著者

町原 英 白旗 慎吾

出版者

一般社団法人日本PDA製薬学会

雑誌

日本PDA学術誌 GMPとバリデーション (ISSN:13444891)

巻号頁・発行日

vol.1, no.2, pp.77-81, 1999 (Released:2006-08-03)

参考文献数

4

---

100% inspection can not completely eliminate all of defect qualities from the lot, because of the limitation of detection rate. Three models of 100% inspection methods, which have different stopping rules, have been statistically compared by taking the detection rate into consideration. The three models of inspection methods are defined as follows. (A) Stop when we detect no defects for the first time. (B) Stop when we detect no defects for the second time. (C) Stop when we meet the first successive two no-detections of defects. It is found that, in view of the average number of total insepction times, the rule A is the best whereas in view of the average number of undetected defects the rule C is the best. The rule B stands between the rules A and C and is close to rule C. If we also consider the costs of inspection and undefected defects, the rule B is robust and when the probability to detect the defects is high then the rule A is excellent. The rule C is found to be too conservative.