, Volume 229, Issue 1, pp 209–218

Cigarette craving and withdrawal symptoms during temporary abstinence and the effect of nicotine gum


    • Cancer Research UK Health Behaviour Research CentreUniversity College London
    • Health Behaviour Research Centre, Department of Epidemiology and Public HealthUniversity College London
  • Peter Hajek
    • Wolfson Institute of Preventive Medicine, Barts and The London School of Medicine and Dentistry, Queen MaryUniversity of London
  • Hayden McRobbie
    • Wolfson Institute of Preventive Medicine, Barts and The London School of Medicine and Dentistry, Queen MaryUniversity of London
  • Jo Locker
    • National Centre for Smoking Cessation and Training
  • Fiona Gillison
    • Department for HealthUniversity of Bath
  • Andy McEwen
    • Cancer Research UK Health Behaviour Research CentreUniversity College London
    • National Centre for Smoking Cessation and Training
  • Emma Beard
    • Cancer Research UK Health Behaviour Research CentreUniversity College London
  • Robert West
    • Cancer Research UK Health Behaviour Research CentreUniversity College London
    • National Centre for Smoking Cessation and Training
Original Investigation

DOI: 10.1007/s00213-013-3100-2

Cite this article as:
Brown, J., Hajek, P., McRobbie, H. et al. Psychopharmacology (2013) 229: 209. doi:10.1007/s00213-013-3100-2



It is widely believed that nicotine withdrawal symptoms appear within a few hours of stopping smoking, but few data exist documenting their emergence in naturalistic settings. In several countries, nicotine replacement products are licensed for relief of withdrawal symptoms during temporary abstinence, but again, there are no data supporting this from naturalistic settings.


To examine the emergence of cigarette craving and withdrawal symptoms during temporary abstinence in a naturalistic setting while using either nicotine or placebo gum.


Double-blind, randomised, placebo-controlled study in which 132 dependent smokers abstained for 6 h with the assistance of either nicotine (2 mg, n = 42 or 4 mg, n = 24) or placebo (n = 66) gum while travelling on a non-smoking train. Outcome measures were ratings of craving and mood withdrawal symptoms prior to treatment and at regular intervals during abstinence.


In a multivariate analysis of all symptoms, there was no interaction between treatment and time [F(21,110) = 1.28, p = 0.20, \( \eta_{\mathrm{p}}^2 \) = 0.20] nor an effect of treatment [F(7,124) = 0.45, p = 0.87, \( \eta_{\mathrm{p}}^2 \) = 0.03]. There was an effect of time [F(21,110) = 11.59, p < 0.001, \( \eta_{\mathrm{p}}^2 \) = 0.69) and univariate analyses revealed that the majority of symptoms increased linearly throughout the period of abstinence with detectable onsets typically between the first 60 and 180 min of abstinence.


Smokers who temporarily abstain in naturalistic settings experience craving and withdrawal symptoms that emerge linearly over the first 6 h of abstinence. Changes in craving and several mood withdrawal symptoms can be detected within the first 3 h. Nicotine gum may not have an acute effect on the development of these symptoms.


