Full-scale Randomized Trial Comparing Acceptance and Commitment Therapy Telephone-Delivered Coaching With Standard Telephone-Delivered Coaching Among Medicare/Uninsured Quitline Callers

Abstract

Introduction

The aim of this study was to compare Acceptance and Commitment Therapy (ACT) telephone-delivered coaching with standard quitline (QL) telephone-delivered coaching.

Methods

Medicare/uninsured adults (analyzable sample N = 1170) who smoked at least 10 cigarettes per day were recruited from Optum, a major US provider of QL services, in a two-arm stratified double-blind randomized trial with main outcome of self-reported missing = smoking 30-day point prevalence abstinence (PPA) at the 12-month follow-up. Participants were mean (SD) age 47.4 (12.7), 61% female, and 72% white race. Five sessions of telephone-delivered ACT or QL interventions were offered. Both arms included combined nicotine patch (4 weeks) and gum or lozenge (2 weeks).

Results

The 12-month follow-up data retention rate was 67.8%. ACT participants reported their treatment was more useful for quitting smoking (92.0% for ACT vs. 82.3% for QL; odds ratio [OR] = 2.48; 95% confidence interval [CI]: 1.53 to 4.00). Both arms had similar 12-month cessation outcomes (missing = smoking 30-day PPA: 24.6% for ACT vs. 28.8% for QL; OR =.81; 95% CI: 0.62 to 1.05) and the ACT arm trended toward greater reductions in number cigarettes smoked per day (−5.6 for ACT vs. −1.7 QL, among smokers; p = .075).

Conclusions

ACT telephone-delivered coaching was more satisfying, engaging, and was as effective as standard QL telephone-delivered coaching. ACT may help those who fail to quit after standard coaching or who choose not to use nicotine replacement therapy.

Implications

In a sample of Medicare and uninsured QL callers, a large randomized trial with long-term follow-up showed that ACT) telephone-delivered coaching was more satisfying, engaging, and was as effective as standard QL telephone-delivered coaching—which has followed the same behavior change approach since the 1990s. This newer model of coaching might be a welcome addition to QL services.

Introduction

Cigarette smoking is the second leading cause of early death and disability globally.¹ Barriers to accessing face-to-face smoking cessation interventions include lack of reimbursement and low demand.² To address access problems, cessation quitlines (QLs) have become an integral component of tobacco control.^3,4

The QL recommended standard of care is the combination of proactive phone coaching and Nicotine Replacement Therapy (NRT).⁴ Since their inception in the early 1990s, QLs’ coaching approach has been based on US Clinical Practice Guidelines (USCPG).⁴ While the success rates of QLs’ coaching with NRT is considerably higher than the 4% success rates of quitting on one’s own,⁴ their weighted average 30-day quit rate has consistently remained at 14% (range 8% to 20%) at 12 months post-randomization.⁴ Improving the effectiveness of the QLs’ intervention could increase their public health impact.

Acceptance and Commitment Therapy (ACT)⁵ has potential to address the need for improving USCPG-based QL quit rates. Both ACT and the USCPG share a focus on teaching skills to cope with urges to smoke. However, the philosophy and content of those skills follow fundamentally different paradigms. Unlike the USCPG, ACT does not focus on changing the content of thoughts (eg, replacing an inaccurate thought with an accurate thought) but rather one’s relationship to them via active awareness and observation of thoughts.⁶ ACT’s first key innovation is a focus on increasing a person’s willingness to experience urges to smoke.^7,8 In contrast, the USCPG use problem solving skills to avoid and control urges to smoke.^4,9 ACT’s second key innovation is promoting value-driven behavior change ^7,8,10,11 (eg, asking: “deep down what really matters to you?”) whereas the USCPG promote reason-driven behavior change^4,9 (eg, asking: “would you list all your reasons for quitting smoking”). For the change approach, ACT teaches skills primarily through metaphors and experiential exercises^7,8,11 whereas the USCPG teach skills primarily through logical and literal explanation.^3,9 In ACT, metaphors and experiential exercises help people step back from literal thought, feeling, and urges. Finally, ACT differs from mindfulness-based therapies in that: (1) being mindful is only one of numerous ACT strategies for increasing willingness to experience urges and (2) ACT focuses on values.¹²

To date, fifteen controlled trials totaling 6991 study participants have tested ACT for smoking cessation in the modalities of group therapy, telephone coaching, website, and smartphone application.^7,8,13–25 The full-scale randomized trials show that group-delivered²⁴ and website-delivered¹⁶ ACT are strong alternatives to USCPG standard behavioral interventions for smoking cessation delivered in these same modalities. Smartphone-delivered ACT is more efficacious than USCPG and may have high public health impact if broadly disseminated.¹⁸

In a pilot randomized trial of telephone-delivered intervention (N =121), we compared ACT to QL, with both interventions offering five sessions. Participants received 2 weeks of nicotine patch or gum (participant’s choice). The missing equals smoking 30-day point prevalence abstinence (PPA) rates at 6 months post-randomization were 31% for ACT versus 22% for QL (odds ratio [OR] = 1.5; 95% confidence interval [CI] = 0.7 to 3.4).²⁶ Coaches’ teaching of ACT skills (eg, awareness of cravings) were highly predictive of lowered odds of smoking.²⁷ These results motivated the current study.

