This paper is in the following e-collection/theme issue:
An Internet-Based Epidemiological Investigation of the Outbreak of H7N9 Avian Influenza A in China Since Early 2013
1School of Public Health and Primary Care, The Chinese University of Hong Kong, Hong Kong, Hong Kong, China (Hong Kong)
2The Shenzhen Municipal Key Laboratory for Health Risk Analysis, Shenzhen Research Institute of The Chinese University of Hong Kong, Shenzhen, China
*these authors contributed equally
Jin-Ling Tang, PhD
School of Public Health and Primary Care, The Chinese University of Hong Kong, Hong Kong
4/F, School of Public Health and Primary Care, Prince of Wales Hospital, Shatin, New Territories
China (Hong Kong)
Phone: 852 22528779
Fax: 852 26453098
1]. The Internet has removed a major constraint to accessing and sharing data, information, and knowledge. Unlike traditional media, it provides an open platform that allows various people to report, confirm, and correct details regarding issues of public concern. For example, “Internet mass hunting,” which literally means to uncover the true identity of a particular person or matter through the coordinated efforts of all “netizens,” is an approach to using the Internet for conducting investigations . In China, this approach has been increasingly employed by the public to elucidate issues of public concern, such as cases of corruption .
The Internet also provides novel opportunities for health research. Eysenbach developed a method for analyzing Google queries to track cases of influenza-type illnesses in a given population [4,5]. That method was employed to accurately predict weekly influenza activity in each region in the United States with a reporting lag of approximately 1 day . Health surveys and clinical trials have also been conducted through the Internet [7-10]. Other examples of Internet-based research in the health sciences include Infovigil, HealthMap, MedISys, BioCaster, and EpiSPIDER [11-15]. Compared with traditional field research methods, such as face-to-face interviews and paper-based questionnaires, Internet-based methods involve considerably fewer resources, including money, time, and human resources. The Internet is particularly useful for obtaining information and influencing behavioral responses during public crises , such as during outbreaks of new infectious diseases.
In early 2013, a new type of avian influenza, H7N9, emerged in China, becoming a matter of strong public concern. After the first case was reported, H7N9 rapidly became a widely discussed topic on the Internet. A considerable volume of relevant information was made publicly available on the Internet through various channels, including news reports, discussion forums, personal blogs, and reports from hospitals and government authorities. In this study, we aimed to describe the outbreak of H7N9 in China based on data available on the Internet by February 10, 2014, and to explore the methods, feasibility, validity, advantages, and limitations of employing the Internet-based approach to investigating public health issues by comparing our findings with those presented in a well-conducted conventional field epidemiologic study on the H7N9 outbreak .
Study Structure and Definitions
We collected publicly available Internet data related to the H7N9 outbreak from reliable websites. The field epidemiologic study by Li and colleagues  was based on the data they reported to the National Center for Disease Control and Prevention of the country (China CDC) and was employed as a reference for validating our study. In our study, we first summarized the epidemiological and clinical characteristics of H7N9 cases based on the data from the Internet updated until February 10, 2014. To validate our study, we compared our findings with those in Li’s study . Li’s results were compared with ours based on data with the same cutoff date of investigation (December 1, 2013) and with our results updated to February 10, 2014.
In Li’s study, suspected and confirmed cases of H7N9 virus infection were defined according to the definition of H5N1 cases recommended by the World Health Organization (WHO) in 2006. Suspected cases were identified through China’s surveillance system for cases of pneumonia of unexplained origin. If laboratory tests, such as real-time reverse transcriptase–polymerase chain reaction assay, viral isolation, and serological testing, indicated the presence of the H7N9 virus, the case was considered confirmed. Once a suspected case was identified, initial field investigations were conducted and respiratory specimens were obtained by local CDCs. Information on each confirmed case was collected through field investigation until December 1, 2013.
In our study, confirmed H7N9 cases were those that claimed to be newly identified H7N9 cases reported on either government or nongovernment websites, and verified either by laboratory tests or nationally or provincially organized specialists.
Sources of Internet Data
Various H7N9-related data were reported through numerous Internet-based resources. The data in this study were obtained from only 2 website categories: government and major media websites. These categories are detailed as follows, in the order of their trustworthiness:
Websites of government organizations providing H7N9-related information in China were the primary source of data. The 5 most representative government organizations that provided our needed information were (1) municipal health bureaus; (2) national, provincial, and prefectural CDCs; (3) the National Health and Family Planning Commission; (4) the Ministry of Agriculture; and (5) the WHO.
