یک مقایسه استراتژی تنظیم احساسات در پاسخ به شناخت ولع مصرف: اثر بر رفتار سیگار، ولع مصرف و در افراد سیگاری وابسته

Abstract Aim The effects of three emotion regulation strategies that targeted smoking-related thoughts were compared on outcomes relevant to smoking cessation. Method Daily smokers applied defusion (n = 25), reappraisal (n = 25) or suppression (n = 23) to thoughts associated with smoking during a cue-induced craving procedure. Smoking behaviour, approach/avoidance behavioural bias, and subjective measures of experiential avoidance, craving, and affect were assessed during the experimental session, with additional behavioural and subjective outcomes assessed at 24 h and seven day follow-up. The influence of baseline group differences in smoking level and nicotine dependence were explored statistically. Results Defusion and reappraisal were associated with greater restraint in smoking behaviour in the immediate post-session period as well as reduction in smoking at seven day follow-up compared to suppression. Relative to suppression, reduced subjective craving was seen in the reappraisal group, and reduced experiential avoidance in the defusion group. Differences in approach/avoidance responses to smoking and neutral cues were observed only between the suppression and reappraisal groups. Although suppression was rated as lower in both credibility and strategy-expectancy compared to defusion and reappraisal, neither credibility nor expectancy mediated the effect of any strategy on changes in levels of smoking. Conclusion Defusion and reappraisal produced similar benefits in smoking-related behavioural outcomes but, relative to suppression, were associated with distinctive outcomes on experiential avoidance and craving. The effects appear to be independent of perceived expectancy and credibility of the different strategies. Overall, the results suggest a role for reappraisal and defusion strategies in the development of psychological treatments for addiction-related disorders.

1. Introduction Despite the success of health campaigns, tobacco addiction remains a significant and costly public health problem. The powerful motivational-affective experience of craving, which reflects the co-ordinated activation of a motivational system that controls attention and behaviour (Sayette, Martin, Hull, Wertz, & Perrott, 2003), is central to the intractability of cigarette addiction. Furthermore, craving is accompanied by self-referential verbal thoughts supported by propositional networks (Tiffany, 1990), and behaviours that are biased towards approaching smoking-related cues in preference to other stimuli (Mogg et al., 2003 and Stacy and Wiers, 2010). Such cognitive and motivational biases themselves increase responsivity to smoking cues (e.g. craving) in a feed-forward mechanism which increases drug-taking behaviour (Franken, 2003 and Robinson and Berridge, 2000). However, emerging evidence suggests that the use of certain ‘emotion regulation’ strategies can subvert this vicious cycle and reduce the intensity of craving and/or smoking behaviour. Emotion regulation refers to the use of cognitive, behavioural or emotional strategies (e.g. avoidance, reappraisal, rumination, escape, suppression, distraction and problem-focused coping; Gross, 1998) to alter the form, frequency, intensity or situational occurrence of emotional experiences. Among these strategies, reappraisal has consistently been shown to reduce the emotional impact of aversive experiences ( Gross, 1998, Gross, 2002 and Jackson et al., 2000). Reappraisal presumably involves modification of the propositional networks that underlie verbal statements that relate, for example, to the desirability of drug-use, self-efficacy in managing intense craving and positive expectancies regarding drug effects. The deliberative use of reappraisal is a central feature of Cognitive Behavioural Therapy (CBT) for addictive disorders ( Marlatt & Gordon, 1985). Alternatively, while commonly used as a spontaneous coping strategy, suppression of aversive emotional experiences can paradoxically enhance unpleasant emotional reactions (e.g. Jackson et al., 2000). In the case of addiction, suppression of drug-related thoughts and feelings might therefore be expected to increase responsivity to drug cues. In contrast to reappraisal, as used in CBT, recently developed psychological therapies such as Acceptance and Commitment Therapy (ACT; Hayes, Strosahl, & Wilson, 1999) emphasise an individual's relationship towards their thoughts, rather than thought content ( Hayes, 2004 and Segal et al., 2004). This approach highlights the role of two broad trans-diagnostic factors in psychiatric disorders: experiential avoidance and psychological inflexibility ( Hayes et al., 1999). Experiential avoidance refers to the habitual tendency to strategically or unconsciously avoid, suppress or otherwise minimize aversive internal sensations (thoughts, emotions and somatic experiences). Psychological inflexibility is the tendency to engage in repetitive and maladaptive cognitive and behavioural strategies despite changing circumstances, often in the service of experiential avoidance. In smokers, higher levels of experiential avoidance in response to stress are associated with higher levels of smoking behaviour ( Pirkle & Richter, 2006) and greater likelihood of relapse ( Gifford et al., 2004). ACT aims to decrease experiential avoidance and increase