- Click here for explanation of classification scheme.

Primary purpose: Report 10-year treatment outcomes of four randomized treatment studies.

Secondary purpose: Identify variables associated with long-term weight regulation.

Inclusion criteria for this review: participation in one of the four previous studies. In general, the inclusion criteria included:

• child 6-12 years of age
• child 20-100% overweight for age, sex, and height
• No current psyciatric diagnosis or reament
• No learning disability
• One parent willing to participate in treatment with the child
• (for exercise studies) subjects could not have a medical problems which prohibited exercise

For specific inclusion critiera see:

Study 1: Epstein LH, Valoski A, Wing RR, McCurley J. Ten-year follow-up of behavioral, family-based treatment for obese children. JAMA 1990; 264: 2519-2523.

Study 2: Epstein LH, Wing RR, Koeske R, Valoski A. Effects of diet plus exercise on weight change in parents and children. Journal of Consulting and Clinical Psychology 1984; 52:429-437.

Study 3: Epstein LH, Wing RR, Koeske R, and Valoski A. Effect of parent weight on weight loss in obese children.  Journal of Consulting and Clinical Psychology.  54(3):  400-401, 1986.

Study 4: Epstein LH, Koeske R, Wing RR. Adherence to exercise in obese children. Journal of Cardiac Rehabilitation. 4:185-195.

(Note, Study 4 was not analyzed as it deals primarily with exercise rather than diet aspects of childhood weight loss.)

For this article, participants were not included if they were not part of the above 4 studies.

For the exclusion criteria for each study, see the above references.

Recruitment

Subjects recruited for weight-control studies at the University of Pittsburgh, PA.

Design

The basic designs of the four included studies are described in the following table (adapted from Table 1 in the article).

Question / Comparison Groups	Intervention Target	Goals for Change	Amount of Exercise (target calorie expenditure)
Study 1: Are treatments designed to target and reinforce habit change and weight loss in obese parents and children together superior to treatments targeting children alone?
Parent and Child	Parent and child	Habits Eating Weight	Self determined
Child only	Child only	Habits Eating Weight	Self determined
Nonspecific	Not specified	Attendance	Self determined
Study 2: Comparison of intervention focusing on diet alone versus diet-plus-life-style exercise in a sample of overweight children and parents
Diet and information	Child only	Habits Eating Weight	Not specified
Diet and lifestyle exercise	Child only	Habits Eating Weight Exercise	1400-2800
Study 3: What is the role of parent weight in the treatment of childhood obesity?
At least one obese parent	Child only	Habits Eating Weight Exercise	300-750
No obese parents	Child only	Habits Eating Weight Exercise	300-750
Study 4: What form of exercise is best suited for long-term weight maintenance?
Diet and aerobic	Parent and child	Habits Eating Exercise	300-900
Diet and lifestyle	Parent and child	Habits Eating Exercise	300-900
Diet and calisthenics	Parent and child	Habits Eating Exercise	Not specified

Blinding used (if applicable)

Blinding not used.

Intervention (if applicable)

Intervention consisted of the basic diet and family intervention approach with different formats for including parents and different exercise regimes.

Statistical Analysis

Data were combined across studies. Different contrasts required different statistical procedures:

1. Analysis of covariance was used to assess interstudy changes from year 5 to year 10.
2. Chi-square analysis used to assess group differences of sustained increases or decreases of 20% overweight and the percent of children obese or overweight.
3. A combination of correlational analysis and analysis of variance techniques used to determine the effects of confounding variables (such as smoking).
4. Multiple regression technique (backward regression procedure) used to identify variables contributing to 10-year outcomes.
5. Basic descriptive statistics were used to compare baseline values for comparing subjects who developed psychiatric disorders with those who did not.

Timing of Measurements

For this study, the key timing was the 5-year and 10-year outcome measurements.

Dependent Variables

• Child weight outcomes: BMI and percentage over BMI (BMI referenced to the 50th percentile for age)
• Percentage of children with at least 20% increase or decrease in weight over 10 years.

Independent Variables

• Diet (measured using modified version of the Food Frequency Questionnaire)
• Activity (caloric expenditure measured using the Paffenbarger Index)
• Program Adherence (measured how often subjects used program techniques, measured using a questionnaire used for this purpose)
• Environmental support for eating and exercise (measured using scales developed by Sallis, Growwman, Pinski, Patterson and Nader (1987))
• Environmental factors, including:
  • subject's current living situation
  • subject's meal pattern
  • weight of people with whom the subject normally eats
  • exercise pattern of people with whom the subject normally exercises
  • subject's perception of the general activity level of the people with whom the subject lives

Control Variables

• smoking
• participation in other weight managment programs
• psychiatric disorders

 

Initial N: 185 families

Attrition (final N): Data from 158 families were available at 10 year follow-up (85%)

Age: children aged 6-12 years at time of treatment

Family Structure: All families were intact.

