Background:

Behavioural weight management interventions are the primary treatment for obesity in the United Kingdom. These interventions focus on diet, physical activity and behaviour change, and are typically delivered over a period of 12 weeks. Although National Institute for Health and Care Excellence guidance makes recommendations on the content of behavioural weight management interventions, there are substantial variations in practice. As a result, what constitutes the most effective composition of behavioural weight management interventions is unclear.

Objective:

To determine the effectiveness of different types of behavioural weight management interventions in achieving weight loss, using individual participant data from randomised controlled trials and real-world services.

Design:

A network meta-analysis of individual participant data.

Setting:

Behavioural weight management interventions delivered in the community.

Participants:

Anonymous individual participant data of adults (> 18 years), living in the United Kingdom and attending behavioural weight management interventions in the real world (n = 76,201) and randomised controlled trial’s (n = 4051).

Main outcome measure:

Mean change in weight at 12 weeks.

Methods:

Two-staged Bayesian network meta-analysis of individual participant data from included randomised controlled trials and real-world services was performed. Risk of bias was assessed for randomised controlled trials using Cochrane Risk of Bias 2.0. Prior to analysis, received data were checked, for consistency with the requests and cleaned for all anomalies.

Results:

All behavioural weight management interventions resulted in weight loss compared to usual care. In the randomised controlled trials, the 52-week weightloss programme referrals for adults in primary care (WRAP) with participants attending intervention achieved the greatest weight reduction at 12 weeks (mean difference = −2.58 kg, 95% credible interval −3.19 to −1.96). However, when a male-only intervention (football fans in training) was included in a sensitivity analysis, it demonstrated the largest short-term weight loss (mean difference = −4.65 kg, credible interval −5.24 to −4.07). In the real-world services, several programmes achieved substantial weight loss, with greater programme attendance associated with improved outcomes.

Conclusions:

The behavioural weight management intervention in both real-world services and randomised controlled trials are effective for weight loss, but there is a variation in the weight loss achieved at the end of active weight loss period depending upon the structure of intervention and participant engagement.

Future work:

Dismantling the interventions into component parts will help determine which components or combination of components are associated with greater weight loss.

Funding:

This article presents independent research funded by the National Institute for Health and Care Research (NIHR) Health Technology Assessment programme as award number NIHR129523.

Eligibility criteria

Data from RCTs and RWSs on adults > 18 years with a body mass index (BMI) > 25, that would allow the assessment of effectiveness of BWMIs, were included.² All BWMIs were UK-based and were compliant with NICE guidance (multicomponent – diet, physical activity, behaviour change techniques, duration ≥ 3 months/12 weeks, weigh-ins at least every 2 weeks, dietary targets). BWMIs were excluded if designed exclusively for a specific single medical condition (e.g. diabetes, pregnancy, post partum or families).

Only RCTs which were UK based, completed within the past 10 years and agreed to share IPD data, were included. A minimum of 40% completion rate of the active weight loss (AWL) sessions and a minimum 60% of participants’ 12-week weights recorded were required.

For data from RWSs, only data from participants who had completed the programme prior to the coronavirus 2019 (COVID-19) pandemic (March 2020) were included. Only data sets containing ≥ 500 participants (data sets with fewer participants were included if they were iterations of the same service) were included.

Identifying studies

Specification of outcomes and effect measures

The primary and secondary outcomes and definitions were derived from the outcome and instrument set developed by expert consensus (including patient and public involvement) for analysis by Mackenzie et al.⁸ (COMET registration 1056) (see Appendix 1, Table 3). These were applied to assess weight loss, attendance and completion over 12 weeks of active intervention.

Risk of bias and quality appraisal

Within-study bias was assessed by two researchers using the Cochrane Risk of Bias tool 2 for RCTs.¹¹ Consistency between IPD and trial publications and any important baseline imbalance across the trials were examined. Two researchers independently examined received data from RWSs for missing, duplicated, or outlying values, and for internal consistency. (Missing information for weight at 12 weeks for both RCTs and RWSs is detailed in Appendix 5, and Tables 8 and 9)

Synthesis

A two-stage Bayesian IPD network meta-analysis (NMA) was conducted to estimate the mean change in weight (kilogram), when comparing different types of BWMI and usual care. In addition, NMA using RCT data were also conducted on other outcomes (see Appendix 3, Figure 11 and Appendix 6, Figures 12–14):

• mean change in BMI
• percentage change in weight
• proportion of participants achieving ≥ 5% weight change
• proportion of participants achieving ≥ 10% weight change.

