Efficacy of telehealth acceptance and commitment therapy for weight loss: a pilot randomized clinical trial

Abstract

Telehealth coaching for weight loss has high population-level reach but limited efficacy. To potentially improve on this limitation, the purpose of this study was to determine the preliminary efficacy of the first known telephone coaching acceptance and commitment therapy (ACT) intervention for weight loss. A two-arm, stratified, individually randomized pilot trial comparing ACT (n = 53) with standard behavioral therapy (SBT; n = 52) was used for this study. Both interventions were delivered in 25 telephone coaching calls (15–20 min each) over a 12 month period. Weight was measured at baseline and 3, 6, and 12 month postrandomization follow-ups. Recruited from 32 U.S. states, participants were of mean age 40.7, 42% male, 34% racial/ethnic minority, and mean baseline body mass index of 34.3. Fractions of 10% or more scale-reported weight loss: 15% for ACT versus 4% for SBT at 3 month follow-up (N = 86; odds ratio [OR] = 4.61; 95% confidence interval [CI]: 0.79, 26.83), 24% for ACT versus 13% for SBT at 6 month follow-up (N = 72; OR = 2.45; 95% CI: 0.65, 9.23), 30% for ACT versus 30% for SBT at 12 month follow-up (N = 57; OR = 0.93; 95% CI: 0.28, 3.09). Fractions of 10% or more self-reported weight loss at 12 month follow-up: 25% for ACT versus 15% for SBT (N = 75; OR = 2.38; 95% CI: 0.68, 8.34). The conclusion of the study was the preliminary evidence that telephone coaching ACT may be efficacious for weight loss.

Lay Summary

Telephone coaching for weight loss reaches many people but has achieved very limited levels of weight loss. A program called acceptance and commitment therapy (ACT) is promising as an in-person weight loss program. However, ACT had not been tested before as a telephone coaching program. We conducted a 1 year telephone coaching study of 105 adults with overweight/obesity comparing ACT with standard behavioral therapy (SBT) coaching. The results showed that ACT telephone coaching had very encouraging levels of weight loss success compared to (SBT) coaching. These results suggest that telephone coaching ACT may be useful for weight loss.

Implications.

Practice: Telephone-delivered acceptance and commitment therapy (ACT) may be useful to improve the effectiveness of weight loss programs.

Policy: Policymakers who want to improve the population-level impact of weight loss programs should explore the use of telephone-delivered ACT that could be implemented on a broad scale.

Research: Future research should be aimed at conducting large-scale randomized trials of telephone-delivered ACT for weight loss.

INTRODUCTION

Telephone coaching interventions for weight loss are delivered by healthcare systems, employee wellness programs, and weight loss centers—reaching over 1.2 million U.S. adults per year as compared to 300,000 for in-person interventions [1–3]. Since the COVID-19 pandemic began, the reach of telehealth services, including telephone coaching for weight loss, has expanded exponentially [4]. Their greater geographic reach yields higher fractions of minorities, men, and those with lower income, lower education, and depression [1,2,5]. Telephone interventions are tailored to participants’ triggers for overeating/sedentary behavior and motivations to lose weight and provide individualized feedback. By contrast, group interventions provide little or no tailoring [6–8]. Telephone coaching uses standard behavior therapy (SBT) techniques, including avoiding triggers to eat and setting caloric intake and physical activity goals [9].

The problem with SBT telephone coaching interventions for weight loss is their small effect size: in randomized clinical trials (RCTs), a mere 0.46 mean difference in body mass index (BMI) between telephone coaching and passive control arms (e.g., usual care or waitlist) [2,5,9]. Effect sizes were even smaller with active treatment controls [2,5,9]. Low effect sizes likely reflect the fact that these interventions have been using the same behavior change techniques for several decades [9]. Small effect sizes stifle the ability of telephone coaching interventions to have a high population-level impact.

