Baseline Characteristics in the Remote Diet Intervention to REduce long-COVID Symptoms Trial (ReDIRECT)

Background The persistence of symptoms for ≥12 weeks after a COVID-19 infection is known as Long COVID (LC), a condition with unclear pathophysiology and no proven treatments to date. Living with obesity is a risk factor for LC and has symptoms which may overlap with and aggravate LC. Methods ReDIRECT is a remotely delivered trial assessing whether weight management can reduce LC symptoms. We recruited people with LC and BMI >27kg/m 2. The intervention was delivered remotely by dietitians, with online data collection (medical and dietary history, COVID-19 infection and vaccination, body composition, LC history/symptoms, blood pressure, quality of life, sociodemographic data). Participants self-selected the dominant LC symptoms they most wanted to improve from the intervention. Results Participants (n=234) in England (64%) and Scotland (30%) were mainly women (85%) of white ethnicity (90%), with 13% living in the 20% most deprived areas, a mean age of 46 (SD10) years, and median BMI of 35kg/m 2 (IQR 32-40). Before starting the study, 30% reported more than one COVID-19 infection (82% confirmed with one or more positive tests). LC Diagnosis was mainly by GPs (71%), other healthcare professionals (9%), or self-diagnosed (21%). The median total number of symptoms was 6 (IQR 4–8). Self-selected dominant LC symptoms included fatigue (54%), breathlessness (16%), pain (12%), anxiety/depression (1%) and "other" (17%). At baseline, 82% were taking medication, 57% reported 1+ other medical conditions. Quality of life was poor; 20% were on long-term sick leave or reduced working hours. Most (92%) reported having gained weight since contracting COVID-19 (median weight change +11.5 kg, range -11.5 to +45.3 kg). Conclusions Symptoms linked to LC and overweight are diverse and complex. Remote trial delivery enabled rapid recruitment across the UK yet certain groups (e.g. men and those from ethnic minority groups) were under-represented. Trial registration ISRCTN registry ( ISRCTN12595520, 25/11/2021).

wanted to improve from the intervention.

Conclusions
Symptoms linked to LC and overweight are diverse and complex.Remote trial delivery enabled rapid recruitment across the UK yet certain groups (e.g.men and those from ethnic minority groups) were under-represented.

Plain Language summary
Long COVID (LC, symptoms lasting 12 weeks or more after a COVID-19 infection) is a poorly understood condition, with no proven treatments.Living with obesity increases the risk of developing LC; symptoms of obesity overlap and aggravate those of LC.
The ReDIRECT study tests, in people living with both LC and overweight, whether weight management can reduce LC symptoms.The study involves total diet replacement (with porridge, soups and shakes) for 12 weeks and is delivered remotely, with dietitian support via internet and/or phone.Researchers collected all data via online forms (medical and diet history, COVID-19 infection and vaccination, weight, height, LC history and symptoms, blood pressure, quality of life, and other demographic data).Each participant selected the LC symptom they most wanted to see improve.
Participants (n=234) lived across the UK, were mainly women (85%) of white ethnicity (90%), with 13% living in the 20% most deprived areas.Their average age was 46 years old with an average body mass index (BMI) of 35kg/m2.
LC is characterised by a constellation of symptoms, with some studies phenotyping the condition based on symptom clusters (Evans et al., 2022).For people living with LC, symptoms can vary in duration, severity, and impact on daily functioning.LC can follow an unpredictable, relapsing and remitting course, with 'flare-ups' following particular triggers (e.g.physical exertion) (Brown & O'Brien, 2021;Hastie et al., 2023).Among the most common symptoms are fatigue, breathlessness, pains, cognitive dysfunction, loss of taste/smell, and anxiety/depression (Davis et al., 2021;Sudre et al., 2021), though over 50 symptoms have been identified (Lopez-Leon et al., 2021).The long-term impact of LC symptoms on the wellbeing and quality of life of people affected by this condition is unknown, nor is the economic impact.
Comorbidities, such as cardiovascular disease, diabetes, obesity, and respiratory conditions, have been identified as risk factors for COVID-19.However, their influence on LC symptoms is not well understood.Moreover, there are also substantially higher odds of LC based on sex (female), higher deprivation and occupation (especially education or health-related occupations) (Shabnam et al., 2023).
Lifestyle factors, such as physical activity levels and dietary patterns, may influence LC symptoms, yet their association remains understudied.At the time of writing (02/11/2023), 501 trials investigating LC treatments were registered on clinicaltrials.govand ISRCTN since 2020 (search terms included Long COVID and synonyms named above).Around 2/3 of the studies were interventional, focusing mainly on rehabilitation and alleviating individual symptoms.With only 26 studies investigating dietary approaches, mostly dietary supplements, dietary strategies to manage LC lack a solid evidence base.Results are awaited for most of these trials.
Weight increase was widely reported across populations following the first COVID-19 lockdown period (March-May 2020) (Bakaloudi et al., 2022), compounding an existing high prevalence of overweight and obesity in the general populationin the UK, 64% of all adults were reported to have a BMI greater than 25kg/m 2 in 2021 (Organisation for Economic Co-operation and Development, 2021).Generally, body composition and weight changes post-COVID have not been thoroughly described in people living with LC.However, the EPILOC study highlighted a consistent association between increasing BMI and post-COVID fatigue, neurocognitive impairment, and chest symptoms, with the greatest recovery rate in those with BMI between 21 and 22 kg/m 2 (Peter et al., 2023).
Weight management programmes in adults with overweight/ obesity have been reported to reduce symptoms such as fatigue, breathlessness, and pain, which are also common with LC (pain andQoL (Höchsmann et al., 2022), dyspnoea (Bernhardt &Babb, 2014;Riddle & Stratford, 2013;Stenius-Aarniala et al., 2000).However, the effectiveness of intentional weight loss to reduce symptoms of LC and prevent future cardiometabolic ill-health has not been studied despite that group's specific needs.
The Remote Diet Intervention to Reduce Long-COVID Symptoms Trial (ReDIRECT) tests the impact of an evidence-based dietary weight management programme on LC symptoms in people with overweight or obesity.Here, we describe the baseline characteristics of ReDIRECT participants, focusing on socioeconomic and demographic factors as well as dietary and medical history, including LC diagnosis and symptoms.

