Pilot Randomized Controlled Trial of Web-Delivered Acceptance and Commitment Therapy Versus Smokefree.gov for Smokers With Bipolar Disorder

Abstract

Introduction

Smokers with bipolar disorder (BD) are less successful at quitting than the general population. In this study, we evaluated in a pilot randomized controlled trial a novel, targeted, web-based intervention for smokers with BD based on acceptance and commitment therapy (ACT) and designed for reach and disseminability.

Aims and Methods

Daily smokers (n = 51) with bipolar I or II disorder were recruited from four US sites and randomly assigned to one of two web-based smoking cessation interventions—ACT-based WebQuit Plus (n = 25) or Smokefree.gov (n = 26) over a 10-week treatment period. All participants received nicotine patch for 8 weeks. Key outcomes were trial design feasibility, intervention acceptability, and cessation at end of treatment and 1-month follow-up.

Results

We screened 119 to enroll 51 participants (target sample size = 60) over 24 months. The most common reason for ineligibility was the inability to attend study appointments. Retention was 73% at end of treatment and 80% at follow-up, with no differences by arm. The mean number of logins was twice as high for WebQuit Plus (10.3 vs. 5.3). The usefulness of program skills was rated higher for WebQuit Plus (75% vs. 29%). Biochemically confirmed, 7-day abstinence at end of treatment was 12% in WebQuit Plus versus 8% in Smokefree.gov (odds ratio = 1.46, 95% confidence interval = 0.21 to 9.97). At follow-up, abstinence rates were 8% in both arms.

Conclusions

Trial design produced favorable retention rates, although alternative recruitment methods will be needed for a larger trial. At end of treatment, acceptability and estimated effect size of WebQuit Plus relative to Smokefree.gov were promising and support continued program refinement and evaluation.

Implications

In this first randomized controlled trial of a targeted intervention for smokers with BD, we found that the ACT-based WebQuit Plus intervention, delivered in combination with the nicotine patch, had promising acceptability and cessation outcomes relative to Smokefree.gov. The observed signals for acceptability and cessation suggest that the WebQuit Plus program should be refined based on participant feedback and evaluated in a larger trial. Feasibility findings from this study also provide direction for refining trial procedures to enhance the recruitment of smokers with BD.

Introduction

The prevalence of smoking among people with bipolar disorder (BD) is two to three times higher than in the general population,¹ leading to disproportionately high rates of tobacco-related disease and mortality among this group.^2,3 Additionally, quit rates for smokers with BD are 30%–40% lower than for smokers without psychiatric disorders.^4,5 Key barriers to improving quit rates are (1) unique challenges for smokers with BD that are not addressed in standard behavioral interventions, such as the common belief that smoking helps manage BD symptoms or its psychosocial sequelae^6,7 and (2) limited access to tobacco treatment within the mental health care system.^8–10 New approaches are greatly needed to address these barriers.

Regarding the unique challenges of smokers with BD, Acceptance and Commitment Therapy (ACT) provides a new treatment lens through which to conceptualize and address these challenges. The treatment goals in ACT are to cultivate awareness and acceptance, rather than avoidance, of emotional, physical, and cognitive cues to smoke and to foster commitment to value-guided action. Overall, ACT focuses on helping individuals pursue what is important to them (eg, physical health) by increasing willingness to experience affective (eg, low mood), physical (eg, strong urges, withdrawal symptoms), and cognitive (eg, belief that smoking helps manage symptoms of BD) triggers for smoking. We previously developed a targeted ACT intervention for smokers with BD and evaluated its acceptability and preliminary efficacy as both a face-to-face and telephone-delivered intervention, in combination with the nicotine patch.¹¹ Results showed high acceptability and a signal for efficacy.

Given this proof-of-concept evidence that smokers with BD responded well to the targeted ACT intervention, we translated the program into a web-based intervention, for several reasons. First, web-based interventions have high potential reach, as 85% of adults in the United States now use the Internet.¹² Because so many people are accessing health information and health behavior change assistance via the Internet,¹³ web-based interventions for smoking cessation are proliferating rapidly.^14,15 Second, web-based treatments have strong appeal to consumers,¹⁶ including those with serious mental illness.¹⁷ Third, the cost-effectiveness of web-based treatments is also greater than other treatment delivery models such as clinic-, workplace-, and telephone-based interventions.¹⁸ Low-cost treatment is particularly important in mental health settings where tobacco treatment may not be reimbursed.^9,10 Finally, the Internet is an ideal medium for making targeted interventions available to the subgroups of smokers for which they are designed through existing online communities (eg, Depression and Bipolar Support Alliance). Thus, web-based interventions have great potential to improve the accessibility of targeted interventions for smokers with BD and therefore increase population-level quit rates.

