A

wrdncarq©

K

1

eu(op(g

GS

0d

Available online at www.sciencedirect.com

Drug and Alcohol Dependence 91 (2007) 159–168

Prediction of lapse from associations between smoking and situationalantecedents assessed by ecological momentary assessment

Saul Shiffman a,d,∗, Mark H. Balabanis a,b, Chad J. Gwaltney a,c, Jean A. Paty a,d,Maryann Gnys a,e,1, Jon D. Kassel a,f, Mary Hickcox a,g, Stephanie M. Paton a,h

a Department of Psychology, University of Pittsburgh, United Statesb San Francisco Bay Area Center for Cognitive Therapy, 5435 College Avenue, Oakland, CA 94618, United States

c Department of Community Health, Box G-S121-5, Brown University, Providence, RI 02912, United Statesd invivodata Inc., 2100 Wharton Street, Suite 505, Pittsburgh, PA 15203, United States

e Department of Veterans Affairs (116), 830 Chalkstone Avenue, Providence, RI 02908, United Statesf University of Illinois at Chicago, Department of Psychology (MC 285), 1007 W. Harrison Street, Chicago, IL 60607-7137, United States

g 2480 Yellow Creek Road, Akron, OH 44333, United Statesh Centers for Behavioral & Preventive Medicine, The Miriam Hospital, 1 Hoppin Street, Providence, RI 02903, United States

Received 11 March 2007; received in revised form 17 May 2007; accepted 21 May 2007

bstract

Smoking is associated with particular moods and activities, but it is not known whether there are individual differences in these associations andhether these differences are associated with success in smoking cessation. We assessed such associations using ecological momentary assessment:

eal-world, real-time data, collected by palm-top computer. Two hundred and fourteen smokers participating in a smoking cessation study providedata during ad lib smoking at baseline. Participants recorded moods and activities each time they smoked and, for comparison, at randomly selectedon-smoking occasions. Situational associations with smoking were captured by examining the associations between smoking and antecedentsonsidered relevant to lapse risk: negative affect (NA), arousal, socializing with others, the presence of others smoking, and consumption of coffee

nd alcohol. The associations varied across participants, confirming individual differences in situational smoking associations. Survival analysesevealed that only the NA pattern predicted first lapse. The effect was only seen in EMA assessments of NA smoking, and was not captured byuestionnaire measures of negative affect smoking, which did not predict lapse risk. Moreover, the effect was not mediated by nicotine dependence.

2007 Elsevier Ireland Ltd. All rights reserved.

colog

apatt

1

eywords: Smoking; Smoking patterns; Cessation; Relapse; Negative affect; E

. Background

Models of smoking suggest that smoking behavior is linked toxternal stimuli and internal states. Associations between stim-li and the probability of smoking, or situational associationsas we will call them), are thought to be important indicatorsf smoking motives. For example, most smokers say they are

articularly likely to smoke when experiencing negative affectBrandon, 1994; McKennell, 1970), and this is seen as sug-estive evidence that smoking or nicotine may reduce negative

∗ Corresponding author at: University of Pittsburgh, Smoking Researchroup, 130 North Bellefield Avenue, Suite 510, Pittsburgh, PA 15213, Unitedtates. Tel.: +1 412 687 5677; fax: +1 412 383 2041.

E-mail address: shiffman@pitt.edu (S. Shiffman).1 Present address: Providence VA Medical Center, United States.

tssaper

376-8716/$ – see front matter © 2007 Elsevier Ireland Ltd. All rights reserved.oi:10.1016/j.drugalcdep.2007.05.017

ical momentary assessment

ffect (e.g., Pomerleau and Pomerleau, 1987). Other smokingatterns, such as smoking when seeing others smoke (Drobesnd Tiffany, 1997), are seen as indicating the influence of condi-ioning processes on smoking, and are thought to be very relevanto risk of relapse after cessation.

.1. Negative affect smoking

Smoking theory and treatment have particularly focused onhe situational association of smoking and negative affect (NAmoking), for good reason. Not only do the vast majority ofmokers report that they smoke when experiencing negative

ffect (McKennell, 1970), but relief of negative affect is oftenosited to be a major negative reinforcer of smoking (Copelandt al., 1995). Moreover, negative affect precipitates increases ineported desire to smoke (Payne et al., 1991), and plays a strong

1 ohol D

rs1WaswTasa

aaiwahaiNiN(w1ei

1

(aatssTao

tmteIsmlitso2nd

tTbs

enaaHtWsentioat

1

muSbSse(la(wt2s

aqbmSftodwsap

60 S. Shiffman et al. / Drug and Alc

ole in precipitating resumption of smoking among abstinentmokers (Baer and Lichtenstein, 1988; O’Connell and Martin,987; Shiffman, 1982; Shiffman et al., 1996a,c; Shiffman andaters, 2004). This raises the question of whether NA smoking

ssociations might predispose smokers to failure in smoking ces-ation. Indeed, Pomerleau et al. (1978) reported that smokersho reported NA smoking were at increased risk for relapse.his important demonstration of the influence of situationalssociations on treatment outcome was replicated in some sub-equent studies (Niaura et al., 1989), but not others (Mothersill etl., 1988; O’Connell and Shiffman, 1988; Kreitler et al., 1976).

If NA smoking is associated with relapse risk, several mech-nisms might be involved. Since smokers experience negativeffect during abstinence, and tend to lapse when experienc-ng distress, a conditioning mechanism could be operative, inhich smokers have learned to associate smoking with neg-

tive affect, which then comes to promote relapse. If smokersave learned that smoking relieves negative affect, then negativeffect situations may promote smoking by providing additionalncentive for smoking. It is also possible that the link betweenA smoking and relapse risk is not causal, but that NA smok-

ng instead serves as a marker for dependence, since reports ofA smoking are associated with greater nicotine dependence

McKennell, 1970; Shiffman, 1993) and more severe nicotineithdrawal (Niaura et al., 1989; though see West and Russell,985; Hughes and Hatsukami, 1986). It could be that NA smok-rs report smoking under negative affect because their smokings withdrawal-driven.

.2. Assessing smoking associations

An important shortcoming of the research on NA smokingand situational associations in general) is that associations weressessed using global questionnaires that ask smokers to char-cterize their smoking patterns. There is substantial evidencehat such assessments are invalid and do not accurately reflectituational associations with smoking (Shiffman, 1993). Retro-pective data are subject to many biases (Hammersley, 1994).he task of characterizing patterns of smoking—summarizingn act smokers undertake, often unconsciously, many thousandsf times a year, may simply be unrealistic (Shiffman, 1993).

