Predictors of Smoking Cessation Attempts and Success Following Motivation-Phase Interventions Among People Initially Unwilling to Quit Smoking

Abstract

Introduction

Most people who smoke cigarettes are not willing (ie, not ready) to make a quit attempt (QA) at any given time. Unfortunately, interventions intended to increase QAs and the success of QAs are only modestly effective. Identifying processes leading to QAs and quitting success could guide intervention development.

Aims and Methods

This is a secondary analysis of a randomized factorial trial of 6 weeks of motivation-phase interventions among primary care patients (N = 517) who were initially unwilling to quit but were willing to reduce their smoking. Using logistic regression, we controlled for treatment condition and tested whether baseline or change in smoking-related constructs after 6 weeks of treatment predicted (1) making an at least 24 h QA between weeks 6 and 26 and (2) quitting success at week 26 (7-day point-prevalence abstinence among those who made a QA). Predictors included cigarettes/day, time to first cigarette, motivation to quit, quitting self-efficacy, anticipated urges to smoke if quit, positive affect, negative affect, and time spent around others who smoke.

Results

In multivariable models that included all smoking-related constructs, changes in the following variables predicted initiating a QA above and beyond other variables: greater baseline time to first cigarette (odds ratio [OR] = 1.60), increases in time to first cigarette (OR = 1.27), and increases in quitting self-efficacy (OR = 1.14). Increased motivation to quit predicted conversion of a QA into quitting success at 26 weeks (OR = 1.36).

Conclusion

Predictors of making a QA differed from predictors of quitting success. Predictors of QAs and success could each serve as important treatment targets of motivation-phase interventions.

Implications

Motivation-phase interventions for people initially unwilling to quit smoking cigarettes may be improved by striving to increase their (1) time to first cigarette and quitting self-efficacy to promote QAs and (2) motivation to quit to promote quit success. Future experimental tests of such interventions are needed to identify causal determinants of QAs and quitting success.

Introduction

Most people who smoke cigarettes in the United States are unwilling (ie, not ready, not motivated, or not able) to make a quit attempt in the near future.¹ Motivation-phase treatments have been developed to increase quitting in those initially not willing to quit smoking.^2–6 Although research has identified motivation-phase interventions that promote abstinence in people initially unwilling to make a quit attempt,⁵ such interventions are only modestly effective. For example, in a review on the effectiveness of nicotine replacement therapy (NRT) for people not willing to make a quit attempt, the estimated sustained abstinence rate was 6.75% at 6 months.⁷ Tests of motivation-phase interventions designed using the Phase-Based Model of Cessation,⁸ yielded evidence that NRT gum might effectively spur quit attempts and behavioral reduction counseling might increase long-term abstinence.^2,9 However, these effects were small. A recent Cochrane review⁶ concluded that there was inconclusive evidence to determine whether motivational interviewing¹⁰ improves smoking cessation outcomes compared to no treatment. In sum, there is a need to substantially improve the effectiveness of motivation-phase treatments.

Identification of the processes that lead to abstinence in those initially unwilling to quit could provide important targets for motivation-phase treatments, leading to improved effectiveness. The person who is unwilling to make a quit attempt must first become motivated to make an attempt and then be supported in their quit attempt. Consistent with the Phase-Based Model,⁸ different smoking-related constructs (ie, tobacco dependence, quitting self-efficacy, withdrawal) may be associated with different phases of smoking behavior change. Indeed, a prior review on baseline smoking characteristics concluded that predictors of quit attempts differ from predictors of successful abstinence in general population samples.¹¹ However, more work is needed to understand how baseline levels and change in smoking-related constructs influence initiating a quit attempt and maintaining quitting success among those not initially willing to try to quit. This analysis tested which smoking-related constructs are significantly associated with the initiation of a quit attempt and the likelihood of maintaining abstinence (quitting success) once an attempt has been made among people who were unwilling to quit at baseline.

