Boosting the Immune System, From Science to Myth: Analysis the Infosphere With Google

Abstract

Background: The concept that one can "boost" immunity is a popular one. Although the only evidence-based approach to this is vaccination, the lay public is exposed to a wide range of information on how to boost immunity. The aim of this study was to analyze such information available on the Internet. Methods and findings: We visited 185 webpages returned from a Google search on "boost immunity" and classified them by typology (blogs, commercial, government, no-profit, news, professional, scientific journals) and by using standard indicators of health information quality (JAMA score, HONCode). We then analyzed their content in terms of disease and "boosters" mentioned. Commercial and news websites represented one third of the results each. Of the 37 approaches to boost immunity recorded, the top ones were diet (77% of webpages), fruit (69%), vitamins (67%), antioxidants (52%), probiotics (51%), minerals (50%), and vitamin C (49%). Interestingly, vaccines ranked 27th, with only 12% of webpages mentioning them. Conclusions: Commercial websites are an important component of the information available to the public on the topic, and thus contribute providing biased information.

Keywords: antioxidants; complementary and alternative medicine; google; immunity; internet; vaccines; vitamins.

Figure 1

JAMA score of webpages of different typologies. Data are median, IQR, minimum, maximum. Number of webpages for each group was: commercial, 60; News, 59; Blog, 18; Professionals, 16; Health portal, 10; Non-profit organization, 9; Other, 7; Government, 3; Scientific journals, 3.

Figure 2

Content analysis of the whole SERP (A) and the top-10 webpages (B). Data show the percentage of webpages (A, n = 185; B, n = 10) mentioning a specific approach in the context of boosting immunity. Only the approaches that were mentioned in at least 10% of total webpages are shown.

Figure 3

Differential representation of approaches to boost immunity by commercial webpages (A) and news webpages (B). Data are calculated as: 100^*(percent mention in commercial or news webpages – percent mention in the total 185 websites/percent mention in the total 185 websites. The expected value, in case of no bias should be zero. A value >0 denotes (blue) an overrepresentation (positive bias), a value <0 (red) an underrepresentation (negative bias). ^*Significantly over-represented by a Fisher test corrected for multiple comparison by the method of Benjamini–Hochberg with a false discovery rate set at 5%. For significance testing, the frequency a boost is mentioned by commercial or news webpage is compared with its frequency in the remaining webpages in the SERP, rather than with that in the whole SERP, to avoid comparing two groups with a significant amount of overlapping data.