A Novel Personalised NUtrition Intervention With Tailored Behavioural Support to Promote Sustainable and Healthy Diets in University Students: the P-NUTS Randomised Controlled Trial
A Novel Personalised NUtrition Intervention With Tailored Behavioural Support to Promote Sustainable and Healthy Diets in University Students: the P-NUTS Randomised Controlled Trial
The aim of this study is to test the efficacy of a novel personalised nutrition intervention with tailored behavioural support compared to a control personalised nutrition intervention for improving adherence to sustainable and healthy dietary recommendations in young (18-30 years), healthy university students.
Recruitment:
Participants will be recruited from the University College Dublin (UCD) PLAN'EAT Living Lab (LL) Citizen Panel (https://clinicaltrials.gov/study/NCT06939231), as well as from the wider eligible UCD university population. Eligible participants will be invited to participate in a 4-week study.
Power analysis:
The trial was powered for the group main effect in an analysis of covariance (ANCOVA) of the endpoint primary outcome variable adjusted for the baseline value of the primary outcome variable. Using G*Power 3.1 (F tests → ANCOVA: fixed effects, main effects and interactions), we set α = 0.05, power = 0.80, numerator df = 1, number of groups = 2, and number of covariates = 1. The effect size was taken from the primary outcome of the pilot trial (legume intake, g/d): partial eta squared (η²p) = 0.134, converted to Cohen's f, using the formula:
f = √η²p/(1- η²p)
where η²p = 0.134, resulting in a Cohen's f of 0.393. The power calculation indicated an estimated required total sample size of 53 participants. Allowing for 20% attrition (pilot trial attrition rate: 17%), the estimated required total sample size increased to 66 participants (n=33 per group). Pilot trial registration: https://clinicaltrials.gov/study/NCT06631469.
Randomisation:
After completion of successful screening and informed consent, participants will be randomised in a 1:1 ratio to either the control or intervention group using block randomisation stratified by sex, using randomly permuted blocks of size four within each sex group to maintain equal allocation (using the blockrand package in R). This randomisation method was chosen as it resulted in a balanced distribution of demographics and baseline dietary intake data across intervention and control groups in the pilot trial.
Intervention overview:
Each group will receive different types of personalised nutrition interventions using decision tree algorithms to support them to consume a healthier and more sustainable diet. The control group will receive a personalised dietary plan and standardised behavioural support (based on the standardised support provided in the pilot trial, which was based on previous personalised nutrition trials and usual care in dietetic practice) to meet their personalised dietary recommendations. The intervention group will receive the same intervention as the control group, with additional tailored behavioural support to help them meet their personalised dietary recommendations.
Baseline:
Participants will be asked to attend a study visit at the UCD Institute of Food and Health where they will complete baseline questionnaires (i.e., consent form, demographics, health, lifestyle, medical history, medication use, supplement use, dietary behaviour questions) and a study nutritionist-guided 24-hour dietary recall via the web-based tool, Foodbook24. Upon giving consent, the participant will also be provided with two urine collection kits (one for study baseline and endpoint) and instructed to collect their urine first thing in the morning (i.e., first-void) and bring it with them on the day of their second baseline in-person visit at the UCD Institute of Food and Health. The participant will be asked to self-complete two further dietary recalls on non-consecutive days before attending the second in-person visit (for a total of three); however, participants will be allowed to attend the second in-person visit in the case that only two 24-hour dietary recalls are completed (in total).
During the second baseline in-person visit at the UCD Institute of Food and Health, a member of the research team will measure participant anthropometrics and blood pressure, while a trained nurse will take a blood sample (if consent is given). Thereafter, the study nutritionist will deliver the intervention to participants via a diet counselling session. During this session, the study nutritionist will communicate the personalised dietary recommendations developed for the participant, providing information and support on how to achieve their recommendations, along with a resource booklet with supporting information and materials (e.g., recipes, shopping lists, etc.), while asking the participant to sign a behavioural contract, where they affirm their commitment to achieving their dietary recommendations throughout the 4-week study period. Participants will be provided with a maximum of three dietary recommendations in total, which can be any combination of the following: 1) increase legume intakes; 2) increase fruit and vegetable intakes; 3) decrease meat intakes; and 4) decrease sweet/savoury discretionary food intakes.
