MEDICAL SELF-DIAGNOSIS AND THE WEB: RESULTS FROM ENTERING PLAIN LANGUAGE SYMPTOMS INTO A SEARCH ENGINE

Monday, October 24, 2011
Poster Board # 16
(Scientific Abstracts should report the results of original research related to diagnostic error in medicine and must contain quantitative or qualitative data. Each abstract should be 400 words or less, have a descriptive title, and the following 4 sections: background, methods, results, and conclusion; may include 1 table or figure. ) Scientific Abstract

Mary Moore, PhD, JoAnn Van Schaik, MLS and Kimberly A. Loper, MLIS, University of Miami Miller School of Medicine, Miami, FL

Background:  News articles frequently describe how Web search engines are used by health consumers to enter symptoms and seek self-diagnoses. The current study explored what happens when symptoms in lay language are entered into a commonly used Web search engine.

Methods: Researchers reviewed symptoms for 20 commonly occurring diseases and conditions listed in the Merck Manual and Lange's Current Diagnosis.  The most common symptoms for those diseases and conditions were converted to plain language by consulting a focus group of health consumers. Searches were conducted using a commonly used Web search engine, and the first three Web pages (first 30 postings) of results for each disease or condition were downloaded for analysis.  Contents of the sites were searched for the expected (but certainly not the only possible) diagnosis.

Results were analyzed for relevance, currency, and authority, consistent with the Health on the Net Code of Conduct (HONcode).  Additional notes were made for results that offered especially unusual suggestions, advice, or diagnoses.  Results: Of the 600 Websites analyzed, approximately 1/6th of the sites yielded terms consistent with the expected diagnoses.  However, the expected diagnoses did not routinely appear on the first page of results.  Most of the sites yielding expected diagnoses were .com sites and many attempted to sell the reader goods or services.  Of the sites displaying the expected diagnostic terms but not relevant for the purpose of relating symptoms to diagnosis, many were sites related to drug side effects.  Each search included some irrelevant and often misleading results. Some, but certainly not all, sites included disclaimers about using the site for self-diagnosis. 

Conclusion: This preliminary exploration indicated that these Web searches using common lay terms for symptoms did yield relevant information, however the reader must carefully evaluate results, cannot be assured of relevancy, and cannot assume the most relevant results will appear early in the list of search results.  Consumers must beware, weighing the relevance, authority, and currency of each Web result and hopefully seeking expert consultation from healthcare providers.