Temporary abstinenceCravingWithdrawal symptomsNicotine gumSmoking


Withdrawal symptoms such as craving are important in nicotine dependence as reflected by the Diagnostic and Statistical Manual of Mental Disorders (DSM) and International Classification of Diseases (ICD) diagnostic criteria (American Psychiatric Association 1994; Baker et al. 2012; World Health Organisation 1992), most theoretical accounts of cigarette addiction (Edwards et al. 1976; Koob and Le Moal 2008; Skinner and Aubin 2010; Tiffany 1990; West 2006) and their predictive relationship with relapse to smoking (Baker et al. 2004; Fidler et al. 2010; Killen and Fortmann 1997; Piasecki et al. 2000; Shiffman et al. 1996; Swan et al. 1996). Withdrawal symptoms are temporary physical and psychological changes occurring as a result of abstinence from a substance or activity to which adaptation has occurred. Although there remains a debate as to which symptoms constitute the withdrawal syndrome, commonly assessed features include irritability, poor concentration, restlessness, depressed mood, anxiety, increased appetite and craving (Hughes et al. 1990, 1991; West and Hajek 2004). Cigarette craving is commonly defined as the subjective experience of a strong motivation to smoke (West 2006). Under this operationalisation, craving and the urge to smoke are interchangeable, which reflects the experience of smokers (Sayette et al. 2000). Craving undoubtedly increases following abstinence and is a critical feature of cigarette addiction (West and Schneider 1987), which is underlined by craving appearing the most reliable predictor of relapse (Fidler et al. 2010; Killen and Fortmann 1997; Swan et al. 1996; West and Ussher 2010). However, the labelling of craving as a part of withdrawal is somewhat contentious because it is typically less related to other symptoms than they are to each other (Hughes 2007a, b) and persists for a longer period of time (Ussher et al. 2012). For example, DSM-IV does not list craving as a symptom of withdrawal—only as one of dependence. It appears likely that DSM-V will amend this position (Baker et al. 2012; Hughes et al. 2011).

It is widely believed that withdrawal symptoms emerge within a few hours of stopping smoking; however, there is inconsistency between different researchers who specify timeframes varying between 2 h (Shiffman et al. 2002) and 6–12 h (Hughes et al. 1994). The reason is that despite the presumed importance of withdrawal symptoms in addiction and the extent of research into its time course over the days, weeks and months following abstinence (Hughes 2007a; McCarthy et al. 2006; Ussher et al. 2012), there is relatively little evidence concerning the acute development of symptoms (Hughes 2007a). In the few studies that have examined the issue directly, different withdrawal symptoms appear to emerge variously between 1 and 6 h after the onset of abstinence (Gross et al. 1997; Hendricks et al. 2006; Parrott et al. 1996; Schuh and Stitzer 1995; Tiffany and Drobes 1991). This speed of onset is consistent with rapidly declining nicotine levels in the brain as a result of the 10-min distributional half-life of nicotine (Russell 1988). However, all these studies were conducted in laboratory settings, and given the importance of cues to the onset of withdrawal symptoms, particularly cravings (Bedi et al. 2011; Freeman et al. 2012; West 2009), there is a need to document the emergence of symptoms in naturalistic settings.

In several countries including Australia, Brazil, Canada and the UK, nicotine replacement products are licensed for relief of withdrawal symptoms during periods of temporary abstinence (TA) (Medicines and Healthcare Products Regulatory Agency 2010), with as many as one in eight smokers using it for this purpose (Beard et al. 2013a, b). TA can occur when smoking is not permitted, for example while visiting a hospital or travelling on a train or aeroplane. It can also be self-imposed, for example indoors at home (Beard et al. 2013b). The licensing represents a harm reduction strategy, which involves any attempt to reduce the psychological or physical harm from tobacco use without complete cessation. The strategy is supported by the laboratory findings on the early onset of withdrawal symptoms and the short half-life of nicotine, both of which suggest withdrawal symptoms would emerge during a period of TA. Indeed, theory and research into the importance of contextual cues in nicotine addiction imply that craving and withdrawal symptoms would be exacerbated in natural settings (Bedi et al. 2011; Freeman et al. 2012; West 2009). There is also indirect evidence to suggest that nicotine replacement therapy (NRT) would provide acute relief from these symptoms during TA: NRT relieves withdrawal symptoms that result from longer periods of abstinence (e.g., West et al. 1984), and its use is associated with cessation among smokers attempting to quit smoking altogether (Brose et al. 2011; Stead et al. 2008). However, current licensing is somewhat unsupported by the lack of direct data to demonstrate an effect of NRT on craving and withdrawal symptoms during TA in naturalistic settings.

This study addressed the following research questions:
  1. 1.

    How is the development of cigarette craving and withdrawal symptoms during abstinence over several hours in a naturalistic setting affected by nicotine gum?

  2. 2.

