Family Partners for Health is a 5-year cluster randomized controlled trial with a two-group repeated measures design (See Figure ). The sample consists of African American (63%), non-Hispanic white (32%), and bilingual Hispanic (5%) low-income, 2nd to 4th grade overweight or obese children and their overweight or obese parents from small towns and rural areas of North Carolina. An experimental group of children and parents are receiving a two-phase intervention with follow-up. In Phase I (Intensive Intervention) they receive 60 minutes of nutrition and exercise education and coping skills training and 45 minutes of exercise training once a week for 12 consecutive weeks. In Phase II (Continued Support) they meet for 9 monthly meetings. They are then followed for 6 months after the completion of Phase II to assess the maintenance of results (follow-up), for a total of 18 months in the trial. The experimental children and parents have data collected at 4 time points: Time 1 (Baseline), Time 2 (12 weeks: Post Phase I-Intensive Intervention), Time 3 (12 months: Post Phase II-Continued Support), and Time 4 (18 months: 6-months of follow-up).
A wait-listed control group of children and parents are receiving usual care, and they have data collected at the same time intervals as the experimental children and parents (Times 1-4). After they complete the Time 4 data collection, they will be offered the nutrition and exercise education, coping skills training, and exercise intervention (Phase I only). During the course of the trial they are receiving monthly cards to thank them for their continued participation in the trial and remind them when they will be eligible to receive the intervention.
Eight elementary schools are being used as sites during early evening hours for recruitment and delivery of the intervention. Four schools are located in Wilson, North Carolina, and four in Burlington, North Carolina. The sites are similar in structure, size, and rural, ethnic and racial mix. Each site has classroom space with tables and chairs that are being used for the education and coping skills training classes and a gymnasium that is used for the exercise classes.
There have been eight induction periods over the course of the trial. Forty-four to 45 children and 44-45 parents in two schools (experimental and wait-listed control group) have been enrolled in each period, for a total of 358 children and 358 parents. Each site was randomly assigned to either the experimental or wait-listed control group the first time a cohort was enrolled. The second time that the site had children and parents enrolled, the opposite condition was assigned.
Power calculations were performed with POWERLIB20 SAS/IML modules [38
]. These methods calculate power for the general linear multivariate model, which includes repeated measures data structures, of which a two-group longitudinal design is a special case. Power was based on a separate multivariate model for each outcome addressed in the aims, incorporating measurements from all time points. Models were derived separately for children and parents. The power analysis focused on one representative variable in each group of variables, BMI percentiles (children) and BMI (parents) for weight status, waist circumference (children and parents) for adiposity, the Child and Adolescent Trial for Cardiovascular Health (CATCH) nutrition knowledge subscale (children) and the Health Promoting Lifestyle Profile II nutrition knowledge subscale (parents) for health behaviors, and the Child and Adolescent Trial for Cardiovascular Health (CATCH) self-efficacy subscale (children) and the Eating Self-Efficacy Scale subscale (parents) for self-efficacy. On the basis of pilot data, the autoregressive correlation parameter was chosen as 0.99 for parents' BMI, 0.90 for children's BMI and parents' and children's waist circumference, and 0.84 for the health behavior and self-efficacy variables. Effect sizes used were based on studies with similar designs using the dependent variables of interest. Given the clustered nature of the design (with each site-by-condition classification considered a cluster), an intracluster correlation coefficient of 0.01 was selected (based on BMI). A total of 179 child-parent pairs per group or a total of 358 children and 358 parents have been inducted.
The trial was approved by the Institutional Review Board at The University of North Carolina at Chapel Hill. Two months before each induction period, the project manager contacted the principals in the two schools to set up a time for study staff to do classroom presentations and hand out backpack flyers to the children containing information about the benefits of participating in the trial. The students took the flyers home and shared them with parents. Interested parents filled out the flyers with their name, telephone number, and a convenient time for study staff to call. The flyers were then returned to school and put in a box in each teacher's room. The study staff picked up the flyers twice a week and then conducted an initial screening over the telephone. Inclusion criteria for children were ability to speak, write, and read in English; a BMI ≥ 85th percentile for age and gender; a parent or guardian with a BMI ≥ 25 kg/m2; and residence with that parent or guardian. The parents were asked their child's gender, age, height, and weight over the telephone, and a BMI percentile was calculated. Inclusion criteria for parents or guardians included an ability to speak, write, and read in English; a BMI ≥ 25 kg/m2 and residence with a 2nd, 3rd or 4th grade child with a BMI ≥ 85th percentile for age and gender. Parents were asked their height and weight over the telephone, and a BMI was calculated. If the child and parent met the trial criteria, an appointment was made that was convenient for them to meet at the school after school hours to confirm eligibility and review the consent and assent. At times, more than one parent or child from the same family wanted to join the trial. If this occurred, we explained that both parents and/or another 2nd, 3rd, or 4th grade child could participate; however, the same parent and same child had to consistently complete the data forms. Parents could choose among themselves. If there were two children from the same family, the child chosen to fill out the data forms was picked by a coin toss. The project manager confirmed eligibility, explained the trial, and answered all questions before asking the parent to consent and the child to assent. Next a nurse conducted a basic sports history and physical examination on both the child and parent to ensure that they did not have any conditions that would exclude them from participating in the trial. They were excluded if either had a history of a heart murmur, congenital heart disease, family history of sudden death or psychological problems such as claustrophobia that would prevent participation in group classes, or if they were participating in another weight management or prevention program or trial.