In the first stage, the IPD from each RCT were used to fit generalised linear regression models to estimate adjusted study-specific intervention effect. We examined potential baseline imbalances between treatment arms, specifically testing for differences in baseline weight, age, sex and height using a stepwise approach. The final model included only covariates that significantly contributed to explaining the variance in weight change; the final adjusted model accounted for baseline weight and treatment arm.

In the second stage, these adjusted estimates were synthesised in a NMA using fixed-effect model. The analysis also accounted for the correlation structure produced by trials with > 2 arms. Pooled estimates of intervention effects were calculated as mean differences (MDs) and odd ratios along with their corresponding 95% credible intervals (CrIs). These were ranked and the probability that each intervention is best was estimated for each outcome.

Consistency was assessed based on the difference between direct and indirect estimates using node-splitting approach.¹² The goodness of fit for the model to the data was assessed using posterior mean residual deviance and leverage plots.

Underlying assumptions

Key assumptions in a NMA are that the indirect comparisons made between two treatments are a feasible comparison and that the indirect comparisons are consistent with existing direct evidence. It is assumed that for all the treatments in the NMA, it is feasible to randomise them in a same trial and if they are not the treatment arm in a trial, are missing at random.¹³

The additional underlying assumption for NMA of RCTs was a homogeneous control arm, that is the control group in the trials were combined as usual care. An additional underlying assumption for the NMA of RWSs, which were single-arm studies, was a control arm with no change in weight to simulate a comparative scenario. This represents the change in weight from baseline following enrolment to the weight management programmes.

Dealing with missing data for weight measures at 12 weeks

All outcomes were measured at 12 weeks. Twelve-week outcomes cover the AWL phase; generally, interventions only run for 12 weeks, and longer-term weight loss maintenance is not funded.

Dropout and missing data from weight management interventions is not a random event; generally, people stop attending when they are not finding the programme effective, either initially or due to later weight regain. Data collection is often linked to reimbursement for the provider. Participants who did not have data for age and height were excluded from all analyses. Lack of a recorded weight was assumed to mean ‘did not attend’ and only sessions with recorded weights were counted as attended.

For RCTs, the information of weights for the participants was available for baseline and 12 weeks only. Assumption of baseline observation carried forward (BOCF) was used for imputing missing weight at 12 weeks and presented as the main analysis.

For the RWSs, the information regarding recording of weight was available for baseline and final weight was recorded at the last attended session which varied from individual levels. For RWSs, missing data were addressed with BOCF in the main analysis alongside a sensitivity analysis addressing the missing final weight with last observation carried forward (LOCF), that is when 12-week weight was missing, weight recorded any time before 12 weeks was used to impute for weight at 12 weeks.

Sensitivity analysis

For the RCTs NMA, the following additional sensitivity analyses were conducted.

• Analysis to explore the impact of including RCTs that potentially contribute to substantial heterogeneity within the network.
• Network meta-analysis using trial calibration method, where the estimates from RCTs were adapted to reflect a target population with similar characteristics to that of those who participated in the RWSs. The mean participant-level covariate information from the RWSs was used to adjust for participant-level covariables in the RCTs.¹⁴ Specifically, the overall mean age, baseline weight, height and sex from the RWS populations were computed and used to centre the corresponding covariates in the RCT populations. A stepwise modelling approach, consistent with the main analysis, was then applied to estimate the main effects.

For RWSs, sensitivity analysis was conducted after imputing missing weights with the LOCF assumption.

Results were reported according to PRISMA guidance.⁹

Ethical approval

Favourable ethnical opinion was provided by Lancaster University (FHMREC reference: FHMREC20085).

