Participants are provided with 8 weeks of nicotine patches and nicotine lozenges, with dosages depending on the number of cigarettes per day and time to first cigarette upon waking. Participants receive 4 weeks of nicotine patches and lozenges at baseline, and they are mailed the additional 4 weeks of NRT when needed.

All participants are given evidence-based self-help materials for smoking cessation (based on the Treating Tobacco Use and Dependence Clinical Practice Guideline) [48]. Materials include the “Clearing the Air” booklet published by the National Cancer Institute and a referral to the Tobacco Cessation Quitline (1-800-QUIT-NOW).

Participants in the MBAT and MBAT+iQuit Mindfully conditions receive videoconference group counseling (using the Zoom platform) based on the MBAT manual (D Wetter, unpublished data, June 2009). Instructors are certified MBSR teachers and licensed clinicians who received additional training on tobacco cessation. MBAT consists of 8 weekly 2-hour sessions. Participants are encouraged to choose their own quit date between day 7 and day 30 of the program. MBAT sessions teach participants to notice the tendency for mindlessness (“automatic pilot”) and focus attention on the present moment, including observing sensations of craving and difficult emotions, so that they can choose to cope in adaptive ways other than smoking. MBAT emphasizes daily mindfulness practice in several forms including sitting meditation, body scan meditation, walking meditation, eating meditation, and gentle yoga. The program also teaches cognitive behavioral strategies for smoking cessation, including clearing the environment of smoking cues, recognizing and coping with high-risk situations, and managing stress [11].

Several strategies are being implemented to promote engagement with telehealth-delivered MBAT among adults of low-income populations. For example, participants in the MBAT and MBAT+iQuit conditions can borrow a study tablet for the 8 weeks of the program. Participants are also given the option to join videoconference groups in an individual room at our research office, in cases in which they do not have a private place to join groups. Finally, the research team provides guidance and technical support for using Zoom through an in-person baseline session, Zoom group orientation session, and ongoing troubleshooting as needed.

Participants receive text messages throughout the 8-week treatment period as well as less frequent messages during the follow-up period, depending on their preferences. All messages are automated using the Upland Mobile Commons platform. Participants are asked to set a quit date at baseline, which is integrated into text messages and can be changed by the participant. Text messages encourage participants to practice mindfulness and other strategies for smoking cessation (for more details, refer to the study by Spears et al [29]). The messages are designed to be interactive. That is, participants are asked questions through a series of flow logic (eg, “Would you like to practice mindfulness right now?” If participants replies “yes,” they are provided a mindfulness technique, later asked about how it went, and are encouraged to continue practice). Participants can also text CRAVE, STRESS, SLIP, or FACT at any point to receive additional text message support for coping with cravings, stress, smoking lapses, or to receive facts about the effects of smoking, respectively. Participants can answer “group poll” questions so that they later receive a text with the most common (deidentified) responses. Text messages include quotes sharing the anonymous experiences of former smokers in our program. Participants can also text keywords (MIND, BODY, or 3MIN) to receive a phone call with a short recording of a mindfulness practice. Participants are given small pocket cards with basic information on the text messaging program and reminders about the text keywords.

Text messages are personalized based on first names, personal reasons for quitting, and the amount of money to be saved based on individual smoking habits and price paid per pack. Picture messages are included based on our initial qualitative work [30]. On the basis of the feedback from our previous message testing [29], message timing and frequency are flexible and personalized. Participants choose the frequency of their choice (ie, very low to very high, ranging from 1 to 2 to 5 to 6/day) as well as a 12-hour time slot of their choice (7 AM to 7 PM or 10 AM to 10 PM). Text messages periodically ask participants about their preferred text message frequency and timing so that participants can change their message schedule as needed.

Individuals who are interested in the study either call the research office to complete telephone screening or click on a link to take a web-based screener survey. Eligible individuals are then scheduled for an in-person session to finalize eligibility, engage in further discussion about the study, and provide written informed consent. To confirm eligibility, the participants provide a breath sample for the assessment of expired CO. Individuals who decline or are ineligible are given self-help materials and referred to other cessation programs. In addition, potential participants complete the Patient Health Questionnaire-2 [49] and Severity of Dependence Scale [50] to screen for clinically significant depressive symptoms and substance dependence, respectively. Individuals who are ineligible for these reasons are provided with appropriate mental health referrals in addition to smoking cessation referrals.

Eligible individuals who provide written informed consent are then asked to complete baseline questionnaires, followed by randomization. Permutated block randomization is implemented, with stratification by race and poverty status. Participants are provided with information and materials specific to their intervention condition and scheduled for their next study visits. Web-based surveys are administered at weeks 1, 3, and 5. In addition, electronic diary assessments (brief surveys sent via text message or email depending on participant preference) are administered in the evening on every other day from week 2 to week 8, thus capturing key processes surrounding quitting smoking. Diary assessments assess smoking behavior (“How many cigarettes did you smoke today?”) and each of the hypothesized mechanisms (mindfulness, emotions, craving and withdrawal, weakened associations of negative affect and craving with smoking, self-efficacy, and social support). In-person assessments take place at weeks 8, 12, and 24, including assessment of expired CO (all in-person visits) and salivary cotinine (week 24) for biochemical confirmation of smoking behavior. The descriptions of the questionnaires are provided below.