Predictors of Quit Attempts

Baseline Measures of Quit Attempts

Prior research has identified several smoking-related constructs that when measured at baseline are associated with future quit attempts (Supplementary Table S1). A review by Vangeli et al.¹¹ found that greater baseline cigarettes per day (CPD) was associated with a decreased likelihood of initiating a quit attempt at long-term follow-up (12–28 months). However, this review¹¹ and other evidence¹² suggest inconsistent or modest relations between other tobacco dependence indices and quit attempts. Baseline motivation to quit¹³ has predicted initiating a quit attempt in most population studies¹¹ whereas baseline self-efficacy (belief in the ability to quit) is a less consistent predictor of quit attempts.¹¹ Thus, there is modest evidence that a baseline measure of CPD predicts future quit attempt occurrence; the evidence is more convincing with regard to motivation to quit.

Change Measures That Predict Quit Attempts

Quit attempt occurrence has been associated with a reduction in CPD among those who did not initially intend to quit at a 6-month follow-up^4,14 and in a fine-grained analysis of daily changes in smoking.¹⁵ In addition, decreases in dependence¹⁴ and increases in motivation^4,14 predicted making a quit attempt at a 6-month follow-up. Self-efficacy predicted quit attempts in one¹⁴ but not another⁴ trial of people initially unmotivated to quit at a 6-month follow-up. Finally, in an international population-based sample, reduction in the number of friends who smoke (ie, contact with smokers) more than a year was associated with an increased likelihood of quit attempts.¹⁶

Predicting Quitting Success

Baseline Predictors of Quitting Success

Baseline measures predicting the likelihood of quitting success (abstinence) included the dependence measure time to first cigarette upon awakening,¹⁷ with meaningful prediction occurring in population studies¹¹ and in clinical trials at 1-week to 1-year follow-ups.^12,18–20 Similarly, most¹¹ but not all²¹ population studies with 6- to 28-month follow-ups found smoking fewer CPD at baseline was associated with quitting success. There was considerable evidence across multiple cessation trials and longitudinal studies and at multiple follow-up intervals, that stronger urge to smoke at baseline is also associated with a decreased likelihood of achieving abstinence.^20,22,23 The Vangeli review found conflicting evidence concerning the relation of baseline self-efficacy with quitting success across follow-up periods of 12–28 months.¹¹ Moreover, motivation to quit has not shown consistent relations with quitting success in population studies^11,24 or in a trial of people attempting to quit at 1-, 6-, and 12-month follow-ups.¹⁹ Baseline positive affect (PA) was associated with successful abstinence in trials with 1- to 4-week^25,26 but not 24-week²⁷ follow-ups. Baseline negative affect (NA) and depression symptoms were associated with relapse in studies with 12-week²⁸ and 3-year²⁹ follow-ups. Finally, greater contact with others who smoke appears to be associated with a decreased likelihood of quit success in studies with 1- to 4-week^20,30 and 1-year follow-ups.³¹

Change Measures That Predict Quitting Success

Among people initially unmotivated to quit smoking, changes that have predicted abstinence include decreases in dependence¹⁴ and increases in motivation^4,14 at a 6-month follow-up (Supplementary Table S1). Among participants motivated to quit, reduced urge to smoke was associated with quit success at 1- to 8-week follow-ups.^26,28,32 Reduced NA was associated with quit success in two^26,32 of three²⁸ studies with 1- to 8-week follow-ups. Similarly, increased PA was associated with quit success in two^26,28 of three²⁵ studies with 1- to 8-week follow-ups. Finally, a reduction in the number of friends who smoke (ie, contact with those who smoke) more than a year was associated with an increased likelihood of maintaining abstinence in an international population-based sample.¹⁶ In sum, modest evidence suggests that individuals are more likely to show abstinence at follow-up if over time they show decreases in dependence, NA, urges to smoke, and contact with smokers and increases in PA.