Anonymisation

Full blinding was not possible in this study due to the nature of the data collected and the support required for data transfer. However, each intervention was provided with a unique site ID and the advisory group and management group only received results by study ID, with the principal investigator and research associate the only team members who had the ability to unblind the results.

Description of studies

Randomised controlled trials

Of the trials identified, six trials agreed to share participant-level data. Therefore, the analysis included six UK-based multicentred RCTs totalling 4051 adult participants living with obesity or overweight (BMI > 25 kg/m²). The mean age of the participants across the trials ranged from 46 to 64 years. For mixed-gender trials, the male percentage ranged from 30% to 74%, with one trial enrolling female-only participants and one trial aiming for male-only participants. The mean height ranged from 1.65 to 1.75 m and mean baseline weight ranged from 80.8 to 110 kg (Table 1).

TABLE 1

Characteristics of included randomised controlled trials (RCTs)

These trials evaluated the effectiveness of BWMI targeting individuals living with overweight or obesity. The interventions included a web-based behavioural weight management programme with either remote (POWeR + R)³ or face-to-face (POWeR + F)³ nurse support, a task-based group programme tailored for disadvantaged communities [weight action plan (WAP)],⁴ a programme engaging men through professional football clubs [football fans in training (FFIT)],⁵ weightloss programme referrals for adults in primary care with participants attending either 12 (WRAP 12) or 52 (WRAP 52) weeks of sessions, personalised lifestyle coaching through trained volunteers in leisure centres, initiated at breast screening clinics (ActWELL)⁶ and face-to-face lifestyle counselling sessions (BeWEL).¹ These interventions were compared with a range of control interventions including waiting lists, self-help weight management booklets, cancer prevention booklets with information about lifestyle modification and web-based information about healthy foods. These were combined as usual care (see Appendix 4, Table 6).

Real-world services

The participant-level data were shared on 19 programmes for 76,201 adult participants living with overweight or obesity (BMI > 25 kg/m²). The mean age of participants attending the programmes ranged from 36 to 60 years and mean height ranged between 1.62 and 1.79 m. For mixed-gender programmes, percentage of male participants varied from 5% to 33% with majority of the services having higher female participation. There were two programmes (6 and 10) with male-only participants and one programme (5) with female participants only (Table 2). The participant attendance varied from 5.15 to 33 weeks, highlighting the fact that not all the participants stayed for 12 weeks in all the services. The programmes’ interventions provided dietary and physical activity advice and targeted behaviour change through various behaviour change techniques over 12-week AWL period. The interventions differed in content and delivery mode (see Appendix 4, Table 7).

TABLE 2

Characteristics of Real World Services (RWSs)

Risk-of-bias assessment for randomised controlled trials

Based on the publication, two trials^15,16 had an overall high risk of bias for not blinding the outcome assessment and for concerns due to higher numbers of participants missing from intervention arm of one study. Another study¹⁷ did not report the results after imputing the missing data and was rated as having some concerns for the risk of bias (see Appendix 2, Figures 8 and 9).

Overall, IPD shared by RCTs were found to be consistent with published RCTs.

Network meta-analysis

Randomised controlled trials

Eligible participant-level data were shared by six randomised trials, which formed a disconnected network (Figure 1a) with 10 interventions including the usual care intervention. One study¹⁸ was disconnected from the network as the comparator arm could not be combined with the usual care group. The connected network (see Figure 1b) is a star-shaped network of five trials evaluating eight interventions including the usual care group. There are two loops in the network formed by the three-arm studies. However, one study within the connected network¹⁷ evaluated intervention designed to target a specific population and was excluded from the primary analysis. The impact of this study was examined in the sensitivity analysis.

FIGURE 1

(a) Disconnected network of RCTs with six included studies; (b) network map of RCTs that were included in the main analysis nodes represent the weight loss intervention programmes used in the RCTs. PNI, practice nurse intervention. Numbers on the edges (more...)

Real-world services

Participant-level data eligible for the analysis were shared by 19 services. After assuming a control group with no change in weight from baseline, to simulate a comparative scenario, NMA was performed. Network of RWSs (Figure 2) was a star-shaped network with no loops and included all 19 interventions.