In this study, we compared, in a pilot randomized controlled trial, a targeted, web-based ACT intervention for smokers with BD (“WebQuit Plus”) to Smokefree.gov, both delivered with nicotine patches. One primary aim of the study was to determine feasible procedures for conducting a larger trial, with primary feasibility outcomes including recruitment (eg, number screened, eligible, and consented; reasons for ineligibility or refusal) and adherence to the trial protocol (eg, retention rates). The other primary aim was to compare WebQuit Plus to Smokefree.gov on treatment satisfaction and utilization, ACT’s theory-based mechanisms of change (ie, acceptance of smoking triggers, commitment to quitting), smoking cessation outcomes, and BD symptoms at the end of the 10-week treatment period and at 1-month posttreatment follow-up. We also report, for the WebQuit Plus arm only, usage of key intervention features and rate of program completion.

Methods

Participants

Participants (n = 51) were treatment-seeking adult smokers with BD. Inclusion criteria were (1) age at least 18 years, (2) daily smoker for the past year, (3) averages at least 5 cigarettes/day for the past 30 days, (4) expired-air carbon monoxide (CO) at least 4 ppm, (5) desire to quit smoking within 30 days, (6) meets lifetime criteria for bipolar I or II disorder, based on DSM-5 criteria, (7) has at least weekly access to high-speed Internet via a tablet, laptop, or desktop computer, as the site was designed to optimize usability on larger screens, and (8) being under the care of a clinician for the treatment of BD and willing to authorize communication between study staff and clinician regarding study participation and clinical deterioration. Exclusion criteria were (1) participating in other smoking interventions or taking FDA-approved medications for smoking cessation, (2) previously used the Smokefree.gov Web site, (3) meets current (past 30 days) DSM-5 criteria for mania, a major depressive episode with symptoms that are currently severe (PHQ-9 > 14), or a moderate or severe substance use disorder, (4) current (past 30 days) psychotic symptoms, (5) psychiatric hospitalization in past 30 days, (6) current suicidal or homicidal ideation, (7) unstable medical or psychiatric condition, and (8) conditions that would increase risk of nicotine patch use (pregnant, breastfeeding, recent cardiac events, and previous allergic reaction to patch).

Procedures

Participants were recruited at four sites: Jefferson Center for Mental Health in Wheat Ridge, CO; VA Palo Alto Health Care System and Stanford University in Palo Alto, CA; Edith Nourse Rogers Memorial VA Hospital in Bedford, MA; and VA Central Western Massachusetts Healthcare System in Northampton, MA. Trial recruitment methods included provider referral at all study sites. Three of the four sites conducted outreach by E-mail and phone to potentially eligible participants identified via electronic medical records. Additionally, flyers, brochures, and information sessions were presented at strategic locations at each site such as clinics, hospitals, support groups, and online forums. One site engaged peer specialists to recruit participants.

Potential participants were prescreened by phone to determine whether basic eligibility criteria were met. Eligible individuals after the prescreen were invited to attend a more comprehensive face-to-face screening where they were provided detailed study information and written informed consent was obtained following the guidelines of each site’s Institutional Review Board. Following informed consent, participants completed screening assessments (see Measures section). Those who screened ineligible were provided information about other resources to help them quit smoking. Participants who remained eligible following in-person screening were enrolled in the study.

Approximately 1 week following the in-person screening (Visit 1), enrolled participants were randomized and completed a baseline battery of assessments (Visit 2). At this visit, participants were encouraged to set a quit date close to Visit 3 (telephone contact) approximately 2 weeks later, and nicotine patches were dispensed, with instructions to begin using the patch on the quit date. Instructions for using the assigned web program were provided, and study staff assisted participants in logging in for the first time. A common introductory page following the log-in screen was included in both Web sites so that this did not unblind study staff to treatment assignment. Following Visit 2, participants were contacted by phone at 2-week intervals in order to (1) administer assessments of smoking, BD symptoms, adverse events (AEs), and concomitant treatment and (2) promote retention in the study. Visits at end of treatment (Visit 7, ~10 weeks post-randomization) and 1-month posttreatment follow-up (Visit 8, ~14 weeks post-randomization) were in person to allow for CO testing to confirm smoking abstinence. Participants were compensated up to $140 for completing all study visits. The trial was registered on ClinicalTrials.gov (NCT#02750904).

Although we considered testing the intervention using an entirely web-based design in order to maximize reach, as in our previous evaluation of WebQuit,¹⁹ we selected this clinic-based management plus web intervention for three major reasons: (1) the ability to carefully monitor safety issues that may arise in these smokers with BD, (2) the ability to validate self-reported psychiatric diagnoses using semi-structured assessment instruments administered face-to-face, and (3) because of the frequent contact and combination of telephone and in-person visits, this design was expected to reduce attrition compared to a web-only design, which is valuable for a proof-of-concept trial.

