Chapther 3: Mindulness based eating therapies
3.2 The efficacy of mindfulness based eating programs
6.1.1 Objectives and hypothesis Aims of the present study were:
− To explore the relationship between mindfulness and mindful eating with others constructs such as emotional overeating, binge eating, psychological distress, body disatisfaction and kental well-being, as well as BMI as a continuous variable.
− To investigate the differences between groups of people with various BMI (normal, overweight and obese) on mindfulness, mindful eating, emotional overeating, body dissatisfaction, and mental well-being.
− To test whether mindfulness and mindful eating might represent respectively a mediator and a moderator in the relationship between overeating and negative outcome (psychological distress, body dissatisfaction and poor mental well-being). Mediator variables, in general, are those that account for the relation between the predictor and the criterion, whereas moderator variables are those that affect the direction and/or strength of the relation between a predictor variable and a criterion variable (Baron & Kenny, 1986).
In particular, based on the literature, we expected that:
− mindfulness and mindful eating were positively associated with mental well-being and negatively associated with BMI, overeating behaviours, psychological distress, and body dissatisfaction;
− compared to normal weight people, the obese reported higher frequency of overeating or emotional eating, higher body dissatisfaction, lower level of quality of life and also lower level of mindfulness and mindful eating;
− mindfulness was a mediator in the relationship between overeating and psychological distress, while mindful eating might more likely represent a moderator in the same relatioship.
6. 2. METHODS
6.2.1 Participants
A total sample of 522 adults (males and females) were recruited from both clinical and general populations using mixed methods. Some data were obtained using an on-line psychometric assessment through exponential non-discriminative snowball sampling. A pool of 17 initial informants (psychology students) were asked to nominate, through their social networks, about 17 other participants who met the eligibility criteria and could potentially contribute to the study (N= 289). Even though the researcher has little control over this sampling method, snowball sampling is a useful tool for building networks and increasing the number of participants, in addition the process is cheap, simple and cost-efficient (Goodman, 1961).
Other data were obtained using a paper version of the same questionnaires among mindfulness centers (N= 116) and weight-loss centers (N= 117).
Eligibility criteria were: ≥18 years of age, male or female, and a large variety of body weight. Underweight people were a-posteriori excluded from the study, as their comparisons with the other weight groups was outside the scope of this study. After the exclusion of 20 subjects from the general population who reported a BMI < 18, a final sample of 502 subjects was considered for the present study.
Sample size was established based on epidemiological data on obesity and overweight in Italy (Palmieri et al., 2010). To allow the recruitment of at least 50 obese (50% female), we considered it necessary having at least 200 subjects involved in the study. It is noteworth to specify that not all subjects received the entire tests battery; therefore, depending on the variables selected for each analysis, the sample size varied from 319 to 502.
6.2.2. Procedures
We obtained permission for this study from the Ethical Research Committee of the University of Bologna. Informed consent to participate in the assessment was obtained before entry into the study. Data were collected from January 2013 to September 2014.
6.2.3 Measures
The assessment included socio-demographic variables, self-reported weight and height, and self-report questionnaires assessing eating behaviour, psychological distress and quality of life.
Socio-demographics included gender, age, and school education, factors often found to be associated with disordered eating symptoms (Striegel-Moore & Bulik, 2007). School education was categorized into primary, secondary (high school) and higher (university degree and more). Self-reported height and weight data were used to calculated BMI. Although the use of self- report measures is associated to a general overestimation of height and underestimation of weight resulting in an underestimation of BMI (Krul, Daanen, & Choi, 2011), self-reported measures are valid for identifying relationships in epidemiological studies as they were found to highly correlate with measured height and weight (r >0.90, p <0.001; Spencer et al., 2002). Participants were categorized in three groups according to BMI levels: normal weight (18.5<BMI<25; 52.99%), overweight (25<BMI<30; 23.11%) and obese (BMI>30; 23.90%) (WHO, 1995). BMI was also taken as a continuous variable.
Binge and emotional overeating
Binge Eating Scale (BES; Gordmally et al., 1982) was used to measure binge eating. A full description of the questionnaire has been presented in study 1.