Thus, this literature review identifies multiple predictors of quit attempts or quitting success. However, results show considerable heterogeneity across the association baseline and change variables’ association with quit attempts and quit success. For instance, quit attempt occurrence was fairly consistently predicted by both baseline motivation to quit and increases in such motivation over time. By contrast, quitting success was instead fairly consistently predicted by baseline measures of urges, time to first cigarette, and amount of contact with other smokers. The change measures that predicted successful quitting included decreases in dependence, urges to smoke, NA, and contact with smokers and increases in PA and motivation. In general, few measures were predictive of both outcomes.

It is important to note two caveats regarding the relations just reviewed. First, the great majority of studies, especially those examining quitting success, were clinical cessation trials; three studies involved individuals initially unwilling to quit smoking.^4,14,15 Second, the reviewed studies of change often do not permit strong inferences regarding temporal priority of change in smoking versus change in the predictor (eg, increases in self-efficacy or reductions in dependence may be affected by reduced smoking). In sum, at present, there is considerable uncertainty regarding the factors that influence quit attempts and successfully quitting that may constitute important targets for motivation-phase treatments.

The parent trial² for this secondary analysis was a factorial randomized trial of nicotine gum, nicotine patch, behavioral reduction, and motivational interviewing with people initially unwilling to attempt to quit but willing to reduce smoking. Nicotine gum was associated with making a quit attempt during treatment but none of the intervention components produced significant main effects on quit attempts during follow-up.⁹ Behavioral reduction increased abstinence at the 12-week follow-up, and a complex interaction showed that nicotine gum with behavioral reduction and without motivational interviewing increased cessation success.² In the current analyses, we tested whether baseline values or changes in smoking-related constructs during motivation-phase treatment influence initiating a quit attempt or maintaining quitting success during follow-up. Specifically, we tested baseline values and change in the following constructs as predictors of making a quit attempt and successfully quitting: CPD, time to first cigarette, motivation to quit, quitting self-efficacy, anticipated urge to smoke if quit, PA, NA, and time spent around others who smoke. After testing each variable in individual models, variables were tested in multivariable models that included all smoking-related constructs (primary analysis). The predictor variables were selected because of their relevance to the motivation phase and because they are often cited as related to smoking cessation success.

Methods

The full study procedure has been reported elsewhere.² Briefly, participants (N = 517) were recruited from 11 primary care clinics in Wisconsin between June 2010 and October 2013. Patients who smoke were identified via the electronic health record and were invited by medical assistants to participate in a research program to help them either quit or reduce their smoking. Interested patients were electronically referred to the research office for screening. Primary inclusion criteria were adults who smoked at least 5 CPD during the last 6 months, were unwilling to make a quit attempt in the next 30 days but were willing to reduce, were not taking varenicline or bupropion, agreed to use only smoking medications provided by the study, had no medical contraindications to NRT, and, for women with childbearing potential, agreed to use approved contraceptives.

Motivation-phase treatment lasted 6 weeks. Participants were randomized to receive 1 of the 16 possible combinations of the four treatment factors: (1) Nicotine Patch (one 14-mg patch per day) versus none, (2) Nicotine Gum (at least five pieces of 2-mg gum per day) versus none, (3) Behavioral Reduction Counseling (initial 20-min in-person session followed by 6 weekly 10-min calls) versus none, and (4) Motivational Interviewing¹⁰ strategies (initial 20-min in-person session followed by three 10-min biweekly calls) versus none. At the end of treatment, all participants had the option to repeat the same treatment for an additional 6 weeks. Participants also had the option to engage in cessation treatment at any point during the 26-week study period. Cessation treatment, which was the same for all participants, consisted of 8 weeks of a combination of nicotine patch and nicotine gum and two brief telephone counseling sessions.