Patient and Public Involvement
People living with Long COVID were involved throughout this research.Before submitting the funding application for the research, in March 2021, we conducted an online consultation via the Long Covid Scotland patient group (n=34) to assess interest in the study and to define priorities.Over 75% (n=26) were very interested, with 85% (n=30) prioritising the relief of fatigue and breathlessness as long COVID symptoms they would most like to improve.This informed our decision to use a novel personalised primary outcome measure.
The lay co-investigator (JO) has lived experience of Long COVID since March 2020 and is an active member of the Long Covid Scotland action group.JO and another co-author with lived experience of Long COVID (CW) were part of the study team and members of the monthly trial management team.A designated UoG staff member (TI) coordinated PPI input throughout, including input from a separate COVID-19 PPI group (6 people).
Throughout the study, the group regularly provided advice relating to the intervention, recruitment, data collection and topics for qualitative interviews, and also shaped dissemination activities, contributed to publications and the development of an animation aimed at the general public.Throughout the trial, the impact of involvement was tracked using the Guidance for Reporting Involvement of Patients and the Public' (GRIPP2) short form, including highlighting where patient perspectives influenced study decision-making.
Ethical approval.Ethical approval was obtained from the South-East Scotland Research Ethics Committee 01 (reference number: 21/SS/0077) on 19 th November 2021.The REC favourable opinion was subject to the following conditions being met prior to the start of the study.Capacity and Capability (in England, Northern Ireland and Wales) or NHS management permission (in Scotland) should be sought from all NHS organisations involved in the study in accordance with NHS research governance arrangements.Each NHS organisation must confirm through the signing of agreements and/or other documents that it has given permission for the research to proceed (except where explicitly specified otherwise).

Confirmation of
At the time of writing, the latest version of the protocol was v1.3, dated June 15th, 2022 and the study status is ongoing, with the final participant visit scheduled in March 2024.
This trial was registered on ISRCTN registry (ISRCTN12595520), on the 25th of November 2021.Study opened to recruitment on the 20th of December 2021.
Study design and participants.The full protocol has been previously reported (Haag et al., 2022).In brief, ReDIRECT is a randomised, wait-list controlled, open-label study.The study is conducted entirely remotely, including recruitment, screening, written informed consent, randomisation, intervention, and outcome assessments.
The main inclusion criteria were people living with self-reported LC (symptoms persisting > 3 months before first recruitment contact, BMI > 27 kg/m 2 (BMI > 25 kg/m 2 for South Asians), and age ≥ 18 years.Study exclusion included lengthy hospitalisations (> 10 days) or intensive care unit (ICU) admissions related to COVID-19, people on insulin or anti-obesity drugs, proven myocardial infarction within the last 6 months, severe mental illness (including severe depression and eating disorder), women who are pregnant or considering pregnancy, history of substance abuse, active illness likely to cause a weight change, people who underwent bariatric surgery within the last 3 years or are planning bariatric surgery, advanced kidney problems (eGFR < 50 ml/min/1.73m 2), gallstones or pancreatitis, participating in another clinical research trial likely to affect diet or weight change, learning disability, and inability to understand English (written or spoken).
A total of 240 participants based in the UK were recruited through a wide range of avenues including social media (Facebook, Twitter), via support from individuals and organisations who disseminated the information within their networks (including Long Covid Scotland, Long Covid Podcast, GP practices, LC clinics, clinical research networks, councils and occupational health services across the UK, as well as colleagues in NHS Health Boards, Scottish government, and charities, including British Heart Foundation and Chest, Heart and Stroke Northern Ireland, The Wheatley Group, Men's Shed Govan, Glasgow Weight Management Service, newspaper ads (Metro and Daily Mail), and local recruitment activities (e.g., posters and flyers at Glasgow libraries and leisure centres, local pharmacies, GP practices, supermarkets and banks, advert post on the University of Glasgow Yammer platform, community health promotion events at local places of worship, and local football clubs).
Research staff assessed eligibility during a screening telephone call, and participants gave informed consent electronically before enrolment.
After completing a baseline assessment, participants were randomised to either i) treatment: Counterweight-Plus programme (a professionally supported and evidence-based weight loss programme delivered entirely remotely), or ii) control: usual health care.The primary outcome is evaluated at the sixmonth time point, after which the control group gains delayed entry to the Counterweight-Plus intervention.Participants who did not complete the baseline assessment were withdrawn (n=5), and 235 were randomised.One participant withdrew consent following randomisation.In total, baseline data was available for 234 participants.
The published protocol describes the diet intervention (Haag et al., 2022).In brief, the Counterweight-Plus programme involves a very low-calorie total diet replacement (about 850 kcal/d) and behaviour change.The formula diet is consumed for 12 weeks, followed by food reintroduction and weight management up to the end of one year.Trained dietitians provide personalised support tailored to the needs of the individual via phone or video calls and text chats.Further support is provided through in-app weekly monitoring and nudges, personalised messaging, and group support.

