The REFORM study included men (n = 16) and women (n = 34) aged 20–49 year who were not underweight or obese (BMI between 18.5 and 30 kg/m2) who were recruited from the local community of Reading, UK, in one cohort between March and September 2012. The intervention visits ran between May 2012 and March 2013. Recruitment strategies included targeted mailings to volunteers on local participant databases. Posters and flyers were circulated around the university campus as well as social and community groups in the Reading area. Key exclusion criteria were as follows: clinical diagnosis of diabetes, CVD, anti-inflammatory or hypertensive medication, smoking, excessive alcohol consumption (>21 units per week for men and >14 units per week for women), participation in regular vigorous exercise or fitness training (≥20 min × 3 times/week), pregnancy or lactating. The Joint British Societies’ JBS-2 total CVD risk calculator was used to estimate the probability (percentage chance) that a participant would experience a CVD event over the next 10-year period . Dietary restraint was quantified using the Three-Factor Eating Questionnaire, with a score >11 (Factor I) indicating restrained eating [23, 24]; this score was used to characterize the cohort, but it was not used as an exclusion criteria for study participation.
Participants were not aware that the primary outcome of the study was body weight, as this could have influenced the study outcome. They were told that the purpose of the study was to examine the effect of reformulated fat, salt or sugar-reduced beverage and food items on risk factors for CVD. All procedures involving human participants were approved by the University of Reading’s Research Ethics Committee (12/03). Written informed consent was obtained from all participants before participation. The REFORM study was registered a clinical trial (Clinicaltrials.Gov ID: NCT01645995).
The REFORM study was performed using a randomized, controlled, double-blind, crossover design and consisted of two 8-week dietary exchange intervention periods. Participants attended the Hugh Sinclair Unit of Human Nutrition on four separate occasions (before and after the two intervention phases). A computer-based minimization procedure (minimized on age, gender, BMI) was used to randomly allocate participants to consume either regular or sugar-reduced products (i.e., diet A or diet B) during their first 8-week dietary exchange period . This was subsequently followed by a 4-week washout period after which the participants began the alternate study intervention.
Participants were advised prior to each visit to avoid strenuous physical activity (PA) and alcohol consumption for a 24-h period. Following a standardized low-fat evening meal (<400 kcal; <9 g fat), participants completed a 12-h fast.
Dietary exchange model
For the REFORM dietary intervention, a flexible food exchange model, based on average UK food intake data of adults aged 19–64 year, was developed using a well-established approach that allows for minimal disruption to the habitual diet of free-living individuals . The amount of exchangeable sugar in the free-living UK diet was calculated from the NDNS database  as the total NMES present in the following easily accessible food sources: breakfast cereals, baked beans, puddings, chocolate, sweet confectionary, sweet spreads, savory sauces, condiments, soft drinks, fruit juice, yoghurt, ice cream and other milks (Table 1). It was estimated that 60 % of the sugar-containing processed foods and beverages could be exchanged so as to manipulate the overall NMES composition of the habitual diet. Commercially available regular and sugar-reduced products, with a specific NMES profile, were tested in the dietary exchange model; it was quantified that the minimum daily NMES difference between the regular and reformulated dietary exchange periods was 32 g/day, with the reformulated arm consuming <7 % EI as NMES.
Dietary exchange intervention and compliance monitoring
For each 8-week dietary exchange period, participants exchanged ≥1 beverage and ≥1 food portion per day from their habitual diet with equivalent sugar-containing or sugar-reformulated products (for details of study foods, see Online Resource 1). After completing baseline study visits (at the beginning of each dietary assessment period), participants were given personalized advice about the dietary exchange intervention and how to incorporate the study products into their diet. Participants were given adequate study product supplies for a 4-week period at the beginning of each dietary assessment period. They also attended the Hugh Sinclair Unit of Human Nutrition for a food collection visit and dietary compliance assessment at week 4, the midway point in each dietary exchange period. During this visit, completed compliance sheets were assessed and any issues with dietary adherence were addressed and rectified. To encourage dietary adherence and reduce the likelihood of ‘product boredom’ , participants were provided with sufficient study products from a choice of 7 beverages including juice drinks and soft drinks, and 19 foods, including pasta sauce, baked beans, muesli, puddings and sweet confectionary (for details, see Online Resource 1). Participants were asked to replace habitually used sugar and condiments with those provided by the investigators ad libitum throughout the dietary exchange periods (Online Resource 1). Study product selection was based on each participants’ response to a food preference screening list that itemized the beverage, food, sugar and condiments options that were available to them as part of the dietary exchange intervention. They were asked to consume similar types of REFORM beverages/foods during their second arm of the study so that the regular and reformulated product types would be matched. Participants completed a daily dietary compliance sheet to assess their intake of the study products during the two 8-week dietary assessment periods (i.e., for a 56-day period). At the end of the first diet period, participants were asked to return unopened leftover products to the Nutrition Unit. They were instructed not to consume leftover study products during their washout period and were advised to return to their habitual diet.
