Appendix C

Literature Search Approach

The committee conducted a literature search to review the methodologic approaches to data collection and analysis presented in recent reports on obesity prevalence and trends. The purpose of the literature search was not to be exhaustive, provide quantitative assessments, or answer the question “What is the current prevalence of, and trends in, obesity?” Rather, this review of literature was intended to provide the committee with insight into the range of approaches that are used. The literature selected does not encompass every methodology, but does point to commonalities and differences across reports and datasets. The evaluated literature also highlights specific issues related to assessing the weight status of children, adolescents, and young adults (defined by the Statement of Task as age 18 years, with consideration up to age 21 years).

Because trends encompass prevalence estimates, the initial approach to identifying reports consisted of a PubMed search for the words (“trend” OR “trends”) and “obesity” in the title or abstract. The results were reduced by using the following preset PubMed filters: published in the past 10 years, human, English, Child: birth-18 years, and Young Adult: 19-24 years. The search resulted in 1,748 publications.

To further explore approaches and methodologies used in publications presenting prevalence estimates, an additional, modified PubMed search was used. In initial assessment of the trends reports, the committee recognized that the terminology shift that occurred in 2007 in classifying body mass index (BMI)-for-age percentiles (i.e., “at risk of overweight” and “overweight” becoming “overweight” and “obese,” respectively) limited the ability to search for 2005-2007 reports assessing what is now defined as

“obesity” in children. For that reason, the prevalence searches were filtered to include only those published in the past 5 years. Because the volume of literature on the topic is expansive, the search approach was restricted to reports with “obesity” and “children” and “prevalence” and (“cross section” or “cross sectional”) in the title or abstract of the publication. The search was also rerun, replacing “children” with (“adolescent” or “adolescence”). The prevalence searches generated 590 results.

Titles and abstracts from these 2,338 publications were reviewed and hand-sorted to identify those reporting on or comparing obesity prevalence or trends in U.S. children and young adults that used BMI to define obesity. Reports assessing obesity prevalence before and after a policy change or other local- or state-based initiative were considered natural experiments and were retained, while publications reporting on the effectiveness or efficacy of a small-scale trial or other interventions were excluded. A total of 137 articles were included from this literature search, and served as the committee’s basis for determining what data collection and analytic approaches appear in reports on obesity prevalence and trends.