Exploring Predictors of Relapse and Maintenance Among Smokers

Little is known about the mechanisms behind relapse to different pre-Action stages of the Transtheoretical Model of Behavior Change (TTM) among ex-smokers. This study provides a preliminary investigation of the possible role static and dynamic variables, including demographic characteristics, smoking behavior and severity, and TTM effort variables, have in two ways: 1) As potential predictors of relapse to Precontemplation/Contemplation stages vs. Preparation; and 2) as potential predictors of relapse to any pre-Action stage vs. maintenance at follow-up. The study sample was derived from an integrated dataset of four TTM population-based smoking cessation interventions conducted in the United States. Unlike forward movement between adjacent stages, participants appeared to be equally likely to relapse to all three preAction stages. Being part of a treatment group was a salient predictor of being a maintainer at follow-up. Scoring higher on certain components of the Situational Temptations and Processes of Change measures differentiated those who relapsed from those who maintained at follow-up. Implications towards improving interventions and research concerning backward stage transitions are discussed.


INTRODUCTION
Cigarette smoking continues to be the leading cause of preventable disease, general morbidity, and mortality in the United States (Centers for Disease Control and Prevention, 2010;Lindson, Aveyard, & Hughes, 2010). Smoking and its consequences are a significant public health concern given the multiple negative effects they impose on an individual and population level. It accounts for almost half a million deaths each year in the U.S. and 30% of all cancer deaths (CDC, 2002). Specifically, smoking has been highly linked to numerous physical conditions such as heart disease, at least fifteen types of cancer (American Cancer Society, 2010), and chronic lung disease among numerous other acute and chronic maladies (USDHHS, 2010). Furthermore, smoking costs American citizens $193 billion in healthcare and lost productivity at the workplace (CDC, 2010).
It is still estimated that, approximately, 1 in 5 adults continue to smoke (CDC, 2010;Saad, 2010) yielding no significant changes in smoking prevalence among American adults over the past five years (Dube, McClave, James, Caraballo, Kaufmann, & Pechacek, 2010;Saad, 2010). On a positive note, it is estimated that 53.1% of smokers report that they have tried to quit smoking and stopped smoking for at least 24 hours in the previous year (CDC, 2008). However, before becoming completely abstinent, most smokers make a number of quit attempts (usually between 4 and14) (Kaida et al., 2004;Communiquenz, 2007). This implies that relapse is a common factor; much more so than complete abstinence after the first quit attempt in the behavior change process (DiClemente, 2006;Piasecki, Fiore, McCarthy, and Baker, 2002). It is also estimated that approximately 75% of those who become abstinent eventually relapse (Agboola, Coleman, Leonardi-Bee, McEwen, & McNeill, 2010;Brownell, Marlatt, Lichtenstein, & Wilson, 1986;Miller & Hester, 1980) days or weeks after the first quit attempt (Garvey, Bliss, Hitchcock, Heinoldr, & Rosner, 1992).
Relapse can be defined as "the return to the problematic pattern of behavior" (DiClemente, 2006). Based on the Transtheoretical Model of Behavior Change (TTM) (Prochaska & DiClemente, 1983), relapse is better defined as "recycling" in which an individual transitions backwards through the pre-Action Stages of Change (SOC; Precontemplation, Contemplation, or Preparation) before moving forward to the Action stage again (DiClemente, 2006) where one quits smoking. Relapse is defined as a type of regression in which an individual moves back from the Action or Maintenance stages to any pre-Action stage, whereas, regression takes place when an individual moves back to an earlier SOC from any stage. In an action paradigm most relapsers are considered the same, as failures to take effective action. In the TTM, relapse to the Preparation stage where smokers are immediately preparing to take action again would be qualitatively and quantitatively different from relapse to Precontemplation where smokers can become demoralized about their abilities to quit.
Quantitatively, relapse to Preparation involves less stage regression than relapse to Precontemplation.