An intervention model with the potential to improve on SBT telephone coaching is acceptance and commitment therapy (ACT; [10]). ACT focuses on increasing willingness to experience physical cravings, emotions, and thoughts while making values-guided committed behavior changes. For weight loss, the ACT-based approach has only been tested in RCTs as in-person interventions [8,11–16]. An in-person group-delivered RCT (N = 190) randomizing adults (overweight/obese) to 12 months group-delivered sessions of ACT or SBT was recently reported [16]. ACT participants were 1.7 times more likely to lose 10% or more weight (odds ratio [OR] = 1.70; 95% confidence interval [CI]: 0.94, 3.08) at the 6 month follow-up and were 1.86 times more likely to lose 10% or more weight (OR = 1.86; 95% CI: 1.04, 3.23) at the 12 month follow-up.

Current study

Building on this group-delivered ACT-based intervention, this paper reports on the development of the first-known telephone-delivered ACT intervention for weight loss and its testing in a pilot randomized trial, including recruitment, retention, satisfaction, potential mediators and moderators, and preliminary efficacy for weight loss.

METHODS

Development of ACT telephone coaching for weight loss

We used an iterative, user-centered design approach [17] to adapt our group-delivered ACT-based intervention (R01 DK095069) [16]. We adapted the ACT group intervention [8] following the same number (25) and timing of sessions as the group program. Consistent with telephone coaching interventions for weight loss, the length of each call was 25–30 min for the first call and 15–20 min for all subsequent calls [2,5,9].

To explore the receptivity to this beta version of the ACT telephone coaching intervention, we conducted a diary study of the program with 10 adults with overweight or obesity recruited from around the USA (60% female, 30% minority; mean age 42, mean BMI 34.1). A diary study with 6–12 participants is recommended to obtain this feedback [18]. After each call, participants completed within 24 hr an online structured diary entry evaluating the content of each call. Ninety percent of all diary entries were completed. We also interviewed study participants during three phases of the intervention (after Calls 7, 14, and 25). Even though the focus was on usability, by Call 25, participants had lost 9.73 kg on average (standard deviation [SD] = 5.43). Ratings of call satisfaction and usefulness of skills were very high: 4.6 and 4.8, respectively, on a 0 (not all) to 5 (extremely) scale. In each of the 25 calls, 90%–100% of participants reported having subsequently practiced a skill covered in a given call. Results included the following strengths: (a) the ACT exercises were helpful, (b) the overall intervention stance of accepting cravings and focusing on values-driven food choices and physical activity was helpful, and (c) phone coaching provided accountability. For 6 of the 33 ACT exercises, some participants were unsure about when and how to use them. Thus, we clarified that these exercises could be used quickly, at the moment, in any place, whenever the participant had a craving. Following these minor changes, the revised version of intervention was ready for testing in the pilot RCT described herein.

Participants, recruitment, and enrollment

Eligibility criteria were similar to the group-delivered ACT intervention trial [8]. Inclusion criteria: (a) age 18 or older, (b) BMI ≥27 but not heavier than BMI of 45.5 (measured by weight and height [19,20]), (c) wants to lose weight in the next 30 days, (d) interested in learning skills to lose weight, (e) willing to be randomly assigned to either condition, (f) resides in USA, (g) has daily access to their own phone and email, (h) does not have a medical or psychiatric condition that would limit the ability to comply with the behavioral recommendations of the program or pose a risk to the participant, including meeting criteria for binge eating disorder, (i) not pregnant, planning to become pregnant, or breastfeeding in the next 12 months, (j) in the past 3 months, not used prescription medications that can cause a significant change in weight, (k) have not lost more than 5% of their weight in the past 6 months, (l) willing and able to read in English, (m) not participating in or planning to participate in other weight loss programs, (n) has not participated in our other interventions, (o) willing to complete follow-up surveys, and (p) providing email, phone, and mailing address. Exclusion criteria: opposite of inclusion criteria.

