Randomized controlled trial of a novel lifestyle intervention used with or without meal replacements in worksites

Abstract

Objective:

Lifestyle interventions have had limited effectiveness in worksites when evaluated in randomized trials. We assessed the effectiveness of a novel lifestyle intervention for weight loss (Healthy Weight for Living, HWL) implemented with or without meal replacements (MR) in worksites. HWL used a new behavioral approach emphasizing reducing hunger and building healthy food preferences, and, unlike traditional lifestyle interventions, did not require calorie counting.

Methods:

Twelve worksites were randomized to an 18-month intervention (N=8; randomization within worksites to HWL, HWL+MR) or 6-month wait-listed control (N=4). Participants were employees with overweight or obesity (N=335; 48 [SD 10] years; BMI 33 [6] kg/m²; 83% female). HWL was group-delivered in person or by videoconference. The primary outcome was 6-month weight change; secondary outcomes included weight and cardiometabolic risk factors measured at 6, 12 and 18 months in intervention groups.

Results:

Mean 6-month weight change was -8.8% (95%CI: -11.2,-6.4) for enrollees in HWL and -8.0% (-10.4,-5.5) for HWL+MR (P<0.001 for both groups vs controls), with no difference between interventions (P=0.40). Clinically meaningful weight loss (≥5%) was maintained at 18 months in both groups (P<0.001).

Conclusions:

A new lifestyle intervention approach, deliverable by videoconference with or without MR, supported clinically impactful weight loss in employees.

Introduction

The Centers for Disease Control and Prevention (CDC) recommend offering lifestyle interventions for healthy weight management in worksites (1) because the majority of adults are employed and rising health care costs incentivize employers to support employee health. However, systematic reviews indicate that most lifestyle interventions tested in employees, as well as commercial lifestyle interventions suitable for scaled use in worksites, have not achieved significant weight loss (2, 3). To our knowledge, other than 1 pilot worksite trial by our team (4), no randomized trial of a U.S. weight loss intervention has demonstrated the 5% mean weight loss that confers clinically important health benefits (5). One potential explanation for the general limited effectiveness of lifestyle interventions is that burdensome core features such as daily food logging limit participant adherence, with an additional factor being the widespread availability of unhealthy food at worksites (6).

We developed a novel lifestyle intervention for weight management (Healthy Weight for Living, HWL) based on a revised health behavior change model (4, 7). The new approach emphasizes addressing biological antecedents of food intake (hunger, food cravings, and food preferences), and developing intrinsic motivation for behavior changes leading to weight loss, topics that receive limited emphasis in traditional lifestyle interventions. HWL is also operationalized without requiring daily food and physical activity logging. In a pilot cluster-randomized trial of 4 worksites (4), participants achieved 8% mean weight loss over 6 months compared to wait-listed controls, which is 3 times greater than typically achieved in worksite interventions (2, 3, 8, 9), and only 11% dropout in the 6-month intensive intervention phase. HWL participants also had no significant weight regain at 1 year. Subsequent programmatic adaptations introduced a flexible delivery structure (in person or by videoconference) and a variety of features to support scaling (10). Nevertheless the original randomized pilot (4) was relatively small, and a subsequent evaluation of a scalable prototype in a real-life setting (10) was non-randomized.

We conducted a cluster-randomized controlled trial to evaluate the effects of HWL on weight loss, cardiometabolic outcomes, weight-related behaviors and health measures compared to wait-listed controls. The HWL program was delivered by a commercial provider (10) to evaluate effectiveness in a scalable model. As meal replacements (MR) have been reported to facilitate portion control and increase weight loss during lifestyle interventions (11, 12), we additionally compared outcomes for participants randomized to HWL+MR versus HWL only, at 12 and 18 months.

METHODS

Study design

The trial (Figure 1) was an 18-month, parallel-group randomized controlled weight loss trial of HWL with or without provided MR (ClinicalTrials.gov identifier: NCT02593240) and included a separate intervention offering that was unrelated to weight loss (13). The trial described here was conducted in 12 worksites in Greater Boston area. Recruitment occurred between September 2015 and February 2016 and the trial was completed in February 2018.

Figure 1:

CONSORT diagram.

The Tufts Health Sciences Institutional Review Board approved the study. A worksite representative gave consent for each site, and enrolling employees provided written informed consent.

Worksites were cluster-randomized in a 2:1 ratio (8 intervention sites; 4 control sites) stratified by worksite type (commercial sites, non-profit sites). Participants within intervention sites were individually randomized to HWL or HWL+MR. Site and participant randomizations were performed within our database manager (StudyTRAX LLC, Macon, GA) using Microsoft.Net Random, with the assignments available to the study manager and Dr. Das, who oversaw study logistics. Randomization information was revealed after baseline testing. The intervention provider was not provided with randomization information. Participants were provided with a $25 stipend for completion of outcomes at 6, 12 and 18 months.

The primary hypothesis was that there would be significant weight loss in HWL and HWL+MR enrollees compared to controls at 6 months. Six months was chosen as wait-list period for controls to support participant retention in this group. A secondary hypothesis was that HWL+MR would result in greater weight loss than HWL alone. Additional study outcomes included changes in cardiometabolic risk factors, reported health and quality of life, and biobehavioral variables relevant to weight management in all participants at 6 months and in intervention participants at 12 and 18 months. The primary analysis was an intention-to-treat (ITT) analysis of enrolled participants; a completers analysis was also preplanned.

Worksites and participants

A multi-stage screening process was used, starting with an email to 155 worksites in the target geographical area. Worksites were informed that the programs would be provided cost-free to participants and meetings would be held at lunchtime or outside work hours to minimize worksite burden. Inclusion criteria for worksites: ≥3 years in operation; ≥300 employees, with an annual personnel turnover of ≤15%; and availability of a representative to give company consent and coordinate study logistics. Exclusion criteria for worksites: other recent, current, or impending on-site commercial programs for general well-being or weight loss. The first 12 interested and eligible sites were enrolled.

To recruit employees, worksites sent an email to employees informing them about the study, and the research team conducted an on-site information session with prospective participant screenings and, for eligible individuals, the opportunity to enroll. Inclusion criteria for employees: full-time employment; willing to accept randomization; ≥21 years; body mass index (BMI) ≥ 25 and <50 kg/m² at screening; a provided email address for study communications; and a signed physician release for study participation to address potential safety concerns of weight loss. Exclusion criteria for employees: non-English speakers; prior weight loss surgery; medical diagnoses that would limit participation; pregnancy or lactation; severe mobility limitations; multiple severe dietary intolerances; >15-lb weight loss in past 6 months; >14 hours/week of vigorous activity; or concurrent participation in another health intervention. Enrollment was capped at ~30 participants for most worksites, with an additional allowance for high demand larger worksites.

