A quasi-experimental trial comparing weight-related anthropometric changes following a six months intervention against a control group receiving no intervention, with a follow-up at 12 months was planned. The rationale and method for the WEHI trial are described in detail elsewhere . This section briefly describes the methodology.
Participants and sample size
Three distinctly different geographical regions in NZ were selected for recruitment: an urban Māori population (Palmerston North), a small town/rural Māori population (Northland) and a Pacific Island community in a major NZ city (Auckland). Seven teams of seven members for each region (total N = 147) were recruited for the intervention. We planned to recruit 150 wait list control participants, who did not receive any intervention (see below).
Eligibility and exclusion criteria
The eligibility criteria were Māori or Pacific people, aged 16 years of age and above, having a body mass index (BMI) ≥ 30 kg/m2, and being at risk of or having developed T2DM or cardiovascular disease (CVD). People who were of any other ethnicity, younger than 16 years, pregnant or were breastfeeding and had type 1 diabetes were excluded. To control for potentially confounding factors, using nicotine from any source (a known appetite suppressant ) or smoking cannabis (with debated effects on weight ) were additional exclusion parameters.
The intervention group was purposively recruited by Māori and Pacific health providers. They used convenience sampling to find participants, advertising through their existing networks to staff and in their communities. Though recruitment of intervention and control participants occurred concurrently, recruitment for control participants was done over an extended time period (four months). Potential participants who declined the intervention were invited to enrol as a control participant, or they responded directly to invites to be a control participant. In lieu of receiving the intervention, control participants were offered one entry in a prize draw (one per region) for petrol vouchers (up to $50) for completing a questionnaire at baseline, six-month follow-up and 12-month follow-up.
The WEHI intervention included four main components: group support, competition, financial incentives and Internet-delivered education and support. Teams were to meet regularly to facilitate their completion of as many of the competition’s weekly and nine daily challenges as they could to amass points. The activities were designed to prompt physical activity, increase consumption of vegetables, reduce consumption of added-sugar drinks and foods, and encourage retention in the competition (see Glover et al.  for a fuller description and evaluation). Each team had seven members and were self-directing, however, they could receive support from regional intervention workers. Support was also provided via the intervention website which provided weight-loss tips and answers to questions on, for example, staying motivated, making choices and increasing physical activity. Each team also had a dedicated team page where they could post photos, recipes and comments. This was also publicly visible as was a competition scoreboard displaying the progress of each team. In each region, three cash prizes were offered for: the greatest progress at two months (NZ$1000), greatest progress at four months (NZ$1000) and greatest progress at six months (NZ$3000). Progress was based on the number of team members who had lost ≥4 kg weight in the preceding two months plus the number of team members who had lost ≥3 cm in waist circumference during the same period, plus the team’s position on the competition scoreboard, which was calculated by tracking team participation and completion of daily and weekly challenges. The prizes were paid to the team’s nominated charity or community organisation.
Anthropometric measurements (height, weight and waist circumference) of all participants were performed by the researchers or research assistants. The equipment used for measurements were a SECA813 digital floor scale , a SECA portable stadiometer height rod  and a SECA ergonomic girth measuring tape . Based on those measures BMI was calculated as (weight in kilograms (kg) / (height meter (m)2)).
Questionnaires (see supplementary file) were self-administered at baseline, 6 months and 12 months to measure changes in Ministry of Health Eating and Activity Guidelines  eating goals, such as eating two servings of fruit a day, activity levels, perceived acceptability of WEHI. Other questions at baseline asked about previous use of dieting/weight-loss programmes, demographic characteristics and food security. In addition to food security, being a holder of a community services card was used as a proxy indicator of socioeconomic status. Community services cards enable people on low incomes to receive discounts, for instance on their healthcare and cost of medications. The questions were drawn or adapted from other surveys, such as the Adult Nutrition Survey, Ministry of Health adult health survey, food security index and relevant literature. The questionnaire was pilot tested with six Māori and Pacific people with BMI > 30 from among the researchers’ networks.
The primary outcome was mean weight loss at six-month follow-up. Secondary outcomes included the proportion of participants losing at least 5 and 10% of baseline weight, change in waist circumference and BMI at six-month follow-up. We also looked at weight-loss outcomes in the intervention group at 12-month follow-up.
A simple descriptive analysis, calculating counts and percentages, was performed on ordinal data variables. Baseline weight, waist circumference and BMI were not normally distributed and so non-parametric tests were used to examine differences between groups. Where indicated, additional analyses were calculated for subcategories. For continuous variables, mean, standard deviation and maximum and minimum values were calculated. Differences between groups were compared using an univariate general linear model, adjusting for baseline body weight. To assess differences in the proportion of baseline body weight lost, a chi-square test was used. An intention-to-treat (ITT) analysis was used. We used a baseline-observation-carried-forward-analysis for those who were lost to follow-up. We also undertook a complete-case analysis. To calculate change in eating behaviour a non-parametric Related-Samples Wilcoxon Signed Rank Test was performed to detect if there was a significant difference (p < 0.05) between eating behaviour at baseline versus six months. All analyses were undertaken using IBM SPSS.
This study was approved by the NZ Ministry of Health’s Northern B Health and Disability Ethics Committee (16/NTB/101).