2.1. Study Participants and Design
Detailed descriptions of the participants and the study design for the parent study have been published previously [30
]. The goal of the parent study was to determine whether four weeks of dietary CHO-restriction would reduce vascular stiffness assessed by aortic pulse wave velocity. The present paper focusses on results obtained from measures of cravings and eating behaviors assessed during this trial. The overall study was approved by the University of Missouri Health Science Institutional Review Board (Protocol #2004733) and written informed consent was obtained from all participants. This trial was registered on ClinicalTrials.gov Identifier: NCT00427193. Men and premenopausal women were recruited who were overweight or obese (BMI: 27–40 kg/m2
), aged 30–55 years, and had characteristics of insulin resistance [31
] or the metabolic syndrome [30
]. With regard to physical activity, subjects had low to medium levels (5000 to 10,000 steps/day). Exclusion criteria included having diabetes, alcohol use of more than 5 standard drinks/week (70 g) for women and more than 10 standard drinks/week (140 g) for men, use of tobacco products, taking prescription medication for clinically significant endocrine, gastrointestinal, cardiovascular, hematological, hepatic, renal, respiratory, or genitourinary abnormalities or diseases, or being on any dietary regimen that would hinder adherence to the low-CHO diet.
As shown in Figure 1
, 64 participants were assessed for eligibility and, as per the original study design’s plan, 20 were assigned the CHO-restricted intervention [30
]. To compare cravings with eating behaviors, WL, and biochemical outcomes (insulin and glucose concentrations), complete data were available for 19 of the original 20 subjects for food cravings and 18 out of the original 20 subjects for eating behaviors, TFEQ. The low-CHO diet provided 1500 kcal/day of energy and included 20–25 g of net CHO/day [30
]. Subjects were provided all food for the first two weeks of the study, during which time they received comprehensive education to be able to cook and prepare their own low-CHO meals during weeks 3 and 4. Thus, for weeks 3 and 4, subjects consumed a diet similar to that eaten during the first two weeks by following the guidelines and using the educational materials given them by study investigators. Study staff were in contact with research participants on a near-daily basis to monitor dietary intake and assess food acceptance. Dietary adherence was assessed by inspection of returned food containers (first two weeks), respiratory quotient, and measurement of plasma ketones during week 4 [30
]. Body weight was measured to the nearest 0.1 kg and fasting glucose and insulin concentrations were measured by enzymatic assay and ELISA, respectively [30
2.2. Food Cravings
The Food Craving Inventory (FCI), originally developed by White et al. [32
], has been previously validated and demonstrates internal consistency (Cronbach’s α = 0.76–0.93). The FCI is a 33-item questionnaire to assess cravings across five subscales: (1) high-fat foods (e.g., fried chicken, gravy, sausages, hot dogs), (2) sweets (e.g., cake, cinnamon rolls, ice cream, cookies), (3) CHO/starches (e.g., sandwich bread, rice, biscuits, pasta, pancakes), (4) fast-food fats (pizza, french fries, hamburger, chips), and (5) fruit/vegetable subscale (raw fruits, raw vegetables, canned fruit, fruit juice, and cooked vegetables). The fruit/vegetable subscale was not included in the original FCI by White and colleagues but has been previously used by others [9
]. Items on the questionnaire are scored on a 1–5 scale with 1 = never and 5 = always/almost every day. A total score for each subscale is calculated by taking the average of scores for individual foods within that subscale. A total craving score is calculated from the aggregate or mean of all five subscales. For the present analysis, internal consistency (Cronbach’s α) was found for baseline and post low-CHO diet, respectively, for the high-fat food subscale (0.85 and 0.83), sweets (0.87 and 0.66), CHO/starches (0.88 and 0.80), fast-food fats (0.80 and 0.81), fruits/vegetables (0.80 and 0.76), and overall cravings (0.94 and 0.92). These values represent acceptable to good internal consistency of the scales used to measure food cravings.
2.4. Statistical Analyses
All statistical analyses were performed by using the Statistical Package for Social Sciences (IBM SPSS Statistics, version 25, SPSS Inc., Armonk, NY, USA). Participant characteristics are expressed as mean ± SD, while mean cravings, and eating behavior scores (dietary restraint, dietary disinhibition, and hunger) are expressed as mean ± SE. Statistical significance was accepted at p < 0.05. For the food cravings data, analyses focused on each participant’s aggregated cravings ratings, which were computed by averaging an outcome rating for foods in the same category (high fat, sweet, CHO/starches, fast-food fats, fruits/vegetables) for baseline, and week 4 (post low-CHO diet). For each of the 33 items measuring food cravings on the FCI, a total craving score was calculated as the average of the overall score for high fat, sweet, CHO/starches, fast-food fats, and fruits/vegetables. Thus, six craving variables (high fat, sweets, CHO/starches, fast-food fats, fruits/vegetables, and total cravings) were included as outcomes. Similarly, for the eating behaviors, an aggregated score for each of three constructs (dietary restraint, disinhibition, and hunger) was calculated by averaging the scores obtained by each subject for each of the three different constructs. A paired-sample t-test was used to assess differences in mean cravings scores, dietary restraint, dietary disinhibition, and hunger between baseline and post low-CHO diet (week 4).
A repeated-measures ANOVA with sex as the between-subject factor and time (baseline, post low-CHO diet) as the within-subject factor was carried out to test the main effect of sex on cravings and eating behaviors. Pearson correlations were used to assess the relationships between change in cravings and change in body weight, change in eating behavior, and change in insulin and glucose concentrations. Change variables were calculated by subtracting the value at baseline from the value at week 4. Thus, a positive value indicates increased cravings, restraint, disinhibition, hunger, weight gain, glucose and insulin concentrations, while a negative value implies decreased cravings, restraint, disinhibition, hunger, weight loss, glucose and insulin concentrations. Percent changes were calculated by dividing the difference between baseline and week 4 values by the baseline value and then multiplying the answer by 100 for each subject. The average was then taken as the percent change for the variable in question. This approach allowed us to capture variability in responses within individual subjects.