Section
7
Measuring Body Composition in Population Health Research: Case Studies
Case Study 1 School-based, Cluster Randomized Control Trial to Prevent Childhood Obesity
Background
A cluster randomized controlled trial is designed to evaluate the effects of a multi-component intervention that increases the weekly minutes of physical education offered to elementary students and introduces a nutrition education curriculum, culinary training for food service staff, and modified menu items for school meal programs. All 3rd to 5th grade children attending the 40 intervention schools will be exposed to these interventions for 3 years. Children in the 40 control schools will experience the standard school curricula for health education, the state-mandated minimum amounts of physical education, and the cafeteria menu that meets school nutrition standards. The study team recognizes that children are growing during these ages (8 to 11 years old), and height and weight are therefore expected to increase as part of normal growth during the study. Healthy weight gain with favorable changes in body composition (FM relative to muscle mass) and prevention of inappropriate weight gain rather than weight loss are the primary goals of the intervention. Therefore, the project team would like to evaluate the extent to which the intervention impacts obesity over the 3 years of the study.
Considerations
The study team plans to measure body composition at baseline and at multiple time points during the study, so the measurement method must be able detect changes over the 3-year study period. School-based interventions are often not intense enough to lead to large changes in body composition. Thus, the study needs to enroll and measure a large number (about 5,000) of children in order to have enough power to detect small changes in body composition.
Given the scale of the intervention and need for many participants, measurements must be quick and feasible to administer in a school setting. Equipment must be sufficiently portable between schools and/or affordable to allow for the purchase of several sets of equipment for measurements at multiple schools at the same data collection time points.
Schools have significant curriculum time requirements for learning as well as statewide student testing periods, which can limit the time available to conduct research study measurements on children. Furthermore, schools have many open public spaces and limited privacy for the study team to conduct measurements. Measurements must always be taken with children wearing their own light clothing. It is important that the measurement process does not embarrass children or contribute to weight stigmatization. For this reason, the study team may want to consider purchasing screens to set up private areas to take measurements. Lastly, the study team must choose assessment methods that will be acceptable to parents who will provide consent, children who will provide assent, and school leaders (e.g., district superintendent, school principal) who stipulate their terms for participating in research studies. In addition, some school districts have their own institutional review boards that must review and approve research protocols.
Measure Selection
The study team is aware that height and weight are most frequently used in school contexts as a surrogate measure for body composition. They are the least intrusive of adiposity assessment methods; require little specialized training and equipment; can be conducted quickly, minimizing the time the child is absent from the classroom; and are relatively low in cost. In addition, because height and weight are commonly used across pediatric practices to track growth, parents and children are familiar with them. When height and weight are collected, BMI can be calculated, and some index of BMI (BMI z-score or BMI percentile) is used as the outcome. However, the study team recognizes that BMI and related indices do not provide information on body composition, specifically FM and FFM. Furthermore, the current version of the CDC growth charts are not intended to be used for BMI z-scores and percentiles above the 97th percentile because changes in extreme values are compressed into a narrow range of associated z-scores or percentiles that do not reflect meaningful changes.62
If the participating student population in your study is known to have a higher prevalence of severe obesity be sure to consider this when selecting a method. The current version of the CDC growth charts are not intended to be used for BMI z-scores and percentiles above the 97th percentile.
The study team considers the feasibility of other potential sources of data and methods. Researchers could consider using secondary data or existing BMI surveillance data as part of their evaluations if the data collection periods sufficiently align with desired outcome time-points for the study and can link individuals’ BMI across time.115 Some states and districts require annual BMI data collection in schools, but policies vary in terms of which grades are required to participate. Use of secondary BMI data also prevents the study team from ensuring standardized data collection protocols. Use of longitudinal measures of skinfold thicknesses or waist circumferences are also considered. However, both methods require greater skill and training to conduct than does BMI, require more time to collect measurements, and can be intrusive and embarrassing to the child and parent. Bioelectrical impedance analysis (BIA) also may be considered, as it is non-invasive, uses portable instrumentation, and is relatively inexpensive, simple, and quick. It is important for researchers to confirm whether the selected BIA instrument has predictive equations that have been developed and validated in a population of children with similar characteristics to those in the study. BIA requires standardized environmental conditions and participant preparation that may be difficult to consistently implement with large samples of young children. It also will be difficult to compare the study’s results to other studies if BIA is selected. Methods that can estimate whole-body FM, such as DXA and ADP, would not be feasible in the school setting for many reasons including lack of accessibility, cost, the need for trained/technical staff, and time.
Ultimately, the study team decides to use measures of height and weight to calculate BMI because, of the possible methods to choose from for this study, it is the least intrusive, requires little specialized training and equipment, and can be conducted quickly. This method of adiposity assessment is typically more feasible and acceptable with children in a school setting.