Measures

We examined smoking-related constructs during motivation-phase treatment because this is when we expected to observe the greatest change among people initially unwilling to quit. The following smoking-related constructs were measured at baseline and weekly during the 6-week motivation-phase treatment. We calculated change scores from baseline to the end of treatment (6 weeks) so that greater scores would be associated with clinical improvement for all measures (eg, positive odds ratio indicates the direction of the desired effect). Cigarettes per day were assessed by asking “On average, how many cigarettes have you smoked each day this week?” Time to first cigarette was assessed with the following item from the Fagerström Test for Cigarette Dependence¹⁷: “How soon after you wake up do you smoke your first cigarette? 1=Within 5 minutes, 2=6-30 minutes, 3=31-60 minutes, or 4=After 60 minutes.” As in prior research,^33,34 motivation to quit was assessed by asking “How motivated are you to quit smoking? (1=not at all to 10=extremely),” self-efficacy was assessed by asking “How confident are you that you can quit smoking successfully in the next month? (1=not at all to 10=extremely),” and anticipated urges if quit was assessed by asking “If you do try to quit smoking for good, how severe do you think your urges to smoke will be?” (1=not at all to 10=extremely). PA and NA were assessed with the Positive and Negative Affect Schedule (composite PA and NA scores on a 1–10 scale).^35,36 As in prior research,³⁷ the time spent around others who smoke was assessed by asking “Over the past week, how often have you been with people while they are smoking? 1=Daily or almost daily, 2=A few times, 3=Rarely, 4=Never.”

Outcomes

Participants completed weekly phone assessments during the 6-week treatment period and follow-up phone assessments at 12 and 26 weeks post-enrollment. In our primary analysis, we defined having made a quit attempt as self-reported intentional abstinence for more than 24 h between the end of treatment (6 weeks) and the final follow-up visit (26 weeks). In addition, we used timeline follow-back data to assess whether participants who missed the week 12 but completed the week 26 assessment made a quit attempt after the 6-week treatment period. Analyzing quit attempts following the end of treatment allowed us to capture maximum changes in the proposed mechanisms, as well as to establish temporal precedence necessary for determining whether such mechanisms lead to desired outcomes (ie, quit attempts, quit success). We defined quit success as self-reported 7-day point-prevalence abstinence at the 26-week follow-up among those who reported making an intentional quit attempt lasting at least 24 h.

Analysis

We conducted all analyses with SPSS (IBM SPSS Statistics for Windows, Version 26.0; IBM Corp., Armonk, NY). We used logistic regression to test the influence of both baseline and change in smoking-related constructs on (1) whether a quit attempt was made after treatment (between 6 and 26 weeks) and (2) quitting success at the 26-week follow-up among those who made a quit attempt. Analysis of quit attempts included all participants (N = 517). Analysis of quit success tested whether those who made a quit attempt after week 6 (n = 269) succeeded in achieving point-prevalence abstinence at the 26-week follow-up. When testing baseline scores individually, we controlled for (1) each of the four treatment factors (1 = participant received the “on” level of that factor [ie, received the intervention component]; −1 = participant received the “off” level of that factor [ie, did not receive the intervention component]), (2) whether participants opted for an additional 6 weeks of cessation treatment, and (3) whether participants made a quit attempt during the initial 6-week treatment period. When testing change scores individually, we controlled for the covariates listed above as well as baseline values for the construct being tested. We used multivariable logistic regressions, which included all smoking-related constructs, to create models with independent predictors of outcomes (primary analysis). We included all of the previously mentioned covariates as well as baseline and change scores for all of the smoking-related constructs in the multivariable models. We did not conduct mediation analysis because we are primarily interested in identifying the influence of smoking-related constructs on quit attempts and success.

Participants who did not provide outcome information were assumed to have not made a quit attempt and to be smoking. At week 26, 44 participants (8.5%) were missing outcome data (quit attempt or abstinence). For the 142 participants (27.5%) missing data on at least one smoking-related construct at week 6, we used the last available weekly measurement during treatment to calculate change scores in our primary analyses.

