Premium
Decimal reduction value ( D ) from fraction negative experiments via maximum likelihood estimation: An enabling spreadsheet and its implications on methodology and current standards
Author(s) -
Moruzzi Guido,
Lou Yuqian,
Fritz Ronald D.
Publication year - 2022
Publication title -
journal of food process engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.507
H-Index - 45
eISSN - 1745-4530
pISSN - 0145-8876
DOI - 10.1111/jfpe.14164
Subject(s) - fraction (chemistry) , computer science , statistics , confidence interval , mathematics , reliability engineering , engineering , chemistry , chromatography
Resistance of biological indicators for sterilization in pharma and food applications is often determined via fraction negative experiments. The ISO 11138 standards prescribe the Holcomb–Spearman–Karber procedure (HSKP) for calculating D value using these methods. However, HSKP imposes limitations on experimental designs. While ISO 11138 requires inactivation kinetics to not significantly deviate from first order, the HSKP does not provide a step to check such departure, or its extent. Maximum likelihood estimation (MLE) is an alternative approach for D value determination from fraction negative experiments. Compared to HSKP, it allows researchers to tailor the experimental design to practical needs, potentially reducing overall sample size in the process. This work presents a development of a 2004 published worksheet for MLE calculation, providing tighter confidence intervals and an improved way for Improbability calculation. Ways to exploit the design flexibility of MLE approach are presented, including simulated experiments and examples on how to optimize experimental design according to various situational needs, also in cases where expected D values are of the order of one or a few seconds as is commonly encountered in the sterilization of aseptic fillers for food. Detection of deviation from first‐order kinetics using the enhanced Improbability methods is discussed; one of the examples provided by ISO 11138‐1 for D value determination by HSKP contains data that deviate significantly from first order.