Research exploring the treatment of adolescent overweight and obesity has largely focused on physiological (e.g., body composition) and behavioural (e.g., eating and activity habits) outcomes of treatment (Oude Luttikhuis et al., Reference Oude Luttikhuis, Baur, Jansen, Shrewsbury, O'Malley, Stolk and Summerbell2008). Psychosocial consequences are likely to be more immediate than physiological consequences of excess weight in adolescents (Lobstein, Baur, & Uauy, Reference Lobstein, Baur and Uauy2004) and are often the reason for seeking intervention (Lobstein, et al.), yet few studies have comprehensively explored the impact of treatment on adolescent psychosocial wellbeing. The absence of research exploring psychosocial outcomes in adolescent obesity has been highlighted in recent reviews (e.g., Oude Luttikhuis et al.).
Psychosocial outcomes of adolescent obesity interventions are of interest for two reasons. First, research is needed to ensure that treatment does not have a detrimental effect on adolescent psychosocial wellbeing. Parental concern about negative impacts of weight management interventions are commonly cited as reasons for not seeking treatment for overweight and obese adolescents (Reilly & Wilson, Reference Reilly and Wilson2006). Second, research is needed to assess the impact of treatment on promoting improved psychosocial functioning. Weight loss alone is unlikely to result in improved psychosocial wellbeing and weight management interventions need to directly target improved psychosocial outcomes (Reilly & Wilson, Reference Reilly and Wilson2006). Additionally, poor psychosocial wellbeing contributes to further weight gain (Stice, Cameron, Killen, Howard, & Taylor, Reference Stice, Cameron, Killen, Howard and Taylor1999) and negatively impacts on treatment adherence, completion and outcome (Zeller et al., Reference Zeller, Kirk, Claytor, Khoury, Grieme, Santangelo and Daniels2004).
The Choose Health Program (Brennan, Walkley, Fraser, Greenway, & Wilks, Reference Brennan, Walkley, Fraser, Greenway and Wilks2008), a healthy lifestyle program for adolescents, was developed to meet the need for effective interventions for adolescent overweight and obesity and was designed to be consistent with recommendations of systematic reviews (Oude Luttikhuis et al., Reference Oude Luttikhuis, Baur, Jansen, Shrewsbury, O'Malley, Stolk and Summerbell2008) and clinical guidelines (National Health and Medical Research Council, 2003). Previous research suggests that this program is effective in improving body composition (Brennan, Walkley, Wilks, Fraser, & Greenway, in press). This study aimed to explore the impact of The Choose Health Program on psychosocial wellbeing in overweight and obese adolescents. Central to the study was the intention to investigate both desirable and detrimental impacts on adolescent psychosocial wellbeing. It was hypothesised that participation in the program would result in sustained improvements in psychopathology, psychosocial functioning and family functioning.
Methods
This article reports a randomised controlled trial with a single cross-over design. The full methodology of this trial has been described in detail elsewhere (Brennan et al., Reference Brennan, Walkley, Fraser, Greenway and Wilks2008). The trial is registered with the Australian New Zealand Clinical Trials Registry (ACTRN12610000111077). Ethical approval for the project was obtained from relevant Human Research Ethics Committees. Voluntary written informed consent was obtained from both parent and adolescent participants.
Recruitment
Information about the Choose Health Program was circulated throughout the community, requesting that interested parents call to register. Parents calling about the program were first assessed for eligibility. Adolescents were eligible to participate if they met the following inclusion criteria: (a) aged 11 to 19 years, (b) overweight or obese according to the international cut-off points for BMI in children (Cole, Bellizzi, Flegal, & Dietz, Reference Cole, Bellizzi, Flegal and Dietz2000), and (c) living with a parent or adult caregiver who was prepared to be fully involved in treatment. Participants were excluded from the study if they had a disability or illness that prevented them from participating in the program. Eligible families were provided with a detailed outline of the research procedures and administered a phone-based intake survey collecting basic demographic information. They were then mailed parent and adolescent plain language statements and a consent form and were asked to return this form to participate in the study.
Allocation to Condition
Following completion of pretreatment assessments, participants were randomly allocated to a treatment or wait-list control condition. Participants were asked to select a number from 1 to 3 and this number was compared to an investigator randomly selected number and was then used to allocate the participant to the treatment or control condition. Twice as many participants were allocated to treatment and they commenced treatment immediately. The treatment group was larger than the control group as a secondary aim of the larger study was to explore the impact of a pretreatment motivational interview on CBT treatment outcomes. Of those allocated to treatment, half received a standard assessment interview and half a motivational interview. These two groups did not differ in terms of baseline characteristics, retention or outcomes (Brennan, Reference Brennan, Walkley, Fraser, Greenway and Wilks2008) and thus were combined for all analysis.
Assessments were repeated 6 months after allocation, and wait-listed participants were then offered treatment while those in the treatment condition participated in a maintenance program and were followed up 6 months later. This allowed for comparisons between treatment and control conditions immediately following the treatment period (post), and for an analysis of changes from pre to post to maintenance (maintenance) in the treatment condition.