This article presents the fidelity, treatment receptivity, and cessation outcomes of a full-scale randomized trial comparing ACT telephone-delivered coaching with standard QL telephone-delivered coaching with the main outcome of self-reported missing = smoking 30-day PPA at the 12-month follow-up among Medicare/uninsured QL callers.

Methods

Design

A two-arm stratified randomized controlled trial (RCT). Participants were surveyed at 12 months, with interim surveys at 3 and 6 months. The 12-month primary endpoint accounted for the high relapse rates that occur by 12 months²⁸ and is directly comparable to the most rigorous trials of telephone-delivered smoking cessation.²⁹ Based on the 6-month quit rates observed in the pilot trial²⁶ and relapse rates occurring between 6 and 12 months after randomization,²⁸ the study was 80% powered for a two-tailed significant (ie, α = 0.05) difference between a 26.5% ACT quit rate and a 19.6% QL quit rate.

Procedures

Participants

Adult smokers (n = 1275) were recruited from November 2015 to August 2018 (Figure 1). Participants were callers to the South Carolina Tobacco Quitline (SCQL) and Louisiana Tobacco Quitline (LQL), operated by Optum, a US provider of QL services. South Carolina and Louisiana were chosen because their 20.7% to 21.0% adult smoking prevalence is substantially higher than the national average.³⁰ To prevent bias, recruitment communications stated that “the purpose of the study is to learn which of two telephone coaching programs for quitting smoking works better.” There were no references to ACT.

Figure 1.

Participant flow diagram.

Eligibility criteria: (1) aged 18 and older; (2) smoked at least ten cigarettes per day (to be eligible for NRT) and had done so for at least the past 12 months; (3) wanted to quit smoking in the next 30 days; (4) if concurrently using any other nicotine or tobacco products including e-cigarettes/vaping, wants to quit using them within the next 30 days; (5) willing to be randomly assigned; (6) willing and able to speak and read in English; (7) willing and medically eligible to use NRT, (8) resided in the United States, and expected to continue for at least 12 months; (9) not participating in other smoking cessation interventions (including our own interventions); (10) had regular access to a telephone; (11) if female, not currently pregnant, breastfeeding, or planning to become pregnant in the next 3 months; (12) did not have Medicaid or private insurance that provided NRT. Participants not eligible were offered the standard QL intervention. Except for the requirement of 12-month availability (instead of 6 months) and the addition of the LQL (to ensure adequate recruitment), these eligibility criteria were identical to our prior pilot RCT.²⁶ Baseline characteristics of the randomized participants were similar to those of SCQL and LQL callers not enrolled in this study.³¹

Recruitment

Optum staff offered the study to the SCQL and LQL callers. Interested callers were (1) screened for eligibility at Optum (N = 2898) and transferred to the Fred Hutchinson Cancer Research Center (Fred Hutch)’s research team who, within two business days, administered by phone the (2) verbal consent, and (3) baseline survey. Those who consented and completed the baseline survey were verified by Optum within 24 h and, if confirmed, were randomly assigned to treatment (N = 1275). Callers not enrolled were referred to their state’s QL program. All study activities were approved by Fred Hutch’s Institutional Review Board.

Randomization

Participants were randomized (1:1) to either ACT or QL using an automated algorithm stratifying on these factors empirically known to predict smoking cessation^9,26,32: gender (male vs. female), daily smoking frequency (<21 vs. ≥ 21), age (<39 vs. ≥ 39; because median age from the pilot trial was 38²⁶), and baseline acceptance of cravings (Avoidance and Inflexibility Scale score <2.88 vs. ≥2.88,³³ based on the 2.88 median value in the pilot RCT²⁶). Random assignments were concealed from participants. Neither research staff nor study participants had access to upcoming study arm assignments.

Blinding and Contamination

For participant blinding, neither intervention mentioned ACT and instead were generically described as “telephone coaching for quitting smoking.”. Outcome assessors were blinded to study arm assignment. Contamination was avoided by having an eligibility criterion of not having family, friends, or other household members participating.

Follow-up

Participants completed follow-up surveys at 3, 6, and 12 months post-randomization. Participants received $25 for completing each survey and an additional $10 bonus if the online survey was completed within 24 h of the survey email invitation. Those who did not complete the survey online were sequentially offered opportunities to do so by phone, mailed survey, and, for main outcomes only, by postcard.

Interventions

Both the ACT and QL intervention were delivered as five-session coaching protocols in combination with NRT. All participants were mailed a booklet specific to their assigned condition, and instructions for NRT use. Both interventions asked about e-cigarettes/vaping, and noted that the skills taught in their intervention can be applied to cessation of all nicotine and tobacco-containing products. Intervention length was the same in both arms: the first call was approximately 20 min and each subsequent call was approximately 10–15 min. Coaches in both arms made up to five call attempts over separate days for each of the five coaching calls before ceasing further contact with the participants. Intervention calls were offered for up to 90 days after randomization. Participants who did not answer their phones when the coaches called had the opportunity to call back and complete their intervention calls for up to 90 days after randomization.

