Study protocol of a pragmatic randomized controlled trial incorporated into the Group Lifestyle Balance™ program: the nutrigenomics, overweight/obesity and weight management trial (the NOW trial)

Seventy-four participants (n = 37 per group) are needed in this study to detect a clinically meaningful difference of 4% in body fat percentage, assuming 80% power, an alpha of 5%, and a standard deviation of 6.1% [43]. We aimed to recruit 88 participants (n = 44 per group) to account for the potential dropout rate of 20%. While minimal research exists outlining a clinically meaningful change in body fat percentage, a 5% change in weight (which would come from fat mass, water weight and/or muscle mass) is often reported to be clinically meaningful [44]. Furthermore, clinical experience from the registered dietitians involved in this study helped to determine the clinically meaningful 4% difference mentioned above. This difference has also been supported in published reference standards of body fat percentage in Caucasian adults which indicate that a 4% change in body fat percentage is associated with a 1 – 2 decile change on the reference standards charts [45].

A cohort randomization model was used rather than subject randomization to allow all participants in a given GLB™ group to obtain the same intervention. At the time of randomization, 12 cohorts (GLB™ groups) were randomized 1:1 to either the personalized lifestyle intervention (PLI) based on genetics, or the standard lifestyle intervention (SLI) based on population-based guidelines. It was anticipated that 12 groups of approximately 7 participants each would be needed to obtain the desired sample size of 88 participants. Prior to obtaining informed consent from participants, randomly permuted blocks were generated using the original generator on an internet-based randomization program [46]. Since participant recruitment was quicker than anticipated and there was an even 1:1 split of a PLI and a SLI group in the last two randomized groups, only 10 of the randomized groups (5 PLI groups, 5 SLI groups) were used, resulting in a total of N = 140 participants in the study (mean number of study participants per group ± SD = 14.0 ± 4.1).

Participants were recruited from the GLB™ program at the East Elgin Family Health Team (EEFHT). There were two primary methods of recruitment into this program: [1] adults from Elgin and Middlesex Counties in Ontario, Canada were referred to the GLB™ program by healthcare professionals in the area such as registered dietitians (RDs), physicians, nurses, nurse practitioners, and physical therapists; and [2] adults joined the program through word-of-mouth referrals from members of the community. Participants expressing interest in joining the GLB™ program were invited to the EEFHT for an in-person meeting to learn about the NOW Trial, and to provide written, informed consent if they decided to take part in the study. Therefore, participants are highly reflective of typical patients in the GLB™ program. Recruitment occurred from April 2017 until September 2018. This study is registered with clinicaltrials.​gov (NCT03015012) and was approved by the Western University Research Ethics Board (108511).

Screening and informed consent were completed in person at the EEFHT during the in-person meeting. Inclusion criteria were as follows: BMI ≥ 25.0 kg/m², ≥18 years of age, English-speaking, willing to undergo genetic testing, having access to a computer with internet at least one day per week, and not seeing another healthcare provider for weight loss advice outside of the study. Pregnancy and lactation were considered exclusion criteria.

Upon obtaining written, informed consent, participants were scheduled for in-person baseline data collection, within approximately 14 days (mean ± SD = 9.3 ± 5.7) prior to the intervention start date. Participants were not given any lifestyle advice during the run-in period.

All data are entered into the database using unique study codes for each participant and is securely stored in a locked cabinet, in a locked office. Baseline data consisted of a combination of in-person, online, and telephone data collection methods.

Trained research assistants collected 3-day food records over the phone using the validated multiple-pass method [47]. To reduce participant burden, each participant chose to have either 3 separate phone calls (one for each day of intake), or 1 phone call (for all three days of intake). One weekend day and two weekdays were collected. In rare cases where research assistants were unable to reach participants over the phone, the food records were collected in-person at the EEFHT. The food records were then analyzed using the Canadian Nutrient File within the nutritional analysis software program ESHA Food Processor (version 11.1).

Participants also completed a self-administered past-month, semi-quantitative, online food frequency questionnaire, the Canadian Diet History Questionnaire II (CDHQII). This questionnaire is a modified version of the United States Diet History Questionnaire adapted for Canadian data [48]. Most participants completed this questionnaire away from the EEFHT, but in cases where participants did not have internet access at home (n = 3), the CDHQII was self-administered at the EEFHT.

In-person baseline data collection included: measured height and weight (used to calculate BMI), a BIA assessment to obtain body composition data (using the Body Stat 1500MDD), a past-week physical activity recall (used to calculate metabolic equivalents), a baseline demographic questionnaire, a list of medications, and a TPB questionnaire. To optimize reliability, weight and height measurements were taken on the same Health O Meter Professional weigh scale and stadiometer, and body composition was assessed using the same BIA machine. The TPB questionnaire was developed based on Ajzen’s Guide to Constructing a TPB Questionnaire [49]. The results for weight, body fat percentage, body fat (kg), lean weight, and water weight from the BIA were communicated to participants during their in-person baseline data collection visit.

