The prevalence of overweight and obesity has risen dramatically worldwide and is considered to be a major global health concern(1). Overweight and obesity can put individuals at risk of physical (e.g. hypertension and type 2 diabetes)(Reference Adams, Schatzkin and Harris2,Reference Afshin, Forouzanfar and Reitsma3) and psychological consequences (e.g. depression and anxiety disorders)(Reference Avila, Holloway and Hahn4). Evidence shows that losing even a small amount of body weight (5–10 % of body weight) can improve health outcomes such as blood pressure and total cholesterol(Reference Wing, Lang and Wadden5). However, traditional weight management approaches that promote dietary restriction seldom lead to weight loss in the long term (i.e. >12 months)(Reference Haines and Neumark-Sztainer6). In addition, rigid dietary restriction can increase the risk of weight cycling and psychological problems, such as disordered eating(Reference Haines and Neumark-Sztainer6). Negative body image, disordered eating, depression and anxiety have been linked to poor compliance with weight loss programmes(Reference Lazzeretti, Rotella and Pala7), but these psychological factors are not typically addressed in many of these programmes(Reference Collins, Meng and Eng8,Reference Lopresti and Drummond9) . A growing body of literature recommends encouraging healthy dietary habits rather than weight loss, and that targeting psychological factors that are linked to body weight could be beneficial for physical and mental health(Reference Bacon and Aphramor10,Reference Tylka, Annunziato and Burgard11) .
A novel and promising approach to addressing these psychological barriers to promoting healthy dietary habits could be enhancing individuals’ self-compassion. Self-compassion, which involves cultivating a kind and compassionate mindset towards oneself(Reference Neff12), is associated with decreased disordered eating and body dissatisfaction(Reference Braun, Park and Gorin13) and with increased mental well-being(Reference MacBeth and Gumley14). Therefore, including self-compassion in nutrition interventions might lead to better outcomes. However, most of the previous studies that examined the effects of self-compassion interventions have focused on weight loss rather than improved dietary habits. These studies have also suffered from some methodological limitations, such as selection bias (e.g. Mantzios and Wilson(Reference Mantzios and Wilson15)), and none of them has examined participants’ experiences in depth regarding the acceptability of a self-compassion intervention for improving dietary habits(Reference Horan and Taylor16–Reference Palmeira, Pinto-Gouveia and Cunha18). Examining the acceptability of nutrition interventions that include a new approach such as self-compassion could guide the development and facilitation of effective nutrition programmes.
Two additional strategies that could be useful in the promotion of healthy dietary behaviours are goal-setting and self-monitoring(Reference Lara, Evans and O’Brien19) with effectiveness demonstrated in both short-term (≤6 months) and long-term studies (≥12 months)(Reference Samdal, Eide and Barth20,Reference Saperstein, Atkinson and Gold21) . However, factors such as lack of motivation and negative reaction to minor failure could derail individuals from their goals and eventually lead to goal abandonment(Reference Mann, De Ridder and Fujita22). Recent theoretical evidence(Reference Terry and Leary23) and empirical studies(Reference Biber and Ellis24,Reference Sirois25) show that self-compassion might be beneficial in addressing these barriers through increasing motivation and regulating negative emotions related to goal lapses(Reference Terry and Leary23,Reference Adams and Leary26) . Thus, integrating self-compassion into goal-setting and self-monitoring interventions for improving dietary habits might improve goal-striving and intervention outcomes.
Finally, online interventions could provide useful tools for promoting healthy dietary habits. Online technologies, such as internet-based interventions and mobile health, have increasingly been used to facilitate the delivery of dietary change interventions(Reference Okorodudu, Bosworth and Corsino27). These interventions provide a platform for individualised feedback and support, which could also improve goal attainment(Reference Krukowski, Harvey-Berino and Bursac28). Studies indicate that using online technologies in nutrition programmes can be of low cost, have a wide reach to clients and could improve dietary habits(Reference Okorodudu, Bosworth and Corsino27). However, the development and assessment of online tools for dietary behaviour change are in their infancy(Reference Dennison, Morrison and Conway29). To ensure higher acceptability and adherence in future online nutrition interventions, exploring people’s perceptions of the acceptability of these interventions is crucial(Reference Dennison, Morrison and Conway29).