Lifestyle intervention

Like traditional lifestyle interventions, HWL (also called iDiet, www.theidiet.com) was informed by multiple theories of health behavior change, including goal-setting theory and Social Cognitive Theory (SCT) (14). These theories support standard program components, including goal setting, development of self-regulation skills, flexible eating restraint, and increasing self-efficacy through incremental accomplishments. Problem solving around stimulus control is another typical program element (15) also included in HWL. The specific goal-setting activities for HWL focused on daily self-weighing rather than the daily logging of food and physical activities that are core requirements of traditional interventions. In addition, HWL’s revised interpretation of SCT included biological antecedents of food intake as factors influencing program adherence (4, 7). Thus, while many lifestyle interventions in the clinical setting (16) are not focused on hunger, and are aimed at managing—rather than reducing, food cravings (17) —core HWL elements aimed to reduce hunger and food cravings and develop healthier food preferences. HWL also emphasized developing intrinsic motivation (18) for the behavior changes that cause negative energy balance in contrast to the traditional approach of building extrinsic motivation for weight loss by reducing energy and fat intakes and increasing physical activity. In other words, HWL emphasized changing types of foods, eating patterns, and activity for their inherent satisfactions (19), with weight loss viewed as a secondary consequence.

Operationally, the revised health behavior change model underlying HWL was implemented in ways that had both similarities and differences from the traditional approaches embodied in the national Diabetes Prevention Program (DPP) (20). Similarities included identical goals for weight loss (0.5–1.0 kg/weight loss/week) through the combination of a reduced-energy diet (deficit of 500–1000 kcal/day) and increased physical activity, as well as the general importance of goal setting, self-monitoring, problem solving, and planning for weight loss success.

Principle operational differences between HWL and traditional behavioral interventions such the National Diabetes Prevention Program (20) were as follows:

• The principle self-accountability tracking in HWL was for daily weight measurements, and daily calorie and activity logging was not required.

• Energy-adjusted self-selection structured menu plans (1000, 1200, 1500, 1800, or 2000 kcal/day depending on participant weight and age) were provided in lieu of required calorie logging. In contrast to the general portion recommendations used in DPP, these menus were very specific for types of foods, recipes, and brands, so that menu adherence automatically created the planned calorie deficit without the need for calorie logging. Food variety was also restricted during the first 2 weeks to simplify start-up; greater variety was introduced starting in week 3, and a 700 kcal/week of discretionary foods was also added at this time.

• In contrast to the DPP emphasis on low-energy diets with 25% of energy from fat, the dietary composition principles of HWL were designed to reduce hunger and increase satiety (4) with 4 dietary factors identified as supporting this goal (4): high fiber (target > 14 g/1000 kcal), moderately high protein (25% energy), and moderately low carbohydrate (45% energy), with high volume and a predominance of low-glycemic-index carbohydrates (21, 22, 23, 24, 25). The research literature is unclear over whether these dietary factors have independent or additive effects (26); thus, the provided meals and snack recipes used pair-wise combinations of dietary factors (e.g., high protein and high fiber, low glycemic index and low energy density), which also allowed participants to choose a range of dietary composition based on their selections. Additional strageties to manage hunger included optional use of “free foods” ad libitum that were very low in calories and high in fiber (e.g., high fiber cereal with low-calorie milk, green vegetables).

• The types of foods offered in the HWL menus and recipes were designed specifically to have similar tastes to generally popular foods (e.g., hamburgers, pizza, chocolate desserts). This was not only to allow participants to enjoy the types of foods they were familiar with, but also as an experimental technique to manage cravings. The technique was based on the principle of devaluating associative conditioning (27, 28), and involved substituting foods with similar flavors to preferred items but with both a reduced post-ingestive energy load and rate of energy absorption (i.e., fewer calories digested at a slower rate). Additional experimental craving reduction techniques included the forehead tapping method (27) for acute craving management.

• HWL did not implement exercise recommendations at the beginning of the intervention but started to encourage exercise around week 6 after the major dietary changes had been implemented.

• In addition to emphasizing methods for active hunger management and craving reduction techniques to a greater extent than the DPP lesson plans, HWL lesson plans also differed from DPP in not including sessions on the calorie content of specific foods, mindful eating, or importance of logging exercise.

HWL was implemented by a commercial provider (www.theidiet.com) as a cost-free group-based program with 1-hour meetings led by interventionists. Interventionists were hired and trained by the company and had a variety of educational backgrounds. Based on a previous evaluation indicating no difference in program effectiveness or drop-out between in-person and videoconference delivery (10), worksites could select either option. Meetings were held weekly for 24 weeks and monthly thereafter for intervention participants, while controls received a 12-meeting program after the 6-month outcome assessment indicating the end of their study participation. Each meeting consisted of a check-in, a pre-recorded video-format education unit, and time for questions and support. Participants could also communicate with their interventionist and each other through an online message board, were sent a mid-week email check-in from the counselor, and were encouraged to log their weights daily (29) in the website, which could be seen by their interventionist. Food and activity logging were not required, but counselors reviewed any logs that participants sent to them.

Meal replacements

Following randomization, HWL+MR participants received an allowance of calorie-controlled MR products with recommendations to consume 2 per day for the first 6 months and 1 per day thereafter. Due to budgetary and geographic constraints the MR was provided in bulk at the beginning of the study (400 packets of products, i.e., 2/day x 6 months plus some extra) and again at the 6-month timepoint (200 packets of products, i.e., 1/day x 6 months plus some extra) All provided products were shelf-stable powders that could be reconstituted with water to make shakes, hot cereals, and complete meals (Nutrient Foods LLC, Reno, NV) (30). They contained an average of 240 kcal/meal, with compositions designed to support general health (31). Specific composition features included low sugar, sodium, and saturated fat contents as well as fortification with vitamins, minerals, and omega-3 fatty acids. MR adherence was not specifically tracked to minimize participant burden. The composition data are summarized in Supporting Information Table S1. Intervention participants who were not randomized to MR were provided with a nominal $63 gift card on 6 occasions. This nominal gift card was not intended to compensate for the market equivalent costs of MR and was only provided during the first 6 months.

Outcome assessments

Physical outcomes were measured at the worksites by a research team that had no role in intervention delivery. Assessments were conducted at baseline and 6 months in all participants and at 12 and 18 months in intervention participants.

