Title: Exploratory randomised controlled trial of a
mindfulness based weight loss intervention for women
Authors: Katy Tapper, Christine Shaw, Joanne Ilsley, Andrew
J. Hill, Frank W. Bond, Laurence Moore
To appear in:
Please cite this article as: Tapper, K., Shaw, C., Ilsley, J., Hill, A. J., Bond, F. W., &
Moore, L., Exploratory randomised controlled trial of a mindfulness based weight loss
intervention for women, Appetite (2008), doi:10.1016/j.appet.2008.11.012
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
Page 1 of 37
Mindfulness based weight loss 1
Running head: MINDFULNESS BASED WEIGHT LOSS
Exploratory randomised controlled trial of a mindfulness based weight loss intervention for
Katy Tapper1, Christine Shaw2, Joanne Ilsley3, Andrew J. Hill4, Frank W. Bond5& Laurence
1Department of Psychology, Swansea University, United Kingdom
2Department of Care Sciences, University of Glamorgan, United Kingdom
3Cardiff Institute of Society, Health and Ethics, Cardiff University, United Kingdom
4Academic Unit of Psychiatry and Behavioural Sciences, Leeds University, United Kingdom
5Department of Psychology, Goldsmiths College, University of London, United Kingdom
Dr Katy Tapper
Department of Psychology
University of Wales, Swansea
Tel. +44 (0)2920 569103
Page 2 of 37
Mindfulness based weight loss 2
Objective: To explore the efficacy of a mindfulness based weight loss intervention for women.
Design: Sixty-two women (ages 19-64; BMI 22.5-52.1) who were attempting to lose weight
were randomised to an intervention or control condition. The former were invited to attend four
2-hour workshops, the latter were asked to continue with their normal diets. Data were collected
at baseline, 4 and 6 months.
Main outcome measures: BMI, physical activity, mental health.
Results: At 6 months intervention participants showed significantly greater increases in physical
activity compared to controls (p<.05) but no significant differences in weight loss or mental
health. However, when intervention participants who reported ‘never’ applying the workshop
principles at 6 months (n=7) were excluded, results showed both significantly greater increases
in physical activity (3.1 sessions per week relative to controls, p<.05) and significantly greater
reductions in BMI (0.96 relative to controls, equivalent to 2.32kg, p<0.5). Reductions in BMI
were mediated primarily by reductions in binge eating.
Conclusion: Despite its brevity, the intervention was successful at bringing about change.
Further refinements should increase its efficacy.
Keywords: obesity, weight loss, mindfulness, physical activity, binge eating, ACT
Page 3 of 37
Health Committee, 2004).
Mindfulness based weight loss 3
Exploratory randomised controlled trial of a mindfulness based weight loss intervention for
Over the last two decades levels of obesity among British and American adults have trebled
(World Health Organisation, 2003). Since obesity is associated with a wide range of health
problems (Must et al., 1999), it impacts not only on quality of life, but also represents a
substantial economic burden. Unfortunately, weight loss is difficult to achieve and even harder to
maintain. For example, it is estimated that less than 5% of those who lose weight will have
maintained these losses after 4 to 5 years (Kramer, Jeffery, Forster & Snell 1989). Research
suggests that this is a result of the individual failing to maintain healthy eating and exercise
habits (Jeffery et al., 2000; McGuire, Wing, Klem, Lang & Hill, 1999). Thus knowing how to
lose weight is simply not sufficient, we also need to tackle the psychological processes that lead
to behaviours associated with weight gain. Indeed, experts are increasingly recognising the need
to address the psychological aspects of obesity (e.g., Byrne, 2002; Cooper & Fairburn, 2001;
House of Commons Health Committee, 2004). Although a growing number of interventions are
now incorporating components aimed at this (e.g., Cooper & Fairburn, 2001; Rapoport, Clark &
Wardle, 2000), the development of these still falls far short of that achieved in areas such as
nicotine and alcohol dependence to which obesity has been compared (House of Commons
Interventions to effect behaviour change in obesity management often draw on cognitive
behaviour therapy (CBT). Such interventions generally incorporate both behavioural elements
Page 4 of 37
avoid or control them. In doing so, the individual is able to abandon maladaptive behaviours
Mindfulness based weight loss 4
such as cue avoidance, and cognitive elements such as challenging dysfunctional thoughts (e.g.,
Nauta, Hosers & Jansen, 2001; Rapoport, Clark & Wardle, 2000). Most recently motivational
interviewing techniques have been included to increase the efficacy of traditional weight loss
programmes (e.g., Carels et al., 2007).
