The acute effects of exercise on cigarette cravings, withdrawal symptoms, affect, and smoking behaviour: systematic review update and meta-analysis

Transcription

1 Psychopharmacology (2012) 222:1 15 DOI /s z REVIEW The acute effects of exercise on cigarette cravings, withdrawal symptoms, affect, and smoking behaviour: systematic review update and meta-analysis Vaughan Roberts & Ralph Maddison & Caroline Simpson & Chris Bullen & Harry Prapavessis Received: 9 October 2011 /Accepted: 23 April 2012 /Published online: 15 May 2012 # Springer-Verlag 2012 Abstract Rationale Smoking cessation is associated with cigarette cravings and tobacco withdrawal symptoms (TWS), and exercise appears to ameliorate many of these negative effects. A number of studies have examined the relationships between exercise, cigarette cravings, and TWS. Objectives The objectives of this study were (a) to review and update the literature examining the effects of short bouts of exercise on cigarette cravings, TWS, affect, and smoking behaviour and (b) to conduct meta-analyses of the effect of exercise on cigarette cravings. Methods A systematic review of all studies published between January 2006 and June 2011 was conducted. Results Fifteen new studies were identified, 12 of which found a positive effect of exercise on cigarette cravings. The magnitude of statistically significant effect sizes for desire to smoke and strength of desire to smoke ranged from 0.4 to 1.98 in favour of exercise compared to passive control conditions, and peaked either during or soon after treatment. Effects were found up to 30 min post-exercise. Cigarette cravings were reduced following exercise with a wide range of intensities from isometric exercise and yoga to activity as high as % heart rate reserve. Metaanalyses revealed weighted mean differences of 1.90 and 2.41 in desire to smoke and strength of desire to smoke outcomes, respectively. Measures of TWS and V. Roberts (*) : R. Maddison : C. Simpson : C. Bullen National Institute for Health Innovation, University of Auckland, Private Bag 92019, Auckland 1142, New Zealand v.roberts@nihi.auckland.ac.nz H. Prapavessis School of Kinesiology, University of Western Ontario, 1151 Richmond Street, London, ON, Canada N6A 3K7 negative affect were reduced following light moderate intensity exercise, but increased during vigorous exercise. Conclusions Exercise can have a positive effect on cigarette cravings and TWS. However, the most effective exercise intensity to reduce cravings and the underlying mechanisms associated with this effect remain unclear. Keywords Nicotine. Exercise. Physical activity. Smoking cessation. Affect. Smoking topography. Tobacco withdrawal symptoms. Cigarette cravings. Behavioural intervention Background Smoking cessation is associated with tobacco withdrawal symptoms (TWS) such as sleep disturbance, irritability, poor concentration, and depressed mood, as well as intense craving for a cigarette (Hughes 2007). Craving, the presence of withdrawal symptoms, and weight gain are associated with an increased risk of smoking relapse and impact negatively on attempts to quit (West et al. 1989; Borrelli et al. 2001). Therefore, identifying ways to reduce these negative effects of smoking abstinence may be important in increasing the success of quit attempts. Participation in regular exercise appears to ameliorate the intensity and frequency of many of the TWS and cravings associated with smoking cessation. However, findings from intervention studies designed to investigate if exercise assists individuals to quit smoking have shown mixed results (Ussher et al. 2012). A clearer understanding of the relationships between exercise, TWS, and cravings during temporary smoking abstinence may improve the design, and ideally the effectiveness, of future exercise-based smoking cessation interventions.

2 2 Psychopharmacology (2012) 222:1 15 Since the publication in 2007 of a systematic review of studies examining the acute effects of exercise during temporary smoking abstinence (Taylor et al. 2007), a number of studies have been published that address this issue. The purpose of this paper is twofold. First, we update the literature on the acute effects of exercise on TWS, cravings, and affect during temporary smoking abstinence, and examine potential mediators of the exercise TWS relationship. For the purpose of comparing current knowledge to the conclusions of an earlier review, we include a summary of the findings of the previous review and employ methods consistent with that review (Taylor et al. 2007). We highlight the findings of studies published since the previous review, discuss progress made, and the priorities and implications for future research. Second, we conduct a meta-analysis of all published studies to date to determine more robust estimates of the strength of the effect of exercise on selected cigarette cravings outcomes. Summary of the previous review The systematic review by Taylor et al. (2007) identified 14 studies published between 1983 and 2006 that examined the effect of exercise on TWS. Of the 12 studies that compared exercise with a control condition, all showed at least one positive effect on withdrawal (Taylor et al. 2007). This was the case for both brief (5 to 10 min) bouts of moderate intensity exercise among smokers who were abstinent overnight (Daniel et al. 2004; Ussher et al. 2001) and for 30 to 40 min bouts of vigorous intensity among smokers trying to quit (Bock et al. 1999). Withdrawal symptoms found to be affected by brief bouts of exercise included anxiety, stress, poor concentration, tension, restlessness, and irritability. Nine (Bock et al. 1999; Danieletal.2004; Taylorand Katomeri 2007; Tayloretal.2005, 2006; Thayeretal. 