The primary outcome was fasting weight measured to ±0.1 kg together with body composition using an impedance scale (Tanita TBF300A, TANITA Corp., Tokyo, Japan). Height was measured at screening to ±0.1 cm (Seca 213, seca GmbH & Co KG, Hamburg, Germany). Baseline and follow-up assessments for all anthropometry measurements were made in duplicate using standard techniques, with an additional measurement for out-of-range variability. Blood pressure (BP) was measured to ±0.3 mm Hg in triplicate at 5-minute intervals after 5 minutes of quiet sitting (3 OMRON HEM-705CP Digital Blood Pressure Monitor, OMRON Healthcare Ltd, Muko, Japan) and the mean of the nearest 2 measures was used. Blood was collected in the fasted state by a finger stick and analyzed with point-of-care instruments (Alere Cholestech LDX System, Alere San Diego, Inc., San Diego, CA; Siemens DCA Vantage^® Analyzer, Siemens Healthcare Point of Care Diagnostics, Norwood, MA) for high-density lipoprotein (HDL), low-density lipoprotein (LDL), total cholesterol, triglycerides, and glucose.

Questionnaires included the RAND Medical Outcome Survey (MOS) 36-Item Short Form for quality-of-life subscales (32, 33), the RAND MOS Sleep Scale (34), the International Physical Activity Questionnaire (35), the Three-Factor Eating Questionnaire (with subscales for hunger, restraint, and disinhibition) (36), and the Food Craving Questionnaire—Trait (37). Dietary intake was assessed by unannounced 24-hour multiple pass recalls (38) for 3 days at baseline, 6 months, and 18 months. Adverse events were reported at site visits.

Statistical analysis

Based on our pilot study (4), we aimed to detect an 8.9 kg difference between intervention and control participants (pooled standard deviation [SD]: 5.0) with 80% power. With n=205 (intervention) and n=72 (control) completing participants in 12 sites for a cluster-randomized trial (average of n=23 per site), with ICC=0.01, and a two-sided alpha level = 0.05, we were powered >99% to detect the projected difference in weight between groups pairwise. For a 2.74-kg difference in 6-month weight loss (within-group SD: 5.0) between groups, we had 80% power.

Baseline characteristics were calculated for the three randomized groups. Associations between treatment and 6-month changes in each outcome were evaluated by multiple imputation using a Markov chain Monte Carlo (MCMC) method with 20 imputations (39) using linear mixed models adjusted for age, sex, ethnicity, worksite type, treatment, time, and a treatment-time interaction. Models of self-reported health outcomes and dietary and physical activity behaviors were additionally adjusted for baseline BMI. Models accounted for repeated measurements within person across time points and variation between worksites as a random effect to account for the cluster-randomized study design. The associations between 18-month changes in each outcome for the 2 intervention groups were evaluated by the same method. Adjusted means and 95% confidence intervals (CIs) were calculated for the HWL, HWL+MR, and control groups. Between-group differences were also calculated. Models of anthropometry are presented as absolute changes in each measurement from the baseline visit to facilitate interpretation of these results. All other models present absolute measurements at each time point. The primary analyses were ITT with multiple imputation. The range of data missingness for each outcome at each time point is provided in Supporting Information Table S2.

Model assumptions were assessed for each outcome variable. Eight variables had violations of either normality (skewness > 2; total: HDL cholesterol, triglycerides, and self-reported physical activity) or equal variances (significant Levene’s test and ratio of variance (σ¹/σ² > 1.5); total cholesterol, LDL cholesterol, glucose, fiber per 1000 kcal/day, and total food cravings) based on model residuals (40, 41). Natural log and Box-Cox transformations were applied for violations of normality and equal variances, respectively, for the variables reported above. Back-transformations were applied to adjusted means and 95% confidence intervals at each time point for presentation in tables to facilitate interpretation of these results, with P values corresponding to the models on transformed data.

A sensitivity analysis excluding observations with studentized residuals > 3 and <-3 was conducted to assess the impact of outliers. Two additional analyses were also conducted: 1) for the combined intervention groups compared to the control group at six months; and 2) for participants who completed a weight measurement at each study time point. Analyses were conducted using SAS version 9.4 (SAS Institute Inc, Cary, NC). All statistical tests were 2-sided, and a P value < 0.05 was considered statistically significant. Because only one comparison was made for the primary outcome and all comparisons for secondary outcomes were considered exploratory, no adjustment for multiple comparisons was applied. A full statistical analysis plan is provided in Supporting Information.

Results

Four universities, 1 non-profit organization and 3 for-profit companies were randomized to the intervention arm, and 1 university, 1 non-profit organization and 2 for-profit companies to the wait-listed control. All worksites performed their required functions for the trial (facilitating recruitment by providing information and facilitating on-site study visits by coordinating logistics with study staff), and no worksite dropped out after randomization. A total of 259 intervention participants and 76 controls were enrolled. Individuals who were laid off, resigned, or became pregnant were excluded from subsequent time points and the calculation of retention statistics (Figure 1). There were 5 pregnancies (3 in HWL, 2 in HWL+MR) and 5 medical withdrawals unrelated to study participation (1 in HWL, 3 in HWL+MR, 1 in control): a broken leg, a neck injury, a stomach virus, and medical leave with unspecified health problems in 2 cases.

Baseline characteristics of enrollees were comparable across groups with a broad range of education, marital status, and household income (Table 1). No P values for formal hypothesis testing are reported as they are not needed during a clinical trial.(42) Of the enrollees, 90.4% (n=303) identified as non-Hispanic (with 84.2% identifying as White, 7.6% as Asian, 5.9% as Black, and 2.3% as having more than one race), 3.6% (n=12) identified as Hispanic (with 42% identifying as White, 33% as having more than one race, and 25% declining to report their race), and 4% (n=20) declined to report their ethnicity (with 45% declining to report their race, 15% identifying as Asian, 15% as White, 15% as having more than one race, and 10% as Black). Participant retention decreased over time and differed between controls and intervention participants but not between intervention groups. Retention at 6 months was 95% for controls, 85% for HWL, and 75% for HWL+MR (P=0.002). Retention at 18 months was 72% in HWL and 64% HWL+MR (P=0.13). These values include individuals who were no longer employed by the worksite and women who became pregnant and thus were ineligible to continue (approximately 7% of the ITT cohort). Net retention at 18 months excluding these individuals was 80% and 68% for HWL and HWL+MR, respectively (P=0.03). Participants lost to follow-up before 6 months did not differ in age or BMI from retained participants (Supporting Information Table S2 and S3), with consistent results for participants who were lost prior to 18 months (Supporting Information Table S4).

TABLE 1.

Baseline demographic and anthropometric data for participants in three randomized groups during an 18-month behavioral intervention for weight loss.