The current study explored a different approach. Recent advances in psychotherapy
suggest that mindfulness-based techniques and therapies may be an effective alternative to CBT
for a wide range of clinical and non-clinical problems (Bishop et al., 2004; Hayes et al., 2006;
Hayes, Masuda et al., 2004; Teasdale et al., 2000). Mindfulness can be described as a process
whereby the individual observes their immediate experience using an open and non-judgemental
stance (Bishop et al., 2004). Acceptance and Commitment Therapy (ACT) is a mindfulness-
based therapy that has also been successfully used to treat addictive behaviours such as drug
abuse and smoking. Of particular note is that in these cases ACT appeared to be as or more
effective than traditional approaches when it came to the maintenance of behavioural changes
(Gifford et al., 2004; Hayes, Wilson et al., 2004). Given the high relapse rates amongst dieters,
an ACT based approach to weight loss may have potential.
What does ACT involve? ACT employs mindfulness strategies to target experiential
avoidance. Experiential avoidance refers to attempts to avoid or control certain private events
such as negative emotions, thoughts or bodily sensations (Hayes, Strosahl & Wilson, 1999).
ACT interventions draw on a variety of mindfulness-based techniques and exercises to bring
about a willingness to experience difficult thoughts, feelings and sensations rather than trying to
normally used for avoidance and control and instead focus on behaviours that move them
towards valued outcomes (Hayes et al., 1999).
Page 5 of 37
Mindfulness based weight loss 5
How might ACT apply to obesity? There is evidence to suggest that obesity is associated
with both emotional eating and external eating. Emotional eating refers to a tendency to overeat
in response to negative emotions such as boredom, stress and unhappiness, whilst external eating
refers to a tendency to overeat in response to food-related stimuli such as the taste, sight or smell
of a palatable food (Van Strien, Schippers & Cox, 1995). Research shows that questionnaire
measures of these types of eating behaviours are positively associated with BMI and obesity
(Blair, Lewis & Booth, 1990; Braet & Van Strien, 1997; Delehanty et al., 2002; Hays et al.,
2002; Wardle, 1987)1. Such measures have also been shown to be associated with retrospective
accounts of adult weight gain (Hays et al., 2002, see also Kayman, Bruvold & Stern, 1990) and
to predict weight regain following weight loss (McGuire et al., 1999). In addition, a study by
Blair et al., (1990) found significant associations between levels of emotional eating and weight
loss success; successful weight control was associated with decreases in emotional eating
between baseline and a 1 year follow-up and with low levels of emotional eating at both time
points. In contrast, unsuccessful weight control was associated with increases between baseline
and follow-up and with high levels at both time points.
It is likely that experiential avoidance is involved in both emotional and external eating
behaviours. Emotional eating occurs in response to negative emotions and there is evidence to
suggest that it may be an attempt to distract attention from, or alleviate, these feelings (Tice &
Bratslavsky, 2000, see also House of Commons Health Committee, 2004). If this is the case,
emotional eating can be viewed as a form of experiential avoidance. In contrast, external eating
1Some of these measure disinhibition of control rather than emotional and external eating per se. Disinhibition is
measured using the Three-Factor Eating Questionnaire (TFEQ) and refers to a tendency to overeat in response to
disinhibiting stimuli such as negative emotions or the presence of palatable foods. The scale encompasses both
emotional eating items and externality items (see Stunkard & Messick, 1985 and Van Strien, Frijters, Bergers &
Defares, 1986) and there are highly significant correlations between measures of disinhibition and measures of
emotional and external eating (Hill, Weaver & Blundell, 1991).
Page 6 of 37
trial of the effectiveness of a brief ACT-based group intervention. Given possible sex differences
Mindfulness based weight loss 6
occurs in response to food cues and is therefore not necessarily prompted by an attempt to avoid
or control negative feelings. However, where an individual is trying to lose weight, or eat
healthily, and is attempting to resist overeating in response to these cues, it is likely that he or she
will experience difficult thoughts, feelings and/or bodily sensations. For example, attempting to
resist desert at a restaurant my elicit uncomfortable cravings. Failure to resist desert may
therefore be viewed as an attempt to avoid or control these cravings and thus also a form of
experiential avoidance. Since ACT directly targets experiential avoidance it may therefore be
effective in bringing about reductions in emotional and external eating behaviours.