A randomized controlled pilot study was conducted to test the feasibility of the child and parent components of the intervention [39
]. The intervention was designed to deliver nutrition and exercise information that was applicable across all cultures and sensitive to individual participants' needs, using examples and handouts depicting all three ethnicities and both genders. The nutrition education classes, which focused on making better food choices and controlling portions, and included ethnic menu plans that reduced fat and calories; exercise education focused on increasing activity and decreasing sedentary behaviors. Exercise physiologists taught the exercise education classes and the exercise training classes in the gymnasium, which included basketball, dancing, tag, exercise bicycles, rowing machines, and stair climbing. Parents also attended coping skills training classes.
From the pilot study, several important lessons were learned that informed the main trial [39
]. First, parents felt it was important for their children to receive the coping skills training. Second, children and parents felt more hands-on activities and food models would make the classes more interesting. Third, parents felt coming to the exercise class twice a week was difficult with their schedules and suggested getting information on how to build a home exercise program. Using this information from the pilot, we revised the intervention for the main study.
Main trial: phase I (intensive intervention)
The nutrition and exercise education in Classes 1-5 were designed to teach children and parents to work together to develop healthy eating habits and increase exercise. Classes 6-11 were designed to teach children and parents to practice new coping skills, using cognitive restructuring, social problem solving, assertiveness training, and conflict resolution in regard to nutrition and exercise issues [37
]. The exercise training classes were designed to reinforce the importance of increasing activity and decreasing sedentary behaviors. At all classes a light healthy meal and beverages were available. Childcare and homework help were available for other children who came with parents and the children enrolled in the study. Transportation vouchers were available for parents to assist with gas to get to and from the intervention.
Class 1 focused on understanding calories, proteins, carbohydrates, and fats [44
]. The interventionists used an interactive strategy with pictures in a game format to show participants how calories added up, what foods contained healthy and unhealthy calories, and what foods contained proteins, carbohydrates (simple and complex), and fats (saturated and unsaturated). Foods that were culturally preferred were used. Class 2 taught how portion control could make a difference and what a usual portion size should be [44
]. The interventionists used food models that children and parents could handle so they could learn to estimate normal portion sizes. Children and parents used measuring cups, bowls, and plates to create a portion-controlled meal balanced in protein, carbohydrates and fats. Each child and parent received a set of measuring cups to take home. Class 3 focused on how to make healthy substitutes with food [44
]. The interventionists used a card game with pictures of healthy foods that could be substituted for unhealthy foods, including skim or 1% milk for whole milk; lean ground turkey for ground beef; baked or broiled chicken or fish for fried chicken or fish; egg whites for regular eggs; and unsweetened cereals for sweetened cereals. Class 4 used an interactive game to teach children and parents how to choose healthy foods when eating out [44
]. They learned that it was important not to eat super-size meals, to avoid breaded and fried meats, and to drink water, low fat milk, or sugar-free drinks instead of regular soda, juice, or milkshakes. They also learned how to ask for nutritional information in fast food restaurants and to check for information online. Children and parents then competed in teams. Each team received a menu with nutritional information from a fast food restaurant and worked together to create a balanced meal. Class 5 taught children and parents the importance of exercise and current exercise guidelines [14
]. Children and parents learned what moderate-intensity exercise was and how they could use breathing or heart rate to monitor exercise intensity [48
The remainder of the class sessions focused on coping skills training. Class 6 taught children and parents how to use cognitive restructuring, using recognition of thoughts and feelings, problem solving, and guided self-dialogue to find ways to increase daily exercise [43
]. Class 7 used social problem-solving to improve nutrition behaviors. Children and parents learned the problem solving steps, including identifying the problem, determining goals, generating alternative solutions, examining consequences, choosing the solution, and evaluating outcomes [37
]. Class 8 used assertiveness training with role playing to teach children and parents how they could motivate each other in a positive manner and provide each other with positive reinforcement [41
]. Class 9 used the social problem solving steps used in Class 7 to help children and parents improve exercise behaviors [37
]. Class 10 used assertiveness training to teach children and parents how they could get back on track after relapsing from healthy eating and exercise behaviors [41
]. Class 11 focused on conflict resolution: children and parents were taught respectful ways to work through conflicts around nutrition and exercise behaviors [42
]. Class 12, the final class, included a jeopardy game with questions from previous classes and provided an opportunity for children and parents to review the important components of the intervention.