Tobacco History assesses the onset of regular smoking, previous quit attempts, abstinence history, smoking rate, and partner smoking status. This includes the Heaviness of Smoking Index (HSI), which comprises two items from the Fagerström Test for Nicotine Dependence [51]: self-reported average number of cigarettes smoked per day and time to first cigarette upon waking (“time to first cigarette”). The HSI is a strong indicator of nicotine dependence [52]. Smoking status surveys tobacco use, use of other tobacco products, and nicotine replacement medications. Smoking abstinence is assessed according to the Society for Research on Nicotine and Tobacco guidelines [53]. The use of novel tobacco products (eg, e-cigarettes) is also assessed, and participants indicate which forms of NRT and/or pharmacotherapy (if any) they used in the past week. Additional resources used for smoking cessation assesses the use of various strategies for quitting smoking (eg, acupuncture, hypnosis, other text messaging programs, or mobile apps) outside of the treatment provided by the study.

The Brief Wisconsin Dependence Motives Questionnaire [54] is a 37-item measure that yields an overall dependence score and subscale scores for other dimensions (ie, cognitive enhancement, affective enhancement, automaticity, affiliative attachment, loss of control, craving, cue exposure or associative processes, social or environmental goals, taste or sensory processes, weight control, and tolerance). The Wisconsin Smoking Withdrawal Scale [55] includes subscales for anger, anxiety, sadness, concentration difficulty, craving, hunger, and sleep. The Self-Efficacy Scale assesses confidence in resisting smoking urges in specific situations (eg, when feeling stressed or when with friends) [56]. Subscales include negative affect, pleasure, social image, social influence, and diet.

The Mindful Attention Awareness Scale [57] is a 15-item self-report measure of dispositional mindfulness [57]. The Five-Facet Mindfulness Questionnaire-Short Form [58] is a 24-item self-report questionnaire on facets of dispositional mindfulness (nonreactivity, observing, acting with awareness, describing or labeling with words, and nonjudging of experience). Participants also complete a Mindfulness Practice Log [11] to indicate how frequently they practice each mindfulness technique taught in treatment. The Self-Compassion Scale-Short Form [59] is a 12-item self-report measure of self-compassion. Subscales include self-kindness, self-judgment, common humanity, isolation, mindfulness, and overidentification.

The Perceived Stress Scale [60] is a 10-item self-report measure of the extent to which individuals view their lives as stressful. The Positive and Negative Affect Schedule [61] is a 20-item self-report measure of affective experience, yielding two factors (positive and negative affect).

The Multidimensional Scale of Perceived Social Support [62] is a 12-item scale, with subscales assessing perceived social support from family, friends, and significant others. The Group Climate Questionnaire [63] is a 12-item self-report scale that measures the climate and cohesion of the group (only applicable for MBAT participants).

Participants provide feedback about their experiences receiving the text messages (text message feedback). They also complete 4 items to indicate perceived benefits of the smoking cessation program. As described by Hoeppner et al [64], participants rate the extent to which the program gave them confidence to quit smoking, made them think it was worthwhile to quit, made them feel that someone cared if they quit, and made them feel that they knew the right steps to take to quit. Items are rated from 1 (completely disagree) to 5 (completely agree). Participants also complete program evaluations to provide feedback and suggestions for improving MBAT and iQuit Mindfully. Usual care participants answer one program evaluation question: “On the scale below, please circle the number that best represents whether you would recommend this quit smoking program (or something similar) for others who are interested in quitting smoking” (rated on a 1 to 10 scale).

Participants receive financial compensation for the time and inconvenience associated with participation as well as parking validation or train or bus vouchers to defray the cost of travel. Participants are compensated up to US $30 for web-based surveys at weeks 1, 3, and 5 (US $25 for each survey plus US $5 bonus if completed within 24 hours). They are also compensated US $40 for week 8, US $60 for week 12, and US $70 for week 24 (all in-person visits). In addition, participants are compensated US $5 per each of the every-other-day surveys (maximum US $120 for all 6 weeks). The maximum compensation per person for completing all aspects of the study is US $380.

Other procedures to increase adherence include the following: (1) reminder phone calls, e-mails, and text messages; (2) requiring a phone number and valid home address so that participants can be contacted; (3) loaning participants a mobile phone for the study duration if they do not have one; and (4) obtaining names, addresses, and phone numbers of up to 3 collaterals who can provide information on participants’ whereabouts if necessary. In addition to traditional procedures for the informed consent process, research staff members engage eligible individuals in a discussion of the pros and cons of participating as well as participants’ ambivalence about behavior change and research participation. These methods have been suggested to improve retention [65].

