Le concept des coûts indirects des accidents du travail a de tout temps intéressé les intervenants en santé et sécurité au travail. Les auteurs tracent son évolution depuis la première étude d'Heinrich en 1931 jusqu'à ce jour. Ils constatent une très grande hétérogénéité dans les résultats obtenus par les études répertoriées. Ces écarts peuvent s'expliquer par des dissemblances au niveau des définitions, des méthodologies de recherche utilisées et des populations visées.
Sixty years ago Heinrich discovered that employers' costs of work accidents far exceeded the amounts typically attributed to such events. It therefore became accepted practice for occupational health and safety (OHS) specialists to add to known or «direct» costs, a multiple, representing hidden, invisible or «indirect» costs which are entirely absorbed by the firm.
This approach contains an implicit behavioral assumption that employer awareness of his total accident costs will increase the perceived returns on prevention activities and consequently lead to greater investment in accident reduction expenditures. The present description and analysis of the literature traces the concept of indirect costs from its origins up to the current period. The first section deals with the evolution of the notion of indirect costs. Then the economic aspects are surveyed. Finally, the definitions, methodologies and results of the empirical studies are analyzed.
Work accident cost research may be divided into two types: «primary» and «secondary». In the first, original cost data on individual accidents are collected using a questionnaire distributed to employers in one or more industrial sectors. The second type measures total accident costs by applying the information (ratios) generated in primary studies to other, available data on direct or known costs.
Many primary studies generated ratios (indirect costs/direct costs) to express the magnitude of accident costs. The size of the ratios found vary from 1:1 to 10:1 and more; the most common, that of Heinrich (1931), being 4:1. Other authors, observing no linear relationship between indirect and direct costs, have preferred different methodologies to express the importance of indirect costs (e.g. Simonds and Grimaldi, 1956). In addition, the most recent ratio studies suggest much more moderate results than those found previously.
1. Very few primary empirical studies have been carried out, the vast majority being of the secondary type. Of those primary surveys reported many are deficient since they reveal precious little on the population studied, methodologies, frequency tables and the actual statistical analysis of the data. These are often found as a chapter in a textbook or handbook on the management of OHS and therefore represent more of an exploratory treatment of the indirect cost concept than a rigorous, scientific procedure of systematic data-gathering, reporting and analysis. The application of their results for estimating real costs must thus be treated with a certain degree of caution.
2. The indirect cost coefficients generally in use today by OHS practitioners and by authors of secondary studies are based on research dating from as far as 1931 (Heinrich) and 1955 (Simonds). The ensuing technical and organisational changes in the workplace have rendered such results largely obsolete.
3. There are substantial variations in the definitions of direct and indirect costs as well as other aspects of research methodologies, making comparisons of the results difficult. Standardization of these aspects would permit more useful comparative studies.
4. Many studies are based on very small populations of accidents or on limited intra- or inter-sectorial representativeness. Sector-specific coefficients generated from multi-sector data would permit more accurate estimates of the extent of indirect costs since these can be expected to vary with firm size, technology, victim characteristics, production standards and organization, etc.
5. No broadly-based, scientific study of indirect work accident costs has ever been carried out in Canada. This is a serious weakness in our ability to estimate total accident costs particularly since relatively high Canadian incidence and severity rates warrant such an undertaking (see Brody, Rohan and Rompre, 1985). The present authors are currently carrying out a study of this kind on some 400 work accidents in Quebec using advanced statistical techniques such as multiple regression analysis.