Abstract

Occupational burden of disease studies to identify priorities for regulatory interventions, and to estimate the impact of these interventions, are becoming more frequent. There are various ways to estimate burden of disease in a population, each method’s appropriateness depends on available data and the research question to be answered. We review the available methods, and assess their usefulness for predicting future occupational disease burden and for testing population-level interventions.

There are several approaches to disease burden estimation: Population Attributable Fraction (PAF) applied to incident or prevalent disease cases, Lifetime Risk/Future Excess Fraction and Disease Projection (Age-Period-Cohort and structural model/’g’ formula) approaches. The PAF takes into account past exposures, with intervention effects estimated from the point of intervention onwards. Lifetime risk estimates disease that workers, currently exposed to a disease-causing agent, will contract in their lifetimes; no latency assumptions are required, with interventions assumed to take effect immediately. Disease projection methods use current and past disease rates to predict rates into the future. The ‘g-formula’, a generalised regression model, is used to estimate the joint distribution of outcome and risk factors; Monte Carlo methods can then simulate disease risk following intervention, compared to no intervention. The methods are tested on examples of occupational disease/exposure pairings, including lung cancer from respirable crystalline silica exposure and COPD from vapour, gases, dust and fume exposure, and results are compared.

The results will provide critical information on the sensitivity of some examples of occupational burden of disease and health impact studies on the modelling approach.