In sensitivity analyses, we tested baseline scores individually without controlling for any other covariates, and we tested change scores individually after controlling only for the construct’s baseline value. In addition, we tested all models when missing data for smoking-related constructs were treated as missing. Finally, we tested all models using a quit attempt outcome variable that included quit attempts that occurred between baseline and week 6 (ie, during the initial treatment).

Results

Participants

Participant characteristics are described in detail elsewhere.² Briefly, participants averaged 47 years old (SD = 14.4), were mostly women (63.4%), white (91.1%), had attended at least some college (61.6%), and had a mean Fagerström Test for Cigarette Dependence¹⁷ score of 4.8 (SD = 2.1) at baseline. Smoking-related constructs at baseline and week 6 are reported in Table 1. In total, 52.0% (n = 269) made a quit attempt after week 6 and 7.4% (n = 38) made their only quit attempt during the treatment period (before week 6). Of those who made a quit attempt after week 6 (n = 269), 16.0% (n = 43) reported achieving 7-day point-prevalence abstinence at week 26. There were multiple significant correlations between smoking-related constructs with small or medium effect sizes (Supplementary Table S2).

Table 1.

Mean (SD) Smoking-Related Constructs at Baseline and Week 6

	Baseline	Week 6
Cigarettes per day	17.4 (8.0)	10.8 (7.1)
Time to first cigarette	1.9 (0.8)	2.5 (1.1)
Motivation to quit	6.1 (2.3)	6.3 (2.5)
Quitting self-efficacy	3.5 (2.2)	4.6 (2.8)
Anticipated urges to smoke if quit	7.9 (2.0)	6.9 (2.2)
Positive affect	6.9 (2.0)	6.8 (2.1)
Negative affect	5.4 (2.2)	5.0 (2.3)
Time spent around others who smoke	1.5 (0.8)	1.8 (1.1)

Cigarettes per day were assessed by asking “On average, how many cigarettes have you smoked each day this week?” Time to first cigarette was assessed with the following item from the Fagerström Test for Cigarette Dependence¹⁷: “How soon after you wake up do you smoke your first cigarette? 1=Within 5 minutes, 2=6-30 minutes, 3=31-60 minutes, or 4=After 60 minutes.” Motivation to quit was assessed by asking “How motivated are you to quit smoking? (1-10 scale).” Quitting self-efficacy was assessed by asking “How confident are you that you can quit smoking successfully in the next month? (1-10 scale).” Anticipated urges if quit was assessed by asking “If you do try to quit smoking for good, how severe do you think your urges to smoke will be? (1-10 scale).” Positive Affect (PA) and Negative Affect (NA) were assessed with the Positive and Negative Affect Schedule (PA and NA scores on a 1–10 scale).^35,36 Time spent around others who smoke was assessed by asking “Over the past week, how often have you been with people while they are smoking? 1=Daily or almost daily, 2=A few times, 3=Rarely, 4=Never.”

Baseline Smoking-Related Constructs as Predictors of a Quit Attempt or Success

Models testing baseline smoking-related constructs separately showed that fewer CPD and longer time to first cigarette predicted making a quit attempt by 26 weeks (Table 2). Greater baseline quitting self-efficacy predicted successfully converting a quit attempt into abstinence by 26 weeks (ie, quitting success). In multivariable models including all smoking-related constructs, greater time to first cigarette was the only baseline predictor of making a quit attempt and no baseline smoking-related constructs predicted quitting success (Table 3).

Table 2.