The Transtheoretical Model of Behavior Change (TTM) is a comprehensive model which lays out a blueprint for intentional behavior change (Prochaska & DiClemente, 1983;Prochaska, DiClemente, & Norcross, 1992;Prochaska & Velicer, 1997). Stage of Change, one of the core constructs of the TTM, provides a useful approach to conceptualizing readiness to change any particular behavior (Prochaska & Velicer, 1997). The SOC construct for smoking cessation is used to assess an individual's readiness to quit smoking. In the Precontemplation stage (PC), individuals are not intending to take action to change a given behavior in the next six months.
Their reluctance may be due to unawareness, misinformation, or resistance to change.
In the next stage, Contemplation (C), individuals tend to be ambivalent about change but at the same time are intending to take action in their behavior in the next six months. In Preparation (PR), individuals have a clear intention of changing their behavior in the next 30 days and may have even started taking steps towards behavior change. In the action stage (A), individuals are in the process of changing their behavior for at least 24 hours but have done so for less than six months. In the Maintenance (M) stage, individuals work on maintaining the acquired healthy behavior which they have managed for at least 6 months whilst also focusing on curtailing setbacks. Transitions between stages are variable as some individuals stay in certain stages for some time while others move backwards (regress) or recycle through earlier stages before moving forwards and becoming abstinent Velicer, Norman, Fava, & Prochaska, 1999).
Another construct within the TTM is Decisional Balance (DB), which is derived from Janis and Mann (1977). It was adapted and initially applied to smoking cessation . SOC is linked to an individual's weighing of the benefits (Pros) and costs (Cons) of smoking . DB has been found to be valuable in predicting transitions between stages and overall behavior change (Prochaska, Velicer, DiClemente, Guadagnoli, & Rossi, 1991).
Based on Bandura's theory (1977), the Self-Efficacy construct for smoking cessation represents the confidence in one's ability to manage and cope with situational temptations to smoke (Prochaska et al., 1997;Velicer et al., 1990).
Temptations are manifested as the converse of confidence in the context of smoking cessation. In TTM-based studies, three factors emerge as reflecting the most common types of tempting situations: negative affect or emotional distress (Negative/Affective), positive social situations (Positive/Social), and craving (Habit Strength/Addictive). The Situational Temptations measure appears to be receptive to changes in forward transitions particularly through the later stages of change and is an effective predictor of relapse (Prochaska, DiClemente, Velicer, Ginpil& Norcross, 1985;Velicer, DiClemente, Rossi & Prochaska, 1990). For health behaviors, while confidence scores have been shown to increase from PC to M (Prochaska, Velicer, Guadagnoli, Rossi, & DiClemente, 1991;Velicer, Prochaska, Fava, Norman, & Redding, 1998), temptation scores tend to decrease as stage transitions occur from PC to M (Prochaska & DiClemente, 1984;Velicer et al., 1990).
Lastly, Processes of Change (POC) encompass covert and overt strategies individuals utilize to move forward through SOC (Prochaska & DiClemente, 1983). In this case, the processes assess how people proceed to smoking cessation. Each process consists of a variety of techniques that are linked to different theoretical orientations (Prochaska, DiClemente, & Norcross, 1992). TTM research suggests that successful self-changers utilize different processes at each SOC. The processes are categorized under two higher order factors, experiential and behavioral, each consisting of five subscales (Prochaska, Velicer, DiClemente, & Fava, 1988 & Fava, 1988).
It has been found that each POC is highly linked to an individual's SOC Prochaska et al., 1991). In other words, some processes are used more often within certain SOC. As such, experiential processes are typically used more often in earlier pre-Action stages while behavioral processes are used more often in Action and Maintenance (Prochaska & Velicer, 1997). In regards to smoking cessation, process use increases while moving forward and decreases as one moves backwards through the SOC. Those in Precontemplation are found to use processes the least compared to individuals in other stages. Behavioral processes are found to be utilized the most in Action and tend to decrease as one regresses back to earlier stages.