ACT

The telephone-delivered ACT intervention^26,34 was delivered by the Fred Hutch-based team. Sessions covered the six core ACT processes: Values, Committed Action, Awareness, Being Present, Cognitive Defusion, and Self-as-Context. Overall, the acceptance components taught skills in (1) increasing willingness to experience urges that cue smoking, and (2) responding differently to smoking urges (eg, noticing and not acting on urges). The commitment components focused on helping individuals articulate the values guiding quitting (eg, the love of one’s children) and taking actions to quit smoking guided by those values.

QL

The telephone-delivered QL intervention was delivered by Optum. This intervention generally followed Agency for Healthcare Research and Quality practice guidelines for smoking cessation^1,4 and has demonstrated efficacy.³⁵ Sessions generally focused on assisting participants to commit to quitting by setting a quit date, develop a quit plan, discuss reasons for quitting, use NRT correctly, practice coping with urges (eg, deep breathing, listening to nature, or taking a walk), enlist social support, and remove all tobacco and associated items from their environment. (Differences between the two interventions are shown in Supplementary Table 1.)

Coach Qualifications and Training

All coaches had at least 2 years of general coaching experience, consistent with prior QL^35,36 and ACT intervention trials.^{37, 38} Coaches had various professional backgrounds (eg, hospice, social work) before entering training. Coaches completed at least 100 h of training and demonstrated acceptable adherence to the intervention protocol using standardized ratings. Quality control procedures included ongoing monitoring of protocol adherence and supervision. We followed the expert recommendation of having two teams of equally competent coaches each deliver one of the treatment arms.³⁹

Nicotine Replacement Therapy

NRT provision is a recommended standard of care for QLs,⁴ and thus was used in both interventions. The SCQL provided combined nicotine patch and gum or lozenge: 4 weeks of patch and 2 weeks of lozenge or gum, as compared to their previous practice during the pilot study of offering 2 weeks of NRT monotherapy.²⁶ SCQL was one of the few state QLs to offer combination NRT. To keep NRT levels equal across the study, we increased the NRT for LQL participants to match the SCQL offering. The combination of slow-acting NRT (eg, nicotine patch) with fast-acting NRT (eg, gum or lozenge) is more effective than either alone.^40,41 Coaches presented NRT as a tool for coping with urges and nicotine withdrawal while participants were learning their respective treatment’s behavioral skills. Optum’s medical team provided NRT oversight for both study conditions. Side effects/adverse events did not vary by treatment arm.

Coaching Fidelity

For coaching fidelity, we audio-recorded calls and rated coaching on adherence to each protocol and implementation quality in a 20% random sample of calls. Each ACT call was rated by two trained Fred Hutch-based raters and each QL call was rated by a team of Optum-based raters. The Fred Hutch team had the same two independent raters, a “gold standard” third rater who rated a 20% random sample of their sessions, and any score discrepancies among them were resolved with re-review and re-rating. The Optum team primarily used a team of call auditors who rated a 20% random sample of their sessions. These coders changed over the course of the study. Reliability was not assessed or corrected for during the study. The QL treatment was the control condition and was meant to be “treatment as usual.” Both treatments were rated for the common metrics of overall adherence to the components of the specific treatment model (1 to 5 scale; 1 = “not at all consistent with designated treatment model” to 5 = “entirely consistent with designated treatment model”) and treatment implementation, which is the skillfulness with which the coach implemented the specific treatment model (1 to 5 scale; 1 = “the treatment model was not at all implemented skillfully” to 5 = “the treatment model was extensively implemented skillfully”).

Measures

Baseline Measures

Participants reported on demographics, depression (Center for Epidemiological Studies Depression Scale; CES-D-20⁴²), alcohol use (Quick Drinking Screen; QDS⁴³), smoking behavior, commitment to quitting (Commitment to Quitting Scale; CQS⁴⁴), nicotine dependence (Fagerström Test for Nicotine Dependence; FTND⁴⁵), and smoking in their social environment.

Coaching and NRT Treatment Engagement

The total number of telephone coaching calls completed was the measure of coaching treatment engagement. Measures of nicotine of replacement therapy, self-reported at the 3-month interim survey, were: (1) whether participant used the nicotine patch (Yes/No), (2) whether participant used the nicotine gum (Yes/No), (3) whether participant used the nicotine lozenge (Yes/No), and (4) number of weeks used the combination of nicotine patch and gum or lozenge.

Treatment Satisfaction

Treatment satisfaction, measured in the 3-month interim survey, was the extent to which participants: (1) rated assigned coaching program as useful for quitting, (2) were satisfied with assigned coaching program, (3) would recommend assigned coaching program to a friend, and (4) and rated their assigned coaching program’s skills and exercises as useful. Example item: “How useful was your assigned telephone coaching program for quitting smoking?” Response choices for all items ranged from “Not at all” (1) to “Very much” (5) and were dichotomized at a threshold of “Somewhat” (3) or higher.