The study CONSORT participant flow diagram is shown in Fig. 1. Participants were recruited nationally from August 29, 2018 to March 31, 2019 through Facebook advertisements with ongoing monitoring and adjustment of recruitment yield, based on our successful methods for recruitment [21]. Some Facebook advertisements were designed for racial and ethnic minorities as well as men. Enrollment was limited to no more than 75% non-Hispanic White and 75% female participants to ensure racial/ethnic minority and male inclusion. All interested individuals were directed to the study website to learn about the study and complete an encrypted web-based screening survey. Those who were eligible were sent an email inviting them to provide informed consent and complete the encrypted baseline assessment. Actions to ensure that individuals were eligible for participation included CAPTCHA authentication, review of IP addresses for duplicates or non-U.S. origin, review of survey logs for suspicious response times (<90 s to complete screening or <10 min to complete baseline survey), and review of mailing addresses and phone numbers to check for prior enrollment in one of our previous studies. Those completing the online enrollment process were reviewed for eligibility by a phone call from a research staff member, who made three call attempts over a 14 day period to confirm their interest in participation. Those passing this step were randomized, sent an email notifying them that they were enrolled, mailed an information packet, calorie reference book, food scale, measuring cups, measuring spoons, and BodyTrace scale [22]. Anyone not confirmed eligible for any reason were sent a link to the Centers for Disease Control and Prevention Weight Loss and Management resource list and the Nutrition.gov weight loss resource.

Fig 1.

| CONSORT participant flow diagram.

All study activities were reviewed and approved by the institutional review board. The trial was registered on Clinical Trials.gov (NCT03738540).

There were 4265 unique ad clicks, 1,378 screened, 299 eligible, 194 consented, and 105 randomized participants. The most common reasons for ineligibility were BMI too high/too low (26%) and screened positive for binge eating disorder (19%).

Randomization and blinding procedures

Enrolled participants were randomized (1:1; ACT, n = 53; SBT, n = 52). We stratified on two factors with well-documented differences in weight loss: (a) sex (male vs. female) and (b) race/ethnicity (minority race or ethnicity vs. non-Hispanic White) [23,24] using permuted block randomization [25]. Randomization assignment was concealed from participants throughout the entire trial. Neither research staff nor study participants had access to upcoming randomized study arm assignments. Data collectors and investigators were blind to random assignment throughout the entire duration of the trial. To ensure that participants were blinded, each intervention was branded as “WeLNES: Weight Loss, Nutrition, Exercise Study.”

Interventions

The length of intervention was the same in both arms: 25 calls delivered over the course of 12 months. Calls 1–16 were weekly, Calls 17–23 were biweekly, and Calls 24–25 were monthly.

Each of the research team’s three coaches had behavioral weight loss coaching experience, basic clinical skills (e.g., active listening), and a master’s degree. They were supervised by a PhD-level licensed clinical psychologist. The coach’s qualifications and level of training are consistent with those of large-scale telephone programs in the community, such as the Kaiser Permanente Wellness Coaching Program [26] as well as similar clinical research trials [1–3,26]. Each coach delivered both interventions to control for therapist treatment effects [27].

Calls began 5–7 days after randomization to allow time to practice the use of the BodyTrace scale [22]. For each scheduled call, coaches made four call attempts and sent one email reminder per week for up to 4 weeks.

Shared treatment components

(a) Nutritional education (e.g., fruits and vegetables, lean protein sources, and reading food labels); (b) 1,200–1,800 calorie goal, depending on weight; (c) physical activity education (e.g., types of exercises); (d) gradual increases up to 250 min per week of aerobic activity; (e) intention formation, including setting specific, actionable, and time-limited goals for eating and physical activity; (f) self-monitoring of diet, exercise, and weight (e.g., weighing all foods); (g) providing feedback on progress; (h) stimulus control (e.g., removal of problematic foods from home and work); (i) relapse prevention (e.g., identifying high-risk situations for overeating/sedentary behavior); (j) social support (e.g., specifying how people you trust will support you). These components were similar to those used in Look AHEAD and the Diabetes Prevention Program (DPP) [28,29]. Weight data were securely viewed by coaches for discussion of progress during calls.

SBT-only components

The SBT-only telephone coaching components were derived from Look AHEAD and DPP [28,29]. Overall, the intervention was guided by a “Control What You Can” framework that focused on aspects of the participants’ experience that can be directly modified (i.e., their personal food environment). The approach for coping with cravings, emotions, and thoughts that trigger eating and sedentary behavior was to reinforce and support avoidance (e.g., ignore cravings) of these experiences. The approach for addressing motivation was to focus on one’s personal weight loss goals.