To assess possible short-term changes in components of the TPB (e.g. attitudes towards nutrition, physical activity, genetic testing, etc.), the TPB questionnaire was administered twice during the baseline assessment period: once during the one-on-one, in-person meeting (pre-intervention), and once immediately after the first group-based intervention session was delivered (post-intervention).

During informed consent meetings, baseline data collection and the run-in period, participants were blinded to their group assignment. However, participants were not blinded to their group assignment during the administration of the second baseline TPB survey (completed immediately after the first group session in order to assess possible changes short-term in key components of the TPB upon receiving either population-based advice or genetic-based advice). Research assistants collecting and analyzing food intake data were also blinded to the study group of the participants and the statistician will be blinded to the group assignments. Since our outcomes included changes in attitudes related to genetic testing for personalized lifestyle advice, as well as change in nutrition and physical activity habits, it was inappropriate to blind the participants throughout the entire duration of the study. Therefore, participants were informed of their group assignment during the first group intervention meeting, as further outlined in section 4.9, below. One author generated the allocation sequence, enrolled participants, facilitated group and one-on-one interventions, collected data, entered data into the database and scheduled participants, and therefore could not be blinded. Allocation was concealed for the other five co-authors.

Staggered cohorts have been used to reduce the impact of confounding by indication and have previously been successful in studies comparing active and passive treatment groups [50]. In the present study, staggered cohorts were pre-planned in order to maximize study efficiency and effectiveness. Seasonality and timing of groups were considered in the planning phase to ensure that there was a similar amount of SLI groups and PLI groups offered during the day and evening. Three SLI groups were offered during the day, and 2 SLI groups were offered in the evenings. Likewise, 3 PLI groups were offered during the day, and 2 PLI groups were offered in the evenings. 1 SLI group began in the spring, 2 in the summer, and 2 in the fall. Similarly, 1 PLI group began in the spring, 2 in the summer, 1 in the fall, and 1 in the winter.

Given its previously documented success [5, 8, 9], the GLB™ program was chosen as the gold standard comparator for this RCT. Furthermore, given that this study is taking place within routine community/clinical practice, it is highly pragmatic with a mean overall PRECIS-2 score of 4.4 (Table 1) [51].

Participants joined the GLB™ group session that best suited their availability, and were blinded to the group intervention assignment at this time. As previously detailed, groups were pre-randomized 1:1 to receive either the standard 12-month GLB™ Program curriculum + a summary report of population-based lifestyle recommendations (SLI/Control Group), or a modified 12-month GLB™ Program + a summary report of DNA-based lifestyle recommendations (PLI Group). All participants underwent a group-based weight management program in addition to four one-on-one sessions (one baseline and three follow-up) with a RD. Group sizes ranged from 7 to 20 participants per group at baseline, with a mean group size of 14 participants. At the three follow-up one-on-one sessions, the RD reviewed the information provided in the summary report (population-based recommendations for the SLI group and DNA-based recommendations for the PLI group; refer to Additional files 2 and 3, respectively, for sample reports). One-on-one sessions lasted approximately 30 min. The same RD who completed the one-on-one sessions was also the lead trained lifestyle coach for the GLB™ Program group sessions. This allowed for optimization of intervention reliability in all group and one-on-one sessions. These sessions were highly standardized as outlined in Additional file 4. No additional healthcare professionals above and beyond standard practice were hired to run the intervention at the EEFHT. Interventions took place between May 2017 and September 2019.

We plan to use SPSS Version 23.0 to conduct all statistical analyses, and the data will be analyzed on an intention-to-treat basis. Generalized linear mixed-effects models will be used to test between group differences from baseline to each follow-up period for each outcome indicator. If significant mean differences are detected, a Tukey’s post hoc test will be used to compare differences by group. General linear regression models will be used to assess interactions between a given genotype of interest and dietary component of interest on BMI and body composition. General linear regression models will further be used to assess interactions between TPB components, study group, and anthropometric measures of weight and body composition. No interim analyses will be completed.

The primary outcome in this study is change in percent body fat. Secondary outcome measures include changes in: dietary intake (calories, fat, protein, carbohydrates, unsaturated fat including mono- and poly-unsaturated fat, saturated fat, and sodium), physical activity, attitudes, subjective norms, behavioural control, intention to make lifestyle changes, weight and BMI.

We plan to disseminate the findings from this trial through: a community presentation to the participants involved in the study, presentations at relevant conferences for researchers and healthcare professionals, as well as in peer-reviewed publications.