The current pilot study aimed to investigate the efficacy and acceptability of a combined online and face-to-face behavioural intervention that used self-compassion, goal-setting and self-monitoring strategies for improving dietary behaviour, as well as psychological factors associated with dietary behaviour, in subjects with overweight or obesity. Considering that the majority of nutrition programmes have poor outcomes(Reference Hafekost, Lawrence and Mitrou30), the insights gained from this research will be useful in guiding future dietary interventions.
Methods
This pilot study had two main aims. The first aim was to investigate the efficacy of the intervention. The primary assessable outcomes of this aim were levels of self-compassion; eating pathology; depression, anxiety and stress; and dietary intake (e.g. fibre intake). Secondary outcomes were body weight, BMI and waist and hip circumferences. The second aim of the study was to examine participants’ perceptions about the usefulness and acceptability of the intervention.
An embedded mixed methods design(Reference Creswell and Clark31) was used for the pilot study. This design involves collecting qualitative data and quantitative data on the same topic to answer different questions that require different types of data. In this study, the qualitative component was embedded within a quantitative design. The two qualitative and quantitative components were given equal weight. The quantitative phase was a 4-week before-after trial, and the qualitative phase of the study was a structured one-on-one in-person interview conducted after the intervention to provide data on participant perceptions about the acceptability of the intervention.
A convenience sampling strategy was used to recruit fifteen participants from the student and staff population of UNSW Sydney during early 2016. Online means such as emails and physical posters on the university campus were used to recruit participants. Participant inclusion criteria were: aged 18–55 years, BMI 25–40 kg/m2, access to a computer/tablet/smartphone that can run an internet browser for at least 1 h/week, able to read and write English and being open to changing eating habits and potentially lose weight. The exclusion criteria were: taking any weight-loss medications or previous use of weight-loss medications during the past 6 months, currently using medication which has substantial weight gain, suffering from any major medical illness or having a history of major medical illness (in the last 5 years), pregnancy or lactation, current participation in any other nutrition or weight loss programme or seeing a nutrition professional, currently smoking, and weight loss of more than 4·5 kg (10 pounds) during the past 6 months. Participants who finished the study received two personal care items (a hand cream and sanitizer) and were entered into a prize draw to win one of the three packages of prizes that included three or two of the following items: a meditation course voucher, an organic fruit and vegetables box and/or a book.
Quantitative data collection
At the beginning and end of the intervention (i.e. at Week Zero and Week Four), participants completed several questionnaires online to obtain demographic information and levels of self-compassion, eating pathology, and depression anxiety and stress. Participants also completed 3-d food diaries and had anthropometric measures taken at Week Zero and Week Four during one-on-one in-person meetings.
Self-Compassion Scale
The Self-Compassion Scale (SCS) is a twenty-six-item self-reported measure designed to assess typical thoughts, emotions and behaviours associated with different components of self-compassion(Reference Neff32). The SCS consists of six subscales: Self-Kindness, Self-Judgement, Common Humanity, Isolation, Mindfulness and Over-Identification. Responses are made on a five-point scale from 1 (Almost never) to 5 (Almost always). Subscale scores are computed as the mean of items in the subscale. For the computation of the overall self-compassion score, negatively worded items were reverse-coded and an average of all items was calculated so that higher scores represent greater levels of self-compassion(Reference Neff32). Internal consistency reliability for overall SCS was excellent (Cronbach’s α = 0·93) in the current study and was good for most of the SCS subscales (Cronbach’s α’s ranged from 0·69 to 0·94).
Eating Disorder Examination Questionnaire
The Eating Disorder Examination Questionnaire (EDE-Q) is a twenty-eight-item questionnaire that asks about maladaptive eating behaviours over the previous 4 weeks(Reference Fairburn33) and provides two types of data. First, it generates a frequency of occurrence of the main behavioural traits of eating disorders such as binge eating (six questions). Second, it has subscale scores that provide the severity of eating-related psychopathology(Reference Fairburn33). These items are responded to on a scale that ranges from 0 (No days) to 6 (All days). The four subscales are: Restraint, Eating Concern, Shape Concern and Weight Concern. The score for each subscale is obtained by calculating the mean of all items for that subscale. The measure also produces a Global score for overall eating pathology which is obtained by averaging the four subscale scores. Higher EDE-Q scores reflect a greater severity of eating psychopathology. In the current study, Cronbach’s alpha for EDE-Q Global was 0·90 and for the subscale scores of Restraint, Eating Concern, Shape Concern and Weight Concern were 0·71, 0·75, 0·85 and 0·58, respectively.