Assessment
Assessment interviews were conducted prior to commencing the study. All other assessments were completed: (1) after the assessment interview and prior to randomisation; (2) after the completion of the treatment phase of the intervention (or the end of wait-list); and (3) after completion of the maintenance phase of the intervention. All assessments were conducted by trained independent assessors blind to participant's group allocation, treatment adherence, and stage of intervention.
Assessment Interview
Assessment interviews were conducted over a 60-minute period with both the adolescent and the parent present. At the completion of the initial interview, adolescent and parent participants were given the option of continuing with the assessments or terminating their involvement in the project.
Physical and Behavioural Assessment
Adolescent participants completed a range of assessments of body composition and function, and self-monitored their eating and physical activity for 7 days at each assessment period. Details of these assessments are provided elsewhere (Brennan, et al., Reference Brennan, Walkley, Fraser, Greenway and Wilks2008) and results have been reported in a previous publication (Brennan et al., in press).
Psychosocial Assessments
Adolescent and parent participants completed comprehensive survey assessments at each assessment period. All the survey assessments were scored by an independent research assistant who was blind to treatment stage and condition. The Parent and Adolescent Health and Weight History surveys (PAHWH; Brennan, Reference Brennan2006) were used to obtain general demographic information. The PAHWH surveys were designed specifically for the purpose of this study to provide general demographic information such as age and ethnic background, dieting and weight history, mental and physical health history.
Adolescents completed a range of measures of psychosocial functioning and psychopathology. The Rosenberg Self-Esteem Scale (RSE, Rosenberg, Reference Rosenberg1965) was included as the measure of self-esteem. The RSE includes 10 items. Respondents indicate their level of agreement to each item on a 4-point Likert scale ranging from 1 (Strongly agree) to 4 (Strongly disagree). A high score on the RSE indicates a positive self-attitude while a low score indicates a negative attitude towards the self. The RSE has been extensively used in both the adolescent and adult literature. The RSE has acceptable psychometric properties (McCarthy & Hoge, Reference McCarthy and Hoge1982; Rosenberg, Reference Rosenberg1965). The Perceived Social Support Scale (PSS, Procidano & Heller, Reference Procidano and Heller1983) was included as a measure of the individual's perceptions of the adequacy of their social support. The PSS consists of two scales, Social Support from Friends (PSS-Fr), and Social Support From Family (PSS-Fa). Each scale contains 20 statements describing feelings or experiences common in relationships. Respondents answer ‘yes’, ‘no’ or ‘don't know’ to indicate whether they believed each of these forms of support occur in their family and friend relationships. A high score on a PSS scale indicates high perceived social support. This scale has acceptable psychometric properties (Lyons, Perrotta, & Hancher-Kvam, Reference Lyons, Perrotta and Hancher-Kvam1988; Procidano & Heller) and has been widely used in a range of populations including adolescents (e.g., Gavazzi, Reference Gavazzi1994).
The Family and Friend Influence on Health Behaviour Scale (FFHB, based on Redding et al., Reference Redding, Rossi, Pallonen, Prochaska, Abrams, Velicer and Rossi1999) was used as a measure of the involvement of family and friends on adolescents’ adoption of health behaviour. It includes four subscales: Family Support for Healthy Eating (FFHB–FaHE), Family Support for Physical Activity (FFHB–FaPA), Friend Support for Healthy Eating (FFHB–FrHE), and Friend Support for Physical Activity (FFHB–FrPA). Each scale contains four items describing supporting behaviours. Respondents indicate their level of agreement to each item on a 5-point Likert scale ranging from 1 (Almost never) to 4 (Almost always). A high score on the FFHB scales indicates strong support.
General psychopathology was measured using the Depression Anxiety Stress Scale (DASS; Lovibond & Lovibond, Reference Lovibond and Lovibond1995). The DASS is a self-report measure of state depression (DASS–D), anxiety (DASS–A) and stress (DASS–S). Each scale includes 14 items and respondents are asked to use a 4-point severity/frequency scale to rate the extent to which they had experienced each of these states over the past week. A high score on any scale indicates a high prevalence of the core symptoms of the emotional disturbance measured by the scale. The DASS has acceptable psychometric properties (Brown, Chorpita, Korotitsch, & Barlow, Reference Brown, Chorpita, Korotitsch and Barlow1997; Lovibond & Lovibond, Reference Lovibond and Lovibond1995), has been used with both nonclinical and clinical samples (Antony, Bieling, Cox, Enns, & Swinson, Reference Antony, Bieling, Cox, Enns and Swinson1998; Brown et al., Reference Brown, Chorpita, Korotitsch and Barlow1997), and is appropriate for use as a screening tool with adolescents (Lovibond & Lovibond, Reference Lovibond and Lovibond1995).
