An analysis of the quality of research reports in the Journal of General Internal Medicine.

Forty-four "original articles" published in the Journal of General Internal Medicine were reviewed using a rating instrument with 13 items developed from previously published standards for the design, analysis, and presentation of medical research. Each item was rated on a four-point (poor-to-excellent) scale. Inter-rater agreement was good (86% agreement; kappa = 0.65). Seven items received excellent or good ratings in 80% or more of the articles, but six problem areas were identified: description of study scope (generalizability), informed consent procedures, reliability assessments, references for statistical procedures or computer programs, and the use of descriptive and comparative statistics. Guidelines for these and other criteria are offered to help researchers prepare clear and informative reports.