Depression Anxiety and Stress Scale-21
Depression Anxiety and Stress Scale-21 (DASS-21) is a twenty-one-item self-administered instrument assessing psychological distress(Reference Lovibond and Lovibond34). It is composed of three subscales: Depression, Anxiety and Stress. Respondents indicate the extent to which they experienced negative emotional states over the past week, ranging from 0 (Did not apply to me) to 3 (Applied to me very much)(Reference Lovibond and Lovibond34). To attain a score for each subscale, the ratings for the subscale items are summed. Cronbach’s alpha for the three subscales ranged from 0·77 to 0·92 in the current study.
Estimated food diary
Participants were asked to record every item of the food and drink consumed for three consecutive days (two weekdays and one weekend day)(Reference Rutishauser35). To collect the 3-d food diary data, online Google Sheets with instructions on how to record food intake were shared with the participants. At the end of each 3-d recording period, the first author reviewed the food diaries with the respondent during the in-person meetings to clarify entries and to probe for forgotten items. Data from the food diaries were entered into the FoodWorks 7(36) software programme for nutrient analysis. Average daily energy intake and nutrient intakes (protein, carbohydrate, fat, alcohol and fibre) that were most likely to be associated with body weight regulation(Reference Astrup and Brand-Miller37) were obtained from the software outputs.
Anthropometry
Body weight, height, and waist and hip circumferences were measured objectively. Weight was measured without shoes and in light clothing using a calibrated digital standing scale (SECA 817), with a precision of ±0·1 kg. Standing height was measured without shoes, using a portable stadiometer (SECA 213). BMI was calculated from these measurements using the formula weight (kg)/height2 (m)(1). Waist circumference was measured directly on the skin using a measuring tape (SECA 201) at the midpoint between the margin of the last palpable rib and the top of the iliac crest(38). Hip circumference was measured with light clothing at the widest area of the buttocks(38). Waist and hip circumferences were assessed in duplicate, and the averages were calculated.
Structured interview
One-on-one, in-person structured interviews were conducted during the Week Four meeting. The interview also included some quantitative questions about the participant’s satisfaction with the intervention. Closed-ended questions included questions such as, ‘Which aspects of the program did you find most useful?’, and participants were provided with a list of answers by the interviewer to select from (see online Supplementary Appendix 1 for the interview guide). Participants were also asked to rate their satisfaction with the intervention using a five-point rating scale that ranged from 1 (very dissatisfied) to 5 (very satisfied). These questions were then followed by open-ended questions such as, ‘Can you describe why you found ‘X’ aspect of the study useful?’ to probe reasons for participants’ opinions. All participants who completed the study (n 14) were interviewed in order to capture as much diverse insight as possible. Interviews were conducted by the first author in a private room at UNSW and lasted between 20 and 35 min each. To add to the study’s trustworthiness (study credibility), additional data resources such as participant goal sheets and the email correspondence between participants and the first author were reviewed to verify findings from participants’ interviews.
The interviews were audio recorded using digital dictation voice recorders (Olympus DS-2500) and transcribed verbatim by a professional transcription service. To ensure the veracity of data, participants were provided with an opportunity to review and check whether the transcripts accurately reflected what they said (i.e. respondent validation); none of the participants expressed any concern.
Intervention
During the 4 weeks of the intervention, participants received information at the beginning of each week. The first information pack was given verbally as well as in printed handouts during the baseline face-to-face meeting. The rest of the information was sent in PDF documents via email. Each information pack had two sections: one providing information on nutrition and the other providing information on self-compassion. Participants were advised to set goals based on the information provided and to track their performance online over the intervention period.
Goal-setting
The goal-setting protocol was based on Locke and Latham’s goal-setting theory(Reference Locke and Latham39). Participants were encouraged to set proximal (short-term), timely, specific goals, and to reward themselves for any success. In addition, factors that may facilitate achievement of health-related goals, such as promoting self-efficacy(Reference Mann, De Ridder and Fujita22), were included in the goal-setting instructions.
