Abstract
Introduction:
analyzing the temporal trend of occupational accidents incidence is important for long-term prognostics. However, cyclical and seasonal changes, such as informal work and the underreporting of accidents, can affect it and need to be considered. The Hodrick-Prescott (HP) filter method, used to assess the dynamics of macroeconomic variables, can provide a more precise analysis of the evolution of variables subject to these fluctuations than methods commonly used in epidemiology, such as that of moving averages (MA).
Objective:
to test the HP filter as trend extraction method for the variable “incidence of occupational accidents”.
Methods:
comparative analysis between the HP filter and the MA method to extract the long-term trend of the following variables: typical accident, commuting accident, occupational disease, and total accidents in Brazil between 2008 and 2013.
Results:
both methodologies indicated similar trends of the variables over the period. However, the analysis of the long-term trend by the MA method may be more affected by short-term fluctuations. Conclusion: the HP method allows to more consistently infer the medium-and long-term trend of the variable “incidence of occupational accidents”, which can make it more suitable for proposing and managing public policies.
Keywords:
work accident; occupational disease; statistical analysis; time series studies; mathematical computing