    What is the time course of craving and withdrawal symptoms while using either nicotine or placebo gum during abstinence in a naturalistic setting?



Study design

The study was a double blind, randomised, placebo-controlled trial of the effect of nicotine gum on the development of craving and withdrawal symptoms during 6 h periods of TA undertaken in the naturalistic setting of train journeys from London to Glasgow, and then returning the following day.


To assist with the 6 h period of TA, participants all received either Nicorette ® gum randomly containing either 2 or 4 mg of nicotine or a placebo gum prepared to be similar in appearance and taste. The intention was to allocate 2 or 4 mg gum (or matching placebo) based on cigarette consumption whereby participants smoking more than 20 cigarettes per day would be allocated the stronger 4 mg gum. However, there was a labelling error. Within each batch of gum provided to the research team, the labels were meant to systematically refer to either ‘2 mg or placebo’ or ‘4 mg or placebo’. Instead, whole batches of gum were either all 2 mg or placebo, or all 4 mg or placebo, and the systematic labelling within batches actually bore no relationship to dose. The mistake was only revealed after the labels were unblinded. The result was that participants were in fact randomly block allocated to either the 2 mg or placebo gum, or to the 4 mg or placebo gum, with 60 % of participants on active treatment receiving 2 mg gum. The nicotine and placebo gum were allocated according to a 1:1 ratio. All gums were supplied free of charge by Pharmacia. The nicotine gum contained nicotine bound to an ion-exchange resin to permit slow release of nicotine and an alkaline hydrocarbonate buffer to increase the buccal absorption of nicotine by increasing salivary pH to 8.5. The gum also contained gum base, sorbitol and flavour. All participants were asked to take at least one piece of gum every hour beginning after the first half-hour of abstinence.


The outcome measures were mean ratings of strength of urges and a variety of other withdrawal symptoms (poor concentration, restlessness, irritability, hunger, depressed mood and anxiety) during the past 30 min, which were assessed using the validated Mood and Physical Symptoms Scale (MPSS, West and Hajek 2004). The relatively brief MPSS has been found to be at least as reliable and sensitive to abstinence as longer measures such as the Questionnaire of Smoking Urges (West and Ussher 2010). During abstinence, assessment occurred every half an hour for 5.5 h (330 min) beginning half an hour after the onset of abstinence (see Fig. 1). For the purposes of analysis, the measurements were averaged across four separate time periods: 0–30 min (prior to medication), 31–120 min, 121–240 minutes and 241–360 min. Baseline assessment evaluated age, sex, ethnicity, difficulty with temporary abstinence (‘During your normal day, how hard do you find it not to smoke for more than three hours? 1 = not at all difficult to 5 = almost impossible’), previous experience with nicotine replacement therapy, the Heaviness of Smoking Index (Heatherton et al. 1989), age of smoking initiation, previous attempts to quit smoking and carbon monoxide (CO) levels in exhaled breath.
Fig. 1

Timing of assessment. The dashed dot vertical line indicates the onset of treatment


Participants were recruited by advertisements placed in noticeboards and newspapers local to London, UK. All participants who were enrolled in the study met the following inclusion criteria: were aged 18 or older; reported regular smoking for 3 years or more, current daily smoking of at least 15 cigarettes and some difficulty with temporary abstinence for more than 3 h (3 or more on 5-point scale used in baseline assessment); registered a CO of 10 ppm or more after at least 15 smoke-free minutes; and were willing to refrain from smoking for the abstinence periods during the study, to use nicotine or placebo gum, to fill in questionnaires for the study, and able to provide informed consent. Participants were excluded from enrolling in the study if they reported any of the following: current use of NRT or other nicotine containing products (other NRT, snuff, cigars and pipe) or behavioural or pharmacological smoking cessation program; daily use of nicotine gum during the last year; pregnancy, lactation or planned pregnancy; participation in another clinical study during the last 3 months; unstable angina pectoris, myocardial infarction within the last 3 months; under psychiatric care or medication that might interfere with the trial; an alcohol or other drug problem; and significant mouth pathology; and/or dental problems that precluded use of gum. Of those participants who enrolled in the trial, participants were subsequently excluded if they failed to refrain from smoking during the periods requiring abstinence, failed to provide an outcome rating prior to treatment or completed <95 % of outcome measurements in total.