ACT-only components

The ACT-only telephone coaching components were derived from our group-delivered intervention [8]. Overall, the intervention was guided by a “Control What You Can and Accept What You Can’t” framework that taught participants to distinguish between the aspects of their experience that can be directly modified (i.e., their personal food environment) versus those aspects of their experience that are not under voluntary control (e.g., thoughts, emotions, and cravings). The approach for coping with cravings, emotions, and thoughts that trigger eating and sedentary behavior was to accept these experiences (e.g., notice and allow urges). An example of an acceptance exercise was to pause when participants have an urge, notice their breath and body, with the goal of not acting on the urge but instead just letting it be as it is. The approach for addressing motivation was to focus on one’s values (i.e., chosen life directions) that guide eating behaviors and physical activity. An example of a values exercise was asking what is inspiring the participant to lose weight and revisiting those answers throughout the course of the program.

Assessment

Coaching implementation quality

Two independent raters assessed a 20% random sample of audiorecorded calls on adherence to and competence of delivering shared, SBT-only, and ACT-only components. Calibration was by an expert PhD-level psychologist. Any adherence issues noted were immediately addressed in ongoing supervision.

Treatment satisfaction

Treatment satisfaction outcomes were the extent to which a participant: (a) was overall satisfied with the assigned treatment, (b) felt assigned treatment was helpful for losing weight, and (c) felt the assigned coach was helpful to them. Response choices for all items ranged from “Not at all” (1) to “Very much” (5) and were dichotomized such that a threshold of “Somewhat” (3) or higher represented satisfaction.

Scale-reported weight

BodyTrace body weight scale [22] wirelessly uploaded participants’ weight measures to our secure internal database. In-home cellular-enabled scales are now common in weight loss intervention trials [30–32] and the BodyTrace scale [22] has high concordance with in-person scale assessments (r = .999, p < .0001 [33]). We asked participants to weigh at the same time each assessment day, similarly clothed [33]. BodyTrace weight data were collected at baseline, 3 months (retention: 82%; 86/105), 6 months (retention: 69%; 72/105), and 12 months (retention: 54%; 57/105) postrandomization. The very low 12 month BodyTrace scale data retention rate of 54% was due to technical problems with the scale (e.g., not transmitting scale data).

Self-reported weight

For the baseline survey, all participants were asked to self-report their weight. For the follow-up surveys, self-reported weight was added late in the trial (April 15, 2019) due to the technical problems with the BodyTrace scale. Consequently, only the 12 month follow-up survey asked all participants to self-report their weight, of whom 74 reported (retention: 70%; 75/105). Self-reported weight at the 12 month follow-up was a secondary outcome of the study, supported by the high correlation between self-reported and scale-reported weight data at 12 months (r = .99; p < .001). Self-reported weight was also provided by 17 participants in the 3 month survey (retention: 16%; 17/105; 22 were asked) and by 32 participants in the 6 month survey (retention: 30%; 32/105; 32 were asked). None of the scale-reported or self-reported scale data retention rates differed by study arm at any follow-up point (all p > .05). For both the scale-reported and self-reported weight measures, the primary outcome was 10% or more weight loss because it is clinically significant and rigorous and more likely to prevent diabetes and heart disease than other common weight loss outcomes [7,34–36].

Baseline moderators of weight loss

(a) Values-guided motivation to change behavior (i.e., actions guided by life purpose) was measured with the 10-item Valuing Questionnaire [37], which has good reliability (α = 0.87) and predicts weight loss [38]; (b) psychological acceptance of food cravings was measured using the 10‐item Food Craving Acceptance and Action Questionnaire (FAAQ), which has good reliability (α = 0.93) and validity [39]; (3) minority race or ethnicity; and (d) gender.

ACT theory-based mediators

(a) Values-guided motivation to change behavior was measured with the Valuing Questionnaire [37]; (2) psychological acceptance of food cravings was measured with the FAAQ [39]; (3) mindful eating was measured with the 28-item Mindful Eating Questionnaire, which has good reliability (α = 0.84) and predicts lower weight [40].