Outcomes and assessments.
Outcomes were assessed for all participants at baseline prior to randomisation.Data were selfreported via bespoke online questionnaires, verbally over the telephone, email and/or text messages, and collected using a web-based bespoke electronic case report form (eCRF).The questions had pre-defined answer options (yes/no) and free text boxes for additional comments.The questionnaire can be found as Extended data to the published protocol (Combet et al., 2022).Participants received digital scales (Model UC-502, A&D Instruments Ltd, Abingdon, UK) and automatic blood pressure monitors (arm type monitor TMB-1970, Kinetik Medical Devices Ltd, Redhill, UK) for measurements of body weight and blood pressure.Self-reported outcomes included COVID-19 and LC-related outcomes, body composition measurements, comorbidities, prescribed medication, wellbeing and health outcomes, work productivity, healthcare use, and food expenditure Table 1.
The CFS is an 11-item scale with four answer options: 'less than usual', 'no more than usual', 'more than usual', and 'much more than usual'.A bimodal scoring system for each item (0, 0, 1, 1) was used and summed to a total score ranging from 0 to 11.As established in the validation study (Chalder et al., 1993), a score of at least 4 was used to indicate the caseness of fatigue.A continuous score (0, 1, 2, 3) can also be used to assess the intensity of symptoms.For both scoring systems, a higher score indicates greater fatigue severity.The CFS can be subdivided into two components to separately measure physical fatigue (items 1 to 7) and mental fatigue (items 8 to 11).
The HADS questionnaire comprises two subscales for anxiety and depression, with seven items, each scored 0 to 3. A total sum of 11 or more for each subscale was used to indicate the caseness of anxiety or depression (Zigmond & Snaith, 1983).
The P4 pain scale consists of four items which assess pain levels on a numeric scale of 0 to 10 in the morning, afternoon, and evening, and with activity with a total score ranging from 0 (no pain) to 40 (pain as bad as it can be).A caseness for pain was considered with a score of 1 or more.(Haag et al., 2022).CFQ: Chalder Fatigue Scale, EQ-5D-5L: EuroQuol 5-dimensional 5-level questionnaire, LC: Long COVID, MRC: Medical Research Council.

Scales / Tools References
Self The mMRC is a tool to assess the degree of breathlessness on a scale of 0 (no breathlessness) to 4 (extreme breathlessness).
A score of at least 1 ('I get short of breath when hurrying on level ground or walking up a slight hill' or worse) indicated breathlessness.
Participants could add further LC symptoms in free text boxes along with a rating on how troublesome the symptoms were on a 10-point numeric scale.All symptoms entered at baseline were considered active.

Health status
Health status on the day of the baseline assessment was measured using the EQ-5D-5L questionnaire (Buchholz et al., 2018).This validated tool indicates health status across five dimensions for mobility, self-care, usual activities, pain/discomfort and anxiety/depression with five levels each (no, slight, moderate, severe or extreme problems).The combination of these five domains can be mapped to a health utility score, which ranges from 0 (a state as bad as being dead) to 1 (perfect health) (Hernandez Alava et al., 2023).An additional visual analogue scale asks the participant to rate their health on the day, ranging from 0 ('The worst health you can imagine') to 100 ('The best health you can imagine').Data were analysed by individually considering each dimension and the VAS and combining the five dimensions to a composite health status.

Statistical analysis.
Baseline data is presented as n (%) for categorical variables and mean (SD) or median (IQR) for continuous variables.Percentages for recruitment are based on the total number of participants recruited (n=240).All other percentages are based on the available data (n=234).
Free text responses were evaluated and grouped into categories using Microsoft Excel following a defined coding dictionary (see extended data).Free text data and pre-defined answer options were combined where they overlapped to avoid duplication.
Medication entries were analysed as medications (Table 3) if included in the British National Formulary (BNF) catalogue and grouped following the main BNF categories.Entries not considered in the medication analysis (n=15) include Androfeme, Sterimar nasal spray, several dietary supplements (multivitamins, vitamin C, cod liver oil, flushing niacin, quercetin, co-enzyme Q10, probiotics, blackcurrant seed oil, flaxseed oil, bromelain), and CBD oil.All dietary supplements entered as medications were also included in the nutritional supplement analysis (Figure 6H).CBD oil was included in private healthcare.
Group differences between those who had COVID-19 before and after vaccination were tested using the Mann-Whitney-U test in R for the median number of symptoms and the chi-squared test for the type of symptoms.A p-value of 0.05 was considered statistically significant.
All graphs were created in R using the ggplot2 package and HH package.Plot panels were compiled using Inkscape.
BioRender.com was used to partially created one of the figures.