There is also evidence that bouts of overeating can be triggered by particular thoughts,
for example about having broken ones diet (Ogden & Wardle, 1991). Likewise, failure to adhere
to exercise and healthy eating plans may be prompted by rationalisations about, for example,
being more conscientious the next day or there being exceptional circumstances that justify the
relapse. An important component of ACT is cognitive defusion, helping the individual to see
thoughts simply as thoughts, rather than as things that should necessarily be believed and
followed. This technique helps individuals relate differently to their thoughts enabling them to
choose to act in accordance with their personal values and life goals. Thus applied to the above
cognitions it may help individuals refrain from bouts of overeating and adhere to exercise and
Following recommendations for the development and evaluation of complex health
interventions (Campbell et al., 2000), the aim of the current study was to conduct an exploratory
in psychological determinants of weight gain and loss, with females potentially engaging in more
Page 7 of 37
Mindfulness based weight loss 7
emotional eating than males (Tanofsky, Wilfey, Spurrell, Welch & Brownell, 1997; Wardle et
al., 1992; Wardle, 1987), the trial was restricted to females only.
Sample size and recruitment
The target sample size was 60. There were no studies directly comparable to the present research
but given attrition rates and effect sizes obtained in previous analogous investigations it was
estimated this would provide an acceptable level of statistical power.
Participants were recruited by the third author (JI) via advertisements and articles in local
newspapers, community and leisure centres, and on the university website. Eligibility criteria
were a BMI of over 20 (this cut-off was selected to avoid excluding those who had been dieting
for some time), over 18 years of age, actively attempting to lose weight, not pregnant, not using
medications that influence weight, able to attend at least three of the four intervention workshops
and no more than one participant per household. Figure 1 shows the flow of participants through
the study. Eligible participants who attended a pre-trial appointment at the university and
returned baseline questionnaires were entered into the randomisation process (62 in total). As a
token of appreciation all participants were sent cheques for £25 (approximately 50 US dollars)
on receipt of both the second and the final sets of questionnaires (see below). The study received
ethical approval from the Cardiff School of Social Science Ethics Committee and informed
consent was obtained.
INSERT FIGURE 1 HERE
Page 8 of 37
Participant details (i.e. participant number, BMI and any medical conditions) were
Mindfulness based weight loss 8
Study design and randomisation
The study employed a randomised controlled trial design with quantitative data (questionnaires
and BMI) collected at baseline, 4 months and 6 months. It also incorporated a qualitative
evaluation, details of which are reported elsewhere (Tapper, Shaw, Ilsley & Moore, 2007).
Participants were allocated to the intervention and control conditions using a stratified
randomisation protocol on the basis of BMI and the existence of medical conditions likely to
effect weight (since these variables were considered most likely to influence study outcomes).
The randomisation list was prepared by the first author (KT) using information provided by
participants who returned registration forms (n=91). These participants were first divided into
‘medical’ and ‘non-medical’ groups according to whether, on the basis of British Heart
Foundation guidelines (British Heart Foundation, 2004), they had an existing medical condition
likely to affect their weight. Of the 91 participants, 10 were classified as having a medical
condition and 81 as having no medical condition. Within the medical group, two strata were then
formed using the median (self-reported) BMI and four strata were formed for the non-medical
group by splitting participants into quartiles, again using self reported BMI. This resulted in a
total of six strata, two for those with medical conditions (above and below a BMI of 28.81) and
four for those without medical conditions (BMI ranges of ≤27.47, 27.48 to 30.24, 30.25 to 34.50
and ≥34.51). A block size of two was used for the medical group and a block size of four for the
non-medical group. Following Pocock (1983), computer-generated random numbers were used
to order intervention and control group allocation within each block.
provided by JI in the order in which she received completed questionnaires. JI was blind to the
number of strata and block sizes employed within the randomisation list. KT was not involved in
Page 9 of 37
dietary advice was provided. This was explained to participants at the start of the workshop
Mindfulness based weight loss 9
recruitment and was blind to participant identities. All participants were informed of the
frequency of weight assessments (i.e. one at baseline and two follow-ups). Participants in the
intervention condition were invited to attend the intervention workshops whilst controls were
simply asked to continue their weight loss attempt as normal. No further information was
provided to control participants but they were given the opportunity to attend a one-day weight
loss workshop at the end of the study.