Existing Research on Relapse and Maintenance among Smokers
The literature suggests that relapse prevention efforts have had 'modest' success and fall short of laying out a consistent formula to curb any and all types of relapse (DiClemente, 2006). Some suggest that efforts need to focus less on relapse prevention and more on "promoting recycling" which can yield important information regarding what smokers learn during their relapse that may provide insight into their long-term abstinence (DiClemente, 2006). As such, it is important to examine all patterns of individuals' change over time (Sun, Prochaska, Velicer & Laforge, 2007).
There is little research on types of relapse and their predictors to pre-Action stages compared to those of forward transitions from pre-Action stages.
More specifically, most TTM-based stage sequence studies focus on forward transitions within the pre-Action stages and from the pre-Action stages to the Action stage. One study (Sun, Prochaska, Velicer & Laforge, 2007) looked at patterns of the 14 TTM variables among three identified groups defined by their pattern of change over time (stable quitter, relapsers, and stable smokers). Relapsers, on average, were found to use five of the Processes of Change the most (Dramatic Relief, Self Reevaluation, Environmental Reevaluation, Helping Relationships and Self Liberation). The authors concluded that relapsers were in fact working hard, or just as much as maintainers, but rather may have lacked adequate preparation for long-term cessation. In contrast to maintainers, relapsers did not decrease their use of SR and did not increase reliance on SC and CC. Relapsers also failed to reduce the utilization of the Pros and Cons of smoking and their overall Situational Temptations cues which reiterates the "successive approximation" or trial-and-error approach to learning to sustain behavior that occurs in recycling (DiClemente, 2006).
Using the same smoking cessation data as Sun et al. (2007), Blissmer et al. (2010) found no significant evidence for effects of demographic characteristics on long-term changes among smokers. The largest effect sizes were found for Stage of Change. Furthermore, Situational Temptation scores were significantly higher at baseline for stable smokers. Problem severity baseline scores were lower for those who were in the Action and Maintenance stages at 24-months. For the latter group of participants, the Pros of smoking were significantly lower as well.
No studies could be found that assessed regression from Action and Maintenance to specific pre-Action stages. One study (Hoving, Mudde, & deVries, 2006) focused just on regression within the pre-Action stages. Overall, Hoving and colleagues found that smokers were more likely to move to an adjacent stage rather than skipping over a stage, yet cited their limitation in testing the differences due to a limited sample size. They also reported that they did not find any evidence to confirm their hypothesis on lower perception of Pros of smoking predicting a backward transition from the Contemplation stage. Specifically Within the TTM framework, significant predictors of successful cessation or abstinence have been found to include problem severity, age, education (Velicer, Redding, Sun, & Prochaska, 2007), and Stage of Change and TTM effort (Decisional Balance, Situational Temptations, Processes of Change) variables (Blissmer et al., 2010;Velicer et al., 2007;Prochaska, Velicer, Prochaska, & Johnson, 2004).
There are no studies looking at predictors of relapse from the Action/Maintenance stages to specific pre-Action stages within TTM framework. This is pertinent information to be aware of because the lack of such research in this area provides a large gap in our understanding of relapse. If we know that each stage holds unique characteristics that pertain to the use of TTM variables, then looking at differences between the stages that pertain to relapse is imperative to our overall understanding of relapse and long-term abstinence. Consequently, we need to better understand the variable patterns of change individuals exhibit over time .

The Present Study
While TTM has primarily been used to look at forward transitions from the pre-Action stages to Action and Maintenance, to our knowledge, there appears to be no literature on relapse from the latter two stages to pre-Action stages. Looking at such transitions would be valuable given that the Action and Maintenance stages hold valuable information about what smokers are doing that can lead to long-term compared to relapse to pre-Action stages. Furthermore, transitions through the Stages of Change reflect differences in cognition, experience, and behavior which suggest that each of these is used at different times throughout the "change process" (Heckhausen, & Gollwitzer, 1987).
Relapse and maintenance patterns are important to assess simultaneously as they each contribute different, though equally important information about the behavior change process. Given that most smoking research focuses on the transition from being a smoker to a non-smoker, relapse tends to be viewed as a failure (Redding, Prochaska, Paiva, Rossi, Velicer, Blissmer et al., 2011). Relapse is a natural part of the quitting process and the goals of the present study are to elucidate potential patters of relapse and maintenance and, hence, are two-fold: 1) To identify variables that are more likely to predict relapse to specific pre-Action stages, PC/C vs. PR; and 2) to explore variables that differentiate those who relapse (to any pre-Action stage) from those who remain quit. The current study recruited only current smokers (in the pre-Action stages) at baseline, therefore we focus on the participants who reported being smoke-free (in A/M) at 12-months post-baseline and who went on to complete the 24-month follow-up assessment.

METHOD Intervention
This study involved secondary data analysis on a combined dataset of four Participants were assessed at 6 month intervals post-baseline through 30 months. The sample, recruitment, inclusion and exclusion criteria, and outcomes for all samples were determined by principal investigators for each study.

