OBJECTIVE: Childhood lifestyle interventions usually involve the families. However, knowledge of family characteristics that promote or constrain a child's weight-reduction outcome is limited. Candidates for such factors might be family characteristics that have proven to be associated with social adjustment (development) in childhood. Thus, we analyzed whether family adversity, maternal depression, and attachment insecurity predict long-term success in children's weight reduction.
PATIENTS AND METHODS: Participants in the study were 111 parent-child dyads with overweight and obese children/adolescents (BMI mean: 29.07 [SD: 4.7] [range: 21.4–44.9]; BMI SD score mean: 2.43 [SD: 0.44] [range: 1.31–3.54]) aged between 7 and 15 years. The families took part in a best-practice lifestyle intervention of 1 year's duration. A longitudinal analysis with 3 assessment waves (baseline, conclusion, 1-year follow-up) was conducted.
RESULTS: Hierarchical regression analyses revealed that long-term success (at least 5% weight reduction by the 1-year follow-up) versus failure (dropping out or less weight reduction) was significantly predicted by the set of psychosocial variables (family adversity, maternal depression, and attachment insecurity) when we controlled for familial obesity, preintervention overweight, age, and gender of the index child and parental educational level. Maternal depression proved to be the best predictor. Maintenance of weight reduction between the conclusion of the program and the 1-year-follow-up was also predicted by the set of psychosocial variables. Maternal insecure-anxious attachment attitudes best predicted this criterion.
CONCLUSIONS: Although cross-validation is required, our results are the first evidence for proximal and distal family characteristics linked to long-term weight-reduction outcomes. The results suggest a need to create tailored intervention modules that address the difficulties of these families.
WHAT'S KNOWN ON THIS SUBJECT:
Lifestyle interventions for overweight children/adolescents have proven effective. However, the small effect sizes indicate that this therapeutic strategy does not fit all children. Knowledge is scarce on the characteristics of those children who do not respond.
WHAT THIS STUDY ADDS:
The results of this study reveal that maternal insecure attachment attitudes, maternal depression, and psychosocial family risks predict long-term failure in weight reduction. The results provide a basis for the design of more tailored therapeutic approaches to reduce the rates of therapeutic failure.
Pediatric overweight and obesity have dramatically increased over the years, and current prevalence rates have reached alarming proportions.1,2 Because of the impact the condition exerts on physical and psychosocial health,3,–,5 effective treatment is urgently needed. Research results have shown that family based treatment programs that combine physical activity, dietary, and behavior therapy components effectively lead to reduced overweight in children and adolescents.6 In addition to short-term effects, long-term continuance of weight reduction was demonstrated.7,8 However, although the results were clinically relevant, the effect sizes of the lifestyle intervention programs were rather small. A relatively high number of participants thus failed to reduce overweight or could not sustain weight reduction.6 This situation calls for efforts to develop better-tailored therapeutic approaches through definition of distinct subgroups of participants whose response can be predicted with high probability. Those participants who probably will not respond should be offered additional or other interventions to prevent the discouraging and potentially harmful experience of failure in weight reduction.
Research on baseline predictors of success in obesity treatment has consistently revealed that children from “obese families” (ie, with the presence of obesity in parents/siblings) show less weight reduction in the lifestyle intervention programs.9,–,12 This association is possibly mediated by genetic factors. Psychosocial factors, however, might also contribute and moreover might be independently predictive of a child's success or failure in weight reduction. However, although most lifestyle interventions for children and adolescents involve the families, knowledge of familial characteristics that promote or constrain weight reduction is limited.6
Familial factors that heighten the risk of failure to respond to intervention might be those that generally compromise social adjustment (development) in childhood and adolescence. Researchers in this field have identified specific psychosocial risk factors. Depression and attachment insecurity of the primary caregiver and more distal family adversity factors (such as incomplete schooling or vocational training of parents, high person-to-room ratio, early parenthood, and broken-home history of parents) were found to best predict inadequate parenting13,14 and precede the development of a child's low compliance with parents, low effortful control, and behavior problems.13,15,–,17 These psychosocial familial characteristics might also constrain the transfer of program contents into everyday family life and the maintenance of modified behaviors after the conclusion of the programs. Indeed, results of previous studies have shown that depressive or psychopathological symptoms in the primary caregiver predict less weight reduction within the course of an intervention.9,18,19 Nothing is known, however, of psychosocial factors that promote or constrain long-term outcomes of the interventions. In the present study we aimed to identify predictors of long-term success in a family-based lifestyle intervention for overweight and obese children. In particular, we sought to determine whether family adversity, maternal depression, and attachment insecurity are predictive for long-term success in weight reduction. We expected that preintervention family adversity factors, maternal depression, and attachment insecurity would predict poor long-term weight-reduction outcomes, and we explored the extent to which the prediction is independent from familial obesity.