	Control n=76	HWL n=130	HWL + MR n=129
Age	51.10 (8.7)	46.6 (10.2)	47.3 (11.0)
Female Sex	63 (82.9)	107 (82.3)	107 (83.0)
Height (m)	1.66 (0.08)	1.65 (0.08)	1.66 (0.08
BMI (kg/m 2 )	33.42 (5.66)	32.93 (5.94)	32.47 (5.08
Race
White	64 (84.2)	100 (76.9)	99 (76.7)
Black / African American	2 (2.6)	11 (8.5)	7 (5.4)
Other	6 (7.9)	16 (12.3)	18 (14.0)
Unknown	4 (5.3)	3 (2.3)	5 (3.9)
Hispanic
Non-Hispanic	66 (86.8)	117 (90.0)	120 (93.0)
Hispanic	1 (1.3)	8 (6.2)	3 (2.3)
Unknown	9 (11.8)	5 (3.9)	6 (4.7)
Education
< College degree	23 (30.3)	35 (26.9)	36 (27.9)
College degree	39 (51.3)	48 (36.9)	48 (37.2)
Graduate degree	11 (14.5)	46 (35.4)	44 (34.1)
Unknown	3 (4.0)	1 (0.8)	1 (0.8)
Marital status
Married	49 (64.5)	73 (56.2)	76 (58.9)
Previously married	13 (17.1)	20 (15.4)	19 (14.7)
Single	9 (11.8)	25 (19.2)	24 (18.6)
Living together	2 (2.6)	11 (8.5)	9 (7.0)
Unknown	3 (4.0)	1 (0.8)	1 (0.8)
Household income
$0 to $59,999	8 (10.5)	11 (8.5)	23 (17.8)
$60,000 to $99,999	22 (29.0)	50 (38.5)	35 (27.1)
$100,000 or more	38 (50.0)	66 (50.8)	70 (54.3)
Unknown	8 (10.5)	3 (2.3)	1 (0.8)
Worksite type
Commercial site	32 (42.1)	48 (37.2)	49 (37.7)
Nonprofit site	44 (57.9)	81 (62.8)	81 (62.3)

Abbreviations: BMI, body mass index; HWL, Healthy Weight for Living; MR, meal replacements.

¹Data reported as mean (standard deviation) for continuous variables and n (%) for categorical variables.

Primary outcome

Changes in measures of anthropometry are shown in Table 2. Mean weight loss at 6 months (primary outcome) was significantly greater in HWL (-8.8%; [-11.2, -6.4]; P<0.001) and HWL+MR (-8.0%; [-10.4, -5.5]; P<0.001) groups than the control group (+0.03%; [-2.5, 2.5]). Mean weight loss for both intervention groups combined was 8.4%. Comparable weight loss was observed in an analysis of participants who completed the 6-month intervention (-8.3% and -7.6% for HWL and HWL+MR, respectively). Decreases in BMI were comparable for HWL and HWL+MR (-2.9 and -2.7 kg/m², respectively) and were significantly greater than the controls (-0.05 kg/m²; P<0.001 for both comparisons). Consistent reductions in body fat percentage were also observed.

TABLE 2.

Changes in body weight and anthropometric measurements for participants in three randomized groups during an 18-month behavioral intervention for weight loss.

	Control	HWL		HWL + MR
	N=76	N=130		N=129
	Mean difference (95% CI)1	Mean difference (95% CI)1	P HWL v. C2	Mean difference (95% CI)	P HWL + MR v. C 2	P HWL v. HWL+MR2
Anthropometric outcomes
Weight (kg)
Baseline	95.44 (88.84, 102.04)	91.71 (85.97, 97.45)		93.07 (87.27, 98.87)
Δ 6 months	−0.15 (−1.59, 1.28)	−7.93 (−9.14, −6.73) ***	<0.001	−7.55 (−8.77, −6.33) ***	<0.001	0.66
Δ 12 months		−6.47 (−8.05, −4.90) ***		−6.54 (−8.35, −4.73) ***		0.96
Δ 18 months		−5.01 (−6.69, −3.34) ***		−5.06 (−7.05, −3.07) ***		0.97
Weight (% change)
Baseline	0	0	--	0	--	--
Δ 6 months	0.03 (−2.46, 2.52)	−8.79 (−11.20, −6.38) ***	<0.001	−7.99 (−10.44, −5.54) ***	<0.001	0.40
Δ 12 months		−7.37 (−11.69, −3.05) ***		−7.61 (−11.81, −3.41) ***		0.89
Δ 18 months		−5.57 (−9.80, −1.33) ***		−6.03 (−10.41, −1.64) ***		0.79
Body mass index (kg/m 2 )
Baseline	33.41 (31.04, 35.79)	32.15 (30.13, 34.18)		33.09 (31.05, 35.14)
Δ 6 months	−0.05 (−0.57, 0.46)	−2.85 (−3.29, −2.42) ***	<0.001	−2.74 (−3.18, −2.3) ***	<0.001	0.72
Δ 12 months		−2.3 (−2.87, −1.74) ***		−2.38 (−3.04, −1.71) ***		0.86
Δ 18 months		−1.73 (−2.33, −1.14) ***		−1.84 (−2.58, −1.1) ***		0.81
Body fat (%)
Baseline	37.4 (34.57, 40.22)	36.33 (33.93, 38.72)		36.87 (34.47, 39.28)
Δ 6 months	0.03 (−0.92, 0.98)	−3.68 (−4.49, −2.87) ***	<0.001	−3.09 (−3.94, −2.23) ***	<0.001	0.30
Δ 12 months		−3.55 (−4.45, −2.65) ***		−2.97 (−4.22, −1.73) ***		0.44
Δ 18 months		−2.33 (−3.25, −1.4) ***		−2.26 (−3.59, −0.93) **		0.94

Abbreviations: HWL, Healthy Weight for Living; MR, meal replacements.

¹Adjusted means and 95% confidence intervals are calculated by multiple imputation with 20 imputations using linear mixed models adjusted for age, sex, ethnicity, worksite type, worksite (as a random effect), randomized group, time, and the interaction between randomized group and time. The baseline values and six-month changes presented are derived from linear mixed models that include repeated measurements for baseline and six months only. The 12- and 18- month changes presented are derived from linear mixed models that include repeated measurements for baseline, six months, 12 months, and 18 months. The significance of changes from baseline within-group (e.g., HWL 6 months vs. HWL baseline) are represented by asterisks

^*P < 0.05

^**P < 0.01

^***P < 0.001.

²Pairwise P values for delta 6-, 12-, and 18-month measurements are for difference-in-difference (e.g., HWL 0 to 6 vs. Control 0 to 6).

Mean and individual values for percent weight loss of completing participants at 6, 12, and 18 months are illustrated by group in Figure 2; percentages achieving ≥5.0% and ≥10.0% weight loss are also shown. There was no significant difference between intervention groups at any time point with respect to the weight loss achieved. At 12 months, 52% of participants achieved ≥5% weight loss and 35% achieved ≥10%.

