Study design and participants
The COPTR Consortium is comprised of four independent randomized controlled trials (RCT) of childhood obesity prevention or treatment. Each RCT implemented a 3-year intervention that was unique and different, but used common and standardized data collection procedures. Additional details of the COPTR Consortium and each intervention study have previously been published [14,15,16,17,18]. The University of North Carolina at Chapel Hill serves as the Research Coordinating Center and receives all common data for the sites. The two childhood obesity prevention trials are located in Minneapolis, MN (University of Minnesota, Now Everyone Together for Healthy and Amazing Kids (NET-Works) Study) and Nashville, TN (Vanderbilt University, Growing Right Onto Wellness (GROW) Study), and the two childhood obesity treatment trials are located in Cleveland, OH (Case Western Reserve University, Ideas Moving Parents and Adolescents to Change Together (IMPACT Study) and Bay Area, CA (Stanford University, Stanford GOALS Study). This study was approved by the Institutional Review Boards on research involving human subjects at University of North Carolina at Chapel Hill, University of Minnesota, Vanderbilt University, Stanford University and Case Western Reserve University.
All four studies recruited predominantly minority populations from households with low socioeconomic status. The sample size, recruitment age range and weight status varied for the four studies. The NET-Works Study recruited 534 2–4 year old children at or above 50th BMI percentile. The GROW Study recruited 610 3–5 year old children between the 50th and 94.9th BMI percentile. The IMPACT Study recruited 360 rising 6th graders at or above the 85th BMI percentile. The Stanford GOALS Study recruited 241 7–11 year old children at or above the 85th BMI percentile. If the household had more than one child that met the eligibility criteria (e.g. two children between 7 and 11 years of age above the 85th BMI percentile) then only one child was randomly selected to be in the study. These analyses are conducted with baseline data from each site. Baseline data were collected between May 2012 and June 2014. For each study, parental consent was obtained for minor child to participate in the study. The two studies with older children also obtained written assent from children.
Dietary assessments
Dietary intakes at all sites were measured using 24-h recalls that were collected on two weekdays and one weekend day using the Nutrition Data System for Research (NDSR) software [19,20,21]. NDSR versions 2011, 2012 and 2013 were used. Dietary recalls were conducted by trained and certified NDSR interviewers. Bilingual (English and Spanish) interviewers conducted dietary recalls in Spanish when requested. The first dietary recall was conducted in-person (except for GROW) and the second (except for NET-Works) and third dietary recalls were conducted over the telephone. In older children (IMPACT and GOALS), the child self-reported their dietary intake with parental assistance, when needed (e.g. provided details on how a food item was prepared). In the preschool-aged samples (NET-Works and GROW), the parent/guardian served as a proxy for the child to report the child’s previous day intake. Food amounts booklets were used by the respondent to assist in identifying portion sizes. For children in childcare, food records were given to the childcare provider and the completed form was used by the parent to report foods the child consumed while in childcare. School menus were also used when needed. The percentage of participants with three dietary recalls was 97.6% in NET-Works, 64.4% in GROW, 96.1% in IMPACT and 100% in GOALS. Average intakes of energy, macro- and micronutrients and food groups were calculated based on the average of each participant’s diet recalls (2 or 3 days).
Dietary quality was measured using the 2010 Healthy Eating Index [10]. The HEI-2010 was used since it aligns with the 2010–2015 Dietary Guidelines for Americans and the data were collected during this time period. The HEI is comprised of nine adequacy and three moderation food components with a predefined maximum score (5, 10, 20) per component. The maximum scores assigned for each component are based on the USDA recommended daily intake per 1,000 kcal (exception: fatty acid component and empty calories component are not standardized to 1,000 kcal). The overall HEI score is the summation of the 12 component scores and ranges from 0 to 100 points with higher scores indicating better dietary quality. Based on the USDA HEI-2005 grading scale, the child’s diet quality was categorized into three groups, 0–50, 51–80 and 81–100) [11]. The Nutrition Coordinating Center (NCC) guidelines and SAS macro for NDSR data were used to calculate the HEI-2010 scores with one exception [22]. Empty calories are the calories from solid fats, alcohol and added sugars. Prior to NDSR version 2014, calories from solid fats were not calculated in the NDSR software, therefore, the individual’s food intakes were used to calculate grams and calories from solid fat. Following the NCC guidelines and SAS macro [22], a component score was calculated for each recall then averaged (2 or 3 recalls) to determine the average HEI component scores.
Covariates
Weights and heights of index parents and children were measured with the participant in light clothing, without shoes, using a standardized protocol across all sites. Weight was measured to the nearest 0.1 kg using research precision grade, calibrated, digital scales and height was measured to the nearest 0.1 cm using a free-standing or wall mounted stadiometer. BMI was calculated as weight in kilograms divided by height in meters squared and age and sex specific BMI percentiles were calculated using the CDC macro [23] and used to categorized children as high normal weight (50th – 84.9th BMI percentile), overweight (85th – 94.9th BMI percentile) or obese (≥95th BMI percentile) [24, 25]. The index parent/guardian was categorized as either underweight (< 18.5 kg/m2), normal weight (≥18.5- < 25.0 kg/m2), overweight (≥25.0 - < 30.0 kg/m2) or obese (≥30.0 kg/m2) [26].
Race/ethnicity, age (date of birth) and sex of index child, marital status (married/living as married, single), employment status (full time, part time, not working for pay) of the index parent/guardian, highest level of household education (<high school, high school or equivalent, at least some college) and participation in supplemental nutrition assistance program (SNAP) were self-reported. Children were classified into five mutually exclusive race/ethnicity groups (Non-Hispanic White, Non-Hispanic Black, Hispanic, multi-racial or other) based on their self-reported race and ethnicity. All questionnaires were administered in English or Spanish according to participant’s preference.
Exclusions
Participants with less than two reliable dietary recalls in a 45-day window (GROW: n = 1; IMPACT: n = 1) or BMI percentile was outside the pre-defined recruitment inclusion criteria (GROW: n = 7) were excluded from the analysis. Reliability of the dietary recall was determined by the interviewer based on the interviewee’s ability to recall dietary intake from previous day. The analytical sample size was 534 for NET-Works, 602 for GROW, 241 for GOALS and 359 for IMPACT.
Statistical analysis
The samples were independently recruited so all analyses were conducted separately. For each participant, the mean overall and 12 component HEI scores were calculated for each recall then averaged. In order to identify potential dietary intervention targets, the percent of the maximum component score was calculated by dividing the average HEI component score by the maximum component score. Linear regression models were used to determine if the overall HEI score differed by six key demographic variables -sex, age, BMI percentile, marital status, employment status and SNAP participation. Because there was more variability in race/ethnicity in the NET-Works Study, this variable (NH White, NH Black, Hispanic, Multi/Other) was also included in the regression models for this site. The fully adjusted models included all of the key demographic variables. The least squares mean HEI scores for the levels of the demographic variables were compared. All analyses were conducted in SAS version 9.4 (SAS Institute) [27].