1993; Ussher et al. 2001, 2006; Daniel et al. 2006; Katomeri 2009) of the ten studies that compared the effects of an exercise condition with a passive condition on cravings found a significant reduction in cravings following exercise. Studies that assessed strength of urges to smoke showed that an average reduction of 1.1 points on a seven-point scale could be achieved. Three of the studies (Bock et al. 1999; Taylor et al. 2006; Thayer et al. 1993) found that exercise had a positive effect on mood and affect during smoking abstinence, increasing activation and energy and decreasing negative affect and tension. One study suggested tension was a mediating factor for reductions in desire to smoke (Taylor et al. 2006). The review included four studies that measured time to ad libitum smoking, all of which found increases ranging from 8 to 57 min following exercise (Taylor and Katomeri 2007; Thayer et al. 1993; Reeser 1983; Katomeri 2009). Except for one study (Bock et al. 1999), which measured cravings throughout a 12-week smoking cessation programme, all of the studies measured cravings and TWS during temporary smoking abstinence only, and all were conducted in the laboratory setting. Taylor et al. (2007) highlighted some key issues to be addressed in future research. This included the need to increase the ecologic validity of subsequent findings and to elucidate the mechanisms underpinning the beneficial effects of exercise on TWS. In regard to this, Taylor et al. surmised that distraction is unlikely to be a mediating factor and proposed that other potential mechanisms such as stress reduction and activation, psychobiological mechanisms, and appetite suppression may be important. The authors suggested that exercise may mimic the effects of smoking by both relaxing and activating the individual. Some evidence pointed to reductions in stress levels playing a part in mediating decreased desire to smoke, but the authors considered more research was needed to examine the relationship between exercise and smoking cravings with regard to stress and boredom. Additional research to examine the role of β-endorphins and opioids as well as dopaminergic activity in the context of exercise and cigarette cravings and TWS was also proposed. Taylor et al. (2007) concluded that a brief bout of exercise can reduce cravings for a cigarette at levels comparable to those of glucose and oral nicotine replacement therapy, and should therefore be recommended. Ussher et al. (2012) provide a brief summary of studies conducted in this area since the review by Taylor et al. (2007). The following review builds on the summary by Ussher et al. (2012). Method We undertook a systematic search of the literature using online searches of the following electronic data bases: Sports Discus, MEDLINE, PubMed, Web of Science, EMBASE, PsycINFO, Cochrane Tobacco Addiction Group specialized register, the ETD Digital Library Networked Digital Library of Theses and Dissertations, and Proquest Digital Dissertations. The keywords used were exercise, physical activity, smoking, tobacco, nicotine, smoking cessation, withdrawal, craving, and affect. The search was limited to all studies of human adult participants ( 18 years) published between January 2006 and February There were no specific criteria for study design. The search of electronic databases identified 144 studies. The first and third authors examined the abstracts of all studies. Where a decision to include or exclude a study could not be attained from the abstract, the full article was reviewed. We contacted prominent researchers in the field and asked for any studies currently under review or in press. Four further studies were identified. We also hand-searched the

3 Psychopharmacology (2012) 222: reference lists of relevant articles and abstracts from the Society for Research on Nicotine and Tobacco annual meetings from Decisions to exclude a study were made based on the title of the abstract, and the remaining abstracts were reviewed for inclusion. In accordance with the previous review, we included all journal articles, conference abstracts, theses, and dissertations that studied the effects of a brief bout of exercise on tobacco withdrawal symptoms, cravings, or affect during smoking abstinence. Results The search yielded 15 new studies (Daniel et al. 2007; Everson et al. 2008; Ho 2009; Janse Van Rensburg et al. 2009b; Scerbo et al. 2010; Ussher et al. 2009; Janse Van Rensburg et al. 2009a, 2012; Janse Van Rensburg and Taylor 2008; Faulkner et al.2010; Elibero et al. 2011; Harper 2011; Williams et al. 2011; Arbour-Nicitopoulos et al. 2011). Twelve were published in peer-reviewed journals, and three were from two PhD dissertations. While one of the identified studies (Taylor and Katomeri 2007) was published within the time frame, it was included in the previous review, so was not included in this review. A mix of between-subject parallel design and within-subject crossover design studies was identified. Four studies (Harper 2011; Williams et al. 2011; Arbour-Nicitopoulos et al. 2011) examined the acute effects of exercise on cravings and withdrawal symptoms at certain time points in participants undergoing a quit attempt. All other studies asked participants to temporarily abstain from smoking prior to attending the treatment session. See Table 1 for a detailed summary of study participant characteristics, treatment conditions, study design, measures, and outcomes. Assessment of study quality and risk of bias was based on the recommendations by the Cochrane Collaboration (Higgins and Altman 2008). Adequate random sequence generation was employed in 14 studies, with the exception of the study by Scerbo et al. (2010), which assigned participants sequentially to each condition based on recruitment order. It was unclear in all studies whether allocation to treatment groups was concealed up to the point of randomisation; however, it is likely that the method of concealment was not described, rather than there being no method of concealment. As participants were assigned to particular types/intensities of exercise and/or passive control conditions, and may be aware of the expected psychological benefits of exercise, it is not possible to blind participants to treatment. This potential source of bias could increase the magnitude of the treatment effect in favour of exercise. However, the effect of participant expectation has been examined and is discussed with reference to two studies (Daniel et al. 2007; Harper 2011, Study 2) below. All 15 studies were considered free of other potential