Figure 2:

Weight change in completing participants: (A) Mean % weight loss by group; (B) Individual values for % weight loss at 6 months; (C) Percent attaining ≥5% weight loss; (D) Percent attaining ≥10% weight loss.

Secondary outcomes

There was a small amount of weight regain at 12 months in both intervention groups (-7.4% and -7.6% for HWL and HWL+MR). Partial regain was also observered at 18 months. However, mean body weight at 18 months was significantly reduced compared to baseline (-5.6% and -6.0% for HWL and HWL+MR, respectively). Weight loss at these timepoints was not significantly different between the 2 groups. However, in an analysis of participants who completed the 18-month study, there was weight regain in both intervention groups resulting in a slightly higher total weight loss in the HWL group (-4.7%) compared to the HWL+MR group (-4.4%) (P=0.83). Comparable reductions in each intervention group were observed for BMI and body fat percentage.

Cardiometabolic measurements during the intervention are shown in Table 3. At 6 months, significant improvements were observed in both intervention groups compared to the controls with respect to triglyceride and glucose levels. The HWL+MR additionally exhibited reductions in total cholesterol compared to the control group (-0.20 mmol/L vs. +0.04 mmol/L, respectively; P=0.02) that were not observed in HWL. The HWL+MR group also significantly reduced their LDL cholesterol and BP from baseline, but these reductions were not significantly different from controls. The HWL group also had significantly lower BP at 6 months compared to baseline measurements and exhibited additional improvements in the total:HDL cholesterol ratio (-0.24; P<0.05) and HDL cholesterol (0.08; [0.03, 0.13]; P<0.01). Although improvements were observed within each intervention group from baseline, these changes were not significantly different between the two groups.

Table 3.

Cardiometabolic measurements for participants in three randomized groups during an 18-month behavioral intervention for weight loss.

	Control	HWL		HWL + MR
	N=76	N=130		N=129
	Adjusted mean (95% CI)1	Adjusted mean (95% CI)1	P HWL v. C2	Adjusted mean (95% CI)1	P HWL + MR v. C 2	P HWL v. HWL+MR2
Cardiometabolic outcomes
Total cholesterol (mmol/L) 3
Baseline	4.75 (4.43, 5.09)	4.89 (4.61, 5.19)		4.78 (4.50, 5.07)
6 months	4.79 (4.47, 5.13)	4.88 (4.59, 5.18)	0.61	4.58 (4.30, 4.87) **	0.02	0.04
12 months		5.01 (4.63, 5.41)		4.77 (4.36, 5.2)		0.37
18 months		5.06 (4.68, 5.48)		4.81 (4.41, 5.23)		0.30
Total cholesterol: HDL cholesterol 3
Baseline	3.7 (3.28, 4.19)	3.80 (3.42, 4.23)		3.71 (3.32, 4.14)
6 months	3.61 (3.19, 4.08)	3.56 (3.18, 3.98) *	0.41	3.44 (3.03, 3.90) *	0.37	0.86
12 months		3.71 (3.19, 4.30)		3.63 (3.10, 4.26)		0.97
18 months		3.72 (3.19, 4.33)		3.57 (3.02, 4.22)		0.67
HDL cholesterol (mmol/L)
Baseline	1.31 (1.16, 1.47)	1.35 (1.22, 1.48)		1.34 (1.21, 1.48)
6 months	1.37 (1.22, 1.52)	1.43 (1.3, 1.56) **	0.58	1.37 (1.23, 1.51)	0.48	0.15
12 months		1.41 (1.25, 1.58)		1.37 (1.20, 1.54)		0.49
18 months		1.45 (1.28, 1.61) **		1.42 (1.25, 1.59) *		0.68
LDL cholesterol (mmol/L) 3
Baseline	2.74 (2.45, 3.05)	2.78 (2.53, 3.05)		2.68 (2.44, 2.94)
6 months	2.61 (2.34, 2.92)	2.74 (2.49, 3.02)	0.36	2.7 (2.45, 2.97) **	0.42	0.06
12 months		2.7 (2.35, 3.09)		2.51 (2.26, 2.77)		0.94
18 months		2.71 (2.35, 3.11)		2.6 (2.21, 3.05)		0.22
Triglycerides (mmol/L) 3
Baseline	1.29 (1.05, 1.58)	1.32 (1.10, 1.58)		1.34 (1.12, 1.61)
6 months	1.45 (1.18, 1.78) *	1.25 (1.03, 1.52)	0.02	1.24 (1.01, 1.53)	0.02	0.74
12 months		1.47 (1.15, 1.88)		1.41 (1.13, 1.75)		0.52
18 months		1.53 (1.21, 1.94)		1.67 (1.29, 2.15)		0.55
Systolic blood pressure (mm Hg) 3
Baseline	129.10 (123.53, 134.67)	126.18 (121.35, 131.01)		127.98 (123.07, 132.88)
6 months	125.17 (119.54, 130.80) **	121.39 (116.43, 126.35) ***	0.64	123.10 (117.99, 128.2) ***	0.62	0.96
12 months		122.83 (117.04, 128.62) **		123.27 (116.77, 129.78) ***		0.59
18 months		122.28 (115.71, 128.86) **		128.51 (122.29, 134.74)		0.04
Diastolic blood pressure (mm Hg)
Baseline	82.87 (79.42, 86.31)	82.20 (79.19, 85.20)		83.12 (80.06, 86.18)
6 months	80.66 (77.18, 84.13) *	78.65 (75.59, 81.71) ***	0.26	79.99 (76.83, 83.16) ***	0.45	0.71
12 months		80.45 (76.61, 84.29) **		80.37 (76.25, 84.5) ***		0.44
18 months		80.91 (76.9, 84.92) *		83.35 (79.17, 87.52)		0.33
Glucose (mg/dL) 3
Baseline	98.26 (93.26, 103.92)	98.90 (103.92, 94.40)		98.86 (103.92, 94.40)
6 months	102.78 (109.09, 96.83) **	96.24 (100.89, 92.16) **	0.001	94.37 (99.47, 90.08) ***	<0.001	0.32
12 months		91.49 (99.47, 84.66) **		90.13 (100.89, 82.31) *		0.76
18 months		94.32 (103.92, 87.23)		92.56 (100.89, 86.35) *		0.67

Abbreviations: HDL, high-density lipoprotein; HWL, Healthy Weight for Living; LDL, low-density lipoprotein; MR, meal replacements.

¹Adjusted means and 95% confidence intervals are calculated by multiple imputation with 20 imputations using linear mixed models adjusted for age, sex, ethnicity, worksite type, worksite (as a random effect), randomized group, time, and the interaction between randomized group and time. The baseline values and six-month values presented are derived from linear mixed models that include repeated measurements for baseline and six months only. The 12- and 18- month values presented are derived from linear mixed models that include repeated measurements for baseline, six months, 12 months, and 18 months. The significance of adjusted mean changes from baseline within-group (e.g., HWL 6 months vs. HWL baseline) are represented by asterisks

^*P < 0.05

^**P < 0.01

^***P < 0.001).