Follow-up survey methods

Procedures for follow-up data collection were modeled after procedures that have been successful in our previous trials [21]. Briefly, our follow-up data were collected by our survey research unit that was blind to random assignment. Participants received $25 at each of the 3, 6, and 12 month assessments when they completed the self-reported outcome survey. We: (a) mailed a $2 preincentive letter 7 days before the first online survey invitation; (b) provided a $10 bonus incentive ($10 + $25 = $35) for completing the survey online within 24 hr. Participants who did not complete the survey online within 12 days were sequentially offered opportunities to do so by phone, mailed survey, and, finally, for self-reported weight only, by postcard.

Statistical analysis

The planned sample size of 100 (50 per arm) was sufficient to inform sample sizes for a full-scale trial. Analyses were complete case, with multiple imputation (R package “mice” [41]) as a sensitivity analysis of the 12 month weight loss outcome. Imputation of missing scale-reported data with self-reported scale data were conducted as a further secondary analysis. Treatment adherence items were assessed using R package “epiR” [42]. Continuous treatment utilization measures were compared between arms using generalized linear models. Binary outcomes were compared between study arms using logistic regression models. Potential moderation of the effect of the treatment arm on weight loss was assessed by including the interaction between study arm and baseline moderators in logistic regression models. All models included two covariates used as factors in stratified randomization: biological sex and minority race or ethnicity. Analyses were two-sided, with α=0.05, and were carried out in R Version 3.6.1 [43].

RESULTS

Study population characteristics

The randomized sample was from 32 U.S. states, including seven of the nine states with the highest rates of people with obesity (≥35% obesity) [44]. As shown in Table 1, they were of mean age 40.7, 42% male, 34% minority, 51% had some college (no degree completed) or less education, 15% LGBT, 24% depressed (CES-D; cutoff ≥ 10 [45]), and mean baseline BMI of 34.3. None of the baseline characteristics differed by study arm (all p > .05).

Table 1.

| Baseline characteristics

	Total	SBT	ACT
	(n = 105)	(n = 52)	(n = 53)
Demographics, n (%)
Age, mean (SD)	40.7 (12.8)	39.4 (12.5)	41.9 (13.1)
Male	44 (42%)	22 (42%)	22 (42%)
Non-Hispanic White	69 (66%)	34 (65%)	35 (66%)
Race
White	77 (73%)	36 (69%)	41 (77%)
African American	11 (10%)	6 (12%)	5 (9%)
Asian	5 (5%)	3 (6%)	2 (4%)
Native American or Alaska Native	2 (2%)	1 (2%)	1 (2%)
Native Hawaiian or Pacific Islander	1 (1%)	1 (2%)	0 (0%)
More than one race	7 (7%)	3 (6%)	4 (8%)
Unknown race	2 (2%)	2 (4%)	0 (0%)
Ethnicity
Hispanic	12 (11%)	5 (10%)	7 (13%)
Married	44 (42%)	24 (46%)	20 (38%)
Education
High school or less education	14 (13%)	6 (12%)	8 (15%)
Some college (no degree) or less education	40 (38%)	19 (37%)	21 (40%)
Family income
Less than $20,000	11 (10%)	5 (10%)	6 (11%)
$20,000 to <$55,000	42 (40%)	18 (35%)	24 (45%)
$55,000 or more	52 (50%)	29 (56%)	23 (43%)
Rural residence	11 (10%)	10 (19%)	1 (2%)
Working	82 (78%)	40 (77%)	42 (79%)
LGBT	16 (15%)	5 (10%)	11 (21%)
Depression, n (%)
Current depression symptoms	25 (24%), n = 103	11 (22%), n = 51	14 (27%), n = 52
Weight, M (SD)
BMI, self-reported weight	34.3 (4.7)	34.8 (5.2)	33.8 (4.2)
BMI, scale-reported weight	34.9 (4.8), n = 99	35.2 (5.3), n = 49	34.5 (4.3), n = 50
Mediators, M (SD)
Valuing questionnaire—progress	20.7 (5.6)	20.6 (5.4)	20.8 (5.9)
Food craving acceptance and action questionnaire	36.2 (7.4)	36.4 (7.2)	35.9 (7.6)
Mindful eating scale	2.6 (0.3)	2.6 (0.3)	2.7 (0.3)
Smoking behavior, n (%)
Current smoker	13 (12%)	6 (12%)	7 (13%)
Alcohol use, n (%)
Heavy drinkera	5 (5%)	2 (4%)	3 (6%)

BMI body mass index; SD standard deviation.