Web Interventions

Both interventions employed an engaging, multimedia format. Methods used in both treatment arms to promote engagement were (1) weekly E-mail reminders containing a link to the assigned program and (2) daily text messages for 70 days (with an option to adjust message frequency or end date as desired) with links to the program. Text message content differed between arms, consistent with the different treatment approaches (ie, ACT vs. standard care).

WebQuit Plus

The WebQuit Plus program was built on the foundation of the WebQuit program, the non-targeted components of which are described in a previous publication.¹⁹ Briefly, the program has four parts that help users: (1) make a quit plan, (2) develop awareness of smoking triggers, (3) develop acceptance-based coping skills to handle triggers, and (4) identify and engage personal values and self-compassion to support long-term abstinence. The structure of the Web site for these four parts, and the exercises within them, is such that they are completed in sequential order. The program also prompts users to track smoking, use of cessation medications, values-guided activities, and practice of ACT skills. To increase interactivity and personalization, tracking results are displayed on the main dashboard along with badges earned for completing actions within the program and user-uploaded inspirational photos. WebQuit Plus also contains an ask-the-expert feature that allows users to post questions and view responses to their own and others’ queries. These features are accessible anytime in any order, in contrast to the four-step program that requires sequential completion.

WebQuit Plus builds onto the content of the original WebQuit program by adding the following targeted components: (1) ACT exercises and psychoeducation to address specific challenges to smoking cessation for smokers with BD (eg, anxiety, depression, severe nicotine dependence and withdrawal, low social support for quitting, belief in benefits of smoking and risks of quitting for smokers with BD^7,20) which have shown strong preliminary evidence of acceptability and efficacy when delivered in individual therapy,¹¹ (2) “inspiring stories” (ie, testimonials) derived from our previous qualitative work,⁷ which describe how a person with BD was able to use program skills to overcome the common challenges to quitting,^6,7 (3) one-way text messages to promote adherence to nicotine replacement therapy (NRT), which has been suboptimal in prior studies involving smokers with BD,^11,21 and (4) two-way keyword messaging to request on-demand assistance with mood-specific triggers (ie, MOOD), as well as other common cessation challenges (ie, URGE, SLIP, MOTIVATE). In addition to these targeted components, WebQuit Plus also uses tracking data to display money and minutes of life saved via reducing or quitting smoking on the main dashboard. New content was evaluated in lab-based usability testing with a small sample of smokers with BD (n = 6) and refined prior to the trial.

Smokefree.gov

The National Cancer Institute’s Smokefree.gov Web site was chosen as the comparison intervention. Smokefree.gov is a multimedia Web site that includes (1) guidance on setting a quit date, preparing to quit, identifying and coping with triggers, and staying motivated; (2) interactive content, such as screening questionnaires for depression and nicotine dependence; and (3) information about the health effects of smoking, presented in text and graphic form. Although Smokefree.gov does not follow the same structured, sequential flow of intervention content as in WebQuit Plus, we chose it as the comparator because it matches the multimedia, self-guided format of WebQuit Plus and its content is evidence-based, consistent with current clinical practice guidelines,²²and substantially different from the ACT-based content we aimed to evaluate in terms of its underlying theory and treatment approach. Given its status as the most-used cessation Web site in the world (Erik Augustson, Jan 29, 2014, personal communication), Smokefree.gov constitutes standard care for a web-based intervention. Finally, because it is in the public domain, a copy of the site could be hosted internally by the Fred Hutchinson Cancer Research Center. This was critical since internal hosting of Smokefree.gov allowed for the collection of utilization data for both Web sites, which would not be possible if participants were directed to the National Cancer Institute Web site to access it. Internal hosting also prevented changes to the content of the site during the study.

Nicotine Replacement Therapy

We provided 8 weeks of open-label NRT (nicotine patch) to study participants for two major scientific and ethical reasons: (1) the current standard of care is the combination of medication and behavioral treatment²² and (2) providing standardized medication reduces a major source of variability in treatment outcomes, making it easier to detect a signal for impact of a new behavioral intervention. While there are more effective medication options available, the nicotine patch is the most commonly used cessation medication approach in the United States²³ and is more broadly accessible than prescription medications—important considerations for population-level interventions. Following USDHHS Clinical Practice Guidelines²² for NRT dosing, participants smoking at least 10 cigarettes/day received 4 weeks of 21 mg nicotine patch, 2 weeks of 14 mg patch, and 2 weeks of 7 mg patch. Those smoking 5–9 cigarettes/day received 6 weeks of 14 mg patch and 2 weeks of 7 mg patch.