sources of bias. Four of the 11 studies that examined cravings during temporary abstinence reported conducting a sample size calculation a priori (Everson et al. 2008; Faulkner et al. 2010; Janse Van Rensburg and Taylor 2008; Scerbo et al. 2010), although only two studies (Ho 2009; Arbour- Nicitopoulos et al. 2011) were insufficiently powered to detect a difference between conditions in cravings, suggesting that the target sample sizes in the other studies were based on previous research. Cigarette cravings All of the 15 studies measured cravings to smoke. Twelve of these found that exercise had a positive effect on cigarette cravings (desire to smoke, strength of desire to smoke, or anticipated pleasure from smoking), with reported immediate effects lasting up to 30 min posttreatment (Daniel et al. 2007; Everson et al. 2008; Janse Van Rensburg et al. 2009a, b, 2012; Scerbo et al. 2010; Ussher et al. 2009; Janse Van Rensburg and Taylor 2008; Faulkner et al. 2010; Elibero et al. 2011; Harper 2011). One study showed a positive effect of exercise on cigarette cravings during treatment, but not posttreatment (Faulkner et al. 2010). Three studies failed to show an effect (Ho 2009; Williams et al. 2011; Arbour- Nicitopoulos et al. 2011), although two of these showed trends in favour of exercise (Williams et al. 2011; Arbour- Nicitopoulos et al. 2011). The measure of cigarette cravings varied between studies (see Table 1). Self-report was the predominant method used to measure cravings, but other approaches included picturebased cue reactivity (Elibero et al. 2011), attention bias (increased attention to smoking-related stimuli) (Janse Van Rensburg et al. 2009a), and brain activation using functional MRI (fmri; Janse Van Rensburg et al. 2009b, 2012). Significant reductions in cravings were found following exercise for all of these measures of cravings, relative to control conditions and/or relative to baseline. Ten studies measured either desire to smoke (Tiffany and Drobes 1991) or strength of desire to smoke (SoD; West and Hajek 2004) and compared exercise with a control condition. Calculated Cohen s d effect sizes (Cohen 1992) revealed moderate to large effects of exercise on strength of desire to smoke, with significant effects ranging from 0.40 to 1.48 for four of the studies (Everson et al. 2008; Scerbo et al. 2010; Ussher et al. 2009; Janse Van Rensburg et al. 2012). Moderate to large effects were also observed for desire to smoke, with significant effects ranging from 0.65 (Scerbo et al. 2010) to 1.98 (Janse Van Rensburg et al. 2012). The magnitude of significant posttreatment effect sizes peaked either during or soon after exercise; however, significant effect sizes were found up to 30 min postexercise, including 30 min post 15 min walking [effect size

4 4 Psychopharmacology (2012) 222:1 15 Table 1 Summary of included studies Study Subjects characteristics Abstinence period Exercise characteristics Measures Design Outcome (Title) Daniel et al. (2007) Everson et al. (2008) Janse Van Rensburg and Taylor (2008) Janse Van Rensburg et al. (2009a) Janse Van Rensburg et al. (2009b) Ussher et al. (2009) 22 M and 23 F Mean013 h Read a(a) positive(b) negative(c) Age years ambiguous paragraph about Mean024 the relationship between TWS and exercise prior to 10 min Mean cigs014.4 per day of cycling between 40 and Mean FTND % of HRR Mean baseline SoD04.4 Exercise: VPA <3 times per week for >20 min, or MPA <5 times per week for >30 min 25 M and 20 F Mean017 h 10 min(a) cycled % Mean age021.8 HRR(b) cycled % Mean cigs013.6 per day HRR(c) Sat quietly with no distractions Mean FTND03.4 Mean baseline SoD04.6 Exercise: 3 times per week for >30 min 15 M and 8 F 15 h 15 min(a) Brisk walk (treadmill) Mean age023.1 (mean RPE010.8 (1.67) Mean cigs013.7 per day range 6 20) (b) Passive control Mean FTND M and 5 F 15 h 15 min(a) Cycling (RPE011 13)(b) Mean age Sitting passively with Mean cigs015.6 per day no distractions Mean FTND04.0 Mean baseline desire to smoke M and F 15 h 10 min(a) Cycling (RPE011 13)(b) Age Sitting passively with Mean cigs013.7 per day Mean FTND03.4 Mean baseline desire to smoke04.6 no distractions 31 M and 17 F Mean016.7 h 10 min(a) Body scan(b) Isometric Mean age027.8 exercise (jaw clenching, fist Mean cigs015.5 per day clenching, pushing the palms of the hands together, pushing down Mean FTND05.0 on the thighs, squeezing thighs together, pushing feet into the floor)(c) Listening to an audiorecording of a natural history text Cravings measure Desire to smoke (Tiffany) Expectations Credibility ScaleResponse MPSS Cravings measure SoD (West),Other measures MPSS and SEES Cravings measures Desire to smoke (Tiffany), 10-item QSU.Other measures Stroop Task reaction time Cravings measures Attentional bias measurement, Desire to smoke (Tiffany) Cravings measures fmri brain activation, Desire to smoke (Tiffany) Cravings measure SoD (West)Other measures MAAS, modified MPSS, perceived credibility Between subject (randomly assigned). Assessments Expectations: 1 month pre and 20 min preresponse: 10, 5, and 0 min pre, mid, and IP, 5 and 10 min post. Compared response to expectations Between subject (randomly assigned, stratified for gender). Assessments: pre, mid, and 5 and 30 min post Randomised crossover design. Assessments pre, & IP, 5-, 10- & 15-min post for all measures. Additional assessment at mid for 'desire to smoke' & QSU. Randomised crossover design. Eye tracking protocol pre- and post treatment. Desire to smoke assessed pre-, midand post-treatment, and post-eye tracking protocol Randomised crossover design. 