^aHeavy drinking was defined as ≥4 drinks for females or ≥5 drinks for males on a typical drinking day.

Treatment adherence

Both treatment arms scored similarly on the presence of therapeutic warmth (99% in both arms; p = .999), authenticity (99% in both arms; p = .999), and positivity (100% in both arms; p = .999). For adherence to shared treatment components, both treatments scored similarly on the occurrence of: weight loss goal setting and monitoring (89% in ACT vs. 92% in SBT; p = .556), physical activity goal setting and monitoring (99% in both arms; p = .999), intention formation (99% in both arms; p = .999), stimulus control (i.e., “changing what you can”; 59% in ACT vs. 62% in SBT; p = .828), nutritional education (78% in ACT vs. 87% in SBT; p = .054), social support planning and seeking (18% in ACT vs. 22% in SBT; p = .470), and feedback on ongoing progress (1.8 mean rating in both arms; p = .926; scale range: 0 [“did not occur during call”] to 2 [“occurred at least twice during call”]). For adherence to uniquely ACT (vs. uniquely SBT) components, the occurrence of teaching skills in acceptance (as opposed to avoidance) of triggers for eating and sedentary were 93% for ACT versus 0% for SBT (p < .001) and for values (as opposed to only goals) guiding eating and physical activity were 87% for ACT versus 0% for SBT (p < .001). Overall, interrater agreement was high across both arms (range: 0.84–1.0; mean = 0.95).

Participant utilization and satisfaction

As shown in Table 2, the total number of calls completed was 16.1 (SD = 9.8) on average (14.3 [SD = 10.3] for ACT vs. 18.0 [SD = 9.0] for SBT; p = .056), 43% completed all 25 calls (38% for ACT vs. 48% for SBT; p = .287), and total minutes of intervention call time averaged 318.1 (SD = 210.6) min (336.4 [SD = 243.9] min for ACT vs. 299.8 [SD = 171.6] min for SBT; p = .348). Both treatment arms scored similarly on (a) satisfaction with assigned treatment (90% for ACT vs. 86% for SBT; p = .647), (b) perception that assigned treatment was useful for losing weight (96% ACT vs. 87% for SBT; p = .202), and (c) perception that their assigned coach was helpful to them (86% for ACT vs. 86% for SBT; p = .989).

Table 2.

| Participant utilization and satisfaction

Outcome	Overall (n = 105)	SBT (n = 52)	ABT (n = 53)	Point estimate or ORa (95% CI)	p value
Number of coaching calls completed, M (SD)	16.1 (9.8)	18.0 (9.0)	14.3 (10.3)	Point estimate: −3.7 (−7.4, 0.0)	.056
Completed all 25 calls, n (%)	45 (43%)	25 (48%)	20 (38%)	OR: 0.65 (0.30, 1.43)	.287
Total length of calls completed, minutes, M (SD)	318.1 (210.6), n = 102	299.8 (171.6), n = 51	336.4 (243.9), n = 51	Point estimate: 39.3 (−42.4, 121.0)	.348
Satisfied with program, n (%)	59 (88%), n = 67	32 (86%), n = 37	27 (90%), n = 30	OR: 1.43 (0.31, 6.64)	.647
Program was useful for losing weight, n (%)	60 (91%), n = 66	33 (87%), n = 38	27 (96%), n = 28	OR: 4.22 (0.46, 38.63)	.202
Coach was helpful, n (%)	57 (86%), n = 66	32 (86%), n = 37	25 (86%), n = 29	OR: 0.99 (0.24, 4.12)	.989

ACT acceptance and commitment therapy; CI confidence interval; OR odds ratio; SBT standard behavioral therapy; SD standard deviation.