²Pairwise P values for delta 6-, 12-, and 18-month measurements are for difference-in-difference (e.g., HWL 0 to 6 vs. Control 0 to 6).

³Variables required transformation natural log transformation (total cholesterol to HDL ratio, triglycerides) or as determined by Box-Cox method (total cholesterol^0.25], LDL cholesterol^0.25], systolic blood pressurê-0.75], glucosê-2.25]). Adjusted means and 95% confidence intervals at each time point were appropriately back-transformed to the original scale for ease of interpretation, with P values corresponding to the original models.

At 18 months, the HWL group had a sustained reduction in systolic BP that was significantly different from results observed in the HWL+MR group (-5 vs. -1 mmHg; P=0.05). In addition, both groups sustained significant improvements in HDL from baseline at 18 months. The HWL+MR also had sustained improvements in glucose that were not observed in HWL (-6.3 vs. -4.6).

Self-reported health outcomes are presented in Table 4. At 6 months, both intervention groups had significant improvements in self-reported health outcomes, with higher scores for self-reported general health and vitality and significantly lower sleep scores compared to controls. The HWL+MR group additionally reported improvements in mental health from baseline (2.96; [0.11, 5.8]); however, this improvement was not significantly different from the control group (2.48; [-0.87, 5.83]; P=0.83). Both intervention groups exhibited significant improvements in all self-reported health measures at 18 months compared to baseline values. These improvements did not significantly differ between the groups.

Table 4.

Self-reported health measurements for participants in three randomized groups during an 18-month behavioral intervention for weight loss.

	Control	HWL		HWL + MR
	N=76	N=130		N=129
	Adjusted mean (95% CI)1	Adjusted mean (95% CI)1	P HWL v. C2	Adjusted mean (95% CI)1	P HWL + MR v. C 2	P HWL v. HWL+MR2
Self-reported health outcomes
General Health (31, 32) (range 0–100) 3
Baseline	64.06 (22.23, 105.90)	62.98 (32.81, 93.16)		68.02 (37.84, 98.21)
6 months	63.65 (21.82, 105.50)	70.11 (39.91, 100.30) ***	<0.001	76.59 (46.36, 106.8) ***	<0.001	0.45
12 months		69.07 (−0.25, 138.39) ***		73.87 (4.48, 143.25) ***		0.99
18 months		67.29 (−2.03, 136.61) ***		73.00 (3.61, 142.39) ***		0.69
Mental Health (31, 32) (range 0–100) 3
Baseline	75.26 (−26.12, 176.63)	75.00 (2.30, 147.7)		76.70 (4.00, 149.39)
6 months	77.73 (−23.64, 179.11)	77.41 (4.69, 150.14)	0.98	79.65 (6.95, 152.36) *	0.83	0.79
12 months		77.49 (37.44, 117.55)		80.22 (40.11, 120.32) *		0.66
18 months		78.62 (38.56, 118.67) **		81.01 (40.88, 121.13) **		0.78
Vitality (31, 32) (range 0–100) 3
Baseline	59.18 (44.74, 73.63)	56.67 (45.56, 67.78)		58.3 (47.18, 69.42)
6 months	58.64 (44.28, 73.01)	67.48 (56.23, 78.73) ***	<0.001	69.37 (58.20, 80.54) ***	<0.001	0.91
12 months		62.85 (20.98, 104.70) ***		66.52 (24.63, 108.40) ***		0.48
18 months		61.79 (19.96, 103.60) ***		69.18 (27.27, 111.1) ***		0.10
Sleep Index I (33) (range 0–100) 3
Baseline	29.2 (−2.64, 61.04)	34.86 (24.34, 45.38)		30.81 (7.77, 53.85)
6 months	30.04 (−1.79, 61.88)	27.42 (4.33, 50.51) ***	0.006	27.10 (4.00, 50.19) **	0.04	0.50
12 months		27.96 (17.11, 38.82) ***		24.71 (14.01, 35.40) *		0.42
18 months		29.47 (18.84, 40.11) **		32.36 (9.31, 55.40) ***		0.17

Abbreviations: HWL, Healthy Weight for Living; MR, meal replacements.

¹Adjusted means and 95% confidence intervals are calculated by multiple imputation with 20 imputations using linear mixed models adjusted for age, sex, ethnicity, baseline BMI, worksite type, worksite (as a random effect), randomized group, time, and the interaction between randomized group and time. The baseline values and six-month values presented are derived from linear mixed models that include repeated measurements for baseline and six months only. The 12- and 18- month values presented are derived from linear mixed models that include repeated measurements for baseline, six months, 12 months, and 18 months. The significance of adjusted mean changes from baseline within-group (e.g., HWL 6 months vs. HWL baseline) are represented by asterisks

^*P < 0.05

^**P < 0.01

^***P < 0.001).

²Pairwise P values for delta 6-, 12-, and 18-month measurements are for difference-in-difference (e.g., HWL 0 to 6 vs. Control 0 to 6).

³Positive changes indicate improvements for general health and vitality, and negative changes indicate improvements (i.e., reduced problems) for sleep.

Changes in diet, eating behaviors, and physical activity are shown in Table 5. Changes in dietary composition towards programmatic goals were observed for both groups at 6 months compared to the control group, with significant reductions in total energy (mean change for both groups combined: -534 kcal/day; [-644, -424]; P<0.001), as well as glycemic index (-5.7; [-6.8, -4.6]; P<0.001) and glycemic load (-45.3; [-54, -36]; P<0.001). These changes co-occurred with significant increases in protein and fiber. All dietary improvements were retained at 18 months with no significant differences between HWL and HWL+MR. Significant improvements in cognitive restraint, disinhibition, and hunger were observed in both intervention groups at 6 months compared to controls, with sustained improvements at 18 months in both the HWL and HWL +MR groups. HWL+MR demonstrated an additional significant reduction in food cravings at six months compared to controls that was not observed in HWL (-13 vs. -4; P = 0.02). Self-reported physical activity measurements tended to increase over the intervention for both groups but were not significantly different from controls. The HWL+MR group exhibited significantly greater physical activity (minutes per week) at 18 months on average compared with the baseline measurement (1176 vs. 1917; P<0.001).

TABLE 5.

Dietary, eating behavior, and physical activity measurements participants in three randomized groups during an 18-month behavioral intervention for weight loss.