Eating and weight specific psychopathology was measured using the Adolescent Dieting Scale (ADS; G. C. Patton et al., Reference Patton, Carlin, Shao, Hibbert, Rosier, Selzer and Bowes1997) and the Eating Disorders Inventory- II (EDI-II, Garner, Reference Garner1990). The ADS lists eight common dieting behaviours to which respondents are required to indicate how often they use each of these strategies. High scores on the ADS indicate more dieting behaviours. This survey differs from other dieting measures as it does not include extreme weight control strategies more characteristic of eating disorders, thus it allows for identification of non-clinical dieting weight control behaviours (G. Patton, Selzer, Coffey, Carlin, & Wolfe, Reference Patton, Selzer, Coffey, Carlin and Wolfe1999; Patton et al., Reference Patton, Carlin, Shao, Hibbert, Rosier, Selzer and Bowes1997).
The EDI-II is a measure of the psychological traits or symptoms common in anorexia nervosa and bulimia nervosa. The survey consists of nine scales. Respondents are required to use a 6-point scale, ranging from Always to Never, to respond to 91 items about eating attitudes, feelings and behaviours. The Drive for Thinness scale (EDI-II-DT) assesses excessive concern with dieting, preoccupation with weight, and fear of weight gain. The Bulimia (EDI-II-B) scale measures thoughts about or occurrences of bingeing or uncontrollable overeating. The Body Dissatisfaction scale (EDI-II-BD) provides a measure of dissatisfaction with overall body shape or specific body parts. The Ineffectiveness scale (EDI-II-I) measures feelings of aloneness, emptiness, worthlessness, and inability to control one's life. The Perfectionism scale (EDI-II-P) assesses the belief that only the achievement of perfect standards is acceptable and that the respondent must perform better than others. The Interpersonal Distrust (EDI-II-ID) scale measures the respondent's hesitance to share thoughts and feelings with others. The Interoceptive Awareness scale (EDI-II-IA) assesses lack of understanding and correct response to internal states. The Maturity Fears scale (EDI-II-MF) assesses the fear of adult physical and psychological experiences. The Asceticism scale (EDI-II-A) measures the perceived virtue of self-denial and self-restraint. The Impulse Regulation scale (EDI-II-IR) assesses poor impulse regulation resulting in recklessness, hostility and destructiveness. The Social Insecurity scale (EDI-II-SI) measures beliefs that social relationships are disappointing, unrewarding and insecure. Scale scores are obtained by adding the scores of all the items for the scale, and higher scale scores indicate higher levels of the variable being measured. The EDI-II has acceptable psychometric properties (Garner, Reference Garner1990) and has been used with both obese (Chandarana, Holliday, Conlon, & Deslippe, Reference Chandarana, Holliday, Conlon and Deslippe1988; Garner, Olmstead, & Polivy, Reference Garner, Olmstead and Polivy1983) and adolescent populations (Rosen, Silberg, & Gross, Reference Rosen, Silberg and Gross1988; Shore & Porter, Reference Shore and Porter1990).
The Automatic Thoughts Questionnaire (ATQ; Hollon & Kendall, Reference Hollon and Kendall1980) was included as a general measure of negative cognitions. This survey includes measures of personal maladjustment and desire for change, negative self-concept and expectations, and helplessness. The ATQ consists of 30 items answered on a 5-point scale, indicating the frequency of each thought in the past week. Scores on the survey have been shown to decrease following cognitive therapy. The ATQ has acceptable psychometric properties (Hollon & Kendall).
Both adolescents and parents completed social and family functioning measures. Assessment of social skills was conducted using the pupil and parent versions of the Social Skills Questionnaire (SSQ; Spence, Reference Spence1995) and the Social Competence Questionnaire (SCQ; Spence). The SSQ consists of 30 items and respondents are asked to answer Not true, Sometimes true or Mostly true to each item. The SCQ consists of 10 items focusing on the consequences of social interactions. Respondents are asked to answer Not true, Sometimes true or Mostly true to each item. These surveys have adequate psychometric properties (Spence, Reference Spence1995).
Family interaction and communication was measured using the Parent Adolescent Communication Scale (PAC; H. Barnes & Olsen, Reference Barnes, Olsen, Olson, McCubbin, Barnes, Larsen, Muxen and Wilson1982) and the Family Problem Solving Communication Index (FPSC; McCubbin, McCubbin, & Thompson, Reference McCubbin, McCubbin, Thompson, McCubbin, Thompson and McCubbin1988). The PAC is a 20-item, true/false, adolescent- and parent-rated survey, designed to assess communication between parent and adolescents. The survey provides measures of open communication (PAC-O) and problems in communication (PAC-P). Ten items contribute to each scale, with high scores indicating positive communication patterns. The PAC-O provides an indication of the freedom of expression, understanding and satisfaction during family communications. The PAC-P focuses on negative interaction styles and constraints in communication. The survey has acceptable psychometric properties (Barnes & Olsen, Reference Barnes, Olsen, Olson, McCubbin, Barnes, Larsen, Muxen and Wilson1982, Reference Barnes and Olson1985; Jackson, Bijstra, Oostra, & Bosma, Reference Jackson, Bijstra, Oostra and Bosma1998). The FPSC was included as a measure of problem solving and coping during family interactions. It has two scales measuring incendiary (FPSC-I) and affirming (FPSC-A) communication. A high score on the FPSC-I indicates high levels of problematic communication styles that inflame a difficult interaction. A high score on the FPSC-A indicates high levels of supportive communication styles that calm a difficult interaction. The FPSC has acceptable psychometric properties (Leske & Jiricka, Reference Leske and Jiricka1998; McCubbin et al., Reference McCubbin, McCubbin, Thompson, McCubbin, Thompson and McCubbin1988).
