The correlation between Google trends and salmonellosis

Salmonella infection (salmonellosis) is a common infectious disease caused by Salmonella bacteria. Salmonella bacteria belong to gram-negative bacteria in the Enterobacteriaceae family [1]. This kind of bacteria lives in the intestines and is transmitted by the faecal-oral transmission route [2]. The foods that can be contaminated with Salmonella include raw meat, poultry, seafood, raw eggs, fruits, and vegetables [3]. The symptoms of Salmonella infection are classified as gastroenteritis, including nausea, abdominal pain, diarrhoea, vomiting, headache, fever, and blood in stool [4]. Typhoid fever is a serious intestinal infection caused by S. typhi [5]. The complications of S. typhi infection include dehydration, bacteremia, reactive arthritis, etc. A total of 2269 outbreaks and 48,178 salmonellosis were reported in the USA from 2009 to 2018. Therefore, it is important to prevent Salmonella transmission with appropriate methods, such as washing hands after using toilets, and avoiding eating raw eggs. Because the sources of Salmonella infection are everywhere, the transmission route (faecal-oral transmission) is difficult to break. In addition, healthy people are generally susceptible to this infection. Therefore, there would be many infected people worldwide.

Internet search is a new method to monitor the outbreak of infectious diseases [6]. The internet-based surveillance system using internet data is logistically advantageous and economical to identify term-related diseases. Traditional systems to monitor infectious diseases depend on laboratory tests, physician diagnoses, and data collection by the authorities. Thus, the reporting on the emerging outbreak of infectious diseases is often delayed by one to 2 weeks in the traditional surveillance system. In the last two decades, the internet has become an important medium for the general public, public health practitioners, and doctors to obtain health-related information [7]. Internet-based surveillance systems could forecast the outbreak of infectious diseases by following online web-based activities. Therefore, it seems to be a promising strategy to monitor disease outbreaks based on internet search behaviour. Valdivia et al. used Google Flu Trends to monitor the activity of influenza in Europe [8]. They found a relationship between the disease activity peak and flu-related internet search. Seifter et al. used Google Trends to predict the epidemiology of Lyme disease [9]. Some Google Trends terms such as “tick bite” and “cough” were considered indicators of Lyme disease epidemiology. Ayyoubzadeh et al. used Google Trends to evaluate the COVID-19 incidence in Iran [10]. The researchers concerned the first wave of COVID-19 in Iran. They found that Google search terms, including handwashing, hand sanitizer, and antiseptic topics, were related to COVID-19 prevalence in Iran. The associations between COVID-19 cases and the search mentioned above terms were related to media discourse on non-pharmaceuticals interventions to mitigate the pandemic. Internet-based surveillance systems could be a useful method to evaluate outbreaks of infectious disease. Google Trends analyses the popularity of top search queries in Google Search.

Previous studies have not identified the relationship between salmonellosis and Internet searches. In this study, we examined the relationship between salmonellosis and Google Trends in the USA from January 2004 to December 2017. We hoped to reveal the correlation between Internet search trend and salmonellosis. In addition, we explored the most important Google search terms that indicate salmonellosis with multiple regression analysis. We wish to reveal the predictors of salmonellosis. Furthermore, we discovered the relationship between Google search terms with Shigella and E. coli infection to validate the function of Google Trends. We prospected Internet search trend could provide a useful strategy to monitor the infectious disease outbreak.

Salmonella is a common food borne pathogen in the USA [12]. Foods are the source of most of Salmonella infections. Contaminated foods and water that look and smell untainted can infect healthy people [13]. In addition, Salmonella can be transmitted from pets to people and from people to people [14, 15]. Most Salmonella patients suffer from fever, diarrhoea, stomach cramps, nausea, vomiting, and headache [16]. Infections can be serious for infants and older adults [17, 18]. Because Salmonella infection is foodborne, it is also preventable. First, people should wash hands before and after preparing foods [19]. Second, different kinds of foods, such as raw meat, eggs, and seafood, should be separated.