Participants
Since all participants were current smokers at baseline (in the pre-Action stages at baseline), therefore, this study includes participants who were in the Action/Maintenance stages at 12 months post-baseline (N=661) and who had complete data at 24-months post-baseline (N=521). Participants who reported that they were in any of the pre-Action stages (PC, C, or PR) at 24-months were classified as relapsers, and those who were in Action/Maintenance were classified as maintaining/maintainers.

Measures
Demographics. Single items were used to assess age, gender, race, ethnicity, education, and marital status.

Severity of Smoking.
Severity of smoking for participants were assessed by the number of cigarettes they smoked daily and time until first cigarette, two main parts of the Fagerstrom index that reflect the degree of addiction (Fagerstrom, Heaherton, & Kozlowski, 1990). In addition to these items, previous longest quit attempt in months and number of quit attempts in the past year were also assessed.
Intervention Group. All four studies used a common TTM-tailored expert system intervention that was printed and delivered to participants' homes. Participants also received stage-matched self-help manuals. Control groups received assessments only.
Stage of Change. Stage of Change was measured by a staging algorithm that assessed their readiness to quit smoking, with response options of 1= Precontemplation (not intending to quit smoking within the next six months), 2=Contemplation (intending to use the quit smoking within the next 6 months), 3= Preparation (intending to use the quit smoking within the next 30 days), 4= Action (quit smoking within the last six months), and 5=Maintenance (quit smoking more than six months ago).
Decisional Balance. An 8-item decisional balance measure (Appendix A) assessed the relative importance of various advantages (Pros) and disadvantages (Cons) in an individual's decision to smoke. This measure assessed Pros of smoking with 4-items (α=.87) and Cons of smoking with 4-items (α=.90). Participants were asked to rate the importance of each item on a 5-point Likert scale, ranging from 1 = "Not At All Important" to 5 = "Extremely Important" (Velicer, DiClemente, Prochaska, & Brandenburg, 1985).
Situational Temptation. A 9-item measure (Appendix B) assessed the intensity of urges to engage in a specific behavior when faced with difficult situations. Participants rated their confidence to be able to quit smoking in the presence of temptations on a 5point Likert scale ranging from 1= "Not At All Tempted" to 5="Extremely Tempted" (Velicer, DiClemente, Rossi, & Prochaska, 1990;DiClemente, 1986DiClemente, , 1981.

Processes of Change.
A 20-item measure (Appendix C) assessed the ten Process of Change. Participants rated their frequency of process use in the past 30 days on a 5point Likert scale ranging from 1= "Never" to 5="Repeatedly" (Fava, Rossi, Velicer, & Prochaska, 1991).

Demographic Characteristics
Overall Sample.

Hypotheses and Findings
Hypothesis 1: Based on the assumption that self-changers typically move one stage, participants in the treatment group are expected to relapse to PR (85%) vs. PC/C (15%) at 24-months post-baseline more so than those in the control group. PC/C at 24-months (χ² (1) = 1.46, p = .23) ( Table 5).
Analysis 1b: Logistic regression analysis was conducted to determine the likelihood of regression to PR vs. PC/C between control and treatment group participants.
Results 1b: Logistic regression analysis indicated that participants in the treatment group were not more likely to relapse to PR vs. PC/C at 24-months compared to participants in the control group (OR = 1.60, p = .23) ( Table 5).
Hypothesis 2: Participants in the control group will be more likely to relapse (regress to any pre-Action stage) vs. maintain at 24-months compared to participants in the treatment group.
Analysis 2a: Crosstabs were conducted to compare those who relapsed vs.
maintained at follow-up.
Results 2a: Crosstabs indicated that 35.0% of the control group participants relapsed to any pre-Action stage, while 19.9% of the treatment group participants relapsed to any pre-Action stage at 24-months. There was a statistically significant relationship between being in the treatment group and relapsing vs. maintaining at 24months (χ² (1) = 14.19, p = .00) ( Table 6).
Analysis 2b: Logistic regression analysis was conducted to determine the likelihood of relapse vs. maintenance among control and treatment group participants.
Results 2b: Logistic regression analysis indicated that participants in the control group were more than twice as likely to relapse (OR = 2.17, p = .00) to any pre-Action stage vs. maintain at 24-months compared to participants in the treatment group (see Table 12).
Hypothesis 3: In both the treatment and control groups, participants who reported higher Pros of smoking, lower Cons of smoking, and higher Temptations to smoke at 12-months post-baseline will be more likely to relapse vs. maintain at 24-months.
Analysis 3: Two logistic regression analyses were conducted, one including Pros and Cons and the other including Situational Temptations) at 12-months, to determine the likelihood of being a relapser vs. a maintainer at 24-months.

