SRA-E Lisbon 2017

In order to monitor product-related accidents, the major industrialized countries have constructed injury surveillance system and product recall notification system in recent decades. This study proposes a new tool for describing 12 product-related accident databases stored at the Accident Information Databank (AIDB) of Japan. The tool is named Product-related Risk Information Description Framework (PRIDF), which was developed to integrate these databases. PRIDF is composed of five categories of elements (host, vector, agent, environment, and consequences), 50 attributes (seven of which are supplementary information given in the remarks section), and a set of vocabularies for the description of these attributes. PRIDF combines two conceptual models: an epidemiological model set forth in the World Health Organization’s (WHO) Injury Surveillance Guidelines and a technical model presented in the safety-related standards of International Organization for Standardization (ISO) and International Labour Organization (ILO). The epidemiological framework composed of host, agent, vector and environment provides us a useful framework for better understanding of product-related risks in a society-level context. At the same time, injuries to consumers caused by various types of hazards are technical and microscopic processes in nature, and their proper description requires a detailed technical framework and appropriate vocabularies, including lists of hazards, classification of products and mechanisms of accident/injury. Needed vocabularies are collected from international sources like ISO and WHO, and are tailored for our purpose. In this study, the advantages of the proposed PRIDF are discussed through its application experiments to four major product accident/injury information systems operating in Japan and Europe: Japan’s NCAC reporting system (PIO-NET), Japan’s NITE Accident Database (NITE-ADB), and the EU’s Injury Database (EU-IDB) and RAPEX alerting system. The proposed PRIDF will be useful in the data integration work needed by AIDB in future, and in the data mining work for product accident/injury prevention at two aspects: macroscopic data mining to identify social risk factors and microscopic data mining to identify technical-level risk factors. Although our proposal is based on Japanese experiences and designed primarily for Japanese users, the extensive use of internationally recognized classifications and vocabularies makes PRIDF a worthy candidate for the integration of accident/injury data in other countries. While international consensus exists for the descriptive frameworks of mortality statistics and occupational health and safety statistics, no such unanimity exists for product accident data. The proposed PRIDF can be used as a candidate to fill this gap.