In our study, we discovered the correlation between salmonellosis with Google search terms. We found that a total of 48 Google search terms were associated with salmonellosis. These terms represented different aspects of salmonellosis. The first group of Google search terms are associated with foods and fruits. The search terms in this category included foods, salmon, beef, vegetable, ground beef, chicken, cucumber, pork, fruits, avocado, salad, sandwich, sushi, tuna, cheese, blueberries, coleslaw, mango, tomato, tomatoes, melon, sea food, lettuce, ice cream, carneasada, watermelon, and chicken salad. A variety of foods and fruits were associated with salmonellosis outbreaks. This result indicated that Salmonella was very common in foods and fruits. Foods and fruits are considered one of most important sources of salmonellosis outbreaks [20]. The results of multiple regression analysis indicated that “blueberries” was a predictor of salmonellosis outbreaks. Blueberries have a relatively short shelf life compared with other kinds of fruits, such as melon and watermelon [21]. Previous studies found that Salmonella could grow on harvested blueberries at retail display temperatures, while Salmonella did not grow on strawberries under the same conditions [22]. The complicated production chain of blueberries might contribute to Salmonella contamination. Blueberries rely heavily on humans to harvest, which promotes Salmonella transmission. Many reported outbreaks of Salmonella were associated with blueberries [23]. Various methods have been developed to reduce Salmonella in blueberries. These methods include antimicrobial solution with freezing, ozone, and UV light [24, 25]. Apart from blueberries, “ice cream” was also a predictor of Salmonella outbreaks. Ice cream, especially homemade creams, contains raw eggs and milk. Raw eggs and milk might be contaminated with Salmonella. A nation-wide outbreak of salmonellosis has been reported to be transmitted from ice cream in the USA [26]. The researchers discovered that one brand of ice cream (Schwan’s) was responsible for the nation-wide outbreak of Salmonella infection. The FDA also detected Salmonella in ice cream production facilities in the US [27]. To avoid the risk of Salmonella infection, people should make ice cream with pasteurized egg products or pasteurized shell eggs. Other methods to make safe ice cream are to use cooked egg base or to prepare ice cream without eggs.

Apart from food-related Google search terms, we also found that salmonellosis were correlated with public places and activities that might be contaminated with Salmonella. Such places and activities included hotels, bathrooms, bars, parties, and bbq. In these public places and activities, healthy people might be infected from foods and facilities. For example, it was reported that sauces and salsa prepared at hotels in Dallas and Texas were considered vehicles for salmonella outbreaks. The investigation pointed out that hotel food workers infected with Salmonella were linked this salmonellosis outbreak, which affected 617 people from 46 states in the US [28]. In our study, the multiple regression analysis also showed that “hotel” and “poor sanitation” were predictors of salmonellosis. If public places such as hotels and bars have poor sanitation, people can easily be infected with Salmonella, leading to salmonellosis outbreaks.

Finally, we also found that “hypovolemic shock” was a predictor of salmonellosis outbreaks. Hypovolemic shock is a life-threatening symptom of Salmonella infection. In typhoid fever and paratyphoid fever, severe vomiting and diarrhoea cause electrolyte and liquid loss, leading to microenvironment imbalance [29]. Fluid and electrolyte imbalance was related to a decrease in arterial pressure and circulating blood volume. Hypovolemic shock and septic shock would occur in this situation. People with Salmonella infection suffer from severe vomiting and diarrhoea. They might be afraid that this situation would lead to hypovolemic shock. Therefore, the search term “hypovolemic shock” was correlated with salmonellosis.

Lastly, we tried to predict 2017 salmonellosis outbreaks based on 2004–2016 dataset with machine learning. However, the results of prediction of future outbreak based on the specific search terms were not meets expectation very well. We inferred some reasons might explain the results. Firstly, salmonella was transmitted by fecal-oral transmission route. Human and animals could be infected with salmonella by contaminated food or water. Salmonellosis outbreaks were not regularly. Secondly, our study only included the dataset of Google Trends. Other data might influence the Salmonellosis outbreaks such as weather, health status, population density were not enrolled in our study.

Our study mainly focused on the effect of Google Trends to monitor the salmonellosis outbreaks. The purpose of Big Data utilization is now shifting toward forecasting from monitoring. So, the directions of the studies in the future should focus on accurate and precise forecasting of infectious disease outbreak.

There have some limitations in this study. Firstly, we used the Internet search engine Google Trends to evaluate the results. Other Internet engines such as Twitter, Facebook have not enrolled in our study. Secondly, we analyzed the data monthly owning to the Google Trends provided the data monthly. The seasonal data has not been evaluated in this study. In the future, the analysis could be adjusted to seasonality if the seasonal data were available. Thirdly, the data of US was analyzed in this study. Other regions worldwide such as European, Asia have not been investigated. Lastly, the prediction of salmonellosis outbreaks has not meets expectation in our study. In the future, the prediction of infectious disease based on the specific search terms could be investigated if the raw data was expanded.

In this study, we discovered the correlation between Google search terms and salmonellosis in the US from 2004 to 2017. We investigated related Google terms with salmonellosis and identified most important indicators of salmonellosis outbreak. We found that Google Trend was a useful method to monitor salmonellosis outbreak. We also validated the Google Trends with Shigella and E. coli. Thus, we considered Google Search could be used to monitor infectious disease.