Assessments

Diagnostic/Screening Assessments

The Structured Clinical Interview for DSM-5, Patient Version (SCID-I/P) is a diagnostic interview for major psychiatric disorders.²⁴ The SCID was administered by a trained interviewer and reviewed by the supervising clinician to verify the diagnosis of BD and assess current substance use disorders.

Smoking Assessments

Baseline smoking was assessed using the (1) Fagerström Test for Nicotine Dependence,²⁵ a six-item, self-report scale that assesses the degree of physical dependence on nicotine and (2) smoking Timeline Followback,²⁶ a calendar-based method of obtaining retrospective estimates of daily smoking. For screening purposes, the prior 30 days was used to establish eligibility (ie, averaging ≥5 cigarettes/day). Expired-air CO was measured at baseline, end of treatment, and 1-month follow-up to confirm eligibility and verify self-reported abstinence (cutoff ≤ 4 ppm). The primary smoking cessation outcomes were 7-day point prevalence abstinence (PPA) (no smoking, not even a puff) at end of treatment and 1-month posttreatment. Secondary outcomes included cessation of all nonmedicinal nicotine and tobacco products and a 50% reduction in CO level.

Mechanism of Change Assessment

The 27-item Avoidance and Inflexibility Scale (AIS)²⁷ was administered at baseline and end of treatment to measure changes in acceptance of smoking triggers, based on ACT’s theory-based mechanism of change. Commitment was assessed using the Commitment to Quitting Scale (CQS).²⁸

BD Symptom Assessments

Assessments of BD symptoms were conducted at screening to determine eligibility and at each visit thereafter to assess symptom changes. The Altman Self-Rating Mania Scale²⁹ and the Patient Health Questionnaire-9 item version (PHQ-9)³⁰ were used to assess symptoms of mania and depression, respectively.

Treatment Satisfaction and Utilization

Treatment satisfaction was assessed using a 12-item survey used in our prior work. Sample items include “Overall, how satisfied are you with your assigned website?” and “How useful was your assigned website’s quit plan?” Responses were on a 5-point scale (1 = “not at all” to 5 = “very much”) and, for ease of interpretation, recoded into a binary (yes/no) variable, with “yes” defined as a response of 3 (“somewhat”) or higher. Web site utilization data were server-recorded page views to assess the number of logins, number of unique days of use, and total duration of use from first to the last login. For the WebQuit Plus arm, we also calculated completion rates and usage of program components. NRT utilization was self-reported at the end of the NRT treatment period.

Safety

AEs were assessed at each visit by asking the participant, “How have you been feeling since the last visit?” For each reported event, the start and end dates, severity, relatedness to study treatment, NRT action taken (eg, discontinuation of medication), outcome, and whether study withdrawal was required were recorded in an AE log.

Statistical Power and Analyses

This study was not designed for power to detect statistically significant differences by treatment group. The NIDA Stage Model of behavioral treatment development³¹ suggests that pilot treatment development trials should include approximately 15–30 participants per arm in order to test feasibility. A sample size of 60 (n = 30 randomized to each intervention) was chosen on the basis of this recommendation.

Recruitment and overall retention outcomes are presented descriptively. We used generalized linear mixed effects models to evaluate differences between study arms on smoking cessation, as well as binary satisfaction outcomes. Linear mixed effects models were used to assess changes in acceptance and commitment and mental health symptoms, adjusting for the baseline value of each outcome. Negative binomial models were used to test differences between study arms on utilization count outcomes, which were heavily right-skewed. All models adjusted for the two variables used in stratified randomization: heavy smoking (ie, average ≥20 cigarettes/day during 30 days prior to baseline), which was included as a fixed effect and recruitment site, which was included as a random effect. Statistical tests were two-sided, with α = 0.05, and analyses were completed using R version 3.4.2 and R packages MASS and lme4.^32–34

Results

Trial Design Feasibility

The study recruitment period was 24 months. Across sites, the recruitment methods with the highest yield of participants eligible based on phone prescreen were staff referral (n = 23), outreach by telephone or letter to potentially eligible patients identified via the electronic medical record (n = 22), and placement of flyers in clinic and community settings (n = 14). The yield of any specific recruitment method varied considerably by site. As shown in the CONSORT diagram (see Figure 1), we prescreened 119 individuals, of whom 56 were potentially eligible and 51 were randomized (28 men and 23 women). The most common reasons for ineligibility across both screening phases were inability to make study appointments (n = 11), missing the window to complete in-person screening (n = 9), smoking too few cigarettes (n = 7), and not meeting criteria for BD (n = 6).

Figure 1.

WebQuit Plus participant flow diagram.