10 min exercise then 15 min fmri scanner. Desire to smoke assessed pre-, mid-, and posttreatment Between subjects (randomly assigned). Assessment pre and IP, 5, 10, and 30 min post-intervention in the laboratory and then undertook intervention again during next 3 h in their normal environment assessed pre and IP, 5 and 30 min post-intervention Expectation manipulation was successful;no significant differences in MPSS or SoD between expectation groups; Significant reduction in symptoms and SoD during and after exercise for all groups (ES ranged from 0.4 to 0.8) (a and b)<(c) desire to smoke during and 5 min post. (ES during0.82 (a), 1.15 (b); ES 5 min post0.79 (a), 1.03 (b)(a) reduced TWS and improved mood 5-min post(b) increased PD and MPSS, and reduced happiness scoresno effects at 30 min post (a) and (b) no significant difference for Stroop(a) reduced desire to smoke, QSU Factor 1 and 2 and reduced cravings up to 15 min post cf (b), ES ranged from.86 to 1.02 (a) reduced desire to smoke mid (ES01.07) and post (ES01.06) treatment cf (b). Dwell time and initial fixation towards smoking images were reduced with (a) cf (b) Scanning found decreased activation in areas of the brain associated with reward, motivation, and visuospatial attention after (a) cf (b). (a) reduced desire to smoke mid- (ES0.88) and posttreatment (ES01.14) cf (b) Desire to smoke and withdrawal symptoms reduced in (a and b) cf (c) for up to 30 min post in lab setting (ES ranged from.61 to.94), and up to 5 min post in normal setting. No significant difference between (a) and (b) Acute exercise effects on smoking withdrawal symptoms and desire to smoke are not related to expectation The effects of moderate and vigorous exercise on desire to smoke, withdrawal symptoms, and mood in abstaining young adult smokers The effects of acute exercise on cognitive functioning and cigarette cravings during temporary abstinence from smoking The effects of acute exercise on attentional bias towards smoking-related stimuli during temporary abstinence from smoking Acute exercise modulates cigarette cravings and brain activation in response to smoking-related images: an fmri study Effect of isometric exercise and body scanning on cigarette cravings and withdrawal symptoms

5 Psychopharmacology (2012) 222: Table 1 (continued) Study Subjects characteristics Abstinence period Exercise characteristics Measures Design Outcome (Title) Ho (2009) 8M 24 h Mean age020.1 Smoke 10 per day Inactive Scerbo et al. (2010) Faulkner et al. (2010) Elibero et al. (2011) Harper (2011) Study 1 Harper (2011) Study 2 Arbour- Nicitopoulos et al. (2011) (a) resistance exercises (6 exercises, 3 10 of each)(b) quiet rest(c) rest (and ad libitum smoking) 10 M and 8 F 3 h 15 min(a) running (80 85 % HRR) Mean age026 (b) walking (45 50 % HRR)(c) Mean cigs013.9 per day sitting on a chair on a treadmill Mean FTND04.4 Mean physical activity0 171 min per week 11 M and 8 F Mean08 h 10 min(a) self-paced brisk walking Mean age025 (mean RPE (1.79), range Mean cigs015.2 per day 6 20)(b) passively sitting on a chair beside a treadmill Mean FTND04.5 Mean physical activity0 171 min per week 76 M and F 1 h 30 min(a) walking on a treadmill Mean age029 (65 75 % HRR)(b) Hatha yoga Mean cigs020 per day 119 F (a) 1 week, (b) 7 Mean age041 weeks, and (c) Mean cigs017 per day 10 weeks post quit date 58 F 1 week post Mean age043 quit date Mean cigs018 per day (asanas included bridge, forward bend, table, cow, cobra)(c) view a video about exercise 20 min on choice of treadmill, rowing machine, stair climber, or stationary bike.(a) moderateintensity exercise (50 60 % HRR)(b) vigorous-intensity exercise (>70 % HRR)(c) vigorous-intensity exercise (>70 % HRR) 20 min on choice of treadmill, rowing machine, stair climber, or stationary bikemoderateintensity exercise (50 60 % HRR)Participants categorised into High EX-EXP and Low EX-EXP, and High EX-CRED and Low EX-CRED 14 M and F Mean07.84 h a) 10 min brisk walking on a Mean age050 treadmillb) 10 min sitting Mean cigs010 Mean FTND04.71 passively on a chair beside a treadmill Cravings measuresod (West)Other measures (1) serum cotinine, (2) plasma ACTH, (3) plasmacortisol, (4) saliva cortisol, (5) HR, (6) SBP and (7) DBP, (8) MPSS, (9) PASAT Cravings measuresdesire to smoke (Tiffany), SoD (West)Other measures HR, salivary cortisol Cravings measure Desire to smoke (Tiffany),Other measures Smoking topography (puff volume, puff duration, puff count, interpuff interval, time to first puff) HR Cravings measure(s) QSU-brief, a picturebased cue reactivity assessmentother measures a brief mood form Cravings measure Shiffman Jarvik withdrawal scale Cravings measure Shiffman Jarvik withdrawal scale Other measures ECQ Cravings measure Desire to smoke (Tiffany)Other measures PAR-Q, MPSS, Feeling scale (Hardy and Rejeski 1989), Felt arousal scale (Svebak and Murgatroyd 1985) Randomised, crossover design. Abstained from 4 p.m., then treatment the next morning, then mental challenge in the afternoon. Assessments pre-a.m., IPa.m., 30 min post-a.m., and pre-p.m., IP-p.m., 30 min post-p.m. Randomised crossover design. SoD assessed pre, mid, IP, 10, 20, and 30 min posttreatment. Salivary cortisol assessed pre, IP, and 30 post each treatment Randomised crossover design. Desire to smoke assessed pre, mid, IP, 10, and 20 min posttreatment. Smoking topography assessed 20 min after each treatment Between subject (randomly assigned). Assessment pre and IP, and 20 min post Sub-study of exercise+nrt smoking cessation intervention trial. Assessment pre and IP Sub-study of exercise+nrt smoking cessation intervention trial. ECQ: pre. Withdrawal scale: pre and IP Randomised crossover design. Pilot study among individuals with SMI participating in a smoking cessation programme. All measures pre, mid, IP, 10, and 20 min post (a) elevated (2, 3, 5, and 7) at IP-a.m. cf (b and c)(2, 3, 4) at pre-p.m. no significant difference between (a) and (b)(a) showed no significant difference for SoD, MPSS of PASAT (a and b) reduced SoD cf (c) (sig ES ranged from 0.4 to 1.48). No significant difference between (a) and (b) for SoD, but effects lasted longer with (a). (a) only, attenuated the decline in cortisol concentrations (a) reduced desire to smoke cf (b) mid condition (ES0.98), but not post-condition. (a) sig longer time to first puff cf (b) Trends in favour of (a) for other topography outcomes (a and b) reduced QSU Factor 1 cf (c). (a) showed decreased craving toward smoking pictures but increased toward neutral pictures. (b) showed decreased craving toward both pictures. (c) showed increased craving toward both types of cues Significant reductions in craving were observed following exercise at time points (a), (b), and (c) and psychological withdrawal