Evidence-Based Guidelines Methodology

The decision to develop "evidence-based" guidelines by an expert panel was based on the increased attention being paid to clinical practice guidelines from methodologists, professional associations, third-party payers and policy makers, and the NHLBI's mission to analyze research results with the goal of providing information that may enable health care practitioners to enhance their ability to detect, treat, and prevent disease.

Choosing the expert panel was an important aspect of the methodology. The panel was comprised of 24 members, 8 ex-officio members, and a methodologist consultant. Areas of expertise contributed to by panel members included primary care, epidemiology, clinical nutrition, exercise physiology, psychology, physiology, and pulmonary disease. There were five meetings of the full panel and two additional meetings of the executive committee comprised of the panel chair and four panel members.

The charge to the expert panel was to develop evidence-based clinical guidelines for primary care practitioners, however, the guidelines would also be useful for certain specialists, clinical investigators, and managed care organizations. In keeping with this approach, the panel

• defined topics to be included in the guidelines
• developed an evidence model depicting the strategy of inquiry for each area of scientific interest
• established criteria for searching and abstracting the literature
• constructed and reviewed evidence tables of individual studies and summary tables of studies falling within a specific category of evidence
• identified the level or strength of the evidence that served as the basis for the recommendations.

A complete description of the methodology used to develop the guidelines can be found in Appendix I.A.1.

The guidelines are based on a systematic review of the published scientific literature found in MEDLINE from January 1980 to September 1997 of topics identified by the panel as key to extrapolating the data related to the obesity evidence model. Evidence from approximately 394 randomized controlled trials (RCTs) was considered by the panel.

The San Antonio Cochrane Center assisted the panel in the literature abstraction and in organizing the data into appropriate evidence tables. The center pretested and used a standardized 25-page form or "Critical Review Status Sheet" for the literature abstraction. Ultimately, 236 RCT articles were abstracted and the data were then compiled into individual evidence tables developed for each RCT. The data from these RCTs served as the basis for many of the recommendations contained in the guidelines.

The panel determined the criteria for deciding on the appropriateness of an article. At a minimum, studies had to have a time frame from start to finish of at least 4 months. The only exceptions were a few 3-month studies related to dietary therapy and pharmacotherapy. To consider the question of long-term maintenance, studies with outcome data provided at approximately 1 year or longer were examined. Excluded were studies in which self-reported weights by subjects were the only indicators used to measure weight loss. No exclusions of studies were made by study size. The panel weighed the evidence based on a thorough examination of the threshold or magnitude of the treatment effect. Each evidence statement (other than those with no available evidence) and each recommendation is categorized by a level of evidence which ranges from A to D. Table ES-1 summarizes the categories of evidence by their source and provides a definition for each category.