Baseline characteristics of participants randomized to WebQuit Plus (n = 25) and Smokefree.gov (n = 26) are given in Table 1, along with the characteristics of the total sample. Of note, 24% of the sample identified as racial/ethnic minority, 49% reported a household income of not more than $20 000/year, and 57% reported receiving disability assistance. The majority (88%) met criteria for bipolar I disorder and half (51%) had a lifetime history of a substance use disorder. Regarding smoking, the average Fagerström Test for Nicotine Dependence score was 6.2 (SD = 2.1), indicative of high nicotine dependence, and the average number of cigarettes/day was 19.1 (SD = 8.2). One-third lived with at least one adult smoker. There were no significant differences between arms on any baseline characteristic (all p-values > .05).

Table 1.

Baseline Characteristics, Overall and by Treatment Arm

	Total	Smokefree.gov	WebQuit Plus
	(n = 51)	(n = 26)	(n = 25)	p
Demographics, M (SD) or n (%)
Age, M (SD)	49.0 (10.8)	50.2 (11.3)	47.7 (10.4)	.427
Male gender	28 (55%)	11 (42%)	17 (68%)	.118
Racial or ethnic minority	12 (24%)	8 (31%)	4 (16%)	.361
Married/partnered	9 (18%)	5 (19%)	4 (16%)	>.99
Sexual or gender minority	10 (20%)	7 (27%)	3 (12%)	.323
Working	16 (31%)	8 (31%)	8 (32%)	>.99
High school or less education	12 (24%)	6 (23%)	6 (24%)	>.99
Low income (household income <$20 000)	25 (49%)	14 (54%)	11 (44%)	.672
Receiving disability assistance	29 (57%)	12 (46%)	17 (68%)	.196
Mental health, M (SD) or n (%)
Bipolar I disorder	45 (88%)	23 (88%)	22 (88%)	>.99
ASRM score, M (SD)	9.1 (3.7)	9.4 (3.9)	8.9 (3.4)	.652
Depression screen positive (PHQ-9 ≥ 10)	28 (55%)	15 (58%)	13 (52%)	.899
Substance use disorder, lifetime	26 (51%)	13 (50%)	13 (52%)	>.99
Smoking behavior, M (SD) or n (%)
FTND score, M (SD)	6.2 (2.1)	6.2 (2.2)	6.3 (2.1)	.783
Cigarettes/day	19.1 (8.2)	17.6 (7.9)	20.6 (8.4)	.190
Used e-cigarettes in past month	3 (6%)	3 (12%)	0 (0%)	.248
Used other tobacco in past month	3 (6%)	2 (8%)	1 (4%)	>0.99
CO level, M (SD)	23.1 (15.2)	22.9 (13.2)	23.4 (17.3)	.913
Smoking in household, n (%)
Live with another adult who smokes	17 (33%)	8 (31%)	9 (36%)	.921
Live with partner who smokes (n = 50)	11 (22%)	5 (19%)	6 (25%)	.881
ACT theory-based measures, M (SD)
AIS (acceptance)	2.9 (0.4)	2.9 (0.4)	2.9 (0.4)	.941
CQS (commitment)	3.9 (0.6)	3.9 (0.6)	3.9 (0.6)	.750
Internet use, n (%)
Daily use	42 (82%)	20 (77%)	22 (88%)	.503

ASRM = Altman Self-Rating Mania Scale; PHQ-9 = Patient Health Questionnaire-9; FTND = Fageström Test for Nicotine Dependence; CO = carbon monoxide; AIS = Avoidance and Inflexibility Scale; CQS = Commitment to Quitting Smoking Scale.

Study retention was 73% (37/51) for the end-of-treatment visit and 80% (41/51) at 1-month follow-up, with no differences by treatment arm (p = .693 for end of treatment; p = .578 for follow-up). The average number of study visits completed (of 8) was 7.1 (SD = 1.5) for WebQuit Plus and 6.6 (SD = 2.2) for Smokefree.gov (p = .340).

Intervention Outcomes

Acceptability

As given in Table 2, WebQuit Plus participants reported descriptively higher satisfaction with the content of their assigned Web site than Smokefree.gov participants in on all five content-related comparisons, with one reaching statistical significance (ie, 75% of WebQuit Plus vs. 29% of Smokefree.gov participants reported that the program skills were useful, p = .015). Smokefree.gov was rated descriptively higher on being well-organized (94% vs. 82%, p = .306) and having sections of the site easily accessible (94% vs. 76%, p = .201), although satisfaction ratings in this area were generally high for both groups. Utilization was descriptively higher for WebQuit Plus across all indicators, including number of logins (10.3 vs. 5.3, p = .069), number of unique login days (9.2 vs. 4.9, p = .075), and total number of days in use from first to last login (28.0 vs. 23.7, p = .665).

Table 2.