FIGURE 2

Star-shaped network of RWSs anchored to assumed control arm (0). Numbers represent weight loss interventions provided at 19 different centres that shared IPD. Details of the interventions are mentioned in Appendix 4, Table 7.

The transitivity assumption held, though there were two male-only programmes and one female-only programme, but these were due to selection of participants and not for a specific population.

Change in weight (kilogram)

Network of randomised controlled trials

The connected network of RCTs compared seven treatments including usual care. All the BWMIs led to reduction in weight as compared to usual care. The 52-week WRAP intervention showed the maximum weight loss at 12 weeks (MD = −2.58, CrI −3.19 to −1.96), closely followed by 12-week WRAP¹⁹ (MD = −2.26, CrI −2.88 to −1.65), POWeR + F (MD = −1.69, CrI −2.39 to −1.00) and POWeR + R (MD = −1.50, CrI −2.19 to −0.80) interventions. The BeWEL¹⁶ (MD = −1.29, CrI −1.84 to −0.75) and ActWELL¹⁵ (MD = −0.90, CrI −1.39 to −0.41) interventions showed the least weight loss when compared to usual care (Figure 3; see Appendix 7, Table 11). The surface under the ranking curve (see Appendix 7, Figure 16) depicts WRAP 52-week intervention to have the highest probability of performing better than other interventions in comparison.

FIGURE 3

Forest plot showing the comparison of BWMI from RCTs vs. usual care for change in weight (kilogram) outcome.

Network of real-world services

The network of RWSs, which included 19 interventions from 19 programmes, found participants in programme 1 (MD = −4.03, 95% CrI −4.12 to −3.94), followed by programme 4 (MD = −3.76, 95% CrI −3.93 to −3.59), programme 19 (MD = −3.55, 95% CrI −3.64 to −3.46), programme 15 (MD = −3.39, 95% CrI −3.78 to −3.00) and programme 12 (MD = −3.45, 95% CrI −3.78 to −3.12), which showed higher weight loss than other programmes. Programme 5 (MD = −0.03, 95% CrI −0.07 to 0.00), programme 6 (MD = −0.18, 95% CrI −0.33 to −0.04) and programme 7 (MD = −0.28, 95% CrI −0.40 to −0.15) were those with very low or no weight loss (Figure 4; see Appendix 7, Table 12).

FIGURE 4

Forest plot for change in weight (kilogram) for interventions in RWS programmes. The numbers on the y-axis represent different RWS programme interventions for which IPD were shared (intervention details can be found in Appendix 4, Table 7).

Including the football fans in training trial (tailored for a specific group of male-only population) in the analysis

The sensitivity analysis by adding the FITT study¹⁷ to the connected network of RCTs showed that the intervention FFIT was associated with the largest amount of weight loss (MD = −4.65, CrI −5.24 to −4.07) at 12 weeks. The change in weight estimates for remaining interventions in the network were similar to the main analysis. WRAP¹⁹ 52 weeks [MD = −2.57 (−3.20 to −1.96)] led to maximum weight loss followed by WRAP 12 (MD = −2.26, 95% CrI −2.88 to −1.65) and POWeR²⁰ face to face (MD = −1.69, 95% CrI −2.39 to −0.99) and remote nurse interventions (MD = −1.50, 95% CrI −2.19 to −0.80). ActWELL¹⁵ (MD = −0.90, 95% CrI −1.40 to −0.41) and BeWEL^11,16 (MD = −1.29, 95% CrI −1.84 to −0.74) showed least change in weight at 12 weeks when compared to the usual care (Figure 5; see Appendix 7, Table 13).

FIGURE 5

Forest plot showing the comparison of BWMI from RCTs vs. usual care for change in weight (kilogram) outcome including the FFIT trial (tailored for a specific group of male-only population) in the analysis.

