Evaluation of a scalable behavioural weight loss programme in people with a recent diagnosis of type 2 diabetes
This WP evaluated the 1-year effectiveness and cost-effectiveness of combining an existing scalable weight loss programme with tailored diabetes education versus standard care in adults with a recent diagnosis of T2D. It also included a mixed-methods process evaluation to identify what worked, what did not and why.
Rationale
In the UK, standard care for T2D includes a structured diabetes education programme (DE) such as DESMOND, which achieves small reductions in weight relative to a control group (1.26 kg), with no significant differences in HbA1c at 1-year follow-up.32 Live Well With Diabetes is a tailored DEW which combines the same weight loss programme used in WP1.1 (WW) with structured diabetes education and dietetic support via two telephone calls. In a recent US trial among adults with T2D, participants randomised to WW and advice from a certified diabetes educator had 2 kg greater reductions in weight and 4 mmol/mol greater reductions in HbA1c than those randomised to standard care at 1 year: 24% of participants randomised to WW plus diabetes education achieved good glycaemic control at 12 months, compared with 14% receiving standard care.37 This RCT was needed to evaluate cost-effectiveness of DEW relative to UK standard care (DESMOND or equivalent) for adults with newly diagnosed T2D.
Objectives
- To evaluate the impact of DEW on 1-year changes in body weight, glycaemia, diabetes status, modelled cardiovascular risk and resource use in adults with a new diagnosis of T2D.
- To quantify the short-term cost-effectiveness of DEW.
- To assess the extent to which DEW reaches the target population and meets the needs of patients.
- To qualitatively examine causal mechanisms and identify contextual factors associated with variations in outcome, informing refinement of programme content and implementation.
Glucose Lowering through Weight management: a randomised controlled trial of the clinical and cost-effectiveness of a diabetes education and behavioural weight management programme versus a diabetes education programme in adults with a recent diagnosis of type 2 diabetes
The RCT addressing objectives 1 and 2 of this WP is published in Mueller et al.54
Method
The protocol for the GLoW trial was pre-registered (ISRCTN18399564) and published in Ahern et al.55 In brief, GLoW was a pragmatic, randomised, single-blind, parallel-group trial designed to evaluate the effectiveness of a tailored DEW compared to standard DE for individuals living with overweight or obesity and newly diagnosed T2D. We recruited 577 adults with a BMI ≥ 25 kg/m2 and a diagnosis of T2D in the last 3 years from 159 primary care practices in England. They were randomised to either DEW or DE, with stratification by sex and diabetes duration. DEW participants received a tailored programme, including diabetes education via telephone calls with a dietitian and free membership to WW, while DE participants attended a 6-hour structured diabetes education workshop (DESMOND program). We collected clinical, behavioural, psychosocial and health economic outcomes at baseline, 6 and 12 months.
Coronavirus disease discovered in 2019 (COVID-19) led to major modifications to study. The study paused recruitment in March 2020 due to the COVID-19 pandemic and restarted in January 2021. Both interventions transitioned to online delivery during COVID lockdowns and hybrid thereafter, and data collection was entirely remote after March 2020.
For the primary outcome (change in HbA1c from baseline to 12 months), we estimated the intervention effect using a random intercepts linear regression model, adjusting for sex, diabetes duration and baseline HbA1c. We conducted pre-specified sensitivity analyses to adjust for glucose-lowering medication and duration of follow-up. We used multivariate imputation by chained equations to impute missing values of HbA1c at 12 months under the assumption that data are missing at random. We investigated the impact of departures from this assumption using a pattern mixture model that allows data to be missing not at random. A per-protocol analysis focused on those who fully engaged in their allocated programme, combining self-reported data with attendance records. We conducted a post hoc sensitivity analysis to evaluate the impact of the COVID-19 pandemic. For secondary outcomes, continuous data were analysed using similar random intercepts models, and binary data were analysed using logistic regression.
In the economic evaluation, we conducted a cost-utility analysis to compare the DEW and DE programs from a UK NHS and Personal Social Services (PSS) perspective. Key outcomes included the incremental cost per QALY gained, as well as cost per 1 mmol/mol reduction in HbA1c and cost per 1 kg weight loss over 12 months.
We estimated that the intervention cost per participant of the DEW programme was £325 (£271 + value-added tax), and the cost of the DE programme was £159 [a weighted average of face-to-face (£268) and online (£12) delivery].
Key findings
Between September 2018 and August 2021, we assessed 1161 participants for eligibility and randomised 577. We recruited 204 of these participants after the trial resumed following COVID-19 protocol amendments. Participants had a mean age of 59.8 years and a mean BMI of 34.6 kg/m2, 52% were female, 91% were of White ethnicity, and 39% had post-secondary education.
