The Abortion Web Ecosystem: Cross-Sectional Analysis of Trustworthiness and Bias

Background: People use the internet as a primary source for learning about medical procedures and their associated safety profiles and risks. Although abortion is one of the most common procedures worldwide among women in their reproductive years, it is controversial and highly politicized. Substantial scientific evidence demonstrates that abortion is safe and does not increase a woman’s future risk for depressive disorders or infertility. The extent to which information found on the internet reflects these medical facts in a trustworthy and unbiased manner is not known. Objective: The purpose of this study was to collate and describe the trustworthiness and political slant or bias of web-based information about abortion safety and risks of depression and infertility following abortion. Methods: We performed a cross-sectional study of internet websites using 3 search topics: (1) is abortion safe?, (2) does abortion cause depression?, and (3) does abortion cause infertility? We used the Google Adwords tool to identify the search terms most associated with those topics and Google’s search engine to generate databases of websites related to each topic. We then classified and rated each website in terms of content slant (pro-choice, neutral, anti-choice), clarity of slant (obvious, in-between, or difficult/can’t tell), trustworthiness (rating scale of 1-5, 5=most trustworthy), type (forum, feature, scholarly article, resource page, news article, blog, or video), and top-level domain (.com, .net, .org, .edu, .gov, or international domain). We compared website characteristics by search topic (safety, depression, or infertility) using bivariate tests. We summarized trustworthiness using the median and IQR, and we used box-and-whisker plots to visually compare trustworthiness by slant and domain type. Results: Our search methods yielded a total of 111, 120, and 85 unique sites for safety, depression, and infertility, respectively. Of all the sites (n=316), 57.3% (181/316) were neutral, 35.4% (112/316) were anti-choice, and 7.3% (23/316) were pro-choice. The median trustworthiness score was 2.7 (IQR 1.7-3.7), which did not differ significantly across topics (P=.409). Anti-choice sites were less trustworthy (median score 1.3, IQR 1.0-1.7) than neutral (median score 3.3, IQR 2.7-4.0) and pro-choice (median score 3.7, IQR 3.3-4.3) sites. Anti-choice sites were also more likely to have slant clarity that was “difficult to tell” (41/112, 36.6%) compared with neutral (25/181, 13.8%) or pro-choice (4/23, 17.4%; P<.001) sites. A negative search term used for the topic of safety (eg, “risks”) produced sites with lower trustworthiness scores than search terms with the word “safety” (median score 1.7 versus 3.7, respectively; P<.001). Conclusions: People seeking information about the safety and potential risks of abortion are likely to encounter a substantial amount of untrustworthy and slanted/biased abortion information. Anti-choice sites are prevalent, often difficult to identify as anti-choice, and less trustworthy than neutral or pro-choice sites. Web searches may lead the public to believe abortion is riskier than it is.