An alternative approach to assessing situational associa-ions without relying on retrospective summary uses ecological

omentary assessment (EMA; Shiffman and Stone, 1998)o collect real-time data about smoking episodes in smok-rs’ natural environments, avoiding recall bias and summary.mportantly, to characterize associations between smoking andituational antecedents such as mood, it is essential to also assessood outside of smoking contexts, in order to control for “base-

ine” mood (Paty et al., 1992). (That is, observing that a smokers slightly anxious, on average, when smoking, is uninforma-ive, because he/she may also be slightly anxious all the time,moking or not.) To quantify associations between mood (or

ther antecedents) and smoking, we (Shiffman et al., 2002,004a) contrasted smoking observations with observations ofon-smoking moments, obtained by “beeping” subjects at ran-om (see Larson and Csikszentmihalyi, 1983) to obtain a random

Hp

e

ependence 91 (2007) 159–168

ime-sampling of non-smoking occasions (see Shiffman, 2007).he association between smoking and mood can be estimatedy assessing differences in mood between smoking and non-moking situations.

Using EMA to analyze situational associations in two differ-nt samples of smokers (Shiffman et al., 2002, 2004a), we foundo association between smoking and mood. In both studies, theverage within-subject correlation between smoking and neg-tive affect was less than 0.04 (Shiffman et al., 2002, 2004a).owever, this does not necessarily undermine the hypothesis

hat some individuals may smoke in response to negative affect.hile these EMA data suggest no average association between

moking and negative affect, there may still be individual differ-nces in how the two are associated. Indeed, the mean correlationear 0 suggests that some smokers might have positive correla-ions while others show negative correlations. Thus, variationsn the association between smoking and affect might still predictutcome in smoking cessation. The purpose of this paper is tossess whether individual differences in EMA-assessed situa-ional associations (NA and others) predict the risk of lapsing.

.3. Other situational correlates of smoking

Besides NA smoking, several other situational associationsay relate to cessation outcome. A potentially important sit-

ational association is “stimulation smoking” (Frith, 1971;hiffman, 1993)—smoking during a state of low-arousal (i.e.,oredom, fatigue, etc.), or when one wants to be stimulated.ome studies using questionnaire measures have reported thattimulation smoking is associated with cessation failure (Barnest al., 1985; Mothersill et al., 1988), while others have notNiaura et al., 1989; West and Russell, 1985). As with NA,ow-arousal has been associated with lapse risk (Shiffman etl., 1996c) and reports of low-arousal smoking with dependenceRussell et al., 1974; Shiffman, 1993), suggesting an associationith cessation outcome. EMA data have shown no associa-

ion between arousal and smoking, on average (Shiffman et al.,002, 2004a), but potential individual differences in low-arousalmoking might be relevant to cessation outcome.

Some smoking situational associations are plausibly associ-ted with low dependence and higher probability of successfuluitting. For example, social motives for smoking are thought toe most prominent early in smoking careers, and to fade as otherotives, including nicotine dependence, develop (Ashton andtepney, 1982), suggesting that smokers who continue to smokeor social reasons should find it easier to quit. Consistent withhis, McKennell (1970) found that smokers who smoked withthers were lighter smokers. Although “chippers” (light, non-ependent smokers who report casual cessation from smokingithout experiencing nicotine withdrawal) report more social

moking on questionnaire measures (Shiffman et al., 1994b),n EMA analysis (Shiffman and Paty, 2006) showed that chip-ers were just as likely as heavy smokers to smoke when alone.

ere, we examine smoking when socializing and smoking in theresence of others who are smoking.

EMA analyses of situational associations have identified sev-ral other situational correlates of smoking. As also reported in

ohol D

gt1aTagrw1afw2a1

tta2CmtyaibmSbsgti

1

taoubsiacfislSswdcd

2

qo

2

nwesai2o

9bo(

2

puDaswSasr‘ppm(ott

srEaaiAe

2

2.3.1. EMA situational antecedents of baseline smoking. Smoking patternswere assessed during the baseline period of ad lib smoking, prior to the quitdate. The data were drawn from observations over eight consecutive days of

S. Shiffman et al. / Drug and Alc

lobal measures (McKennell and Thomas, 1967) and in labora-ory studies (e.g., Griffiths et al., 1976; Mello and Mendelson,986), smoking is associated with drinking alcohol (Shiffmannd Balabanis, 1995; Shiffman et al., 1994a,b, 2002, 2004a).he association of smoking and drinking appears to be greatermong non-dependent smokers (Shiffman and Paty, 2006), sug-esting that this situational association may be linked to lowerelapse risk. On the other hand, lapse episodes often occurhile people are drinking (Shiffman, 1982; Shiffman et al.,996c), suggesting that drinking, or having a learned associ-tion between drinking and smoking, might put smokers at riskor relapse. Similar considerations apply to coffee drinking,hich is similarly associated with smoking (Shiffman et al.,002; Istvan and Matarazzo, 1984; McKennell, 1970; Russell etl., 1974) and with relapse (Shiffman et al., 1996c; Shiffman,982).

Another important influence on smoking is craving or urgeo smoke. While the role of urges in smoking has been ques-ioned (Tiffany, 1990), we have found that elevated urge levelsre associated with ad lib smoking (Shiffman et al., 2002,004a), as well as with relapse (Shiffman et al., 1996a,c).onceptually, one might expect that those whose smoking isost associated with craving might be the most dependent and

hus the most likely to lapse after quitting. However, anal-ses of craving data suggest that craving when people arebout to smoke is high for all smokers, and is thus indiscrim-nating; conversely, it is the experience of elevated cravingetween cigarettes, when the person is not smoking, that isost associated with dependence (Shiffman and Paty, 2006;hiffman et al., 2004a,b). Accordingly, lower associationsetween craving and smoking (i.e., smaller differences betweenmoking and non-smoking occasions) may be associated withreater dependence. In any case, assessing how the associa-ions between craving and smoking relate to relapse risk is ofnterest.

.4. The present study

Given the lack of clear findings in the literature on the rela-ionship between smoking associations and success in cessation,nd the literature’s reliance on global retrospective measuresf smoking associations, we sought to assess whether individ-al differences in such associations, as assessed situationallyy EMA methods, would predict the outcome of smoking ces-ation. Analysis of outcomes suggests that failing to achievenitial abstinence, lapsing after having achieved abstinence,nd progressing from lapsing to relapse are different pro-esses (Shiffman et al., 2006), making analyses of aggregatenal outcomes possibly uninformative. Accordingly, we focusedpecifically on the risk of lapsing, since research has estab-ished that lapses are subject to situational influences (e.g.,hiffman, 1982; Shiffman et al., 1996a,c). Where the analy-is showed significant prediction from smoking associations,

e planned to test whether these relationships could also beocumented for questionnaire-based measures of smoking asso-iations, and whether they could be accounted for by nicotineependence.

swtt

ependence 91 (2007) 159–168 161

. Methods

We assessed smoking patterns during ad lib smoking at baseline, prior touitting, and used these as independent variables to predict smoking cessationutcome, using survival analysis to examine time to lapsing.

.1. Participants

Participants were 214 smokers who quit smoking (defined as 24 h of absti-ence) while in a smoking cessation research program. This sample overlapsith that reported in other papers on the characteristics of lapse and temptation

pisodes (e.g., Shiffman et al., 1996a,c), and is a subset (i.e., those who quitmoking) of that on which situational associations were reported in Shiffman etl. (2002). Participants were recruited through advertisements. To be includedn the study, participants had to smoke at least 10 cigarettes a day for the pastyears, and report high motivation and efficacy to quit (combined score of 150n the sum of two 0–100 scales).

Fifty-seven percent of the participants in this analysis were female, and0% were Caucasian. Participants averaged 42 years of age (S.D. = 10.2), hadeen smoking for an average of 23 years (S.D. = 12.4), and smoked a meanf 28 cigarettes per day (S.D. = 13.7). Estimated FTQ scores averaged 6.11S.D. = 1.80, n = 201); 45% had had FTQ scores of 7 or greater.

.2. Procedures

Before participants were asked to quit smoking (baseline period), they com-leted a battery of questionnaire assessments. Participants were also trained tose the electronic diary (ED) to monitor the antecedents of their ad lib smoking.uring baseline, subjects were instructed to continue smoking in their usual rate

nd pattern, and directed to record all occasions of smoking in the ED (beforemoking).1 The ED randomly sampled a subset of these entries for assessmentith the target of collecting data on five cigarettes per day (actual M = 4.39,.D. = 2.63 assessed smoking occasions per day). On these occasions, the EDdministered a series of questions assessing antecedent activities and emotionaltates. The same assessment items were also presented in approximately fiveandom occasions throughout the day (M = 4.76, S.D. = 2.28), when the EDbeeped’ participants while they were not smoking. These assessments wererogrammed not to fall within 10 min of a cigarette entry. Subjects respondedromptly to 91% of the ED’s random assessments prompts within 2 min. Theethods and compliance metrics are described in more detail in Shiffman et al.

2002). Participants were provided group cognitive behavioral treatment, with-ut a pharmacological adjunct, as described in Shiffman et al. (1996c). Thereatment deliberately did not discuss situational associations, to avoid biasinghe data.

After participants quit smoking (≥24 h abstinent; 70% reached this mile-tone), they monitored their ongoing quitting experiences (temptations, and anyeturn to smoking) for up to 4 weeks. Participants were instructed to initiate anD entry when they lapsed, defined as any occasion of smoking, even a puff. Tossess subsequent outcome, participants attended a follow-up session 2 monthsfter ED monitoring and retrospectively reported lapses since the ED monitor-ng period, using the time-line follow-back method (TLFB; Brown et al., 1998).bstinence was verified biochemically through assays of salvia cotinine and

xhaled breath carbon monoxide.

.3. Assessments

1 While there could be concern that the pending quit attempt might changemokers’ behavior, comparison of smoking associations between this sample,hich was planning to quit (Shiffman et al., 2002) and a sample not planning

o quit (Shiffman et al., 2004a) suggested that smoking patterns were similar inhe two groups.

1 ohol D

siwbtTa4dwatse(ea

2bts1tnsTttwsmr

2

6Ep

2

2dafacabposniccddwtcwwew

tsWw(iae

ihol

2sf(

2ibwe

2

a(Wiwae

3

3

3lrcdnb

3(nbsribc

62 S. Shiffman et al. / Drug and Alc

moking, beginning 4 days into monitoring (allowing for adaptation) and end-ng 6 days before the target quit date (avoiding potential changes in smokinghen preparing to quit). Smoking patterns were quantified using data collectedy ED on smoking and random non-smoking occasions. On each occasion, par-icipants rated the intensity of their current urge to smoke using a 0–10 scale.he ED also requested information about antecedent activities, mood statesnd exposure to others’ smoking and to smoking regulations (a sub-group of1 subjects in this sample were randomized to a reduced-burden protocol thatid not assess smoking regulations or others’ smoking). Participants reportedhether they had consumed alcohol or caffeine in the preceding 15 min. They

lso reported whether they were socializing. Participants rated 15 mood adjec-ives derived from the circumplex model of affect (Russell, 1980), using a 4-pointcale (“NO!!, no??, yes??, YES!!”; see Meddis, 1972). As described in Shiffmant al. (1996c), these ratings yielded bipolar factor scores for negative affectα = .87; positive: happy, contented, and overall feeling; negative: irritable, mis-rable, tense, frustrated/angry, sad) and arousal (α = .79; tired, energetic, overallrousal level).

.3.2. Smoking typology and nicotine-dependence questionnaires. During theaseline period, participants also completed the Russell smoking motives ques-ionnaire (Russell et al., 1974), which includes a negative affect smokingcale, and the Smoking Occasions Questionnaire (McKennell and Thomas,967), which includes a similar Nervous Irritation scale. They also completedhe Nicotine-Dependence Syndrome Scale (NDSS; M = −0.05, S.D. = 1.0), aicotine-dependence measure based on Edward’s (1986) concept of dependenceyndrome (Shiffman et al., 2004b), and a modified version of the Fagerstromolerance Questionnaire (FTQ; Fagerstrom, 1978), which, like the Fagerstrom

est of nicotine dependence (Heatherton et al., 1991), used expanded scalingo yield more variance; variations on some items were also added. Consistentith Payne et al. (1994), factor analysis yielded two factors (z-scores): Morning

moking (time to first cigarette [reversed], preferring the first cigarette, smokingore in the morning, craving cigarettes more in the morning), and Difficulty

efraining (smoking when ill, smoking where forbidden).

.4. Electronic diary (ED)

The ED was a computerized system which employs a PSION Organizer II LZ4 hand-held portable computer (5.6 in.× 1.1 in., 8.8 oz; PSION Ltd., London,ngland). The system runs from software devised specifically for this researchrogram and is described in prior papers (Shiffman et al., 1996a,b,c, 2002).

.5. Data reduction

.5.1. Situational associations. Situational associations with smoking wereefined conceptually as linkages between smoking, in contrast to non-smoking,nd situational antecedents such as alcohol consumption. We tested several dif-erent ways to specify or express the association between smoking and situationalntecedents. The primary method used for analysis contrasted the mean (forontinuous variables: affect and urge) or proportion (for dichotomously codedntecedents, such as alcohol) observed in smoking and non-smoking situations,y taking the difference between the two. So, for each participant, we com-uted, for example, the difference in mean rated negative affect in smokingccasions and in non-smoking occasions (i.e., the smoking mean minus the non-moking mean). Similarly, we computed the difference between smoking versuson-smoking situations in the proportion of occasions the participant was drink-ng (i.e., the smoking proportion minus the non-smoking proportion). Becauseontrast of proportions are typically expressed as an odds ratio (OR), we alsoomputed the OR for each participant, expressing the association between, e.g.,rinking and smoking, and analyzed the log (OR), which is more nearly normallyistributed. As another summary expression of association, we also computed,ithin each subject, the correlation between dichotomously coded observation

ype (smoking versus non-smoking) and the antecedent. For dichotomously

oded antecedents, these were phi coefficients; for continuous variables, theseere point-biserial coefficients. All of these expressions were highly correlatedith each other. We repeated all the survival analyses using all of the forms of

xpression for the situational smoking associations. We also ran models in whiche entered both means (continuous variables) or proportions (dichotomies) into

abwc

ependence 91 (2007) 159–168

he survival model simultaneously, evaluating the presence of the antecedent inmoking episodes, while controlling for its presence in non-smoking episodes.e present as primary the analysis based on difference scores described above,hich is simplest to understand and derives most directly from the raw data.

This also retained more subjects: The OR is not defined when either proportions 0.) The results of analyses using other expressions of the smoking associationsre also summarized, to assess the sensitivity of the analysis to the choice ofxpression.

To capture non-linear, quadratic effects of NA and arousal – i.e., where smok-ng might be associated with both good and bad mood, and with both low andigh arousal – we also computed associations between smoking and the squaref these variables. These were tested in the same way as the untransformed,inear variables

.5.2. Smoking restrictions. To assess the possible confounding influence ofmoking restrictions, a subset of cases was identified in which smoking was notorbidden, and in which the smoker had not changed locations in order to smokefor smoking observations).

.5.3. Abstinence/lapses. If a lapse occurred, the date of lapse was the firstnstance of smoking as recorded (by ED during the ED monitoring period ory TLFB thereafter). If a participant had completed or dropped out of the studyithout a lapse, the survival analysis treated the outcome as censored data (Curry

t al., 1988).

.6. Data analysis

Cox Survival Analysis was used to test the effect of each smoking pattern onbstinence (survival). Survival models were tested using the various expressionsabove) for the participant’s association between smoking and each antecedent.

e also ran a multivariate survival model with all the situational associationndices included. Where a variable predicted lapse risk, we also (1) assessedhether parallel questionnaire scales of smoking patterns were also predictive,

nd (2) assessed whether nicotine dependence might be related to the observedffect.

. Results

.1. Description

.1.1. Lapse rate. As reported in Shiffman et al. (1996c), theapse rates were highest early in abstinence and slow thereafter,esulting in a negatively decelerating survival curve. Thirty per-ent of the participants lapsed into smoking within the first threeays of their quit date; 50% of the participants had lapsed by theinth day post-quit; 76% of the participants had reported a lapsey the end of the 90-day observation period.

.1.2. ED entries. Participants provided an average of 33.49S.D. = 9.70) smoking assessments and 37.11 (S.D. = 11.84)on-smoking assessments per participant. The minimum num-er of assessments for any subject was 28; removing the 6ubjects who had less than 40 observations did not change theesults, so we report on the full sample. To assess whether rat-ngs of smoking and non-smoking situations, which formed theasis of the situational association indices, were reliable, weorrelated participant-level estimates of each variable (e.g., NA,

lcohol drinking) obtained from odd versus even days duringaseline (Table 1). Most of the correlations were 0.80 and higher,ith a few as low as 0.60. Reliabilities for the composite (i.e.,

ombining both odd and even days) were computed using the

S. Shiffman et al. / Drug and Alcohol D

Table 1Correlations between average ratings or proportions estimated on odd vs. evendays, and estimated reliabilities

Correlation (reliability)

Non-smoking Smoking

Urge 0.94 (0.97) 0.97 (0.98)Negative affect (NA) 0.88 (0.94) 0.87 (0.93)Arousal 0.82 (0.90) 0.82 (0.90)Socializing 0.60 (0.75) 0.67 (0.80)Coffee 0.79 (0.88) 0.81 (0.89)Alcohol 0.60 (0.75) 0.79 (0.88)

Nf

Sw

3dpafTuat

ttnbTcedt(satwnnsf

3cp

ote: Reliability estimated from correlation coefficients by Spearman–Brownormula for split-half reliability.

pearman–Brown formula (Spearman, 1910). All reliabilitiesere 0.75 or better.

.1.3. Situational associations: distributions and individualifferences. For each participant and each antecedent, we com-uted the differences in means or proportions between smokingnd non-smoking occasions. Fig. 1 shows the distribution of dif-erences in means; Fig. 2 shows the differences in proportions.

he distributions suggest that, even though all the curves exceptrge were centered around 0 (i.e., there was at best only a slightssociation, on average, between smoking and the antecedents),here was considerable variability around the mean. In all cases,

wbaa

Fig. 1. Distribution of differences in mean ratings for smokin

ependence 91 (2007) 159–168 163

he distributions indicate a range of associations, with some par-icipants’ data showing positive associations and others, showingegative associations. All of the distributions appear roughlyell-shaped, though not necessarily fitting a normal distribution.o assess whether the between-subject variation was statisti-ally meaningful, we conducted tests of the between-subjectsffects for each antecedent. With the antecedent as the depen-ent variable, and individual observations as the unit of analysis,hese models tested whether the differences by observation typesmoking/non-smoking) varied significantly by subject (i.e., theubject by situation-type interaction). All but one of the inter-ctions were highly significant (p’s < 0.00001), indicating thathere were reliable between-subject differences. The exceptionas situations in which others were smoking, which appearedot to differ across subjects. Since this indicated that there wereo reliable individual differences in the relation between othersmoking and subjects’ smoking, this variable was dropped fromurther analysis.

.1.4. Correlations among situational associations. To furtherharacterize situational associations, correlations were com-uted among indicators (Table 2). Participants whose smoking

as associated with NA also tended to have associationsetween smoking and arousal. There were intercorrelationsmong patterns where smoking was associated with socializingnd drinking alcohol. Those for whom it was associated with

g and non-smoking occasions, for continuous variables.

164 S. Shiffman et al. / Drug and Alcohol Dependence 91 (2007) 159–168

king

cs

3

eidNti

TC

UNaASAC

iqr

3tions might have distorted situational associations with smoking,we re-tested all the survival models using a subsample of situa-tions in which smoking was not forbidden; the results were thesame.

Fig. 2. Distribution of differences in proportions for smo

offee drinking were also likely to report higher urges whenmoking.

.2. Prediction of survival from situational associations

Table 3 shows the results from Cox Survival Analysis forach smoking pattern. Only NA smoking associations signif-cantly predicted lapse risk. Fig. 3 illustrates this relationship,

isplaying the observed survival curve for smokers with averageA pattern, and the modeled curve for those 1 standard devia-

ion higher and lower than the mean. A 1S.D. deviation increasen NA smoking resulted in a doubling (increase of 108%)

able 2orrelation matrix of situational association indices

N Urge Negativeaffect (NA)

Arousal Socializing Alcohol

rge 214 –egative

ffect (NA)214 0.09 –

rousal 214 −0.02 0.34* –ocializing 214 0.01 −0.13 0.09 –lcohol 104 0.03 −0.05 0.16 0.35* –offee 203 0.15* 0.14 0.04 −0.08 −0.13

* p < 0.05.

TR

S

D

and non-smoking occasions, for dichotomous variables.

n the daily hazard of lapsing. Other variables, including theuadratic expressions of NA and arousal, did not predict lapseisk.

.2.1. Sensitivity testing. To assess whether smoking restric-

able 3esults of univariate survival models for situational associations and covariates

N Parameterestimate (beta)

Standarderror

Univariatehazard ratio

ituational associationUrge 214 −0.00 0.05 0.99Negative affect 214 0.74 0.27 2.08*

Arousal 214 0.13 0.29 1.14Socializing 214 −0.22 0.64 0.80Alcohol 104 2.51 1.25 2.50Coffee 203 0.31 0.63 1.36

ependenceNDSS-T 202 0.16* 0.08 1.18*

Morning smoking 213 0.02 0.08 1.02Difficulty refraining 213 0.13 0.08 1.14

* p < 0.05.

S. Shiffman et al. / Drug and Alcohol D

Fig. 3. Observed survival curve for smokers at average levels of negative affect((t

avuN

smptT(edO

3

rrNararmt

3a

nptn

n(d(nCal

4

setafvaw

4

arwctwfiNd

rnP2(firnb

t1taoetm

NA) smoking at baseline, with modeled curves for those with low NA smoking−1S.D.) and high NA smoking (+1S.D.), modeled based on coefficients fromhe survival analysis.

To test for sensitivity of the models to how the situationalssociations were expressed and quantified, we re-tested the sur-ival models using each of the different expressions describednder data reduction. All produced the same pattern of results:A associations, and only NA associations, predicted lapse risk.

Because NA was a continuous variable, and most of theituational variables that failed to predict lapse risk were dichoto-ous, we considered whether the differences in their ability to

redict lapsing might have come about because of the disadvan-age in statistical power associated with dichotomous variables.o assess this, we expressed NA as a dichotomous variablebased on a median split), and re-tested the survival model. Theffect was the same (whether we expressed the association withichotomous NA as the difference in proportions, phi, or logR).

.3. Mediation by dependence

We tested whether measures of dependence predicted lapseisk. As shown in Table 3, only the NDSS-T score predicted lapseisk. A one-point increase (i.e., one standard deviation) in theDSS-T score corresponded to an 18% increase in the daily haz-

rd of lapsing. Accounting for the influence of NDSS-T on lapseisk did not, however, change the direct association between NAssociations and lapse risk. (Not surprising, as NDSS-T was notelated to NA association, r = 0.10, n.s.) Estimating multivariateodels including gender and dependence also did not change

he result for any other situational associations.

.4. ED-assessed versus questionnaire-assessed NAssociations

Since NA smoking associations assessed by EMA sig-

ificantly predicted survival, we considered whether similarrediction could be obtained from a smoking typology ques-ionnaire measure of NA smoking. We first observed thateither the Russell NA Smoking score (r(n = 214) = 0.05, n.s.)

satc

ependence 91 (2007) 159–168 165

or the McKennell and Thomas Nervous Irritation scorer(n = 214) = 0.02, n.s.) was significantly correlated with EMA-erived NA scores. Neither the Russell NA smoking scoreHR = 1.14, n.s.), nor the Nervous Irritation score (HR = 1.06,.s.) predicted survival, and entering them into a multivariateox equation did not affect the magnitude or significance of thessociation between EMA-derived NA smoking association andapse risk.

. Discussion

This is the first study to use real-time EMA methods to assessituational associations with smoking and analyze their influ-nce on cessation outcome. The data show that smokers whoend to smoke when they are experiencing negative affect haveheightened risk of lapsing. This effect could not be accounted

or by nicotine dependence. Although we observed individualariation in the association of smoking with urge levels, arousal,lcohol, coffee consumption, and social smoking, none of theseas related to lapse risk.

.1. Negative affect smoking

The finding that people who smoke under conditions of neg-tive affect are more prone to failure in cessation mirrors theeports of Pomerleau et al. (1978) and Niaura et al. (1989), whichere based on smoking typology questionnaires. Our analysis

ontradicted one plausible mechanism—that the effect was dueo an association between NA smoking and nicotine dependence,hich is, in turn, associated with failure in cessation. We didnd that more nicotine-dependent smokers (as assessed by theDSS) were at greater risk for lapsing. However, dependenceid not mediate the influence of NA associations on lapse risk.

What, then, accounts for the link between NA smoking andelapse? The literature on smoking has long speculated thaticotine may ameliorate negative affect (Kassel et al., 2003;omerleau and Pomerleau, 1992; cf.: Conklin and Perkins,005). Perhaps smokers who have come to appreciate this effectif it is indeed present) find smoking more reinforcing and there-ore harder to give up. The incentive to smoke to blunt affectn an upsetting situation may be the proximal cue that triggerselapse. Importantly, this might operate even if smoking doesot in fact reduce distress—smokers’ beliefs that it does so maye sufficient to drive behavior.

Previous studies have indicated that smokers who are proneo negative affect have a high risk of cessation failure (Brandon,994; Carmody, 1992), and may benefit from treatment to reduceheir affective distress (Hall et al., 1994). However, our findingsre not about smokers’ level of distress per se: Our measuref NA smoking is independent of whether smokers generallyxperience negative affect; it assesses more specifically whetherhey were more likely to smoke when in negative affect states. Our

easure of association inherently controlled for any between-

ubject variations in NA levels per se. It is not clear what clinicalpproach is suited to smokers who smoke when upset. Breakinghe association between affect and smoking – e.g., by stimulusontrol strategies (Cinciripini et al., 1994) – could prove helpful.