The study sample consisted of 111 parent-child/adolescent dyads with children/adolescents aged 7 to 15 years (mean: 11.5 [SD: 1.84] years). Inclusion criteria were (1) BMI > 97th percentile for age and gender or BMI > 90th percentile20 with the presence of additional risk factors or diseases (eg, hypertension, dyslipidemia, diabetes, orthopedic problems), (2) referral for weight-reduction treatment by a local pediatrician, and (3) attendance at a regular school.
To analyze whether baseline psychosocial variables can be used to predict weight change up to a 12-month follow-up examination in children and adolescents who attend a “best-practice” routine-care lifestyle intervention, we conducted a longitudinal analysis with 3 assessment waves: at baseline (T0: within 3 weeks before the start of the intervention) body weight and height of participants and family members and the psychosocial family characteristics were assessed; at the conclusion of the program (T1: 1 year after T0) and 1 year after conclusion (T2: 2 years after T0), body weights and heights of participants were reassessed.
Of the 111 children and adolescents who started the intervention, 95 (85.6%) completed the 12-month program and took part in the T1 assessment wave. Of these children, 78 (82.1%) attended the 1-year follow-up (T2). Thus, the dropout rate from the intervention was 14.4% (n = 16), and an additional 15.3% (n = 17) of study participants were lost to follow-up at the 24-month examination. These rates correspond to those of other lifestyle interventions.6
The outpatient lifestyle intervention Fit Kids for overweight and obese children and adolescents is a routine care program (certified by the German Obesity Association) that is based on the evidence provided by Summerbell et al,21 the outpatient lifestyle intervention program by Reinehr et al,22 and the guidelines of the German Nutrition Society. The program has been described in detail elsewhere.9 In short, the program combines behavioral therapy for eating and physical activity behaviors, physical exercise, and dietary courses. The families are continuously involved. The program consists of 2 phases: a first intervention phase of 3 months' duration comprising (1) behavioral therapy sessions, (2) a dietary training course, (3) a parents' course to facilitate transfer into everyday family life, and (4) a physical exercise course for the children and adolescents. Each course is 90 minutes in duration and takes place biweekly. During this 3-month intervention phase the families visit the hospital-based program once a week for two 90-minute sessions conducted in succession. The second repetition and maintenance phase is of 9 months' duration. This phase consists of the weekly physical activity course and a monthly parent group.
The study protocol was approved by the ethics committee of the University Medical Centre, Giessen. Children and parents agreed to participate in the study. Parents gave their informed, written consent.
Overweight at Baseline and Weight (Reduction) Outcomes
At the 3 assessment waves (T0, T1, and T2), body weight and height of the child were measured by using the same calibrated scale, a Seca 701 digital column scale (Seca, Hamburg, Germany) and a wall-mounted Seca 222 stadiometer. Children were weighed in their underwear. BMI was calculated. On the basis of German reference data for children,20 BMIs were transformed into SD scores (SDSs) by using the least-mean-square method by Cole et al,23 which normalizes the resulting distribution.