Study products were presented to volunteers in a blinded manner. An investigator not involved with assessing study outcomes de-branded, relabeled or repackaged the study products so that they appeared identical between interventions, with the two dietary exchange products identifiable by diet A or B. The minimum beverage or food portion size was stated by weight and in general household measures.
Assessment of intervention foods and blinding strategy
At the end of each 8-week dietary period, the study products were assessed for visual appeal, smell, taste, aftertaste and palatability using visual analogue scale (VAS) questionnaires . Each VAS assessed a sensation on a 100-mm horizontal line, anchored at the beginning and end by opposing statements. On completion of the study, a retrospective questionnaire assessed participants’ perceived awareness of the order in which they had been assigned to the regular and reformulated study products.
Assessment of dietary intake
Four-day weighed food diaries were completed prior to commencing and during the final week of the dietary exchange periods (week 0, 8, 12, 20). Participants completed their food diaries over three weekdays and one weekend day, with the same days repeated for subsequent food diaries. Detailed verbal and written instructions for completing their 4-day weighed food diaries were given to participants ≥7 days prior to their first study visit. A set of sample diaries and a set of digital scales were also provided, and the importance of not changing habitual dietary patterns was emphasized. Participants were asked to record recipes used during cooking and retain packaging from ready meals so that the items could be added to the dietary database. During study visits, food diaries were assessed for completeness. If necessary, further detail was collected to facilitate precise data entry.
Dietplan software (version 6.60; Forestfield Software Ltd.) was used to calculate specific energy and macronutrient intake. The nutrient composition of foods consumed was based on NDS Nutrient Database or McCance and Widdowson’s food tables. Dietary fiber intake was defined as non-starch polysaccharide (NSP) using the technique of Englyst and Cummings . The composition of the study products was also added to Dietplan, based on manufacturer details.
Assessment of physical activity
Participants were instructed to wear a triaxial accelerometer (GT3X + activity monitor; Actigraph, LLC) directly above the right iliac crest during sleeping and waking hours (except during water-based activities) for seven consecutive days, over the same time period that dietary intake was assessed. Device initialization, data processing and analysis were conducted using Actilife data analysis software (version 6.4.5). For analysis inclusion, participants were required to have produced counts on their activity monitor for ≥4 days (>600 min/day of wear time) . Non-wear time was defined as ≥60 min of zero activity counts . Data were summarized in 60-sec epochs, and mean EE from PA (EEPA) was calculated (kcal/day). Total daily step count, based on accelerometry, was also calculated and was used to estimate PA levels with <5000, 5000–7400, 7500–9999 and 10,000 steps/day categorizing participants as sedentary, low active, somewhat active and active, respectively [32, 33].
Basal metabolic rate
Basal metabolic rate (BMR) for each participant was calculated based on age, gender and body weight using the Henry equation .
Upon arrival at the Nutrition Unit, height was measured to the nearest 0.1 cm using a wall-mounted stadiometer (first visit only). While wearing light garments, body mass (kg) and body fat (%) were assessed by Tanita Body Composition Analyzer (BC-418 MA; Ill, USA) using standard settings (normal adult body type and 1 kg for clothing) at week 0, 8, 12 and 20.