Study population
In total, 240 participants from across the UK were enrolled in the ReDIRECT study, of whom 235 (98%) were randomised to either the diet intervention Counterweight-Plus (n=117, treatment group) or usual care (n=118, control group).One participant in the treatment group withdrew consent after randomisation, resulting in n=116 in the treatment group.
The main source of recruitment was social media (n=104, 43%), followed by other online sources, including e-newsletters, websites, and podcasts (n=40, 17%), communication through health care professionals (n=33, 14%), word of mouth (n=31, 13%), LC support groups (n=13, 5%), and newspapers (n=8, 3%).Local recruitment activities (at Glasgow Gurdwaras and football clubs) and GP practice searches conducted in 14 practices in the NHS Greater Glasgow and Clyde health board (32 invitations sent to participants) were least successful with only n=4 (2%) and n=3 (1%) recruited, respectively.One person in the Public Patient Involvement group participated; the recruitment source was unknown for 3 participants.These methods facilitated rapid recruitment of 240 participants in slightly over 6 months.
Baseline demographic characteristics are described in Table 2.
Including people who worked part-time, 44 (19%) participants were on long-term sick leave, and 7 (3%) were on reduced hours due to LC at baseline.
Based on the International Standard Classification of Occupations (ISCO-08), occupations ranged across seven major occupational groups (Figure 1B).Healthcare and teaching professionals were the most common occupation groups, with over one-third of participants working in healthcare and over 10% in the educational sector.No participant worked as a skilled agricultural, forestry and fishery worker in elementary occupations or the armed forces.
A third of participants (n=71, 30%) contracted COVID-19 at least a second time before the start of the study, which mostly occurred in the winter wave in 2021/2022 (Figure 2A).Nearly all (n=70, 99%) had at least one positive test (PCR 68%, LFT 77%, antibody test 7%).Hospital admission rates following the second infection were slightly lower than the first (n=5, 7%), and the mean stay was 3 days (SD 3).
Of the 65 participants who contracted COVID-19 after the first vaccination, 8 (3%) became ill before the second vaccination, 28 (12%) became ill after the second but before the third vaccination, and 26 (11%) became ill after the third vaccination.Data on any further infections prior to the study start were not collected.
A small proportion reported having had LC over two years (n=17, 7%).
Similarly, both breathlessness (n=228, 97%) and pain (n=219, 94%) were highly prevalent among participants (Figure 2E).The median for breathlessness was 2 (IQR 1 -2) of a maximum of 4, i.e. most participants had moderate breathing difficulties ('On level ground, I walk slower than people of the same age because of breathlessness or have to stop for breath when walking at my own pace').
Pain severity was spread across the entire range from no pain at all (score 0) to pain as bad as it could be (score 40) (Figure 2E).The median for pain was 20 (IQR 12 -28).
At baseline, most participants (n=191, 82%) were taking medication, with the total number of medications per participant ranging from 1 to 16 (median (IQR) 4 (2 -6)).Over half of the participants had been prescribed at least one medication to treat symptoms of LC (n = 124, 53%), while 28 (12%) took  at least one medication for which they reported not knowing whether they were taking it for LC symptoms or a different condition.
In total, 818 different medications were reported (Table 3), of which 42% were taken to treat LC and 51% were prescribed for other health conditions.The treated condition was unknown to the participant for 6% of medications.
A full list of medications can be found in the supplementary data as Extended data (Haag et al., 2023b).
Lifestyle changes were common after COVID.Notably, physical activity was reduced for 9 out of 10 participants (n=212, 91%) (Figure 6D).On average, median physical activity levels reduced from exercising occasionally (< 3 times per week) to walking only.The number of people who did not exercise at all rose from n=1 (<1%) before COVID-19 to n=54 (23%) at baseline.