Main trial: exercise training classes
Exercise training classes were held weekly for 45 min after the classroom sessions, providing experiences that did not require sophisticated equipment, so that the children and parents could develop a home exercise program. All sessions started with a warm-up, followed by strength circuit training and aerobics, and activities such as small-group team sports (basketball, soccer, floor hockey, and bucketball), chase games, and cardio kickboxing to a video, walking, and Dance Dance Revolution. The exercise interventionists reinforced ways to increase exercise such as taking a walk with a partner, spouse, and/or child, walking the dog, using the stairs instead of the elevator, parking farther from the store, and walking to the store. Each child and parent received a pedometer to measure steps. They were asked to increase their steps by 500 a day for the first week and then by 500 steps a week until they were averaging 10,000 steps a day [48
Main trial: phase II (continued support)
During Phase II, the experimental group came back to the school in the early evening once a month for 9 months for continued support. The sessions focused on discussing and problem solving any nutrition and exercise issues participants were having difficulty with. Two weeks after each monthly meeting, the interventionists made a brief telephone call or sent an email to check in and answer any questions the children or parents had.
Table shows the variables and measures being used in the study, the data source, and measurement times. Data are being collected at Time 1 (Baseline), Time 2 (Post Phase I-Intensive Intervention), Time 3 (Post Phase II-Continued Support), and Time 4 (6-Months After Completion of Phase II). All instruments were evaluated for psychometric properties before the pilot study, were re-evaluated during the pilot study, and will be re-evaluated in this study. Completion of measures takes approximately 60 min for most children and parents.
To ensure inter-rater reliability during the course of the study, RAs were trained and tested for inter-rater reliability prior to each data collection on height, weight, waist circumference, triceps and subscapular skinfolds. During data collection, a duplicate measures program was implemented for quality control testing. Every 10th participant had height, weight, waist circumference, triceps and subscapular skinfolds repeated by a second RA. These data were analyzed monthly to assess reliability and if reliability was not adequate, the RAs were retrained. All data collection RAs were blinded to the study group assignment.
Parents filled out a demographic sheet for themselves and their children, on age, gender, ethnicity, and race. Additional questions for parents included marital and employment status, socioeconomic status and education level. Additional questions for parents to answer about their children included their birth order and health status.
Height was measured on all children and parents in street clothes without shoes, using a stadiometer, calibrated in 1/8-cm (cm) intervals. Height was measured twice and averaged. Weight on all children and parents was measured in a private room, in street clothes without shoes, to the nearest 0.1 kg using a Tanita WB-110A Digital Scale.
For children, BMI percentiles were calculated twice by entering height, weight, age, and gender [49
]. Children with a BMI ≥ 85th
and < 95th
percentile for age and gender were considered overweight, and those at or above the 95th
percentile were obese [49
]. BMI of parents was calculated twice by entering height and weight (kg/m2
]. In adults age 20 years and older, overweight was defined as a BMI between 25.0 and 29.9, and obesity was defined as a BMI equal to or greater than 30.0 [49
Waist circumference was measured in a privately screened area by two RAs, following the procedure used in the Insulin Resistance Atherosclerosis Study, using a Figure Finder measuring tape with lock (Novel Products Inc., Rockton, IL) [50
]. All measurements were performed three times and averaged according to the National Health and Nutrition Examination survey procedures [51
Using Lange skinfold calipers, triceps and subscapular skinfolds were measured in children and parents on the right side of the body three times and averaged, also according to the National Health and Nutrition Examination survey procedures [51
]. To ensure reliability, prior to each data collection, RAs were tested for inter-rater reliability by calculating correlations when measuring skinfolds on the same participants.
