Comparing rates of the health outcomes of interest between individuals with high exposure and those with low or no exposures. This is what we did in our study since the external reference group was not ideal. In such situation it must be a way of gradients in exposure. However, it is not likely that all occupational hazards pose gradient effects on human health (many industries show uniform exposure).