Treatment Acceptability, Efficacy for Cessation, Mechanisms of Change, and Change in Psychiatric Symptoms

Outcome variable	Smokefree.gov n = 26	WebQuit Plus n = 25	OR or IRR (95% CI)	p
Satisfaction, n (%)
Satisfied with program	10 (56%), n = 18	16 (84%), n = 19	4.25 (0.91, 19.93)	.066
Would recommend to a friend	11 (55%), n = 20	14 (67%), n = 21	1.74 (0.47, 6.41)	.407
Skills were useful	4 (29%), n = 14	12 (75%), n = 16	7.48 (1.48, 37.82)	.015
Plans for quitting were useful	6 (43%), n = 14	11 (69%), n = 16	2.94 (0.65, 13.32)	.161
Support forum was useful	5 (45%), n = 11	9 (69%), n = 13	2.71 (0.51, 14.48)	.243
Web site was well-organized	16 (94%), n = 17	14 (82%), n = 17	0.29 (0.03, 3.12)	.306
Easy to access Web site sections	15 (94%), n = 16	13 (76%), n = 17	0.22 (0.02, 2.25)	.201
Web site was made for me	9 (53%), n = 17	10 (59%), n = 17	1.27 (0.33, 4.96)	.728
Utilization during the 10-week treatment period (prior to Visit 7), M (SD)
Number of log-ins	5.3 (11.3)	10.3 (14.2)	2.00 (0.95, 4.24)	.069
Number of days logged in	4.9 (10.5)	9.2 (12.3)	1.95 (0.94, 4.06)	.075
Time in days from first to last use	23.7 (26.1)	28.0 (26.9)	1.21 (0.52, 2.82)	.665
Primary cessation outcomes (CO-confirmed, missing = smoking)
7-day PPA from smoking, EOT	2 (8%)	3 (12%)	1.46 (0.21, 9.97)	.701
7-day PPA from smoking, 1-month FU	2 (8%)	2 (8%)	0.96 (0.12, 7.57)	.967
Other cessation outcomes (CO-confirmed, missing = smoking)
7-day PPA from all nicotine/tobacco, EOT	2 (8%)	3 (12%)	1.46 (0.21, 9.97)	.701
7-day PPA from all nicotine/tobacco, 1-month FU	2 (8%)	2 (8%)	0.96 (0.12, 7.57)	.967
50% reduction in CO level, EOT	4 (15%)	6 (24%)	1.52 (0.33, 7.06)	.593
50% reduction in CO level, 1-month FU	7 (27%)	7 (28%)	0.89 (0.24, 3.32)	.860
Mechanisms of change, M (SD) (n = 40)
Change in AIS (acceptance) score	0.16 (0.57)	0.18 (0.60)	0.02 (−0.33, .37)	.916
Change in CQS (commitment) score	−0.57 (0.94)	−0.26 (0.81)	0.36 (−0.13, .84)	.174
Mental health symptoms, M (SD) (n = 40)
Change in PHQ-9 (depression) score	−0.32 (8.74)	0.00 (5.74)	0.29 (−3.03, 3.60)	.871
Change in ASMR (mania) score	−1.11 (4.47)	−1.62 (3.65)	−0.51 (−2.09, 1.02)	.527

CO = carbon monoxide; EOT = end of treatment; FU = follow-up; PPA = point prevalence abstinence; AIS = Avoidance and Inflexibility Scale: CQS = Commitment to Quitting Smoking Scale; PHQ-9 = Patient Health Questionnaire-9; ASRM = Altman Self-Rating Mania Scale; OR = odds ratio; CI = confidence interval.

All models are adjusted for two factors used in stratified randomization: heavy smoking (≥20 cigarettes/day in the past month) and random effect of study site. Change scores are calculated as follow-up minus baseline and models adjusted for baseline values of continuous outcomes.

As an exploratory analysis, we examined detailed utilization metrics in the WebQuit Plus arm to understand (1) what proportion of participants completed each step of the intervention and what proportion earned a completion certificate (completion of all exercises in the program plus using the tracking feature at least once) and (2) how other features of the intervention were used. Eighty percent (n = 20) completed Step 1 (Quit Plan), 40% (n = 10) completed Step 2 (Awareness), 28% (n = 7) completed Step 3 (Acceptance Skills), 20% (n = 5) completed Step 4 (Values and Self-Compassion), 64% (n = 16) used the tracking feature at least once, and 20% (n = 5) earned the completion certificate. None of the participants accessed the Anytime Tools section, where a significant portion of the targeted content for smokers with BD was located. Regarding the text messaging, 28% (n = 7) of WebQuit Plus participants used the two-way keyword message feature to receive advice on urges (n = 8 messages, six participants), mood (n = 7 messages, five participants), motivation (n = 2 messages, two participants), or slips (n = 1 message, one participant). Message frequency changes by participants were primarily to increase frequency (n = 12). Fewer participants chose to decrease message frequency (n = 4) or stop them altogether (n = 2). These outcomes were similar to the Smokefree.gov arm (11 increased, 2 decreased, and 2 stopped messages).