We will conduct both intent-to-treat (ITT) and per-protocol analyses. ITT includes all randomized participants, regardless of compliance, withdrawal, and other events after randomization. A strength of ITT is that it is based on the original randomization. However, effect estimation using ITT may be conservative and misleading with increasing attrition. Per-protocol analysis considers only participants who fully complied and completed the study. Per-protocol analysis is less conservative and may reflect true treatment differences for patients with full compliance. Including both approaches will provide a more complete understanding of the treatment effects.

Analyses will be performed using the SAS software. All analyses will control for demographics (age, gender, race, education level, and partner status) and baseline nicotine dependence (cigarettes per day and time to first cigarette).

The relationship between the treatment condition and each dichotomous study outcome (smoking cessation and lapse recovery) will be assessed using a generalized linear mixed model with a binomial distribution and logit link function. This is an appropriate statistical model with repeated measures data collected over time [66]. A 2-level multilevel generalized linear mixed model will be specified, with time nested within the subject. This approach will properly account for the multilevel data structure. The fixed effects considered in each model will include demographics, baseline nicotine dependence, treatment condition and time as main effects, and a treatment condition by time interaction. Testing of the covariance structure will be conducted, and information criteria will be used to assess model fit and selection. Linear contrasts will be built into the analysis to estimate the specific comparative effects of all treatment conditions. To test treatment effects on primary smoking cessation outcomes, repeated measures outcomes will be biochemically confirmed at 7-day point prevalence abstinence at weeks 8, 12, and 24. To test the effects of treatment on lapse recovery, this analysis will also be conducted among participants classified as smoking at the last treatment session [11]. In addition, gender, race and ethnicity will be assessed as potential moderators of the treatment effects.

The moderating effect of poverty status will be examined with the addition of a poverty status by treatment condition interaction into the models described above. This will be tested using both dichotomous (ie, below vs at or above the federal poverty threshold) and continuous poverty status variables (ie, depth of poverty as indicated by the ratio of income to poverty and income deficit or surplus) according to the United States Census Guidelines [67,68].

Hypotheses regarding the mechanisms will be tested separately using both traditional questionnaire data (collected from baseline to 24 weeks after the start of treatment) and electronic diary data (collected for 6 continuous weeks during treatment). Mediation testing will be conducted regardless of the statistical significance of the intervention effects because of statistical, conceptual, and practical reasons for testing mediation in such cases [69]. Volatility will be calculated using the mean square successive difference approach [70] to capture both within-person variability and temporal instability. Mediation will be examined using multilevel mediation analyses for binary outcomes in the developed generalized linear mixed model. Bootstrapping with 5000 replications will be used to estimate the standard errors and P values for indirect effects. In addition to the mediators of associations between treatment and abstinence outcomes, we will test more fine-grained day-to-day associations using electronic diary assessments. We will test moderated mediation (ie, the hypothesis that mindfulness not only reduces stress, negative affect, and craving but also weakens the associations between these variables and smoking) using a model proposed by Preacher et al [71], as the independent variable (mindfulness training) also functions as a moderator. Kline [72] termed this “second-stage moderation,” as the second path of the indirect effect of the predictor on the dependent variable depends on the moderator. Conditional mediation and moderation effects will be tested using bootstrapping methods [71].

Although there is no complete, comprehensive method for handling missing data for generalized linear mixed models, several different approaches can be used to address this issue [66]. We will use multiple imputations of missing data whereby missing observations for individuals are estimated based on baseline responses and other study covariates, and we will perform additional sensitivity analyses regarding various missing data assumptions for smoking outcomes [73].

The sample size was estimated based on the number of participants required to detect meaningful treatment effects of the iQuit Mindfully and MBAT interventions. We obtained effect sizes (converted to Cohen d for comparability across measures) from relevant meta-analyses and used the Optimal Design program to estimate the sample size for the generalized linear mixed models [74]. Smoking abstinence in MBIs versus usual care was quantified as Cohen d=0.35 in a meta-analysis [12], and a Cochrane review [75] of primarily text messaging–based interventions for smoking cessation concluded an effect size of Cohen d=0.33. Assuming 80% power, a level of significance of .05, 2-tailed statistical tests, an error variance of 1, a coefficient variance of 2, and oversampling to account for 20% attrition, this results in 145 participants in each treatment group. An unequal allocation ratio was used to allocate fewer participants to usual care (n=50) [76,77] for an overall sample size of 485 participants (145 per treatment group and 50 in usual care). We expect this sample size to adequately allow for mediational testing (aim 2) at the 80% power level. Fritz and MacKinnon [78] offer a guide regarding the sample size required to detect mediated effects and report that 79.9% of studies in their literature review applied mediational testing with <400 participants, suggesting that our study plan with 485 participants will be ample to achieve all study aims.