Medical Research Council Hot Topic workshop report: Planning a UK Nutrition and Healthy Life Expectancy Trial

There is a drive in the UK to harness findings from novel fundamental and efficacy nutritional research and, through inter-disciplinary and multi-agency collaborations, to improve eating behaviour for the benefit of population health. This report summarises the progress made during the Medical Research Council-funded Hot Topic workshop on the planning for a potential UK-wide nutrition primary prevention randomised controlled trial with incident disease as the study endpoint: the UK Nutrition and Healthy Life Expectancy (NuLifE) Trial. Through two workshops, along with online discussions and a systematic evidence synthesis, over 40 experts from a range of disciplines came together over 6 months. The workshop reached a consensus and delivered a three-stage plan with the ultimate ambitious aim of providing effective eating behaviour change strategies to address the growing inequalities in the UK and contribute to both a reduced risk of prevalent diet-related chronic disease and an increase in healthy life expectancy.

Fairweather-Tait (SF) and Naveed Sattar (NS), with Amy Jennings (AJ) conducting a systematic review to inform the NuLifE trial design.The workshops were attended by 42 participants from 19 organisations with expertise in nutritional epidemiology, nutritional science, behavioural science, ageing, public health, chronic disease pathology, weight loss, nutrigenomics, health inequalities, clinical trial design and delivery, business strategy, public policy, health economics and digital technology (Appendix A).Senior researchers were encouraged to identify early career researchers to attend.

R ATIONALE FOR THE WORKSHOP
Although beginning to plateau, life expectancy (LE) in the UK has increased 2-3 years every decade for the last century and currently stands at 79 and 83 years for males and females.However, healthy life expectancy (HALE) is not increasing at the same rate, with an increased rate for HALE:LE of 0.8 (ONS, 2018).This is in effect creating one additional year of ill health per person every 15 years.Furthermore, there are large inequalities evident, with 18 years difference in HALE evident in England between the lowest and highest Index of Multiple Deprivation (IMD) groups (ONS, 2018).This trajectory is unsustainable for NHS and social care delivery.The UK Government's 'Ageing Society' Grand Challenge set a mission to increase HALE by 5 years (from 63 to 68 years) by 2035, with the planned NuLifE programme in direct response to this ambitious target (BEIS, 2019).
Suboptimal diet is the number one modifiable determinant of HALE and chronic disease risk globally, and, in the UK, is responsible for 15-20% of the population attributable fraction of years of life lost (Afshin et al., 2019;Steel et al., 2018).Current UK dietary recommendations and public health policy are based largely on evidence from prospective cohort studies.There is limited evidence from efficacy trials which typically examine the short-term impact of select dietary components or foods on biochemical, functional and imaging-based surrogate markers of disease, as the primary endpoint.The actual impact and size effect of a whole diet intervention on incident disease is completely unknown.In addition, the approaches necessary to implement such an intervention by achieving long-term changes in individual and community eating behaviour are unidentified and untested.
The remit for a NuLifE primary prevention trial is therefore important and timely.The long-term aim of the programme is to establish the impact of a wholediet nutrition intervention on the risk of transition from health to a clinical diagnosis of disease in 'at-risk' UK adults.

AIMS OF THE WORKSHOP
1. Fully consider the need for and impact of the NuLifE trial.2. Develop linkages between disciplines (nutrition, trialists, primary and secondary care, health economics, digital health, behavioural science, food industry/retailer, the public) and identify discipline gaps.3. Advance the research questions and trial design.4. Develop a plan and timeline for delivering the NuLifE programme and funding strategy.

WORKSHOPS STRUCTURE
The workshop was delivered as two half-day virtual events, held on the 28th September 2020 and the 17th February 2021.Prior to the first meeting, the following documents were provided to all registered participants to show the relevant policy and scientific background to the NuLifE remit.