Findings From Covariate-Adjusted Models Testing Smoking-Related Constructs Separately as Predictors of a Quit Attempt or Success

	Quit attempt OR (95% CI)	Quit success OR (95% CI)
Cigarettes per day
Baseline	0.97 (0.95, 0.996)	1.00 (0.96, 1.05)
Reduction: baseline to week 6	1.03 (0.99, 1.07)	1.10 (1.01, 1.19)
Time to first cigarette
Baseline	1.28 (1.03, 1.59)	1.07 (0.73, 1.55)
Increase: baseline to week 6	1.23 (1.04, 1.45)	0.99 (0.74, 1.35)
Motivation to quit
Baseline	1.07 (0.99, 1.15)	1.03 (0.89, 1.20)
Increase: baseline to week 6	1.11 (1.02, 1.20)	1.41 (1.16, 1.72)
Quitting self-efficacy
Baseline	1.06 (0.98, 1.15)	1.20 (1.04, 1.38)
Increase: baseline to week 6	1.13 (1.05, 1.21)	1.18 (1.03, 1.36)
Anticipated urge to smoke if quit
Baseline	0.94 (0.86, 1.02)	0.93 (0.79, 1.09)
Reduction: baseline to week 6	1.02 (0.93, 1.12)	1.23 (1.04, 1.47)
Positive affect
Baseline	1.08 (0.99, 1.18)	1.05 (0.89, 1.23)
Increase: baseline to week 6	1.07 (0.98, 1.18)	1.00 (0.83, 1.19)
Negative affect
Baseline	0.98 (0.91, 1.06)	0.93 (0.80, 1.09)
Reduction: baseline to week 6	1.00 (0.92, 1.09)	1.00 (0.85, 1.17)
Time spent around others who smoke
Baseline	1.07 (0.87, 1.33)	1.28 (0.88, 1.86)
Reduction: baseline to week 6	0.91 (0.73, 1.14)	1.57 (1.06, 2.30)

Quit attempt = self-reported intentional abstinence for more than 24 h after week 6. Quit success = self-reported 7-day point-prevalence abstinence at 26 weeks post-enrollment. OR = odds ratios, adjusted for assigned treatment condition, whether participants attended an additional 6 weeks of treatment, whether participants made a quit attempt during treatment between weeks 0 and 6, and baseline values of the smoking-related construct. Bold values represent significant findings (p < .05).

Table 3.

Findings From Multivariable Models That Included All Smoking-Related Constructs as Predictors of Quit Attempts and Success

	Quit attempt OR (95% CI)	Quit success OR (95% CI)
Cigarettes per day
Baseline	1.0 (0.95, 1.02)	0.99 (0.91, 1.07)
Reduction: baseline to week 6	0.99 (0.95, 1.04)	1.07 (0.97, 1.18)
Time to first cigarette
Baseline	1.60 (1.17, 2.17)	1.07 (0.63, 1.82)
Increase: baseline to week 6	1.27 (1.06, 1.53)	0.89 (0.63, 1.25)
Motivation to quit
Baseline	1.04 (0.93, 1.17)	1.19 (0.91, 1.55)
Increase: baseline to week 6	1.05 (0.95, 1.17)	1.36 (1.06, 1.76)
Quitting self-efficacy
Baseline	1.10 (0.98, 1.26)	1.10 (0.88, 1.37)
Increase: baseline to week 6	1.14 (1.03, 1.26)	0.96 (0.80, 1.16)
Anticipated urge to smoke if quit
Baseline	0.98 (0.87, 1.11)	0.87 (0.70, 1.10)
Reduction: baseline to week 6	0.97 (0.87, 1.08)	1.15 (0.93, 1.43)
Positive affect
Baseline	1.10 (0.97, 1.25)	0.98 (0.76, 1.26)
Increase: baseline to week 6	1.07 (0.95, 1.20)	0.92 (0.73, 1.15)
Negative affect
Baseline	1.04 (0.93, 1.17)	0.97 (0.78, 1.22)
Reduction: baseline to week 6	0.96 (0.87, 1.06)	0.97 (0.81, 1.17)
Time spent around others who smoke
Baseline	1.05 (0.83, 1.34)	1.42 (0.90, 2.24)
Reduction: baseline to week 6	0.86 (0.67, 1.09)	1.48 (0.97, 2.25)

Quit attempt = self-reported intentional abstinence for more than 24 h after week 6. Quit success = self-reported 7-day point-prevalence abstinence at 26 weeks post-enrolment. OR = odds ratios from multivariable models including all baseline and change scores for the smoking-related constructs as well as variables to adjust for assigned treatment condition, whether participants attended an additional 6 weeks of treatment, and whether a quit attempt was made during treatment (ie, before week 6). Bold values represent significant findings (p < .05).