Summary of key findings
The baseline characteristics of participants in the Remote Diet Intervention to Reduce Long-COVID Symptoms Trial (ReDIRECT) describe a diverse group of adults across the UK who are predominantly middle-aged women of white ethnicity.
The remote nature of the study allowed for rapid recruitment, with a broad geographic spread of participants across the UK.While participants were from a range of socioeconomic backgrounds, those living in the most deprived areas were underrepresented compared to those from the least deprived areas (13% vs. 27%).Two-thirds of participants had at least an undergraduate degree, with a third working in healthcare.
Representation of healthcare professionals reflects some of our recruitment strategies, including dissemination through NHS Health Boards in Scotland and electronic bulletins aimed at healthcare professionals, and the fact that front-line staff were markedly more exposed to COVID-19 in the earlier parts of the pandemic (Nguyen et al., 2020).
A critical methodological feature of the ReDIRECT study is the personalised primary outcome, with participants nominating at baseline the LC symptom that they would most like to see improve (Haag et al., 2022).The LC symptom nominated by most participants was fatigue, followed by breathlessness, pain, and cognitive impairment.A small proportion of participants named other symptoms (such as anxiety/depression, loss of taste and smell, tremors, and tinnitus) as their main n=48, 20%).The overall health utility score (combining all five domains) was mean (SD) 0.48 (0.243).The mean EQ-5D VAS was 46 (SD 17) (Supplementary Following COVID, high weight gain was prevalent across all areas of deprivation (Figure 6A).Over 90% of participants reported weight gain (n=215, 92%).A few participants reported not changing in weight (n=10, 4%), losing weight (n=6, 3%), or did not know how their weight changed in comparison to before 1%).The highest reported weight gain was 45.3 kg, and the highest weight loss was 11.5 kg.In total, the median (IQR) change in weight from before COVID-19 to the study baseline was 11.5 kg (6.9 -16.7).
Before starting the study, most participants already had experience with weight management (n=194, 83%), of which 80% had successfully lost weight in the past (Figure 6C).Commercial weight loss programmes were the most used method (n=159, 68%) and the most successful for weight loss (n=115, 72% of those who used commercial weight management).Other methods included increased physical activity, self-directed weight loss programmes (e.g., the 5:2 diet), diet change and other methods (bariatric surgery and practising meditation).Pre-COVID-19 BMI and BMI at baseline were higher in those with prior weight management experience (median pre-COVID BMI: 29.2 vs 25.6 kg/m 2 , median BMI at  16, which can be found as Extended data (Haag et al., 2023b).BMI: body mass index, IMD: index of multiple deprivation, WM: weight management.
Fatigue was reported by nearly all participants, with a high median score on the Chalder Fatigue Scale.Similarly, both breathlessness and pain were highly prevalent, affecting close to all participants.Anxiety and depression were present in approximately half of the participants, reflecting a high level of illness burden and distress across the study population, with corresponding impacts on quality of life.There is a well-established relationship between obesity and health-related quality of life.However, the EQ-5D VAS mean score of 46 (SD 17) et al., 2023;Tran et al., 2023).
Nearly all participants received healthcare support for LC through the NHS, mainly through GPs, Long COVID clinics, hospital specialist services and physiotherapy.A quarter of participants had turned to private healthcare, anecdotally because of difficulties accessing support through the NHS.Our qualitative analysis will further explore the reasons for accessing private healthcare and the additional burden on participants and their families.
Non-healthcare-related support was sought by every 3 of 5 participants, mostly through online LC communities or local support groups.Online LC communities evolved early in the pandemic, with patients sharing experiences of the after-effects of COVID-19 on social media and collectively coining the term "Long Covid" (Callard & Perego, 2021).At a time of social distancing and with evidence about LC first emerging through these patient groups, online communities were an important source of support for LC patients, providing a safe space to speak about symptoms and concerns, filling a gap in care, and helping patients find validation in others sharing similar experiences (Day, 2022;Russell et al., 2022).
In the ReDIRECT study, the number of pre-existing comorbidities and the prevalence of overweight and obesity before LC corresponded to the average in the UK population (Office for National Statistics UK Health Indicators, 2022; Organisation for Economic Co-operation and Development).Pre-existing asthma, hypertension and depression have been associated with increased risk for LC (Loosen et al., 2022;Sudre et al., 2021;Tsampasian et al., 2023), which may explain that the proportion of people with asthma in our study was greater than the age-standardised rate in the UK (18% vs 10% for men and 12% for women).While pre-COVID-19 weight data was available for only 79% of our participants, at least 71% were living with overweight or obesity prior to becoming ill with COVID-19.Interestingly, the proportion of pre-existing conditions generally associated with overweight and obesity, such as hypertension, dyslipidaemia, and type 2 diabetes, was about 5% (diabetes) to 10% (hypertension and dyslipidaemia) lower compared to the general population (Office for National Statistics UK Health Indicators, 2022).Whether this is due to the lower mean age (46), the largely female population or other factors is unclear.
Previous studies have highlighted the impacts of the COVID-19 pandemic on health behaviours that influence weight (including eating habits, physical activity, stress, and sleep), with most reporting weight gain across populations (Alah et al., 2021;Almandoz et al., 2022;Flanagan et al., 2021;Khubchandani et al., 2020;Khubchandani et al., 2022;Mason et al., 2022).The ReDIRECT study provides novel insight into body composition and weight management in this participant group living with both LC and overweight.Whilst a significant proportion of participants already lived with overweight or obesity prior to contracting COVID-19 (over two-thirds), a majority (68%) reported having gained a substantial amount of weight (5kg or more) following infection.The reasons for this weight gain may be multiple, including behavioural changes during confinement periods, compounded by fatigue and mobility issues, both common symptoms reported in this study.
There is currently limited evidence exploring this in LC, and this aspect will be further explored in the ReDIRECT trial qualitative analysis.
Most participants were already familiar with weight management before enrolling in ReDIRECT, with 4 of 5 participants having previously tried losing weight and 3 of 5 having successfully lost weight at some point in their lifetime.
Commercial weight loss programmes were the most popular and effective for losing weight, followed by increased physical activity and self-directed diet regimes.BMI was higher pre-COVID and at baseline for those who had previously attempted weight loss than those who had not.However, the weight gained post-COVID was not different between the two groups.
Behaviour changes were common after contracting COVID-19 infection.Notably, physical activity levels reduced from exercising occasionally (< 3 times per week) to walking only, with nearly a quarter of participants not exercising at all, at baseline.Attempts at dietary modifications were common, with over half modifying their diet after developing LC, commonly using approaches including calorie restriction and low carbohydrate diets as well as other diet modifications including anti-inflammatory, vegetarian/vegan, dairy-free, glutenfree, low sugar, low FODMAP, and low histamine diets, despite the lack of evidence base in this space.Similarly, dietary supplementation was common and varied.Changes and supplementation were often self-led, short-term, and without professional support.By baseline, most participants had returned to an omnivorous diet.
One of the proposed mechanisms underpinning LC is immune dysregulation and long-term systemic inflammation with elevated levels of inflammatory markers, such as interleukin (IL)-1β, IL-6, and TNF-α (Peluso et al., 2021;Schultheiss et al., 2022;Schultheiss et al., 2023;Zhang et al., 2023).Anti-inflammatory therapies in LC patients are potential treatment strategies currently being studied in several clinical trials (Bonilla et al., 2023).Weight loss has been shown to reduce pro-inflammatory markers present in the chronic low-grade inflammatory state associated with obesity, including CRP, IL-6 and TNF-a (Christiansen et al., 2010;Marfella et al., 2004;Moeller et al., 2016) and may thereby help manage LC symptoms.Weight loss may also benefit people living with both LC and overweight/obesity beyond the effect on inflammation.The symptoms commonly reported in LC, including fatigue, shortness of breath, pain, and mood disorders, are also experienced by people living with obesity.