Trial calibration using participant characteristics from real-world services

After centring and adjusting for the covariates using the RWS data, effect estimates for change in weight obtained from stage 1 were used in stage 2 to run NMA. The analysis found participants randomised to FFIT¹⁷ intervention arm showed maximum weight loss (MD = −4.39, 95% CrI −5.43 to −3.35) followed by BeWEL¹⁶ [−2.91 (−4.76 to −1.07)], WRAP¹⁹ 52-week (MD = −2.75, 95% CrI −3.75 to −1.75), WRAP 12 weeks (MD = −2.75, 95% CrI −3.75 to −1.75) and POWeR+²⁰ with face-to-face nurse support interventions (MD = 2.42, 95% CrI −4.16 to −0.68). The interventions ActWELL¹⁵ (MD = −1.36, 95% CrI −2.77 to 0.05) and POWeR²⁰ with remote nurse support (MD = −1.68, 95% CrI −3.52 to 0.17) did not show a meaningful change in weight when compared to the usual care interventions (Figure 6, right; see Appendix 7, Table 15).

FIGURE 6

Forest plot showing the comparison of BWMI from RCTs vs. usual care for change in weight (kilogram) outcome using trial calibration method.

Real-world services

Last observation carried forward was used for imputing 12 weeks weight for the primary outcome change for RWSs. The analysis showed intervention for programme 6 (which enrolled male participants only) to have the highest weight loss (MD = −4.90, 95% CrI −7.00 to −2.80) followed by interventions of programme 8 (−4.55, 95% CrI −4.59 to −4.51), programme 19 (MD = −4.39, 95% CrI −4.49 to −4.29) and programme 13 (MD = −4.44, 95% CrI −4.81 to −4.07). Intervention for programme 7 showed the least decrease in weight (MD = −0.32, 95% CrI −0.45 to −0.19) (Figure 7; see Appendix 7, Table 17).

FIGURE 7

Forest plot change in weight (kilogram) at 12 weeks in RWS using the LOCF assumption.

Equality, diversity and inclusion

Individuals were not recruited for this research; however, the inclusion criteria for the IPD data from RCTs and RWSs were kept broad to include a diversity of services from across the UK.

The main determinant of diversity are the criteria chosen by the included trials and RWSs. Typically weight management services are accessed more by a female population, age range 40–60 and of a white British ethnic background.

Twenty-six per cent of adults in the UK are living with obesity, with the prevalence being highest in lower socioeconomic groups. This research is identifying effective components of weight management, and, in doing so, will assist in increasing access to these services, as mentioned in the 2019 NHS long-term plan.

CRediT contribution statement

Nishant Jaiswal (https://orcid.org/0000-0001-5511-4572): Data curation (equal), Formal analysis (lead), Investigation (lead), Writing – original draft (lead).

Rebecca Gregg (https://orcid.org/0000-0001-9119-1997): Data curation (equal), Investigation (supporting), Project administration, Writing – original draft (supporting).

Neil Hawkins (https://orcid.org/0000-0003-3199-221X): Formal analysis (supporting), Writing – editing and reviewing (supporting).

Sahar Sharif-Hurst (https://orcid.org/0000-0001-6885-0456): Formal analysis (supporting).

Alison Avenell (https://orcid.org/0000-0003-4813-5628): Conceptualisation (supporting), Funding acquisition (supporting), Methodology (supporting), Writing – editing and reviewing (supporting).

Louisa Ells (https://orcid.org/0000-0003-0559-4832): Conceptualisation (supporting), Funding acquisition (supporting), Methodology (supporting), Writing – editing and reviewing (supporting).

Sandra Jayacodi (https://orcid.org/0000-0001-5028-3465): Funding acquisition (supporting), Methodology (supporting), Writing – editing and reviewing (supporting).

Ruth Mackenzie (https://orcid.org/0000-0003-2041-5038): Conceptualisation (supporting), Funding acquisition (supporting), Methodology (supporting), Writing – editing and reviewing (supporting).

Sharon Anne Simpson (https://orcid.org/0000-0002-6219-1768): Conceptualisation (supporting), Funding acquisition (supporting), Methodology (supporting), Writing – editing and reviewing (supporting).

Olivia Wu (https://orcid.org/0000-0002-0570-6016): Conceptualisation (supporting), Funding acquisition (supporting), Methodology (lead), Writing – editing and reviewing (supporting).

