Introduction

Countries across the spectrum of development are grappling with burgeoning rates of childhood obesity [1,2] although recent evidence of plateauing prevalence in few countries has emerged [3]. Chronic obesity-related diseases previously diagnosed among adults have now emerged among paediatric populations [1,4,5]. Malaysia, a developing, middle-income and ethnically diverse country situated in South-East Asia, is likewise experiencing a concomitant rise in paediatric obesity [6,7]. The National Health and Morbidity Survey (NHMS IV) in 2011 reported that on a national level, 6.1% of Malaysian children and adolescents aged below 18 years were obese [8].

While eating is a necessity for survival, food intake beyond energy requirements in today’s ‘obesogenic’ environment is common, leading to consumption largely stimulated by pleasure (‘hedonic hunger’) rather than physiological hunger [9]. Interestingly, not all individuals become obese, some remain lean despite increased food intake. Differences in heritable neurobehavioural and psychological traits affecting eating behaviours may explain a greater predisposition or resistance to weight gain observed among individuals [10].

Eating behaviour is influenced by both internal and external determinants including food availability, knowledge, attitudes, emotional state and experiences of the individual [11]. Profiles of eating behaviours that predict weight status is a growing area of research. Although experimental laboratory studies are objective measures of eating behaviour, it is often impractical for large-scale research due to external factors present during the test, a limited number of participants allowed and the cost of setup [12]. The development of several psychometric tools has made epidemiological studies on eating behaviour possible. The most comprehensive tool currently identified is the Children’s Eating Behaviour Questionnaire (CEBQ) developed by Wardle and colleagues [13].

The CEBQ is a 35-item questionnaire, originally designed as a parent-report to measure eating behaviour traits related to obesity risk among children aged below 12 years old. All items were developed from a combination of literature review and interviews with parents [13]. The CEBQ contains eight dimensions of eating styles; four subscales that measure food-approach behaviours [Food Responsiveness (FR), Enjoyment of Food (EF), Emotional Overeating (EOE), Desire to Drink (DD)] and another four subscales that reflect food-avoidant behaviours [Satiety Responsiveness (SR), Slowness in Eating (SE), Emotional Undereating (EUE) and Food Fussiness (FF)].

FR reflects response to environmental food cues while EF refers to a general interest in food, traits strongly observed in overweight or obese children [12,13]. EOE and EUE represent fluctuating food intake during negative emotional states. BMI in children has been positively associated with EOE and negatively associated with EUE [14-17]. DD represents the desire to drink liquids, particularly sugar-sweetened beverages (SSBs) [13]. Frequent consumption of SSBs have shown strong positive associations with BMI [18]. SR reflects the ability to regulate food intake according to internal satiety cues. Overweight and obese individuals tend to allow external food cues to override internal satiety signals [12]. A reduced eating speed due to a lack of enjoyment and decreased interest in food characterises SE. Obese children tend to eat faster towards the end of a meal [19]. FF is described as being highly selective of foods resulting in a limited variety of foods consumed. To date, associations between fussy eaters and BMI among children have been inconclusive [14,15,17].

The CEBQ has demonstrated good internal consistency, test-retest reliability and has been validated in the United Kingdom, Portugal, Chile, the Netherlands, Sweden, China and Canada; investigating eating behaviours of toddlers and children [13-16,20-22]. This instrument has been successfully employed as a parent-report in various studies including assessing the continuity and stability in eating behaviour of children, appetite preference in children of lean and obese parents, investigating relationships between temperament and eating behaviours and associations between appetitive traits with weight and dietary intake [23-26]. As such, this study intended to extend the CEBQ to adolescents as a self-report.

Food habits, eating patterns, dietary intake and risk of eating disorders among children and adolescents of varying weight status across Malaysia have been extensively studied [27-29], mirroring the unhealthy habits of their Western counterparts. However, investigations focused on appetitive traits associated with weight gain is unexplored amongst Malaysian children or adolescents. Studies in different cultures are necessary to provide insights into the similarities and differences inherent across cultural borders with regards to eating behaviours related to obesity. This present study describes the adaptation of the CEBQ as a self-report among adolescents in the local setting. The factor structure and psychometric properties of the Malay version of the self-reported CEBQ were examined. Gender and ethnic differences in eating behaviour were explored and associations between eating behaviour and BMI z-scores among multi-ethnic adolescents were further investigated.