	Control	HWL		HWL + MR
	N=76	N=130		N=129
	Adjusted mean (95% CI)1	Adjusted mean (95% CI)1	P HWL v. C2	Adjusted mean (95% CI)1	P HWL + MR v. C 2	P HWL v. HWL+MR2
Diet
Energy intake (kcal/day)
Baseline	2332.03 (1398.88, 3265.19)	2308.81 (1629.82, 2987.8)		2223.35 (1544.26, 2902.44)
6 months	2180.83 (1246.72, 3114.93)	1724.51 (1041.61, 2407.42) ***	<0.001	1752.18 (1070.06, 2434.31) ***	0.002	0.22
18 months		1874.43 (928.12, 2820.75) ***		1855.61 (914.29, 2796.94) ***		0.40
Energy from carbohydrate (%)
Baseline	43.92 (30.55, 57.3)	45.12 (35.36, 54.88)		46 (36.27, 55.73)
6 months	44.15 (30.76, 57.55)	44.86 (35.08, 54.65)	0.74	44.96 (35.18, 54.74)	0.40	0.55
18 months		45.14 (37.97, 52.31)		46.77 (39.61, 53.94)		0.82
Energy from fat (%)
Baseline	35.33 (13.36, 57.31)	34.77 (18.97, 50.57)		34.71 (18.91, 50.51)
6 months	36.04 (14.06, 58.03)	32.01 (16.20, 47.82) ***	0.008	31.81 (15.99, 47.63) ***	0.007	0.91
18 months		33.58 (18.73, 48.44)		32.83 (17.94, 47.72)		0.69
Energy from protein (%)
Baseline	16.84 (10.55, 23.12)	16.47 (11.86, 21.08)		16.36 (11.74, 20.98)
6 months	16.59 (10.3, 22.89)	21.20 (16.55, 25.84) ***	<0.001	21.77 (17.07, 26.47) ***	<0.001	0.42
18 months		20.05 (15.09, 25.02) ***		18.86 (14.06, 23.66) **		0.29
Total fiber intake (grams/1000 kcal)
Baseline	9.27 (4.9, 19.8)	9.27 (15.91, 5.76)		9.84 (17.03, 6.08)
6 months	9.72 (21.01, 5.09)	20.29 (40.11, 11.34) ***	<0.001	18.71 (36.15, 10.63) ***	<0.001	0.17
18 months		13.98 (38.50, 6.23) ***		14.91 (42.22, 6.53) ***		0.87
Glycemic index
Baseline	58.91 (31.69, 86.13)	59.93 (40.40, 79.46)		59.73 (40.20, 79.26)
6 months	59.62 (32.4, 86.85)	53.99 (34.43, 73.56) ***	<0.001	54.36 (34.82, 73.90) ***	<0.001	0.62
18 months		55.53 (44.23, 66.83) ***		57.29 (45.98, 68.60) *		0.15
Glycemic load
Baseline	136.70 (104.3, 169.14)	145.37 (120.8, 169.98)		139.77 (115.1, 164.44)
6 months	130.83 (98.11, 163.54)	96.20 (70.82, 121.59) ***	<0.001	99.12 (74.10, 124.14) ***	<0.001	0.28
18 months		108.50 (78.02, 138.97) ***		112.60 (83.25, 141.95) ***		0.20
Eating behavior
Cognitive restraint (35)
Baseline	9.22 (2.95, 15.50)	9.66 (5.10, 14.23)		10.67 (6.10, 15.24)
6 months	8.92 (2.65, 15.19)	14.14 (9.56, 18.72) ***	<0.001	14.61 (10.01, 19.20) ***	<0.001	0.41
12 months		12.74 (−0.19, 25.67) ***		13.43 (0.49, 26.36) ***		0.79
18 months		12.07 (−0.85, 25.00) ***		12.99 (0.05, 25.93) ***		0.96
Disinhibition (35)
Baseline	7.65 (2.69, 12.62)	8.41 (4.79, 12.04)		7.61 (3.99, 11.24)
6 months	7.59 (2.63, 12.55)	8.41 (4.79, 12.04) ***	<0.001	6.06 (2.41, 9.72) ***	0.002	0.64
12 months		6.63 (3.00, 10.27) ***		6.61 (4.68, 8.53) *		0.32
18 months		6.72 (4.63, 8.82) ***		6.09 (3.75, 8.44) *		0.94
Hunger (35)
Baseline	5.61 (−1.16, 12.37)	5.75 (0.85, 10.64)		5.32 (0.43, 10.22)
6 months	5.28 (−1.49, 12.04)	5.75 (0.85, 10.64) ***	0.01	3.19 (−1.72, 8.11) ***	<0.001	0.29
12 months		4.12 (−0.79, 9.03) ***		3.63 (1.43, 5.82) ***		0.41
18 months		4.25 (2.17, 6.34) *		3.56 (1.36, 5.76) ***		0.17
Total food cravings score (36)
Baseline	100.15 (20.47, 239.98)	99.48 (36.42, 193.58)		99.13 (36.20, 193.11)
6 months	96.44 (18.82, 234.17)	90.87 (31.25, 181.57) ***	0.20	87.04 (29.02, 176.16) ***	0.02	0.33
12 months		92.28 (0.18, 385.49) ***		88.97 (0.36, 378.73) ***		0.57
18 months		91.92 (0.19, 384.75) ***		87.60 (0.45, 375.89) ***		0.44
Physical activity
Total physical activity (min/week) (34)
Baseline	1215.09 (192.56, 7667.45)	1108.23 (292.17, 4203.57)		1176.27 (310.13, 4461.34)
6 months	1179.92 (187.21, 7436.68)	1295.59 (340.11, 4935.39)	0.30	1355.63 (355.93, 5163.15)	0.34	0.93
12 months		1319.06 (211.88, 8212.01)		1645.56 (262.99, 10296.30) **		0.30
18 months		1380.36 (220.73, 8632.23) *		1917.2 (307.01, 11972.50) ***		0.10

Abbreviations: HWL, Healthy Weight for Living; MR, meal replacements.

¹Adjusted means and 95% confidence intervals are calculated by multiple imputation with 20 imputations using linear mixed models adjusted for age, sex, ethnicity, worksite (as a random effect), worksite type, randomized group, time, and the interaction between randomized group and time. The baseline values and six-month values presented are derived from linear mixed models that include repeated measurements for baseline and six months only. The 12- and 18- month values presented are derived from linear mixed models that include repeated measurements for baseline, six months, 12 months, and 18 months. The significance of adjusted mean changes from baseline within-group (e.g., HWL 6 months vs. HWL baseline) are represented by asterisks

^*P < 0.05

^**P < 0.01

^***P < 0.001).

²Pairwise P values for delta 6-, 12-, and 18-month measurements are for difference-in-difference (e.g., HWL 0 to 6 vs. Control 0 to 6).