Jennifer Logue (https://orcid.org/0000-0001-9549-2738): Conceptualisation (lead), Funding acquisition (lead), Methodology (supporting), Writing – editing and reviewing (lead).

Acknowledgements

The authors would like to acknowledge the support of the eDRIS Team (Public Health Scotland) for their involvement in obtaining approvals, provisioning and linking data and the use of the secure analytical platform within the National Safe Haven.

We thank Dr Ryan Field (HEHTA, University of Glasgow) for help with R (The R Foundation for Statistical Computing, Vienna, Austria) software.

Patient data statement

This work uses data provided by the public and collected by Local Authorities as part of their care and support. Using public data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it is important that there are safeguards to make sure that those are stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation

Data-sharing statement

All data requests should be submitted to RDM@lancaster.ac.uk. Access to anonymised data may be granted following review.

Ethics statement

This research was conducted accordance with the World Medical Association Declaration of Helsinki. Ethical approval was provided by Faculty of Health and Medicine Research Ethics Committee (FHMREC), Lancaster University on 26/01/21 (FHMREC reference: FHMREC20085). It was determined via the online criteria check that, due to the nature of the data request, NHS REC approval was not required.

Information governance statement

University of Lancaster is committed to handling all personal information in line with the UK Data Protection Act (2018) and the General Data Protection Regulation (EU GDPR) 2016/679. Under the Data Protection legislation, University of Lancaster the Data Controller, and you can find out more about how we handle personal data, including how to exercise your individual rights and the contact details for our Data Protection Officer here.

Disclosure of interests

Full disclosure of interests: Completed ICMJE forms for all authors, including all related interests, are available in the toolkit on the NIHR Journals Library report publication page at https://doi.org/10.3310/GJJL0404.

Primary conflicts of interest: Louisa Ells is specialist academic advisor to the Office of Health Improvement and Disparities.

Sharon Anne Simpson was on the NIHR Health Technology Assessment Clinical Evaluations and Trials Committee, the Commissioning Panel for the NIHR Policy Research Programme and the Chief Scientist Office HIPS committee. She is also President of the UK Society of Behavioural Medicine, a Society which has an interest in these findings.

Jennifer Logue is an employee of AstraZeneca with an honorary contract at Lancaster University. AstraZeneca has no involvement in the funding, design or outputs of this research. She was a member of the NIHR Health Technology Assessment Clinical Evaluations and Trials Committee from 2016 to 2020.

Ruth Mackenzie has no conflicts of interests to declare.

Rebecca Gregg has no conflicts of interests to declare.

Nishant Jaiswal has no conflicts of interests to declare.

Alison Avenell has no conflicts of interests to declare.

Sahar Sharif has no conflict of interest to declare.

Olivia Wu reports COVID-19 Reviewing 1 June 2020–30 September 2020; Systematic Reviews NIHR Cochrane Incentive Awards (no dates provided); HTA Remit and Competitiveness Group 1 February 2020–31 December 2021; HTA General Committee 1 August 2016–30 November 2019; HTA Post-Funding Committee teleconference (POC members to attend) 1 February 2020–31 December 2021; HTA Funding Committee Policy Group (formerly CSG) 1 February 2020–31 December 2021; HTA Clinical Evaluation and Trials Committee 1 November 2020–30 November 2023; HTA Programme Oversight Committee (no dates provided); ESP – Evidence Synthesis Programme Grants Committee (no dates provided); ESP – NIHR Incentive Awards Committee (no dates provided); ESP – Evidence Synthesis Programme Advisory Group 1 July 2015–1 June 2021; TARs Contract Retender 2020 Committee (no dates provided).

Department of Health and Social Care disclaimer

This publication presents independent research commissioned by the National Institute for Health and Care Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, MRC, NIHR Coordinating Centre, the Health Technology Assessment programme or the Department of Health and Social Care.

This article was published based on current knowledge at the time and date of publication. NIHR is committed to being inclusive and will continually monitor best practice and guidance in relation to terminology and language to ensure that we remain relevant to our stakeholders.

Study registration

This study is registered as PROSPERO CRD42020183949.