Methods

Study procedures

An overview of the study procedures is illustrated in Figure 1. In Phase 1, the final version of the Malay self-reported CEBQ was distributed to 386 Form One students in two conveniently selected government-funded secondary schools in Kuala Lumpur. A total of 362 participants completed the self-administered questionnaire in school without the guidance of the researcher or teachers, yielding a response rate of 93.8%.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Flowchart of study procedures.

https://doi.org/10.1371/journal.pone.0082885.g001

A test-retest study was conducted among a subsample of 151 students from one school that consented. The questionnaire was self-administered twice within a 7 day interval. A total of 133 adolescents completed the questionnaire on both occasions.

In Phase 2, a total of 652 adolescents aged 13 years were recruited via multi-stage sampling from 23 government-funded national secondary schools in Kuala Lumpur. The Malay self-reported CEBQ was self-administered by the adolescents with 646 completed questionnaires returned (99.08%). Parents were requested to provide demographic information including ethnicity, parental educational level, total monthly household income and medical history of their child prior to the study. Ethnicity was self-reported (Malay, Chinese, Indian, Others).

Anthropometric measurements on the adolescents were performed by trained enumerators following a standard protocol. Body weight was measured in light clothing and with shoes removed, to the nearest 0.01 kg using a digital calibrated floor scale (SECA 813, Hamburg, Germany). Height was measured without shoes to the nearest 0.1 cm with a portable stadiometer (SECA 813, Hamburg, Germany). BMI was calculated in kg/m² and was converted to standardised z-scores, adjusted for gender and age based on the International Obesity Task Force (IOTF) cut-off points [34]. Waist circumference was measured to the nearest 0.1 cm at the umbilicus, between the tenth rib and the iliac crest. Hip circumference was measured to the nearest 0.1 cm at the point of maximum extension of the buttocks using a flexible tape measure (SECA 203, Hamburg, Germany). The 90^th percentile was used as the cut-off point to identify abdominal obesity among Malaysian adolescents [35]. This translated to waist circumferences above 83.8 cm for boys and above 78.8 cm for girls, aged 13 years old.

Results

Gender and ethnic differences

Statistical significance in eating behaviours was not detected between genders (Figure 2). Ethnic differences in eating behaviours among the three major ethnic groups are graphically represented in individual radar charts (Figure 3). The four food-approach subscales were indicated along the points of the polygon in clockwise direction followed by the four food-avoidant subscales. Malay adolescents reported significantly higher FR, EF, EOE, SE, EUE and FF1 mean scores compared to their Chinese and Indian peers.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Eating behaviour mean scores of 13 year old Malaysian adolescents by gender.

Error bars indicate 95% CI. Mean scores were obtained with the self-reported CEBQ. Statistical significance was not detected.

https://doi.org/10.1371/journal.pone.0082885.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Eating behaviour mean scores among 13 year old Malaysian adolescents by ethnicity.

(Clockwise) The four ‘food-approach’ subscales are indicated along the polygons followed by the four ‘food-avoidant’ subscales. Markers indicate point estimates of the eating behaviour scores on the self-reported CEBQ among the Malay, Chinese and Indian adolescents. ^* Statistical significance.

https://doi.org/10.1371/journal.pone.0082885.g003

Discussion

This present study pioneers the adaptation of the CEBQ as a self-report among adolescents compared to the original parent report for toddlers and younger children used in previous studies. To the best of our knowledge, our study is the first to validate an instrument that investigates eating behaviours associated with weight gain among multi-ethnic urban Malaysian adolescents.

In our sample, a nine-factor structure solution explained 54% of the variance. The SR scale was dropped due to low-loading items. CFA supported the 8-factor model which contained variations from the original CEBQ. The four items from different subscales that loaded on the FR scale were retained on their original, theoretically determined scales. The EF scale split into two factors, separating these pair of items, ‘I am interested in food’ and ‘I look forward to mealtimes’ from ‘I love food’ and ‘I enjoy eating’ but was retained as a single factor for subsequent analysis. Enjoyment of food can be attributed to the allure of taste, aroma, appearance and variety of food available. These make eating a pleasurable experience such as sharing conversations over a gastronomical delight. That said, individuals have an innate preference for energy-dense foods that are more satiating which thereby, increases overall energy intake [41].