Glycated haemoglobin data were available for 528 participants at baseline, 358 at 6 months, and 398 at 12 months. Baseline characteristics were similar for those with and without missing data.
Intervention engagement data were available for 289 participants in the DEW group and 179 in the DE group, with 61% of DEW participants and 50% of DE participants engaging in the intervention.
No significant intervention effect was observed on the primary outcome of change in HbA1c from baseline to 12 months (difference: −0.84 mmol/mol, 95% CI −2.99 to 1.31; p = 0.44). Sensitivity analyses and per-protocol analysis also showed no effect.
Diabetes education and BWMP participants lost more weight than DE participants: 1.77 kg more at 6 months and 1.38 kg more at 12 months. At 6 months, DEW participants were significantly more likely to achieve ≥ 5% and ≥ 10% weight loss compared to DE. These effects were less pronounced at 12 months.
Diabetes education and BWMP participants had higher odds of achieving diabetes remission at both 6 and 12 months compared to DE.
There were no significant differences in psychosocial or behavioural outcomes, except for a small effect on dietary restraint at 6 months.
Diabetes education and BWMP had lower mean costs than DE over 12 months but marginally fewer QALYs gained. The mean incremental costs for DEW versus DE were −£232, with an incremental net monetary benefit (iNMB) that was positive at cost-effectiveness thresholds of £13,000, £20,000 and £30,000 per QALY gained. The probability of DEW being cost-effective within 12 months was approximately 0.62–0.66 at these thresholds.
Mixed-methods process evaluation of the Glucose Lowering through Weight management trial
Method
We assessed intervention uptake, engagement and acceptability using self-reported data (e.g. perceived usefulness), objectively measured data (e.g. app engagement metrics) and semi-structured interviews with participants (n = 19 from the intervention group, n = 8 from the control group) and providers (n = 7).
Key findings
Intervention uptake (i.e. completing at least one dietitian call, and using at least one WW resource) was 61%. Baseline characteristics were not associated with uptake. Perceived facilitators of engagement included the convenience of digital access, trust in the dietitians and free membership. Barriers included negative preconceptions about WW/commercial weight management, lack of digital skills/access, lack of trust in the WW coaches (= trained lay people), social anxiety and competing responsibilities.
Forty-seven per cent of participants received their programme post onset of the COVID-19 pandemic, when WW workshops were offered virtually instead of in-person. Uptake was reduced post onset in both groups (DEW: 68 to 53%, DE: 57 to 40%). Older age was associated with lower odds of engaging digitally [OR = 0.94 (0.90–0.99)]. While some participants reported perceived benefits from the intervention, others found it difficult to implement behavioural changes; participants felt the intervention required high levels of enthusiasm/motivation, which was difficult to maintain.
Strengths and limitations
Recruitment for the GLoW trial was extremely challenging, particularly during the COVID-19 pandemic, and in the end, it took 3 years to recruit 577 participants. We achieved diversity in gender (48% male) and education (< 40% had a post-secondary education) and had a good spread of participants across IMD quintiles; however, 91% were of White ethnicity. While this reflects the ethnic makeup of the UK population and is common across similar trials, it does limit generalisation of findings to other ethnic groups.
The COVID-19 pandemic led to disruptions of planned study procedures and intervention delivery and resulted in higher-than-expected attrition. This reduced power for our primary analysis and limited our analyses of secondary outcomes. Although we found no statistical evidence that the onset of the COVID-19 pandemic influenced intervention effects, there is some suggestion that effects may have differed pre and post onset.
Conclusions
We found no evidence that combining a scalable behavioural weight loss programme with remote diabetes education was better than standard diabetes education at reducing HbA1c at 12 months. However, high attrition for the primary outcome and the impact of the COVID-19 pandemic suggest that we should be cautious in interpreting this finding. We found that the intervention led to more weight loss and could be cost saving within 12 months because of reductions in diabetes medication and other healthcare costs. The intervention was generally well received by participants. While most participants successfully engaged digitally with the programme, future implementation should consider the challenges of digital exclusion and consider how to increase perceived trustworthiness, credibility, and relevance of the programme to diabetes, and how to support long-term engagement and behaviour change.
Interrelation with other work packages
Data on metabolic trajectories was used in WP1.3 to model the longer-term impacts of the programme and the process evaluation data were used to support intervention costing.
We had intended that qualitative data from this WP would inform intervention development in WP2.1. However, recruitment delays meant that this was not possible. Instead, the views and experiences of people with T2D after weight loss were included in WP2.1 via our patient and public involvement and engagement (PPIE) group.