Key presentations
The first NuLifE workshop began with an overview of the workshop justification from AMM.To inform the possible disease end-point focus for NuLifE, AMM presented the UK LE, HALE and morbidity and mortality statistics.Dementia is the number one cause of mortality in England and Wales accounting for 12.8% of total deaths in 2017, with dementia and cardiovascular diseases collectively responsible for 40% of total deaths (ONS, 2017).Two-thirds of the UK population are either overweight or obese, with a greater prevalence in those who are most deprived (NHS Digital, 2020).Reflecting on key UK nutrition and dietary guidelines as operationalised in the Eatwell Guide, and current population diet and nutrition data, large proportions of the population are not meeting current dietary targets and there is an evident socio-economic gradient (Bates et al., 2019).In a 2020 analysis using data from 2012 to 2017 in the National Dietary and Nutrition Survey (NDNS), population adherence for nine of the Eatwell Guide recommendations ranged from 7% to 80%, with the lowest adherence for fibre, oily fish and sugar (Scheelbeek et al., 2020).
To fuel breakout room discussions, and put the subsequent key presentations into perspective, AMM presented some initial thoughts (and related questions) around the potential PICO (i.e.population, intervention, control, outcome) elements of the NuLifE trial (Richardson et al., 1995).For the 'Population' element, the trial would likely need to recruit participants with an 'at-risk' profile in order to achieve the required number of disease cases during the intervention period (but how do we define 'at-risk'?).For the 'Intervention', should we adopt a whole-diet approach, rather than focus on specific nutrients, dietary derived bioactives or select foods/food groups?Should the intervention be bespoke and personalised?Should the trial be a pragmatic, effectiveness trial?For the 'Control' arm, should it be usual care?For the 'Outcome' element, do we focus on incident cardio-metabolic disease, should we include dementia, and should our composite end-point include other major chronic diseases such as diet-sensitive cancers?
Lastly, AMM set out a crude sample size calculation to set the expectations of attendees to the likely scope of the trial.Using event rates from the large prospective UK Biobank study of 0.5 million adults in the UK, with a high-risk population over the age of 55 years, a composite outcome of cardiovascular disease and type 2 diabetes with an event rate of approximately 3% a year would give a simple two-arm trial over 5 years a power of 90% to detect a 10% relative risk reduction if there were approximately 16 000 participants (UK Biobank, 2007).
The following speakers and breakout room discussions expanded on these key elements of trial design and the necessary research infrastructure that would be needed for a nutrition trial that was much larger than any that had been previously attempted in the UK (with only a few examples globally; Estruch et al., 2018).
Jane Armitage from the University of Oxford spoke about how new opportunities with digital technologies and data linkage could enhance the ability to run streamlined trials at scale with lower costs and improved quality.Giving examples from the largescale ORION-4 and ASCEND trials (ClinicalTrials.govIdentifier: NCT03705234; NCT00135226), she highlighted the importance of simple inclusion criteria; and the use of electronic health records to identify large numbers of participants, screen and consent them efficiently, and follow them up more completely for a wide range of events (Mafham et al., 2020).
Falko Sniehotta from the University of Newcastle gave a summary of the particular challenges of Finally, Paul Aveyard from the University of Oxford spoke about the trial design options.This started with a reminder of the MRC Framework for complex intervention development and reinforcing the value of defining the behavioural theory underpinning the putative intervention (Craig et al., 2008).Given this is a trial aiming for a long-term health outcome, it is vital to ensure the intervention had the best chance of achieving sustained dietary change.The presentation then passed on to novel trial designs that may be employed in early-stage development work, such as the multiphase optimisation strategy trial design, that can help determine the effect of particular components of an intervention, and the value of adaptive designs in finding ways to maximise the benefit of a behavioural support programme (Wyrick et al., 2014).Finally, he presented an explanatory-pragmatic trial continuum to fuel debate about which end of the spectrum we would want to situate this work.

Breakout rooms and general discussion
Four breakout rooms then discussed the PICO elements on the study design in depth, and fed back to the entire group their main points.See Table 1 for a summary of the discussed PICO elements.
When the breakout rooms fed back to the final discussion, a consensus emerged that there was a need and a will to proceed with the NuLifE programme.It was noted that it was especially important to consider how such a trial might help address health inequalities, perhaps by recruiting specifically in areas of high deprivation.The other main area of discussion was whether NuLifE would be an efficacy trial, focussed on answering the question of the strength of the impact of a healthy diet on the risk of disease outcomes; or an effectiveness trial, demonstrating the impact of a pragmatic dietary intervention on a representative sample of the UK.
Subsequently between workshops, comments were sought online on three possible study designs.
1.An individual-level efficacy (explanatory) RCT aiming to change dietary behaviours and reduce incident disease in UK adults.