An interesting finding was that the adolescents perceived food fussiness as two different concepts; dislike towards food (FF1) and trying new food (FF2). A picky or fussy eater is often someone who is selective of foods, consumes an inadequate variety of food [42] and therefore, is at risk of nutritional deficiency. Food preferences commonly involve sensory properties including taste, texture, colour, appearance and aroma [43]. Rozin and Fallon (1987) postulated three motivational dimensions for acceptance or rejection of foods – anticipated consequences (danger), sensory perception of food (distaste) and ideational reaction towards food (disgust) [44]. Thus, an individual resolves a dislike towards a certain food not when the food is tasted rather when the food is within sight. The FF2 scale comprised of items that revolve around trying new foods. Food neophobia refers to the reluctance to eat or the avoidance of new foods [19]. Similar to food rejection, willingness to try novel foods can be decided solely on visual cues. Previous negative experiences and unfamiliarity can likewise invoke a cautionary response.

Two items from the SR scales clustered with the FF1 items and therefore, were combined into the FF1 scale. Reaching or displaying early satiety and not having a clean plate may possibly be related to being a picky eater, showing distaste or disgust or perceived harm. Leaving food on the plate at the end of a meal may be regarded as being wasteful and is culturally frowned upon due to superstitious beliefs and seen as disrespect or lack of gratitude in the local culture [45]. The remaining items on the SR scale were omitted due to low factor loadings. It is plausible that the adolescents could not identify with these items.

The psychometric properties of the Malay self-reported CEBQ (30-items, 8-factor) were generally acceptable. Most scales showed strong internal consistency similar to previous works [13,14,16,20,22] except for EUE, FF1 and FF2 scales. The extent to which an individual experiences hyperphagia (overeating) or hypophagia (undereating) during emotional distress may vary and evidence is still inconclusive [11]. Cultural beliefs and norms towards being fussy about food should be given due consideration. Exhibiting satiety before finishing a meal and food fussiness is often regarded as discontentment in our local culture.

The self-reported CEBQ demonstrated good reliability between the two administrations. Prior to this, a test-retest was performed only with the original CEBQ and was analysed using Pearson’s correlation coefficient [13]. The test on the second occasion was intentionally administered under conditions as similar as possible to the first test. Several factors such as the interval between Test 1 and Test 2, recall bias, true changes and intrinsic feelings must be acknowledged as these may have potentially affected test-retest reliability. In addition, a moderate correlation resulted from the split-half method indicated an acceptable level of reliability.

The four food-approach scales were positively correlated as each scale independently and collectively measured increased appetitive traits. Unlike prior findings whereby EF, EOE, FR, SR, SE and FF scales differed significantly between genders [15,16,21], no statistically significant differences in eating behaviour were detected between the male and female adolescents in our study. This is interesting and warrants further investigation since female adolescents; in particular, tend to exhibit weight concerns which often results in dietary restraint or restrictive eating. Adolescents are also vulnerable to eating disorders often fuelled by negative body image perceptions or low self-esteem [19]. It is worth mentioning that a trend of EF, EOE and SE was observed among female adolescents, similar to Santos and colleagues [15]. Food-approach behaviours may predispose them to a higher risk of developing obesity. This interest in eating should be nurtured towards self-efficacy in the preparation, selection and consumption of healthy eating. The food-avoidant behaviours identified including sensitivity to satiety signals and a healthy eating pace should be encouraged to prevent unnecessary weight gain among these adolescents.