Blood pressure and arterial stiffness
Participants rested semi-recumbent in a temperature-controlled vascular suite (22 ± 1 °C). After a 20-min rest and familiarization period, BP was recorded in triplicate using an ambulatory upper arm BP monitor (ABPM TM-243; A&D Instruments Ltd.). Digital volume pulse (DVP) and pulse wave analysis (PWA) were assessed as markers of vascular stiffness. DVP was determined by placing a photoplethysmographic transducer on the left index finger (Pulse Trace PCA 2 device; Micro Medical Ltd.). Measurements were taken in triplicate for calculation of mean reflection index (DVP–RI) and stiffness index (DVP–SI) . PWA was carried out by a single trained operator using the SphygmoCor (ScanMed Medical; see  for details). The pulse pressure wave at the radial artery was recorded using applanation tonometry . The mean of two values with the highest operator index (≥80) was averaged to calculate the augmentation index (AIx), a measure of overall systemic arterial stiffness and AIx adjusted to a standard HR of 75 bpm (AIx HR75).
Blood sampling and biochemical analysis
Venous blood samples were collected and centrifuged at 3000 rpm for 15 min at 4 °C. Serum and plasma were separated and stored at −80 °C until subsequent analysis. Concentrations of serum lipids [total cholesterol, high-density lipoprotein cholesterol and triacylglycerol (TAG)], C-reactive protein, non-esterified fatty acid and glucose were determined using enzyme-based colorimetric tests on an ILab 600 Clinical Chemistry Analyzer (Instrumentation Laboratories Ltd.). The ratio of total:HDL cholesterol was calculated. LDL cholesterol concentrations were measured according to the Friedewald equation . Plasma insulin concentrations were determined with enzyme immunoassay kits (Alere Ltd.). Samples from each participant were analyzed in a single run to minimize inter-batch variation.
Energy intake and body weight simulation tool
For each participant, baseline variables including height, weight, age and gender were entered into the NIDDK human body weight simulation tool . PA level, predicted from mean daily step count categories which were coded to match corresponding activity levels 1.4, 1.5, 1.6 and 1.9 in model [16, 33], was also entered. Collectively, these variables were used to provide a prediction of baseline EI (kcal/day).
For the body weight simulation, it was a required assumption that the PA level remained constant for the duration of each dietary exchange period. For body weight simulation, the value of 181 kcal/day was chosen to represent the mean energy deficit of the group during the reformulated dietary exchange period. This was calculated from the average of each participant’s 56-day caloric difference between the regular and reformulated dietary exchange periods (Online Resource 2). The energy deficit value was subtracted from the predicted baseline EI and predicted weight, after a 56-day body weight simulation was recorded. For body weight simulation following the regular dietary exchange period, it was assumed that EI remained constant, i.e., the same as that predicted at baseline.
Power calculations were based on body weight, our primary outcome measure. An estimated weight loss of 1.2 kg was predicted with a daily conservative target difference of 181 kcal/day between the regular and reformulated dietary exchange phases for an 8-week period. The SD of weight reduction was estimated as 1.1 kg. Based on a previous calculation , an estimated recruitment of n = 37 participants was deemed necessary to give sufficient power to detect significant changes in our secondary outcome measures including biochemical analysis, if energy compensation did not occur, with P < 0.05 and 80 % power. With the allowance for a 25 % dropout rate, we aimed to recruit a minimum of 47 participants.
All statistical analysis was performed on primary (body weight) and secondary outcome measures using SPSS (version 19.0; SPSS Inc.). Data that were not normally distributed, as assessed by the Shapiro–Wilk test, were logarithmically transformed. The potential impact of an intervention order was assessed by adding it as a main effect to our repeated measures ANOVA model. However, because no significant interactions between intervention order and our other main effects (time and diet) were detected, a two-way ANOVA was used to identify significant time X diet (T X D) interactions. Bonferroni post hoc corrections were applied to all data to control for multiple comparisons. For our primary outcome measure, multiple adjustments were made for age, gender and dietary restraint using a two-way ANCOVA. The association between observed and predicted (using the body weight simulation model ) body weights was analyzed using Pearson’s correlations. Paired t tests were used to compare dietary compliance data. Differences in baseline characteristics between participants randomly assigned to the regular and reformulated dietary exchange arms were assessed by using the independent t tests and the Chi square test for continuous and categorical variables, respectively (Online Resource 3). Data are presented as mean ± SD. P values < 0.05 were deemed to be significant.