Validation of exposure information in occupational epidemiology.

Exposure data may be inadequate mainly for two reasons; they may not comply with a reasonable exposure-effect model or they may be biased. In the use of historical data it is essential to keep track of what the data were supposed to mean when collected in the first place. Most measurements of air contaminants in work places are probably carried out to establish compliance or non-compliance with limit values. Sometimes measurements are also made to evaluate control measures. Unfortunately, measurement data acquired in this way may have very little to do with the true average exposure of employees. The influence of trends and variations in production on the exposure level is substantial. This influence is similar and coincidental in widely different industries. In addition the variations attributed to production levels, there can be important variations due to season.