Overall, 91% of the sample (91% in WebQuit Plus; 90% in Smokefree.gov, p = .923) reported using the patch at least once during the study. The mean number of weeks of daily patch use was 4.9 (SD = 3.1) overall, with no difference between arms: 5.3 (SD = 3.1) for WebQuit Plus versus 4.6 (SD = 3.2) for Smokefree.gov, p = .874.

Cessation Outcomes

Table 2 includes comparisons on the primary and secondary smoking outcomes. Missing = smoking, CO-confirmed, 7-day PPA at end of treatment was 12% for WebQuit Plus versus 8% for Smokefree.gov (odds ratio [OR] = 1.46, 95% confidence interval [CI] = 0.21 to 9.97, p = .701). Missing = smoking, CO-confirmed, 7-day PPA at 1-month follow-up was 8% in both arms (OR = 0.96, 95% CI = 0.12 to 7.57, p = .967). Findings were unchanged in secondary analyses considering all forms of tobacco and (nonmedicinal) nicotine use rather than smoking alone. On the secondary outcome of 50% reduction in CO level, similar to the primary cessation outcomes, there was a descriptively greater number of participants who reduced by 50% in the WebQuit arm compared to the Smokefree.gov arm (24% vs. 15%, p = .593), but there was little evidence of a difference by 1-month follow-up (28% vs. 27%, p=.860).

Mechanisms of Change

Change in mean AIS (acceptance) score from baseline to end of treatment did not differ between arms nor did change in CQS (commitment) scores (see Table 2).

Change in Mental Health Symptoms

Average depression and mania scale scores remained unchanged or slightly improved (0 to 1.6-point decreases) and were not significantly different between arms (see Table 2).

Safety

A total of 18 AEs were recorded during the trial, of which 9 were serious adverse events (SAEs). Seven events were categorized as psychiatric, and all seven of these were SAEs by virtue of requiring urgent psychiatric care or hospitalization (n = 5 unique participants: 3 in the Smokefree.gov arm, 2 in the WebQuit Plus arm). Only one of these SAEs was judged to have any potential relationship with the intervention (ie, one participant in the WebQuit Plus arm had increased impulsivity and suicidal ideation that required psychiatric hospitalization and was deemed “possibly related” due to the temporal association with treatment initiation). Overall, 13 of the 18 AEs were coded as “definitely not related” to study intervention, 3 were coded as “possibly related,” and 2 were coded as “probably related.” With the exception of the one psychiatric SAE noted above, all of the “possibly” or “probably” related AEs were either nicotine withdrawal symptoms or side effects of nicotine patch and were mild to moderate in severity.

Discussion

The two primary aims of this pilot trial were to assess trial design feasibility and to preliminarily evaluate the acceptability and cessation outcomes of the novel, targeted WebQuit Plus program compared with Smokefree.gov, both delivered with 8 weeks of nicotine patch. Recruitment was challenging, with the total number randomized falling short of the target sample size (51 of 60). Attending study visits was one of the main impediments from the participants’ perspective. Qualitative work to better understand barriers to participation in cessation trials could help to inform improved recruitment methods for this group. To conduct adequately powered trials focusing on smokers with BD, researchers may also need to consider broadening the pool of potentially eligible participants via any or all of the following methods: (1) minimizing eligibility criteria, as in a pragmatic trial design, (2) utilizing remote methods of conducting the trial, either by telephone or web, including online recruitment strategies, to increase reach of recruitment outside of narrow geographic regions and reduce participation burden (as in our previous remotely conducted study of web-delivered ACT for a general population of smokers that enrolled 221 participants with self-reported BD over 18 months without any targeted outreach⁵), and (3) testing a staged intervention approach that, as a first step, focuses on increasing motivation to make a quit attempt among those who are not yet ready to quit.^35–37

Once enrolled, participants were highly adherent to the trial protocol, attending the majority of sessions. Data retention rates of 73% at end of treatment and 80% at 1-month follow-up suggest that this is a feasible protocol to implement and that our retention plan of maintaining regular contact with participants during the treatment period (every 2 weeks by phone), reminder phone calls prior to in-person appointments, and compensation for time spent on research-specific activities was successful.