In addition to the above, participants in the intervention group will receive tailored behavioural support to further help them to achieve their personalised dietary recommendations. The tailored behavioural supports will be developed to overcome participants' main barriers to achieving their specific dietary recommendations towards a sustainable and healthy diet. The specific tailored interventions for participants may include: provision of foods, cooking demonstration videos, habit-building support, motivation-building support, advice to seek social support, or additional knowledge provision (depending on the participant's main barriers). The main barriers to achieving specific dietary recommendations towards a sustainable and healthy diet were identified through quantitative and qualitative data collection in young (18-30 years), healthy UCD students, as part of work conducted within the PLAN'EAT Project and the UCD PLAN'EAT Living Lab (https://clinicaltrials.gov/study/NCT06939231).
Midpoint:
All participants will be contacted by their study nutritionist via email or phone call or message at the midpoint of the study (2 weeks) to check-in on their progression and to address any queries they may have.
Endpoint:
At the end of the 4-week study, participants will complete an online diet-related questionnaire and will be asked to self-complete two 24-hour dietary recalls (via the web-based tool, Foodbook24) on non-consecutive days before attending the final in-person visit at the UCD Institute of Food and Health. However, participants will be allowed to attend the final visit in the case that only one 24-hour dietary recall is completed prior to the visit. The participant will again be instructed to collect their urine first thing in the morning (i.e., first-void) and bring it with them on the day of the final in-person visit. During the final visit, a member of the research team will re-measure participant anthropometrics and blood pressure, while a trained nurse will take a blood sample (if consent is given). Finally, the participant will complete their final study nutritionist-guided 24-hour dietary recall (via the web-based tool, Foodbook24) before leaving.
Statistical analysis plan:
Continuous outcomes will be analysed using ANCOVA (with robust (Huber-White sandwich) standard errors if model residuals are non-normal and/or heteroskedasticity is present) on an intention-to-treat principle, including the baseline variable as a covariate, via the full information maximum likelihood (FIML) method, which treats missing endpoint data as missing at random (MAR). If residual distributions deviate markedly from normality, variables will be transformed prior to use in ANCOVA. Where a group x baseline covariate interaction is detected, treatment effects will be estimated at the mean value, the mean + 1 SD value, and the mean - 1 SD value of the baseline covariate.
Sensitivity analyses for both primary and secondary estimands will be conducted per protocol, with complete cases only (i.e., ≥2 24-hour dietary recalls at endpoint), which treats missing endpoint data as missing completely at random (MCAR). Data will be analysed using ANCOVA (with robust (Huber-White sandwich) standard errors if model residuals are non-normal and/or heteroskedasticity is present) or with robust regression (via M-estimation, Huber method) if residual distributions deviate markedly from normality, including the baseline variable as a covariate. Additional sensitivity analyses for both primary and secondary estimands will be performed excluding suspected dietary under-reporters.
Exploratory analyses for continuous primary and secondary outcome measures will be conducted using linear mixed models (LMMs) across three timepoints (baseline, endpoint (4 weeks), and follow-up (8 weeks)), with fixed effects for group, time, and their interaction, and a random intercept for participants to account for within-subject correlation. The time x group interaction at follow-up (8 weeks) will be the exploratory estimand of interest for the outcome measures, and between-group post-hoc tests will be conducted where a time x group interaction is significant. Models will be estimated using Restricted Maximum Likelihood (REML) under an MAR assumption. Robust (Huber-White sandwich) standard errors will be used if model residuals are non-normal and/or heteroskedasticity is present. If residual distributions deviate markedly from normality, outcomes will be transformed prior to analysis. A per-protocol sensitivity analysis will be performed using only complete cases, which assumes missingness is MCAR. An additional sensitivity analysis will be performed excluding suspected dietary under-reporters. These data will be analysed using the same LMM specification.
Categorical/ordinal outcomes will be analysed using a cumulative link model (CLM) with a logit link, including the baseline variable as a covariate. If model convergence issues arise (e.g., due to sparse outcome categories or separation), a cumulative link mixed model (CLMM) will be fitted instead, including a random intercept for participant ID and a fixed time x group interaction to assess differential change over time.
Inclusion Criteria:
Exclusion Criteria: