This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Public Health and Surveillance, is properly cited. The complete bibliographic information, a link to the original publication on http://publichealth.jmir.org, as well as this copyright and license information must be included.
Plague is a highly infectious zoonotic disease caused by the bacillus
In this study, we aimed to assess public reaction regarding the recent plague outbreak in Madagascar by quantitatively characterizing the public’s interest.
We captured public interest using Google Trends (GT) and correlated it to epidemiological real-world data in terms of incidence rate and spread pattern.
Statistically significant positive correlations were found between GT search data and confirmed (
An earlier digital Web search reaction could potentially contribute to better management of outbreaks, for example, by designing ad hoc interventions that could contain the infection both locally and at the international level, reducing its spread.
Plague and history have always been strongly interrelated since the earliest description of plague pandemic [
This highly infectious zoonotic disease is caused by the bacillus
Diagnosis of plague is made based on the above-mentioned clinical findings accompanied with suspected history of exposure. Identification by culture as well as increased paired serum titers are suitable diagnostic methods. Plague can be fatal if left untreated. Besides supportive therapy for severely ill patients, treatment is composed of systemic antibiotics of aminoglycoside-based therapy.
Though highly associated with the past, plague still represents a global public health concern. Cases of plague continue to be reported worldwide, especially in Africa but also in Asia, South America, and even in the United States. According to the World Health Organization (WHO), 3248 cases of plague were reported from 2010 until 2015, with 584 deaths [
However, despite such a long-standing and rich history, it is rather difficult to get a comprehensive overview of the general situation. Within the framework of electronic health (eHealth), in which people surf the internet more and more looking for health-related material, new information and communication technologies, such as Web 2.0, portable computers, mobile phones and devices, as well as social media and social networks, could enable researchers to get a wealth of data, which could complement traditional surveillance of infectious diseases.
In this study, we aimed to assess public reaction regarding the recent plague outbreak in Madagascar by quantitatively characterizing this interest and correlating it to epidemiological
Google Trends (GT) is a free open-source tool used to track and observe internet search activity [
The GT search tool has two options for searching keywords: searches can be performed by
The results given by GT are output as normalized values (ie, relative search volumes [RSVs]) rather than absolute, raw values. Every query is divided by the total searches performed in a given geographic region and time range and normalized to a scale between 0 and 100 based on the topic’s popularity in comparison to all searches carried out in that region and time frame. For further details concerning GT, the reader is referred to Nuti et al’s review of GT and its potential applications in the medical field [
In this study, the second search option (ie, searching by topic with related terms) was used. Specifically, we searched for “Plague (Topic).” Searches were geographically limited to Madagascar. In Madagascar, the language spoken is Malagasy, while the second official language is French. English is spoken by less than 20% of the population. It is widely known that the sample when using online queries cannot be representative; however, in our case, searching for “Plague” and selecting the
Correlational analysis and multivariate regression models for nowcasting and forecasting, with lags up to 7 days, were performed based on the GT results with the number of confirmed, suspected, and probable cases of plague as reported by the WHO situation report. Different regression models were run, computing the different fitting parameters, including
Statistical analyses were performed with the commercial software XLSTAT 2017 (Addinsoft). All values with
Average plague-related search activities, expressed as RSVs, are shown superimposed on the trends of new suspected, probable, and confirmed cases of plague in
The best nowcasting model in terms of AIC values (see
Time trends of plague cases (confirmed, probable, and suspected) and plague-related Google Trends (GT)–generated data. All data are normalized for comparison purposes. RSV: relative search volume.
Scatterplots of the correlations between epidemiological values and the Google Trends (GT)–generated data related to the recent plague outbreak in Madagascar. RSV: relative search volume.
Spatial heat map of Google Trends–based plague-related data. The figure is based on the map provided by Google Trends. 1: Antsiranana Province; 2: Mahajanga Province; 3: Toamasina Province; 4: Antananarivo Province; 5: Toliara Province; and 6: Fianarantsoa Province. Color gradient correlates with the volume of plague-related Web searches.
Plague outbreak in Madagascar has drawn wide public attention shown here by our findings based on large Web search activity data analysis. Madagascar has been known as an endemic area of plague in its bubonic form with annually reported cases from April to September, generally across rural areas. However, the recent outbreak is characterized by pneumonic plague occurring in larger and more crowded cities. The current outbreak, known to be a highly contagious form of plague and in combination with the recent spread to the Seychelles islands, has its own distinctive features.
Monitoring and analyzing Web search activity manifested by novel data streams (NDS), especially during outbreaks, is of great importance in terms of surveillance as shown by O'Shea [
GT, based on Google search, is a freely accessed website tool, which provides data on how often a specific search item (ie, plague) is searched relative to total search volume worldwide or in specific areas and in different languages. For instance, in 2009, during the peanut butter-associated outbreak of
In our study, bursts of searches of plague-related topics corresponded both spatially and temporally with the outbreak’s spatiotemporal trends across the region studied (ie, Madagascar). The role of Web-based NDS for outbreak surveillance is crucial for workers in the field of public health and safety. Plague-related digital behavior as captured by GT analysis reflected rapid public response to the pneumonic plague outbreak in Madagascar, with some minor search peaks occurring even before the formal declaration by the WHO. Moreover, this reaction seemed to decline rapidly afterward, whereas the WHO continued to release the report of additional confirmed plague cases.
In our study, it is interesting to note that the potential influence of prior awareness of a clinical case of plague, which occurred on September 13, 2017, on search behaviors of a population was reflected by the rapid increase of searches found on September 14, 2017. From September 30, 2017, people were probably more able to recognize specific signs and symptoms related to plague due to news or public campaigns. In this case, the suspicion of disease may lead people to seek confirmatory Web information, contributing to the increase of the activity of internet users. These arguments could be used to explain the highest value of
Findings from the regression analyses showed the feasibility of exploiting NDS for predicting (ie, nowcasting and forecasting) plague cases. Extant predictive models of plague are usually built within the
Despite its novelties, which are among the major strengths of the current investigation, our study suffers from some limitations, which should be properly recognized. The shortcomings include the fact that GT provides relative and not absolute values, thus hindering the possibility of further refining and processing them. Moreover, GT captures only Web searches carried out with the Google search engine, which is, on the other hand, the most utilized search tool. Another drawback was the relatively low values of
Our study has shown an increase in digital Web searches with a unique pattern induced by the recent outbreak of plague in Madagascar. GT plays a highly important role in outbreak tracking and monitoring, in that it can capture public reaction and interest toward infectious disorders in real time before cases are formally communicated by the WHO. This earlier digital Web search reaction could potentially contribute to better management of outbreaks, for example, by designing ad hoc interventions that could contain the infection both locally and at the international level, reducing its spread.
Regression analyses for the nowcasting models.
Fitting parameters of the nowcasting models.
Regression analysis of the best forecasting model.
Fitting parameters of the forecasting models.
Akaike Information Criterion
electronic health
Google Trends
novel data streams
relative search volume
World Health Organization
None declared.