Task 5.1. Pilot intervention study (PI: D Tomé (PNCA), M Leclerc (Micalis))
Aims. This pilot study aims at estimating how a personalized food incentive, applied to a well-defined small cohort of human subjects, can reverse their T2D at-risk phenotype. An adapted personalized nutrition combined with an improved adherence should help healthy subjects exhibiting high risk factors for later T2D incidence and dispersed phenotypes to get back to a non-at-risk status. This status encompasses indexes such as body weight, body composition, clinical biomarkers, and microbial profiles. Urine metabolites and gut microbiota have been shown to respond within 24h, then be statistically associated with long-term dietary changes/patterns, especially when switching from a fat-rich to a plant-based diet (Wu et al, 2011). The objective is to evaluate if individual phenotyping and adapted personalized diet can be applied (12 months), if the dispersion of the individual specific risk can be identified and if adapted diet can be associated to these individual risks (36 months).
Methods and data. A number of 50 healthy subjects will be recruited and their dietary intake, physical activity level and energy requirement determined during 2 months. They will then be separated into two groups, and subjected to a second period of 4 months. Subjects from the first group will not change their diet, while for the second group the diet will be modified according to nutritional standards and individually tailored according to individual’s phenotype determined after the first two months. At the end of each period, saliva, fecal, fasting blood, and fasting urine samples will be collected and subjects will be characterized for body weight, body composition, usual clinical biomarkers, and additional specific biomarkers identified in WP1.3. The subjects’ adherence to their diet will be evaluated at the end of each period by interviews and questionnaires. Monthly longitudinal sampling of saliva, fecal, and serum samples will be collected from day 0 to 6 months. Our working hypothesis is that if followed, the personalized dietary guidelines should lead to an overtime change of the microbiota with a reduction of species associated with T2D risks in the feces or the blood, and a microbial profile towards the healthy one.
The microbiota will be analyzed as in WP 1.3, followed by a two-step data analysis i) a fast in silico assessment of the presence then absence of T2D microbial markers detected in WP1.3 ii) an emphasize on individual microbiota trajectory, by building Jensen Shannon Distance maps as performed in ANR-AlimIntest. Each individual microbiota vectors will be assessed for correlation with their tailored diet and metabolic features. The link between microbial markers and metabolic features will lead to networks, which will quantify and visualize the individual dispersion over time.
Expected results. Impact of a four-month personalized diet on the metabolic and microbial signatures, individual trajectory and data dispersion. The results should allow evaluating the influence of diet on the dispersion of phenotype, biological markers, microbial index, and subject compliance to the diet.
Risk. No main risk as it is a pilot study of feasibility.
Task 5.2 Impact of tailored recommendation patterns on individual’s food and physical activity behavior (PIs: O. Allais (Alisss), G. Fagherazzi (CESP))
Aims. Moving from generic to tailored dietary recommendations is supposed to facilitate dietary changes and the adoption of preventive behaviors. The goal of this task is to test this assumption by delivering different types of messages and diet recommendations, either generic or adjusted according to experiment participants. Messages and recommendations will be based on WP1 to 4 results, and designed so that they potentially induce health gains for not too high changes costs. In addition, we will determine to what extent information on personal risk (rather than general information about food-related risks) affects individuals’ decisions to modify food behavior (Godino et al., 2012 and 2014; Nielsen and El-Sohemy, 2014; Grant et al., 2013).
Methods. A five-arm, parallel, RCT over 6 months will be conducted. The number of participants (around 800) will be chosen to guarantee the statistical power of the evaluation and part of them will be participants selected from the E4N cohort. Two information/recommendation patterns will be evaluated: i) weekly summary feedbacks on the nutritional quality of participant’s diet (calorie and main macronutrients intakes) and PA, and ii) weekly dietary recommendations. For each recommendation pattern, we will either reveal or not individual’s phenotypic risk for T2D. All recommendations will be disseminated using WP4 mobile app. After a two-month baseline data collections (gender, socio-demographics, marital status, number of children, BMI, smoking status, diabetes family history, alcohol, food and PA habits, health-related risk and ambiguity attitudes, and QALY evaluation of T2D), participants, including T2D at-risks participants, will be randomized to one of the five groups (i.e. control; feedback; feedback + T2D risk information; feedback + tailored recommendation; and feedback + tailored recommendation + T2D risk information). Our primary evaluation criteria will be calorie and main macronutrient intakes consumed and PA level per week. Difference-in-differences estimations will be implemented to get the causal estimates of each recommendation content.
