[A methodological approach for measuring personal exposure to noise].

A criterion is proposed for planning personal noise exposure surveys. For measurements taken at random over entire work shifts, the aim is to achieve preset confidence limits of the arithmetic mean of daily personal exposure (i.e., +/- 2 dBA or +/- 1 dBA). Five or six measurements are sufficient to estimate the standard deviation. By following simple rules to calculate the confidence limits for a population with unknown variance, it is possible to fix the number of measurements N with sufficient accuracy to achieve this goal. This number is a function of daily personal exposure variability. The choice of the confidence limits determines the accuracy of the personal noise exposure assessment, provided that it is the exponential mean of N measurements. This method allows the survey to be planned according to the desired accuracy of the final result. An example is provided of a survey carried out in a refractory brick factory, which shows that if the daily personal exposure range is 4-6 dBA, a 5-measurement survey is sufficient to achieve a +/- 2 dBA confidence interval, while 10-12 measurements are necessary to achieve a +/- 1 dBA confidence interval. If the range is 14-15 dBA the same results are achieved with 12-14 measurement in the first case and with about 50 in the second. The latter results shows that if the survey is stopped after 5 measurements, the probability of accepting a value of integrated personal noise exposure outside the N-integrated +/- 1 dBA range may reach 50%, mainly among groups of workers with greater variability in daily exposure.