Results 3:
Decisional Balance (Pros and Cons). Participants who reported higher on the Pros of smoking and lower Cons of smoking at 12-months were not significant at the .05 level set for predicting the likelihood of relapse vs. maintenance at 24-months. Odds ratios were 1.03, p = .16 for Pros and 1.00, p = .94 for Cons (see Table 7).
Situational Temptations. Participants who reported higher Habit Strength scores were more likely to relapse vs. maintain at 24-months (OR = 1.05, p = .02) (see Table 7).
Hypothesis 4: Hypothesis 4 was a repeat of Hypothesis 1 in the Thesis Proposal; as a result, it has been eliminated from the Thesis due to redundancy.
Hypothesis 5: If regression works the way it is expected with forward transitions, then participants who reported higher Pros of smoking, lower Cons of smoking, and higher Temptations to smoke at 12-months will be more likely to regress back to PC/C rather than PR at 24-months.
Analysis 5a: Two logistic regression analyses were conducted, one including Pros and Cons and the other including Situational Temptations)at 12-months, to determine the likelihood of regression to PC/C vs. PR.

Results 5a:
Pros, Cons, and Situational Temptations were converted into T-scores (M = 50, SD=10), and then entered into two separate logistic regression analyses.

Decisional Balance (Pros and Cons).
Reporting higher Pros of smoking (OR = .90, p = .06), and lower Cons of smoking (OR = .99, p = .71) at 12-months were not statistically significant predictors of relapse to PC/C vs. PR at 24-months (Table 8). .05) was not a statistically significant predictor of relapse to PC/C vs. PR at 24-months (Table 8).

Exploratory Analyses
The final step of the study consisted of exploratory analyses evaluating findings on the Processes of Change construct of TTM, and across different demographic groups and smoking behaviors and severity.

Processes of Change. Based on Sun et al.'s findings (2007), all Processes of Change
items were evaluated to see whether using DR, SR, ER, HR, and SL processes at 12months would increase the likelihood of relapse vs. maintenance at 24-months. It was predicted that those who use less DR, SR, ER, HR, and SL processes at 12-months would relapse to the earlier stages, PC/C vs. PR, at 24-months.
Separate scores for the ten Processes of Change (CR, CC, DR, ER, HR, RM, SC, SL, SO, SR), the Experiential subscale score, the Behavioral subscale score, and the Total Processes of Change score at 12-month were converted into T-scores (M = 50, SD = 10) and entered into logistic regression analyses. Next, to avoid collinearity between the subscales and the total scores for Processes of Change variables, three separate logistic regression analyses were conducted for all ten Processes, the Experiential and Behavioral subscale scores and the Total Processes of Change score at 12-months.
Two of the Processes of Change were significant predictors of relapse to PC/C vs. PR at 24-months at the .05 level: CR (OR = 1.17, p = .03) and SR (OR = .79, p = .03). Specifically, those who had higher scores of CR were more likely to relapse to PC/C compared to PR. Those who had higher scores of SR were less likely to relapse to PC/C compared to PR. The remaining eight processes were not found to be significant predictors (Table 9).
Two of the Processes of Change were significant predictors of relapse vs.
maintenance at 24-months at the .05 level: RM (OR = 1.05, p = .04) and SR (OR = 1.08, p = .01). Specifically, those who had higher scores of RM were more likely to relapse compared to maintain, and those who had higher scores of SR were more likely to relapse compared to maintain. The remaining eight processes were not found to be significant predictors (Table 10).
Demographic Variables. None of the baseline demographic variables were significant at the .05 level set for predicting the likelihood of relapsing to PR vs. PC/C at 24months. Furthermore, their corresponding confidence intervals were fairly wide (Table   11).
With regards to baseline demographic variables as predictors of relapse to any pre-Action stage vs. maintenance at 24-months (  (Table 13).
In contrast, participants who had a previous longest quit attempt last between 36-72 months compared to one month were less likely to relapse vs. maintain (OR = 0.42, p = .02). In addition, participants who had made 3-10 quit attempts compared to no attempts prior to baseline were more likely to relapse vs. maintain (OR = 1.70, p = .03). Participants who had been quit 36-72 months at some point pre-baseline compared to one month in the past were less likely to relapse vs. maintain (OR = .42, p = .02) (Table 14) during this timeframe.
Other Study Timepoints. For further data exploration, all aforementioned static and dynamic independent variables were evaluated at between 6-18 months and 18-30 months of the larger study. Logistic regressions performed at these timepoints; however, due to small sample sizes did not have adequate power to detect significant findings (Wright, 1995). Therefore, meaningful comparisons between timepoints could not be made.