Health behavior outcomes
The 23-item Adult Health Behavior Survey [53
] and the 20-item Child Health Behavior Survey [53
] were used to collect information on intake of sugared beverages, juices, fruits, vegetables, whole grains, fried, baked and fast foods on a daily or weekly basis. Responses are scored from 0 = none or never to 5 = 5 or more. Alpha coefficients in children range from 0.79 to 0.80 and in parents from 0.80 to 0.86 [53
The Health Promoting Lifestyle Profile II (HPLP II) was used to measure health promoting lifestyle behaviors in parents [54
]. This 48-item, 4-point Likert scale questionnaire with 4 response choices: never, sometimes, often, or routinely, measures the frequency of health promoting behaviors in six subscales. Only four subscales (health responsibility, exercise, nutrition, and stress management) were used in this study; the communication and spirituality subscales were not used. The instrument has been used with both minority and white populations [54
]. Alpha coefficients ranged from 0.78 to 0.93 for the subscales. Test-retest reliabilities in African American women have ranged from 0.70 to 0.74 [57
The Child and Adolescent Trial for Cardiovascular Health (CATCH) questionnaire was used to measure children's diet and exercise health behaviors and self-efficacy [55
]. The instrument contains 130 forced-choice items on a 3-point Likert scale in seven subscales: exercise, dietary knowledge, dietary intentions, dietary choices, support, social reinforcement, and self-efficacy. Alpha coefficients ranged from 0.76 to 0.84 in a sample of 5,000 children [55
]. The questionnaire is at a 2rd grade reading level and takes approximately 15 min to complete [40
]. Alpha coefficients for the subscales ranged from 0.87 to 0.90 in our pilot study [40
The Eating Self-Efficacy Scale [56
] was used to measure self-efficacy related to dietary changes in parents. This 25-item instrument asks participants to rate their difficulty in controlling eating from 1 (no difficulty) to 7 (difficulty) on two subscales, negative affect (NA) and socially acceptable circumstances (SAC). Negative affect eating is related to emotional eating and the triggers that cause it (e.g., anger or anxiety). Socially acceptable eating is related to overeating at parties, family events, or holidays. Scores range from 25 to 175, with higher scores indicating more difficulty in controlling eating. Alpha coefficients were 0.94 for the NA subscale and 0.85 for the SAC subscale. Test-retest reliability was 0.70 in a sample of 600 women and men [56
Exercise self-efficacy in parents was measured using Bandura's Exercise Self-Efficacy Scale with 18 questions on a 100-point scale, ranging in 10-unit intervals from 0 (cannot do at all) through intermediate degrees of assurance such as 50 (moderately certain can do) to 100 (certain can do) [33
]. The questions are added up and divided by 18 to calculate a total Exercise Self-Efficacy score [33
]. A higher score indicates greater self-efficacy. Alpha coefficients of the total scale ranged from 0.86 to 0.92 in adult men and women [58
Data analysis plan
An intent-to-treat analysis will be used in which all participants are analyzed according to their initial randomized assignment, whether they receive the intervention regularly or not, to preserve the balance of covariates due to randomization and to provide a conservative analysis that does not overestimate intervention effectiveness. Random baseline differences will be accounted for with this approach.
To determine the effects of the intervention on weight status, adiposity, health behaviors and self-efficacy, general linear mixed models will be used. Separate random coefficients models will be used to test each outcome for longitudinal differences between the experimental and wait-listed control groups. Advantages of the mixed model approach include its ability to incorporate data from visits that are mistimed as well as certain missing data structures, so that participants who miss one or more visits need not be excluded from the analysis. In addition, random coefficient models accommodate both categorical and continuous covariates.
If a characteristic is found to differ between the groups at baseline, it will be included as a covariate to account for this random imbalance. Additionally, recognizing that certain factors could influence our dependent variables, whether or not they are randomly imbalanced in the two groups, we will control for them as covariates in the models. These variables include asthma, diabetes, low-dose steroids, psychiatric medications (e.g., amitriptyline), cardiac medications (e.g., beta blockers), and recent smoking cessation.
The relationships between experimental child changes in weight status, adiposity, health behaviors, and self-efficacy and experimental parent changes in weight status, adiposity, health behaviors, and self-efficacy will also be addressed via the general linear mixed model. Each outcome will be analyzed through a separate random coefficient model. The models will be structured in a fashion similar to that described for Aim 1, but with a different set of predictor variables, in order to examine the associations between improvements in parents and their children. For each model, the children's baseline values for the outcome will be included as a covariate.
To determine whether African American, non-Hispanic white, and bilingual Hispanic children and parents in the experimental group benefit differentially from the intervention, two indicator variables for African American and Hispanic children and adults, respectively, will be created, and non-Hispanic whites will serve as the referent group. The models described above for Aim 1 will be refitted, with these indicators added as main effect terms. Additionally, pairwise interaction terms between the intervention indicator and each of these ethnic indicators will be added to the models to assess the degree to which the impact of the intervention differs among these three ethnicities.