During the initial in-person meeting, participants set two goals with the first author’s guidance. One goal was about dietary habits (e.g. ‘I aim to eat three serves of vegetables every day’), and the other goal was about self-compassion behaviours (e.g. ‘I aim to treat myself like a good friend under challenging situations this week’). They were also advised to set new goals every week based on the new information they would receive. Participants had the option of carrying forward their nutrition goals to subsequent weeks or setting new ones. Participants were asked to set or retain a maximum of three nutrition goals and one self-compassion goal per week (i.e. a total of four goals in any week).
Participants were shown how to use a personalised online Google goal sheet for self-monitoring as well for interacting with the first author. Goal sheets were structured as weekly calendars with space to enter their goals and then track daily progress (i.e. indicating whether or not they completed the goal with a ‘Yes’ or ‘No’). During the study, participants could contact the first author for further guidance. The first author reviewed each participant’s goal sheet at the end of each week, and feedback was emailed to the participant. Email reminders were also sent to participants if they did not complete their goal sheets for three consecutive days.
Nutrition information
Nutrition information was based on the Australian Dietary Guidelines(40). Guidance on the ideal intake of foods was tailored to focus on the regulation of body weight and hunger. Therefore, the information encouraged a diet with foods high in protein, fibre and carbohydrates low in glycaemic index and low in energy density. Each week, two or three of the food categories listed in the Australian Dietary Guidelines were introduced to participants along with some goal options related to these food categories.
Self-compassion information
Self-compassion information and goal options were partially based on Neff’s website(Reference Neff41). The website teaches ‘mindful self-compassion’(Reference Neff41). The investigators partially modified the information to focus more on how self-compassion may be related to nutrition and dietary behaviour change (e.g. emotional eating or goal relapse)(Reference Adams and Leary26,Reference Terry, Leary and Mehta42) . Goal options provided were either formal practices (i.e. guided meditation) or informal practices (i.e. self-compassionate thoughts in daily life, such as repeating self-compassionate phrases to oneself in moments of suffering). The informal practices were related to sufferings either in general life or relating to body image and diet. One of the goal examples relating to the distress associated with dietary habits was: ‘If I do not accomplish my nutrition goals as much as I would like, and I won’t feel guilty. Instead, I will motivate myself to do better in the future with encouraging language.’
Statistical analysis
Descriptive statistics were used to describe the baseline characteristics of the study sample. To compare changes before and after the intervention, paired samples t tests and Wilcoxon-paired rank tests were used for the normally distributed and non-parametric data, respectively. Simple linear regressions were carried out to examine if changes in self-compassion predicted the Week Four values of each outcome variable. The regression models were adjusted for baseline values of those outcome variables. Data analysis was performed using SPSS (version 22). Differences were considered to be statistically significant at P < 0·05. Cohen’s d effect size was used for the effect size calculation, with effect sizes of 0·2, 0·5 and 0·8 representing small, medium and large effects, respectively(Reference Cohen43).
Qualitative analysis
Qualitative content analysis was used to analyse the qualitative interview data. A deductive approach was used to code the data and assess the conceptual and theoretical underpinnings of the study(Reference Elo and Kyngäs44,Reference Marshall and Rossman45) . An inductive approach was then applied to develop higher order categories or data that did not fit into the unconstrained matrix. The latter approach is taken when there is not enough information about the topic to be analysed(Reference Elo and Kyngäs44).
The transcripts were read several times by the first author before coding. After initial open coding based on a few transcripts, the first author consulted with a qualitative expert to confirm the validity of the generated codes. Codes related to similar or dissimilar opinions on the same topic were collapsed into broader categories to reduce the total number of categories. QRS International Nvivo 11 software was used for coding and managing the data. After coding all transcripts, refining codes, categorisation and abstraction, a list of categories and subcategories was generated and their definitions were discussed between the first and fourth authors.
Results
Response rate and participant characteristics
Out of forty-six people (forty-three women and three men) who initially responded to the advertisements, eighteen female participants were interested and eligible; of those, fourteen completed the study. Figure 1 presents the recruitment process and the numbers of participants involved at each stage of the intervention. The average age of the sample was 37·9 (sd 9·8) years, and the average BMI was 30·58 (sd 3·44) kg/m2. All participants had some university education with 71 % having postgraduate education. The ethnic composition of the sample was diverse with 29 % of the participants being Oceanian, 14 % European, 7 % African and Middle Eastern, 22 % Asian, 14 % American and 14 % others.