and sedation at (a) and (b) High EX-EXP>reduction in craving following exercise cf Low EX-EXP. High EX-CRED> reduction in craving following exercise cf Low EX-CRED No significant main effects were found for time or condition for cravings. Sig time condition interaction for affective valence: (a) felt more pleasant than (b). (a) > activation than (b). No sig effects on MPSS outcomes Effects of resistance exercise of the HPA axis and cardiovascular responses to psychological stress during short-term smoking abstinence in men Effects of exercise on cravings to smoke: the role of exercise intensity and cortisol Cutting down one puff at a time: the acute effects of exercise on smoking behaviour Acute effects of aerobic exercise and hatha yoga on craving to smoke Exercise provides further craving and withdrawal relief for recently quit women who are using nicotine replacement therapy Ex-Exp and Ex-Cred affects craving but not TWS A pilot study examining the acute effects of exercise on cigarette cravings and affect among individuals with serious mental illness

6 6 Psychopharmacology (2012) 222:1 15 Table 1 (continued) Study Subjects characteristics Abstinence period Exercise characteristics Measures Design Outcome (Title) Acute effects of moderateintensity aerobic exercise on affective withdrawal symptoms and cravings among women smokers Nonsignificant trend toward time treatment interaction for cravings: (a)<(b).significant time treatment interaction for energy and tiredness in favour of (a). No effect on tension or calmness Between-subjects design RCT. Assessed cravings and affect pre and IP treatment at each session throughout an 8-week smoking cessation programe Cravings measure Desire to smoke (Shiffman et al. 2003) Other measures: The ADACL (a) 50 min brisk walking on a treadmill, 3 per week(b) 30 min watching videos on health and lifestyle issues, 3 per week 60 F Each contact Mean age042 (3 per week) Smoke 5 per day throughout an 8-week smoking Mean FTND04.8 cessation programme Williams et al. (2011) The acute effects of exercise on cigarette cravings and brain activation in response to smoking-related images: a block-design fmri study Scanning found increased activation in areas of the brain associated with visual processing after (b) cf (a). (a) reduced desire to smoke mid (ES01.96) and post (ES0 1.98) treatment, and SoD mid (ES01.27) and post (ES01.31) treatment, cf (b) Randomised crossover design; 10 min exercise then 15 min fmri scanner. Desire to smoke, SoD assessed pre-, mid-, and posttreatment Cravings measures fmri brain activation, Desire to smoke (Tiffany), SoD (West) 20 M and F 15 h 10 min(a) Cycling (RPE011 13) Age (b) Sitting passively with Mean cigs012.3 per day no distractions Janse Van Rensburg et al. (2012) Mean FTND02.3 Mean baseline desire to smoke05.3 M male, F female, RPE rating of perceived exertion, SoD Strength of Desire to smoke, MPSS Mood and Physical Symptoms Scale, SEES Subjective Exercise Experience Scale (PD psychological distress, PWB positive well-being), IP immediately post, ES effect size, HRR heart rate reserve, TWS tobacco withdrawal symptoms, 10-QSU 10-question Questionnaire on Smoking Urges, fmri functional magnetic resonance imaging, MAAS Mindful Attention Awareness Scale, ACTH adrenocorticotropic hormone, SBP systolic blood pressure, DBP diastolic blood pressure, SV-POMS Shortened Version of the Profile of Mood State Questionnaire, PASAT Paced Auditory Serial Addition Task, HR heart rate, ECQ treatment expectancy and credibility questionnaire, High EX-EXP high exercise expectancy group, Low EX-EXP low exercise expectancy group, High EX-CRED high exercise credibility group, Low EX-CRED low exercise credibility group, PAR-Q Physical Activity Readiness Questionnaire, SMI serious mental illness, ADACL The Activation Deactivation Adjective Checklist (ES)00.4 (desire to smoke) and 0.92 (SoD) (Scerbo et al. 2010)] and 30 min post-isometric exercise [ES00.69 (SoD) (Ussher et al. 2009)]. A number of studies compared different intensities of exercise. Two studies (Everson et al. 2008; Scerbo et al. 2010) compared the effect of moderate intensity versus vigorous intensity exercise on cravings (see Table 1 for details of treatment protocols for the exercise intensity in each study) and showed similar effects on desire to smoke. There were no significant differences between the effects of moderate and vigorous intensity exercise on cravings in either study, although both studies found significant differences between both exercise conditions and the passive controls. Harper (study 1) also assessed the effects of acute bouts of both moderate and vigorous intensity exercise on cravings, but did not conduct a direct comparison between the two intensities. In a sample of female smokers participating in a 14-week exercise-aided nicotine replacement therapy (NRT) programme for smoking cessation, cravings (measured with the Shiffman Jarvik withdrawal scale; Shiffman and Jarvik 1976) were assessed before and after an exercise treatment session (see Table 1 for details of each exercise session) at week 5 (1-week post quit date and on 21-mg NRT patch), week 11 (7 weeks post quit date and on 14-mg NRT patch), and week 13 (9 weeks post quit date and on 7-mg NRT patch). Significant reductions in cravings from pre- to post-exercise were found at all three time points (i.e. week 5, ǹ ; week 11 ǹ ; and week 13, ǹ ). In an actual quit attempt involving NRT, craving relief following an acute bout of exercise can be achieved. Two studies examined light-intensity yoga (Elibero et al. 2011) and isometric exercise (Ussher et al. 2009) and found statistically significant reductions on cravings to smoke. Elibero et al. (2011) compared three conditions, 30 min of yoga, 30 min of moderate intensity [70 % of maximum heart rate (HR)] walking on a treadmill, and a passive control group, and showed that both exercise groups significantly decreased urges to smoke (measured with the QSU-brief, Cox et al. 2001) compared to the control group. Relative to the control group, who increased cravings when presented with both smoking and neutral images, cravings were reduced in both exercise groups in response to smokingrelated images, and in the yoga group in response to neutral images. Ussher et al. (2009) showed a significant reduction in cravings following a 10-min isometric exercise. One study examined the effect of resistance-based exercise (back squat, bench press, bent-over row, arm curl, Romanian deadlift, and sit ups) and reported no significant difference in cravings to smoke, compared to a passive control group (Ho 2009); although with small to moderate effect sizes ( ) posttreatment, this was perhaps due to being underpowered. This is the only study to date to