Strengths and limitations
A key strength of the ReDIRECT study is its design, which relies on remote data collection and intervention delivery.This is particularly important in the context of the population studied since people living with LC face a high fatigue and mobility burden, sometimes compounded by careful avoidance of higher-risk situations where COVID-19 exposure may be likely.Remote delivery allowed for a flexible approach to recruitment across the UK, increased sample diversity and potentially facilitated greater inclusion by removing the burden on participants linked to attending study centres or clinics to take part.The intervention was a repurposing of the evidencebased weight loss intervention used in the DiRECT study (Lean et al., 2018), known to be safe and effective in people with Type 2 diabetes and overweight/obesity.Using a pragmatic trial design, with professional support, also recognises that dietary choices and adherence are complex and need to be evaluated individually in real-life settings.
The ReDIRECT study relied, from its conception through to its execution, on in-depth co-production with patient representatives.This co-production is fully reflected in the breadth and depth of tools to support data capture and the trial's design to reflect the requirement for personalisation and adjusting how the intervention was delivered.As such, a unique feature of the study is the selection of a personalised primary outcome, whereby participants nominate the LC symptom they would most like to improve.To our knowledge, no other trial has used such an approach.
There is currently no proven treatment or rehabilitation for LC and limited research on dietary interventions.To our knowledge, this is the first study to investigate a dietary management approach to improve LC symptoms.
The study also has limitations.While the remote nature of data collection eased the burden of not attending study visits in person, it also increased the need for computer literacy and the burden linked to engaging with data capture forms for participants, linked to anecdotal study fatigue.ReDIRECT relies on self-measurement of weight and blood pressure and self-reports.Flexibility was built into questionnaire completion timelines to minimise missing data and alleviate participant burden with assistance via email, text messages and phone calls if required.Questionnaires were also carefully checked for missing answers, and entry mistakes were followed up with participants.
In total, five participants in our study reported their first COVID-19 infection before or around the time of the first COVID-19 confirmed case in the UK (January 26, 2020) (The-nCoV Outbreak Joint Field Epidemiology Investigation Team & Li, 2020) in October and November 2019, and in January 2020.At that time, there was no testing available to confirm these infections.In all five cases, however, participants reported awareness of LC symptoms within 7 to 14 months of infection with LC diagnosed by healthcare professionals and also had a second (confirmed) infection in 2020-2022 before joining the study.
Despite focused efforts to reach out to a broad segment of the population, men are under-represented in this study, as are people from more socioeconomically disadvantaged backgrounds and minority ethnic groups.Recruitment through general practices was challenging, possibly linked to the limited use of clinical codes and challenges in identifying patients with LC (Jeffrey et al., 2023;Walker et al., 2021).Several factors may also have contributed to our study's high proportion of women.Being a woman is associated with 50% higher odds of developing LC (Su et al., 2022;Subramanian et al., 2022;Thompson et al., 2022;Tsampasian et al., 2023;Whitaker et al., 2022).The PHOSP study also found that women were less likely to have recovered from COVID-19 after one year (OR 0.68) (Evans et al., 2022).Finally, past studies have shown that structured weight loss programmes, such as the Counterweight-Plus programme, are largely used by women (Robertson et al., 2014;Stubbs et al., 2015;Tudor et al., 2021).

Conclusions
The baseline characteristics of the ReDIRECT study participants outline a complex presentation of symptoms attributable to LC but also to having excess weight, which increased substantially during the pandemic.With excess weight being a risk factor for LC and a large proportion of the participants having attempted dietary changes, often without support, in the context of their infection and subsequent LC, there is an urgent need to develop evidence supporting dietary strategies for people living with both LC and overweight, recognising that such approach will need a high level of personalisation to meet the needs of the individuals.

Underlying data
As per our study protocol, access to the raw data is restricted to the primary research team whilst the research is being conducted (until the end of the project, May 2024) and publication of the primary research papers.Upon publication of these papers, fully anonymised data will be placed on a research data repository with access given to researchers on request to the corresponding authors and subject to appropriate data sharing agreements.

Jessica Campbell
Auckland University of Technology, Auckland, New Zealand Agreeing with peer reviewer 1, this is a well written and interesting paper which I enjoyed reading.
It would however benefit from some clarification regarding the aim of the paper.In particular, when reading the abstract, it was not initially clear that results would not be presented, though the title of the paper is clear.Additionally, below are a few comments/suggestions for the authors to consider.Add details regarding which version of R was used.1.
Clarify the use of statistical tests.For example, were Mann-Whitney tests chosen due to data failing the assumptions of parametric testing, or was this always intended to be the test of choice?

2.
Throughout the paper it would be helpful to make a clearer distinction between acute Covid-19 infection and Long Covid.If the authors have data, it would be interesting to see more detailed information about when LC started (rather than just how long individuals had been experiencing LC symptoms).Data are presented relating to the dates of each acute Covid-19 infection and about these infections in relation to vaccination; however, it would have been nice to see these events in relation to LC.For example, was LC more likely after the first or second infection?Was it more likely before or after vaccination?This wasn't entirely clear.