At end of treatment, there was a signal for greater acceptability of WebQuit Plus relative to Smokefree.gov. While there was a promising estimated effect size for cessation (OR = 1.46), the study was not designed or powered to detect a treatment difference between groups. Based on these findings, WebQuit Plus appears to be at least as effective as Smokefree.gov in the short term and likely to be more acceptable. It is also worth noting that the 7-day PPA for WebQuit Plus in combination with nicotine patch (12% at end of treatment) was very similar to the end-of-treatment 7-day PPA for 12 weeks of individual, face-to-face counseling (using a standard treatment approach) in combination with nicotine patch in the EAGLES trial (ie, 11%),⁴ suggesting that web-delivered ACT had outcomes similar to in-person brief counseling combined with nicotine patch. In our prior pilot work testing more intensive, in-person (n = 10) and telephone delivery of ACT (n = 6) in combination with nicotine patch for smokers with BD, we saw higher estimated quit rates: 40% for in person and 33% for telephone.¹¹ In addition to the different modality of treatment delivery, these higher quit rates may be, at least in part, a function of the differing characteristics of the samples, as our prior work had more restrictive psychiatric eligibility criteria and excluded those with elevated depressive symptoms and/or a major depressive episode at baseline, which can decrease the odds of successful cessation by as much as 50%.³⁸ In the present study, over half (55%) of participants had significant elevations in depressive symptoms (as indicated by their PHQ-9 scores) at baseline.

Refinements to WebQuit Plus may further improve acceptability and cessation outcomes. One clear future direction for enhancing the program is to alter the site’s organization to increase interaction with the targeted content. Although there was targeted content in several areas, the amount of targeted content in the Anytime Tools section that was not viewed at all was substantial. Adding individualized coaching (via text messaging or telephone) to help participants apply ACT skills and provide support and accountability⁷ may improve the development of acceptance skills and enhance commitment to quitting, potentially addressing the lack of observed impact on these theory-based mechanisms of change.

Low efficacy of the adjunctive pharmacotherapy may have negatively impacted the overall quit rates for these highly dependent smokers, as a recent study (published after this study began) suggested that nicotine patch alone may not be the most effective treatment for smokers with BD compared to other pharmacological interventions, especially varenicline.⁴ A different pharmacotherapy approach may be needed, particularly given the high severity of nicotine dependence observed in this sample as well as in other studies involving smokers with BD.^4,11,39 Varenicline is the only medication that has shown promise for smoking cessation in BD so far,⁴ but the efficacy of combination NRT has not been evaluated in this population.

Limitations of this study include small sample size with low power to detect small differences in quit rates, particularly given that differences as small as 1% can be clinically meaningful for low-cost web-based interventions with high potential reach.⁴⁰ While the sample size for this pilot trial is consistent with expert recommendations,³¹ precision of estimated effects is consequently low, resulting in wide confidence intervals across outcome measures. The characteristics of the sample (eg, low socioeconomic status) have been associated with lower quit rates,⁴¹ so results might be different in a different population of smokers with BD. Finally, the two programs differ in several respects, including the treatment approach (ie, ACT vs. standard care), BD-specific targeting (present vs. absent), and information architecture (sequential for the core ACT content of WebQuit Plus vs. unstructured for Smokefree.gov). Consequently, it is not possible to determine the independent effects of any of these characteristics on acceptability or cessation outcomes.

Despite its limitations, the study has a number of strengths. To our knowledge, this is the first randomized controlled trial of a behavioral smoking intervention targeted specifically for smokers with BD. It is also one of the largest studies examining cessation outcomes among smokers with a confirmed diagnosis of BD, with the EAGLES pharmacotherapy trial⁴² (n = 285 with BD⁴) being the largest and another pharmacotherapy-focused trial³⁹ being the second largest (n = 60). Given that a substantial proportion of the literature on smoking cessation studies focusing on the BD population have had sample sizes less than 20,^{11,21,43–45} this is clearly an underdeveloped area of research that is important to pursue given the potentially different pattern of response to pharmacotherapies and lower quit rates observed among smokers with BD.^4,5 Importantly, our findings that psychiatric symptoms showed either no change or slight improvement over time on average and that psychiatric AEs showed no clear relationship with the intervention further extend the evidence that cessation is not detrimental to mental health among smokers with BD,⁴ addressing a harmful myth that impedes cessation efforts for this group^6,20 and providing further impetus to address tobacco use as a major threat to the health of people living with BD. Preliminary acceptability and efficacy of WebQuit Plus, along with a growing literature supporting the use of remote methods of tobacco treatment delivery^46,47 to increase accessibility, support continued program refinement and evaluation in a larger trial with longer-term follow-up. If successful, WebQuit Plus could provide a highly accessible new treatment option for smokers with BD.

Funding

This study was funded by a grant from the National Institute on Drug Abuse, National Institutes of Health (#R34DA040119, to JLH and MO), with additional financial support provided by the National Network of Depression Centers.

Declaration of Interests

JLH has received research support from Pfizer. JBB has served as a consultant for GlaxoSmithKline and serves on the advisory board for Chrono Therapeutics. None of the other authors have competing interests to disclose.