4.2. Headline Analysis

We captured headlines from articles published from 1999 through 2020 using the “US Newsstream” database as provided by proquest.com (accessed date: 15 September 2021; ProQuest LLC, Ann Arbor, MI, USA). The full description of the search is provided in Figure S1. To reduce redundancy the pool of articles was limited to those with a full text. The most important search parameters were the presence of the terms “West Nile Virus” or “Zika Virus” in the full text or the headline of the news article; the WNV and ZIKV searches were conducted independently. The results were imported to R objects for further scrubbing and analysis. The scrubbing strategy is included with the supplemental material (Figure S1).

Regarding “duplication”, we considered that a news article was not duplicated but a reprint, if the article was published in several news outlets (e.g., syndication) or if the article was published on different dates in the same news outlet because this “republication” represented the public’s interest in the topic as determined by the outlet’s editor. For example, the news article titled “Rains propel West Nile Virus surge in San Bernardino County” by Jim Steinberg [40] was published in “The Sun” on 7 October 2015 and 15 December 2015, and in the “Inland Valley Daily Bulletin” on 20 December 2015. We considered this situation to be three publications of a WNV article.

The frequency of news publications that mentioned WNV and ZIKV was evaluated by frequency overtime; their corresponding headlines using a word cloud. A word cloud (tag cloud or wordle) is a visual representation of terms and their relative importance. The terms are usually single words, in our case we joined the words “West”, “Nile”, and “Virus” when found together to form “wnv”. The same was true for “Zika” and “Virus” to form “zv”. We presented the frequency of the terms using different font sizes.