Tweets were catalogued based on mentions of hydroxychloroquine, chloroquine, azithromycin, and remdesivir. Factba.se was used to note if any Twitter posts were archived or otherwise hidden by Donald J Trump [11]. Tweet Binder was used to assess average analytics data on key metrics, such as likes, retweets, and comments [12]. Tweet Binder further aggregated all tweets and averaged longitudinal tweeting patterns.

The time Donald J Trump spent discussing each treatment on television was recorded from the 5 largest US television stations: CNN (Cable News Network), C-SPAN (Cable-Satellite Public Affairs Network), Fox News, MSNBC (Microsoft/National Broadcasting Company), and BBC (British Broadcasting Corporation) News. The Global Database of Events, Language, and Tone was used to note White House briefings and the total monitored airtime of each station for broadcasts that aired from March 1 to April 30, 2020 [13]. Airtime data are presented as percentages of 15-second airtime blocks, which is the maximum the software can crawl its historical database. Manual crawling of television airtimes or news headings was not performed.

Google searches from March 1 to April 30, 2020 were indexed with Google Trends, focusing on searches performed within the United States [14]. Searches including the words “hydroxychloroquine,” “chloroquine,” “azithromycin,” “remdesivir,” and “covid treatment” were catalogued, along with the Google purchasing patterns for these items. Analyses for these searches were performed concurrently with cataloguing. However, the results for this period were not large enough to report. The data obtained were the proportional patterns of searches instead of exact amounts, given Google’s data agreement [15].

Amazon purchases were indexed based on the following words: “hydroxychloroquine,” “chloroquine,” “azithromycin,” and “remdesivir.” Search volume was assessed using Helium 10 software [16]. Estimated purchase amounts were calculated using Ahrefs [17].

Data were analyzed with Microsoft Excel. This included aggregating the data, cataloguing data, and formatting data with key points, such as Donald J Trump’s first press conference on March 19, 2020 or his numerous tweeting dates.

The wide variety of sampling, numerous software-defined metrics, and methods of measuring outcomes, such as gross estimates and proportions instead of true samples, did not lend themselves to statistical analysis [18]. The key assumption of transitivity, which assumes no important differences in the distribution of either potential effect modifiers or sampling techniques (eg, individual comparisons of price points between Google and Amazon), would be high for indirect comparisons.

Table 1 notes the characteristics of US president Donald J Trump’s tweets. Donald J Trump mainly focused on hydroxychloroquine, as per his own tweets (eg, his tweets on March 21, 2020) and his retweets of Twitter posts from other people who supported his claims regarding the efficacy of hydroxychloroquine. Various tweets included news articles (eg, the tweet on April 4, 2020), which were mostly from Fox News (4/6, 67%).

Based on our analysis with Tweet Binder [19], with an average proportion of nearly 100,000 likes and 20,000 retweets, Donald J Trump’s first tweet on March 21, 2020 advocating for the use of hydroxychloroquine and azithromycin was one of his most popular tweets. With 385,700 likes and 103,200 retweets at the time of this study, the tweet had a potential estimated impression reach (ie, the number of potential user views on the individual tweet) of 78,800,580.

Table 2 notes the number of times Donald J Trump mentioned COVID-19 treatments during White House briefings. He mentions hydroxychloroquine 37 times in total, most frequently on April 4, 2020 (9 times); chloroquine 12 times in total, most frequently on March 19, 2020 (7 times); azithromycin 8 times in total, most frequently on March 19, 2020 (3 times, while also mentioning hydroxychloroquine); and remdesivir 8 times in total, most frequently on March 19, 2020 (4 times).

Figure 1 displays the television airtime of broadcasts that mentioned the treatments. Airtime for mentions of chloroquine (Figure 1A) peaked on March 24, 2020 on Fox News (0.59%), MSNBC (0.78%), and CNN (0.61%), following the chloroquine-related death. Following Donald J Trump’s tweet advocating for the use of hydroxychloroquine, airtime for mentions of hydroxychloroquine (Figure 1B) peaked on April 4 and April 6, 2020 on Fox News (1.94% and 2.46%, respectively), while peaking on CNN on April 22, 2020 (2.88%). Fox News was one of the single television networks that continually provided coverage on azithromycin (Figure 1C), as airtime for its mention peaked at 0.45% on April 7, 2020. Remdesivir (Figure 1D) was not mentioned on any news organizations until April 30, 2020, followed by the decision by the US Food and Drug Administration to approve remdesivir as a COVID-19 treatment on May 1, 2020 [28]. At this point, airtime for mentions of remdesivir experienced peaks comparable to those of the other treatments. COVID-19 treatment coverage has been increasing on all networks, especially after the March 19, 2020 press conference.

Searches for and purchases (Figure 2A) of chloroquine on Google peaked on March 19, 2020, followed by a second peak on March 23, 2020, after the story of chloroquine-poisoning was made public. Searches for and purchases of hydroxychloroquine followed a similar trend, though there was a second peak on April 4, 2020 when Donald J Trump retweeted a studying discussing the apparent efficacy of hydroxychloroquine. Peaks in COVID-19 treatment searches (Figure 1E) largely coincided with Donald J Trump’s White House briefing on March 19, 2020, with 3 peaks correlating with his tweet history in Table 1 (March 19, March 21, and April 4, 2020).

Our Amazon purchase analysis (Figure 3) showed a peak in purchases for chloroquine, hydroxychloroquine, and azithromycin after the March 19, 2020 White House briefing. This increase in purchases was observed for various forms of these medications, whether it was for purchases of a book about azithromycin [29], which increased by 10 sales; herbal elements claiming to have hydroxychloroquine as an active element [30], which increased by an estimated 50 sales; or texts on alternative chloroquine-containing compounds, such as chloroquine phosphate, which increased by an estimated 30 sales [31]. Secondary peaks for searches on and purchases for hydroxychloroquine were observed after the tweet on April 4, 2020.