Acceptance of Cravings to Smoke

Acceptance of cravings to smoke was measured at baseline and 3-month follow-up via the Avoidance and Inflexibility Scales’ 9-item subscale measuring one’s willingness to experience physical cravings that cue smoking (adapted from Gifford et al.⁴⁶). The items were rated on a 5-point scale from (1) “Not at all” to (5) “Very willing” and averaged, with higher scores indicating greater acceptance.

Primary Cessation Outcome

For direct comparability with the most rigorous telephone-delivered intervention randomized trials to date,²⁹ the primary outcome of the study was missing = smoking 30-day PPA (ie, no smoking at all in the past 30 days) at 12-months post-randomization. Due to low demand characteristics for false reporting, the Society for Research on Nicotine and Tobacco Subcommittee on Biochemical Verification recommended biochemical confirmation as unnecessary in population-based studies with no face-to-face contact and studies where data are optimally collected through the web, telephone, or mail.⁴⁷ Self-reported smoking is a standard method for assessing the efficacy of telephone-delivered interventions.²⁹ We considered biochemical data collection and elected not to because of three major methodological problems: (1) high attrition, (2) problems with identifying the person providing the sample, and (3) the high cost relative to the likely low percentage of falsifying from a high reach-low intensity intervention (Ref. 51, Supplementary Material).^48–50 Therefore, smoking status was the self-reported response to the question “When was the last time you smoked, or even tried, a cigarette?”

Secondary Outcomes

The 12-month post-randomization secondary outcomes were: 30-day PPA with multiple imputations of missing outcomes, 30-day PPA at 12 months post-randomization among those who did not use fast-acting NRT, 7-day PPA missing = smoking, 7-day PPA with multiple imputations of missing outcomes, and 30-day cessation of all tobacco products (including e-cigarettes, vaping, smokeless tobacco, chewing tobacco, snus, a pipe, bidis, cigarillos, cigars, hookahs, kreteks, and cloves). The 6-month post-randomization secondary outcomes were: 30-day PPA missing = smoking and 7-day PPA missing = smoking. The baseline to 12-month post-randomization progress outcomes, among all participants and among 12-month daily smokers, were: whether participant reduced the number of cigarettes smoked, change in number of cigarettes smoked per day, whether participant made at least one quit attempt.

Statistical Analyses

Specified a priori as the primary outcome was the 12-month follow-up 30-day PPA using a missing equal smoking imputation, which assumes that participants with missing smoking status at follow-up are still smoking (Ref. 52, Supplementary Material). While research suggests that missing = smoking outcomes may be biased (Ref. 53, Supplementary Material), they are recommended by the Russell Standard (Ref. 52 in Supplementary Material), allowing for comparison of results with prior telephone-delivered intervention trials. All secondary outcomes described above were also examined. Multiple imputations by chained equations was used to create ten multiply imputed data sets and pool the analysis results using the R package “mice” (Ref. 54, Supplementary Material). We used logistic regression models for the cessation outcome as well as secondary binary outcomes related to cessation and treatment satisfaction. Generalized linear models were used to assess differences between treatment arms for continuous outcomes (eg, number of completed calls). In all models, we adjusted for the stratification variables used in randomization to avoid losing power and to improve CI estimates (Ref. 55, Supplementary Material). Finally, we evaluated the Bayes factor for the primary cessation outcome to provide a summary of the presence and magnitude of the treatment effect (Refs. 56, 57 in Supplementary Material). All statistical tests were two-sided, with α = 0.05, and analyses were completed using R 3.6.1 (Ref. 58, Supplementary Material) and R packages “BayesFactor” (Ref. 59, Supplementary Material) and “MASS” (Ref. 60, Supplementary Material).

Results

Recruitment, Enrollment, and Retention

As shown in Figure 1, 7196 individuals were screened, 2898 were eligible, 1338 consented, and 1275 were randomized (671 to ACT; 604 to QL). Due to Optum technical errors, 84 participants randomized to the ACT condition were provided access to the Optum web intervention, 24 of whom logged into the web program. All 84 participants were excluded. Once these contamination errors were discovered and reported to the Institutional Review Board, they were corrected and study enrollment continued until participants affected by the errors were replaced. Thus, the full analyzable sample was 1170 (586 for ACT; 584 for QL). The 12-month data retention rate was 68% (793/1170) and did not differ between arms (ACT: 67% [391/586]; QL: 69% [402/584]; OR = 0.91 [0.71, 1.17], p = .475). Participants completed the 12-month follow-up survey via web (60.8% of respondents), via telephone (24.5%), via mailed survey (10.3%), and via postcard short survey of primary and selected secondary outcomes (4.4%). Forty-two percent of those who completed the survey did so within 24 h of receiving the email invitation that noted the $10 bonus incentive. Follow-up data retention rates were 68.5% overall at 3-months ([802/1170); 69.1% [405/586] for ACT vs. 67.9% [397/584] for QL, p = .670) and 69.4% overall at 6 months ([812/1170]; 67.6% [396/586] for ACT vs. 71.2% [416/584] for QL, p = .184].

Baseline Participant Demographics and Smoking Behavior

As shown in Table 1, the mean (SD) age at enrollment was 47.4 (12.7) years. Participants were 72.4% women and 27.6% (868/2415) racial/ethnic minority. There were 52.5% with high school or less education. Regarding smoking, 87.1% had smoked for ≥10 years and 86.1% smoked more than a half pack (at least 11 cigarettes) per day. There were no statistically significant differences between the two arms on any baseline variable (all p > .05). Overall, these characteristics were very similar to those of prior published telephone-delivered smoking intervention trials.²⁹

Table 1.