Fig. 1 Participant flow chart
Participants’ earlier exposure to self-compassion
Some information about participants’ earlier exposure to self-compassion was collected because the early exposure might affect participants’ ability to develop a self-compassion mindset and their perception of the study acceptability(Reference Neff and Germer46). Eight participants reported that they had already heard of or were familiar with the concept of self-compassion and some of the participants were familiar with some similar concepts, such as mindfulness.
Changes between Week Zero and Week Four
Table 1 provides within-participant comparisons on the following outcomes: self-compassion; eating pathology; depression, anxiety and stress; and anthropometry. With respect to self-compassion, there were significant improvements in the total scores on the SCS as well as some of its subscales. Global scores on the EDE-Q did not show any significant change, but there was a significant decrease for two of the subscales and for the frequency of binge days (days on which binge eating occurred). There was also an increase in levels of the Restraint subscale of the EDE-Q that fell just short of significance. Further analysis of the five items that make up the Restraint subscale revealed that the scores on items related to food avoidance (P = 0·01) and dietary rules (P = 0·02) increased significantly after the intervention, while scores on the other three items (restraint over eating, avoidance of eating and empty stomach) did not change significantly (Ps > 0·60). Among the DASS subscales, only a decrease in Depression scores approached significance. There was no significant change in any anthropometric variables after the intervention.
Table 1 Anthropometry, eating behaviours, depression, anxiety, stress and self-compassion at Week Zero and Week Four (n 14)
SCS, Self-Compassion Scale; EDE-Q, Eating Disorder Examination Questionnaire; DASS, Depression Anxiety and Stress Scale.
* Non-parametric analysis.
Week Zero and Week Four comparisons for dietary outcomes, such as energy and macronutrient intake, are presented in Table 2. Decreases in average daily energy intake and some macronutrients’ intake were significant after 4 weeks of intervention. There was no change in fibre consumption; however, after adjusting for energy intake, fibre intake showed a significant increase from 2·9 to 3·5 g/MJ. The proportions of energy provided by the different macronutrients did not change significantly over the course of the intervention.
Table 2 Dietary intake at Week Zero and Week Four
* Non-parametric analysis.
Self-compassion change as a predictor of study outcomes
Table 3 shows coefficients of simple linear regression predicting the study outcomes at Week Four based on changes in self-compassion total score and subscale scores. Note that, because the impact of the intervention was similar for the three positively worded items and for the three negatively worded items, these were combined to form positive and negative subscales of self-compassion. The regression analyses were adjusted for baseline values of the respective outcomes. Changes in positively worded self-compassion subscales predicted Week Four scores on the Stress subscale of DASS and marginally predicted scores on the Week Four Anxiety subscale. Changes in negatively worded self-compassion subscales did not significantly predict any Week Four scores for eating pathology or depression/anxiety/stress.
Table 3 Regression coefficients for changes in self-compassion predicting outcomes at Week Four
SCS, total self-compassion score; SCS pos, self-compassion positively worded items; SCS neg, self-compassion negatively worded items; DASS, Depression Anxiety and Stress Scale; EDE-Q, Eating Disorder Examination Questionnaire.
†P < 0·10, *P < 0·05.
Goal adherence
To determine participants’ goal adherence, participants’ online goal sheets were reviewed. Goal adherence was calculated by the number of times a ‘Yes’ was recorded for reaching a goal divided by the number of times the goal was supposed to be accomplished. On average, participants accomplished 73 % (sd 14·8) and 67 % (sd 18·9) of their nutrition and self-compassion goals, respectively, during the 4 weeks of the intervention.
Acceptability and satisfaction with the study
The majority of participants stated that the study matched their expectations. Average satisfaction with the study was 3·9 out of 5 (where 4 = ‘satisfied’), and 78·5 % of participants rated the study as 4 or 5. Participants were also asked to list verbally the most useful and the least useful aspects of the study using four categories that represented the main study components (i.e. self-compassion, goal-setting, nutrition guidance and online support). Participants stated that they found self-compassion and goal-setting to be the two most useful aspects of the study.
Analysis of the transcripts from the Week Four interviews generated six major categories describing features that may have an impact on the acceptability of the study and participant adherence. Table 4 presents these six major categories and their subcategories with example quotes supporting these findings. Participants had mixed opinions and feelings about the presence or absence of these features in changing their dietary behaviours.