7 Psychopharmacology (2012) 222: have examined resistance exercise in this context, and further research is required to examine the effects of resistance exercise. Two studies sought to determine whether reductions in cravings post-exercise were related to participant s outcome expectancy by either manipulating (Daniel et al. 2007) or predetermining (Harper 2011, Study 2) participants expectancy of an effect of exercise on cravings (see Table 1 for study designs). Both studies demonstrated significant differences between pre- and post-exercise craving scores. However, whereas Daniel et al. found no differences between groups, Harper showed significantly greater reductions in cravings in those categorised as high in exercise expectancy compared to those considered low in exercise expectancy. Significantly greater reductions were also observed for those classified as high versus low credibility. Tobacco withdrawal symptoms Withdrawal symptoms known to be reduced by exercise include irritability, depression, tension, restlessness, difficulty concentrating, and stress. Seven studies measured TWS using either the Mood and Physical Symptoms Scale (MPSS; West and Hajek 2004) (Daniel et al. 2007; Everson et al. 2008; Ho 2009; Ussher et al. 2009; Arbour- Nicitopoulos et al. 2011) or the Shiffman Jarvik withdrawal scale (Shiffman and Jarvik 1976); (Harper 2011). Three (Daniel et al. 2007; Everson et al. 2008; Ussheretal. 2009) out of the five studies (Daniel et al. 2007; Everson et al. 2008; Ho 2009; Ussher et al. 2009; Arbour- Nicitopoulos et al. 2011) that measured TWS with the MPSS (West and Hajek 2004) found a positive effect of exercise on at least one withdrawal symptom. Of these three studies, two compared exercise with a passive control condition, and both found a significant difference between conditions in favour of exercise: isometric exercise (Ussher et al. 2009) and moderate-intensity cycling (Everson et al. 2008). In contrast, increased composite MPSS scores were found during bouts of vigorous intensity exercise, suggesting an adverse effect on symptoms; however, these adverse effects were not evident after exercise (Everson et al. 2008). Harper (Study 1, 2011) showed that 20 min bouts of moderate intensity exercise at week 5 of an exercise-aided NRT smoking cessation programme and vigorous intensity exercise at week 11 of the programme significantly reduced psychological and sedation withdrawal symptoms from pre- to post-exercise. Differences between pre- and postexercise scores for these withdrawal symptoms at week 13 also approached significance (p00.083). Two studies examined whether reductions in TWS postexercise were related to participant s outcome expectancy (Daniel et al. 2007; Harper 2011, Study 2). Both studies demonstrated significant differences between pre- and postexercise TWS scores; however, there were no differences between groups for outcome expectancy. It is worth noting, however, that in the study by Harper, those in the high exercise expectancy and high exercise credibility groups experienced greater reductions in withdrawal symptoms following exercise than those in the low expectancy and credibility groups, although these were not statistically significant. Janse Van Rensburg and Taylor (2008) assessed the effects of exercise (15 min self-paced walk) on impaired concentration by measuring cognitive function using the Stroop Task (Stroop 1935). While no significant group by time interaction or main effect of condition for Stroop Task reaction time was found, between condition t tests at each time point revealed significantly lower mean reaction time scores at 10 and 15 min post-exercise compared to those in the passive control group. This study represented a methodological improvement compared to previous self-reported measures of concentration. Affect Four (Elibero et al. 2011; Everson et al. 2008; Williams et al. 2011; Arbour-Nicitopoulos et al. 2011) of the five studies (Elibero et al. 2011; Everson et al. 2008;Ho2009; Williams et al. 2011; Arbour-Nicitopoulos et al. 2011) that examined the effect of exercise on affect found a positive effect of exercise on at least one measure of affect. Elibero et al. (2011) showed that positive affect increased and negative affect decreased immediately after both 30 min of yoga and 30 min of moderate-intensity walking, and Everson et al. (2008) found that positive well-being was increased and psychological distress was decreased 5 min after 10 min of moderate-intensity cycling. These results are consistent with findings from the previous review (Taylor et al. 2007). However, Everson et al. (2008) also found that positive well-being decreased and psychological distress increased during the vigorous exercise condition, mimicking the trend for composite MPSS score, described above. Using the Activation Deactivation Adjective Checklist (ADACL; Thayer, 1989), Williams et al. (2011) found significant time treatment interaction effects for energy and tiredness, such that exercise (50 min brisk walking) participants reported higher mean energy and lower mean tiredness compared to control participants posttreatment, compared with no differences pretreatment. Conversely, they found no such effects for tension or calmness. Finally, Arbour- Nicitopoulos et al. (2011) found a significant time condition interaction effect for affective valence (pleasure displeasure) using the one-item Feeling Scale (Hardy and Rejeski 1989), but no interaction effect for activation (measured using the one-item Felt Arousal Scale(Svebak and Murgatroyd 1985)).

