Abstract
Safety performance evaluation is an important issue of industrial risk management. If this statement is well accepted by the community, how to evaluate this performance does not meet consensus. The traditional incident-based approach has been shaken in 2007, after the Texas City refinery explosion. Accident report has pointed out confusion between personal safety and process safety, and between lagging and lagging indicators. Large number of thoughts and discussions has emerged regarding the use of safety performance indicators to measure level of risk. An indicator can be defined as a measurement, qualitative or quantitative, which provides information, in a specific objective of interpretation, in a particular context, for a specific actor. Almost 10 years after the Texas City accident, literature is now enriched with many proposals of set of indicators, emerging from scientific articles and industrial guidelines. Faced with this diversity of sources and management tools, we highlight the following question. What are the criteria for comparison, differentiation, and selection of these tools? Our study is based on an in-depth analytical review of the literature. As a result, several criteria seem to emerge: models used to design indicators, object of study, objectives of interpretation, terms of implementation and revision, actors affected by the outcome, and terms to validate the scientific relevance and operational feasibility. Multiplication of tools gradually improves the monitoring and driving of safety. However, our study shows that several criteria for indicators still need to be discussed, both theoretically and practically.