The target sample size was determined on the basis of detecting a difference in mean ratings of craving and withdrawal symptoms between treatment groups. Previous studies indicated that a standard deviation of about 1 in the computed score was likely and that a difference in group means is of 0.5 of clinical importance (West and Hajek 2004). Sixty-four participants in each of the two groups would have provided 80 % power to detect such an effect with alpha set to 0.05. To allow for uncertainty in the assumptions, the target sample size was set to 150 participants.


Following telephone screening, eligible participants were invited for baseline assessment. Between 1 and 5 days after baseline assessment, participants boarded a train travelling between London and Glasgow and were asked to abstain from smoking for 6 h after which they changed trains for the final part of the journey. Participants were provided with 30 pieces of either nicotine or placebo gum, instructed as to how best to use the gum and then asked to take at least one piece of gum every hour beginning after the first half-hour of abstinence. During the period of abstinence, participants were asked to record the extent of their withdrawal symptoms during the past 30 min every half an hour after the first 30 min had passed. A research team travelled with participants and reminded them whenever they were due to use a piece of gum or complete a questionnaire. In the summer of 2002, participants travelled together in three groups of approximately 55 accompanied by the research team on a non-smoking train to Glasgow and then on a train to Edinburgh, which allowed smoking where they stayed overnight (the time in Edinburgh was a part of the incentive for taking part in the study). On the train from Glasgow to Edinburgh, participants were allowed to smoke, and the effect on their CO was measured and shown to them (this part of the study is the focus of another experiment examining the effect of a ‘CO boost’ on longer-term smoking cessation outcomes and will be reported separately). The procedure was repeated on the return journey the following day, without re-randomisation—participants received the same gum to which they had been randomly allocated on the previous day—and without the final CO boost element. Participants received free train travel, a free night in university accommodation and an additional £25.


The emergence of withdrawal symptoms during TA in naturalistic settings and the effect of nicotine gum were assessed by a mixed MANOVA model including the within-subject effects of time and journey and the between-subject effect of treatment and their interactions. Significant omnibus results were investigated by univariate analyses and Sidak-adjusted t test comparisons. All analyses were conducted with missing data imputed from measurements taken within the same interval of time for which data was missing (with the exception that all participants who failed to provide a rating prior to treatment or who missed more than 5 % of outcome measurements were excluded). In two sensitivity analyses, (1) the effect of treatment with three levels (placebo vs 2 mg vs 4 mg nicotine gum) was examined and (2) all analyses were repeated using only complete cases. Effect sizes were reported as either partial eta-squared or Cohen’s d. Partial eta-squared was provided when a relative measure of effect size was more appropriate.


One hundred and fifty-four adult smokers were enrolled into the study and randomised to receive placebo or nicotine gum. Four reported smoking during the course of the two journeys, 15 failed to provide a rating prior to treatment on one of the two journeys and seven missed more than 5 % of the outcome measurements. The number excluded for each reason was similar between the conditions (smoking: nicotine gum, n = 2; placebo gum, n = 2; missing data on the first collection on either journey: nicotine gum, n = 8; placebo gum, n = 7; missing more than 5 % of data: nicotine gum, n = 3; placebo gum, n = 4). One hundred and thirty-two participants who both successfully abstained and provided sufficient outcome data were included in the study. Of these participants, the mean number of outcome measurements provided was 166 out of 168 although only 83 provided all measurements. The extent of missing data was a procedural consequence of collecting a large amount of time-sensitive, self-report data over an extended period in a busy and naturalistic setting.