Table 4 Six major categories describing features that may have an impact on the acceptability of the study
Participants generally appreciated the idea of goal-setting (including self-monitoring) and self-compassion for dietary behaviour change. They believed that goal-setting and self-monitoring increased their motivation and emphasised that aiming for a realistic level of change could increase goal adherence. Participants also stated that self-compassion helped them to regulate the negative affect associated with eating and dietary change but noted that cultivating a self-compassion mindset could be challenging.
In addition to the study intervention components, participants talked about other aspects that may impact their adherence to the study. One of these aspects was the simplicity and efficiency of the current study. While some participants found the study to be time-efficient, easy to understand and accessible, others thought that the study needed to be simpler and more time-efficient. The informative aspects of the study were also something that the participants found useful. In addition, some participants highlighted that the study could have been more flexible and more interactive to address participants’ needs more favourably. Finally, feedback from the ‘expert’ (i.e. the investigator) and reminders were considered as factors that increased engagement with the study. However, some participants also suggested that opportunities for interaction with peers and more frequent interaction with experts and feedback might have improved engagement with the study.
Discussion
This pilot study aimed to investigate the efficacy and acceptability of a combined online and face-to-face behavioural intervention that used self-compassion, goal-setting and self-monitoring strategies for improving dietary behaviour in fourteen adults with overweight or obesity. The quantitative data show that the study was effective in improving self-compassion, some aspects of eating behaviour and some aspects of dietary intake. The qualitative data indicated that participants generally liked the idea of self-compassion and goal-setting for promoting healthier dietary behaviour and found the intervention to be partly acceptable and feasible. However, change scores in self-compassion did not predict any outcomes measure at Week Four, except for stress levels.
In the current study, the level of total self-compassion improved significantly over 4 weeks. The effect size (Cohen’s d = 0·61) attained in our study was comparable to previous short-term (e.g. 5 d)(Reference Palmeira, Pinto-Gouveia and Cunha18,Reference Braun, Park and Conboy47,Reference Toole and Craighead48) or online self-guided self-compassion interventions(Reference Rodgers, Donovan and Cousineau17,Reference Toole and Craighead48) . These effect sizes are smaller than those reported in interventions with a longer period (i.e. 3–8 weeks) or group sessions (average effect size Cohen’s d = 1·4)(Reference Neff and Germer46,Reference Smeets, Neff and Alberts49) . In those studies, group sessions might have given participants a better understanding of self-compassion as well as more opportunities for formal practice.
The current intervention significantly decreased some aspects of disordered eating as measured by the EDE-Q. Similar studies report comparable findings to the current study, such as significant decreases in disordered eating and body image concerns in clinical and non-clinical samples(Reference Palmeira, Pinto-Gouveia and Cunha18,Reference Adams and Leary26,Reference Albertson, Neff and Dill-Shackleford50–Reference Kelly, Wisniewski and Martin-Wagar53) . Scores on the Restraint subscale showed a slight (but non-significant) tendency to increase over the intervention (P = 0·06, d = 0·34). However, this might not be considered as a detrimental outcome given that the increase was observed on two items of ‘food avoidance’ (limiting specific foods) and ‘dietary rules’ (having some rules for eating) rather than items with more concerning patterns including ‘desire for empty stomach’ or a ‘long period of fasting’. There is evidence that self-control and some degree of restraint in people with overweight and obesity can result in less binge eating and more success in weight loss compared with subjects with lower levels of restraint(Reference Lazzeretti, Rotella and Pala7,Reference Johnson, Pratt and Wardle54) . Thus, the slight increase in the Restraint subscale might be considered as a positive outcome in this context.
Significant improvements were also observed in some aspects of dietary intake over the 4 weeks of the intervention. Energy, carbohydrate, protein and alcohol intakes decreased; and fibre intake per unit of energy increased. Our findings are consistent with the few other studies that have examined the efficacy of self-compassion on dietary intake. Those studies have also reported improvements in some aspects of the diet, such as increased scores on nutrition subscale of a health questionnaire(Reference Braun, Park and Conboy47), and reduced frequency of dietary fat consumption(Reference Horan and Taylor16).