Funding

This article presents independent research funded by the National Institute for Health and Care Research (NIHR) Health Technology Assessment programme as award number NIHR129523.

Box

This article reports on one component of the research award BEhavioural Weight Management: COMponents of Effectiveness (BE:COME). For more information about this research please view the award page (www.fundingawards.nihr.ac.uk/award/NIHR129523).

About this article

The contractual start date for this research was in December 2020. This article began editorial review in January 2024 and was accepted for publication in April 2025. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The Health Technology Assessment editors and publisher have tried to ensure the accuracy of the authors’ article and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this article.

Copyright

Copyright © 2026 Jaiswal et al. This work was produced by Jaiswal et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This is an Open Access publication distributed under the terms of the Creative Commons Attribution CC BY 4.0 licence, which permits unrestricted use, distribution, reproduction and adaptation in any medium and for any purpose provided that it is properly attributed. See: https://creativecommons.org/licenses/by/4.0/. For attribution the title, original author(s), the publication source – NIHR Journals Library, and the DOI of the publication must be cited.

Disclaimer

Every effort has been made to obtain the necessary permissions for reproduction, to credit original sources appropriately and to respect copyright requirements. However, despite our diligence, we acknowledge the possibility of unintentional omissions or errors and we welcome notifications of any concerns regarding copyright or permissions.

Members of the BE:COME study group

Avni Vyas, Amy Ahern, Olivia Wu, Jamie Blackshaw, Jackie Smith, Adrian Coggins, Hidde van der Ploeg, Margaret Ogden, Sandra Jayacodi, Nicola Cooper, Jamie Hartmann-Boyce, Johan Parry, Colin Greaves.

RWSs – Optional authorship on the results paper(s) (up to two people per service). Published ‘on behalf of the study group’ – collaborators not identified unless they wish to be, following review of all the evidence.

Data requests and data handling

A signed collaboration agreement was required before services and RCTs could proceed with participation. The RWS agreement covered the blinding of the analysis, the return of the results to each individual intervention and that no programme would be identifiable within any publication. While participating interventions had the option to be listed on any publication, due to the potential for poor outcomes that could lead to commercial concerns, there was the option to not be listed. The RCT agreement ensured proper standards of use and anonymisation for individuals in the data.

Data requests were discussed during the contract period; this was an iterative process that allowed the researchers to determine if the service had access to suitable data and could meet the requirements of the research which included the completion of an intervention description template. Appendix 1, Tables 4 and 5 outline the required format for the data set from participating RWSs and RCTs. Services were asked to prepare the data (which in the case of RWSs had mostly not previously been used in research before) to meet the requirements of the data request; this was also an iterative process and, as anticipated, services required more support to prepare their data then RCTs in the form of online meetings and e-mail. Services were compensated for the time taken to do this as outlined in the data-sharing agreement (DSA). Payment was made once data have been transferred and checked for correctness.

Services were paid £1789 per real-world weight management intervention to cover the costs of data preparation and transfer, completion of the intervention description template and availability to respond to queries. While they were not NHS interventions, this was based on 10 days of band 8 Agenda for Change salary as this is the appropriate NHS grade for a senior data analyst or dietetic service lead. RWSs signed a contract detailing the DSA, and financial compensation for preparing data.

Risk-of-bias assessment

The summary risk-of-bias graph (Figure 8) showing 30% of the included RCTs at high risk of bias and Figure 9 showing the risk-of-bias assessment²⁵ of individual studies for each of the domain.

FIGURE 8

Summary risk-of-bias assessment for RCTs.

FIGURE 9

Risk-of-bias assessment for individual RCT included in the analysis.

Assessment of model fit

Though convergence was satisfactory for both fixed- and random-effects model, fixed-effects model provided a better fit. We compared the posterior means and DIC and constructed the leverage plots for the primary outcome – change in weight (kilogram) and found no significant differences between the two models.

FIGURE 10

Leverage plot with DIC and posterior means for fixed.

FIGURE 11

Leverage plot with DIC and posterior means for random-effects model.

Additional analysis for randomised controlled trials

In addition to primary outcome, additional NMA using RCT data were performed for outcomes associated with weight loss.

