Mayaro virus (MAYV) was first discovered in Trinidad in 1954 and isolated by Charles Anderson and collaborators from the blood of 5 febrile forest workers [1-3]. MAYV is similar to the chikungunya and Semliki Forest viruses, being a linear, positive-sense, single-stranded, enveloped RNA virus, more specifically an arbovirus of the family Togaviridae and of the genus Alphavirus) [4]. MAYV is generally spread by the Haemagogus mosquito [5], but it can also be spread by Aedes aegypti and Aedes albopticus mosquitoes, which appear to be competent vectors as well [6,7]. Rubber workers are particularly at risk of developing MAYV infection.

In the past, small and occasional outbreaks have been described mainly in the Amazon basin of Brazil and a few other parts of South America [8-11]. The first epidemics were reported in 1955 in Brazil and Bolivia [12]. The reemergence of MAYV is cause for great concern for both public health organizations and communities. Anthropogenic changes in ecosystems and environments, due to a variety of phenomena, including urbanization, globalization and migration, agricultural intensification, and deforestation, together with displacement of populations and invasion of wildlife habitats by humans and domestic animals, are playing a major role in MAYV reemergence. Climate changes, economic downturns, and poverty are further drivers of the reemergence of neglected tropical diseases [13]. Therefore, eliminating mosquito breeding sites constitutes an important preventive measure. Efforts to strengthen and improve pathogen surveillance technologies are also fundamental in programs to control disease.

MAYV infection is a nonfatal and generally self-limiting disease causing arthritis in the knee, ankle, and small joints of the extremities, generalized myalgia, frontal headaches and photophobia, vertigo, nausea and epigastric pain, and chills, followed in two-thirds of patients by a fine maculopapular rash affecting the trunk and the extremities. In some cases, MAYV infection can persist up to 2 months.

In May 2007, an outbreak occurred in Chuquisaca Department, Bolivia, and involved 12 persons [14]. In January 2010, a French tourist, after a 15-day trip in the Amazon forest, Brazil, reported MAYV infection [15]. In 2011, MAYV was diagnosed in a 27-year-old male Swiss tourist returning from Peru [16]. MAYV disease was also imported into the United States by 2 infected people who had visited eastern Peru [17] and, more recently, into the Netherlands by a couple infected during their holidays in Suriname [18].

In June 2010, an outbreak occurred in Venezuela, with 69 cases in Ospino, Portuguesa state, and 2 additional cases in San Fernando de Apure, Apure state, on June 7, for a total of 71 confirmed cases as of June 8 (out of the initially reported 77 cumulative cases) [19]. A single case of MAYV infection in an 8-year-old child with fever and abdominal pain was described in rural Haiti (in the Gressier-Léogâne area, 20 miles west of Port-au-Prince) in 2015 [20].

In conclusion, MAYV has been so far isolated in humans, wild animals, and mosquitoes in Bolivia, Brazil [21,22], Colombia, Costa Rica, French Guiana [23], Guatemala, Guyana, Panama, Peru [24], Suriname, Trinidad, and Venezuela [25].

Given the emerging importance of MAYV [26-28], we here aimed to explore the feasibility of exploiting a Web-based tool for monitoring and tracking cases of MAYV infection.

Google Trends (Google Inc) is a freely available, online tracking system that, properly using keyword(s), enables a visualization of hit-search volumes in terms of relative search volumes (RSVs). In more detail, for each keyword or string of keywords, searches can be performed using search term or search topic strategies. With the first option, Google Trends tracks and monitors the exact text typed by the user. The second strategy, instead, consists of an exhaustive and systematic collection of all searches semantically related to the given query. Generally, the second search option results in broader findings.

A Google-based approach seems to be particularly useful to monitor infectious diseases epidemics [29]. Pelat and collaborators [30], as well as Valdivia and Monge-Corella [31], documented the usefulness of using Google Trends in capturing influenza and chickenpox outbreaks. The nowcasting or forecasting approach has been used also for other tropical diseases, such as malaria [32], Ebola [33], West Nile virus [34], and dengue [35].

In this study, we systematically searched Google Trends from its inception (January 2004 through to May 2017), using as keywords “virus Mayaro,” “Mayaro virus,” “virus de Mayaro,” and “virus del Mayaro.” We carried out this investigation according to the guidelines and recommendations put forth by Nuti and coworkers [36].

We built an ad hoc database of cases of MAYV by extensively mining Google, Google Scholar, the scholarly literature (PubMed or MEDLINE, Scopus, Scientific Electronic Library Online, and Latin American and Caribbean Health Sciences Literature), epidemiological alerts (from the US Centers for Disease Control and Prevention, European Centre for Disease Prevention and Control, World Health Organization, and Pan American Health Organization), HealthMap, and ProMED-mail reports.

Since Web searches can be prompted by different external or environmental cues (media coverage, education system, etc), we carried out a multivariate regression analysis according to the following predictive model:

We measured “scientific interest” by counting the number of MAYV-related articles indexed in PubMed or MEDLINE in the study period, using scientific production as a proxy of the interest of the scientific community toward MAYV; “epidemiology” was the number of confirmed MAYV cases; we assessed “media impact” as the number of MAYV-related news items released in the openly available news aggregator Google News as a proxy of the media coverage and influence over public opinion; and ε is the intercept of the model.

This model was theoretically inspired by the extant scholarly literature on infodemiology and infoveillance [37-40] and, in particular, on Google Trends, as well as by a study by Segev and Baram-Tsabari [41], which systematically investigated different search patterns in terms of the roles of the media and the education system.

We chose the best model according to the goodness-of-fit statistical model.

All statistical analyses were performed with IBM SPSS (version 24.0; IBM Corporation). Figures with a P value less than .05 were considered statistically significant.

Figure 1 shows the MAYV-related RSV trend. A burst in search volumes can be noticed in the period from July 2016 (RSV=13%) to December 2016 (RSV=18%), with a peak in September 2016 (RSV=100%). Before this burst, the average search activity related to MAYV was very low (median 1%).

Figure 2 and Table 1 show the countries with major search volumes. MAYV-related queries were concentrated in the Caribbean.

According to the best multivariate regression model, both scientific interest and media coverage had an impact on seeking behavior, with negative and positive effects, respectively (Table 2,Figure 3,Figure 4).

Textbox 1 lists the top MAYV-related and rising queries. These mainly related to MAYV infection symptoms (eg, fever, influenza-like symptoms), its carriers and vectors of transmission (eg, mosquitoes of the Aedes genus), and other similar tropical diseases (eg, yellow fever, malaria, dengue, chikungunya).

Table 3 reports the different models and their goodness-of-fit statistics.