Baseline Demographics and Smoking Behavior

	Total	QL	ACT
	(n = 1170)	(n = 584)	(n = 586)
Demographics
Age, mean (SD)	47.4 (12.7)	47.2 (12.4)	47.5 (13.0)
Male, No. (%)	454 (38.8%)	222 (38.0%)	232 (39.6%)
White, No. (%)	836 (72.4%), n = 1154	412 (71.8%), n = 574	424 (73.1%), n = 580
African American, No. (%)	263 (22.8%), n = 1154	133 (23.2%), n = 574	130 (22.4%), n = 580
Asian, No. (%)	1 (0.1%), n = 1154	0 (0.0%), n = 574	1 (0.2%), n = 580
Native American or Alaska Native, No. (%)	18 (1.6%), n = 1154	7 (1.2%), n = 574	11 (1.9%), n = 580
Native Hawaiian or Pacific Islander, No. (%)	2 (0.2%), n = 1154	2 (0.3%), n = 574	0 (0.0%), n = 580
More than one race, No. (%)	34 (2.0%), n = 1154	20 (3.5%), n = 574	14 (2.4%), n = 580
Hispanic, No. (%)	20 (1.7%), n = 1164	12 (2.1%), n = 579	8 (1.4%), n = 585
Married, No. (%)	369 (31.6%), n = 1169	176 (30.2%), n = 583	193 (32.9%)
Working, No. (%)	488 (41.7%)	247 (42.3%)	241 (41.1%)
HS or less education, No. (%)	613 (52.5%), n = 1168	307 (52.7%), n = 582	306 (52.2%)
LGBT, No. (%)	80 (6.9%), n = 1161	41 (7.1%), n = 580	39 (6.7%), n = 581
Heavy drinker,a No. (%)	98 (8.4%), n = 1169	57 (9.8%)	41 (7.0%), n = 585
Positive depression screen, No. (%)	427 (36.6%), n = 1167	213 (36.5%), n = 583	214 (36.6%), n = 584
Insurance status, No. (%)
Fully uninsured	588 (51%), n = 1162	296 (51%), n = 583	292 (50%), n = 579
Partially uninsured (no NR coverage)	287 (25%), n = 1162	147 (25%), n = 583	140 (24%), n = 579
Medicare	287 (25%), n = 1162	140 (24%), n = 583	147 (25%), n = 579
Smoking behavior
FTND score, mean (SD)	5.7 (2.0), n = 1147	5.7 (2.0), n = 574	5.7 (2.0), n = 573
High nicotine dependence (FTND ≥ 6), No. (%)	641 (55.0%), n = 1165	309 (53.2%), n = 581	332 (56.8%), n = 584
Smokes more than half pack per day, No. (%)	1,009 (86.2%)	509 (87.2%)	500 (85.3%)
Smokes more than one pack per day, No. (%)	347 (29.7%)	175 (30.0%)	172 (29.4%)
First cigarette within 5 min of waking, No. (%)	610 (52.1%)	308 (52.7%)	302 (51.5%)
Smoked for 10 or more years, No. (%)	1,014 (87.1%), n = 1164	510 (87.8%), n = 581	504 (86.4%), n = 583
Used e-cigarettes at least once in past month, No. (%)	119 (10.2%)	52 (8.9%)	67 (11.4%)
At least one quit attempt in past 12 months, No. (%)	419 (35.9%), n = 1166	202 (34.6%), n = 583	217 (37.2%), n = 583
Commitment to quitting smoking score, mean (SD)	4.5 (0.6), n = 1168	4.4 (0.6)	4.5 (0.6), n = 584
Friend and partner smoking
Close friends who smoke, mean (SD)	2.5 (1.8), n = 1163	2.5 (1.8), n = 581	2.5 (1.8), n = 582
Number of adults in home who smoke, mean (SD)	1.5 (1.0), n = 1162	1.5 (0.8), n = 581	1.5 (1.1), n = 581
Living with partner who smokes, No. (%)	301 (25.8%), n = 1166	155 (26.6%), n = 582	146 (25.0%), n = 584

ACT = Acceptance and Commitment Therapy; FTND = Fagerström Test for Nicotine Dependence; LGBT = lesbian, gay, bisexual, and transgender; QL = quitline.

^aHeavy drinking is defined as ≥4 drinks per day for females and ≥5 drinks per day for males within the past 30 days.