An individual-level effectiveness (pragmatic) RCT
aiming to change dietary behaviours and reduce incident disease in UK adults.3. A population-level cluster (area/town/city) RCT aiming to influence structural (social, cultural, economic, geographical and other environmental) determinants of eating behaviour, improving eating behaviour and nutrient status in UK adults, and reducing the risk of incident disease in UK adults.

NULIFE WORKSHOP 2
At the start of the second NuLifE workshop, AMM began with a summary of the first workshop and the online working documents.The online working document of potential study designs between sessions had reached an agreement that there was insufficient justification to proceed with an explanatory RCT (#1 above) for the following reasons: • To run a true whole-diet efficacy RCT with long-term outcomes, we would need to be very confident of the ability to implement an intervention with high fidelity, and with a compliant (pre-screened, selected and monitored) population group, to have the power to detect outcomes.It would be practically very challenging to get the compliance needed for the level of certainty required.• Furthermore, such an approach would likely divert attention from populations where there is greatest need (i.e.participants in communities with high levels of deprivation).TA B L E 2 (Continued) • An efficacy trial would need to be followed by effectiveness trials if the results are going to inform public health policy.• As a result, the higher priority and best use of resources would be to conduct a pragmatic trial in deprived communities where the poorest adherence to dietary guidelines is evident.Such a trial is likely to be more informative and impactful than a similarly sized trial in the general population.
Although there was a strong rationale for a cluster intervention approach (#3 above), which focussed on contextual modulators of eating behaviour, the development of such an intervention appeared currently unrealistic as it would be reliant on enactment of several fiscal, food and health-related policy changes; and a much greater understanding of the relative structural determinants of food purchasing and eating behaviour than is currently available.
The consensus was clear, that NuLifE would proceed with planning an effectiveness (pragmatic) trial focussed on examining the impact of an eating behaviour intervention on incident disease in communities with a high score on the IMD.
Such an approach is aligned with the public health policy need as reflected in: 1. the UK Government Ageing Society Grand Challenge mission (BEIS, 2019); 2. the Public Health England's Strategy 2020-2025, around healthy inequalities and healthy ageing (PHE, 2019); 3. current health and behaviour inequalities, aggravated and highlighted by the COVID-19 pandemic and its impact on COVID-19 prognosis (Bambra et al., 2020); 4. reports such as Health equity in England: The Marmot review 10 years on (Marmot, 2020).
Prior to the second meeting, the working group agreed that it was clear we were lacking a consensus on the intervention approach for NuLifE.It was decided that we needed an evidence synthesis of eating behaviour interventions to help inform the discussion.

Evidence synthesis of eating behaviour interventions
Changing eating behaviour is hugely challenging, with numerous social, cultural, economic, environment and geographical variables influencing food purchase and consumption.Prior to workshop 2, a systematic search of the empirical evidence was conducted by AJ (University of East Anglia) to inform the discussion and to establish the evidence for the impact of various interventions on attainment and maintenance of eating behaviour change.Search terms looked for any RCT on diet and disease in adults that did not involve supplements, which was published since 2015; the search was extended to also look for any RCTs on diet conducted in the UK since 2010.Results were excluded if the trials did not involve behavioural change or were lacking diet-only treatment arms (see Appendix B for more details).The following intervention approaches were focussed on and summarised, along with their results: 1. food/meal provision 2. cash incentives, food vouchers, supermarket delivery 3. digital intervention 4. personalisation 5. education and cookery skills 6. Behaviour Change Technique (BCT)-based 7. group sessions and peer support 8. community-based intervention.
Results from trials run in the last 5 years are summarised in Table 2 (further results are in Appendix B).Many trials were conducted in healthy individuals who were not selected on the basis of the quality of their diet and hence many had adequate diets at baseline.A range of dietary components were targeted and, in many cases, an exact definition of dietary improvement was not detailed.It was therefore difficult to conclude if the studies that reported no effect on health outcomes were due to poor adherence or a true lack of effect.With such heterogeneity in the results, it was not possible to reach a consensus on which whole diet intervention

Presentations and breakout rooms focussed on intervention
The presentations and breakout room discussions for the second workshop all focussed around defining an effective eating behaviour intervention.First, spokespeople from previous whole-diet interventions (JM from Food4Me, AJ from NuAge, Wendy Hall from CRESSIDA, AMM from MedEx) summarised key insights from their studies (details of these studies are summarised in Appendix B).Breakout rooms chaired by members of the working group (NS, SJ, JC, JM) then discussed the intervention components as a first priority, and the study population or key endpoints as a second priority.Feedback and general discussion amongst the entire group revealed that there was a general enthusiasm for targeting groups with high IMD scores of deprivation as a unique component of NuLife.However, there was also a consensus that the NuLifE trial would need to be conducted in stages in order to achieve its ultimate goal, requiring a strong community co-creation, pilot and feasibility phase in order to proceed with confidence with an eating behaviour intervention strategy which is fit for purpose.

