Objective quantification and analysis of eating behaviour associated with obesity development : from lab to real-life
Sammanfattning: Introduction: The last four decades have seen a marked increase in childhood and adult obesity prevalence, attributed to an “obesogenic” environment. Several genetical, environmental and behavioural factors have been identified that increase the risk of obesity, but treatment outcomes are usually modest and the risk of relapse high. One limitation responsible for these moderate results could be methodological, with researchers questioning both the external validity of eating behaviour measures in the laboratory (controlled) and the internal validity of eating behaviour measures in free-living (real-life) settings. Technological advances could solve some of these issues, allowing for accurate methods, similar to those used in controlled settings, to be used in real- life. Deploying accurate methods in both controlled and real-life settings would in turn enable the estimation of external validity, determining the limits of generalization between settings. In turn enabling the deployment of these methods in settings which allow large scale screening, for early identification of individuals at risk of becoming obese. Aim: The overarching aim of the thesis was to: i) evaluate the stability of human eating behaviour and ii) investigate the usability and feasibility of methods developed for controlled settings, when deployed in semi-controlled and real-life settings. Paper I – Determine if individuals maintain their eating behaviour, in relation to the group, despite experimental manipulations to meal conditions (i.e., unit sizes and serving occasion); Paper II – Feasibility of employing novel technology for baseline eating behaviour collection in adolescents eating school lunches in a school cafeteria setting (semi-controlled); Paper III – Feasibility of employing novel technology in an experimental manipulation study, to determine the effect of proximity in a semi-controlled school setting; and Paper IV – By use of novel technology, examine the maintenance of eating behaviours in adolescents, from semi-controlled to real-life settings, both at group- and individual-level. Methods: Paper I – Three randomised crossover studies, of which two compared eating behaviour across different unit sizes, while one compared eating behaviour between lunch and dinner in healthy young adults. Performed in a controlled setting, employing traditional laboratory methods; Paper II – An observational study of healthy adolescents, performed at lunch in a school cafeteria, employing traditional laboratory methods in a semi-controlled setting; Paper III – A randomised experimental study of healthy adolescents, performed in a semi- controlled, comparing the eating behaviour between two groups seated at different proximity to food items; and Paper IV – An observational study on eating behaviour of healthy adolescents, divided into two parts; i) collection of eating behaviour data, performed at lunch in a school cafeteria, using a similar protocol to that of Paper II; and ii) collection of eating behaviour data by the participants in real-life settings, using the same devices as in the controlled setting. Results: In all papers the distribution of eating behaviour values between individuals were large. In Paper I, the largest increase in unit size significantly increased meal duration and chews and while there was a trend for both increased meal duration and number of chews the larger the food unit sizes were, it did not lead to a significant reduction in food intake. Meanwhile, the correlation coefficient of all eating behaviours across all conditions was high (except for number of bites between the largest and smallest food unit size condition). In Paper II, male participants ate significantly more, mediated by significantly larger bites. The bite sizes of both men and women were reduced as the meal progressed. In Paper III, increased distance to food led to a significant reduction in intake, caused by individuals taking less chocolate. In Paper IV, there was no significant difference in eating behaviour characteristics between the semi- controlled and real-life meals. In addition, the correlation coefficient of food intake and eating rate was high between settings, while the correlation of meal duration was low. Also, on an individual level, 50%, 32% and 27% of the food intake, eating rate and meal duration measures, respectively, from the semi-controlled meal fell within the confidence interval of the real-life meals. In the semi-controlled and real-life settings (Papers II-IV), the agreement between subjective and objective eating behaviour measures were very low. Meanwhile, in both semi- controlled and real-life settings the method could be deployed within the time schedule imposed by the school, with high data retention. Also, participants rated the comfortability participating in the semi-controlled and real-life settings very high and the usability of the system as “Good” or higher. Conclusions: Human eating behaviour appears stable in comparison to the group when unit size and serving occasion is manipulated in a controlled setting and when eating in different settings (semi- controlled and real-life). Suggesting generalisations can be made between settings and conditions and that risk behaviours may be measured in settings other than real-life, at least on group level. However, although individual prediction rates of eating behaviour characteristics from semi-controlled setting to real-life settings appears higher than subjective ratings, they are still too low for use in the design of tailored interventions. In addition, compared to controlled studies, the method allowed recruitment of a younger age group, since it enabled measurements in a different location. The thesis also provides evidence that the employed methods are usable, feasible and acceptable, with high data retention in adolescent users, in semi-controlled and real-life settings. Methods similar to the ones used in this thesis can provide previously unattainable information (primarily temporal) in settings that are less controlled than the laboratory, such as semi-controlled and real-life.
Denna avhandling är EVENTUELLT nedladdningsbar som PDF. Kolla denna länk för att se om den går att ladda ner.