1 ohol D

4

iewlp“s

ts2w(idiInesbpof

4

ntrwanalbbl

ielINnsaseTctq

pcstEapw

cessaaf

cD1srttIsnE“tmqte

4

asmiitnpoEei2vt

66 S. Shiffman et al. / Drug and Alc

.2. Other smoking associations

Contrary to other hypotheses, no other individual differencen situational smoking associations predicted lapse risk. Smok-rs who displayed more “indulgent” smoking (i.e., who smokeith others, and when drinking alcohol or coffee) were not less

ikely to lapse. Given that this was a treatment-seeking sam-le, we may not have observed real extremes of “indulgent” orsocial” smoking that might be seen among low-dependencemokers (Shiffman and Paty, 2006).

Also contrary to hypothesis, we did not observe any rela-ionship between urge-driven smoking and lapse risk. This wasomewhat surprising, as two prior analyses (Shiffman et al.,004a,b; Shiffman and Paty, 2006) have suggested that smokersho experience high urges in the intervals between cigarettes

thus resulting in a lower association between urges and smok-ng) were more dependent. Particularly as experiencing urgesuring abstinence is thought to be an impediment to maintain-ng abstinence, this was expected to be a predictor of lapse risk.t may be that the dynamics of urges change when total absti-ence is achieved. For example, the high urges some smokersxperience between cigarettes may be due to priming by the just-moked cigarettes (Shaham et al., 2003) or to the expectation ofeing able to smoke soon (Wertz and Sayette, 2001). Both thesehenomena would be absent during abstinence. The dynamicsf urges during ad lib smoking and following cessation deserveurther exploration.

.3. Methodological issues

While smoking restrictions have the potential to disruptatural situational associations with smoking, we found essen-ially identical results for samples of observations that includedestricted settings and those limited to occasions when smokingas not forbidden. This is consistent with our prior finding that

ssociations between smoking and situational antecedents areot substantially changed by smoking restrictions (Shiffman etl., 2002, 2004a). However, smoking restrictions may neverthe-ess disrupt situational associations that might otherwise haveeen observed. This is not a limitation of the study’s methods,ut reflects the reality that, in many countries, smoking is noonger an unrestricted ad libitum behavior.

Methodological limitations may have precluded detection ofnfluences on relapse. With regard to drinking and smoking, forxample, many participants were never observed while drinking,eaving only 104 participants for the analysis of alcohol effects.n addition, the quantity of alcohol consumed was not assessed.ull results may also have occurred because the sample didot include lighter or less dependent smokers, who may demon-trate a wider range of patterns, particularly relating to social andlcohol cues. However, the data from this sample demonstratedignificant between-person variability: all the associations wexamined showed variation in both magnitude and direction.

he smokers in the sample were preparing to quit, and couldonceivably have altered their smoking patterns just before quit-ing. However, smoking patterns assessed in a group not trying touit (Shiffman et al., 2004a) were quite similar. Although com-

(pnt

ependence 91 (2007) 159–168

liance with randomly prompted assessments was very high, weould not assess compliance with cigarette assessments with theame degree of certainty, and some smokers may have failedo record some cigarettes, thus undermining the validity of theD-assessed patterns. However, the fact that ED-assessed neg-tive affect smoking did predict relapse, and that the observedatterns had a coherent structure, suggests that the assessmentsere valid.Our analytic approach used an explicit two-stage hierarchi-

al design, in which we computed subject-level statistics forach subject, and then analyzed these in a second-stage between-ubject analysis of survival. Newer methods for combining bothtages into a single analysis are becoming available (Reardon etl., 2002). These may have advantages in terms of power andppropriate treatment of error variance, and may be useful inuture analyses.

In any case, the present approach to assessing situational asso-iations of smoking improves significantly over prior research.ata were collected via EMA methods (Stone and Shiffman,994), in near-real-time, occasion-by-occasion, in real-worldettings, ensuring ecological validity and avoiding retrospectiveecall and summary. Further, smoking episodes were contrastedo randomly sampled non-smoking occasions to provide “con-rols” for the base rate of situational variables (Paty et al., 1992).n this study, two questionnaire measures of negative affectmoking failed to show any relationship to lapse risk. Indeed,either had any relationship with a measure based on real-timeMA data. This is consistent with the evidence that smokingtypology” questionnaires do not accurately assess smoking pat-erns (Shiffman, 1993). Conversely, the finding that an EMA

easure of smoking associations predicts outcome, even whereuestionnaire measures do not, adds to evidence for the predic-ive and incremental validity of EMA assessment (see Kamarckt al., 2007).

.4. Summary and implications

The finding that smokers who tend to smoke when emotion-lly distressed have a heightened risk of relapse suggests thatuch smokers may need particular clinical attention and treat-ent. However, providing treatment tailored to these smokers

s currently challenging, for several reasons. First, our find-ngs showed that this smoking pattern could only be identifiedhrough real-time EMA monitoring, and not through question-aire assessment. Although EMA is not currently used in clinicalractice, it may, in time, become easier to implement these meth-ds in clinical settings to obtain more valid patient assessment.lectronic devices that facilitate real-time monitoring of smok-rs’ state may also be useful in delivering tailored real-timenterventions that address patient needs as they arise (Shiffman,006; Carter et al., 2007). It may also be possible to improve thealidity of questionnaires assessing smoking patterns, althoughhe limitations imposed by autobiographical memory processes

Hammersley, 1994) will be difficult to overcome. A final,erhaps most profound challenge to specialized treatment foregative affect smokers is that it is not clear what interven-ions would effectively address their needs. Given that their

ohol D

saHfmsp

R

A

B

B

B

B

C

C

C

C

C

C

D

F

F

G

H

H

H

H

I

K

K

K

L

M

M

M

M

M

N

N

O

O

P

P

P

P

P

P

R

R

R

S

S. Shiffman et al. / Drug and Alc

pecial challenges arise from an association between smokingnd mood, an extinction approach seems conceptually suitable.owever, extinction treatments have yet to demonstrate efficacy

or smoking cessation (Niaura et al., 1999). Thus, our findingsay not lead directly to improvements in treatment, but they do

uggest the need for further careful attention to stimulus-drivenatterns of smoking.

eferences

shton, H., Stepney, R., 1982. Smoking: Psychology and Pharmacology. Rout-ledge, Chapman & Hall, Inc., New York.

aer, J., Lichtenstein, E., 1988. Classification and prediction of smoking relapseepisodes: an exploration of individual differences. J. Consult. Clin. Psychol.56, 104–110.

arnes, G.E., Vulcano, B.A., Greaves, L., 1985. Characteristics affecting suc-cessful outcome in the cessation of smoking. Int. J. Addict. 20, 1429–1434.