psychological flexibility through the use of strategies that include mindfulness, acceptance and ‘defusion’. As with reappraisal in CBT, the primary target of these ACT-based therapeutic (emotion regulation) strategies is propositional thinking (i.e. self-defeating verbal statements). While a growing body of evidence suggests that ACT is a promising therapeutic approach for a variety of disorders – include substance use disorders – the active components of this complex treatment remain unclear. Experimental studies in the tradition of ‘component research’ can help parse the effects/effectiveness of individual component strategies within complex psychological interventions (Levin, Hildebrandt, Lillis, & Hayes, 2012). The role of defusion for example, has been investigated in isolation from other aspects of ACT using experimental instructions that aim to overcome the literal believability of thoughts by generating a sense of ‘psychological distance’ from them (Twohig, Masuda, Varra, & Hayes, 2005). These studies suggest that, like reappraisal, defusion techniques can reliably be taught to participants in experimental settings (Deacon et al., 2011, Hooper and McHugh, 2013 and Levin et al., 2012). Most studies on defusion have investigated its effects on self-critical thoughts (Healy et al., 2008, Masuda et al., 2010a, Masuda et al., 2004 and Masuda et al., 2009). Other studies with more direct relevance to substance use disorders have examined the effects of defusion on food cravings. These show, for example, that defusion results in greater reductions in chocolate consumption compared to suppression (Hooper, Sandoz, Ashton, Clarke, & McHugh, 2012), reappraisal (Moffitt, Brinkworth, Noakes, & Mohr, 2012), acceptance and relaxation (Jenkins & Tapper, 2013). Ideally, studies comparing CBT- and ACT-based emotion regulation strategies should include measures that tap the emotional, cognitive and behavioural processes that are predicted to change in response to the respective strategies used in these therapies. However, recent experimental studies of experiential acceptance have tended to use outcome measures which tap acute changes in the intensity of negative emotion or craving, consistent with the aims of CBT rather than ACT (Hofmann et al., 2009, Szasz et al., 2011, Szasz et al., 2012 and Wolgast et al., 2011). On the other hand, studies comparing defusion with other emotion regulation strategies have tended to include outcome measures guided by the ‘psychological flexibility’ model that underpins ACT (e.g. believability of thoughts). The latter studies provide preliminary support for the idea that defusion is an effective strategy for regulating the effects of self-defeating thoughts and therefore has clinical utility in its own right. However, important questions remain, not least about the effectiveness of defusion techniques beyond addressing negative self-referential thoughts (self-criticism) and food craving in non-clinical populations. The effects of defusion on drug-use-related thoughts as well as somatovisceral craving sensations, remain unclear. Moreover, studies of emotion regulation rarely assess the credibility and expectancy effects of tested strategies. Of the studies referred to above, only one examined credibility of the interventions tested (Masuda et al., 2004). This is a fundamental limitation of extant research as it is not known whether comparisons are being made between equally credible strategies, and if not, whether treatment-related appraisals (credibility and treatment expectancies) have an effect on outcomes. The current study seeks to contribute to our understanding of adaptive emotion regulation strategies and their utility in substance use disorders by examining the comparative effectiveness of brief standardised defusion and reappraisal instructions on smoking-relevant and theory-consistent outcomes, using suppression instructions as the comparator. In particular we examined the effects of these instructions on smoking behaviour, implicit behavioural approach/avoidance tendencies, and subjective measures of experiential avoidance, cue-induced craving, and negative affect. In line with previous research, we predicted that thought suppression would adversely affect smoking-related outcomes through its well-established rebound effects on unwanted thoughts and feelings (Gross & Thompson, 2007). In addition, theoretical and empirical studies suggest beneficial but distinct effects of reappraisal and defusion in some domains (Segal et al., 2004). Specifically, emotion regulation and cognitive behavioural theories would suggest that reappraisal will produce relatively immediate reductions in subjective craving and negative affect (Gross, 2002 and Perkins et al., 2007). Alternatively, since the techniques originating from the psychological flexibility model do not focus on producing immediate reduction in the intensity of specific subjective experiences, craving and negative affect are not predicted to change acutely in response to defusion instructions. Rather, defusion is predicted to be associated with changes in participants’ relationship to their craving-related thoughts as well as overt smoking behaviour. In addition to testing these predictions, we also examine the effects of reappraisal, suppression and defusion on a smoking approach-avoidance task which assesses a more implicit, non-verbal level of processing of smoking stimuli.

