Preference for behavior change strategies in randomized controlled trials: Evidence from weight management

Abstract

Background:

Randomized controlled trial participants are expected to embrace assignment to any of the study arms, yet individuals’ relative preference for the study arms invariably affects who participates in trials and for how long.

Methods:

Our ongoing Avoid/Resist trial (1R01DK130851) tests two strategies to bridge the intention-behavior gap in a weight management intervention. Avoid combines pantry makeover and online grocery shopping. Resist involves gamified, inhibitory control training. During screening, individuals rate Avoid and Resist on affective valence (I don’t like this – I like this) and instrumental utility (This will not benefit me – This will benefit me) using a 0–100 analogue scale. K-means clustering was used to identify clusters of individuals based on their liking and perceived benefits of the tested strategies before randomization.

Results:

Among respondents who completed the screener between January 2024 and January 2025 (n = 306; 64 % Female; 40 % Hispanic/Latino), the correlations between liking and perceived benefit ratings were high (>0.70). Median scores of liking and perceived benefits were 90 and 88 for Resist, and 91.5 and 90 for Avoid. K-means clustering revealed 3 groups: (1) highly favorable to Avoid and Resist (all ratings >90; 61 %); (2) relative preference for Resist (22 %); (3) relative preference for Avoid (17 %).

Conclusions:

Even among individuals willing to be randomized, nearly 40 % had a relative preference for one of the study arms. Additional work is needed to understand the role of relative preference on retention, adherence, and outcomes in weight management trials.

Trial registration:

The study is registered with ClinicalTrials.gov (NCT05143931).

1. Introduction

In randomized control trials, participants are expected to embrace their assigned study arm and adhere to protocols. Individuals’ relative preference for the study arms, however, invariably affects who participates in trials, how well participants adhere to components and for how long [1]. Those who do not expect benefits from the tested interventions are unlikely to participate, and participants assigned to their non-preferred condition may not complete study follow-ups [2,3] or stop adhering to the intervention [4,5].

Available evidence generally supports a strong correlation between emotional and cognitive drivers of relative preference among alternatives [6]. With some exceptions, people tend to have positive feelings towards what is useful to them, and they anticipate benefit from what they like. Yet, few studies have examined the respective and/or combined influence of emotion and cognition in the context of clinical trials [7,8]. The emotional dimension of relative preference reflects the affective valuation of a particular option (e.g., “I like this”; “I don’t care for it”) [9]. These initial feelings, which may result from prior experiences with similar interventions or components, tend to predict adherence to trial requirements [10]. Conversely, the cognitive or utilitarian dimension of relative preference reflects the benefit an individual expects to derive from the alternative (e.g., “This will be useful for me”; ‘I don’t think it will benefit me”) [11]. Perceived utility has been linked to trial retention [12], and misalignments between participants’ expected and actual outcomes can lead to premature disengagement [13,14].

Our ongoing Avoid/Resist trial (1R01DK130851) tests two strategies to bridge the intention-behavior gap in a weight management intervention. Avoid combines pantry makeover and online grocery delivery. Resist involves gamified, inhibitory control training. All participants are enrolled in the Weight Watchers© (WW) digital program for 12 months and are randomized to (1) WW, (2) WW + Avoid, (3) WW + Resist or (4) WW + Avoid+Resist. Since we initiated recruitment, we noticed that some individuals seem to prefer one of the study arms. Some individuals simply opted out because they were unwilling to be randomized (n = 22), while others expressed which strategy they believed would work for them or voiced skepticism for a component. In exit interviews with withdrawn participants, 17 % explicitly stated that the assigned study arm was the reason for their withdrawing consent, while 45 % invoked lifestyle incompatibility with the assigned intervention [15].

Despite its potential influence on study engagement, intervention adherence, and resulting outcomes, initial preference is often overlooked in clinical trials [16]. This observational study leverages screening data from the Avoid/Resist trial to examine initial, relative preference for the study arms among individuals interested in participating in a weight management randomized controlled trial. We were especially interested in capturing the proportion of individuals with initial preferences, as well as the relationships between affective valence (liking) and instrumental utility (benefit) towards the behavior change strategies.

2. Methods

Paper flyers and digital advertisements targeting healthy adults (age ≥ 18 years old) who were seeking to lose weight were disseminated through a variety of community, clinical, and educational settings and platforms across Los Angeles County, CA. Interested individuals were asked to scan the flyer QR code to complete an eligibility screening questionnaire. The screener captured medical history, willingness and ability to engage with the study procedures, as well as Hispanic ethnicity and sex as birth, which were used as stratifying variables in the parent trial.

The recruitment for the Avoid/Resist trial was initiated in June 2021. In January 2024, we began assessing relative preference for the Avoid and Resist strategies in the screening questionnaire. Individuals were presented with brief descriptions of the study arm requirements and their underlying rationales. Specifically, Avoid was described as “pantry makeover and online grocery delivery to remove temptation from the home food environment”. Resist was described as “daily gamified inhibitory control training to strengthen impulse control and temptation management”. To minimize order effects, the presentation of the Avoid and Resist descriptions was randomized. Individuals were then asked to rate Avoid and Resist on affective valuation (I don’t like this – I like this) and perceived benefit (This will not benefit me – This will benefit me) on a 0–100 analogue scale. Descriptive statistics were used to explore patterns and trends in self-reported preference. K-means clustering and the elbow method were used to identify clusters of individuals based on their liking and perceived benefits of the tested strategies.