randon, T.H., 1994. Negative affect as motivation to smoke. Curr. Dir. Psychol.Sci. 3, 33–37.

rown, R.A., Burgess, E.S., Sales, S.D., Whiteley, J.A., Evans, D.M., Miller,I.W., 1998. Reliability and validity of a smoking timeline follow-back inter-view. Psychol. Addict. Behav. 12, 101–112.

armody, T.P., 1992. Affect regulation, nicotine addiction, and smoking cessa-tion. J. Psychoactive Drugs 24, 111–122.

arter, B.L., Day, S.X., Cinciripini, P.M., Wetter, D.W., 2007. Momentary healthinterventions: where are we and where are we going? In: Stone, A.A., Shiff-man, S., Atienza, A., Nebeling, L. (Eds.), The Science of Real-time DataCapture: Self-Reports in Health Research. Oxford University Press, NewYork, 289–307.

inciripini, P.M., Lapitsky, L.G., Wallfisch, A., Mace, R., Nezami, E., 1994.An evaluation of a multicomponent treatment program involving scheduledsmoking and relapse prevention procedures: initial findings. Addict. Behav.19, 13–22.

onklin, C.A., Perkins, K.A., 2005. Subjective and reinforcing effects of smok-ing during negative mood induction. J. Abnorm. Psychol. 114, 153–164.

opeland, A.L., Brandon, T.H., Quinn, T.B., 1995. The Smoking ConsequencesQuestionnaire-Adult: measurement of smoking outcome expectancies ofexperienced smokers. Psychol. Assess. 7, 484–494.

urry, S., Marlatt, G.A., Peterson Jr., A.V., Lutton, J., 1988. Survival analysisand assessment of relapse rates. In: Donovan, D.M., Marlatt, G.A. (Eds.),Assessment of Addictive Behaviors. Guilford Press, New York, 454–473.

robes, D.J., Tiffany, S.T., 1997. Induction of smoking urge through imagi-nal and in vivo procedures: physiological and self-report manifestations. J.Abnorm. Psychol. 106, 15–25.

agerstrom, K.O., 1978. Measuring degree of physical dependence to tobaccosmoking with reference to individualization of treatment. Addict. Behav. 3,235–241.

rith, C.D., 1971. Smoking behaviour and its relation to the smoker’s immediateexperience. Br. J. Soc. Clin. Psychol. 10, 73–78.

riffiths, R.R., Bigelow, G.E., Liebson, I., 1976. Facilitation of human tobaccoself-administration by ethanol: a behavioral analysis. J. Exp. Anal. Behav.25, 279–292.

all, S.M., Munoz, R.F., Reus, V.I., 1994. Cognitive-behavioral interventionincreases abstinence rates for depressive-history smokers. J. Consult. Clin.Psychol. 62, 141–146.

ammersley, R., 1994. A digest of memory phenomena for addiction research.Addiction 89, 283–293.

eatherton, T.F., Kozlowski, L.T., Frecker, R.C., Fagerstrom, K.O., 1991. Thefagerstrom test for nicotine dependence: a revision of the Fagerstrom Toler-ance Questionnaire. Br. J. Addict. 86, 1119–1127.

ughes, J.R., Hatsukami, D., 1986. Signs and symptoms of tobacco withdrawal.

Arch. Gen. Psychiatry 43, 289–294.

stvan, J., Matarazzo, J., 1984. Tobacco, alcohol, and caffeine use: a review oftheir interrelationships. Psychol. Bull. 95, 301–326.

amarck, T.W., Shiffman, S., Muldoon, M.F., Sutton-Tyrell, K., Gwaltney, C.J.,Janicki, D.L., Schwartz, J., 2007. Ecological Momentary Assessment as a

S

S

ependence 91 (2007) 159–168 167

resource for social epidemiology. In: Stone, A.A., Shiffman, S., Atienza, A.,Nebeling, L. (Eds.), The Science of Real-time Data Capture: Self-Report inHealth Research. Oxford University Press, New York.

assel, J.D., Stroud, L.R., Paronis, C.A., 2003. Smoking, stress, and negativeaffect: correlation, causation, and context across stages of smoking. Psychol.Bull. 129, 270–304.

reitler, S.G., Shahar, A., Kreitler, H., 1976. Cognitive orientation, typeof smoker and behavior therapy of smoking. Br. J. Med. Psychol. 49,167–175.

arson, R., Csikszentmihalyi, M., 1983. The experience sampling method. NewDir. Methodol. Soc. Behav. Sci. 15, 41–56.

cKennell, A.C., 1970. Smoking motivation factors. Br. J. Soc. Clin. Psychol.9, 8–22.

cKennell, A.C., Thomas, R.K., 1967. Adults’ and Adolescents’ SmokingHabits and Attitudes. British Ministry of Health, London.

eddis, R., 1972. Bipolar factors in mood adjective checklists. Br. J. Soc. Clin.Psychol. 11, 178–184.

ello, N.K., Mendelson, J.H., 1986. Cigarette smoking: interactions withalcohol, opiates, and marijuana. In: Braide, M.C., Ginsburg, H.M. (Eds.),Strategies for Research on the Interactions of Drug Abuse. National Insti-tute on Drug Abuse Research Monograph No. 68. U.S. Government PrintingOffice, Washington, DC, 154–180.

othersill, K.J., McDowell, I., Rosser, W., 1988. Subject characteristics andlong term post-program smoking cessation. Addict. Behav. 13, 29–36.

iaura, R., Abrams, D.B., Shadel, W.G., Rohsenow, D.J., Monti, P.M., Sirota,A.D., 1999. Cue exposure treatment for smoking relapse prevention: a con-trolled clinical trial. Addiction 94, 685–695.

iaura, R., Goldstein, M.G., Ward, K.D., Abrams, D.B., 1989. Reasons forsmoking and severity of residual nicotine withdrawal symptoms when usingnicotine chewing gum. Br. J. Addict. 84, 681–687.

’Connell, K.A., Martin, E.J., 1987. Highly tempting situations associated withabstinence, temporary lapse, and relapse among participants in smokingcessation programs. J. Consult. Clin. Psychol. 55, 367–371.

