Methods
Fig 1 summarises the multi-stage approach used to develop and test SCREEN-IT. Participants aged 18 years and over were recruited (using word of mouth, departmental databases, community groups, posters distributed in the local area) in line with the inclusion criteria (no health conditions impacting food intake, not following a weight-loss diet, not pregnant or breastfeeding, no cognitive impairments, food allergies or intolerances, living in the UK and having access to a digital device). If interested, participants read the information sheet (described as a study on testing different communication strategies to prevent bias), had an opportunity to ask any questions and provided informed consent. The research received a favourable ethical opinion for conduct by the University of Reading School of Chemistry, Food and Pharmacy Research Ethics Committee (study number: 26/2024) and was performed in accordance with the Declaration of Helsinki as well as registered with clinicaltrials.gov (NCT06675630).
Development of SCREEN-IT
SCREEN-IT development was a five-step process as summarised in Fig. 2 with the overall goal of creating a quick and easy tool to capture dietary fibre intake in the UK population.
Step one: identification of key dietary fibre-rich food categories was based on commonly consumed foods within the UK diet and/or likely to resonate (user-friendly language) to help develop the tool with food-level insights captured from the NDNS [2]. This was supported by: (a) a previous Dutch dietary fibre-related screening tool for design purposes; (b) analysis of commonly consumed dietary fibre-rich foods (all selected foods groups were within this ranking apart from vegetables); and (c) dietary fibre categories were initially piloted for suitability in a real-world setting and based on this feedback an extra category (dietary fibre snacks) was included to capture food items such as cereal bars, oatcakes, flapjack and popcorn [28, 30, 31]. This resulted in selection of ten dietary fibre-rich food groups: (1) wholemeal bread; (2) dietary fibre-rich cereals; (3) wholemeal pasta and rice; (4) potatoes; (5) dried fruits; (6) fresh and frozen fruits; (7) vegetables; (8) pulses; (9) nuts and seeds; and (10) dietary fibre-rich snacks (Table S1).
Step two: determination of the ten most frequently consumed food items within each category were determinated utilising the NDNS food-level data (Rolling Programme Years 6–9) to enable direct comparison with the selected UK-based FFQ [32]. The dietary fibre content (AOAC) per 100 g was collated from the McCance and Widdowson’s Composition of Foods Integrated Dataset [33] to calculate a weighted mean dietary fibre content per food item and category based on frequency of consumption.
Step three: focused on scale development; a six-point scale (never to always) was used to capture consumption habits during the previous two weeks [28]. The resulting daily frequencies were defined quantitively as never (no; 0), rarely (less than once; 0.14), sometimes (one to two days; 0.29), often (three to four days; 0.57), mostly (five to six days; 0.86) to always (daily; 1) [28].
Step four: selected two representative food-based examples per dietary fibre-rich category to support portion size estimation derived from healthy eating guidance and this determined dietary fibre (g) per typical portion size [34]. For example, 80 g of baked beans and chickpeas were presented for pulses (Figure S1) [35, 36]. The selected foods (n = 20) were prepared, cooked (as required), weighed and plated for this purpose. A series of digital photographs were taken using an agreed protocol by a professional photographer (Bennetto Photography, Reading, UK) with previous food-based photography experience [32, 37, 38]. The number of portions was only visible for SCREEN-IT if a frequency between “rarely to always” was selected. Five options were presented from “one portion to five or more portions” (with a comment box if outside this range) where this information was then defined quantitively (e.g., one portion is one and so on).
Step five: developed the dietary fibre calculation adapted from Rijnaarts et al. [28]: frequency [six-point scale] x portion size [five-point scale] x dietary fibre content [g per typical portion size]; all ten dietary fibre categories were then aggregated to provide an overall dietary fibre intake (g/d) (Figure S2).
SCREEN-IT validity
This stage estimated dietary fibre intake from SCREEN-IT compared with a UK based eNutri-FFQ in line with previous research (country-specific FFQ and two-week timeframe) [27, 28]. Cade et al. [39] noted that a mimimum of 50 participants was a suitable sample size for FFQ validation studies. To cater for potential dropouts or implausible data, at least 77 participants were invited to complete SCREEN-IT or eNutri-FFQ and then seven days later fill out eNutri-FFQ or SCREEN-IT using a single-blinded randomised crossover design.