3. Results

The analytic sample includes all individuals who completed the Avoid/Resist eligibility screening questionnaire (n = 306; 64 % Female; 40 % Hispanic/Latino) between January 2024 and January 2025, regardless of whether individuals were ultimately enrolled into the study. Among the 306 individuals screened, 152 were deemed eligible (49.7 %) based on (1) body mass index (25–45 kg/m2) and (2) availability of a household member to complete study assessments.

Overall, individuals rated positively both intervention strategies, with median liking and perceived benefit ratings of 90 (IQR 75–100) and 88 (IQR 71–100) for Resist, and of 91.5 (IQR 75–100) and 90 (IQR 70–100) for Avoid (Fig. 1).

Fig. 1.

Distribution of responses for liking and perceived benefit for the Avoid and Resist strategies.

The correlations between Resist liking and Resist perceived benefit ratings (r = 0.72; 95 % CI: 0.66 to 0.77) and between Avoid liking and Avoid perceived benefits (r = 0.77; 95 % CI 0.73 to 0.82) were generally high. Liking scores for Resist were, however, only moderately correlated to Avoid liking ratings (r = 0.43; 95 % CI 0.33 to 0.51), with a similar correlation between Avoid and Resist perceived benefit ratings (r = 0.48; 95 % CI 0.39 to 0.56). There was no significant difference in liking and perceived benefit ratings by sex at birth, ethnicity, or trial eligibility.

Despite the overall high liking and perceived benefit ratings for both Avoid and Resist, relative preference emerged among 39 % of respondents. Specifically, a K-means clustering analysis revealed three groups of individuals: (1) highly favorable to Avoid and Resist (all ratings >90; 61 %); (2) relative preference for Resist (22 %); (3) relative preference for Avoid (17 %). Fig. 2 presents a simplified visualization of these clusters based on liking ratings.

Fig. 2.

Simplified visualization of the three factors based on liking ratings alone.

4. Discussion

This observational study leveraged screening data from the Avoid/Resist trial to examine initial, relative preference for the study arms among individuals interested in participating in a weight management randomized controlled trial. While the overall liking and perceived benefit ratings for Avoid and Resist were equally high (>85/100), nearly 40 % of respondents had a relative preference for one of the study arms. This relative preference was further supported by the moderate correlations between Avoid and Resist ratings, suggesting that some respondents who rated one option highly only gave moderate ratings to the other option. These findings may help explain variability in weight management outcomes [17–20]. Conceivably, participants assigned to their non-preferred condition are more likely to opt out of randomization, prematurely terminate their involvement, and/or fail to adhere to intervention recommendations or study protocols. By contrast, those assigned to their preferred treatment condition may adhere better than expected, which in turn may lead to more optimal outcomes. These issues threaten the validity of our findings as treatment effects may not reflect the effectiveness of the strategies or interventions tested. Research supports the benefit of participants’ treatment choice in specific areas, such as pharmacotherapy versus psychotherapy for depression, or surgery versus lifestyle interventions for back pain [21]. When strong preferences for treatment exist, patient-preference designs, in which some patients choose their intervention assignment, might be preferred to traditional randomized controlled trials [22].

Our findings indicated strong correlations between affective and utility ratings for a given strategy, suggesting that respondents had positive feelings towards the option they believed would be most useful to them, and/or anticipated more positive outcomes from the intervention they liked. These results are consistent with the strong correlation between emotion and cognition in decision-making [6]. In other areas, however, affective valence and perceived instrumental utility may differ markedly, possibly resulting in choice conflicts or ambivalence. For instance, individuals undergoing cancer treatment rarely report “liking” chemotherapy, but they generally anticipate benefits [23,24]. Similar choice ambivalence is common in decisions around health behaviors such as food choices (i.e., “I don’t really like spinach, but it provides health benefits”) or decision-making around time allocation (i.e., “running will be very unpleasant, but it will yield positive health outcomes”) [25].

Our findings suggest that, among individuals willing to be randomized in the Avoid/Resist trial, nearly half had an initial preference for one of the behavior change strategies. Strong initial preference likely influences whether someone remains in a trial once arm assignment is revealed. The initial evaluation of treatment options can be influenced by a myriad of factors, including personality traits, prior experiences with similar interventions, and even the description of the experimental conditions provided by the study team [16,21]. Among randomized participants, their lived experience of the intervention may further change how they value or perceive the assigned study arm, both positively and negatively. At a minimum, ensuring that participants’ initial expectations are aligned with the intervention requirements– beyond the oft-general description provided in the informed consent– may minimize premature disengagement or help ensure that enrolled participants are willing and able to complete the study [13,14]. The implication of these findings is not entirely clear at this stage, as additional work is needed to further understand the role of relative preference on retention, adherence, and weight management outcomes, as well as in the context of other behavioral trials [16].

Supplementary Material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.cct.2025.108031.

Data availability

Data will be made available on request.