Objectives: Assessment of worker's exposure is becoming increasingly critical in the pharmaceutical industry as drugs of higher potency are being manufactured. The batch nature of operations often makes it difficult to obtain sufficient numbers of exposure measurements and occupational exposure models may be useful tools in the exposure assessment process. This paper aims to describe further refinement and validation of an existing deterministic occupational exposure model to predict airborne exposure of workers in this industry.
Methods: Workplace exposure assessment data (n = 381) containing all the contextual information required for the exposure model were collated from a multinational pharmaceutical company. The measured exposure levels ranged from 5 × 10⁻⁷ to 200 mg m⁻³ for largely task based samples, and included a range of handling activities, local control measures and abnormal operating conditions. Model input parameters for local control measures and handling activities were refined to reflect pharmaceutical situations.
Results: The refined exposure model resulted in good correlations between the log-transformed model predictions and the actual measured data for the overall dataset (r(s) = 0.61, n = 381, p < 0.001) and at scenario level (r(s) = 0.69, n = 48, p < 0.001). The model overestimated scenarios with measured exposure levels < 0.1 mg m⁻³ (r(s) = 0.69, bias = 0.71, n = 46, p < 0.001), and underestimated scenarios with higher measured concentrations ( > 0.1 mg m⁻³) (r(s) = 0.59, bias = -4.9, n = 33, p < 0.001). Including information on the refined sub-parameters improved the correlations, suggesting the uncertainty in the model parameters was partly responsible for the bias.
Conclusion: Further scientific data from the pharmaceutical industry on model input parameters, particularly on the efficacy of local control measures, may help improve the accuracy of the model predictions. The refined exposure model appears to be a useful exposure assessment screening tool for the pharmaceutical industry.