Study design
This is a quasi-experimental quantitative cohort study design. The general programmes participants will have pre-measurements (measured or self-reported weight at entry to BBM) and multiple post-measurements (measured weight at various intervals), and the FTC participants will have pre-post programme weight measurements. Evidence from the literature of the natural course of weight change in populations like the BBM participants with and without interventions will enable counterfactual comparisons. This study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement and a Standard Protocol Items: Recommendations for Intervention Trials (SPIRIT) checklist has been completed.
Setting
Attendees of BBM programmes in Auckland from 2021.
Participant recruitment
For the general programme, participants will be recruited through study advertisements on BBM’s social media platforms, and as posters in physical locations (the exercise centres). BBM trainers will promote the study to BBM attendees. All attendees over the age of 16 years will be eligible to participate.
For FTC group participants, all people registering for three 12-week programmes will be invited to enrol in the study. All FTC participants will be eligible, with the exclusion of repeat attenders already enrolled in the study.
Participants from both the general and FTC groups will be offered a koha (gift) of a $40 token at the 24-month follow-up. Participants will continue to be enrolled until the target samples are reached. Given recurrent lockdowns and restrictions on class sizes due to COVID-19, it is anticipated that the study will be extended by one year to enable 24 month follow-up of all participants.
Sample size calculations
The study has a finite population and has a quasi-experimental design of an existing intervention with no randomised sample, hence our sample size needs to take a pragmatic approach. We will recruit a many eligible participants as possible in the available timeFor the general group the target is to recruit 1000 BBM members, and for FTC. there are about 20 enrolees in each FTC programme, and some repeat the cycle, so the total number is estimated to be about 50 over three cycles.
Variables
For the general programme group, enrolment measurements will be height in metres and weight in kilograms (kg), and they will complete a Māori scale of health and well-being, Hua oranga, electronically. Their weight in kg will be recorded at six and 12 months, and at conclusion at 24 months, they will repeat Hua oranga and have a final weight taken. The primary outcome will be changes in weight, and the secondary outcome will be changes in Hua oranga scores. Note that first measurement is at enrolment in the evaluation and many participants at this stage will have been in the BBM programme for a period of time already. At enrolment, their self-reported heaviest weight, and weight at their first BBM visit will be recorded.
For the FTC group, in addition to the data collected above, clinical data from their health records will be sought. This includes diagnosis classifications, past history and current problems of any chronic diseases, especially type 2 diabetes, hypertension, gout, dyslipidaemia, cardiovascular disease, stroke, renal dysfunction, liver disease, osteoarthritis, gall bladder disease, chronic obstructive respiratory disease, and obstructive sleep apnoea. Blood and urine analyses laboratory results, prescribed medications, and other referral letters and discharge summaries will also be accessed. Their blood pressure at enrolment and at 12-weeks will also be recorded.
Data collection
Pacific research assistants will undergo training in recruitment and data collection and be provided with detailed step-by-step instructions. They will contact respondents to arrange a time and location to review the Participant Information Sheet (PIS) and gain consent, preferably at a BBM exercise centre. If requested, the PIS and consent form can be provided in te reo (Māori language), Samoan and Tongan. A log of all attempted communications with potential participants will be maintained. Once participants have consented, they will be invited to complete the online survey, which includes length of time since first joining BBM, weight at that time, frequency of attendance at BBM exercise sessions in the last three months, the Hua Oranga survey and demographic details (such as age, gender, ethnicity).
All participants will have their height and weight measured at enrolment by BBM staff and research assistants who will have undergone training using standardised protocols [11]. Participants will be weighed using Seca Scale Column Model 703 300 kg scales without footwear and light clothing only. Height is measured using the study’s stadiometer without shoes. Participants’ retrospective self-recorded weights, for example when they first entered the BBM programme, will also be collected.
Additional data will be collected for FTC participants. BBM trainers already measure and record blood pressure in this group, using an extra-large cuff calibrated sphygmomanometer. All FTC participants will be asked to consent to accessing their National Health Identification number and/or details of their primary healthcare provider / general practitioner to enable extraction of clinical details from their medical records: chronic disease diagnoses, diagnostic laboratory results and reports, clinic letters, e-referrals, prescribed medications, hospital appointments and admissions, and discharge summaries. Most New Zealand patients are enrolled in a Primary Health Organisation (PHO) through their registered general practice. PHOs collect primary care data from their member practices [12]. There are eight geographically dispersed and overlapping PHOs in the Auckland region. It is anticipated that most FTC participants will be enrolled with the three largest PHOs operating in South Auckland (where most BBM members live). Patient data will be extracted by these organisations on application using their data management processes. Where participants attend practices that belong to other PHOs, the clinical data will be sought via direct contact with individual practices.
Routine data for FTC participants collected by BBM will be included in the analyses. These data include age, frequency of exercise per week, whether they suffer from a respiratory condition or asthma, high blood pressure, blood disorder, diabetes, arthritis, epilepsy of seizures, past history of stroke, and whether they have been pregnant, had major surgery or hospitalised in the previous 12 months.
Statistical analysis
All statistical analysis will be done in R [13] using the lme4 package [14]. The primary outcome of interest is sustained weight loss. How successful BBM is at helping participants lose and maintain 5% weight loss will be a critical outcome measure, however, the effectiveness of BBM compared to other counterfactuals will also be assessed.
The general programme cohort will include existing and new entrants. For existing members we will also have self-reported weights at their first BBM session, which will be considered the baseline or ‘time zero (T0)’ weight with subsequent times measured from then. For example, for a person who started at BBM seven months before enrolment in the study, their reported weight at BBM entry will be T0, their enrolment weight (measured by the research team at study enrolment) will be T7, their 6-month weight will be T13, 24-month weight will be T31. Typically, the pattern of weight loss with weight loss programmes is tick-shaped with a weight loss nadir at 6–12 months followed slow return back towards T0 weight or beyond. Estimated duration of the study will be 24 months of follow-up from enrolment.
Determination of the counterfactuals for comparison
Since there is no enrolled comparison group in this study, the counterfactuals to be included to determine how effective BBM is will be threefold:
-
1.
Versus similar New Zealand population. The estimated mean weight change of ethnic/gender/age groups will be derived from the last five years of NZ Health Surveys [15] and these will be compared with the general programme groups participating in BBM. Note that, longitudinally, weight change in a subpopulation is the additive effects of secular changes to weight (ie each age group gets a little heavier each year) and age-related weight change (ie the person moves to an older age group which has a higher average weight).
-
2.
Versus control groups in other similar studies. A literature review [16] has identified five weight loss studies with predominantly Pacific or Māori participants that had control groups (total N = 955, follow-up 3–24 months) [17,18,19,20,21]. Their weight change trajectories will be compared to the BBM general programme participants.
-
3.
Versus other interventions. The literature review noted above [16] identified 18 studies with various interventions with weight change outcomes [17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34]. These combined results show how effective other programmes have been. Since the duration of interventions varies (3–24 months), weight loss is not linear.
A mixed method meta-analysis was conducted using the metaphor package [35] in R. Studies were only included in this analysis where the difference with baseline (time 0) was published together with their standard error, standard deviation, or confidence interval (to enable calculation of the standard error). Figure 1 shows the forest plot of the included studies.
The moderator variables were intervention, time, and interaction intervention and time, and the response variable was the weight change in time from a baseline of zero for both control and intervention. The random variable was the interaction between the individual studies and the intervention. Spline functions are formed by joining polynomials together at fixed points called knots. The number and position of knots in the spline function was found by evaluating the Akaike Information Criterium (AIC) using all possible combinations of placement of two knots [36]. The best AIC was found with one knot at two months. Splines were compared with polynomials, but the fit of a spline function was superior to them. A spline function with one knot could be fitted within a mixed effects meta-analysis to describe the initial weight loss followed by increase in weight (Fig. 2).
A funnel plot of the standard errors against the residual values after fit was symmetrical, with a non-significant Eggers test (p = 0.8505), indicating no evidence of publication bias in the articles used in the meta-analysis (Fig. 3).
Weight for both the general BBM and the FTC cohorts will be analysed as with a general linear mixed model for repeated measures, with person (participant) as the random variable and time as the fixed variable. The two cohorts will be combined because some general BBM members may previously have been FTC participants, and some FTC participants subsequently become general members. The weights will be logarithmic transformed before analysis, if the residual analysis indicates that. The comparison of our study data with the tick shaped curve prediction model generated from the literature (Fig. 2) will be done by comparing specific points in time and calculating the t-probability of being equal on basis of their 95% confidence intervals.
The FTC group will be considered as a series of three cohorts of about 20 participants each entering the FTC programme. An estimate of 80% recruitment into the study will give 50 participants who will have baseline and follow up data. There is also no parallel comparison group for this part of the study. However, their weight change trajectories can also be compared with the counterfactuals noted above. FTC data will be analysed with a Difference in Difference (DID) analysis using the estimated covariance matrix from the repeated measures analysis to obtain the corrected differences (Treatment – Baseline) – (Control – Baseline) as an estimate of the effect of the programme.
Analysis of hua oranga scores
The hua oranga scores will be analysed with a general or generalised linear mixed model with proper distribution (gaussian, binomial or poisson) depending on the characteristics of the distribution of the outcomes. The random term is the person (participant). The scores per domain will be analysed as well as the sum of all the scores. Because we have pre- and post-data, all analyses will be followed with a DID analysis to obtain the corrected differences.
Data management
All data will be stored securely on a password-protected shared drive on a University of Auckland with access restricted to the relevant researchers conducting the data collection and analysis.
Data monitoring
As this is not an intervention study but observation of an existing programme, there is no need for a data monitoring committee nor strategies to stop a trial. Participants can freely chose when they wish to engage with BBM and repeated measurements of their weight and quality of life at the beginning and end of the study has low risk of harm.
Dissemination
The findings will be published in appropriate peer-reviewed publications, as well as disseminated through community fono and hui (meetings) and conferences. BBM will further disseminate the findings via their social network platforms to BBM members and followers. All researchers who meet the criteria will be authors and no professional writers will be used.