’Connell, K.A., Shiffman, S., 1988. Negative affect smoking and smokingrelapse. J. Subst. Abuse 1, 25–33.

aty, J.A., Kassel, J.D., Shiffman, S., 1992. The importance of assessing baserates for clinical studies: an example of stimulus control of smoking. In:DeVries, M. (Ed.), The Experience of Psychopathology. Cambridge Uni-versity Press, Cambridge, 347–352.

ayne, T.J., Schare, M.L., Levis, D.J., Colletti, G., 1991. Exposure to smoking-relevant cues: effects on desire to smoke and topographical components ofsmoking behavior. Addict. Behav. 16, 467–479.

ayne, T.J., Smith, P.O., McCracken, L.M., McSherry, W.C., Anthony, M.M.,1994. Assessing nicotine dependence: a comparison of the FagerstromTolerance Questionnaire (FTQ) with the Fagerstrom Test for NicotineDependence (FTND) in a clinical sample. Addict. Behav. 19, 307–317.

omerleau, O., Adkins, D., Pertschuk, M., 1978. Predictors of outcome andrecidivism in smoking cessation treatment. Addict. Behav. 3, 65–70.

omerleau, O.F., Pomerleau, C.S., 1987. A biobehavioral view of substanceabuse and addiction. J. Drug. Issues 17, 111–131.

omerleau, O.F., Pomerleau, C.S., 1992. Euphoriant effects of nicotine in smok-ers. Psychopharmacology 108, 460–465.

eardon, S.F., Brennan, R., Buka, S.L., 2002. Estimating multi-level discrete-time hazard models using cross-sectional data: neighborhood effects onthe onset of adolescent cigarette use. Multivariate Behav. Res. 37, 297–330.

ussell, J., 1980. A circumplex model of affect. J. Pers. Soc. Psychol. 37,345–356.

ussell, M.A.H., Peto, J., Patel, U.A., 1974. The classification of smoking byfactorial structure of motives. J. R. Stat. Soc. 137, 313–346.

haham, Y., Shalev, U., Lu, L., DeWit, H., Stewart, J., 2003. The reinstate-ment model of drug relapse: history, methodology and major findings.

Psychopharmacology 168, 3–20.

hiffman, S., 1982. Relapse following smoking cessation: a situational analysis.J. Consult. Clin. Psychol. 50, 71–86.

hiffman, S., 1993. Assessing smoking patterns and motives. J. Consult. Clin.Psychol. 61, 732–742.

1 ohol D

S

S

S

S

S

S

S

S

S

S

S

S

S

S

S

S

S

T

68 S. Shiffman et al. / Drug and Alc

hiffman, S., 2007. Designing protocols for Ecological Momentary Assess-ment. In: Stone, A.A., Shiffman, S., Atienza, A., Nebeling, L. (Eds.), TheScience of Real-time Data Capture: Self-reports in Health Research. OxfordUniversity Press, New York, 27–53.

hiffman, S., 2006. Reflections on smoking relapse research. Drug Alcohol Rev.25, 15–20.

hiffman, S., Balabanis, M., 1995. Associations between alcohol and tobacco.In: Fertig, J.B., Allen, J.P. (Eds.), Alcohol and Tobacco: From Basic Scienceto Clinical Practice. NIH Research Monograph No. 30. National Institutesof Health, Bethesda, MD, 17–36.

hiffman, S., Fischer, L.A., Paty, J.A., Gnys, M., Kassel, J.D., Hickcox, M.,Perz, W., 1994a. Drinking and smoking: a field study of their association.Ann. Behav. Med. 16, 203–209.

hiffman, S., Gnys, M., Richards, T.J., Paty, J.A., Hickcox, M., Kassel, J.D.,1996a. Temptations to smoke after quitting: a comparison of lapsers andmaintainers. Health Psychol. 15, 455–461.

hiffman, S., Gwaltney, C.J., Balabanis, M.H., Liu, K.S., Paty, J.A., Kassel,J.D., Hickcox, M.M., Gnys, M., 2002. Immediate antecedents of cigarettesmoking: an analysis from Ecological Momentary Assessment. J. Abnorm.Psychol. 111, 531–545.

hiffman, S., Hickcox, M., Paty, J.A., Gnys, M., Kassel, J.D., Richards, T.,1996b. Progression from a smoking lapse to relapse: prediction from absti-nence violation effects, nicotine dependence, and lapse characteristics. J.Consult. Clin. Psychol. 64, 993–1002.

hiffman, S., Kassel, J.D., Paty, J.A., Gnys, M., Zettler-Segal, M., 1994b. Smok-

ing typology profiles of chippers and regular smokers. J. Subst. Abuse 6,21–35.

hiffman, S., Paty, J.A., 2006. Smoking patterns of non-dependent smokers:contrasting chippers and dependent smokers. J. Abnorm. Psychol. 115,509–523.

W

W

ependence 91 (2007) 159–168

hiffman, S., Paty, J.A., Gnys, M., Kassel, J.D., Hickcox, M., 1996c. First lapsesto smoking: within-subjects analysis of real time reports. J. Consult. Clin.Psychol. 64, 366–379.

hiffman, S., Paty, J.A., Gwaltney, C.J., Dang, Q., 2004a. Immediate antecedentsof cigarette smoking: an analysis of unrestricted smoking patterns. J.Abnorm. Psychol. 113, 166–171.

hiffman, S., Scharf, D.M., Shadel, W.G., Gwaltney, C.J., Dang, Q., Paton, S.M.,Clark, D.B., 2006. Analyzing milestones in smoking cessation: illustrationin a nicotine patch trial in adult smokers. J. Clin. Consult. Psychol. 74,276–285.

hiffman, S., Stone, A.A., 1998. Ecological Momentary Assessment: a newtool for behavioral medicine research. In: Krantz, D., Baum, A. (Eds.),Technology and Methods in Behavioral Medicine. Erlbaum, Mahwah, NJ,117–131.

hiffman, S., Waters, A.J., Hickcox, M., 2004b. The Nicotine Dependence Syn-drome Scale: a multidimensional measure of nicotine dependence. NicotineTob. Res. 6, 327–348.

hiffman, S., Waters, A.J., 2004. Negative affect and smoking lapses: a prospec-tive analysis. J. Consult. Clin. Psychol. 72, 192–201.

pearman, C., 1910. Correlation calculated from faulty data. Br. J. Psychol. 3,271–295.

tone, A.A., Shiffman, S., 1994. Ecological Momentary Assessment in behav-ioral medicine. Ann. Behav. Med. 16, 199–202.

iffany, S.T., 1990. A cognitive model of drug urges and drug-use behavior: roleof automatic and nonautomatic processes. Psychol. Rev. 97, 147–168.

ertz, J.M., Sayette, M.A., 2001. A review of the effects of perceived drug useopportunity on self-reported urge. Exp. Clin. Psychopharmacol. 9, 3–13.

est, R.J., Russell, M.A.H., 1985. Pre-abstinence smoke intake and smok-ing motivation as predictors of severity of cigarette withdrawal symptoms.Psychopharmacology 87, 334–336.