Objectives: To estimate a threshold value for the respirable quartz dust concentration and silicosis incidence (1/1, ILO 1980/2000) in the German porcelain worker cohort.
Methods: Forty incident cases occurred among 17,144 workers. Estimation of the relationship to cumulative respirable quartz dust exposures by using Cox regression with restricted cubic splines (7 knots or less) and fractional polynomials (degrees 5 or less). Threshold estimates were obtained by subtracting candidate values from the time-dependent concentration data (setting the result to zero if negative) and recalculating cumulative exposure.
Results: Best threshold estimate was 0.25 mg/m (95% confidence interval: 0.15 to 0.30 mg/m).
Conclusions: A concentration threshold for silicosis risk is plausible. A threshold Cox model fitted the data significantly better than a nonthreshold model, summarized the cohort information without a loss in extracted information and more simply than restricted cubic splines and fractional polynomials.