Bright Bodies Weight Management Program

Study 1

Summary

Savoye et al. (2007, 2011) and Shaw et al. (2009) used a parallel group, randomized control trial to examine 209 youths aged 8 and 16 with a body mass index (BMI) above 95th percentile. Participants were randomly assigned to the treatment group or a traditional clinical weight management control group. However, due to discontinuation of a part of the program, 105 participants ended up in the treatment group and 69 participants in the control group. Measures collected at baseline, six months, 12 months and 24 months examined body weight, blood pressure, and glucose levels.

Savoye et al. (2007, 2011) and Shaw et al. (2009) found that, relative to control participants, intervention participants experienced:

• minimal weight gain over 12 months, despite increases in height resulting in a reduction in BMI
• a 4% reduction in body fat; and reductions in total cholesterol and insulin sensitivity at 12 months
• maintenance of the improvements for 1 year after the intervention (2 year study)

A subsample of participants was also measured on glucose and insulin sensitivity and participants in the treatment group experienced the following outcomes after 12 months:

• a 53% reduction in insulin levels
• a 42% increase in whole body insulin sensitivity index
• a small, but statistically significant, decrease in glucose levels
• a significant conversion from prediabetes to normal glucose for intervention and conversion from normal to prediabetes in control group

Evaluation Methodology

Design: Bright Bodies was evaluated using a parallel group, randomized control trial that recruited participants from a Pediatric Obesity Clinic who met the following criteria: body mass index (BMI) above 95th percentile based on the Center for Disease Control growth chart; between the ages of 8 and 16; able to speak English; caregiver willing to participate in the educational component of the program; did not have a major mental illness or serious medical condition; were not taking medication that could potentially cause weight gain.

A total of 284 participants were assessed for eligibility; 62 declined to participate and 13 did not meet criteria. Therefore, 209 participants were randomly assigned to a weight management group and a control group at a 2:1 ratio and then within the weight management group were randomly assigned at a 1:1 ratio to Bright Bodies using a structured meal plan and Bright Bodies using better food choices as part of the nutritional component. However, due to an 83% dropout rate by 6 months from the structured meal plan group (n=35), this portion of the program was discontinued and randomization to control and Bright Bodies continued at the 2:1 ratio. Therefore, a total of 174 participants remained in the study, 105 participants in the treatment group and 69 participants in the control group. Control group participants and their caregivers received diet and exercise counseling by dieticians and physicians once every 6 months at a pediatric obesity clinic. They also received brief psychosocial counseling by a social worker at these appointments.

A subset of 34 participants was further randomized to undergo an oral glucose tolerance test (OGTT) to evaluate the program's impact on glucose metabolism and insulin sensitivity, however, only 23 participants (13 in the treatment group and 10 in the control group) completed these measures at baseline and 12 months.

By the end of the 12 month program, 29% of the participants in the treatment group and 33% of the participants in the control group dropped out. At 24-month follow up, 57% of the treatment group and 55% of the control group had dropped out.

The program was delivered for one year; measures were collected at baseline, 6 months, 12 months and 24 months for both the treatment and control groups at the same facility using the same scales and instruments.

Sample: Of the 174 participants in the study, 61% were female, 37% were White, 38% were Black and 24% were Hispanic. The average age was 12 years old.

Measures: Measures included weight, body fat, body mass index, and blood pressure. Blood samples were also taken to determine lipid and insulin levels. Homeostasis Model Assessment of Insulin Resistance (HOMA-IR) was used to measure insulin sensitivity and was calculated using fasting plasma and glucose levels from the blood samples. All of these measures were collected at baseline, 6 months, 12 months and 24 months for both the treatment and control groups at the same facility using the same scales and instruments. The additional measure for the subsample was an oral glucose tolerance test (OGTT) which measures glucose and insulin sensitivity. From this test, whole body insulin sensitivity index is also calculated.

Analysis: Procedures were consistent with an intent-to-treat analysis. Baseline characteristics were compared using t-tests and chi-square tests. Changes in weight, body fat, BMI, blood pressure, lipid and insulin levels were compared between treatment and control groups using mixed model repeated measures analysis. Savoye et al. (2007) imputed missing data using a multiple imputation method; however, separate analyses were also conducted on models without multiple imputation and no differences were found on the treatment estimates between models with and without multiple imputation of missing data. Therefore, the study only presented results from the multiple imputation analyses. 

Savoye et al. (2009) used mixed model repeated measures analysis incorporating all available data. Prior to analyses, missing data patterns were examined with simple graphical and tabular methods. The authors state that missing levels of outcomes were not related to observed previous levels of outcomes or any baseline covariates for either the treatment group or control group. Additionally, using pattern mixture models with dropout as a covariate, no effect of dropout was found on estimates of treatment, time, and treatment by time.

Outcomes

Implementation Fidelity: No information on fidelity measures was provided in the study.

Baseline Equivalence: Baseline characteristics were compared using t-tests and chi-square tests. Control and treatment groups were statistically equivalent at baseline on all demographic and outcome variables.

Differential Attrition: At the 24-month follow-up, no significant differences were found in dropout rates among ethnic groups, although minority participants had a somewhat greater retention than white participants. Further, no baseline difference in demographic or outcome measures (with the exception of diastolic blood pressure) was found for those who dropped out versus those who didn't. The most common reasons for dropping out were similar for the treatment and control groups - moved with no forwarding address and lost interest in the program. The percentage of participants lost to follow-up was substantial, overall, but similar between groups (57% in the treatment group and 55% in the control group).