. Results 3.1. Demographic and smoking-related characteristics Table 2 provides a summary of key demographic characteristics across the three groups. There were no between-group differences in years spent in education or smoking preferences. Table 2. Participant demographics by emotion regulation group. Values are M (SD) for age and education, and N (%) for all of other variables. Defusion Reappraisal Suppression Age 25.40 (7.49) 24.40 (6.56) 25.20 (7.93) Gender Female 13 (48) 13 (52) 12 (48) Male 12 (52) 12 (48) 13 (52) Ethnicity White 13 (52) 13 (52) 15 (60) Mixed/multiple ethnic groups 1 (4) 2 (8) 1 (4) Asian/Asian British 5 (20) 4 (16) 2 (8) Black/African/Caribbean/Black British 4 (16) 1 (4) 1 (4) Other ethnic group 2 (8) 4 (16) 4 (16) Missing 0 (0) 1 (4) 2 (8) Education (years) 14.6 (2.0) 15.28 (1.72) 14.89 (1.84) Table options Due to random chance, there were baseline differences between the groups in level of nicotine dependence (F [2, 70] = 4.493, p = 0.015, η2 = 0.113) and number of cigarettes smoked in the past seven days (TLFB score; F [2, 70] = 4.217, p = 0.019, η2 = 0.108), which were higher in the reappraisal group than the defusion group (t(48) = 2.852, p = 0.017, d = 0.403 and t(48) = 2.772, p = 0.021, d = 0.392) respectively. No other between-group baseline differences were found in smoking-related characteristics (F values < 2.92, p values > 0.05; Table 3). Table 3. Smoking characteristics separated by emotion regulation group. Values are Mean (SD). Defusion (N = 25) Reappraisal (N = 25) Suppression (N = 23) Motivation to quit smoking 2.08 (0.91) 1.96 (1.14) 1.74 (0.76) Estimated number of cigarettes per day 11.53 (3.96) 14.64 (4.88) 12.81 (4.81) FTND Score 4.58 (1.05)* 5.56 (1.39)* 5.28 (1.28) TLFB baseline 11.04 (4.15)* 14.77 (5.26)* 11.87 (4.80) TLFB seven day follow-up 7.72 (4.40) 9.53 (5.44) 10.65 (4.87) Hours since last cigarette 6.48 (4.53) 5.02 (3.68) 5.32 (3.85) * = group differences are significant at p < 0.05. Table options 3.2. Effects of emotion regulation strategy on smoking behaviour A main effect of Time (pre, post) on number of cigarettes smoked as assessed by the TLFB indicated an overall reduction in the mean number of cigarettes smoked per day at seven day follow-up compared to baseline (F[1,70] = 42.224, p < 0.001, ηp2 = 0.376; see Table 3). This was qualified by a Time × Strategy interaction (F[2,70] = 5.286, p = 0.07, ηp2 = 0.131). Comparisons across time-points within groups found that participants in the defusion (t[24] = 4.169, p < 0.001, d = 0.834) and reappraisal groups (t[24] = 4.616, p < 0.001, d = 1.246) reported a reduction in TLFB-smoking while those in the suppression condition did not (t[22] = 1.644, p = 0.105, d = 0.329). Including baseline FTND score as a covariate in the model produced a significant covariate effect (F[1,69] = 24.435, p < 0.001, ηp2 = 0.262), attenuated the Time main effect (F[1,69] = 2.065, p = 0.115, ηp2 = 0.029), but left the Time × Strategy interaction intact (F[2,69] = 5.795, p = 0.05, ηp2 = 0.144). The baseline chance group differences in level of dependence therefore do not appreciably affect the efficacy of the defusion and reappraisal interventions. There was an effect of Strategy on latency to smoke (K[2, N = 73] = 11.108, p = 0.004). Those in the suppression group reported smoking within a shorter period (in minutes) after leaving the experimental session (M = 18.652, MED = 7, SD = 22.699) than those in the defusion (M = 128.6, MED = 63, SD = 213.876; U [47] = 2.695, p = 0.021) and reappraisal (M = 73.08, MED = 62, SD = 58.58) groups [U[47] = 18.849, p = 0.006]. Baseline smoking level and nicotine dependence did not correlate with latency to smoke [τ (73) = −0.057, p = 0.48], so were not modelled in the analysis. 