Change in Smoking-Related Constructs as Predictors of a Quit Attempt or Success

Models testing the change in smoking-related constructs separately showed that increased time to first cigarette, motivation to quit, and quitting self-efficacy predicted making a quit attempt by 26 weeks (Table 2). Reduced CPD, anticipating lower urges to smoke if quit, decreased time spent around others who smoke, and increased motivation to quit and quitting self-efficacy predicted abstinence at 26 weeks. In multivariable models including all smoking-related constructs, increased time to first cigarette and quitting self-efficacy remained significant predictors of making a quit attempt. Increased motivation to quit was the only predictor of abstinence at week 26 that remained significant (Table 3).

Sensitivity Analyses

Findings remained similar to primary outcomes in a series of sensitivity analyses where missing data for smoking-related constructs were treated as missing. Findings also remained similar when only baseline values were included as covariates in models testing change and when no covariates were included in models testing baseline values. In addition, we tested the multivariable models, which including all smoking-related constructs, when missing data were treated as missing. Findings from these sensitivity analyses were similar except that baseline positive and negative affect predicted making a quit attempt and change in motivation to quit was no longer a significant predictor of abstinence. Finally, when quit attempts that occurred at any point between baseline and week 26 (including during treatment) were included, most primary findings remained significant plus additional smoking-related constructs became significant (Supplementary Tables S3 and S4). Importantly, change in smoking-related constructs and quitting occurred during the same time period for 19.3% (n = 100) participants in this sensitivity analysis, and thus the additional significant findings may be due in part to quit attempts causing a change in smoking-related constructs rather than vice versa.

Discussion

Findings from this secondary analysis showed that different smoking-related constructs predict quit attempts versus quitting success among people initially unwilling (ie, not ready, not motivated, or not able) to make an attempt to quit smoking. Our primary findings were that greater time to first cigarette at baseline, as well as increases in time to first cigarette and quitting self-efficacy, predicted initiating a quit attempt above and beyond other smoking-related measures. Increases in motivation to quit predicted quitting success among those who made a quit attempt above and beyond all other smoking-related measures. In addition, our individual tests of smoking-related constructs revealed statistically significant predictive relations between variables measured at baseline and making quit attempts (greater baseline time to first cigarette, fewer cigarettes smoked per day at baseline) and succeeding in those attempts (greater baseline quitting self-efficacy). Tests of individual predictors also yielded significant predictive relations between change in variables (from baseline to 6 weeks after treatment initiation) and making quit attempts (increased time to first cigarette, increased motivation to quit, increased quitting self-efficacy) and succeeding in those attempts (reduction in cigarettes smoked per day, increased motivation to quit, increased self-efficacy, reduced time spent around other smokers, and decreased anticipated urges to smoke upon quitting).

Our findings that baseline and increases in time to first cigarette, an important measure of cigarette dependence,³⁸ predicted making a quit attempt while increases in motivation predicted quitting success appear to differ from some prior findings¹¹ (also see Refs ^12,19). A review by Vangeli et al.¹¹ concluded that baseline factors associated with motivation predicted making a quit attempt while lower dependence predicted quitting success in population samples. Differences may be due, in part, to the fact that we tested change scores while much of the prior research was limited to baseline smoking characteristics.^11,12,19 In addition, differences could be due to sample differences, length of follow-up, or exposure versus nonexposure to motivation-phase interventions. In our research, we used a sample of participants initially unwilling to quit but willing to reduce smoking. We tested both baseline and change in smoking-related constructs as predictors of outcomes at a 26-week follow-up. Our findings are partly consistent with some prior tests of smoking-related constructs among similar populations with a similar follow-up period. For instance, one trial found that decreased dependence, increased motivation to quit, and increased quitting self-efficacy predicted making a quit attempt.¹⁴ Another study found that increased motivation to quit predicted making a quit attempt (although quitting self-efficacy did not).⁴ Neither study tested quitting success among only those who attempted to quit.

