This project’s aim to build an evidence base is not a novel endeavor, even in the field of injury information. A long time ago, John Graunt (1620–1674) first explored demographic statistics based on church records of baptisms and burials. In the field of statistical history, he is regarded as the “founder of demography”, and may also be considered to be a pioneer in the field of injury data science. Graunt analyzed the number of burials during the 60-year period from 1604 to 1664, and constructed a table that categorized these deaths according to 81 different causes of death. However, the causes of death had been recorded by “uninformed persons in the role of coroner”, and therefore could not be effectively used. Despite this, Graunt was able to discern a form of regularity in the injury data, as he found that several of the causes of death maintained a fixed ratio with the total number of burials. He also constructed a table of casualties, and noted that similar phenomena could be observed in drowning, suicides, and various types of accidents in that there was a sense of regularity in accident-related deaths. His insight that it may be possible to observe a certain degree of regularity in injury occurrence through data analysis may be considered the dawn of injury data science. Furthermore, Graunt also conducted estimates on population vital statistics using what is known as a “life table”. Edmond Halley went on to refine this life table, which formed the mathematical basis of life insurance.

In the middle of the 19th century, approximately 200 years after Graunt’s era, several cities in Europe began conducting population surveys. As ever, the causes of death varied widely, which underscored the need to scientifically develop a universal classification of these causes. In this context, the International Classification of Diseases and Related Health Problems (ICD) was created. The development of this scientific classification system enabled, for the first time, the accurate extraction of information from raw data.

H. W. Heinrich (1886–1962), who is famous for his eponymous Heinrich’s Law, may be considered the one who established a new frontier on the quantitative evaluation of accidents. Heinrich was an engineer in the Travelers Insurance Company, and analyzed big data concerning insurance claims for industrial accidents. He also discovered several other empirical laws in addition to his famous 1:30:300 rule, and emphasized the importance of safety measures to management staff by enabling the visualization of industrial accidents. Since then, evidence data concerning industrial accidents and injuries have become an important policy tool for the labor bureaus of various countries, and have grown to be the foundation of industrial safety administration. The establishment of the International Labour Organization (ILO) has advanced the international standardization of data standards concerning industrial accidents, and enabled the development of industrial accident databases that allow for international comparisons.

When England was gripped by a cholera outbreak in 1831, John Snow (1813–1858), a physician, focused on the relationship between the origin of infected patients and the position of a communal water well. Snow realized that the infection was not airborne, but instead spread through drinking water. As a result, effective measures to prevent the further spread of infection could be implemented.

Shortly after, Florence Nightingale (1820–1910) was a nurse at a hospital in Scutari, which was located at the frontline of the Crimean War. Nightingale perceived that the poor hygiene conditions at the field hospital were responsible for many of the soldiers’ deaths. She prepared mortality statistics at the field hospital, and used these as concrete evidence to protest against the stubborn army commander’s refusal to admit that the hospital was operating under terrible conditions. Nightingale’s proposed measures took immediate effect, and the soldiers’ mortality rates dropped rapidly.

The activities of Snow and Nightingale took place several decades before the discovery of various etiologic agents in succession by Robert Koch and others. In other words, these activities created viable measures based on data concerning the occurrence of diseases combined with environmental factors that preceded medical analyses of causal relationships. Thus, epidemiology was born.

In the field of medicine, elucidating the pathological causes of a disease is conducted in parallel with examining the epidemiological causes of disease. William Haddon (1926–1985), an engineer in the US National Highway Safety Agency, demonstrated that it was possible to adopt an epidemiological approach to identify causes in the elucidation of accident occurrence processes in addition to examining the engineering and human factors. The surge in traffic accidents in the 1960s had become a massive societal problem in the US. To address this problem, Haddon formulated what is now known as the “Haddon Matrix”, which is a framework that enables the comprehensive assessment of traffic accidents and strategies for intervention. This matrix is an epidemiological framework that facilitates an extensive review of accident occurrences, where the vertical rows represent the timeline (pre-accident, accident occurrence, and post-accident) and the horizontal columns represent various factors that contributed to an accident (human factors, vehicle and equipment factors, and environmental factors). Based on this framework, Haddon proposed various measures—such as the use of seatbelts—that have been effective in reducing injuries. He presented startling data on the probabilities of being involved in a traffic accident throughout the life cycle of an automobile, where 1 out of 30 cars would have an accident resulting in bodily harm, and 1 out of 5 cars would have an accident resulting in harm to a passenger. This called attention to the importance of crash safety design. This information was presented at a combined meeting between the Texas branch of the Society of Automotive Engineers (SAE) and the regional medical society. Despite the lack of graphs and figures, the persuasiveness of the report lay in the provision of actual numbers for key points that highlighted the need for improvements in the automobiles themselves. While epidemiological frameworks had been previously limited to diseases and public health, Haddon expanded their application to accidents, and these were later adopted by the World Health Organization.

During the 17th century when Graunt lived, people only had records of baptisms and deaths. However, those of us who live in today’s society are subject to a wide variety of records, including accidents and injuries. Through the application of these records, we are able to obtain highly accurate information on the occurrence of injuries in everyday life.

The US experienced a high number of accidents due to defective products in the 1960s. In response to this situation, the Consumer Product Safety Act was rapidly developed and enacted. This act stipulated the initiation of the National Electronic Injury Surveillance System (NEISS), which was designed to systematically obtain information not only on industrial and traffic accidents, but also on accidents that occur in living spaces (at home, at leisure, at school, etc.). Based on rigorous statistical sample design from approximately 6,000 emergency care hospitals located nationwide, a sample of 130 hospitals was selected; injury data and analysis were then initiated based on a coding manual. (Incidentally, US government statistics are almost entirely based on statistical inference, and rely little on descriptive statistics involving the entire population. This is in marked contrast to Japan, which favors surveys that use complete enumeration.)

