Effect of metasite selection on the quality of World Wide Web information: A collection development approach to the evaluation of Web-based consumer health information on the treatment of hypercholesterolemia

Purpose. Determine if there is a relationship between type of search engine and the quality of Web-based information retrieved on a consumer health topic, using hypercholesterolemia as an example. Test the validity of certain filters (authority of source, recent revision date, quantity of text, and number of hyperlinks to relevant information) commonly used to evaluate the quality of Web-based health information.; Setting/subjects. Two general search engines, Lycos (relevance ranked) and Yahoo (hierarchical classification), and two health portals, HealthAtoZ (relevance ranked) and Healthfinder (hierarchical classification) were used to generate a collection of 120 Web sites on the topic of cholesterol.; Methodology. Prospective, causal-comparative. A multimodal approach was used to measure the quality of this information.; Research questions. (1) How accurate is Web-based information? (Experts evaluated the content using a scoring instrument based on the recommendations of the National Cholesterol Education Program and current evidence-based guidelines.) (2) How comprehensive is this information on the topic of hypercholesterolemia? (MeSH was used to index the Web sites and compare their relative subject coverage.) (3) What is the reading level of this information? (Flesch Reading Ease and Flesch-Kincaid Grade Level were used.) (4) What languages are available to the reader? (5) Is there a significant difference in the quality of information (accuracy, subject coverage, reading level, and language availability) retrieved by the four different types of metasites? (6) Is the accuracy of Web-based cholesterol information related to either the authority of the content provider or the revision date? (7) Is the comprehensiveness of cholesterol information related to either the number of text words or the number of relevant hyperlinks included on the Web site?; Conclusions. Both health portals provided significantly more reliable information than either commonly recommended filters or general, inclusive search engines. The interaction between quality control and subject classification appears to be a desirable feature in a search engine. Healthfinder outperformed all other types of metasites in identifying accurate and complete cholesterol information that was easy to read and understand.