At baseline (T0), after intervention (T1), and at the 1-year follow-up examination (T2) mean BMIs were 29.07 (SD: 4.7; n = 111), 27.94 (SD: 4.7; n = 95), and 29.22 (SD: 5.1; n = 78); and mean BMI SDSs were 2.43 (SD: 0.44; n = 111), 2.13 (SD: 0.60; n = 95), and 2.16 (SD: 0.68; n = 78). Thus, in the 1 year of the intervention (T0 to T1), BMI SDS decreased on average by −0.30 (SD: 0.36 [95% confidence interval: −0.23 to −0.37]). In the 78 children who took part in the follow-up examination, BMI SDSs increased slightly by 0.05 (SD: 0.32 [95% confidence interval: −0.02 to 0.12]) in the 1 year after conclusion of the program (T1 to T2). These rates are comparable to those observed for other reported lifestyle interventions that led to changes in BMI SDS.8,24,–,26
In correspondence with other investigators,10,26 we defined long-term success as a reduction of at least 5% in BMI SDS between baseline (T0) and the 1-year follow-up (T2) and failure as a reduction of <5% in BMI SDS or dropping out of the program prematurely (families that dropped out from intervention were not followed up). Of the 16 children who dropped out (after 3–10 months), 8 children/adolescents had gained up to 4 kg, 7 showed no change, and only 1 child had lost weight. Thus, 41 children/adolescents showed long-term success and 53 children failed to reach the criterion of >5% BMI-SDS reduction (n = 37) or dropped out of the intervention (n = 16).
In addition to this long-term success criterion, we used a second criterion to capture change in BMI SDS specifically in the 12 months after the conclusion of the intervention (difference in BMI SDS between T1 and T2). We thus analyzed whether the psychosocial variables were predictive for long-term success versus failure (T0 to T2) and change in BMI SDS between T1 and T2.
Obesity of Family Members
In a structured interview, parents were asked for weight and height of all household members. On the basis of this information, we identified obesity in parents (BMI ≥ 30) and siblings (BMI ≥ 95th age- and gender-related percentile), which led to 3 dichotomous variables.
Family Adversity Factors
During the structured interview conducted with a parent at baseline, information on educational level of parents and family adversity characteristics was gathered. The Psychosocial Risk Index by Laucht et al15 was used. This index is based on 8 adverse family characteristics: parent without educational qualifications and/or skilled job training; less than 1 room per person in the home; parental psychiatric disorder; parental broken-home history; marital discord (low quality of partnership); parent younger than 19 years at the child's birth; 1-parent family; and severe, chronic life difficulties (eg, chronic disease, financial problems). The family adversity score consists of the number of risks present in the family. High construct validity of the index has been demonstrated by Laucht et al.15 In the present sample, 42 (31%) of the families showed 0 risk factors, 32 (24%) showed 1, 33 (24%) showed 2, and 29 (21%) showed ≥3 risk factors. In accordance with Laucht et al,27 we dichotomized the distribution into low family adversity (0 or 1 risk factor present) and high family adversity (≥2 risk factors).
Depression in the Primary Caregiver
The German version of the Center for Epidemiology Studies Depression Scale28 was applied to test for maternal depression. The German version has been evaluated on the basis of a representative sample of almost 1300 adults and >200 patients with psychiatric disorders. Good internal consistency (Cronbach's α = 0.89) and high validity were demonstrated.28 Of the 105 mothers who filled in the questionnaire, 18 exceeded the recommended cutoff score for depression (>23).
Attachment Insecurity in the Primary Caregiver
Attachment style of the primary caregiver was assessed by using the German version29 of the Adult Attachment Scale by Collins and Read.30 The Adult Attachment Scale is a self-report instrument for measuring attachment-related attitudes. The questionnaire consists of the 3 subscales: depend, close, and anxiety. “Depend” reflects the extent a person trusts and relies on others. “Close” refers to discomfort with intimacy and emotional closeness. “Anxiety” refers to fears of rejection and abandonment in close relationships. Internal consistency (Cronbach's α = 0.72–0.79) and good convergent and discriminant validity of the German version have been shown. The scales close and anxiety that refer to the insecure attachment attitudes (attachment avoidance and anxiety) are largely uncorrelated but both correlate with the depend scale (which reflects the secure attachment attitude). Thus, to avoid redundant predictor variables, we did not consider the depend scale in the data analyses.