The sample had a mean BMI of 31.57 (SD=6.06, range=22.53-52.12) and a mean age of
41 years (SD=13, range=19-64). Table 1 shows the baseline characteristics across intervention
and control groups. As shown, there were no significant differences in BMI, age (though
intervention participants were, on average, slightly older than control participants), level of
education, % of participants with medical conditions affecting their weight, % of participants
attending formal slimming clubs, number of previous diet attempts, or length spent on current
diet. However, participants in the intervention group reported starting dieting at a significantly
older age than those in the control group (25 versus 20 years respectively, p<.05).
INSERT TABLE 1 HERE
The intervention was designed to be used alongside participants’ own weight loss plans. For
these reasons only participants who were already attempting to lose weight were recruited and no
sessions and reiterated when questions relating to specific diet strategies arose.
Page 10 of 37
session. A manual was provided to accompany the workshops. This included details of key
Mindfulness based weight loss 10
The intervention drew on selected concepts, exercises and metaphors previously
employed in ACT interventions (Hayes & Smith, 2005; Hayes et al., 1999) and adapted these to
the context of weight loss. Key intervention components were a) values, to enhance motivation,
b) cognitive defusion, to help break links between food- and exercise-related thoughts and
behaviour, and c) acceptance, to help the individual tolerate negative feelings. A summary of
components employed, and their application to weight loss, is displayed in Table 2.
INSERT TABLE 2 HERE
The intervention was delivered by JI via a series of three workshops conducted over three
consecutive weeks with a fourth follow up session taking place approximately three months later.
The length of the treatment was based on that employed in previous brief ACT interventions that
have produced successful outcomes (Bach & Hayes, 2002; Bond & Bunce, 2000; Metzler,
Biglan, Noell, Ary & Ochs, 2000). Each session lasted two hours and included a powerpoint
presentation and explanation of key concepts using metaphors, exercises and pen and paper
tasks. Questions were encouraged during the session to ensure concepts were understood. JI, KT
and CS had all attended a range of ACT training workshops and, as an informal assessment of
treatment integrity, eight of the 12 workshops were observed by either KT or CS. During these
sessions no problems relating to treatment integrity were identified.
Participants were also asked to complete a series of homework exercises in between each
concepts and exercises, forms for pen and paper based tasks and details of homework.
Participants also received a CD containing the four ‘eyes-closed’ exercises: ‘Leaves on a
Page 11 of 37
Electronic Scales. Participants’ clothing was also recorded at baseline and participants were
Mindfulness based weight loss 11
Stream’, ‘Giving Feelings a Form’, ‘The Tin Can Monster Exercise’ and ‘Being Where You Are’
(Hayes & Smith, 2005; Hayes et al., 1999). This was designed to support participants’ practice at
It was acknowledged that in a real-world setting full workshop attendance by all
participants was unlikely. A number of features were therefore incorporated into the design of
the intervention to both maximise attendance and minimise the effects of non-attendance. First,
three workshops following the same protocol were conducted each week with participants able to
choose to attend either a daytime or evening session and vary the one they attended from week to
week if other commitments intervened. (Sessions ran on Tuesdays at 6pm, Wednesdays at 7pm
and Thursdays at 1pm.) Second, details of topics covered in each session were contained in the
manual and those missing sessions were encouraged to read through these and complete the
homework as usual. Third, each workshop session began with a recap of material previously
covered, and lastly, the fourth follow-up workshop consisted of a more extensive recap of key
Main outcome measures were as follows:
BMI (kg/m2). Height was measured using the Leicester Height Measure (Invicta Plastics
Ltd., Leicester) and recorded to the last completed millimetre. Weight was measured without
footware to the nearest 1/10thof a kilogram using Weight Watchers Heavy Duty Precision
asked to wear similar clothing at later assessments. BMI was computed by dividing weight by
Page 12 of 37
indicating higher levels of the behaviour. The DEBQ has been shown to have satisfactory to
Mindfulness based weight loss 12
Brief Physical Assessment Tool (BPAT) to assess physical activity. The BPAT (Smith,
Marshal & Huang, 2005) consists of three items recording a) the number of 30 minute bouts of
moderate intensity levels of activity within a week, b) the number of 30 minute bouts of walking
within a week, and c) the number of 20 minute bouts of vigorous levels of activity within a week.