SCREEN-IT was self-administered and deployed digitally using the Compusense platform (Compusense Cloud, Ontario, Canada). Participants were presented with ten dietary fibre-rich categories monadically and asked if they consumed each category in the previous two weeks (six-point scale: never to always). If so, the number of portions typically consumed were recorded (five-point scale: one to five or more portions with a comment box option if outside range) supplemented with two portion size photographs per category. The overall dietary fibre value (g/d) was manually calculated using the equation described in step five. eNutri-FFQ is a web-based tool that quantifies food and nutrient intake from the previous four weeks via a validated, up-to-date and comprehensive UK-based FFQ as described elsewhere [32, 40, 41]. Prior validation of eNutri-FFQ demonstrated acceptable agreement of dietary fibre intakes against three-day weighed diet records [32]. In brief, it records how often participants consumed each of the 157-items [food+drink] (not in the last month, less than once a week, once a week, 2–4 times a week, 5–6 times a week, once a day, twice a day, 3–4 times a day or 5 + times a day) and usual portion sizes via selecting one of seven portion size buttons (three photographs and four text descriptions). Using the inputted data, eNutri automatically calculates the mean daily intakes of food items and subsequently estimates intakes of a range of food groups and nutrients.
For both tools, three key parameters were recorded: dietary fibre intake (g/d), time to complete (min; from software derived timestamps) and usability (10-item System Usability Scale (SUS) as outlined by Brooke [42] with a five-point scale (strongly disagree to strongly agree) and converted into a score out of 100). In addition, basic demographic information was captured (age, sex, self-reported weight/height, living status, education level and employment type).
SCREEN-IT reproducibility and usability
This stage focused on assessing tool reproducibility and usability using a larger sample size balanced across three key age-groups (younger: 18–40: middle: 41–64 and older: 65 + years) for representativeness aiming for 50 participants per group [39]. Participants (n = 165 to allow for 10% dropouts) were invited to complete SCREEN-IT on two occasions four-weeks apart (retest approach; SCREEN-IT 1 and SCREEN-IT 2) on the Compusense platform. Five parameters were recorded (i) usability (SUS) [42]; (ii) device screen size (small: smartphone; medium: tablets; and large: laptop/desktop computer); (iii) time to complete from software derived timestamps (min); (iv) feedback on SCREEN-IT (open-ended comment box); and (v) basic demographic information.
Statistical analysis
A pre-defined statistical protocol was utilised in line with recommendations for testing validity of dietary assessment tools and interpretation criteria as described by Lombard et al. [29]. Dietary fibre intake (g/d) was the variable of interest for comparison using the following techniques: (1) intra-class correlation coefficients (ICC) agreement between tools (poor: <0.20, acceptable: 0.20–0.59, good: >0.61) using the inter-rater model; (2) paired samples t-test (agreement at group level); (3) percentage difference (size and direction of error) where the mean difference (A – B) was divided by average of both methods ((A + B)/2) multiplied by 100 (poor: >20.0%; acceptable: 11.0–20.0%; good: 0.0-10.9%); (4) cross-classification (agreement at an individual level) was estimated via quartile analyses to calculate the proportion of participants within exact (same quartile for both methods), adjacent (one quartile apart) or extreme (opposite quartiles) agreement; (5) Cohen’s weighted kappa agreement at an individual level (poor: <0.20; acceptable: 0.20–0.59; good: >0.61) was performed on categorical data computed utilising a contingency table; and (6) Bland-Altman plots (presence, direction and extent of bias) were visualised in accordance with Bland and Altman methodology [43] in XLSTAT (version 2023.3.1.1416, New York, USA) and GraphPad Prism (version 10.4.2 (534), Boston, USA). In addition, age-related differences were explored using analysis of variance (ANOVA) and if significant, Tukey’s Honestly Significant Difference test for pairwise comparisons; p < 0.05 was considered significant for relevant analyses. Feedback on SCREEN-IT was captured by an open-ended question and collated by identifying recurring themes (e.g., at least five mentions) [44].
CC BY 4.0. "Closing the dietary fibre gap - developing a novel dietary fibre screening tool (SCREEN-IT) for the UK population: validity, reproducibility and usability insights" by Norton V, Lovegrove JA, Weech M et al., European journal of nutrition (2026). https://doi.org/10.1007/s00394-026-04033-4. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Peer-reviewed and open access; full text from PubMed Central (PMC13332889). Headings and layout adapted for reading; figures and tables are in the original.