Consistent with the Phase-Based Model of Cessation,⁸ our findings show some differences in predictors of the two outcomes. For instance, multivariable models show that quit attempt likelihood was predicted by baseline time to first cigarette and increases in time to first cigarette and self-efficacy. However, these variables did not predict quitting success above and beyond other constructs. Instead, multivariable models showed that quitting success was predicted by increases in motivation to quit, which did not predict quit attempts. Again, such results conflict with some of the literature cited earlier in which motivation was especially predictive of quit attempts¹¹ while quitting success was more highly associated with urges^20,23 and time spent with smokers.^30,31 However, the single variable models yielded results more in line with prior research; for example, decreases in contact with smokers and anticipated urges strongly predicted quitting success but not quit attempts. These results highlight the need for additional research in smokers initially unwilling to quit; reliable findings could be used to guide phase-based intervention. Thus, time to first cigarette and quitting self-efficacy may be important targets for interventions designed to motivate quit attempts (Motivation-phase intervention). Motivation to quit, decreased time spent around other smokers, and decreased anticipation that urges will be severe while quitting may be important targets for Cessation-phase intervention.

Some of the predictive relations that remained significant in our multivariable models were sufficiently large so as to encourage further study. For example, after accounting for other smoking-related constructs, one-unit increases in time to first cigarette and quitting self-efficacy were associated with 27% and 14% increases in the odds of attempting to quit among those who were initially unwilling to make a quit attempt. A one-unit increase in motivation to quit was associated with a 36% increase in the odds of quitting success at week 26. Perhaps most impressively, a one-unit higher baseline score in time to first cigarette was associated with a 60% greater odds of initiating a quit attempt. Information on baseline characteristics could be used so that interventions are personalized to focus on an individual’s greatest needs.

Limitations

We relied on self-report to assess quit attempts and abstinence. We focused on quit attempts that occurred after week 6 in order to achieve temporal priority of change variables and the quitting outcomes. Thus, the number of participants who made a quit attempt in our primary analyses differs from that reported in a prior paper.⁹ Finally, because our findings differ somewhat from those obtained in some prior studies (eg, Refs ^11,12,19), attempts at replication are warranted.

Conclusions

This secondary analysis identified smoking-related measures associated with quit attempts and quitting success in a sample of patients who were unwilling to try to quit smoking at baseline. Greater time to first cigarette was the only baseline construct that predicted initiating a quit attempt above and beyond other smoking-related measures. In terms of change in smoking-related constructs, increases in time to first cigarette and quitting self-efficacy predicted initiating a quit attempt while increases in motivation to quit predicted quitting success above and beyond all other smoking-related measures. Research is needed to further test such findings. Once reliable predictors are identified, researchers should develop Motivation-phase interventions that meaningfully target such variables and then determine whether treatment-induced change translates into superior transition across the later phases of smoking behavior change.

Supplementary Material

A Contributorship Form detailing each author’s specific involvement with this content, as well as any supplementary data, is available online at https://academic.oup.com/ntr.

Funding

NIDA Institutional Training (grant T32DA007242) and National Cancer Institute (grant numbers 9P50CA143188 and 1P01CA180945).

Declaration of Interests

EMK, RM, TBB, MCC, MEP, TRS, DEJ, LMC, and JWC have nothing to disclose. JRH has received consulting and speaking fees from several companies that develop or market pharmacological and behavioral treatments for smoking cessation or harm reduction and from several nonprofit organizations that promote tobacco control. He also consults for Altria, Philip Morris, and Swedish Match on their harm-reduction products.