^aOutcome models are adjusted for the two factors used in stratified randomization: minority race/ethnicity and biological sex.

Weight loss

As shown in Fig. 2, results for the 10% or more scale-reported weight loss outcome were: 15% for ACT versus 4% for SBT at the 3 month outcome (N = 86; OR = 4.61; 95% CI: 0.79, 26.83), 24% for ACT versus 13% for SBT at the 6 month outcome (N = 72; OR = 2.45; 95% CI: 0.65, 9.23), and 30% for ACT versus 30% for SBT at the 12 month outcome (N = 57; OR = 0.93; 95% CI: 0.28, 3.09). The 12 month 10% or more self-reported weight loss outcome was 25% for ACT versus 15% for SBT (N = 75; OR = 2.38; 95% CI: 0.68, 8.34) and multiple imputation showed a similar result: 29% for ACT versus 18% for SBT (N = 1,050; OR = 2.13; 95% CI: 0.65, 7.01). (For the reader’s information, self-reported weight loss at the 3 month outcome was 12% for ACT versus 17% for SBT [N = 14; OR = 0.43; 95% CI: 0.01, 19.84] and weight loss at the 6 month outcome was 33% for ACT versus 23% for SBT [N = 25; OR = 2.17; 95% CI: 0.29, 16.07]). Results at all follow-up time points were very similar when missing scale-reported data were imputed with available self-reported scale data (results not shown). Among those who completed all 25 coaching calls, the 10% or more scale-reported weight loss was 38% for ACT versus 32% for SBT at the 12 month outcome (N = 35; OR = 2.13; 95% CI: 0.42, 10.84) and the 10% or more self-reported weight loss was 35% for ACT versus 20% for SBT at the 12 month outcome (N = 45; OR = 2.67; 95% CI: 0.62, 11.49).

Fig 2.

| Scale-reported weight loss outcomes for acceptance and commitment therapy versus standard behavioral therapy.

Moderators

Exploring potential moderation of the 12 month scale-reported and self-reported weight loss outcomes, there was no statistically significant evidence of an interaction between treatment arm assignment and these baseline factors: valued living, acceptance of food cravings, minority race or ethnicity, and gender (all p > .05).

Change in ACT theory-based mediators

Compared to SBT participants, ACT participants had greater baseline to 3 month, but not baseline to 6 month, increases in valued living (above median baseline to 3 month change in Valuing Progress [37] score: 57% for ACT vs. 36% for SBT [OR = 3.71; 95% CI: 1.27, 10.82]; above median baseline to 6 month change in Valuing Progress [37] score: 45% for ACT vs. 50% for SBT [OR = 0.87; 95% CI: 0.34, 2.27]) and neither baseline to 3 month changes nor baseline to 6 month changes in valued living predicted the 12 month scale-reported or self-reported weight loss outcomes (all p > .05). For the other ACT theory-based mediators, their baseline to 3 month or baseline to 6 month changes were not significantly different between treatment arms (all p > .05). However, there were descriptive differences with wide CIs. Specifically, the above median baseline to 3 month change in acceptance of food cravings was 53% for ACT versus 42% for SBT but the effect size had a wide CI (OR = 1.51; 95% CI: 0.59, 3.85). Similarly, the percentage of participants who lost 10% or more body weight at 12 months was 22% for those above the median in acceptance of food cravings versus 18% for those below the median in change in acceptance of food cravings—a difference with a wide CI (OR = 2.22; 95% CI: 0.52, 9.54).

DISCUSSION

Methodological strengths

The recruitment method was effective in enrolling a diverse sample of adults with overweight or obesity who were geographically dispersed across the USA, including seven of the nine states with the highest obesity rates (≥35% obesity) [44]. The recruitment yielded high geographic reach and high percentages of men, minorities, younger people, and those with lower education and depression [8,12,14,16]. The 3 and 6 month scale-reported outcome data collection yielded good retention rates. Outcome data retention rates did not differ by study arm. Both treatments were delivered according to their respective intervention models. Participants had a comparable dosage of treatment, each with about 5 hr of total intervention engagement. Participants were highly and similarly satisfied with their assigned treatments. Overall, there was strong evidence that trial was conducted with a high degree of integrity.