The baseline characteristics of included participants are presented in Table 1. The study sample contained a similar number of men and women, was predominantly white and had a mean age in the middle 30s. Nearly a third of the sample had previous experience with using nicotine replacement therapy. Other smoking characteristics were typical of a dependent sample of smokers. Randomisation appeared successful as baseline characteristics were similar between the groups.
Table 1

Characteristics of study participants


Placebo (n = 66)

Nicotine gum (n = 66)

Total (n = 132)

Agea, mean (SD)

34.3 (11.5)

36.3 (12.0)

35.3 (11.8)

Women, % (N)

51.5 (34)

51.5 (34)

51.5 (68)

White, % (N)

89.4 (59)

89.4 (59)

89.4 (118)

Previous experience of NRT, % (N)

25.8 (17)

33.3 (22)

29.5 (39)

Cigarettes per day, mean (SD)

23.1 (6.3)

23.3 (6.7)

23.2 (6.5)

Smoking within 30 min of waking, % (N)

87.9 (58)

86.4 (57)

87.1 (115)

Age of smoking initiationb, mean (SD)

15.9 (3.0)

16.8 (2.9)

16.3 (3.0)

Lifetime quit attempt, % (N)

75.8 (50)

81.8 (54)

78.8 (104)

Carbon monoxide in exhaled breath, mean (SD) (ppm)

28.4 (11.1)

27.0 (10.5)

27.7 (10.8)

aData on age were missing for one participant in the placebo and for two in the nicotine gum group

bData on age of smoking initiation were missing for one participant in the placebo and one in the nicotine gum group

Figure 2 presents the development of craving and withdrawal symptoms over the 360-min periods of TA by treatment group during the different journeys. In a multivariate analysis of craving and withdrawal symptoms, there was no interaction between treatment and time [F(21,110) = 1.28, p = 0.20, \( \eta_{\mathrm{p}}^2 \) = 0.20], nor was there a main effect of treatment [F(7,124) = 0.45, p = 0.87, \( \eta_{\mathrm{p}}^2 \) = 0.03]. There was also no evidence that any treatment effect depended upon whether it was the outgoing or return journey [F(7,124) = 0.53, p = 0.81, \( \eta_{\mathrm{p}}^2 \) = 0.03] or that there was any three-way interaction [F(21,110) = 1.28, p = 0.99, \( \eta_{\mathrm{p}}^2 \) = 0.06]. There was an effect of time [F(21,110) = 11.59, p < 0.001, \( \eta_{\mathrm{p}}^2 \) = 0.69], journey [F(7,124) = 11.08, p < 0.001, \( \eta_{\mathrm{p}}^2 \) = 0.39] and an interaction between them [F(21,110) = 3.93, p < 0.001, \( \eta_{\mathrm{p}}^2 \) = 0.43].
Fig. 2

The development of mean ratings of withdrawal symptoms during a 6-h period of temporary abstinence by treatment group during an a outgoing and then b return train journey. The dashed dot vertical line indicates the onset of treatment. Error bars are SEMs