Coaching Fidelity

Inter-rater agreement for the coaching fidelity was higher for ACT than for QL. Specifically, for ACT calls, the prevalence adjusted kappa internal reliability was 1.00 (95% CI: 0.89 to 1.00) for overall adherence and 0.97 (95% CI: 0.84 to 1.00) for treatment implementation quality whereas in QL this internal reliability was 0.63 (95% CI: 0.38 to 0.81) for overall adherence and 0.39 (95% CI: 0.12, 0.62) for treatment implementation quality. Mean adherence and implementation quality scores were higher for ACT than for QL (overall adherence: 4.96 [SD: 0.20] for ACT vs. 4.33 [SD = 0.77] for QL, p < .001; treatment implementation quality: 4.89 [SD: 0.31] for ACT vs. 4.04 [SD: 0.84] for QL; p < .001). To explore whether any of the QL calls contained ACT-consistent content, in a random sample of 10 QL calls, the mean ratings for ACT’s core therapeutic processes of Awareness (ie, training in awareness of urges that cue smoking behavior), Values (ie, training skills in articulating one’s personal values guiding quitting smoking), and Committed Action (ie, training skills in taking actions to allow discomfort and not smoke, as guided by one’s personal values) were high: 3.7, 2.3, and 4.6, respectively, on a scale of 1 (never explicitly occurred) to 5 (addressed highly in-depth).

Treatment Engagement

As shown in Table 2, behavioral treatment engagement was higher in the ACT arm than the QL arm (mean number of calls completed: 3.1 for ACT vs. 2.4 for QL; p < .001). NRT engagement was similarly high across both arms. Specifically, of the 6 weeks of medications provided, the mean number of weeks of NRT use was 3.8 for ACT vs. 4.1 for QL (p = .228). Overall, 86.4% of participants reported using the nicotine patch provided, 40.7% reported using the nicotine gum provided, and 43.7% reported using the nicotine lozenge provided. There were no differences between treatment arms on NRT usage (all p > .05).

Table 2.

Treatment Engagement and Satisfaction^a

Variable	Overall	QL	ACT	OR, IRR, or point estimate (95% CI)b	p
	(n = 1170)	(n = 584)	(n = 586)
Engagement
Number of telephones coaching calls completed, mean (SD)	2.8 (1.8) median = 2	2.4 (1.6) median = 2	3.1 (1.9) median = 3	Point estimate: 0.6 (0.4 to 0.8)	<.001
Used nicotine patch, No. (%)	616 (86.4%), n = 713	303 (84.9%), n = 357	313 (87.9%), n = 356	OR: 1.28 (0.83 to 1.98)	.257
Used nicotine gum, No. (%)	288 (40.7%), n = 708	144 (40.8%), n = 353	144 (40.6%), n = 355	OR: 0.99 (0.73 to 1.34)	.942
Used nicotine lozenge, No. (%)	310 (43.7%), n = 709	160 (44.8%), n = 357	150 (42.6%), n = 352	OR: 0.90 (0.67 to 1.22)	.506
Number of weeks used NRT daily, mean (SD)	4.0 (3.7), n = 664	4.1 (3.8), n = 329	3.8 (3.5), n = 335	IRR: 0.91 (0.78 to 1.06)	.228
Satisfaction
Coaching program was useful for quitting, No. (%)	595 (87.2%), n = 682	275 (82.3%), n = 334	320 (92.0%), n = 348	OR: 2.48 (1.53 to 4.00)	<.001
Satisfied with assigned coaching program, No. (%)	622 (91.2%), n = 682	291 (87.7%), n = 332	331 (94.6%), n = 350	OR: 2.45 (1.39 to 4.33)	.002
Would recommend assigned coaching program, No. (%)	632 (88.4%), n = 715	308 (86.8%), n = 355	324 (90.0%), n = 360	OR: 1.39 (0.88 to 2.22)	.161
Skills and exercises were helpful, No. (%)	579 (85.8%), n = 675	273 (82.7%), n = 330	306 (88.7%), n = 345	OR: 1.65 (1.06 to 2.57)	.026
Telephone coach was helpful, No. (%)	600 (88.5%), n = 678	274 (83.5%), n = 328	326 (93.1%), n = 350	OR: 2.67 (1.61 to 4.44)	<.001

ACT = Acceptance and Commitment Therapy; CI = confidence interval; IRR = incidence risk ratio; NRT = Nicotine Replacement Therapy; OR = odds ratio; QL = quitline.

^aThe number of calls completed was assessed at 6-month follow-up, while the remaining engagement and satisfaction items were assessed at 3-month follow-up.

^bOR indicates odds ratio in logistic regression for binary variables, IRR indicates incident rate ratio in negative binomial regression for count variables (ie, number of weeks of daily NRT use), and point estimate indicates difference between treatment arms for continuous variables (ie, number of calls completed). Results are adjusted for the four factors used in stratified randomization: daily smoking frequency, gender, age, and acceptance of cravings to smoke.

Treatment Satisfaction

ACT participants were more satisfied with their coaching program than QL participants (Table 2). For example, ACT participants reported their coaching program was more useful for quitting smoking (92.0% for ACT vs. 82.3% for QL; p < .001) and was more satisfying overall (94.6% vs. ACT vs. 87.7% for QL; p = .002).

Impact on Acceptance of Cravings to Smoke

ACT and QL participants had similar and significant baseline to 3-month month follow-up increases in acceptance of cravings to smoke (M = 0.36 [95% CI: 0.28 to 0.44], p < .001 for ACT vs. M = 0.24 [95% CI: 0.16 to 0.32], p < .001 for QL). Baseline to 3-month changes in acceptance of cravings strongly predicted 30-day PPA rates at the 12-month follow-up (OR = 4.93; 95% CI: 3.64 to 6.69; p < .001).