Weight loss and mediation

Results for the weight loss outcomes were generally more favorable for the ACT intervention at 3 and 6 month follow-ups. As retention rates impact the level of bias in the outcome data [46,47], the low (54%) retention rate for the scale-reported 12 month data made it hard to interpret those weight comparisons. By contrast, the higher retention rate (71%) for the self-reported 12 month data lent confidence in the interpretation that the ACT intervention was more favorable than the SBT intervention. The more promising results for ACT are likely not due to an underperforming SBT condition given that the effect sizes for SBT are very similar to those observed in past trials of telephone-delivered SBT [2,5]. The CIs for the weight loss effects from this study are comparable to other weight loss studies [48–50] and thus illustrate that response to weight loss interventions vary from person to person. Moreover, this study’s weight loss results are consistent with prior randomized trials of in-person delivered ACT [8,11–16].

Regarding mediation, comparing ACT with SBT, there was a significant effect of baseline to 3 month increases in valued living. As for remaining mediators, given low power to detect mediator effects due to the pilot sample size, the effect sizes were appreciable based on descriptive analysis. For example, the ACT treatment appeaed to be leading to greater increases in the acceptance of food cravings, which is consistent with prior research [8,15,16]. These results overall comport with the theoretical model underlying ACT for weight loss [8,15,16].

A generalizable quality of these results is that the interventions were delivered by coaches with training and experience comparable to community-based programs [26]. The fact that these results were achieved with a mere 5 hr of intervention contact over a 12 month period suggests that the ACT telephone intervention could be feasible to implement in healthcare systems. The value of the main results was underscored by the growing use of telephone intervention in behavioral healthcare [4].

Methodological limitations

As a pilot trial, the study had a lack of: (a) power to detect weight loss effects, mediation, and moderation; (b) longer-term follow-up given the phenomenon of weight loss relapse over time [51,52]; and (c) diet and physical activity measures as process data that predict weight loss outcomes [53,54]. Technical problems with the BodyTrace scale impacted the 12 month scale data reporting. While outcome data attrition is common in both in-person and remotely assessed outcome studies [5,55–58], we recommend that future studies use a digital scale with better usability/reliability, increase monetary compensation for providing scale data, and ensure that follow-up data compensation is dependent on the completion of scale weighing. We are now addressing these limitations in a fully powered randomized trial with longer-term follow-up (Clinical Trials.gov Identifier: NCT04447313). In conclusion, this pilot RCT conducted with a high degree of rigor showed preliminary evidence that telephone coaching ACT may be effective for weight loss.

Funding: This study was funded by the Fred Hutchinson Cancer Research Center Hutch Award Luncheon. The funder had no role in the trial conduct or interpretation of results.

Compliance with Ethical Standards

Conflicts of Interest: E.M.F. serves on the Scientific Advisory Board of Tivity Health and receives royalties from Oxford University Press for a published treatment manual and workbook. Other authors have no declarations.

Author Contributions: J.B.B. conceived of the study, oversaw every aspect of the intervention protocol development, trial design and implementation, analyses, interpretation, and led the writing of the manuscript. K.E.M. led the trial design and analysis, as well as contributed to every aspect of the trial implementation and manuscript writing. B.M.S. managed every aspect of the trial and contributed to the manuscript. E.M.F. collaborated on the development of the ACT procotol, the training and adherence manuals, collaborated on every aspect of the trial implementation, and contributed to writing the manuscript.

Ethical Approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. This article does not contain any studies with animals performed by any of the authors.

Informed Consent: Informed consent was obtained from all individual participants included in the study.

Data availability: Deidentified data from this study will be made available (as allowable according to institutional review board standards) by emailing the corresponding author.

Trial registration: The trial was preregistered: Clinical Trials.gov registration identification number NCT03738540.

Analytic plan: The specified written plan of analyses were not formally preregistered.

Analytic code availability: Analytic code used to conduct the analyses may be available by emailing the corresponding author.

Materials availability: Materials used to conduct this study may be available by emailing the corresponding author.