Ethnicity: not specified

Other relevant demographics: Family SES was, on average, middle-class (45.2 ± 2.0 on Hollingshead index).

Anthropometrics: In all studies except for Study 3, at least one parent was obese (in Study 3 parent weight was an independent variable).

Location: STudies occurred at the University of Pittsburgh, PA

 

Five types of results were reported:

1. Group changes in percent overweight at 5 and 10 years
2. Maintenance of decrease of at least 20% weight at 5 and 10 years
3. Obese versus non-obese status at 10 years
4. The effect of confounding variables
5. Predictors of change in overweight at 10 years

Group Changes in Percent Overweight

Study/Comparison Groups	Significant Group Differences at 5 years	Significant Group Differences at 10 years
Study 1: parent/child, child only, non-specified control	parent/child group > non-specified control (F(1, 145 = 12.94, p<.001) parent/child > child only (F(1, 145 = 5.15, p=.025)	parent/child group > non-specified control (F(1, 145 = 7.11, p=.009) No difference between parent/child and child only groups parent/child group decreased percentage overweight by 15.3% while controls increased percentage overweight by 7.6%
Study 2: diet and information, diet and lifestyle exercise	no group differences both groups decreased percent overweight	no group differences both groups decreased percent overweight
Study 3: at least one obese parent, no obese parents		trend for children of non-obese parents to do better, though difference was not statistically significant (F(1, 145) = 2.44, p=.12)
Study 4: diet and aerobics, diet and lifestyle exercise, diet and calisthenics		diet and aerobics > calisthenics -19.7% vs +12.2% (F(1, 145 = 7.18, p=.008) diet and lifestyle > calisthenics -10.9% vs +12.2% (F(1, 145 = 3.90, p=.05) Changes from 5 to 10 years also significant between aerobics or lifestyle versus calisthenics group.

Maintenance of At Least 20% Decrease from Baseline Percentage Overweight

At 10 years, the percentage of children who maintained 20% decrease in percent overweight from baseline differed among groups (X²(18, N=158)=31.84, p=.02). The following table compares the percent of children in each treatment group who either maintained at least a 20% decrease from their initial percent overweight or who gained at least 20% over their initial percent overweight.

Study/Comparison Groups	Maintained at least 20% decrease	At least 20% increase
Study 1
Parent/Child	43	14
Child only	22	28
Non-specified control	29	38
Study 2
Diet and information	20	20
Diet and lifestyle exercise	31	0
Study 3
At least one obese parent	33	33
No obese parents	16	6
Study 4
Diet and aerobic	58	8
Diet and lifestyle	64	14
Diet and calisthenics	20	20
Total N (N=158)	53	32
Overall Percent	33.5	20.3

Percent Obese after 10 Years

30% of the children across the four studies were not obese after 10 years. No significant differences were detected by group.

Confounding Variables

Smoking and the presence of other weight-control group participation made no difference in 10 year outcomes.

However, the authors compared 10 year treatment outcomes of subjects who had psychiatric problems with those who did not. Using non-parametric comparisons, the authors found that children who developed psychiatric disorders:

• tended to be lighter at baseline
• changed weight less between 5 and 10 year measurement
• were more likely to have mothers who were less obese
• were more likely to have fathers who were more obese at baseline

Factors Predicting Long-Term (10 year) Weight Loss

The best predictor of maintenance of weight loss at 10 years was maintenance of weight loss at 5 years (r²=.52--accounting for 52% of the variance).

When 5-year change was excluded from the predictive equation, other variables (accounting for 34% of the variance were important for predicting success in maintaining weight loss.

Children were more likely to be successful in maintaining weight loss at 10 years following treatment when:

• they were girls
• they had a greater percent overweight at baseline
• fewer meals were eaten at home
• fewer overweight people were living at home
• they were better at self monitoring
• they received less "good eating" encouragement from family members
• they received more discouragement from friends to eat poorly
• they exercised with their roommate
• they received less rewards from exercising from their roommate.

All the above variables were significant (p<.05) and r²=.341

In summary, the results of these studies provide support for the family-based behavioral treatment of childhood obesity. These studies provide the first evidence that weight regulation in children can be achieved and maintained over extended periods from childhood through adolescence to adulthood.

Study 1 suggested that in intact families effects are improved for children by including a parent with the child in treatment.

Study 4 suggested that exercise enhances the long-term effects of diet interventions.

University/Hospital:

University of Pittsburg's School of Medicine

Strengths

• strong designs across studies
• long-term follow-up

Weaknesses

• Predictive equations at 10 years are probably different than we would see in other programs (e.g., other research indicates that eating outside the home is associated with increased overweight, the fact that the 10 year predictive equations find the opposite is probably due to the over sampling of obese parents in the sample--hence the presence of obesigenic households.
• Concern that the way in which families are recruited and the type of famlies (middle-class, intact families) makes generalizability questionable. Disappointing that the authors do not discuss this.