolescent sample. For other scales, the factor structure of nicotine dependence measures have been found to be unchanged by the presence of AUDs (e.g., reasons for smoking scale: Curie, 2004; FTND: Burling and Burling, 2003). Nevertheless, since the sample size here was not sufficiently large to formally explore the potential influence this characteristic on psychometric characteristics, this sample heterogeneity was a limitation of the study. More systematic sampling would have fostered generalizability. Confirmation of nicotine use with a laboratory measures such as salivary cotinine would have provided verification of nicotine use. Since the primary use of such confirmation is typically to detect nicotine use where abstinence is reported, however, laboratory results would likely have had limited utility here. This sample did not receive systematic smoking cessation inverventions, and the psychometric characteristics and ap-

241

plicability of the NDSS as a treatment outcome measure may be different from those observed here. These findings extend results with adult samples by indicating that the NDSS is a psychometrically sound nicotine dependence measure when applied to adolescents. While a preliminary effort has been made to develop a version of the NDSS specific to adolescence (Nichter et al., 2002), it remains to be seen whether a version of the NDSS developed specifically for adolescents will be psychometrically superior to this version. Nevertheless, should a specific version be developed that is clearly improved for this specific developmental period, the availability of a version applicable to both adolescence and adulthood would still have several advantages. The applicability of this version of the NDSS to adolescents and adults provides a version that could be utilized in mixed samples of adolescents and adults as well as in samples followed from adolescence to adulthood. Furthermore, this version would allow the direct comparison of adolescents and adults on NDSS dimensions. The results suggest that the NDSS complements smoking rate and FTND in comprising a multidimensional assessment of nicotine dependence for adolescents. The NDSS total score may be interpreted as reflecting a general nicotine dependence dimension. In the context of this and previous studies (Shiffman et al., 2004), the NDSS is best conceptualized as having a multidimensional structure with correlated factors reflecting a higher-order construct of dependence. Nonetheless, the psychometric properties of the four NDSS scales examined here need improvement. As has been previously suggested (Shiffman et al., 2004), future studies of the NDSS need to focus on: (1) the potential for refinements for a specific adolescent version of the NDSS to improve the psychometric properties of the measure; (2) examination of the utility of the NDSS for studying the development of nicotine dependence characteristics during the early stages of smoking; (3) relationship between NDSS and DSM-IV nicotine dependence characteristics; (4) construction of nicotine dependence profiles as phenotypes; and (5) prediction of smoking cessation intervention outcomes. While examination of the NDSS in a larger adolescent sample would extend these observations, the results suggest that refinements of the NDSS with specific modifications to enhance validity for adolescent samples may be needed.

Acknowledgments This work was conducted at the Pittsburgh Adolescent Alcohol Research Center and was supported by NIAAA (K02AA00291, K02-AA00249, P50-AA08746, R01-AA13397, R01-AA13370) and NIDA (R01-DA14635, P50-DA05605). A preliminary version of these results was presented at the College on Problems of Drug Dependence: Bal Harbor, FL 6/19/2003. For further information regarding further development of the NDSS contact Saul Shiffman at [email protected].