We calculated χ2 statistics, t tests, and correlation coefficients to analyze the bivariate associations between each potential predictor variable (anthropometric and psychosocial family characteristics) and the 2 criteria of long-term weight change: success versus failure in weight reduction up to the 12-month follow-up and weight change between the conclusion of treatment and the 12-month follow-up.
We then conducted multivariate (logistic) regression analyses to test the overall statistical significance of the complete set of psychosocial variables (thus controlling for multiple comparisons). We thereby analyzed the β coefficients (and Wald statistics, respectively) to assess the unique contribution of each of the single predictor variables to the prediction of the criterion. The overall procedure is as follows: First, baseline BMI SDS, age, gender of child/adolescent, parental education level, and familial (parents, siblings) obesity were introduced into the regression equation (when these data were significantly associated with the criterion in the bivariate analyses). Thereafter, as a second block, the complete set of psychosocial variables was introduced simultaneously to test whether the set of psychosocial variables significantly predicted the long-term outcome over and above the variables of the first block. The “change statistics” reflect the significance of the unique contribution of this set of variables.
Because of the relatively wide age range of the sample population, we further tested the possibility that age influenced the strength of the associations between the psychosocial variables and the weight loss criteria. Therefore, the interaction effects between age and the psychosocial variables were successively introduced into the regression equation.
Finally, in case of redundant predictors (in comparison with the bivariate association low β coefficient/Wald statistic), a second regression analysis with a stepwise selection procedure was run to identify the best independently contributing predictors.
Prediction of Success Versus Failure in Weight Reduction up to the 12-Month Follow-up (T0 to T2)
To exclude the possibility that the presence of somatic risk factors/diseases (eg, hypertension, dyslipidemia, diabetes, orthopedic problems) affected the success in weight reduction, we analyzed this association. Success versus failure in weight reduction was not significantly associated with presence versus absence of somatic risks/diseases (χ2 = 0.93).
The bivariate associations of the social, anthropometric, and psychosocial variables with the success versus failure criterion are shown in Table 1. Children/adolescents who failed to reduce overweight by more than a 5% BMI-SDS reduction or dropped out of the intervention prematurely significantly differed from successful children/adolescents in age, presence of obese siblings, family adversity, and maternal depression. Specifically, those who failed were older, more often had obese siblings, and more often came from families with more pronounced psychosocial problems; their mothers more often reported depressive symptoms that exceeded the clinical cutoff.
In the logistic regression analysis (Table 2), all at least marginally significant social and family obesity variables (ie, age of index patient, presence versus absence of obese siblings, and maternal obesity) were introduced first to control for their influences. In a second step, the complete set of psychosocial variables was introduced simultaneously to test our hypothesis. The amount of variance uniquely explained by the psychosocial variables proved statistically significant. The Wald statistics of the final model indicated that the presence versus absence of obese siblings explains significant unique variance in the criterion over and above the other variables, whereas the psychosocial variables predict shared variance. The complete model with all predictors proved statistically significant and explains ∼30% of variance. None of the interaction effects between age and the psychosocial variables added significant variance (χ2 between 0.03 and 1.35).
With a second stepwise regression analysis we extracted 2 predictors: presence versus absence of obese siblings and maternal depression. These 2 variables explain significant unique variance of the criterion, which indicates that children with obese siblings and children whose mothers report pronounced depression are at an increased risk of not reducing overweight considerably.
Weight Change Within 12 Months After Program Completion
Presence versus absence of somatic risk factors was not significantly associated with weight change after completion of the program (point-biserial correlation coefficient [rp.bis] = −.04). Table 3 shows the bivariate correlations between the social, anthropometric, and psychosocial variables and weight change during the period between conclusion of the intervention (T1) and the 1-year follow-up examination (T2). Maternal obesity, depression, and anxious attachment attitude were significantly associated with the criterion.
