From medical records to clinical science.

Medical records contain an abundance of information, very little of which is extracted and put to clinical use. Increasing the flow of information from medical records to clinical practice requires methods of analysis that are appropriate for large nonintervention studies. The purpose of this article is to explain in nontechnical language what these methods are, how they differ from conventional statistical analyses, and why the latter are generally inappropriate. This is important because of the current volume of nonintervention study analyses that either use incorrect methods or misuse correct methods. A set of guidelines is suggested for use in nonintervention clinical research.