Primary Cessation Outcome

The missing equals smoking 30-day PPA rates at the 12-month follow-up were 24.6% for ACT and 28.8% for QL (Table 3[; OR =.81; 95% CI: 0.62 to 1.05). The Bayes Factor for the primary abstinence outcome was 0.24, indicating “substantial” evidence for the null hypothesis of no difference between treatment arms (Ref. 61, Supplementary Material).

Table 3.

Smoking Cessation and Progress Outcomes.^a

Outcome variable	Overall	QL	ACT	OR or Point estimate	p
	(n = 1,170)	(n = 584)	(n = 586)	(95% CI)b
12-Month cessation outcomes
30-Day PPA, No. (%)	312 (26.7%)	168 (28.8%)	144 (24.6%)	OR: 0.81 (0.62 to 1.05)	.106
30-Day PPA, multiple imputation, No. (%)	4698 (40.2%), n = 11,700	2467 (42.2%), n = 5,840	2231 (38.1%), n = 5,860	OR: 0.84 (0.64 to 1.10)	.200
7-Day PPA, No. (%)	354 (30.3%)	181 (31.0%)	173 (29.5%)	OR: 0.93 (0.73 to 1.20)	.591
7-Day PPA, multiple imputation, No. (%)	5316 (45.4%), n = 11.700	2623 (44.9%), n = 5,840	2623 (46.0%), n = 5,860	OR: 1.04 (0.82 to 1.32)	.749
30-Day PPA of all tobacco products (including e-cigarettes), No. (%)	292 (25.0%)	153 (26.2%)	139 (23.7%)	OR: 0.88 (0.67 to 1.15)	.335
6-Month cessation outcomes
30-day PPA, No. (%)	328 (28.0%)	175 (30.0%)	153 (26.1%)	OR: 0.83 (0.64 to 1.07)	.145
7-day PPA, No. (%)	380 (32.5%)	200 (34.2%)	180 (30.7%)	OR: 0.85 (0.66 to 1.09)	.197
12-Month cessation progress
Reduced number of cigarettes smoked, among all participants, No. (%)	565 (85%), n = 661	279 (84%), n = 332	286 (87%), n = 329	OR: 1.23 (0.80, 1.91)	.349
Change in number of cigarettes per day, among all participants, mean (SD)	−12.8 (15.4), n = 661	−11.5 (15.8), n = 332	−14.1 (14.8), n = 329	Point estimate: −2.0 (−4.0 to 0.1)	.065
At least one quit attempt, among all participants, No. (%)	601 (85%), n = 706	296 (83%), n = 355	305 (87%), n = 351	OR: 1.33 (0.87 to 2.02)	.183
Reduced number of cigarettes smoked, among smokers, No. (%)	190 (66%), n = 286	92 (63%), n = 145	98 (70%), n = 141	OR:1.24 (0.74 to 2.09)	.416
Change in number of cigarettes per day, among smokers, mean (SD)	-3.6 (16.9), n = 286	−1.7 (17.5), n = 145	−5.6 (16.1), n = 141	Point estimate: −3.3 (−6.9 to 0.3)	.075
At least one quit attempt, among smokers, No. (%)	274 (82.8%), n = 331	135 (79.9%), n = 169	139 (85.8%), n = 162	OR:1.57 (0.87 to 2.82)	.132

ACT = Acceptance and Commitment Therapy; CI = confidence interval; OR = odds ratio; PPA, point prevalence abstinence; QL = quitline.

^aPenalized imputation (ie, participants lost to follow-up are assumed to be smokers) was specified a priori as the primary outcome.

^bOR indicates odds ratio in logistic regression for binary variables, and point estimate indicates difference between treatment arms for continuous variables. Results are adjusted for the four factors used in stratified randomization: daily smoking frequency, gender, age, and acceptance of cravings to smoke.

Secondary Cessation Outcomes

As shown in Table 3, the multiple imputations 30-day PPA rates at the 12-month follow-up were 38.1% for ACT and 42.2% for QL (OR = 0.84; 95% CI: 0.64 to 1.10). The missing equals smoking 7-day PPA rate at the 12-month follow-up was 29.5% for ACT and 31.0% for QL (OR = 0.93; 95% CI: 0.73 to 1.20). The above pattern of results was similar for 12-month multiple imputations of 7-day PPA, 30-day PPA of all tobacco products (including e-cigarettes), and the 6-month cessation outcomes.

Cessation Progress From Baseline to 12-Month Follow-up

As shown in Table 3 results among 12-month daily smokers and among all participants, the ACT arm trended toward greater reductions in the mean number of cigarettes smoked per day from baseline to 12-month follow-up: 5.6 cigarettes reduced for ACT versus 1.7 cigarettes reduced for QL among 12-month daily smokers (p =.075); 14.1 cigarettes reduced for ACT versus 11.5 cigarettes reduced for QL among all participants (p = .065).