One of the study’s hypotheses was that the current intervention could decrease depression, anxiety and stress, but there was no statistically significant effect of the intervention on these outcomes. (Depression scores did show a decreasing trend, P = 0·07, d = –0·43.) This lack of statistical significance is not congruent with previous study results, where significant decreases in emotional distress have been reported(Reference MacBeth and Gumley14,Reference Neff and Germer46,Reference Bluth, Gaylord and Campo55,Reference Orellana-Rios, Radbruch and Kern56) . The null results of the current study might be due to the short period of the intervention and the fact that current study observed a smaller effect size in self-compassion levels compared with the previous study and this change might not be sufficient to cause a significant change in the study other outcomes.
In contrast to earlier findings(Reference Mantzios and Wilson15,Reference Palmeira, Pinto-Gouveia and Cunha18,Reference Mantzios and Wilson57) , the current study did not show any change in the secondary outcomes of anthropometry. This is perhaps not surprising given that (a) the current intervention was focused on improving dietary habits rather than weight loss and (b) the intervention was of relatively short duration. Other self-compassion studies that have focused on improving dietary habits similarly either found no change(Reference Horan and Taylor16) or reported only a small change in BMI and waist circumference (Cohen’s d = 0·10)(Reference Palmeira, Pinto-Gouveia and Cunha18).
Although the current intervention successfully improved self-compassion scores, there was very limited evidence that changes in self-compassion account for the changes in other outcomes. Positively worded subscales of SCS (e.g. Self-Kindness) predicted Week Four Stress scores; for Anxiety scores, the correlation was marginally significant (P = 0·08). There were no other significant associations. These findings are in contrast to earlier research which reported that changes in self-compassion negatively predicted changes in psychological distress(Reference Neff58,Reference Neff59) and eating pathology(Reference Pennesi and Wade60). This inconsistency may be explained by the fact that the current self-compassion intervention was in the form of unsupervised self-help that did not result in as large an effect for self-compassion as the other studies have. The small sample size also could be another possible explanation for this inconsistency.
This study had a good retention rate, with 78 % of participants who gave consent completing the study. This retention rate is within the acceptable range of retention for intervention studies (i.e. about 20 % attrition for short-term studies)(Reference Furlan, Pennick and Bombardier61). The ethnic composition of the study sample was also heterogeneous indicating that the acceptability of the current study could be generalisable to more diverse ethnic groups of people. In addition, qualitative exploration also showed that the study is acceptable and promising. Most of the participants found the goal-setting and self-monitoring to be essential aspects of the current study for changing dietary behaviour. Respondents also reflected on the importance of having attainable and short-term goals. Recent studies that examined participant perceptions or expectations from web-based health programmes also reported similar findings(Reference Dennison, Morrison and Conway29,Reference Mummah, King and Gardner62–Reference Waterlander, Whittaker and McRobbie65).
Participants’ perceptions regarding the benefits of self-compassion for dietary change support the self-compassion model of health behaviours, which theorises that self-compassion might be beneficial in regulating undesirable thoughts and emotions(Reference Terry and Leary23,Reference Sirois25) that are associated with emotional eating and goal abandonment(Reference Sirois and Giguère66). However, similar to previous qualitative studies in clinical and non-clinical samples, the findings indicated that understanding the concept of self-compassion or developing a self-compassionate attitude might be difficult, especially when participants have negative thoughts, or they are self-critical(Reference Campion and Glover67,Reference Pauley and McPherson68) .
In addition to the findings related to the conceptual and theoretical underpinnings of the study, qualitative data also showed the importance of simplicity, ease of access and efficiency of online tools for dietary behaviour change. Time efficiency(Reference Papadaki, Thanasoulias and Pound63,Reference Hammarström, Wiklund and Lindahl69) , ease of use and accessibility of the online tools(Reference Papadaki, Thanasoulias and Pound63,Reference Bentley, Otesile and Bacigalupo70,Reference Mann, Riddell and Lim71) have also been discussed by participants in other studies that asked for participants’ perceptions of using mobile or online health applications for dietary change. The importance of having novel information that helps participants with dietary change in nutrition programmes was another finding which is consistent with the findings of studies that sought participants’ experience about dietary habits programmes(Reference Dennison, Morrison and Conway29,Reference Mummah, King and Gardner62,Reference Papadaki, Thanasoulias and Pound63,Reference Bentley, Otesile and Bacigalupo70) .
