Customize your JAMA Network experience by selecting one or more topics from the list below.
Phillipov G, Phillips PJ. Frequency of Health-Related Search Terms on the Internet. JAMA. 2003;290(17):2258–2259. doi:10.1001/jama.290.17.2258
To the Editor: Wolfe et al1 recently reported that 53% of US adults use the Internet to seek health-related information. Measurement of Internet use, however, has generally been obtained by self-report, rather than by assessment of actual search behaviors.1,2 Because search engines represent the principal method by which people seek Internet-based information,3 we analyzed the frequency of use of search engine keywords for health-related subject matter.
During the period from September 9, 2001, through July 20, 2002, we evaluated the 300 most common queries in the Wordtracker Top 500 Keyword Report, which is published via e-mail each Saturday.4 We examined a total of 39 weekly reports, each including a list of the top 300 queries and the total number of queries during the surveyed 24-hour period. The mean (SD) 24-hour query rate was 5 100 000 (900 000) and accounted for all queries from 2 large Internet metacrawlers, Metacrawler and Dogpile. While the top 300 lists we evaluated had been filtered to exclude the bulk of pornography-related queries, such queries were not excluded from the total 24-hour activity rates. Wordtracker claims that 20% to 25% of all search engine queries relate to pornography.4
To standardize for the variation in total queries between individual reports, each report's top 300 keywords was divided by the total 24-hour activity rate and the result multiplied by 1 million, yielding a metric of "hits per million queries." Because of the potentially large number of possible health-related search terms, they were classified as health-related on an intuitive basis. From our initial inspection of the unfiltered top 300 reports, there seemed little likelihood of health-related terms being removed by the pornography filters. Moreover, the unfiltered top 300 lists, which contained many high-ranking pornography-related terms, significantly decreased the occurrence of most health-related terms.
The number of identified health-related terms was small. Such terms appeared infrequently and each typically ranked low within top 300 listings (Table 1). To place these scores in perspective, the maximum 24-hour query rate for a term during the study period was 3409 per million queries (Halloween, March 11, 2001), while queries on horoscopes ranked in all reports, and averaged 180 per million queries. Many of the health-related terms appeared to relate to transient media publicity, or to reflect other short-term publicity. For example, the extremely high anorexia query rate on December 8, 2001 (Table 1), followed a television show on pro-anorexia Internet sites,5 while the single appearance of acrylamide in the top 300 appeared at the same time as media reports about the high levels of this potential carcinogen in fried foods.6 General health-related queries (eg, health, women's health, and PubMed) were highest in the period after September 11, 2001, and markedly declined with time. For example, queries relating to health decreased from 96 per million queries (October 27, 2001) to 48 per million queries (July 20, 2002), while neither women's health nor PubMed appeared in top 300 lists after January 2002.
Our results suggest that information on both general health and specific disorders have a very low priority for people using Internet search engines. The validity of our approach is supported by the expected periodic increases in specific key terms associated with holidays, entertainment, and sporting events, as well as sudden surges in activity in response to news items. These results challenge claims of high consumer demand for Internet-based health information. Strengths of our study include its length and large sample, although we acknowledge that we were unable to assess participants' characteristics and demographics, as well as possible use of other search engines. The fact that only a single medical condition, diabetes, ranked consistently, is unexpected. Diabetes is a prevalent chronic disorder, for which current medical strategy is predominantly directed toward patient self-management and empowerment.7 However, there are other equally prevalent chronic disorders, such as hypertension and arthritis. Our results suggest that health-related terms are uncommon as search terms for Internet queries.
Create a personal account or sign in to: