Close Calls and Lives Saved: A Content Analysis of Exemplars in User-Generated Commentary on Mammography News Coverage

Theoretical Background and research questions/hypothesis:  Changes in breast cancer screening recommendations have created confusion and controversy.  Multiple prior studies have examined the nature of news coverage of mammography but have neglected comments generated by readers.  Comments on online news stories are increasingly common, and research shows that comments may affect readers’ beliefs and behaviors.  Comments with exemplars (example cases) may be particularly persuasive, as suggested by exemplification theory and theories of narrative transportation. Because media shape the health information environment, understanding the composition of comments is necessary to fully understand the messages to which online mammography news consumers are exposed.  This research investigates the nature of news coverage of mammography from 2009 to 2014, with an emphasis on the content of comments on online news stories about mammography.  Specifically, it addresses the following research question: How prevalent and representative are mammography exemplars in mammography news articles comments?

Methods:  All articles and blog posts about mammography published by The New York Times between November 2009 and December 2014 were collected via LexisNexis and the NYT search engine and coded for inclusion (included N = 71), valence toward mammography and existence of exemplars.  All comments from included articles were collected and a stratified random 20% sample (N = 1,185) was coded for inclusion criteria, valence toward mammography, and existence of exemplars.  Acceptable inter-coder reliability (using Krippendorff’s alpha) was achieved for these variables.

Results:  The majority of news articles (54%) were neutral or balanced in their coverage of mammography.  Roughly one-quarter of articles (27%) included a mammography exemplar, with individuals with mammogram-detected breast cancers being the most common type of exemplar (42%). Thirty-one percent of comments included a mammography exemplar.  Of these, 41% included an individual with a mammogram-detected cancer, 19% included a false positive exemplar, and only 7% included exemplars reporting a history of normal mammograms.  This distribution is not representative of actual mammography outcomes for women of screening age.  Comments including a mammogram-detected breast cancer exemplar were mostly pro-mammogram (85%), while comments including a false-positive mammogram exemplar were frequently cautious toward mammography (69%).  Finally, analyses demonstrated that articles with mammography exemplars were more likely to have comments that also included mammography exemplars (OR = 1.65, p = 0.024).

Conclusions:  Many articles and comments about mammography include details from a personal experience with mammography.  Within mammography exemplar comments, mammogram-detected cancer exemplars are over-represented, false positives and normal mammograms are underrepresented, and valence toward mammography is related to exemplar type.  This research also suggests that using exemplars in articles may increase the tendency of readers to include exemplars in comments. 

Implications for research and/or practice:  Because exemplars affect both risk perceptions and behavioral intentions, an over-representation of mammogram-detected cancers may lead to skewed risk perceptions and harmful effects on screening intentions in comment readers. News organizations and health organizations that allow user-generated comments may want to consider the representativeness of comments when moderating and/or highlighting comments.  Additionally, because using exemplars in articles tends to increase the sharing of exemplars in comments, writers should be aware of this effect when creating content.