Post hoc Analysis of Those Not Using NRT

From an ACT perspective, the function of NRT may be to avoid cravings and withdrawal. Therefore, we posited post hoc that participants who elected not to use NRT would benefit more from ACT (than QL) because of its skills exercises focus on accepting cravings and withdrawal. Following this logic, the results showed that, among those who elected not to use any NRT (patch, gum, or lozenge), 30-day PPA rates at the 12-month follow-up were 31.0% for ACT and 23.3% for QL (OR = 1.49; 95% CI: 0.45 to 4.90; N = 59) and the 7-day PPA rates at the 12-month follow-up were 34.5% for ACT and 23.3% for QL (OR = 1.80; 95% CI: 0.55 to 5.86; N = 59), although these results were nonsignificant.

Discussion

This article reported on a large RCT with 12-month follow-up comparing ACT telephone-delivered coaching with standard QL telephone-delivered coaching for smoking cessation. ACT participants were more satisfied and more engaged. ACT participants had quit rates similar to the QL and had suggestive evidence of a greater reduction in the number of cigarettes smoked per day. Post hoc analyses, while not significant, suggested telephone-delivered ACT may be more helpful for smoking cessation among those not using NRT.

The cessation rates were similarly high across both arms, suggesting that both were highly effective and that ACT is a reasonable alternative to QL. On the main outcome of missing equals smoking 30-day PPA at 12-month follow-up, the quit rates (24.6% for ACT vs. 28.8% for QL) were about twice as high as the 14% weighted average (range: 8% to 20%) 30-day PPA at 12-months observed in prior QL trials (Ref. 62, Supplementary Material).^4,36 One reason for these high cessation rates is that both treatments included combination NRT (patch plus gum or lozenge), which is now offered in the SCQL but is only rarely offered in QLs across the United States due to higher costs (Ref. 63, Supplementary Material). Adding combination NRT to a behavioral treatment greatly enhances quit rates.^40,41 However, combination NRT may have overpowered the effects of ACT compared to QL. A study with NRT monotherapy, as was done in the pilot RCT, may have yielded different results.²⁶ While inconclusive, the descriptively higher cessation rates in ACT versus QL among those who did not use NRT suggests that NRT may decrease the relative efficacy of ACT. ACT could thus be an option for those electing not to use NRT. Needed next are randomized trials of telephone-delivered ACT for smoking cessation without the provision of NRT. As many smokers cycle through QLs multiple times (Ref. 64, Supplementary Material), ACT could be also offered to those who do not quit smoking with standard QL coaching (Ref. 64, Supplementary Material).

While ACT was implemented with high fidelity, the QL treatment reliability and fidelity limitations hinder our understanding of the QL content. These limitations led the authors to take a further step to understand the content from a random sample of 10 QL calls. The results of this follow-up analysis, which showed high QL ratings for Awareness, Values, and Committed Action, are difficult to interpret from a small sample. The ratings show that the Awareness and Committed Action had the highest rating whereas the ratings for Values was lower, which would suggest that greatest level of overlap was with Awareness and Commitment and least overlap was with Values. Consistent with this interpretation, both the ACT and QL arms had similar and significant baseline-to-3-month follow-up increases in acceptance of cravings. In turn, acceptance of cravings highly predicted a greater odd of smoking cessation at the 12-month main outcome. Moreover, the prior pilot trial suggested that ACT processes of Awareness are highly predictive of a greater odd of smoking reduction.²⁷ Overall, high ratings for ACT content in the random sample of QL intervention calls as well the evidence for QL’s similar intervention impact on acceptance of cravings are shortcomings that may limit the internal validity of the study.

While we were encouraged that the 68% retention rate at the 12-month follow-up was higher than the 55% average obtained with similar study populations recruited via QLs,²⁹ we are unclear about the causes of the 32% outcome data attrition. Our hypothesis was that the attrition reflected the challenges of the study population being low-income, which may have made it more difficult to track them for outcome surveys (eg, higher rates of residential changes and disconnected phones (Ref. 65, Supplementary Material). However, when identical follow-up methods were employed in our other large scale RCTs of smokers recruited in the general community, we have obtained 12-month retention rates of 88% among low-income smokers (ie, household income of $20 000 or less [Refs. 67, 68, Supplementary Material]).¹⁸ Future research is needed to determine effective means of increasing retention of QL callers. A final limitation of the study sample was the fact that they were on Medicare or were uninsured, which limits generalizability of the findings.

In summary, a large randomized trial with long-term follow-up showed that ACT telephone-delivered coaching was more satisfying, engaging, and was as effective as standard QL telephone-delivered coaching. ACT may help those who fail to quit after standard coaching or who choose not to use NRT.

Supplementary Material

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

Clinical Trials.gov Registration Number: NCT02421991.

Acknowledgments

We gratefully acknowledge the tireless contributions of the entire study staff and the ACT and QL coaches. We are thankful to the South Carolina and Louisiana State Quitlines for their participation in this trial. We are very appreciative of every study participant.

Funding

National Institute on Drug Abuse (R01 DA038411 awarded to JBB). Funder had no role in the trial conduct or interpretation of results.

Declaration of Interests

KMC and AJT are employees of Optum, the provider of quitline services for this study. Other authors have no declarations.

Data Availability

Data for this article will be shared with other researchers upon reasonable request to the corresponding author.