Finally, the last feature that participants considered essential to the study’s acceptability was for it to be engaging. Some participants suggested that the study could have been more engaging and had there been more online and in-person support, feedback and reminders. Participants in other studies also spoke about the value of feedback and reminders to increase engagement in technology-based dietary interventions(Reference Dennison, Morrison and Conway29,Reference Papadaki, Thanasoulias and Pound63) .
Overall, the evidence obtained from this study suggests that nutritionists and healthcare providers could include self-compassion in their counselling for promoting healthy dietary habits as a means of assisting people in dealing with negative emotions related to eating behaviour modification. Self-compassion mindset might have the potential to be beneficial in a broader scope such as promoting healthy eating in different settings such as schools. Schools could be an ideal setting because adolescence and childhood are important periods for growth and cognitive changes(Reference Giedd72) and might facilitate the development of both healthy eating behaviours(Reference de Vlieger, Riley and Miller73) and a self-compassion mindset(Reference Bluth, Gaylord and Campo55). Self-compassion could also be beneficial in addressing other public health issues, such as changing a sedentary lifestyle(Reference Biber and Ellis24). Future studies should explore the application of self-compassion interventions in these contexts to broaden the applicability of the self-compassion concept.
Several limitations to this study need to be acknowledged. First, this study was a short-term, one-armed pilot. Furthermore, because this was a pilot study with a small sample size, the alpha level of significance was not corrected for the multiple comparisons on the effects of the intervention. Therefore, multiple comparisons from the same set of data might have increased the likelihood of type I error (i.e. false rejection of null hypotheses). Thus, quantitative results must be interpreted with caution. In addition, the majority of participants were familiar with concepts similar to self-compassion, such as mindfulness, that could also assist with building self-compassion whereby having prior experience in similar activities is advantageous(Reference Neff and Germer46). Therefore, the current study findings could not be generalised to the general population. Finally, the qualitative interviews were conducted by the same person who provided support to participants throughout the intervention. While this connection might have contributed to a good rapport and allowed participants to feel more comfortable talking with a familiar person, it might also have influenced participants to respond in a positive way to questions about the intervention.
Conclusions
This pilot study provides preliminary but promising findings on the feasibility of the current behavioural intervention. Despite having a small sample size and a short intervention period, the intervention improved some aspects of dietary and eating behaviours. However, changes in self-compassion over the 4 weeks did not significantly predict study outcomes at Week Four, except for the level of stress. The efficacy of the intervention and the mechanism of change in the study outcomes need to be examined in future research with a larger sample, a longer intervention period and a control arm.
The current study also indicates that the combined online and face-to-face behavioural intervention that aimed to improve dietary habits was feasible and acceptable. Overall, participants in the current study found self-compassion, goal-setting and self-monitoring are essential for promoting dietary change. However, some factors such as efficiency, simplicity and the interactivity of the programme should be taken into consideration for future studies. Furthermore, self-compassion researchers might want to consider the challenge of cultivating a self-compassion mindset and explore different methods to facilitate the adoption of a self-compassionate mindset.
Acknowledgements
Acknowledgements: The authors are deeply grateful to Dr Husna Razee who thoughtfully provided invaluable guidance on qualitative data analysis. This research was a part of a PhD project which was supported by an Australian Government Research Training Program (RTP) Scholarship. Special thanks to Ooooby Sydney, Kadampa Meditation Centre Sydney and Thankyou company for donating some incentives to the study participants. Financial support: This research received no specific grant from any funding agency, commercial or not-for-profit sectors. Conflict of interest: None. Authorship: R.C.R. and H.R.-A. formulated the study questions and designed the pilot study. H.R.-A. prepared the intervention materials, and R.C.R. supervised this work. H.A.-R. and R.C.R. contributed to study recruitment. H.A.-R. carried out intervention and data collection (both qualitative and quantitative) and also performed data analysis and interpreted the data (both qualitative and quantitative). L.R.V. supervised quantitative data analysis. A.S. contributed to the final qualitative data interpretation. H.A.-R. wrote the first draft of the manuscript and R.C.R., L.R.V., N.Z. and A.S. contributed to the final manuscript. Ethics of human subject participation: This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving study participants were approved by the UNSW Human Research Ethics Committee. Written informed consent was obtained from all subjects before any data collection.
Supplementary material
For supplementary material accompanying this paper visit https://doi.org/10.1017/S1368980020000658