Comparison of eating behaviours across ethnicities among Malaysian adolescents using the CEBQ is previously unexplored. Food responsiveness, enjoyment of food, emotional overeating, slowness in eating, emotional undereating and food fussiness 1 (‘dislike towards food’) were most significantly reported among the Malay adolescents. Malays originated from across the South East Asian archipelago which infuses diverse and rich flavours into their cooking. Malay cuisine, in particular, is very aromatic, flavourful and comprises of a delectable variety. Within Peninsula Malaysia, traditional Malay food is recognised by their states, each with their unique taste and cooking styles [46,47]. It is possible that growing up eating flavourful foods may have developed strong taste preferences. Moreover, Malays, in general, are very communal, collectivist and hospitable in nature. With the mushrooming of eateries and 24-hour ‘mamak’ (Indian Muslim) restaurants, the accessibility and availability to an abundance of food is unrivalled. In that vein, eating in a hurried pace is discouraged as it equates to gluttony. This may elucidate the relaxed pace of eating observed among these adolescents. This is speculative but worthy of future research to investigate the appetitive traits observed among this ethnic group.

Findings in this study surfaced the influence of emotions on eating behaviours significantly observed among Malay adolescents. Emotional stress has found to increase or decrease eating responses. Emotional eaters or hyperphagics regulate or escape negative emotions by overeating, often sweet, high-fat and energy-dense foods which increases the risk of adiposity [48]. Conversely, intense emotional arousal can reduce gut activity, as observed in emotional undereating or hypophagia [48]. Stemming from a high-context culture that involves emotions and values relationships, the Malays emphasise manners or ‘adab’, propriety and prefer indirect, non-verbal communication. Anthropological studies concur that they are generally polite, obliging, considerate towards the feelings of others and strive to preserve harmony [49,50]. Overeating when experiencing negative emotions may be regarded as a non-disruptive coping mechanism. However, triangulation of data is needed and more work is advisable to elucidate and confirm this finding. Guiding these adolescents on how to channel their emotions into constructive activities is strongly recommended. Empowering these adolescents with strategies to manage ‘hedonic hunger’, particularly in an ‘obesogenic’ environment is likewise necessary to avoid overindulgence in food intake.

Several authors have found that BMI z-scores were positively associated with FR, EF and EOE and were inversely associated with SR and SE scales [14-16] while others have reported no associations [20,21]. We found BMI z-scores were inversely associated with FF1 (‘dislike towards food’). Obese individuals tend to be less fussy eaters compared to their leaner counterparts, however, previous studies have yielded mixed results [14,15]. While food fussiness may appear to be protective against excessive calorie intake and subsequent weight gain, a well-balanced and nutritionally-adequate diet should not be compromised.

Possible reasons that our findings differ from the original CEBQ are three-fold. Firstly, the parent report may reflect the parent’s perception of their child’s eating behaviour and this definitely varies from an adolescent’s self-evaluation of their own eating styles. Secondly, while age difference was not the focus of this study, it should be highlighted that the parent report focused on younger children whereas studies among adolescents were unexplored prior to this. Thirdly, cultural differences in a multi-ethnic Asian setting should be considered since the influence of culture is complex and is mediated by biological, demographic, psychosocial and environmental factors [51].

The sample in Phase 2 (association study) may have under-represented the Chinese ethnic group in the country possibly due to volunteer bias. Causality cannot be determined due to the cross-sectional design. It is recommended to replicate this study in a more heterogeneous sample with a wider age range.

Despite these limitations, this study spearheads the validation of the CEBQ as a self-report measure to investigate eating styles related to obesity among adolescents of various ethnicities in Malaysia. Understanding behavioural traits allows for planning and implementation of obesity prevention and intervention strategies in the local context. This self-report questionnaire can then be used further as an assessment tool to evaluate the effectiveness of those measures.

Conclusions

Although CEBQ is a valuable psychometric instrument, adjustments were required due to age and cultural differences in our sample. While previously used as a parent-report among younger children, the self-reported CEBQ now empowers adolescents to evaluate their own eating behaviours from a range of dimensions. Our findings reveal the influence of gender, ethnicity and weight status on eating behaviour amongst multi-ethnic Malaysian adolescents using the Malay self-reported CEBQ. In addition, a lack of dislike towards food was observed among obese adolescents. Future longitudinal and qualitative studies are warranted to understand behavioural phenotypes of obesity and to guide prevention and intervention strategies.

Acknowledgments

The authors would like to express our appreciation to the Ministry of Education (MOE) and Federal Territory of Kuala Lumpur Education Department (JPWP). We are grateful to all the school administrators, teachers, parents and participants for their co-operation and support throughout the study. Our sincere thanks and gratitude to Professor Jane Wardle for her advice and guidance.