In the univariate analyses to examine the effects of time, journey and their interaction on the different symptoms, the first two time intervals were collapsed to create three equally balanced intervals as there was no effect of treatment from the multivariate analysis (see Fig. 1). Figure 3 presents the development of symptoms over time collapsed across treatment groups during the different journeys. There was a linear increase over time for all the craving and withdrawal symptom symptoms [urge to smoke: F(1,131) = 167.69, p < 0.001, \( \eta_{\mathrm{p}}^2 \) = 0.56; depressed mood: F(1,131) = 15.05, p < 0.001, \( \eta_{\mathrm{p}}^2 \) = 0.10; anxiety: F(1,131) = 49.64, p < 0.001, \( \eta_{\mathrm{p}}^2 \) = 0.28; irritability: F(1,131) = 96.23, p < 0.001, \( \eta_{\mathrm{p}}^2 \) = 0.42; restlessness: F(1,131) = 102.15, p < 0.001, \( \eta_{\mathrm{p}}^2 \) = 0.44; poor concentration: F(1,131) = 44.06, p < 0.001, \( \eta_{\mathrm{p}}^2 \) = 0.25], except hunger [F(1,131) = 3.37, p = 0.07, \( \eta_{\mathrm{p}}^2 \) = 0.03]. While there was only a main effect of journey on urge to smoke [F(1,131) = 25.29, p < 0.001, \( \eta_{\mathrm{p}}^2 \) = 0.16], anxiety [F(1,131) = 6.52, p = 0.01, \( \eta_{\mathrm{p}}^2 \) = 0.05] and restlessness [F(1,131) = 11.95, p < 0.01, \( \eta_{\mathrm{p}}^2 \) = 0.08], there was evidence that for all the symptoms that there was an interaction between time and journey [urge to smoke: F(2,262) = 14.34, p < 0.001, \( \eta_{\mathrm{p}}^2 \) = 0.10; depressed mood: F(2,262) = 6.53, p < 0.01, \( \eta_{\mathrm{p}}^2 \) = 0.05); anxiety: F(2,262) = 11.84, p < 0.001, \( \eta_{\mathrm{p}}^2 \) = 0.08; irritability: F(2,262) = 13.79, p < 0.001, \( \eta_{\mathrm{p}}^2 \) = 0.10; restlessness: F(2,262) = 8.78, p < 0.001, \( \eta_{\mathrm{p}}^2 \) = 0.06; poor concentration: F(2,262) = 4.90, p < 0.01, \( \eta_{\mathrm{p}}^2 \) = 0.04; hunger: F(2,262) = 15.23, p < 0.001, \( \eta_{\mathrm{p}}^2 \) = 0.10]. Simple effect analyses of the interactions revealed that the linear increase over time remained on both journeys for all the symptoms [Fs(1,131) > 11.91, ps < 0.001, \( \eta_{\mathrm{p}}^2 \) > 0.08] with the exception of depressed mood and hunger, which increased linearly on the return journey [depressed mood: F(1,131) = 17.19, p < 0.001, \( \eta_{\mathrm{p}}^2 \) = 0.12; hunger: F(1,131) = 19.98, p < 0.001, \( \eta_{\mathrm{p}}^2 \) = 0.13)] but not the outgoing journey [depressed mood: F(1,131) = 3.20, p = 0.08, \( \eta_{\mathrm{p}}^2 \) = 0.02; hunger: F(1,131) = 3.11, p = 0.08, \( \eta_{\mathrm{p}}^2 \) = 0.02]. In comparisons of the first two time intervals to examine the onset of symptoms on each journey, ratings were significantly higher by 180 min into the train journey (i.e. 121–240 min interval) than in the initial part (first 60 min, i.e. 0–120 min interval) for urge to smoke, irritability, restlessness and concentration on both the outgoing [t(131) = 5.95, p < 0.001, d = 0.47; t(131) = 5.26, p < 0.001, d = 0.39); t(131) = 5.83, p < 0.001, d = 0.48; t(131) = 2.98, p < 0.05, d = 0.22 respectively] and return journeys [t(131) = 8.13, p < 0.001, d = 0.68; t(131) = 5.89, p < 0.001, d = 0.47; t(131) = 5.80, p < 0.001, d = 0.47; t(131) = 3.33, p < 0.01, d = 0.24 respectively]. In the case of anxiety and hunger, the onset was first detected between the first 60 min of abstinence and after 180 min only on the return journey [t(131) = 4.63, p < 0.001, d = 0.37, t(131) = 3.40, p < 0.01, d = 0.31, respectively]. Therefore, the interactions appeared to arise primarily from differences in the overall gradient of symptom development on the different journeys rather than from consistent differences in the onset or shape (see Fig. 3).
Fig. 3

The onset of withdrawal symptoms during a 6-h period of temporary abstinence during an a outgoing and then b return train journey. Stars indicate whether there was a significant difference between the first time two intervals. Error bars are SEMs