3.
Similar to point 3, there is discussion throughout the paper about changes after Covid-19 (for example, weight gain).It would be good to clarify whether this relates to LC specifically, or to the first acute infection, the second acute infection, etc.In some cases, this may be difficult to determine, but acknowledgment of this would be beneficial.

4.
I suggest changing 'vegan' to 'plant-based' given that the former is an ethical stance related to broader changes in lifestyle, whereas the latter refers specifically to a diet which may be adopted for health reasons (as is more appropriate in this context).

5.
It is perhaps worth noting or clarifying that some of the changes that occurred post-Covid may be related to more general pandemic responses rather than being (only) related to illness.For example, weight gain and reductions in exercise were widely reported due to 6. lockdown measures even in otherwise healthy individuals.If it is possible to rule this out, a comment noting this would be beneficial.While (online) LC support groups were undoubtedly helpful for many, it may be worth noting the downsides of these groups e.g.(Turner J., 2017) 7.
While the limitations section notes that self-report measures were used, this could be explained more explicitly.

8.
The discussion mentions that several patients who reported infection before testing was widely available became aware of LC within 7 to 14 months post-infection.This point may need further elaboration.Rather than focusing on the lack of testing, consider discussing the significant time gap between infection and the onset of LC symptoms.This delay makes it challenging to ascertain whether the symptoms are genuinely related to Covid-19 or if they might be idiopathic.Reference missing.This point is important as your hypothesis states that obesity is a risk factor for LC.
○ 5 th paragraph, 1 st sentence Reference missing.This point is important, as your hypothesis is whether a dietary modification may influence LC.

○
Last paragraph.This is the real M&M.This text should also be at the abstract at introduction and M&M sections.○ Study design and participants.Last paragraph, last sentence.One question: "Further support is provided through in-app weekly monitoring and nudges, personalized messaging, and group support" this was also done with the control group in the first 3 months and until de year? ○

RESULTS
Long COVID 1 st paragraph, 2 nd sentence.This is a very important epidemiological finding!Hence, think about writing this finding in the abstract and conclusions sections.
○ 7 th paragraph, 2 nd sentence.Regarding cognitive impairment.This is also a ○ ○ 1.The aim is not clear in the abstract.Please add the aim of the study to the abstract.2. Please clarify aim on page 4, I was confused whether the paper is going to talk about the impact of the intervention on listed outcomes or just describe characteristics of the study population, please clarify the exact aim of the paper.3. Please define abbreviations of "UoG" and "PPI" on page 4. 4. On page 5, you could remove the section that starts with "The REC favorable opinion...." to "except where explicitly specified otherwise".this is by default a requirement by ethics whenever an NHS site is included.5. Could you please whether the answers to the "living situation" in table 2 is "select all that applies" as total number for this part exceeds 234 participants.6. on page 16, the sentence "To our knowledge, this is the first study to investigate a dietary management approach to improve LC symptoms".the paper only reports baseline characteristics of included participants, thus I'm not sure this line fits with this paper, especially that no impacts are mentioned related to the use of the intervention.

Are sufficient details of methods and analysis provided to allow replication by others? Yes
If applicable, is the statistical analysis and its interpretation appropriate?I cannot comment.A qualified statistician is required.

Are the conclusions drawn adequately supported by the results? Yes
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: I'm a research Fellow with experience of conducting research in Long COVID.
I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

Figure 1 .
Figure 1.Employment status and occupation groups before COVID and at baseline.(A) Participants were asked to choose one of the six categories in the figure which best described their employment status before COVID-19 and at baseline.Pre-covid, ' other' includes homemaker and full-time carer/part-time student.At baseline, ' other' includes long-term sick leave, homemaker, disabled, part-time student, and agency worker.(B) Occupations were classified based on the International Standard Classification of Occupations (ISCO-08) and ranged across seven of the ten major occupational groups both before COVID-19 and at study baseline.

Figure 2 .
Figure 2. Covid-19 infection and Long COVID symptoms.(A) COVID-19 infection dates: First COVID infection dates ranged from October 2019 to March 2022, and second infection dates ranged from November 2020 to July 2022.Most infections were confirmed by at least one positive COVID test (PCR, rapid antigen lateral flow test, or antibody test), shown in orange and dark blue.Yellow and light blue bars show the number of infections which were not confirmed by a COVID test.(B) Long COVID duration: Figure shows the length of time in months participants were living with Long COVID at the start of the study.Long COVID was diagnosed either by a GP (yellow), a different healthcare professional (blue), or self-diagnosed (grey).GP: general practitioner, HCP: healthcare professional, LC: Long COVID.(C) Percentage of the Long COVID symptom each participant selected as the one they would most like to see improve.The most selected symptom (fatigue) is highlighted in yellow.(D) Number of Long COVID symptoms (core symptoms and symptom groups) in people who had COVID before (yellow) and after (blue) vaccination.Difference between the groups was significant (p=0.003).(E) Core Long COVID symptoms: Fatigue, breathlessness, pain, anxiety, and depression were assessed using validated questionnaires (Chalder Fatigue Scale, modified MRC Dyspnoea Scale, P4 pain scale, Hospital Anxiety and Depression Scale).Fatigue is represented as total fatigue score and bimodal scoring.The y-axes show complete range of possible scores for each scale (0 to 33, 11, 4, 40, 21, 21 for fatigue total, fatigue bimodal, breathlessness, pain, anxiety, and depression, respectively), with higher scores indicating greater severity.The grey dashed line shows the cutoffs used for caseness of the symptoms.No cutoff for total fatigue score was used to determine caseness.