8 8 Psychopharmacology (2012) 222:1 15 Smoking topography or behaviour Only one study examined the effect of exercise on smoking topography (Faulkner et al. 2010). Using the Clinical Research Support System Pocket, a computer-based handheld unit which automatically measures smoking behaviour parameters (Plowshare Technologies, Borgwalt KC, Inc., VA), Faulkner et al. (2010) measured a variety of smoking behaviour outcomes following either a 10-min brisk walking or passive sitting. The time to first puff was significantly longer following the brisk walking condition than following passive sitting, and there were trends towards significant effects in favour of walking for all of the other smoking topography outcomes. These preliminary findings suggest that participating in regular light-to-moderate exercise may help lengthen the time between each cigarette, and thus decrease the number of cigarettes smoked per day, which may assist in smoking cessation. However, more research is required with larger sample sizes, greater periods of abstinence, and less active smokers, as the authors suggest (Faulkner et al. 2010). Psychobiological biomarkers Based on previous recommendations (Taylor et al. 2007), two studies have examined the effect of exercise on psychobiological variables and their mechanistic role on the exercise craving relationship. Scerbo et al. (2010), reported that the normal cortisol decrease during abstinence from cigarettes was attenuated by a 15-min bout of vigorous intensity running, for up to 30 min post-exercise, and Ho (2009) found that plasma adrenocorticotropic hormone (ACTH), serum cortisol, heart rate, and systolic blood pressure were all elevated post-resistance exercise when compared to both a passive condition during abstinence and in an ad libitum smoking condition. However, neither study found a relationship between these changes in biomarkers and changes in tobacco cravings or withdrawal. Meta-analyses All studies published to date, including those reviewed by Taylor et al. (2007), were reviewed for inclusion in a metaanalysis of the effect of exercise on cigarette cravings during temporary abstinence. Due to the heterogeneity of craving outcomes across studies, it was not possible to combine all trial data in one meta-analysis to provide overall summary statistics for cigarette cravings. However, there were sufficient similar trials to conduct meta-analyses for two outcomes, desire to smoke (Tiffany and Drobes 1991) and strength of desire to smoke (West and Hajek 2004). Authors of studies which measured these outcomes but did not report means and standard deviations were contacted and asked to provide this information. Following exhaustive attempts to contact the authors, we were unable to obtain these essential descriptive data for two studies (Daniel et al. 2004, 2006). No studies reported change in mean (and standard deviation) for desire to smoke or strength of desire to smoke from baseline to posttreatment. These data were therefore imputed according to the methods outlined by the Cochrane Heart Group for handling continuous variables (Cochrane Heart Group). The change in mean from baseline to posttreatment for each condition was calculated by subtracting the mean at follow-up (zero or 5 min posttreatment, depending on when the outcome was measured in each study) from the mean at baseline. The standard deviation of the difference (SD difference) was calculated with the following formula: SD difference ¼ standard error ðseþ difference p n Where, SE difference0 p SD 2 1 n 1 þ SD 2 2 n = 2 ð1 rþ Where, SD 1 n 1 SD 2 n 2 r The standard deviation at baseline The number at baseline The standard deviation at follow-up The number at follow-up The correlation coefficient For the correlation coefficient, in the absence of large epidemiological studies in this area, a value of 0.5 was assumed (Follmann et al. 1992). These data are presented in Table 2. To incorporate within-subject design studies into a metaanalysis with parallel group trials, the mean difference of the treatments and the corresponding standard error are required (Curtin et al. 2002; Elbourneetal.2002; Higgins et al. 2008). Using the imputations above, the mean difference in change from baseline to follow-up between groups was then calculated using the following formula: MðdiffÞ ¼ M E M C Where: M(diff) M E M C The mean difference between groups Exercise group mean (baseline) exercise group mean (follow-up) Control group mean (baseline) control group mean (follow-up) For within-subject design studies, the treatment effect was defined as the mean within-subject difference between conditions in change from baseline to follow-up, assuming no carry-over effect. As the studies did not report the mean difference in change from baseline between conditions, the standard error for the within-subject differences could not be extracted. Therefore, the standard error for the within-

9 Psychopharmacology (2012) 222: Table 2 Mean and standard deviation (SD) at baseline and follow-up, and the calculated change in mean and SD from baseline to follow-up, for exercise and control conditions Exercise Control Baseline Follow-up Difference n Baseline Follow-up Difference n Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD Desire to smoke Taylor et al Janse Van Rensburg et al. 2009a Janse Van Rensburg et al. 2009b Scerbo et al Faulkner et al Janse Van Rensburg et al Arbour-Nicitopoulos et al Janse Van Rensburg and Taylor Taylor and Katomeri Ussher et al Janse Van Rensburg and Taylor Strength of desire to smoke Taylor et al Scerbo et al Janse Van Rensburg et al Ussher et al Everson et al Taylor and Katomeri Ussher Everson et al Ussher et al subject trials was imputed according to the guidelines in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins et al. 2008), which state that when the standard deviation of the difference between groups is not reported, the standard error for