3.3. Cue-induced craving and negative affect There was a significant main effect of Time on cue induced craving (F[3, 210] = 50.612, p < 0.001, ηp2 = 0.420), with the highest QSU-brief scores at baseline. There was a main effect of Strategy (F [2, 70] = 3.406, p = 0.039, ηp2 = 0.089), driven by lower overall craving in the reappraisal than suppression group [t(46) = 2.588, p = 0.035, d = 0.763] with a trend for a Time × Strategy interaction (F[6, 210] = 2.093, p = 0.068, ηp2 = 0.056). Pair-wise comparisons between groups at each time point indicated that after craving-induction (t[46] = 3.181, p = 0.007, d = 0.918) and one day later (t[46] = 2.741, p = 0.023, d = 0.808) only participants in the reappraisal condition reported lower cravings than those in the suppression condition. There were no large correlations between craving and baseline nicotine dependence or smoking levels any time point (all rs (73) < 0.24), so these were not modelled in the analysis. There was no main effect of Time (F [1, 70] = 0.008, p = 0.931, ηp2 < 0.001) or Strategy (F[2, 70] = 1.043, p = 0.358, ηp2 = 0.029) and no interaction (F[2, 70] = 1.143, p = 0.325, ηp2 = 0.032) on negative affect. 3.4. Smoking-specific experiential avoidance There was no main effect of Time (F[1, 72] = 2.139, p = 0.148, ηp2 = 0.03) or Strategy (F[2, 70] = 2.22, p = 0.12, ηp2 = 0.06) on smoking specific-experiential avoidance as assessed by the AIS. There was, however, a Time × Strategy interaction (F[2, 70] = 3.561, p = 0.034, ηp2 = 0.09). Participants in the defusion group reported a significant reduction in smoking specific experiential avoidance (t[24] = 2.24, p = 0.03, d = 0.51) whereas those in the reappraisal (t[24] = 1.69, p = 0.10, d = 0.25) and suppression (t[22] = 0.88, p = 0.39, d = 0.41) conditions did not. However, as can be seen in Fig. 2, there were similar reductions in experiential avoidance between the Defusion and Reappraisal groups. Indeed, follow-up 2 (Time) × 2 (Strategy; Defusion/Reappraisal) ANOVA showed that these slopes did not differ significantly (Time × Strategy interaction (F[1,48] = 0.122, p = 0.729, ηp2 = 0.003)). Smoking-specific experiential avoidance by strategy at pre and post cue-induced ... Fig. 2. Smoking-specific experiential avoidance by strategy at pre and post cue-induced craving. Symbols indicate mean values (SEM). Figure options 3.5. Approach/avoidance behaviour A main effect of Behaviour (approaching, avoiding) on response time (F[1, 70] = 13.928, p < 0.001, ηp2 = 0.166) indicated faster approach trials across strategies and stimuli ( Fig. 3). There was also a main effect of Stimulus (smoking, neutral) on response time (F[1, 70] = 82.928, p < 0.001, ηp2 = 0.542), such that participants were quicker to respond to smoking-related than neutral images. There was no main effect of Strategy, but a significant Behaviour × Stimulus interaction (F[1, 70] = 19.532, p < 0.001, ηp2 = 0.218) such that participants were quicker to approach than avoid smoking-related images, (t[72] = 6.467, p < 0.001, d = 1.52] with no difference for neutral images. A Behaviour × Stimulus × Strategy interaction was found (F [2, 70] = 3.63, p = 0.032, ηp2 = 0.094). Between-groups pairwise comparisons within levels of the behaviour × stimulus interaction showed that this was driven by longer latency to avoid smoking images (t[46] = 2.464, p = 0.048, d = 0.727) and approach neutral images (t[46] = 2.78, p = 0.021, d = 0.819) in the Suppression group than the Reappraisal Group. Approach/avoidance behaviour by strategy. Hatched and solid bars indicate means; ... Fig. 3. Approach/avoidance behaviour by strategy. Hatched and solid bars indicate means; error bars indicate SEM. Figure options Approach-avoidance task performance was not correlated with credibility, expectancy, craving change or change in TLFB smoking (all p values >0.1). 