DISCUSSION
Most smoking relapse research has focused on static individual factors (i.e. demographics and smoking severity) (Ockene et al., 2000;Swan, Jack, & Ward, 1997 in Shiffman, 2005). Given that such characteristics are unchangeable and only provide information regarding who tends to relapse, looking at dynamic variables can also provide valuable information on when and why relapse occurs. To that end, the primary goal and strength of this study was to explore static as well as dynamic variables including demographic characteristics, smoking behavior and addiction severity, and three of the TTM effort variables (Decisional Balance, Situational Temptations, Processes of Change) as potential predictors of relapse to pre-Action stages within a multivariate and longitudinal study design.
Preliminary findings indicated that the majority of participants (71.4%) maintained at follow-up. Disconfirming Hypothesis 1, the majority of relapsers moved back to PC/C (n = 94) vs. PR (n = 55). So, at first glance, those who relapsed tended to relapse to earlier stages where they were not intending to quit again in the next six months or were intending to quit in the next six months but were not actively preparing to engage in the cessation process. However, when participants in PC and C were separated, relapsers were, in fact, fairly equally distributed between all three pre- Participants aged 25-64 were less likely to relapse maintain compared to participants aged 18-24. One interpretation is that even though young adulthood is a time of many transitions, including changes in smoking behavior in which initiation of smoking as well as relapse are common (Tercyak, Rodriguez, & Audrain-McGovern, 2007) most adults who have been longer-term smokers are at increased risk to relapse. This is corroborated by the fact that the older people get, they are more likely to have more quit attempts which increases the likelihood of relapse. In line with previous findings (Velicer et al., 1990) the psychological and physiological aspects of smoking behavior assessed by Habit Strength items as well as the Total Situational Temptations scores predicted that those who scored higher on those items were more likely to relapse vs. maintain at follow-up. This discrepancy in findings between the two sets of measures of addiction severity indicate that a more comprehensive way of assessing addiction via immediate emotional and social factors, also termed as "process-situational," an approach pioneered by Martlatt and Gordon (1985), may be able to better capture the "process" of relapse. Furthermore, this finding adds to one assumption that relapsers tend to relapse not solely due to smoking addiction severity, necessarily, but due to immediate precursor factors such as emotional distress (Shiffman & Waters, 2004;Shiffman, Paty, Gnys, Kassel, & Hickox, 1996). In addition, although psychological as well as physical repercussions of nicotine withdrawal is an established barrier to quitting smoking, it may not play the same role among individuals who have already quit smoking given that the intensity of withdrawal symptoms typically decrease during the first month of quitting (Hatsukami, Stead, & Gupta, 2008). Based on the significance of the Total Situational Temptations score finding, it is also possible that positive social experiences related to smoking, in which a positive affective component is present, also can instigate relapse (Velicer, DiClemente, Rossi, & Prochaska, 1990).
Surprisingly, Pros and Cons did not differentiate between those who relapsed vs. maintained or between those who relapsed to PR vs. PC/C at follow-up. One possible interpretation is that perhaps even though ex-smokers are aware of the Pros and Cons of smoking, they do not find them helpful when faced with situational distress.
Consciousness Raising (CR) and Self-Reevaluation (SR) were two significant predictors of regression to PC/C vs. PR. CR is a key process for self-changers to utilize as they transition from Precontemplation to later pre-Action stages while acquiring new information regarding quitting smoking. The use of the latter process was found to be predictive of relapsing to the earlier Stages of Change. Similarly, SR is also a key experiential process for self-changers to utilize as they transition forward from a non-Action stage to a more action-oriented stage. Similar to what we have observed in previous two studies (Redding et al., 2011;Sun et al., 2007) SR was a key process that differentiated relapsers from maintainers in the present study. In this case, however, we also found that those relapsers were less likely to move back to PC/C compared to PR. So in fact, those who do use SR are more likely to move back only one stage; to PR in which they are still working towards quitting again. As such, once individuals enter the Action and Maintenance stages, they would benefit from decreasing their reliance on SR and increasing their utilization of Behavioral Processes such as Helping Relationships for potential stress management and support, and Stimulus Control for alteration of environmental cues to maintain the cessation process.
Overall, when a continuous measure (e.g. Reinforcement Management) is used, the score range is wider and therefore the interpretation of the odds ratio is different from a dichotomous predictor variable. The odds ratio for RM was 1.05 which means that there was a 5% increase in relapse for each one unit increase in Temptations, and the range for this variable is from 2-10. As a result, even though 5% appears to be a small increase, it is, in fact, larger if a change from, for example, a sum score of 2 to 10 is being considered. In such a case, the odds of relapse would be 40% greater. This suggests that the aforementioned Processes of Change, including SR