Posttest (Savoye et al., 2007): Differences between the two groups in changes in BMI, weight, and body fat at 6 and 12 months were all significant (p<0.001). Specifically, while both groups had equivalent increases in height in the 12-month period, the treatment group gained an average of 0.66 pounds versus the control group that gained an average of 16.94 pounds. The treatment group lost an average of 4% body fat, whereas the control group gained an average of 2% body fat. Differences in total cholesterol measures and changes in insulin sensitivity as measured by the HOMA-IR were also statistically significant at 6 (p=.05 and p<0.001 respectively) and 12 months (p=.005 and p<0.001 respectively). However, there were no significant differences in changes in other lipid levels or blood pressure between treatment and control groups.

Long-term (Savoye et al., 2011): Bright Bodies demonstrated sustained treatment effects on all measures that were significant at post-test. Specifically, differences between the two groups in changes in BMI, weight, and body fat were significant at 24-months (p<0.002). While both groups had equivalent increases in height in the 24-month period, the treatment group gained an average of 13 pounds versus the control group that gained an average of 26.4 pounds. The treatment group lost an average of 3.6% body fat, whereas the control group gained an average of 0.6% body fat. Differences in changes in total cholesterol measures and changes in insulin sensitivity as measured by the HOMA-IR were also statistically significant at 24 months (p=.004 and p<0.001 respectively).

Subsample Outcomes (Shaw et al., 2009): A report on the subsample comparison between treatment and control groups was reported in a third article. Treatment group participants had a 53% reduction in insulin levels and a 42% increase in whole body insulin sensitivity index (derived from glucose and insulin levels tested using OGTT) compared to the control group (p=.0025 and 0.007 respectively). There was also a small, but statistically significant, decrease in glucose levels in the treatment group (p=0.042). Further, in three participants in the treatment group who had pre-diabetes (impaired glucose tolerance or impaired fasting glucose), their levels normalized, whereas the number of pre-diabetic participants in the control group increased from one to five at 12 months (p=0.04).

Study 2

The program had participants attend sessions twice per week for six months.

Summary

Savoye et al. (2014) used a randomized controlled trial to examine 75 youths ages 10-16 with prediabetes (but not diabetes or other serious medical conditions). After assignment to an intervention or standard clinical care control group, the youths completed a six-month assessment of glucose and insulin levels, weight and body mass, and cardiovascular health.

Savoye et al. (2014) found at the six-month posttest that the intervention group, relative to the control group, had:

• lower glucose and insulin levels
• lower weight and body mass
• better cardiovascular health

Evaluation Methodology

Design:

Recruitment: The sample came from youths ages 10-16 who had prediabetes and were recruited from the Yale Pediatric Obesity Clinic. Those having diabetes or other serious medical conditions, taking medications that affect weight, insulin sensitivity, or glucose metabolism, participating in another lifestyle program, or planning to move within the next six months were not eligible. Of 577 assessed, 92 met eligibility criteria and 75 consented.

Assignment: The randomized controlled trial used permuted blocks to randomly assign the 75 participants to the intervention group (n = 38) or a control group (n = 37) that received diet and exercise instruction from the Obesity Clinic and standard clinician care every 2-3 months.

Assessments/Attrition: Assessments occurred at baseline and posttest (i.e., six months later). Six-month posttest data were missing for 23% of the participants.

Sample:

The randomized sample averaged 13.0 years of age and included 65% females. By race, the sample had 28% Blacks, 36% Hispanic Whites, and 33% non-Hispanic Whites. About 80% of children lived in homes with incomes of less than $30,000.

Measures:

The study examined 22 outcome measures that were grouped according to three primary categories: 1) glucose and insulin metabolism, 2) anthropomorphic characteristics (height, weight, body mass), and 3) cardiovascular health. All measures related to physical health or obesity. Change in two-hour plasma glucose was the primary outcome. The measures appear to have been obtained independently with well-validated instruments and procedures.

Analysis:

Changes in outcomes were compared between groups using ANCOVA with adjustment for the baseline outcome and body weight. Analyses also adjusted for baseline factors associated with dropping out.

Missing Data Method: The analysis used multiple imputation for missing data at baseline and posttest, including the posttest outcomes. An additional complete case analysis had little effect on treatment estimates and was not presented.

Intent-to-Treat: The analysis used all randomized participants in their originally assigned condition. Five control participants "were withdrawn from the study because their clinicians started them on metformin," but they were included in the analysis with multiple imputation.

Outcomes

Implementation Fidelity:

Figure 1 in the supplementary document indicated that 95% of the intervention group received the intervention, but the authors offered no details on implementation of the program.

Baseline Equivalence:

Tests in Table 1 for condition differences among the full randomization sample "produced similar distributions of baseline characteristics in both groups," but the authors did not explicitly state that all the differences were non-significant. In addition, the difference between the means for one outcome (2-hour insulin) appears large relative to the standard deviations.

Differential Attrition:

Attrition rates were 18% for the intervention group and 27% for the control group. Dropping out was associated with several baseline measures, but according to the text (p. 320), tests for baseline equivalence using the analysis sample in Table 1 showed no significant differences. Note again that the difference between the means for one outcome (2-hour insulin) appears large relative to the standard deviations.

Posttest:

Of the 22 tests listed in Table 2, 12 reached statistical significance. The intervention group had significantly better outcomes for five of nine measures of glucose and insulin metabolism, five of six measures of weight and body mass, and two of seven measures of cardiovascular health. These effects were the same across gender.

Mediation analysis revealed that from 11-38% of the treatment-related changes in glucose, insulin, and triglycerides were explained by changes in body weight.

Long-Term:

Not examined.