The questionnaire is scored as the total number of activity sessions per week with 20 minute
bouts of vigorous activity counting as two sessions. It has been shown to have moderate test-
retest reliability; fair to moderate concurrent validity and poor to fair criterion validity (Smith et
General Health Questionnaire-12 (GHQ-12) to assess mental health. The GHQ-12
(Goldberg, 1978) is a 12 item screening measure of current mental health status. Participants are
asked to compare their recent experience of a particular symptom or behaviour with their usual
level of functioning on a four point scale (0-3) ranging from ‘less than usual’ to ‘much more than
usual’. The questionnaire is scored from 0-36 with lower scores indicating greater mental health.
Hypothesised mediator measures were as follows:
Dutch Eating Behaviour Questionnaire (DEBQ) to assess emotional and external eating.
The DEBQ (Van Strien, Frijters, Bergers & Defares, 1986) assesses emotional, external and
restrained eating. It contains 33 statements each rated by participants as never / rarely /
sometimes / often / very often. Only the emotional and external eating subscales were included in
the present analysis (13 and 10 items respectively). Scores range from 1-5 with higher scores
good reliability, excellent factorial validity and satisfactory concurrent and discriminant validity
(Van Strien et al., 1986; Wardle, 1987).
Page 13 of 37
Dietary adherence. Participants were asked to rate the extent they had adhered to their
Mindfulness based weight loss 13
Emotional Eating Questionnaire (EEQ) to assess emotional eating. This was a modified
version of the Emotional Overeating Questionnaire (Masheb & Grilo, 2006) and assessed recent
episodes of emotional eating. Participants recorded the number of days out of the last 7 (0-7)
that they had eaten in response to a range of feelings (anxiety, sadness, loneliness, tiredness,
anger, happiness). Internal reliability was moderate at baseline (alpha=0.47, n=62) but acceptable
(alpha=0.62) with the exclusion of item 6 (happiness). Thus item 6 was excluded from all
Binge Eating Scale (BES) subsection to assess binge eating. A shortened version (6
items) of the BES (Gormally, Black, Datson & Rardin, 1982) was used to detect the presence of
any of the cardinal features of binge eating, i.e. rapid consumption of large quantities of food and
feelings and cognitions such as loss of control and fear of being unable to stop eating. The
questionnaire was scored from 0-18 with higher scores indicating more severe symptoms. The
scale showed acceptable internal reliability at baseline (alpha=0.74, n=62).
Acceptance & Action Questionnaire -II (AAQ-II) to assess acceptance and action. The
AAQ-II (Bond et al., submitted) consists of 10 statements relating to the individual’s willingness
to experience difficult thoughts and feelings and the degree to which these interfere with their
lives (psychological flexibility). Participants rated how true each statement was for them on a
scale of 1-7. The AAQ-II has been shown to have good construct, concurrent, predictive and
discriminant validity (Bond et al., submitted). The questionnaire was scored from 10 to 70 with
higher scores indicating greater psychological flexibility.
weight loss strategies over the previous seven days on a five-point scale ranging from ‘never’ to
‘all of the time’. This measure was completed at baseline, and at the two follow-ups.
Page 14 of 37 Download full-text
significantly lower levels of physical activity, 5.3 (4.2) versus 7.3 (3.7) respectively, t(60)=2.04,
Mindfulness based weight loss 14
Hypothesised moderator variables were a) number of workshops attended, b) use of
workshop principles during the programme, c) use of workshop principles at 6 month follow-up,
d) workshop understanding, e) relevant value identification (i.e. motivation; Tapper et al., 2007),
f) homework completion, and g) previous experience of meditation. Data for the number of
workshops attended were obtained from registers whilst the remaining variables were assessed in
a questionnaire administered at 6 months.
All data were collected by assistants who were blind to participant group allocation and
who were otherwise uninvolved in the project. BMI was assessed at the university and
questionnaires were collected by participants for completion in their own time. On receipt of
questionnaires the assistant checked for missing data and, where necessary, contacted the
participant to obtain these details. Five such participants were contacted at 4 months and 12 at 6
Comparison of quantitative questionnaire measures between intervention and control groups at
baseline showed that the intervention group scored significantly higher on the binge eating scale
compared to controls, 9.1 (3.5) versus 7.2 (3.9) respectively, t(60)=2.03, p<.05, and reported
p<.05. Other measures were well-matched (see Table 3).