STAGE 1: TAILORING COMPLE X D IETARY INTERVENTIONS TO COMMUN ITIES WITH A HIGH INDE X OF MULTIPLE DEPRIVATION
Stage 1 of the NuLifE programme will be a communitybased (high IMD communities), Patient and Public Involvement informed, project to co-design, develop and test the NuLifE intervention approaches to increase the agency of residents to enact eating behaviour within the Eatwell guidelines.The intervention will be developed based on the MRC Developing and evaluating complex intervention guidelines (Craig et al., 2008).In addition, the intervention will be co-developed and co-produced with organisations such as the Integrated Care Systems, Community Foundations, FoodBanks/ Trussell Trust, community-embedded food retailers (e.g.Co-op/ALDI) and build on existing community assets.Both personalised (to the individual's baseline health and eating behaviour status) and contextual intervention components will be considered, with the ultimate aim to provide individual and communities with the capability, opportunity and motivation to enact healthier eating behaviours which aim to increase HALE and reduce chronic disease onset (Michie et al., 2011).

STAGE 2: ONE-YE AR EFFECTIVENESS TRIAL ON INTERIM END POINT OF BLOOD PRESSURE
The NuLifE consortium then agreed that Stage 2 of the NuLifE programme would be an interim step to test the effectiveness of a one-year intervention of the NuLifE approach with blood pressure as the primary endpoint.Blood pressure is a robust target due to its strong prognostic value for future cardiovascular diseases, and it is influenced by a wide array of foods and macro-and micro-nutrients and non-nutrient bioactives (Jennings et al., 2019;Ndanuko et al., 2016).

STAGE 3: L ARGE-SCALE EFFECTIVENESS PRIMARY PREVENTION RCT
Finally, Stage 3 would be to run a full effectiveness primary prevention RCT of the NuLifE intervention to assess the effect on incident disease.

CONCLUSION
We thank the MRC for funding this workshop, which provided the impetus for over 40 experts from a range of disciplines to come together over 6 months.Through two workshops, along with online discussions and a systematic evidence synthesis, the first UK nutrition, primary prevention, effectiveness trial took shape.The workshops delivered a three-stage plan with the ultimate ambitious aim of establishing a multi-disciplinary and multi-agency approach, alongside effective eating behaviour change strategies, to address the growing inequalities in the UK and contribute to a reduced risk of prevalent diet-related chronic disease and increased HALE.Such an approach directly addresses the Government's 'Ageing Society' Grand Challenge to 'ensure that people can enjoy at least 5 extra healthy, independent years of life by 2035, while narrowing the gap between the experience of the richest and poorest' (BEIS, 2019).

ACK NOWLED GEM ENT S
JC and AMM drafted the manuscript; JM, SFT, SJ, NS, AJ all assisted with creating, editing and revising the manuscript.

TA B L E 1
Summary of Population, Intervention, Control, Outcome (PICO) elements discussed during breakout rooms in NuLifE workshop 1

PICO breakout room Main discussion points Population
Evidence synthesis of randomised controlled trials using behaviour change to improve diet in the last 5 years • Challenges of working in deprived areas, but possible target for greatest effect • Using GP surgeries to recruit high-risk patients • Potential screening for those with low-quality diets at baseline (vs.using streamlined recruitment methods) • Previous nutrition efficacy trials have shown effects on intermediate mechanisms for cardiometabolic diseases like cardiovascular disease and diabetes • Need a different outcome for an interim pilot study, such as blood pressure, that would demonstrate efficacy of the intervention quickly • What to do about weight loss?Is this an outcome to aim for, or would it confuse the picture?Study design • Effectiveness or efficacy trial?• Factorial design with personalised tailoring TA B L E 2