the withinperson differences [SE (MD)] can be imputed using the following formula: p SEðMDÞ ¼SD diff= n Where, SD diff ¼ p SD 2 E þ SD 2 C ð2 r SD E SD C Þ Where, SD diff SD E SD C r The standard deviation of the within-person differences between conditions The standard deviation of the difference between baseline and follow-up in the experimental (exercise) group The standard deviation of the differences between baseline and follow-up in the control group The correlation coefficient The correlation between treatment outcomes was approximated using a conservative estimate of the correlation coefficient (r00.62) based on the difference between treatment group means, p values, and t statistics, at 0 or 5 min posttreatment, from trials included in the meta-analyses that reported this information. Several different correlation coefficients were imputed as part of sensitivity analyses (Higgins et al. 2008). The corresponding standard error was then calculated, as described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins et al. 2008). The generic inverse variance method, with a random effects model applied, was used to combine both betweenand within-subject design studies in the meta-analyses (Higgins et al. 2008). Data synthesis and statistical analyses were conducted using the Cochrane Collaboration Review Manager (RevMan, version 5.1; The Cochrane Collaboration, Copenhagen, Denmark). When multiple comparisons were conducted within one study (i.e. Everson et al. 2008; Scerbo et al. 2010), the moderate-intensity condition was included in the meta-analyses as moderate-intensity exercise

10 10 Psychopharmacology (2012) 222:1 15 was the predominant exercise intensity examined by the other included studies (Fig. 1). A total of ten trials (Janse Van Rensburg et al. 2009a, b, 2012; Janse Van Rensburg and Taylor 2008; Scerbo et al. 2010; Taylor and Katomeri 2007; Taylor et al. 2006; Ussher et al. 2001; Faulkner et al. 2010; Arbour-Nicitopoulos et al. 2011) compared the difference between exercise and a passive control condition for desire to smoke. The weighted mean difference in self-reported desire to smoke between exercise and control conditions was 1.90 points on a seven-point scale in favour of exercise (95 % confidence interval (CI), 3.06 to 0.75; p00.001). A random effects model was used as significant heterogeneity was indicated (I %, p0< ). Sensitivity analyses revealed that, based on effect size, five studies were responsible for the heterogeneity (Faulkner et al. 2010; Tayloretal.2006; Ussher et al. 2001; Janse Van Rensburg and Taylor 2008; Arbour-Nicitopoulos et al. 2011). When excluding all five studies from the analysis, the I 2 statistic was reduced from 98 to 5 % (p00.38). The pooled reduction in desire to smoke changed only slightly with the five studies excluded ( 2.13, 95 % CI, 2.41 to 1.84; p< ; Fig. 2). A total of nine trials (n0295) (Everson et al. 2006, 2008; Scerbo et al. 2010; Taylor and Katomeri 2007; Taylor et al. 2005; Ussher et al. 2006; Ussher et al. 2001, 2009; Janse Van Rensburg et al. 2012) compared the difference between exercise and a passive control condition for strength of desire to smoke. The weighted mean difference in selfreported strength of desire to smoke between exercise and control conditions was 2.41 points on a seven-point scale in favour of exercise (95 % CI, 3.45 to 1.37; p< ). A random effects model was used as significant heterogeneity was indicated (I %, p< ). Sensitivity analyses revealed that, based on effect size, two studies were responsible for the heterogeneity (Ussher et al. 2001; Taylor et al. 2005), both of which reported large reductions in strength of desire to smoke. When excluding these two studies from the analysis, the I 2 statistic was reduced from 94 to 0 % (p00.63). The pooled change in strength of desire to smoke with the two studies excluded was reduced to 1.74 (95 % CI, 2.05 to 1.44; p< ). One study (Ho 2009), which measured strength of desire to smoke, was not included as a six-point scale was used. Discussion The overall goal of this paper was to update the evidence regarding the effect of exercise on the acute effects of exercise on TWS, cravings, and affect during temporary smoking abstinence, and to examine potential mediators of the exercise TWS relationship. The evidence to date suggests that brief bouts of exercise decrease cravings for cigarettes at rates comparable to nicotine replacement therapy (Taylor et al. 2007). The duration of the effect of exercise on cravings ranged from 5 to 30 min postexercise. This was the case for both moderate- and vigorous-intensity exercise, and very light-intensity exercise. However, the duration of the effect post-exercise needs further investigation, particularly in real-life situations. To date, only one study (Ussher et al. 2009) has examined bouts of exercise performed in the participants normal environment, rather than a controlled laboratory setting. In this study (Ussher et al. 2009), duration of effect post-isometric exercise decreased from 30 min when performed in the laboratory to 5 min in the participants normal environment. Future studies should be conducted in more ecologically valid settings. The magnitude of the effects of exercise on cravings is encouraging. Meta-analyses suggest that exercise is more effective than passive control conditions at reducing both desire to smoke and strength of desire to smoke, with weighted mean differences between exercise and control groups on a seven-point scale of 1.90 and 2.41, respectively. Only two of nine studies that measured strength of desire to smoke on a seven-point scale failed to show a statistically significant difference between exercise and passive control conditions at zero or 5 min post-condition Fig. 1 Meta-analysis of trials comparing exercise and control conditions with change in desire to smoke from baseline to between 0 and 5 min posttreatment as the outcome measured