Abstract
A novel idea of statistical analytical procedure for the level of detection (LOD) was demonstrated by its successful application to qualitative test methods for Salmonella, SALX System and NIHSJ-01. Salmonella cells of a hydrogen sulfide-producing strain FSD287 and a non-hydrogen sulfide-producing strain FSD347 were added to beef and shrimp food samples using a cell sorter to achieve bacterial cell concentration of 1, 5, and 10 cfu/25g-test portion (tp). The colony forming probability (CFP) of the added cells was estimated by means of 10×10 sorting plates. All of test portions containing FSD287 or FSD347 were decided to be positive by SALX System. NIHSJ-01 using CHROMagar® Salmonella (CHS) decided test portions of each of both strains to be positive, while NIHSJ-01 using desoxycholate hydrogen sulfide lactose (DHL) agar decided selectively only FSD287-test portions to be positive. All blank test portions were negative. To evaluate the level of detection at 95% probability (LOD(95)), in addition to test results, we introduced virtual results of blank conditions approaching zero. As a result, LOD(95) for every case was estimated to be lower than 0.326 cfu/tp indicating that both methods were able to detect 1 cfu/tp at higher than 95% probability. Therefore, our protocol for statistical analysis for LOD was feasible for the verification of the test methods that meet the requirement of detecting small number (minimum 1 cfu/25g-tp) of target micro-organisms in food test portions.