Parents also completed the Parenting Scale (PS; Arnold, O'Leary, Wolf, & Acker, Reference Arnold, O'Leary, Wolf and Acker1993; Irvine, Biglan, Smolkowski, & Ary, Reference Irvine, Biglan, Smolkowski and Ary1999), a measure of parenting practices. The PS yields a total score and three factors: laxness (PS-L), over-reactivity (PS-O), and verbosity (PS-V) This 30-item survey is included so the effectiveness of parent training can be examined. Respondents use a 7-point scale to indicate their level of agreement with each of the double-anchored statements. High scores indicate high levels of parenting practice. This scale has been adapted for use with adolescents and has adequate psychometric properties when used with this population (Irvine et al.).
Treatment Acceptability
Adolescent and parent consumer satisfaction were measured at the completion of the treatment phase of the intervention using the Choose Health Consumer Satisfaction Survey designed for the current study. Respondents rated the level of overall progress on 5-point rating scale ranging from 1 (Problems got worse), to 5 (Great deal of progress). They were also asked to rate the program on a number of items measuring satisfaction with the program and the clinician. Scores can range from 1 (Strongly disagree) to 5 (Strongly agree) with a higher score indicating higher treatment acceptability.
Intervention
The Choose Health Program was developed for the purpose of this study. It aims to facilitate small and sustainable lifestyle changes to improve eating and physical activity habits. Cognitive behavioural principles are used to assist the adolescent and their family to manage the environmental, social and psychological barriers to change with the aim of promoting maintained improvements and positive psychosocial wellbeing. The nutritional component of the intervention aimed to promote eating habits consistent with the Australian Guide to Healthy Eating (Smith, Kellett, & Schmerlaib, Reference Smith, Kellett and Schmerlaib1998), which recommends the consumption of a variety of foods from each of the five food groups (cereals, vegetables, fruit, diary and meat products and alternatives), the selection of low fat alternatives, and the consumption of water. The physical activity component of the intervention aimed to promote physical activity habits consistent with the Australian Physical Activity Guidelines for Children and Young People (Department of Health and Ageing, 2004a), which recommends adolescents achieve at least 60 minutes of moderate to vigorous physical activity per day and spend no more than 2 hours per day in noneducational screen activities. Rather than promote rapid unsustainable changes, the program aimed to instigate small maintainable improvements in nutritional intake and physical activity habits to promote sustained changes in body composition, cardiovascular fitness, and psychosocial wellbeing (Brennan et al., Reference Brennan, Walkley, Fraser, Greenway and Wilks2008).
Treatment was conducted on an individual basis. The intervention program consisted of a treatment and a maintenance phase. The treatment phase consisted of twelve 60-minute face-to-face sessions and one phone call session. The first 10 treatment sessions were conducted weekly. The remaining sessions were conducted every second week. Both parents and adolescents were required to attend the first six treatment sessions. Adolescents were then given the choice of attending the remaining sessions alone, or with the support of a parent.
The maintenance phase consisted of two 60-minute maintenance clinic sessions and seven 15-minute maintenance phone call sessions. Following the last treatment session, phone call sessions were completed every second week and a face-to-face session was scheduled 3 months after the last treatment session. This was followed by monthly phone call sessions and a final face-to-face session conducted 6 months after the last treatment sessions.
Treatment was provided by a single clinician, the developer of the program. The clinician based each session on a structured session outline, which listed each of the topics to be discussed and information about how to complete the tasks. Table 1 briefly outlines the content of each treatment session. Parents and adolescents also received a program workbook, which included information, tasks and home practice activities for each treatment session. At the end of each session, adolescents and parents recorded their self-selected behaviour change goals. Adolescents were encouraged to set goals relating to their own behaviour change, and parents were encouraged to set goals relating to them assisting their adolescent to change their behaviour. These behaviour change goals were used to individualise treatment, ensuring it was appropriate for the adolescents’ lifestyle and developmental needs (Brennan et al., Reference Brennan, Walkley, Lukeis, Risteska, Archer, Digre and Greenway2009). Between sessions, adolescents were encouraged to monitor their behaviour change goals and record their eating and physical activity habits. See Brennan et al. (Reference Brennan, Walkley, Fraser, Greenway and Wilks2008) for a more detailed description of this intervention.
