Take Home Messages• There are several open access health datasets that promote effective retrospective comparative effectiveness research.• These datasets hold a varying amount of data with representative variables that are conducive to specific types of research and populations. Understanding these characteristics of the particular dataset will be crucial in appropriately drawing research conclusions. IntroductionSince the appearance of the first EHR in the 1960s, patient driven data accumulated for decades with no clear structure to make it meaningful and usable. With time, institutions began to establish databases that archived and organized data into central repositories. Hospitals were able to combine data from large ancillary services, including pharmacies, laboratories, and radiology studies, with various clinical care components (such as nursing plans, medication administration records, and physician orders). Here we present the reader with several large databases that are publicly available or readily accessible with little difficulty. As the frontier of healthcare research utilizing large datasets moves ahead, it is likely that other sources of data will become accessible in an open source environment. BackgroundInitially, EHRs were designed for archiving and organizing patients' records. They then became coopted for billing and quality improvement purposes. With time, EHR driven databases became more comprehensive, dynamic, and interconnected.

he blind men feeling the elephant had two generally unappre-T ciated advantages. The beast didn't move and it didn't change.Physicians curious about access to online medical information databases are less fortunate. The online beast moves and it changes. Nevertheless, Pbysicians Online: Final Report of the Canadian Medical Association iNet Tral is a valuable asset. It is a report of an eight-month trial on the use of selected online medical information systems by practicing physicians. The systems (read vendors) selected were the American Medical Association/GTE Medical Information Network (MINET) and the Bibliographic Retrieval Systems ( B E ) which served as a portal of entry to the computerized versions of the Index Medicus (MED-LINE) and Excerpta Medica (EMBASE). Begun in early 1985 and published in April, 1986, the report is a compilation of the experiences of 24 users (23 physicians and 1 pharmacist) in 14 practice settings. Telecom Canada offered the iNet 2000 network to facilitate the accessibility by the user-physician to the systems under study. Such networks, called gateways, provide directories that help to identify which specific database contains the information the user is seeking.The evaluation data were obtained by interviews and questionnaires. The report contains several useful appendices: a descriptive list of the subunits in MINET, a step-by-step interaction depicting what the user sees on the video screen when using the BRS, a list of additional medical information vendors and a detailed comparison of the features, including costs, of six of them. Precisely because the medical information access systems are continuing to develop, periodic benchmark analyses are necessary. The Canadian report is an important update for 1986. Others'-3 have reported on the potential usefulness of similar systems at earlier developmental stages. All were in agreement that further studies are clearly needed to determine which sys- Q 1986 by The