In a sensitivity analysis that examined the multivariate effect of treatment with the three levels of placebo (n = 66), 2 mg (n = 42) or 4 mg (n = 24) nicotine gum, there remained no interaction between treatment and time [F(42,220) = 1.07, p = 0.37, \( \eta_{\mathrm{p}}^2 \) = 0.17], no effect of treatment [F(14,248) = 0.67, p = 0.80, \( \eta_{\mathrm{p}}^2 \) = 0.04], no interaction between treatment and journey [F(14,248) = 0.53, p = 0.91, \( \eta_{\mathrm{p}}^2 \) = 0.03] and no three-way interaction [F(42,220) = 0.46, p = 0.99, \( \eta_{\mathrm{p}}^2 \) = 0.08]. In another sensitivity analysis that repeated all analyses using only complete cases (n = 83), the pattern of results was unchanged.


While temporarily abstaining in a naturalistic setting, smokers who were using either nicotine or placebo gum reported a modest linear increase in the strength of their craving and withdrawal symptoms over a 6-h period of TA. There was a detectable change in urge to smoke, irritability, restlessness and concentration within the first 3 h on both journeys. For the majority of symptoms, the linear development and onset were similar between both journeys although there was evidence that the precise gradient of the linear development differed between the two journeys. There was no acute effect of nicotine gum on the development of these symptoms compared with placebo.

The current estimates of the early time course of self-reported withdrawal symptoms obtained in naturalistic settings replicate those obtained by a similar study conducted in a laboratory setting (Hendricks et al. 2006). Both studies detected increases in a number of withdrawal symptoms including craving, irritability/anger, anxiety and concentration within the first 3 h, and all symptoms appeared to emerge linearly. The early emergence of craving also replicates several other laboratory studies that assessed craving but no other withdrawal symptoms (Gross et al. 1997; Schuh and Stitzer 1995; Tiffany and Drobes 1991). Altogether, the replication of laboratory findings on self-reported symptoms in a naturalistic setting confirms the onset of withdrawal symptoms within 3 h of abstinence. This result is strengthened by laboratory findings of similarly early differences in physiological and cognitive performance (Hendricks et al. 2006), which are difficult to examine within naturalistic settings.

The fact that the key characteristics of the early time course for the majority of symptoms—including the linear emergence of symptoms and the timing of the onset—were similar across the different journeys reinforces the generalisability of the result. However, there was evidence of some differences between the journeys in the overall gradient of symptom development, which appeared to be driven by steeper gradients on the return journey. The differences may have resulted from greater reactivity to assessment while the experience was novel at the beginning of the outgoing journey. In particular, unusually high ratings at the start of this type of experiment would have obscured detection of typical symptom trajectories. However, this explanation would need to be examined systematically in future research.

In several countries, nicotine replacement products are licensed for relief of withdrawal symptoms during periods of TA. The current finding that there was no acute effect of nicotine gum on the development of these symptoms within naturalistic settings may be seen to somewhat undermine this position. However, there are a number of mitigating factors, which should limit the generalisability of this result. An acute effect on withdrawal symptoms during TA may not have been detected because (1) the present study used primarily 2 mg gum in predominantly heavy smokers and, although there was no detectable effect of the 4 mg gum either, the 4 mg subsample was small; (2) gum does not replace some of the behavioural or sensory-motor aspects of smoking, is not as fast-acting as some NRT products such as the mouth spray or nasal spray, and has been rated among the least helpful nicotine replacement product for the purpose of TA (Beard et al. 2013b); (3) the sample were largely inexperienced with NRT and may have required more learning episodes given NRT is often incorrectly and under-used at first; and (4) TA occurs in a variety of settings and the use of NRT may not be required or effective in all of them (Beard et al. 2013b). However, this final explanation seems unlikely as the study was conducted in England shortly after smoking on trains was banned on certain journeys, which would have meant a ‘train setting’ contained cues to smoke learned prior to the ban for most smokers, and thereby provided a sensitive context in which to test the effect of nicotine gum. Indeed, 40 % of smokers who use NRT for TA appear to do so for the purpose of ‘travelling’ (Beard et al. 2013b). Nonetheless, future research could systematically investigate each of these four possibilities.