Data. Individuals' dietary intakes and PA levels will be measured using WP4 mobile app. PA will be automatically recorded using app already available in Smartphone. Participants' lifetime risk of developing T2D will be calculated by epidemiologists involved in the project.
Expected results (36th month). It is assumed that the learning and health consciousness effects will push individuals toward a healthier lifestyle, as the level of the personalization of recommendation and the health risk level increase. By measuring the impact of information on participants’ choices and diets, we will also be able to assess the value of information. This value of information is the major step toward a costs-benefits analysis that also integrates the cost of revealing information. From this costs-benefits framework, we will identify the «paths of least resistance», namely solutions giving the highest benefits and the lowest costs to consumers. Another important aspect of our evaluation in terms of public policies would be to identify, thanks to the baseline data collected prior to the intervention, the characteristics of the “compliers” (who strongly respond to the recommendation scheme) versus the non-compliers in each condition of the RCT.
Risk. The success of our evaluation crucially depends on WP4 apps. The first version will be available at month 12 and will be used to collect dietary information. The app designed within WP4.2 will also provide recommendations on healthier items that minimize the wellbeing costs of behavior change. If this feature of the app is not fully performant after 18th month, a dietitian will accomplish this task.
Collaborations. The whole task is a joint work between INRA-Aliss and INSERM-CESP. CESP will bring its medical competences to provide relevant credible and reliable diet recommendations and T2D risk measures. INRA-Aliss will be in charge of the evaluation process.
Task 5.3. Tailored Recommendations, Diversity of Uses and Social Norms (PIs: F Régnier (INRA Aliss), C. Licoppe (Telecom Paris Tech))
The implementation of nutritional recommendations differs according to social milieu (Régnier, Masullo, 2009), gender (Saint Pol, 2010), and cultural context (INSERM, 2014), and uses of digital devices shows a wide diversity (Pharabod et al., 2013). The effectiveness of tailored messages (health and physical activity) sent by digital tools (Lyzwinsky, 2014) has not been examined in relation to individuals’ socioeconomic characteristics. The impact of exposure to tailored information varies greatly (Swan, 2013). Digital devices are a new method of self-monitoring and self-measurement that could lead to a new kind of reflexivity (Cahour, Licoppe, 2010) and empowerment (Fainzang, 2012), but with risks of practices normalization and dependency (Lupton, 2014).
Aim. Evaluation of how people - in real life and in experimental conditions - perceive and apply tailored recommendations disseminated by digital devices (M24 to M36). We will analyze (i) the diversity in how personalized recommendations are received and applied in a domestic setting, and the existence of possible tensions between individual-based approach and family concerns; (ii) to what extent personalization-based approaches lead to a standardization of private behaviors and conflict with personal or social norms?); (iii) the impacts of personalization on individuals’ autonomy (empowerment) versus its possible negative effects (normopathy, medicalization of diet); (iv) the socioeconomic and cultural conditions for how personalized recommendations are perceived, applied and of their impacts on health behaviors.
Methods and Data: this task is based on two complementary field studies on a socially diverse sample in order to evaluate the effectiveness of personalization-based approaches in different social groups:
- Surveys of digital device diet/health app users in real life that will provide context for the experiment, in order to evaluate uses and meaning of diet and physical activity quantification devices (smartphone applications), to assess their effects on dietary practices/weight/body image and perception of health, and to understand the circumstances of recourse to digital communities. We aim approximately 70 interviews (M24 to M36).
- A post-intervention field survey among the sample of task 5.2 to evaluate the impact of the tailored recommendations disseminated during the experimentation (M32 to 36). We will select individuals differing in consumption changes measured in Task 5.2. We aim 50 interviews.
Expected Results. Effectiveness of personalized recommendations on health-related behaviors; meaning of modifications in food consumption; typology of users by pooling experiment and field data; assessment of diet tailored recommendations’ effectiveness in reducing health inequalities and the social gradient in health.
Risks. Surveys in real life have a low level of risk, and if some delays due to the time for recruitment happened, they would not affect the subsequent data exploitation and dissemination.
Collaborations: We will exchange data during and after intervention with Tasks 5.1 and 5.2, to combine the experimental and the comprehensive aspects of the experiments. Food consumption measurements will be crossed with the field data among the sample of 5.2: the economists will determine the main changes or disrupts in consumption, and sociologists will determine the meaning and conditions, for individuals, of these changes in behaviors.