Interventions designed to support individuals to change their health behaviours, such as behavioural weight management interventions, typically require a high amount of personal agency (such as time, money, and education) to be effective. As a result, these interventions may benefit advantaged groups with greater resources more than those who are less advantaged, and consequently may be inequitable and exacerbate health inequalities. However, there is currently limited empirical evidence about whether behavioural weight management interventions contribute to inequalities in overweight and obesity.

This thesis has examined whether there are inequalities in the attendance and effectiveness of behavioural weight management interventions using data from of randomised controlled trials. Across four projects, I considered inequalities in these interventions in a range of individual-level characteristics captured by the PROGRESS-Plus framework (Place of residence, Race/ethnicity, Occupation, Gender/sex, Education, Socioeconomic status, Social capital, Plus other factors where discrimination may occur such as age and sexual orientation) and across a range of different outcomes (uptake, adherence, weight loss, weight loss maintenance, eating behaviour, physical activity, wellbeing, and mental health).

My first project was a systematic review of inequalities in the uptake of, adherence to, and effectiveness of behavioural weight management interventions (thesis Chapter 2). I narratively synthesised data from 103 trials and summarised findings in harvest plots. Fifty-six trials considered inequalities in relation to at least one of intervention or trial uptake (n=15), intervention adherence (n=15), trial attrition (n=32), or weight outcome (n=34). Most trials found no inequalities gradient for trial uptake, intervention adherence and trial attrition. If a gradient was observed, it favoured those considered ‘more advantaged’. A limitation of this systematic review is that the individual analyses in each trial are unlikely to be sufficiently powered to detect inequalities in intervention effectiveness should they exist, and that the data and analyses reported were not similar enough to be meta-analysed. A way to mitigate these limitations is to access individual-level data from trials so that data can be harmonised and appropriate analyses conducted across all trials.

Consequently, my second project was a two-stage individual participant data meta-analysis (IPD-MA) of UK-based trials of behavioural weight management interventions (thesis Chapters 3-5). I obtained and harmonised data from 12 out of 14 eligible trials and reanalysed each trial to consider if there were inequalities in 1) the effectiveness of interventions at 12-month follow-up, and 2) attendance of these interventions. I then meta-analysed the relevant coefficients. The difference in weight loss between the intervention and control groups was 0.98kg larger for male compared to female participants, despite lower overall weight loss in men. The difference between intervention and control weight change was 2kg larger for participants of a White ethnicity compared to those from other ethnic groups, showing that these interventions may contribute to inequalities in obesity by ethnicity. I also found some evidence that behavioural weight management interventions had less effect in those who lived in areas of higher deprivation, although this was not consistent across the deprivation gradient. The intervention effect did not differ by occupation, education, socioeconomic status (annual household income), social capital (marital status), and age. There was no association between attendance and ethnicity, occupation, gender/sex, education, socioeconomic status (IMD and annual household income), social capital (marital status), and age.

The first two studies in this thesis focused on inequalities in effectiveness of behavioural weight management interventions at the 12-month follow up time point. Generally, there is a paucity of follow up data beyond this time point in UK-based randomised controlled trials. However, in the WRAP (Weight loss Referrals for Adults in Primary Care) trial there was an opportunity to consider if there are inequalities in the weight trajectories of participants up to five-year follow up. In the third study of this thesis, I conducted cohort analyses using data from the WRAP trial to consider if there are inequalities in weight change from one- to five-years post intervention baseline (thesis Chapter 6). This is important as weight regain often occurs post intervention. I did not find evidence of inequalities by ethnicity, occupation, gender, education, nor socioeconomic status, though older age was associated with lesser weight regain or greater weight loss between one- and five-years post-intervention.

The final project in this thesis used data from a trial of a web-based intervention to support weight management during Covid-19 (SWiM-C, thesis Chapter 7). I codesigned the baseline demographics questionnaire for this trial; as a result, data across a broad range of PROGRESS-Plus criteria were collected, such as access to green space, food insecurity, and perceived social support. I considered if there were differential effects of the SWiM-C across a range of outcomes including weight, eating behaviour, physical activity, wellbeing, and mental health. I found that the intervention had a differential effect by marital status; it had a greater effect on weight, eating behaviour, and mental health in those who were living as single versus those who were married, in a civil partnership or cohabiting. I also found evidence that the SWiM-C intervention had a greater effect on weight in those who were unable to work due to sickness or disability versus those who were working full-time or were self-employed. I did not find evidence of a differential intervention effect by the other PROGRESS-Plus characteristics on weight, nor did I find consistent evidence of differential intervention effectiveness on the remaining outcomes.

Overall, this thesis makes several important contributions to the field. First, I identified characteristics in which there are inequalities in the effectiveness of behavioural weight management interventions. I found that these interventions have a greater effect in men (versus women) and that this effect was produced by men gaining weight overall in the control group; weight loss in the intervention group was similar between men and women. I also found that these interventions were more effective in participants from White, versus minority, ethnic groups, meaning they likely exacerbate existing health inequalities by ethnicity. Second, there were several of the PROGRESS-Plus characteristics I did not find evidence of inequalities in behavioural weight management interventions by, such as occupation, education, and socioeconomic status (particularly annual household income). Similarly, I did not find evidence of inequalities in attendance to behavioural weight management interventions. Third, I assessed the representativeness of trial participants and reporting of inequality related data in trials of behavioural weight management interventions. I found that, beyond gender or sex and age, the PROGRESS-Plus characteristics were inconsistently reported and that trials of behavioural weight management interventions were not representative of the population living with overweight and obesity. The fourth contribution of my thesis to the field is that it adds data to the field where there was a previous paucity of data; this is particularly the case in Chapters 6 and 7 which were the first analyses of inequalities in weight regain following weight loss intervention (Chapter 6) and in outcomes other than weight (Chapter 7).

Future research should further explore the inequalities by ethnicity to ensure the effects of behavioural weight management interventions do not widen existing health inequalities, and how to better engage men into generic behavioural weight management interventions given their effectiveness.