Problematic relationship to eating and food and its association to body mass index, incident diabetes, metabolic syndrome, and diet in middle-aged adults: The Coronary Artery Risk Development in Young Adults (CARDIA) Study

Abstract

Background: There is a high prevalence among adults of eating behaviors and attitudes that do not meet the threshold or fulfill the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-V) criteria for eating disorders. Examples include overeating with or without the sense of loss of control or distress associated with overeating or sense of loss of control. These subclinical eating behaviors and attitudes may be called a ‘Problematic Relationship to Eating and Food’ (PREF) and may occur on a continuum. That is, unhealthy eating behaviors may vary by severity, ranging from normal eating, through unhealthy behaviors and attitudes that do not require psychiatric or medical treatment, to the most severe clinical eating disorders. However, while problematic eating behaviors have been studied in young people, little is known of the prevalence and long-term effects of problematic eating behaviors in generally healthy middle-aged adults enrolled in a prospective cohort. Thus this thesis adds to the field of eating disorders and eating behaviors by exploring the prevalence of a range of behaviors and attitudes related to eating and food among a large sample of generally healthy middle-aged adults who have been observed in 9 clinical examinations over 30 years in their adult life. Using this long-term follow-up, the objective of this thesis is to investigate the association of a PREF scale with BMI trajectories, risk of incident diabetes and metabolic syndrome, as well as energy intake, dietary patterns and diet quality in the Coronary Artery Risk Development in Young Adults (CARDIA) Study. Methods: This dissertation includes three separate research projects which aimed to describe the prevalence of problematic behaviors and attitudes related to eating and food among generally healthy middle-aged adults. A Problematic Relationship to Eating and Food (PREF) score was defined as the sum of number of behavioral and attitudinal constructs endorsed by participants. The PREF score was assessed based on the Questionnaire on Eating and Weight Patterns-Revised (QEWP-R, originally created to gather data in support of the concept of Binge Eating Disorder), administered in the Coronary Artery Risk Development in Young Adults (CARDIA) study in its Year 10 examination when the community based sample of nearly 4000 participants was aged 27-41 years. PREF captures 8 behaviors and attitudes related to eating and food including anxiety around eating or food, endorsement of unhealthy compensatory behaviors intended to lose weight or to maintain weight loss, consumption of a large amount of food in a relatively short period of time, sense of loss of control when overeating in a relatively short period of time, distress associated with overeating, distress associated with sense of loss of control, preoccupation with dieting, and concern about personal weight and shape. The study then investigates how PREF is prospectively associated, among generally healthy middle-aged adults, with 1) body mass index (BMI), 2) incident diabetes and metabolic syndrome, and 3) energy intake, dietary intake, pattern, and quality. The first project examined how PREF was associated with the trajectories of BMI throughout 15 years of follow-up (through CARDIA Year 25), using repeated measures regression. In addition, PREF associations with BMI classes were examined using multinomial logistic regression. The second project examined how the problematic eating behaviors were associated with incident diabetes and metabolic syndrome using proportional hazards regression, logistic regression and Kaplan-Meier curves. The last project, concerned the examination of associations between PREF and energy intake, dietary pattern, and diet quality (including an objective measure of diet quality, the sum of 4 serum carotenoids), used multiple linear regression. Results: Results for the three research projects were as follows. We identified six categories of problematic relationships to eating and food by giving one point for each PREF construct endorsed. PREF 0 points (called normal eaters in this thesis) and participants with PREF (1, 2, 3, 4-5 points, and 6-8 points) were identified. Almost 60% of the participants endorsed at least one construct (therefore being assigned at least 1 PREF point). Moreover, even after the exclusion of 3.7% of the participants with PREF 6-8 points (i.e., who endorsed 6 to 8 constructs, which presumptively closely corresponds to a diagnosable eating disorders), 54.4% endorsed 1 to 5 constructs (PREF 1-5 points). Therefore, problematic eating attitudes and behaviors were prevalent in middle-aged adults. Greater BMI trajectories were observed in PREF ≥1 point compared to PREF 0 points. Among those with PREF 1-5 points, mean BMI was graded by the number of PREF constructs endorsed. The BMI trajectories further diverged throughout 25 years of follow-up, with approximately 0.6kg/m2 greater BMI per PREF point in CARDIA Year 0, increasing to 1.0 kg/m2 per PREF point in CARDIA Years 10 and 25. Participants had 1.2 times the hazard of diabetes per PREF point after 15 years of follow-up, adjusting for demographic and behaviors. Participants had 1.2 times hazard of metabolic syndrome per PREF point after 10 years of follow-up, adjusting for demographic characteristics and health behaviors. However, both associations were close to null after further adjustment for Year 10 body mass index, suggesting associations between problematic eating behaviors and attitudes were mediated through BMI. A less favorable diet in those with PREF ≥ 6 points was suggested by greater energy intake and greater consumption of artificially-sweetened beverages. However, a healthier A Priori Diet Quality Score among those with PREF ≥ 1 point (p=0.04 in a graded relationship across PREF points,1.9 A Priori Diet Quality Score points higher in PREF 6-8 points than in PREF 0 points) suggested knowledge of and attention to diet composition in those with any PREF points, despite having substantial excess body weight. This finding of slightly better quality diet was supported by greater values of the objective measure, a sum of 4 serum carotenoids, which, after adjustment for BMI, were found in participants with PREF ≥1 point versus those with PREF 0 points. While some of the above associations were not significant among those with PREF 1-5 points when each such category was compared specifically to PREF 0 points, the statistically significant linear trend suggests that the associations of dietary patterns and intakes among PREF 1-5 points follows the general trend and thus are similar to those with presumptive eating disorders. Conclusions: These results indicate that people in all PREF categories above 0 had greater BMI trajectories, were at greater risk of diabetes and metabolic syndrome, had greater energy intake, and consumed more artificially-sweetened beverages, all of which are of potential concern for public health. In contrast, they had greater A Priori Diet Quality Scores (healthier diet). A novel aspect of this study was our ability to examine the serum concentration of the sum of 4 carotenoids, which are not synthesized by humans and are only available through diet from plant foods (especially fruits and vegetables); and which empirically are lower in the presence of oxidative stress, for example when BMI is high. After adjustment for demographics, behavioral variables, and particularly for CARDIA Year 10 BMI, the sum of carotenoids was positively associated with PREF points. These associations with PREF were not seen to differ by race and sex. These findings illustrate the need for early identification and treatment of people with problematic relationships to eating and food, and the need to view unhealthy eating behaviors as a continuum and to investigate the full spectrum of eating behaviors. Obesity in those with PREF is consistent with their higher energy intake and this higher intake likely contributes to their excess BMI. However, the fact that those with PREF did not have worse diet quality suggests that they show interest in eating a healthy diet (by the fact that their diet score is higher, confirmed by their sum of 4 carotenoids). Despite their apparent interest in eating a good composition diet, this disconnect with their ongoing obesity may suggest a window for teaching better ways to relate to food both at the individual and policy levels. Further research is necessary to estimate the magnitude and effect of problematic relationship to eating and food in the US population, and whether interventions might help to reduce this problem.