In the multiple regression analysis (Table 4), we first introduced paternal education level and maternal obesity. The set of psychosocial characteristics explained significant variance over and above these control variables. The complete model proved significant and explained 15% of the criterion variance. The β coefficients of the variables indicate that the anxious attachment attitude of the mother exclusively explains significant unique variance. No interaction effect with age proved statistically significant (Fchange between 0.03 and 3.53).
A second, stepwise regression analysis resulted in the selection of maternal anxious attachment attitude as best predictor (explaining 14% variance), which indicated that children of mothers who describe anxiety of rejection and abandonment in close relationships were at an increased risk of (re)gaining weight in the year after the conclusion of the intervention.
In line with our expectations, we found that psychosocial variables were significantly predictive for long-term success when we controlled for familial obesity, preintervention overweight, age and gender of index child, and parental education level. In particular, a high number of family adversity factors and maternal depression significantly predicted long-term failure, and maternal insecure-avoidant attachment attitude showed a trend in this direction. Maternal depression proved to be the best “marker” of that set of variables. Together with familial obesity, maternal depression explained ∼20% of the criterion variance. This result confirms the role of familial obesity10,–,12 and corresponds with the finding by Favaro and Santonastaso,19 who showed that maternal neuroticism was associated with weight loss in children on a prescribed diet.
The characteristics of maternal depression, insecure-avoidant attachment attitudes, and psychosocial risks are most probably associated with less adequate parenting and a poor parent-child relationship,13,15 which may have led to insufficient support of the child's weight-reduction efforts. Future studies should focus on the specific difficulties in the parent-child relationship. On that basis, it might be possible to offer interventions that address the difficulties of these families.
Moreover, we analyzed which factors are specifically predictive for weight change during the period after completion of the program until the 1 year follow-up (T1 to T2). Here we found maternal insecure-anxious attachment attitude, depression, and obesity to significantly predict this outcome variable. Over and above the other predictors, maternal insecure-anxious attachment attitude explained unique significant variance of that criterion (14%). Interestingly, recent research on physician-patient relationships has shown that insecure attachment attitudes of the patient are associated with a lack of compliance and low satisfaction with therapy.31,32 Thus, mothers with insecure-anxious attachment attitudes may relapse more easily into former habits because of low satisfaction with therapy. Moreover, it might be that specifically after the conclusion of “external control” by the therapists, mothers with an insecure-anxious attachment style13 might fear that the weight-control behaviors threaten their relationship with the child/adolescent. In this area, more research should be conducted that addresses the specific difficulties of this group of caregiver-child/adolescent pairs to better meet their needs.
We found evidence for the role of psychosocial characteristics of families in long-term weight-reduction and maintenance outcomes. If cross-validations confirm these results, they point to the need to analyze the specific mediating difficulties of these families and to create tailored intervention modules or other interventions that address the problems related to parenting and attachment.
- Accepted June 13, 2010.
- Address correspondence to Ursula Pauli-Pott, PhD, Institute of Medical Psychology, University of Giessen, Friedrichstrasse 36, D-35392 Giessen, Germany. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
- SDS —
- SD score
- T0 —
- baseline, within 3 weeks before the start of the intervention
- Tn —
- n year(s) after baseline
- Wang Y,
- Beydoun MA
- Oude Luttikhuis H,
- Baur L,
- Jansen H,
- et al
- Reinehr T,
- Kleber M,
- Lass N,
- Toschke AM
- Summerbell CD,
- Ashton V,
- Campbell KJ,
- Edmunds L,
- Kelly S,
- Waters E
- Golley RK,
- Magarey AM,
- Baur LA,
- Steinbeck KS,
- Daniels LA
- Hughes AR,
- Stewart L,
- Chapple J,
- et al
- Laucht M,
- Becker K,
- Blomeyer D,
- Schmidt MH
- Stiensmeier-Pelster J,
- Schürmann M,
- Duda K
- Ciechanowski PS,
- Russo J,
- Katon W,
- et al
- Copyright © 2011 by the American Academy of Pediatrics