Medical record review (MRR) studies have been reported to make up 25% of all scientific studies published in emergency medical (EM) journals. However, unlike other study designs, there are no standards for reporting MRRs and very little literature on the methodology for conducting them. The purpose of this article is to provide the reader with methodological guidance regarding the strengths and weaknesses of these types of studies. Key words: medical records; retrospective studies; research design; evidence-based medicine. ACADEMIC EMERGENCY MEDICINE 2004; 11:187-192. WHAT IS A MEDICAL RECORD REVIEW STUDY?The term ''medical record review'' (MRR) refers to any study that uses prerecorded, patient-focused data as the primary source of information to answer a research question. The sources of information include: physician and nursing notes; ambulance call reports; diagnostic tests (e.g., electrocardiograms, radiographs, laboratory tests); clinic, industry, administrative, and government records; and/or computerized databases.As with any source of data for scientific study, the medical record must ''be capable of providing data that [are] both reproducible and valid.'' 1 This key point is seen as the primary weakness of medical record reviews.
WHY SELECT AN MRR STUDY DESIGN?Ideally, a study design is selected because it is the best method of answering the research question. The greatest advantage of the MRR design is that the data are already collected. Emergency health services research makes use of routinely collected data (e.g., emergency department [ED] visits and trauma databases) and chart reviews to address questions about the utilization, appropriateness, process, and outcome of care. Physicians often conduct case series studies based on reviews of charts. Most importantly, MRRs allow us to address research questions that cannot be answered in prospective trials such as: 1) the effects of harmful exposures to which people cannot be randomized; 2) the effects of potentially beneficial exposures to which people cannot be randomized; and 3) the occurrence of rare events in exposures to which people cannot be randomized.Also, there are purposes that are impractical to address with prospective studies, such as: 1) studies of patterns of disease or behavior (e.g., ED visits) over prolonged periods; 2) quality assurance studies; 3) studies involving the sharing of cases to create large (e.g., regional trauma) databases; and 4) pilot studies to provide information for planning prospective trials.However, using medical records will not be useful for measuring phenomena that are commonly not documented.2 Therefore, the convenience of precollected data does not make the MRR study the most appropriate design.
HOW ARE STUDY CASES BEST SELECTED FOR MRRs?The two primary differences between MRR and prospective studies are: 1) case selection-in the former, the cases have already occurred and the information on them is mixed with that of all the noneligible subjects; and 2) data quality-the data of MRRs were not o...