At present, the system collects information on approximately 370,000 injuries per year, which indicates that 13,000,000 injuries occur annually when extrapolated to all hospitals in the US. An official of the Consumer Product Safety Commission estimated there to be a total of approximately 34,000,000 injuries when accounting for the injuries not treated at hospitals. These data are available online for download and analysis. While the number of recorded items is limited, the data can be analyzed as big data due to the codification of information such as age, sex, location of accident, type of treatment, injured body part, and category of the product that caused the accident. Other databases include the Injury and Potential Injury Incident (IPII) database, which contains near miss information (approximately 30,000 cases per year); the Death Certificates (DTHS) database, which includes information on fatalities due to products (approximately 8,000 cases per year); and the 4-stage In-Depth Investigation (INDP) database (approximately 8,000 cases per year).

The integration of markets in Europe in 1985 advanced efforts among countries to collaborate and share injury information. Previously, each country had its own surveillance system for traffic, industrial, and living space accidents. These systems were integrated, and all injury information was consolidated as big data. In the European Union (EU), the Rapid Exchange of Information System (RAPEX) is an emergency alert system for product recalls. This system enables the sharing of information throughout the EU, where information on a high-risk product identified in one region can be transmitted to the rest of the union.

In today’s society, a large proportion of daily products (such as textiles, toys, commodities, and home appliances) used throughout the world are manufactured in China. The lifestyles of the world’s consumers are, to a large degree, dependent on Chinese products. Some time back, the book A Year Without "Made in China" was a bestseller in the US, and the author concluded that living without Chinese products was prohibitively difficult. In this context, global consumers have great interest in the question, “Just how safe are Chinese products?”

Authorities responsible for consumer product safety in Japan, the US, and Europe have collaborated with China’s General Administration of Quality Supervision, Inspection and Quarantine to establish a feedback channel for product recall information to Chinese manufacturers concerning their products in overseas markets. It is hoped that this will function as an incentive for improvements.

In addition, consumers within China are also increasingly concerned about product safety, with calls for manufacturers and authorities to also focus on domestic consumers. In response, the government implemented the National Injury Surveillance System (NISS) in 2007 to collect injury information from throughout the country. Similar to the US and EU, the information collection channel is composed of emergency care hospitals. While information had been gathered from 22 hospitals located throughout 18 provinces until 2013, the number of hospitals has currently increased to 54. Although information on 50,000 product-related injury cases has been gathered, the use of these data is limited and not available to all users.

While the previous examples of injury information systems from other countries were based mainly on information from emergency care hospitals, the system in Japan is substantially different from those of the US and Europe in that it uses a wide variety of information sources, such as businesses, consumers, consumer affairs centers, hospitals, and newspaper articles. Since the 1970s, several injury information systems have been developed in Japan, including the National Consumer Affairs Center of Japan’s PIO-NET (information sources include consumer complaint information and hospitals) and the product accident information database of the National Institute of Technology and Evaluation (NITE; information sources include reports from businesses). With the inauguration of Japan’s Consumer Affairs Agency in 2009, the various injury information systems were consolidated under this agency’s accident information databank system. As this system was not established in accordance with a single coordinated design principle, there are variations in the recorded items and classifications. Therefore, while it may be described as unified, this database is, in fact, an amalgam that lacks systematicity. The systems in the US and Europe involve data input at a location (i.e., hospitals) distant from where the accidents occurred, but Japan’s information sources tend to be nearer the accident location, which enables relatively detailed descriptions of the accidents. In addition, the NITE database has the exceptional practice of conducting verification experiment operations.

Under the serious accident reporting system implemented in 2007, businesses are mandated to report serious accidents (such as fatalities, accidents resulting in serious injuries, and fires) to the Consumer Affairs Agency. Despite this mandate, the number of accidents reported to the Consumer Affairs Agency has remained at approximately 20,000 cases per year, which is thought to cover only a portion of all accidents that occur in Japan. The US NEISS receives 370,000 reports per year, and uses sampling design to estimate the total number of accidents throughout the country. In contrast, the data in Japan are not collected based on scientific sampling methods, and therefore cannot be used to produce national estimates.

For the same reason, it is unclear if the aggregate results according to age groups and location are indicative of overall trends. The following is an international comparison of these estimates from claims data.

There is a large amount of big data on injury information that remains dormant and unused in Japan. For example, ambulances are dispatched approximately 6 million times (in 2014) per year; of those, approximately 1.6 million of those dispatches involve patients with injuries or poisoning victims. The condition of each transported patient is recorded by the paramedics, who also receive recorded feedback concerning the patients’ diagnoses and treatments from the hospitals. Some of this information has been digitized and consolidated by the Fire and Disaster Management Agency (FDMA). Next, approximately 60,000 fires are reported each year. After each fire has been contained, an on-site investigation of the cause is conducted; the results are recorded in the detailed fire investigation report. Also, there are approximately 1 million traffic accident reports filed by police stations annually. Information on medical treatments for injuries and poisoning are recorded on patient charts in both outpatients and inpatients. In addition, detailed information in insurance claims data are also recorded. There is a massive amount of such information, with approximately 600 million cases of injured or poisoned patients alone each year. In cases when injuries or poisoning lead to fatalities, physicians must fill in a death certificate and submit a notice of death to the appropriate municipal government. The annual number of these cases exceeds 40–50,000 (Accidental deaths, 2014).

Statistical data based on these records are reported in white papers or annual reports by the FDMA and the police, or as part of population vital statistics. Although FMDA or police white papers include various analyses conducted based on these accident records, the individual records are not released for analysis, and remain dormant within the files and computers of these government agencies.