Spicher G, Peters J
Zentralbl Bakteriol Orig A. 1975 Jan;230(1):112-38.
The subject of control of sterility in homogenous lots by means of random sampling is discussed. The probability that non-sterile lots may be declared as accepted, was calculated by means of hypergeometric, binomial, and Poisson distribution. The following factors were considered: size of samples, size of lots, and degree of contamination of lots. The values obtained were reproduced in graphs (Figs. 1, 2, and 4). Reference is made to the conditions that have to be met in each case to enable a use of the various distribution functions for a calculation of the acceptance probability of non-sterile lots (Table 1). Furthermore, the necessary size of samples to contain at least one contaminated unit was calculated by means of binomial distribution for probabilities of 90, 95, and 99% (Fig. 3). For example, from a lot having a degree of contamination of 10%, 30 units would have to be assayed for sterility to recognize non-sterility of such a lot in 95% of cases. A degree of contamination of 1% would reguire already 300 units to be examined...
本文讨论了通过随机抽样控制同质批次无菌性的问题。利用超几何分布、二项分布和泊松分布计算了可能将非无菌批次判定为合格的概率。考虑了以下因素：样本大小、批次大小和批次污染程度。所得数值以图表形式呈现（图1、图2和图4）。文中提及了在每种情况下为能够使用各种分布函数计算非无菌批次的合格概率所必须满足的条件（表1）。此外，通过二项分布计算了在概率为90%、95%和99%时包含至少一个污染单元所需的样本大小（图3）。例如，对于污染程度为10%的批次，必须检测30个单位的无菌性，才能在95%的情况下识别出该批次的非无菌性。污染程度为1%时，则需要检测300个单位……
