The NCCOR Childhood Obesity Evidence Base (COEB): Test of a Novel Taxonomic Meta-Analytic Method aims to:
The COEB Project aligns with NCCOR’s efforts to identify and evaluate practical and sustainable interventions as well as facilitate the ability of childhood obesity researchers and program evaluators to conduct research and program evaluation.
This Project was published as a special supplement in Childhood Obesity.
As a result of the NCCOR COEB Project, a taxonomic-specific database and several other resources were developed, which can be used to examine additional interventions and research in the field. These products are available on the NCCOR COEB Project Documentation page.
The COEB Project was carried out as part of a collaborative effort with the NCCOR COEB workgroup, Mission Measurement, and an external expert panel. Full acknowledgments are listed on the acknowledgments page.
The NCCOR COEB Project is an example of a novel taxonomic approach to social science evidence aggregation, which classifies childhood obesity prevention interventions by characteristics such as study design, intended recipients, context, and intervention components. Traditionally, conventional meta-analytic approaches have been used to examine the effectiveness of childhood obesity prevention interventions. These methods tend to examine narrowly defined obesity prevention initiatives, for example, focusing on solely school-based interventions and/or those designed as randomized controlled trials. As such, they do not allow the field to draw conclusions across settings, participants, or subjects. In the field of childhood obesity research, there is a need for analytic techniques that would allow synthesis of learning across interventions of varying designs and conducted in various settings. This approach, known as a taxonomic meta-analysis, can be a powerful tool for summarizing the evidence that exists and for generating hypotheses that are worthy of more rigorous testing.
For the NCCOR COEB Project, this principle is applied to childhood obesity prevention interventions using body mass index (BMI) as an outcome measure. In this approach, discrete activities are isolated, which may relate to positive child outcomes in a variety of contexts for the intended recipients, children aged 2–5 years. By using the taxonomic meta-analytic approach, we examined which intervention components may be more effective for specific outcomes, in this case BMI, rather than childhood obesity prevention or public health in general. The full list of intervention components (created by Mission Measurement through a Grounded Theory approach and reviewed by the NCCOR COEB Workgroup and an external expert panel is listed in supplementary materials on the NCCOR COEB Project Documentation page.1, 2
The figure below explains major differences between conventional and taxonomic meta-analysis.
| Conventional Meta-analysis | Taxonomic Meta-analysis | |
|---|---|---|
| Objective | Objective is to identify the effect of one or more types of interventions that are defined in advance. Example Researchers plan a conventional meta-analysis where the objective is to learn the effect of a type of intervention, e.g., increased physical activity, in community-based interventions. | Objective is to model the variation of effects as a function of intervention components, contexts, intended recipients, and methodological characteristics. Example Researchers plan a taxonomic meta-analysis to evaluate the efficacy of multiple intervention components, such as setting, content focus, and duration, which may include efforts to increase physical activity. |
| Hypothesis Orientation | Hypothesis-testing orientation Example At the beginning of the project, researchers want to test the hypothesis: Does the average effect of the treatments support that the treatments are effective in changing the pre-designated outcome? | Hypothesis-generation orientation Example The researchers generate hypotheses following a taxonomic meta-analysis, i.e., are certain patterns of treatment components associated with larger effects in the desired outcome? |
| Procedure | Inclusion rules that limit the set of admissible research designs Fixes a set of coding categories and analyses before coding begins Typically focuses on one particular effect size (regarding the pre-designated outcome) and narrow set of pre-defined contextual characteristics Typically focuses on a narrow range of design types or study quality ratings intended to assure internal validity Typically focuses on average effects and their consistency across studies Example In conventional meta-analyses, researchers begin by specifying rules to decide which studies to include, which variables to code, and what statistical analyses of the components to perform. | Wider range of research designs and sources of evidence Coding categories emerge from “ground up” coding of a training set of documents to develop taxonomies; the final codebook is then applied to a larger set of studies Typically focuses on a wider range of design, contextual, and component characteristics Can include a wider range of study effect indicators that may be defined differently but are targeting the same outcome Typically includes a wider range of design types, but accounts for both internal and external validity by coding study characteristics (e.g. sample and publication bias, data omission, and skewing) that could influence internal validity or external validity Typically focuses on modeling variation as a function of components of interventions, contexts, and methodological characteristics Example In taxonomic meta-analyses, it begins with a broader specification of rules for study inclusion and then develops coding categories for interventions, their intended recipients, and contexts. Statistical meta-analytic methods are adopted to account for variations in these factors. |
For more information on why the NCCOR COEB Project completed a taxonomic meta-analysis, see A Rationale for Taxonomic vs Conventional Meta-Analysis.1
To execute the taxonomic meta-analysis, Mission Measurement and the Systematic Review and Meta-Analysis Research Methods team completed four tasks, which were reviewed and iterated with the NCCOR COEB workgroup:
The first analysis examined reducing rate of increase in BMI over time. Findings using the NCCOR COEB database found interventions focused on early childhood obesity prevention efforts may be successful. Specific intervention components that emerged as promising strategies include 1) training caregivers in healthy behavior change strategies, particularly reducing screen time, 2) engaging health care providers directly in the delivery of obesity prevention efforts, and 3) using health care settings were effective.
For more information on how the NCCOR COEB Project developed taxonomies to be applied in taxonomic meta-analysis, see Methods for Taxonomy Development for Application in Taxonomic Meta-Analysis.4
For more detailed information on the results of the taxonomic meta-analysis itself, see A Systematic Review and Meta-Analysis of a Taxonomy of Intervention Components to Prevent Obesity in Children 2 to 5 Years of Age, 2005 to 2019.5
In addition to allowing childhood obesity researchers to draw conclusions from interventions, the process of developing a taxonomic meta-analysis resulted in several resources that can be used by researchers, such as a literature informed database that includes a project overview, table of contents, document log, dataset, and dataset glossary.
The NCCOR COEB searchable database has the potential to inform future initiatives, taking into account intended recipients, delivery channels, intervention components, and context, thereby facilitating initiative customization and the potential for successful outcomes.
There are numerous, possible analyses using the NCCOR COEB dataset. A few expected uses include:
For more detailed information on the implications of the NCCOR COEB Project, see Building Translational Capacity Through Meta-Analytic Methods.6
The NCCOR COEB searchable database and project documentation can be found here.
