Since its emergence in late 2019 in Wuhan, China, COVID-19, the disease caused by the novel coronavirus SARS-CoV-2 has drastically impacted daily life around the world [1,2]. Among the changes to the public sphere include government-imposed lockdowns of businesses, schools, and universities, designed to mitigate the spread of the disease and to alleviate the significant strain on health care systems [2]. As many continue to stay at home due to the COVID-19 pandemic, internet use has become an increasingly dominant part of daily life. In a recent poll, a majority of Americans considered the internet “essential” during this time [3]. Nearly all major internet services have seen increased traffic since early March 2020 [4]. Given the unprecedented nature of the pandemic, there is naturally much public uncertainty regarding COVID-19, and thus, many are turning to the internet to ask their questions and obtain information about the coronavirus.

Previous studies have shown that patients frequently use the internet to research their conditions and inform their discussions in clinic [5,6]. As physicians, insight into what topics patients are curious or anxious about may help guide and structure our interactions, leading to improved patient rapport and satisfaction. Additionally, with well-publicized recent examples of misinformation originating from many sources, including places of authority, it is paramount for physicians to collectively take responsibility to provide reliable and trustworthy information based on the best available evidence [7,8]. Thus, the aims of the present study were to (1) determine the most commonly searched phrases related to COVID-19 in the United States and (2) identify the sources of information for these web searches. In doing so, we believe that we can distill the collective curiosity of the internet-using public into useful information for physicians in clinic.

Search terms related to COVID-19 were entered into Google web search using a clean-installed Google Chrome browser on May 30, 2020, in New York, NY. Google web search is by far the most widely used internet search engine in the United States [9]. In 2018, Google introduced a natural language processing algorithm, which greatly improved the ability of the search engine to identify clusters of search queries related to any given topic [10]. Due to this new technology, Google redirects all searches related to COVID-19, such as “COVID-19,” “coronavirus,” “coronavirus disease,” etc, to a centralized COVID-19 homepage. This search results page incorporates the location of the user’s search and generates a list of questions and websites that are frequently associated with the initial query. On each results page, the 200 most commonly asked questions were generated. The questions were downloaded to a database using a freely available program (Scraper, version 1.7). The specific question and web address were noted on the webpage by their unique XML Path Language (XPath) strings.

The questions were first categorized according to Rothwell’s classification system by a single trained reviewer [11,12]. This classification system was expanded into one of 6 topics—clinical signs and symptoms, treatment, transmission, cleaning methods, activity modification, and policy—which were chosen based on previously published studies that examined the web and social media concerns of users during the COVID-19 pandemic [13-15] (Table 1).

Activity modification consisted of questions regarding the effectiveness of various activities or lifestyle changes in preventing COVID-19. Policy included questions detailing local or national policy changes enacted in response to COVID-19, including questions about economic support. A full listing of the criteria for each topic category is listed in Table 1.

In line with previous studies, the websites were categorized according to source: World Health Organization (WHO), Centers for Disease Control and Prevention (CDC), non-CDC government, academic, news, and other media [16,17] (Table 1). Specifically, non-CDC government websites consisted of webpages directly maintained by a national governmental entity such as the National Institutes of Health (United States) or the National Health Service (United Kingdom). Academic websites were defined as an organization with a clear academic mission statement. Other media consisted of websites not described by one of the previous categories, including CNET, WebMD, and Wikipedia. A full listing of the criteria for each web source category is listed in Table 1.

In total, 200 questions and their corresponding source of information were extracted; the top 25 questions are listed in Textbox 1. The most common question topic was transmission (n=63, 31.5%), followed by clinical signs and symptoms (n=54, 27.0%) and activity modification (n=31, 15.5%) (Table 2).

Most questions regarding the transmissibility of the coronavirus asked about specific modes of transmission such as spread through food, feces, air conditioning units, delivery packages, and mosquitoes. Interestingly, the clinical signs and symptoms category captured questions about myths associated with the disease, such as whether consuming alcohol stops the coronavirus. In the activity modifications category, there were many questions about staying at home, wearing masks, and managing pre-existing travel plans. The most commonly asked question—“Can antibiotics treat the coronavirus disease?”—was classified as treatment, which, in total, comprised 9% (n=18) of the searched questions.

With respect to sources of information, the most common websites provided were maintained by the CDC (n=73, 36.5%), academic medical organizations (n=48, 24.0%), and the WHO (n=47, 23.5%) (Table 2). With an additional 5% (n=10) of web information provided by a government source, an overwhelming majority of information (n=178, 89%) came from highly trustworthy web sources. However, the remaining 11% (n=22) of information came from either news or other media. In particular, 4.5% (n=9) of information came from web sources classified as other media, which included potentially erroneous sources of information such as Wikipedia.