[Design and validation of a job-exposure matrix to silica].

OBJECTIVE To develop a population-based matrix of job-exposure to crystalline silica in Brazil and to estimate its validity. METHODS An epidemiologist and an industrial hygienist developed a matrix of job-exposure in four stages: coding of occupation variable; coding of industry variable; consensual exposure classification between researchers; and estimate of registered workforce in 1995 for each level of exposure. The cross-tabulation of the variables industry (25 columns) and occupation (347 lines) resulted in 8,675 cells, classified according to silica exposure in four levels: non-exposed, possibly exposed, probably exposed, and definitively exposed. For validating the job-exposure matrix, five industries (mining and quarrying, construction, foundry, management of technical personal and textiles), were re-coded according to exposure by external experts. Reliability of the study and external experts was evaluated by agreement measured using kappa analysis. RESULTS The job-exposure matrix showed high coding agreement, ranging from 64.0% for foundry to 94.0% for mining. Kappa analysis showed good agreement in mining (0.9), and low or average for other sectors (ranging from 0.1 to 0.5). High specificity was found in foundry (86.5%) and mining (100.0%). Construction had 56% specificity. CONCLUSIONS The study job-exposure matrix showed good accuracy and seems to be appropriate for estimating silica exposure among Brazilian workers.