Choose Health Treatment and Maintenance Session Outlines
Note: PC = Phone Call Session
Adapted from Brennan, L., Walkley, J., Fraser, S., F, Greenway, K., & Wilks, R. (Reference Brennan, Walkley, Fraser, Greenway and Wilks2008). Motivational interviewing and cognitive behaviour therapy in the treatment of adolescent overweight and obesity: Study design and methodology. Contemporary Clinical Trials, 29, 359–375. Reprinted with permission.
Statistical Analyses
The limited related research, failure to report effect sizes, and huge variation in the methodologies of studies in this area at the time of the study (Summerbell et al., Reference Summerbell, Achton, Campbell, Edmunds, Kelly and Waters2003) prevented valid a priori sample size requirement calculations. The sample size used in the current study was determined by resources availability. Based on the number of participants randomised, post-hoc power analysis indicated that the study had 49% chance of a medium effect (f = .25)(Faul, Erdfelder, Lang, & Buchner, Reference Faul, Erdfelder, Lang and Buchner2007). Body composition, as measured by DEXA, was the primary outcome for this study and results have been reported previously (Brennan, et al., under review). The psychopathology, psychosocial and family functioning variables reported in this article are secondary outcomes therefore no alpha adjustments have been made to account for multiple comparisons.
All analyses were preceded by data cleaning and checking. Both completer and intention-to-treat analyses were conducted. The first value carried forward method was used to impute missing data for intention-to-treat analysis. Treatment efficacy was assessed by comparing outcomes of the control and treatment conditions post treatment using between subjects Analysis of Covariance (ANCOVA); pretreatment data was the covariate and condition (treatment, control) the factor. For completer analyses, follow-up analyses were conducted to determine statistically significant differences across assessment periods within the treatment conditions where post-treatment group differences were evident. Repeated measures Analyses of Variance, with time (pre-, postmaintenance) as the within subjects factor, and subsequent post hoc analysis were used to assess maintenance. These follow-up analyses were conducted for all intention-to-treat analyses.
Results
Participant Characteristics
The sample comprised 29 male and 34 female adolescents classified as overweight (n = 15) or obese (n = 48) according to the international reference standard BMI cut-off points (Cole, et al., Reference Cole, Bellizzi, Flegal and Dietz2000). Adolescent participants were aged 11.5 to 18.9 years (M = 14.3, SD = 1.9), and parent participants were aged 28 to 61 years (M = 44.8, SD = 5.0). Ninety-four per cent of parent respondents were biological mothers, 5% were biological fathers, and 2% were step-mothers. Seventy-three per cent of parents (n = 43) provided height and weight information. Their BMI ranged from 20.5 kg/m2 to 52.7 kg/m2 (M = 28.5, SD = 6.3); 35% were overweight and 33% were obese. A more detailed description of participant characteristics is provided elsewhere (Brennan et al., Reference Brennan, Walkley, Fraser, Greenway and Wilks2008). All participants were asked to complete all psychosocial assessments; however, there is missing data due to incomplete completion and/or nonreturn of questionnaires.
Forty-two participants were allocated to the treatment condition and 21 to the wait-list control condition. Fourteen (67%) wait-listed participants completed the 6-month wait-list period. Twenty (48%) completed the treatment phase and 17 (40%) completed the maintenance phase (Figure 1). Six (10%) adolescents did not complete the intervention/wait-list as they were referred for clinically significant pre-existing depression, anxiety, disordered eating and/or family difficulties identified during assessment and/or intervention.
Participant flow through the trial.
Participants in the two conditions did not differ significantly in terms of baseline adolescent (gender, age, pubertal status, weight status, ethnicity), parent (relationship with child, age, BMI, ethnicity, education, employment status, hours of employment), or family (family status, partner employment status, hours of employment, education, family income) characteristics (p > .05). The treatment and control condition differed significantly on a number of outcome variables (Table 2), thus an ANCOVA was used to control for pre-treatment differences when comparing outcomes of the control and treatment conditions.
Psychopathology, Psychosocial and Family Functioning Variables at Baseline for Treatment and Control Condition
Note: Eating Disorders Inventory – 2 (EDI-II).
Adolescent Completed Measures
Results of completer analyses are described below. Similar results were obtained with intention-to-treat analysis displayed in Table 3.
Psychopathology, Psychosocial and Family Functioning Variables Post-Treatment and Maintenance Intention to Treat Descriptives and Analyses
Note: Eating Disorders Inventory — 2 (EDI-II).
Psychopathology
After controlling for pretreatment differences, those in the intervention condition obtained significantly increased scores on the EDI-II-IR Impulse Regulation subscale, F(1,29) = 13.43, p = .001, partial η2 < .33, following treatment relative to the control condition. However, analysis of pre, post and maintenance data for those in the treatment condition indicated that the effect of treatment phase was not significant for the EDI-II-IR, Wilks’ λ = .99, F(2, 12) = 0.05, p = .995, partial η2 < .01. These results are displayed in Figure 2.