242

D.B. Clark et al. / Drug and Alcohol Dependence 77 (2005) 235–242

References Burling, A.S., Burling, T.A., 2003. A comparison of self-report measures of nicotine dependence among male drug/alcohol-dependent cigarette smokers. Nicotine Tobacco Res. 5, 625–633. Chabrol, H., Niezborala, M., Chastan, E., Montastruc, J.L., Mullet, E., 2003. A study of the psychometric properties of the Fagerstrom Test for Nicotine Dependence. Addict. Behav. 27, 1–5. Chassin, L., Presson, C., Rose, J., Sherman, S., 1996. The natural history of cigarette smoking from adolescence to adulthood: demographic predictors of continuity and change. Health Psychol. 15, 478–484. Cicchetti, D.V., 1994. Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments for psychology. Psychol. Assess. 6, 284–290. Clark, D.B., Cornelius, J., 2004. Childhood psychopathology and adolescent cigarette smoking: A prospective survival analysis in children at high risk for substance use disorders. Addict. Behav. 29, 837–841. Clark, D.B., De Bellis, M.D., Lynch, K.G., Cornelius, J.R., Martin, C.S., 2003. Physical and sexual abuse, depression and alcohol use disorders in adolescents: onsets and outcomes. Drug Alcohol Depend. 69, 51–60. Clark, D.B., Pollock, N.K., Mezzich, A., Cornelius, J., Martin, C.S., 2001. Diachronic assessment and the emergence of substance use disorders. J. Child Adol. Subst. Abuse. 10, 13–22. Clark, D.B., Winters, K.C., 2002. Measuring risks and outcomes in substance use disorders prevention research. J. Consult. Clin. Psychol. 70 (6), 1207–1223. Colby, S.M., Tiffany, S.T., Shiffman, S., Niaura, R.S., 2000a. Measuring nicotine dependence among youth: a review of available approaches and instruments. Drug Alcohol Depend. 59 (1), s23–s39. Colby, S.M., Tiffany, S.T., Shiffman, S., Niaura, R.S., 2000b. Are adolescent smokers dependent on nicotine ? A review of the evidence. Drug Alcohol Depend. 59 (1), s83–s95. Cornelius, J.R., Lynch, K., Martin, C.S., Cornelius, M.D., Clark, D.B., 2001. Clinical correlates of heavy tobacco use among adolescents. Addict. Behav. 26, 273–277. Curie, S.R., 2004. Confirmatory factor anlaysis of the Reasons for Smoking Scale in alcoholics. Nicotine Tobacco Res. 6, 465–470. Edwards, G., 1986. The alcohol Dependence Syndrome: a concept as stimulus to enquiry. Br. J. Addict. 81, 171–183.

Fagerstrom, K.O., 1978. Measuring degrees of physical dependence to tobacco smoking with reference to individualization of treatment. Addict. Behav. 3, 235–241. Fagerstrom, K.O., Schneider, N.G., 1989. Measuring nicotine dependence” a review of the Fagerstrom Tolerance Questionnaire. J. Behav. Med. 12 (2), 159–182. Heatherton, T.F., Kozlowski, L.T., Frecker, R.C., Fagerstrom, K.O., 1991. The Fagerstrom Test for Nicotine Dependence: a revision of the Fagerstrom Tolerance Questionnaire. Br. J. Addict. 86, 1119–1127. Jaffe, J.H., 1989. Addictions: What does biology have to tell? Int. Rev. Psychiatry 1, 51–61. Mayhew, K.P., Flay, B.R., Mott, J.A., 2000. Stages in the development of adolescent smoking. Drug Alcohol Depend. 59 (s1), s61–s81. Nichter, M., Nichter, M., Thompson, P.J., Shiffman, S., Moscicki, A., 2002. Using qualitative research to inform survey development on nicotine dependence among adolescents. Drug Alcohol Depend. 68s., 41–50. Novy, P., Hughes, J.R., Callas, P., 2001. A comparison of recovering alcoholic and non-alcoholic smokers. Drug Alcohol Depend. 65, 17–23. Payne, T.J., Smoth, P.O., McCracken, L.M., McSherry, W.C., Anthony, M.M., 1994. Assessing nicotine dependence: A comparison of the Fagerstrom Tolerance Questionnaire (FTQ) with the Fagerstrom Test for Nicotine Dependence (FTND) in a clinical sample. Addict. Behav. 19 (3), 307–317. Pomerleau, C.S., Carton, S.M., Lutzke, M.L., Flessland, K.A., Pomerleau, O.F., 1994. Reliability of the Fagerstrom Tolerance Questionnaire and the Fagerstrom Test for Nicotine Dependence. Addict. Behav. 19 (1), 33–39. Radzius, A., Moolchan, E.T., Henningfield, J.E., Heishman, S.J., Gallo, J.J., 2001. A factor analysis of the Fagerstrom Tolerance Questionnaire. Addict. Behav. 26, 303–310. Reise, S.P., Waller, N.G., Comrey, A.L., 2000. Factor analysis and scale revision. Psychol. Assess. 12, 287–297. Shadel, W.G., Shiffman, S., Niaura, R., Nichter, M., Abrams, D.B., 2000. Current models of nicotine dependence: what is known and what is needed to advance understanding of tobacco etiology among youth. Drug Alcohol Depend. 59, s9–s21. Shiffman, S., Waters, A., Hickcox, M., 2004. The Nicotine Dependence Syndrome Scale: A multidimensional measure of nicotine dependence. Nicotine Tobacco Res. 6, 327–348.