3.6. Strategy credibility and expectancy Groups differed on ratings of perceived credibility (F[2,70] = 9.19, p < 0.001, ηp2 = 0.21) and expectancy of effect of the strategy (F[2,70] = 3.61, p = 0.03, ηp2 = 0.09). Strategy credibility was rated lower amongst participants in the suppression group (M = 14.16, SD = 4.84) than the defusion (M = 18.52, SD = 4.00; t[46] = 3.66, p = 0.001, d = 0.98) and reappraisal (M = 18.72, SD = 4.11; t[46] = 3.82, p = 0.001, d = 1.02) groups. Expectancy was also lower in the suppression group (M = 4.36, SD = 1.66) than the reappraisal (M = 5.70, SD = 1.62; t[46] = 2.66, p = 0.03, d = 0.82) group. 3.7. Relationships between outcomes, baseline scores and credibility/expectancy ratings For the majority of outcomes variables, baseline TLFB did not correlate with the dependent variable and so was not appropriate to include in the model. The baseline differences in smoking may be more problematic if they represent heterogeneity in regression slopes between baseline and seven day TLFB scores. This was assessed by correlating these scores overall and across groups. Overall there was a correlation between seven day TLFB smoking and baseline TLFB (r(73) = 0.587, p < 0.001) and FTND (r (73) = 0.341, p = 0.003) scores. Group-wise correlations explored the possibility that group differences were driven by the baseline differences in TLFB smoking and FTND. The correlation coefficient for the association between baseline FTND and seven day TLFB smoking in the reappraisal group was not significantly different to the suppression (z = 0.63, p = 0.529) or defusion groups (z = 1.38, p = 0.168) and the suppression and defusion groups did not differ (z = 0.71, p = 0.477). Similarly the association between baseline TLFB and change in smoking was not significantly different between groups. These findings do not support the idea of heterogeneous regression slopes between baseline and outcome smoking levels among groups and, with the weight of evidence of all analyses, suggests that baseline differences in smoking are unlikely to account for the observed strategy effects. The association between credibility, expectancy and changes in smoking levels (as assessed by the TLFB) baseline to follow-up was also explored. Expectancy (r (73) = 0.261, p = 0.0261) but not credibility (r (73) = −0.198, p = 0.094) was associated with change in TLFB smoking, with higher expectancy associated with greater reductions in smoking. To assess any mediating impact of credibility and expectancy on the relationship between strategy and TLFB change, groups were compared in a pairwise manor (suppression vs. reappraisal; suppression vs. defusion; defusion vs. reappraisal) via a simple mediation model (model 4) via the regression approach implemented by PROCESS for SPSS (Hayes, 2008). The path estimates are based on bias corrected and accelerated bootstrapping using 10,000 bootstrap samples. These models, along with their relevant statistics are presented in Fig. 4A and B. As expected from the observed effect of strategy in the mixed models ANOVAs, strategy predicted variance in TLFB change, but no mediating impact of credibility or expectancy was found for any pairwise comparison, suggesting intervention effects are independent of credibility appraisal and expectancy. A. Statistical mediation models for the relationship between strategy and change ... Fig. 4. A. Statistical mediation models for the relationship between strategy and change in smoking behaviour with expectancy ratings as mediator. B. Statistical mediation models for the relationship between strategy and change in smoking behaviour with credibility ratings as mediator. (Hayes, 2008).