Figure 3 .
Figure 3. Overview of Long COVID symptoms reported in the ReDIRECT study.Participants reported Long COVID symptoms outside of fatigue, breathlessness, pain, and anxiety/depression using free text boxes.Partially created with Biorender.com.

Figure 4 .
Figure 4. Existing comorbidities prior to the first COVID infection.(A) The number of comorbidities per participant and (B) the 16 most prevalent comorbidities which were present before the first COVID infection.

Figure 6 .
Figure 6.Weight management, diet and lifestyle changes from before COVID to baseline.Pre-COVID weight data was provided by 184 (79%) participants.The change in weight from before the first COVID infection to the baseline assessment is shown per IMD quintile (A) and BMI category prior to COVID (B).BMI (kg/m 2 ) was categorised as follows: < 25 = normal weight, 25 -30 = overweight, 30 -35 = obesity class I, 35 -40 = obesity class II, > 40 = obesity class III.The number of participants per BMI category before COVID and how their weight status had changed at baseline is illustrated in (E).(C) Most participants had prior experience with weight management using different methods (yellow) with varying degrees of success in losing weight (blue).A minority did not attempt to lose weight prior to joining the study (grey).(D) Physical activity levels before COVID and at baseline were reported in four levels: regular exercise (≥ 3 times per week), occasional exercise (< 3 times per week), walking only, or no physical activity at all.The black diamond and line show the change in median activity levels before and after COVID.(F) Dietary changes following the development of Long COVID were common.Other diets not explicitly listed include reduction in lactose, gluten, FODMAPs, sugar, as well as vegetarian, pescetarian, anti-obesity medication, and general healthy eating.(G) shows the diet consumed at baseline.Other includes reduction in gluten, dairy, lactose, sugar, FODMAPs and increase in healthy eating.Supplements taken at baseline are listed in (H).For the full list of supplements, see Supplementary Table16, which can be found as Extended data(Haag et al., 2023b).BMI: body mass index, IMD: index of multiple deprivation, WM: weight management.
J: Online Support Groups: The Good, the Bad, and the Motivated.Journal of Consumer Health on the Internet.2017; 21 (1): 11-25 Publisher Full Text Is the work clearly and accurately presented and does it cite the current literature?Yes Is the study design appropriate and is the work technically sound?Yes Are sufficient details of methods and analysis provided to allow replication by others?Yes If applicable, is the statistical analysis and its interpretation appropriate?I cannot comment.A qualified statistician is required.Are all the source data underlying the results available to ensure full reproducibility?Yes Are the conclusions drawn adequately supported by the results?Yes Competing Interests: No competing interests were disclosed.Reviewer Expertise: I have previously published research on Covid-19 and currently focus on studying the effects of nutritional interventions.My work involves exploring how diet and lifestyle changes can impact health outcomes, particularly in the context of chronic disease.I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.Reviewer Report 04 July 2024 https://doi.org/10.3310/nihropenres.14676.r32173© 2024 Tomasa-Irriguible T. This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Teresa Maria Tomasa-Irriguible Hospital Universitari Germans Trias i Pujol, Badalona, Spain Reviewer Comments.Baseline Characteristics in the Remote Diet Intervention to REduce long-COVID Symptoms Trial (ReDIRECT) ABSTRACT Methods section: Does not reflect what the paper is about.What is the objective/s?"Baseline Characteristics"?You could write this sentence: "Here, we describe the baseline characteristics of ReDIRECT participants, focusing on socioeconomic and demographic factors as well as dietary and medical history, including LC diagnosis and symptoms".
sentence.I think that could be a problem, since they may not offer a fairness judgment.○ Ethical approval.3 rd paragraph.Think about moving this paragraph to 3 paragraphs below, in Study design and participants before "A total of 240 participants..."

Table 3 .
Medications.Frequency and type of medications taken at baseline.

confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
Need to clarify what is your rationale for selecting the tools for XYZ symptoms, on the other hand to clarify whether these tools were selected after participants answered the baseline questionnaire and selected the most burdensome symptoms.The first paragraph of page 8 does not explicitly say this.RESULTSNeed to specify for those decreased work hours whether this was due to their own disease or due to lockdown conditions or other reasons, to avoid cofounders or speculation DISCUSSION, limitations and conclusions This is the section I enjoyed the most as you reviewed several papers.More data on the influence of the pandemic on health and wellbeing during lockdowns(Tronco Hernandez et al, 2021)as well as experiences reported by people with long COVID and lessons for professionals and researchers(Tronco Hernandez et al. 2023) would provide more breadth to this body of evidence Good to acknowledge the major limitation self reports and lack of training to self-measure weight and blood pressure The burden of the questionnaire on participants needs to be considered too In general, researchers studied a lot of variables, some may be cofounders and lead to spurious assumptions.I suggest you divide them into primary (weight, blood pressure, etc) and secondary variables (e.g.socioeconomic status, vaccination status) otherwise there is too much data hence the main aim can get easily lost for readers or they could make erroneous assumptions Reviewer Expertise: I am a postdoctoral researcher who has published several papers on the effects of covid on people and how to best support professionals and employers delivering wellbeing and dietary strategies to patients with Long covid I This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.