Using the World Wide Web to answer clinical questions: how efficient are different methods of information retrieval?

The World Wide Web (Web) has the potential to revolutionize information retrieval in medicine. However, the best method of information retrieval from the Web is not known. The purpose of our study was to compare medical search engines, general-purpose search engines, medical meta-lists, and commercial sites on the Web with regard to their efficiency in retrieving medical information. Ten questions were identified from a database of questions posed by primary care clinicians. Authoritative answers were identified. Searches were performed using 1 commercial site, 4 general search engines, 9 medicine-specific search engines, and 2 medical meta-lists. The main outcome measures were the number of questions answered by each Web site, the correctness of the answers, the number of links followed to get an answer, and how well documented the answer was using the Health on the Net criteria. MD Consult, a commercial site, answered 6 of 10 questions. Hardin MD (a meta-list) and Excite and HotBot (general search engines) each answered 5 questions. The medicine-specific search engines performed poorly, answering an average of only 1 question. MD Consult and HotBot required the least number of links to find an answer. MD Consult and Hardin MD had the best documented answers. Medicine-specific search engines on the Web fare poorly in answering clinical questions when compared with general search engines. MD Consult, Excite, HotBot, and Hardin MD found the greatest number of answers.