³Variables required transformation natural log transformation (total physical activity) or Box-Cox transformation (fiber per 1000 kilocalorie^-0.25], total food craving^0.50]). Adjusted means and 95% confidence intervals at each time point were appropriately back-transformed for ease of interpretation, with P values corresponding to the original models.

Comparable results to those reported above were observed in a sensitivity analysis excluding observations with studentized residuals > 3 and <-3 with two exceptions. The statistically significant, within-group improvement in HDL cholestrol in the HWL+MR intervention from baseline to 18 months was attenuated after the exclusion of outliers. Furthermore, in the sensitivity analysis, there was a statistically significant improvement in triglycerides from baseline to 18 months in the same group (P<0.05).

Discussion

This randomized controlled trial evaluated a conceptually new lifestyle intervention approach for weight loss (26), implemented by a commercial provider with or without provided MR. Participants randomized to HWL (both HWL groups combined) achieved a substantial 8.4% mean weight loss at 6 months compared to a 0.03% gain in controls, with clinically significant mean weight loss and health improvements retained throughout the 18-month trial. The provision of MR did not increase weight loss but did reduce total cholesterol compared to HWL alone. The weight loss achieved by both intervention groups is 2–3 times the mean values typically reported for commercial programs previously tested in randomized trials and is judged to have above average effectiveness (3). Importantly, the results here were obtained in worksite employees, a population that previously had particularly low rates of success in randomized trials (2), and the new results are comparable to those obtained by pharmacotherapy (43). As HWL does not require daily food logging, which is a burdensome requirement of traditional lifestyle interventions (44), these results identify an alternative approach to behavioral treatment of obesity that employers can implement for obesity management in the workforce.

The mean weight loss achieved at 12 months was similar in both intervention groups (7.4% in HWL, 7.6% in HWL+MR). These values are also similar to our earlier report on this lifestyle intervention in a 12-month pilot worksite study (4) and an evaluation of the same intervention in a commercial implementation (10). All these values are notably greater than the 2.7% mean weight loss over 12 months typically measured in randomized trials of scalable lifestyle interventions in worksite employees (2, 3), the 3%-5% achieved by other commercial services (3), and the 4.2% reported for a non-randomized analysis of the scaled national DPP in community settings (45). Further, 36% of participants achieved ≥10% weight loss by 12 months, which is a recognized benchmark for strongly beneficial effects on multiple health parameters including remission of type 2 diabetes (5). These strong results suggest that the revised health behavior change model and/or novel implementation features of HWL supported programmatic success. In particular, participants did not have to perform food logging, a core feature of traditional lifestyle interventions (46). It may also be relevant that the default dietary composition of HWL recommendations decreased dietary glycemic load (47), and participants reported significant reductions in hunger, food cravings, and disinhibited eating behaviors. HWL’s emphasis on developing intrinsic motivation for enjoyable healthy food habits, rather than extrinsic motivation for weight loss via calorie counting and exercise, may additionally have played a role.

Further research is needed to determine whether the observed partial weight regain in this study could be attenuated with more behavioral support, especially as intervention meetings were reduced from weekly to monthly after the initial 6-month intensive phase. However, this suggestion does not negate the fact that mean weight loss in both HWL and HWL+MR groups remained in the clinically impactful range (5). A surprising finding was that the MR did not increase weight loss as seen in previous research (11, 12). We speculate that this may have been due to the higher rates of weight loss due to the behavioral intervention in this trial or to the provision of structured menu plans, which have previously been reported to be equivalently effective to provided food (48). Further studies are needed to investigate potential underlying causes.

By using a commercial intervention provider, this trial tested a realistic simulation of a scaled program. In addition, the flexibility to deliver the intervention by videoconference was an important advantage, allowing for interventions to be conducted outside work time. Today, this programmatic feature offers even greater benefits given the recent shift to remote work due to COVID-19. It is also noteworthy that the strong results obtained here were achieved without employees self-funding participation. If such programs could be provided by large employers there could potentially be a net return on investment through decreased health care costs, with additional anticipated benefits to employee productivity.

Study limitations include the fact that the majority of enrollees were non-Hispanic white females (24% of participants identified as other than non-Hispanic white), and despite the inclusion of for-and non-profit worktypes, the worksites were mostly large with a stable population of employees who had income levels above the national median. In addition to the lack of an ‘attention placebo’ for the control group, the study was powered for weight change but not for cardiometabolic or other health outcomes, a weakness mitigated by the well accepted association of intentional weight loss with health benefits (5). Further, relatively little tracking was conducted to determine participant adherence to different intervention components (e.g., meeting attendance, meal replacement use) out of concern for participant attritition, which tends to be high in worksite weight loss interventions (up to 88% in one study of 18 months duration (49)). Additional studies are therefore needed with expanded types of worksites and with data on measured health care expenses to determine health care savings. Nevertheless, this trial’s demonstration of clinically significant weight loss in a new and scalable intervention model in a challenging population is noteworthy, especially because randomized trials typically have low effect sizes relative to single-arm evaluations (50).

In summary, a novel lifestyle intervention implemented in a cost-effective group-distribution model, with in-person groups or videoconference, achieved 8.4% mean weight loss at 6 months in a challenging population of worksite employees, and clinically impactful mean weight loss and health benefits were observed throughout the 18-month trial. These results were achieved despite providing the interventions to participants free of charge. Participants randomized to receive MR in addition to the lifestyle intervention had similar weight loss with an additional benefit for total cholesterol and reduction in cravings. These results identify an effective new approach to weight management that can be used in worksites to reduce high rates of obesity in the workforce.

Study Importance.

What is already known?

• Traditional lifestyle interventions have had limited effectiveness in worksites. A new lifestyle intervention emphasizing hunger management and healthy food preferences without calorie-counting (Healthy Weight for Living, HWL) achieved strong results in a pilot study, indicating the need to test a scalable model.

• Use of meal replacements (MR) increases weight loss in traditional lifestyle interventions, but their effects in the HWL intervention are unknown.

What does this study add?

• In a randomized controlled trial of HWL, HWL+MR versus a wait-list control, with HWL provided to employee groups in person or by videoconference, participants enrolled in HWL and HWL+MR lost 8.8% and 8.0% body weight on average at 6 months, respectively; controls gained 0.03%.

• Some weight was regained by program enrollees, but clinically impactful weight loss was retained at 18 months (5.6% and 6.0% in HWL and HWL+MR, respectively).

How might these results change the direction of research or the focus of clinical practice?

• These results demonstrate that clinically impactful and scalable weight management is achievable in employees with a new lifestyle intervention approach, expanding options for reducing population obesity and health care costs.

• The results were obtained without requiring burdensome daily food logging—long considered a cornerstone of lifestyle interventions—suggesting novel avenues for research to increase the adoption of lifestyle interventions.