Change in body mass index (kg/m²)

The network of RCTs was also analysed for change in BMI and it also resonated with results of change in weight. Twelve-week WRAP intervention showed the highest decrease in BMI (MD = −0.80, 95% CrI −1.01 to −0.58) followed by POWeR²⁰ face-to-face nurse (MD = −0.60, 95% CrI −0.85 to −0.36) and POWeR²⁰ with remote nurse interventions (MD = −0.52, 95% CrI −0.77 to −0.27). BeWEL¹⁶ (MD = −0.45, 95% CrI −0.64 to −0.26) and ActWEL¹⁵ (MD = −0.34, 95% CrI −0.52 to −0.16) also showed a significant decrease in BMI, whereas mean change in BMI was not significant in the arm receiving WRAP¹⁹ 52-week intervention (MD = −0.11, 95% CrI −0.27 to 0.05) when compared to usual care (Figure 12).

Mean percentage change in weight

Mean percentage of weight loss was highest in the arm receiving POWeR²⁰ face-to-face nurse intervention (MD = −3.00, 95% CrI −3.75 to −2.24), WRAP¹⁹ 12-week (MD = −2.95, 95% CrI −3.54 to −2.35) and WRAP 52-week (MD = −2.64, 95% CrI −3.21 to −2.07) intervention when compared to usual care. The ActWELL¹⁵ (MD = −0.89, 95% CrI −1.48 to −0.31) and BeWEL¹⁶ (MD = −1.56, 95% CrI −2.17 to −0.94) interventions showed least percentage of weight loss from baseline weight when compared to usual care (Figure 13; see Table 13).

Number of participants with > 5% weight loss

Comparing the number of participants with more than 5% weight loss, WRAP¹⁹ 52-week intervention arm had maximum number of participants losing more than 5% weight (OR = 10.32, 95% CrI 6.57 to 16.72), The interventions arms where participants lost over 5% weight compared to usual care were WRAP 12-week (OR = 5.75, 95% CrI 3.60 to 9.60), POWeR²⁰ + F (OR = 4.67, 95% CrI 2.67 to 8.65), POWeR + R (OR = 4.26, 95% CrI 2.41 to 7.94) and BeWEL¹⁶ (OR = 3.74, 95% CrI 1.82 to 8.38). Number of participants with more than 5% weight loss was not significant in ActWELL¹⁵ (OR = 1.03, 95% CrI 0.67 to 1.60) when compared to usual care (Figure 14; see Appendix 7, Table 14).

Number of participants with > 10% weight loss

Weight Watchers referrals with participants attending either¹⁹ 12-week intervention arm had maximum number of participants losing over 10% of weight (OR = 8.14, 95% CrI 2.85 to 34.1). The intervention arms where participants lost over 10% weight were WRAP¹⁹ 52-week (OR = 5.73, 95% CrI 1.98 to 24.10), POWeR + R²⁰ (OR = 4.19, 95% CrI 1.75 to 11.80), POWeR + F (OR = 3.26, 95% CrI 1.45 to 10.30) when compared against usual care. Number of participants with more than 10% weight loss were not significant in BeWEL¹⁶ (OR = 3.98, 95% CrI 0.431 to 122.0) and ActWELL¹⁵ (OR = 0.74, 95% CrI 0.24 to 2.19) when compared to usual care (Figure 15; see Appendix 7, Table 14).

FIGURE 12

Forest plot showing comparison of BWMI from RCTs with usual care: change in BMI (kg/m²). Mean percentage change in weight.

FIGURE 13

Forest plot showing comparison of BWMI from RCTs with usual care: mean percentage change in weight.

FIGURE 14

Forest plot showing comparison of BWMI from RCTs with usual care: number of participants with more than 5% weight loss.

FIGURE 15

Forest plot showing comparison of BWMI from RCTs with usual care – number of participants with more than 10% weight loss.

Article history

The contractual start date for this research was in December 2020. This article began editorial review in January 2024 and was accepted for publication in April 2025. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The Health Technology Assessment editors and publisher have tried to ensure the accuracy of the authors’ article and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this article.