Emotional Overeating Questionnaire (EOQ) (Masheb & Grilo, 2006) was used to assess overeating behaviours associated with emotional states. The EOQ is a 6-item self-report questionnaire developed to measure overeating in response to 6 different emotions: anxiety (worry, jittery, nervous), sadness (blue, down, depressed), loneliness (bored, discouraged,
worthless), tiredness (worn-out, fatigued), anger (upset, frustrated, furious), and happiness (good, joyous, excited). The response set for the six items is a 7-point scale reflecting the frequency of days in which the behavior occurred in the past 28 days (i.e., 0= no days, 1=1–5 days, 2= 6–12 days, 3 =13–15 days, 4= 16–22 days, 5 = 23–27 days, and 6 = every day). The total score is obtained by adding the responses of the six items and dividing by six. Higher scores reflect more frequent overeating. The OEQ has a good internal consistency (α= .85), and its items were significantly and moderately correlated (range .32 to .70) with each other, with one factor accounting for 58% of the variance. The EOQ items and total score were characterized by good test-retest reliability (ICCs, ranged from .62 to .73). In the present study Cronbach’s alpha values was α= .73
Psychological distress, body dissatisfaction and mental well-being
Symptoms Checklist 90 Revised, (SCL-90-R) (Derogatis, 1983; Prunas et al., 2012; Derogatis, 2011) was used to assess psychological distress. A complete description of the questionnaire has been presented in study 1. In this study, instead of the 9 scales used in Study 1 and 2, we calculated a general summative index, the Global Severity Index (GSI), which represents a combination of the number of symptoms reported with the intensity of each symptom. It is a global measure of the severity of the psychological distress suffered by the individual. It varies from 0 to 4, with 4 indicating the higher level of distress.
Body Image Avoidance Questionnaire (BIAQ; Rosen, et al, 1991) was used to measure body dissatisfaction. BIAQ is a 19-item questionnaire that assesses the behavioral facets of body image and the frequency of body image avoidance behaviors. Scoring is obtained by adding the item responses and a higher score indicates more avoidant behaviors. BIAQ has proven reliability by both test-retest (r = .87) and internal consistency (Cronbach’s alpha = .89). In the present study Cronbach’s alpha was .85
World Health Organization Well-Being Index (WHO-5; WHO, 1998; Bonsignore, Barkow, Jensen & Heun, 2001) was used to measure mental well-being. It is a selection of five items from
the World Health Organization’s short health related quality of life measure, the WHOQOL- BREF. Each of the five items is rated on a 6-point Likert scale from 0 (= not present) to 5 (= constantly present). Scores are summated, with raw score ranging from 0 to 25. Then the scores are transformed to 0-100 by multiplying by 4, with higher scores meaning better well-being. Evidence suggests a score of 50 or below being indicative of low mood, and a score of 28 or below indicates likely depression. Psychometric properties have been examined in 23 countries with internal consistency coefficients lying between .75 and .87. In the present study Cronbach’s alpha was .87.
Mindfulness and mindful eating
Freiburg Mindfulness Inventory-14 (FMI; Walach, et al., 2006) was used to measure a general dispositional mindfulness, in the Italian adaptation developed for the purposes of this study. See study 2 for a full description.
Mindful eating was measured by the Mindful Eating Questionnaire (MEQ; Framson et al, 2009) in the Italian adaptation of Clementi et al. (submitted). A full description was presented in study 2.
6.2.4 Statistical analysis
A preliminary bivariate correlational analysis was run to investigate the relationship between mindfulness (FMI) and mindful eating (MEQ) with overeating (BES, EOQ), psychological distress (SCL-90R), body disatisfaction (BIAQ), and quality of life (WHO-5) and BMI as a continuous variable.
MANOVA between BMI groups was run using all psychological measures as dependent variables and gender and age as control variables.
A multiple mediator model of regression was used with the PROCESS macro for SPSS (Hayes, 2013; Hayes & Pritcher, 2014) to test if mindfulness (FMI) and mindful eating (MEQ) might be mediators between overeating and psychological distress or well-being, taking into consideration also BMI, gender and age as control variables. Correlations between the predictors
and the criterion variables, between the predictors and the mediator variables, and between the mediator and criterion variables should be significant. The relation between predictors and criterion should be reduced (to zero in the case of total mediation) after controlling the relation between the mediator and criterion variables.
A significance test of the indirect effect was performed through a computer-intensive method, also called “resampling method”. Specifically, a bootstrapping procedure was performed using 5,000 re-samples and a 95% bias corrected and accelerated confidence interval (BCa-CI) was calculated (Preacher & Hayes, 2008). Repeated many times, the distribution of the indirect effects over multiple bootstrap estimations serves as an empirical approximation of the sampling distribution. These estimates are sorted low to high, and a 95% confidence interval for the indirect effect is constructed. Mediation is considered significant if the 95% CI did not include zero (Preacher & Hayes, 2008). We also tested whether there was complete or partial mediation by testing whether the direct effect coefficient was statistically significant, which is a test of whether the association between the independent and dependent variable is completely accounted for by the mediator (James, Mulaik, & Brett, 2006). If the direct effect coefficient is statistically significant and there is significant mediation, then there is evidence for partial mediation. Standardized regression coefficients were presented as they may serve as effect size measures for individual paths in the mediated effect (McKinnon, 2008).
A moderator analysis was also conducted, with mindful eating as a moderator that was expected to affect the correlation between overeating and psychological distress. In this analysis variables such as gender, age and BMI (taken as continuous) were controlled for.
Analyses were conducted using SPSS (version 19.0) and PROCESS, a computational tool available for SPSS developed by Hayes (2012).
6.3. RESULTS