Limitations
This study had several limitations. First, all analyses were based on a predominantly White and female sample. Although sample characteristics were representative of the larger population-based clinical trial, homogeneity of race and gender limit generalizability of findings. Second, the long recall period between baseline and prior year allows for potential recall bias about quit attempts and prior smoking behavior (Gilpin & Pierce, 1994). Although efforts were made in both studies to recruit proactively from the general population, study participants had to be willing to engage in the intervention related to smoking cessation. Also, given fairly small samples, differentiating predictors between pre-Action stages was not robust. It is also important to note that given that an odds ratio is reflective of a one unit increase in the dependent variable, dichotomization may have inflated classification (e.g. treatment vs. control group).

Future Directions
The traditional view has been that biological addiction severity accounts for most of the barriers to quitting smoking. However, we now have preliminary evidence to suggest that this is not entirely true for relapse. Specifically, our findings add to the literature that relapse may be much more of an affective and situational process among ex-smokers. Studies suggest that nicotine craving, an intense desire to smoke, typically lasts around 5-12 minutes, and that cravings, as well as increases in smoking rate and nicotine intake are highly related to acute physical or psychological stress (al'Absi,Wittmers, Erickson, Hatsukami, & Crouse, 2003;al'Absi, Amunrud, &Wittmers, 2002). As such, it is imperative that ex-smokers have the tools to be able to manage stress effectively.
The interaction between craving and stress is important to examine, since stress-induced craving states have been associated with relapse vulnerability (Ng & Jeffery, 2003). As such, one promising approach would be to provide additional expert guidance on how ex-smokers can manage stress effectively when they enroll in treatment at any Stage of Change. If resources are limited, tailored guidance can be provided for those who enter the Action stage given that underlying withdrawal symptoms including anxiety, anger, and irritability (Hughes, 2007) in addition to the physical symptoms appear to most prevalent and severe closer to the time of quitting.
In addition, future research needs to find ways to capitalize on TTM variables over the course of the intervention, as well as after treatment ends given that smoking cessation is a lifelong behavior change. In Sun et al.'s study (2007), relapsers were using five of the processes the most: Dramatic Relief, Self-Revaluation, Environmental Revaluation, Helping Relationship, and Self-Liberation . In the present study, the relapsers were using CR and SR more than the maintainers. As proposed in the latter study, relapsers did not increase their use of Behavioral Processes such as Counterconditioning and Stimulus Control. Furthermore, future research could build upon these findings by tailoring interventions and encouraging evaluating the Cons of smoking when contemplating smoking again.
Another area that may need further exploration is the quantitative and qualitative investigation of the specific decision-making process that goes on between being tempted to smoke and actually lighting up a cigarette as well as the time it takes between those two timepoints. Looking at relapsers over time at more than two timepoints in future studies may provide additional pertinent information about relapse. And finally, the preliminary findings in the present study need further evaluation when data are adequately powered. With friends at a party. Temptation 2 When I first get up in the morning. Temptation 3 When I am very anxious and stressed. Temptation 4 Over coffee while talking and relaxing. Temptation 5 When I feel I need a lift. Temptation 6 When I am very angry about something or someone. Temptation 7 With my spouse or close friend who is smoking. Temptation 8 When I realize I haven't smoked for a while. Temptation 9 When things are not going my way and I am frustrated.
engage in the healthy behavior. work that remind me not to smoke.

Social Liberation
Realizing that the social norms are changing in the direction of supporting the healthy behavior change.
1. I find society changing in ways that make it easier for nonsmokers.