Eating Disorders Inventory-2 Impulse Regulation post-treatment and maintenance outcomes.
After controlling for pretreatment differences, the treatment and control condition did not differ in terms of obtained scores on any other EDI-II subscales following treatment; EDI-II-DT, F(1,29) = 0.08, p = .777, partial η2 < .01; EDI-II- B, F(1,29) = 2.40, p = .133, partial η2 = .08; EDI-II-BD, F(1,29) = 0.13, p = .726, partial η2 < .01; EDI-II-I, F(1,29) = 1.34, p = .257, partial η2 = .05; EDI-II-P, F(1,29) < 0.01, p = .981, partial η2 < .01; EDI-II-ID, F(1,29) = 0.06, p = .803, partial η2 < .01; EDI-II-IA, F(1,29) = 0.43, p = .517, partial η2 = .02; EDI-II- MF; F(1,29) = 1.42, p = .244, partial η2 = .05; EDI-II-A, F(1,29) = 0.01, p = .930, partial η2 < .01; EDI-II-SI, F(1,29) = 0.02, p = .904, partial η2 < .01.
After controlling for pretreatment differences, those in the intervention condition obtained significantly increased scores on the ADS, F(1,29) = 5381, p = .023, partial η2 = .12, following treatment relative to the control condition. Analysis of pre-, post- and maintenance data for those in the treatment condition indicated that weight control behaviour increased significantly from pre- to post-treatment and remained higher at maintenance, Wilks’ λ = 0.23, F(2, 10) = 16.67, p < .001, partial η2 = .77. These results are displayed in Figure 3.
Adolescent Dieting Scale post-treatment and maintenance outcomes.
After controlling for pretreatment differences, the treatment and control condition did not differ in terms of obtained scores on any of the remaining psychopathology measures following treatment; ATQ, F(1,30) = 0.12, p = .732, partial η2 < .01; DASS-D, F(1,31) = 0.33, p = 0.573, partial η2 = .01; DASS-A, F(1,31) = 2.66, p = .114, partial η2 = .08; DASS-S, F(1,31) = 3.51, p = .071, partial η2 = .11.
Psychosocial Functioning
After controlling for pre-treatment differences, those in the intervention condition obtained significantly increased scores on the PSS-Fa scale, F(1,29) = 10.96, p = .003, partial η2 = .30, following treatment relative to the control condition. Analysis of pre, post and maintenance data for those in the treatment condition indicated that the effect of treatment phase was not significant for the PSS-Fa, Wilks’ λ = .61, F(2, 11) = 3.59, p = .063, partial η2 = .40. These results are displayed in Figure 4.
Perceived Social Support from Family Scale post-treatment and maintenance outcomes.
After controlling for pretreatment differences, the treatment and control condition did not differ in terms of obtained scores on any of the remaining psychosocial measures following treatment; RSE, F(1, 31) = 2.97, p = .096, partial η2 = .09; PSS-Fr, F(1,29) = 0.02, p = .908, partial η2 < .01; FFHB-FaHE, F(1, 30) = 0.45, p = .508, partial η2 = .02; or FFHB-FaPA, F(1,30) = 0.72, p = .402, partial η2 = .03; FFHB-FrHE, F(1, 29) = 0.36, p = .556, partial η2 = .01; FFHB-FrPA, F(1,29) = 1.21, p = .281, partial η2 = .04; SSQ, F(1,30) = 0.50, p = .487, partial η2 = .02; or SCQ, F(1,30) = 0.02, p = .902, partial η2 < .01.
Family Functioning
After controlling for pretreatment differences, those in the intervention condition obtained significantly increased scores on the PAC-P, F(1,29) = 5.45, p = .027, partial η2 = .17, following treatment relative to the control condition. Analysis of pre, post and maintenance data for those in the treatment condition indicated that the effect of treatment phase was not significant for the PAC-P, Wilks’ λ = .96, F(2, 11) = 0.23, p = .796, partial η2 = .04. These results are displayed in Figure 5.
Parent Adolescent Communication Problem Scale post-treatment and maintenance outcomes.
After controlling for pretreatment differences, the treatment and control condition did not differ in terms of obtained scores on the remaining family functioning measures following treatment; PAC-O, F(1,29) = 2.31, p = .140, partial η2 = .08; FPSC-A, F(1,28) = 3.25, p = .083, partial η2 = .11; FPSC-I, F(1,28) = 2.05, p = .164, partial η2 = .07.
Parent Completed Measures
Adolescent Psychosocial Functioning
After controlling for pretreatment differences, parents in the treatment and control condition did not differ in terms of obtained scores on any parent reported measures of adolescent psychosocial functioning following treatment; the SSQ, F(1,32) = 0.75, p = .394, partial η2 = .02; SCQ, F(1,32) = 7.51, p = .344, partial η2 = .03; FFHB-FaHE, F(1,32) = 0.25, p = .619, partial η2 = .01; FFHB-FaPA, F(1,31) = 1.23, p = .276, partial η2 = .04.
Family Functioning
After controlling for pretreatment differences, parents in the treatment and control condition did not differ in terms of obtained scores on and parent reported measures of family functioning following treatment; PAC-O, F(1,32) = 0.28, p = .602, partial η2 = .01; PAC-P, F(1,32) = 0.03, p = .877, partial η2 < .01; FPSC-A, F(1,32) = 0.23, p = .634, partial η2 = .01; FPSC-I, F(1,32) = 0.21, p = .655, partial η2 = .01; PS-L, F(1,30) = 0.20, p = .657, partial η2 = .01; PS-O, F(1,30) = 0.18, p = .672, partial η2 = .01.
Treatment Acceptability
Consumer satisfaction questionnaires were returned by 71% of adolescent and 50% of parent intervention completers. All respondents indicated that they made some progress as a result of participating in the program. Adolescents reported an average overall progress rating of 4.06 (0.68) with 19% reporting some progress, 56% reporting considerable progress, and 25% reporting a great deal of progress. Parents reported an average overall progress rating of 3.75 (0.87) with 50% reporting some progress, 25% reporting considerable progress and 25% reporting a great deal of progress. No adolescents or parents reported that the problem did not change or that it got worse. Descriptive statistics for parent and adolescent responses on individual items are displayed in Table 4.
Descriptive Statistics for Consumer Satisfaction Items
Discussion
This study aimed to explore the impact of the Choose Health Program on psychosocial wellbeing in overweight and obese adolescents. Psychosocial outcomes were of interest for two reasons. The first was to ensure that treatment did not have a detrimental effect on adolescent or parent psychosocial wellbeing, and the second was to assess the impact of treatment on promoting improved psychosocial functioning in treatment seeking overweight and obese adolescents.
Treatment did not impact on adolescent depression, anxiety, or stress. This finding indicates that treating adolescent overweight and obesity in a supportive setting does not result in increased adolescent general psychopathology. While adolescent psychopathology was not significantly influenced by treatment participation, the measurement of these factors in adolescent weight loss interventions continues to be important. Six adolescents (10%) in the treatment sample were referred to specialist services for assistance in the management of anxiety, depression, disordered eating and/or family difficulties. This figure is considerably higher than the 3.6% of Australian adolescents identified as having clinical levels of psychopathology (Sawyer et al., Reference Sawyer, Arney, Baghurst, Clark, Graetz, Kosky and Zubrick2000). This outcome highlights the importance of assessment and detection of psychopathology in adolescents participating in weight loss interventions as treatment outcomes may be improved if deficits can be identified and interventions can be modified or supplemented to address these barriers. Future research is needed to determine the optimum combination of interventions for adolescents requiring weight loss and psychological intervention. It may be that psychopathology needs to be addressed prior to weight loss intervention, or that the two can be addressed separately but concurrently, or that a combined weight loss and psychopathology intervention is most effective. Future research should also explore the usefulness of psychosocial wellbeing in predicting adolescent overweight and obesity treatment completion, adherence and outcomes.
Treatment resulted in increased weight control behaviour, which was sustained at maintenance. Weight control behaviour among normal weight adolescents is linked to increased risk of psychopathology (Pesa, Reference Pesa1999). However, non-extreme weight control behaviours is recommended for overweight and obese adolescents and is not associated with negative outcomes (Reilly & Wilson, Reference Reilly and Wilson2006). This is supported by findings of the current study, indicating that disordered eating symptomatology did not increase as a result of treatment, despite increases in weight control behaviours.
Food-related disinhibition and poor impulse control are more prevalent among the overweight and obese (Gallant et al., Reference Gallant, Tremblay, Pérusse, Bouchard, Després and Drapeau2010), and are associated with poorer treatment outcomes (Dalle Grave, Calugi, Corica, Di Domizio, & Marchesini, Reference Dalle Grave, Calugi, Corica, Di Domizio and Marchesini2009). Thus, improvements in impulse regulation evident in the current study are promising. There were no other differences in eating disorder symptomatology, namely, drive for thinness, body dissatisfaction, bulimia, ineffectiveness, perfectionism, interpersonal distrust, interpersonal awareness, maturity fears, asceticism, or social insecurity. Given the high prevalence of disordered eating in the overweight and obese, and the associated negative psychosocial consequences, future research should explore disordered eating in overweight and obese adolescents, particularly in those seeking treatment.
These findings indicate that participation in a weight loss intervention did not increase disordered eating in overweight and obese adolescents. These results are consistent with adult research, indicating that comprehensive multifaceted weight loss interventions do not lead to increases in disordered eating, and may in fact reduce eating psychopathology (Wadden, Womble, Stunkard, & Anderson, Reference Wadden, Womble, Stunkard, Anderson, Wadden and Stunkard2002; Yanovski, Reference Yanovski, Fairburn and Brownell2002). This finding has important implications for health promotion as many parents in the current program reported that fears of causing disordered eating had prevented or delayed them seeking weight loss intervention for their adolescent.
Adolescents in the treatment and control conditions did not differ in terms of self-esteem. Similarly, treatment did not result in changes in adolescent reported family and friend support for healthy eating and physical activity or parent reported family support for healthy eating and physical activity. Improvements were not evident in parent or adolescent reported social skills or social competence. However, adolescents participating in the intervention reported significant improvements in the social support they received from their family. These findings suggest that providing parents with specific strategies to support adolescent behaviour change improves their ability to provide more general social support to their adolescent. Perceived family support has been identified as a protective factor for adolescents (Department of Health and Ageing, 2004b; Sawyer et al., Reference Sawyer, Arney, Baghurst, Clark, Graetz, Kosky and Zubrick2000) and is associated with adolescent coping and wellbeing (Jackson et al., Reference Jackson, Bijstra, Oostra and Bosma1998), and is therefore likely to be associated with improved treatment outcomes and psychosocial wellbeing.
Treatment resulted in statistically significant improvements in adolescent reported problem communication. This finding suggests that joint parent and adolescent participation in an adolescent weight loss intervention may result in improved parent adolescent communication and problem solving from the adolescent's perspective. This is consistent with the previously discussed finding that adolescents receiving treatment perceived improvements in social support from their family. Improved communication and problem solving may be expected as parents and adolescents negotiate the individual and family changes required to improve eating and physical activity habits with the support of the clinician. Given the importance of parents in supporting adolescent behaviour change (Dishion & McMahon, Reference Dishion and McMahon1998; Sanders, Reference Sanders1999), adolescent weight loss treatments may be improved by the inclusion of strategies to improve parent–adolescent communication.
All respondents returning consumer satisfaction questionnaires indicated that they made some progress, and the majority made considerable progress or a great deal of progress as a result of participating in the program. Individual ratings of aspects of the treatment program and the treating clinician were also high. These results indicate that both adolescents and parents who completed the Choose Health Program found it highly acceptable. A number of adolescents and parents did not return consumer satisfaction questionnaires. It may be that these participants were less satisfied with the intervention, therefore future research should attempt to improve response rates to consumer satisfaction questionnaires.
While the current study addresses many limitations of previous research, several additional methodological issues are acknowledged. The requirement that participants allocated to the wait-list condition be offered treatment immediately following the treatment period prevented long-term comparisons between the treatment and control conditions. Measurement of long-term outcomes of overweight and obesity treatment programs is essential as maintenance of treatment effects is generally poor (Oude Luttikhuis et al., Reference Oude Luttikhuis, Baur, Jansen, Shrewsbury, O'Malley, Stolk and Summerbell2008). Future research must consider alternative techniques to allow for evaluation of the long-term effects of adolescent weight loss interventions.
As the sample was recruited from the community, results may not generalise to adolescents seeking treatment in medical or community health settings. Research suggests that adolescent and parents recruited for research have fewer comorbidities than those clinically referred for treatment (Kazdin, Reference Kazdin2003). It is possible that the noncompleters differed from the completers in important ways, and consequently results are not representative of those obtained had the entire sample completed all assessments. Results from completer analysis were, however, consistent with intention-to-treat results, providing support for the validity of completer analyses.
Limited data were available to inform an a priori power analysis for the current study. While the study was sufficiently powered to detect treatment effects on primary outcome variable body composition (Brennan et al., Reference Brennan, Walkley, Fraser, Greenway and Wilks2008) it was likely insufficiently powered to detect treatment impact on psychosocial factors as evidenced by a number of nonsignificant positive trends. The drop-out rate, combined with the incomplete questionnaire data, further reduced the sample size and statistical power of the current study. Future studies should explore psychosocial factors in larger samples. Future research should also include more comprehensive measurement of weight specific psychopathology and consider the impact of intervention on these weight specific difficulties. Methods to improve retention and promote more complete data completion are also required.
This study aimed to explore the impact of the Choose Health Program on psychosocial wellbeing in overweight and obese adolescents. Treatment did not have detrimental effects on psychopathology, psychosocial functioning, or family functioning. Parents and professionals can be assured that a comprehensive, multifaceted parent-supported intervention for overweight and obese adolescents does not cause psychological harm. Further, treatment resulted in significant improvements in impulse regulation, a factor implicated in the development of eating disorders. Adolescents receiving treatment reported improvements in social support from their family, and there was some evidence of improved parent-adolescent communication and problem solving.
Disclosures
LB developed and delivered the Choose Health program and was responsible for design and conceptualisation of the study, data analysis and interpretation and manuscript preparation.
Acknowledgments
This study was conducted as part of LB's PhD program. LB developed and delivered the intervention; she was responsible for design and conceptualisation of the study, data analysis and interpretation and manuscript preparation. RW and JW were LB's research supervisors and provided feedback on the manuscript. LB was funded by a RMIT Postgraduate Award and a VicHealth Postgraduate Scholarship. Thank you to Kylie Murphy for reviewing and editing the final version of the manuscript.