Baseline characteristics of parents and children are summarized in Table . The mean age of the parents was 36.7 years and the great majority were female (93%). Over half (63%) of the parents and children were African American, 32% were non-Hispanic white, and 5% bilingual Hispanic. The mean age of the children was 8.6 years and the majority (56%) was also female.
Demographic Characteristics of Parents and Children
The parents' mean BMI was 37.6, which is classified as obese (Table ). African American parents mean BMI (38.4) was higher than either non-Hispanic white parents' (36.3) or Hispanic parents' BMIs (32.8). The children's mean BMI percentile was 95.2%, which is also classified as obese. African American children (95.5%) and Hispanic children (95.2%) were similar but non-Hispanic white children (94.0%) were lower.
Weight and adiposity by total sample and by ethnicity of parents and children
Parents' mean waist circumference was 109.4 cm, which is classified as obese. African American parents (111.2 cm) had higher waist circumferences than non-Hispanic white (106.5 cm) and Hispanic parents (99.6 cm). Parents' mean subscapular skinfolds were 37.0 mm and triceps skinfolds were 37.3 mm, which are both classified as obese. Again, African American parents had higher subscapular (38.6 mm) and triceps skinfolds (39.4 mm) than non-Hispanic white (34.6 mm; 36.3 mm) and Hispanic parents (30.6 mm; 30.2 mm). The children's mean waist circumference was 77.5 cm, which is classified as obese. African American children (78.0 cm) had higher waist circumferences than non-Hispanic white (76.4 cm) and Hispanic children (76.6 cm). The children's subscapular skinfolds were 22.4 mm, and triceps skinfolds were 24.7 mm, both classified as obese. Again, African American children had higher subscapular skinfolds (22.9 mm) and triceps skinfolds (25.4 mm) than non-Hispanic white children (21.7 mm; 23.5 mm) and Hispanic children (19.1 mm; 23.3 mm).
Health behavior outcomes were measured in children using the Child and Adolescent Trial for Cardiovascular Health (CATCH) [55
] questionnaire and the Child Health Behavior Survey [53
]. On the CATCH mean scores for dietary intention (1.5), usual food choices (1.5), dietary knowledge (1.5), dietary habits (1.4), and support for physical activity (1.4) were moderate. In addition, children reported support for making healthy food choices from parents (1.4), teachers (1.5) and friends (1.6) (See Table ).
CATCH health behavior questionnaire
The Child Health Behavior Survey [53
] measures usual daily and weekly food and beverage intake and found that 52% of the children drank 2 or more sweetened beverages a day, 21% drank 3 or more glasses of milk a day and 45% drank water when thirsty. Eleven percent ate 4 or more vegetable servings a day and 20% had 4 or more servings of fruit a day. When having a snack 46% chose candy, chips, cereals, cookies or cake and 44% chose fruit, vegetables, yogurt or ice cream for a snack (Table ).
Child health behavior survey
Health behavior outcomes were measured in parents using the Adult Health Behavior Survey [53
] and The Health Promoting Lifestyle Profile II [54
]. The Adult Health Behavior Survey [53
] measures daily and weekly food and beverage intake. Over 50% of the parents drank 2 or more sweetened beverages a day, 4% drank 3 or more glasses of milk a day and 46% drank water when thirsty. Six percent ate 4 or more servings of vegetables a day and 3% ate 4 or more servings of fruit a day. When having a snack 61% chose candy, chips, cereal, cookies or cake and 22% chose fruit, vegetables, yogurt or ice cream (Table ).
Adult health behavior survey
The Health Promoting Lifestyle Profile II examines health responsibility, nutrition, exercise, and stress (Table ). On the subscales the parents mean score for health responsibility (1.3) was sometimes, nutrition (2.2) was often, exercise (0.9) was never to sometimes and stress (1.3) was sometimes.
Health promoting lifestyle profile II, eating self-efficacy and exercise self-efficacy
Self-efficacy in children was measured using the CATCH [55
] questionnaire, which examines diet self-efficacy (2.3) and exercise self-efficacy (2.5) (Table ). Self-efficacy in parents was measured using the Eating Self-Efficacy Scale [56
] and the Exercise Self-Efficacy Scale [33
]. See Table . The children had moderate mean scores on diet self-efficacy (2.3) and physical activity self-efficacy (2.5). The parents demonstrated moderate scores in eating self-efficacy for the negative affect scale (43.6) and socially acceptable scale (37.5). In exercise self-efficacy the parents scored moderately (45.0) on the scale.