By contrast, there remains existing evidence to support the licensing position on the use of NRT for TA that should prevent its immediate re-appraisal. First, clinical and population-level data demonstrate that the use of NRT for harm reduction (either TA or as a means of ‘cutting down’) can move smokers towards a successful quit attempt (Beard et al. 2012, 2013b; Beard and West 2012; Moore et al. 2009), and it may be that the use of NRT for TA leads to quit attempts by some mechanism other than experience of effective relief for acute cravings and withdrawal symptoms (Beard et al. 2012). Second, although there is no direct evidence for the relief of withdrawal symptoms by NRT in typical situations requiring TA, there is a substantial body of research demonstrating the effectiveness of NRT for permanent smoking cessation (e.g., Brose et al. 2011), which is assumed to be mediated by withdrawal relief.

The naturalistic conditions were an important feature of the study, but there were limits. The practical need for participants to travel in relatively large groups and to travel exclusively for research purposes could have affected the results. For example, there is evidence that peer influence affects cigarette consumption (Harakeh et al. 2007). Therefore, there could be value in future research that seeks an even closer approximation to reality using emerging technologies for ecological momentary assessment such as smartphone applications to examine craving and withdrawal symptoms in those smokers whose everyday routines would ordinarily involve temporary abstinence and the use of NRT. Another limitation is that at least some data were imputed in approximately 40 % of the participants. Prevalent missing data are a procedural consequence of naturalistic data collection particularly when large amounts of data are required that require active introspection (cannot be automated) and are time sensitive (cannot be provided retrospectively if accidentally omitted). However, on average, participants provided 166 out of 168 outcome measurements and in a sensitivity analysis that included only those providing complete cases, the pattern of results was unchanged. An additional limitation relates to the study design. Ideally, the time course of withdrawal symptoms would have been analysed in comparison with a control group who were not temporarily abstaining, i.e. smoking. However, this is not particularly practical in settings in which TA genuinely occurs (i.e. where smoking is not permitted), and previous research from the laboratory suggests that these symptoms remain relatively stable when normal smoking is permitted, which suggests the within-subject comparison over time used in the current study is valid (Hendricks et al. 2006). A final limitation is that all smokers were using either nicotine or placebo gum and that, as a result, the estimates of the time course may be conservative as there may have been a placebo effect of the gums. The fact that the current findings in naturalistic settings correspond with estimates from laboratory research suggests that this is unlikely; however, further research should explore the possibility.


Smokers who are temporarily abstaining in naturalistic settings while using nicotine or placebo gum experience craving and withdrawal symptoms that emerge linearly over the first 6 h of abstinence. Changes in the majority of symptoms, including the urge to smoke, can be detected within the first 3 h. Nicotine gum may not have an acute effect on the development of these symptoms within a typical period of temporary abstinence. Further research is needed to examine whether this was related to the strength of the gum and whether acute effects would emerge with more experience or different types of NRT.


We are grateful to Pharmacia for funding this study and to Björn Landfeldt and his Pharmacia colleagues for their help in designing the study. JB’s post is funded by the National Prevention Research Initiative. RW is funded by Cancer Research UK. The research team is part of the UK Centre for Tobacco Control Studies. We thank Leonie Brose, Kristjan Laane and Lion Shahab for providing comments on a draft of this manuscript.

Statement of competing interests

Pharmacia both funded the study and manufactured the smoking cessation medication that was investigated. Pharmacia have since merged with Pfizer. Individuals at Pharmacia provided advice on study design and data collection; however, final decisions were taken by the investigators and were unrestricted. RW, AMcE, PH and HM have undertaken research and consultancy for companies that develop and manufacture smoking cessation medications. RW and AMcE have a share of a patent for a novel nicotine delivery device. There are no other relationships or activities that could appear to have influenced the submitted work; in particular, the delay between data collection and publication was not a result of a sponsorship clause.

Copyright information

© Springer-Verlag Berlin Heidelberg 2013