Social media websites were used as supplementary sources of data. Websites that set up a special column or discussion forum for the outbreak of H7N9 Avian Influenza A were given a higher priority than others. In fact, most of our supplementary data were obtained from the 2 most popular public websites in the country: Sina and Sohu. Both had a special section for H7N9 on their websites [18,19]. They provided numerous information, including live reports of new H7N9 cases, individual data for confirmed H7N9 cases, simple summary data of the outbreak, prevention and treatment of H7N9 Avian Influenza A, and comments from netizens.
Governmental data were considered most trustworthy and contributed most of the data in this study. Information was sought from social media websites when some data were missing from government websites.
We included 2 cases reported on Health Authority websites in Hong Kong, but did not search Taiwan-based websites and thus the few cases in Taiwan were not included in this study. No case was reported in Macau.
Case Identification and Data Collection
To avoid relying on or repeating the summary or aggregate results of official reports and other epidemiological studies by those who had access to official individual data on H7N9 cases, we searched and extracted only those data on individual cases from websites with free public access. These data were mainly included in daily reports published on relevant government authority websites. Websites were selected according to usefulness, accessibility, and credibility. Usefulness was determined by the comprehensiveness and up-to-datedness of the information. To obtain data for each confirmed H7N9 case, we searched daily reports published on the CDC and other health bureau websites at the national, provincial, and prefectural level from March 31, 2013 (the date of the first reported case) to February 10, 2014. Complementary searches of social media websites, predominantly Sina and Sohu, were also performed to supplement data obtained from the government websites. Cases were identified based on the date of illness onset, family name, demographic data (gender, age, and region), and exposure history. Duplicate cases were defined as those with an identical date of illness onset, family name, gender, age, and region. When cases were reported on multiple websites, we used the data from government websites of the highest level from national to provincial to prefectural.
Data Extraction and Quality Control
Two researchers (MYD and YYY) extracted the data for each confirmed case independently by using a self-designed data extraction form. Discrepancies were resolved by double-checking the websites and discussions if deemed necessary. Extracted data included (1) demographic data, such as age, sex, rural/urban residency, and occupation; (2) epidemiological data, such as potential exposure history to the H7N9 virus, the number of close contacts, secondary cases, familial aggregation cases, and confirming method for diagnosis; and (3) clinical data, such as hospitalization, intensive care unit (ICU) admission, development of acute respiratory distress syndrome (ARDS), death, and dates for these clinical outcomes (see Multimedia Appendix 1).
We employed descriptive and analytic statistics to summarize the epidemiological and clinical characteristics of confirmed H7N9 cases. SPSS version 18.0 (SPSS Inc, Chicago, IL, USA) was used to perform all the statistical analyses.
Epidemiologic Characteristics of Confirmed Cases by February 10, 2014
From February 17, 2013, to February 10, 2014, a total of 334 cases of H7N9 infection were identified. Cases occurred in the following 15 regions (Figure 1): (1) Zhejiang (130 cases); (2) Guangdong (62 cases); (3) Shanghai (42 cases); (4) Jiangsu (42 cases); (5) Fujian (20 cases); (6) Hunan (10 cases); (7i) Jiangxi (6 cases); (8) Anhui (6 cases); (9) Henan (4 cases); (10) Beijing (3 cases); (11) Guangxi (3 cases); (12) Shandong (2 cases); (13) Hong Kong (2 cases); (14) Hebei (1 case); and (15) Guizhou (1 case).
The majority of cases were obtained from official sources, with 89.5% (299/334) from government websites, 4.5% (15/334) from nongovernment websites, and 6.0% (20/334) from both government and nongovernment websites. Table 1 summarizes the epidemiologic characteristics of the confirmed cases. The median age was 58 years, with 33.9% (111/327) older than 65 years and 1.8% (6/327) younger than 5 years. More men (219/327, 67.0%) than women were reported as confirmed cases of H7N9. Most of the cases (186/278, 66.9%) lived in urban areas. Seven cases (7/201, 3.5%) were poultry workers. Five cases had comorbidities including hypertension, heart disease, diabetes mellitus, and chronic bronchitis. Among the 91 confirmed H7N9 cases with information on the method of diagnosis, 78.0% (71/91) were confirmed using nucleic acid detection and the remaining 20 cases were confirmed by a group of infectious disease specialists.
Information on recent exposure to animals was available for 56 of 334 confirmed cases of H7N9, 39 of which were patients with a history of exposure to animals. Cases with an animal exposure are detailed as follows: (1) 3 patients reported 1 instance of exposure to poultry; (2) 10 patients reported multiple instances of exposure to poultry; (3) 3 patients were exposed to pigeons, quails, and pet birds; and (4) the remaining 23 cases did not report the type of animals to which patients were exposed.