食源性致病菌污染估计中删失数据分析的研究进展

食源性致病菌污染水平的确定是开展微生物定量风险评估的重要前提，而删失数据的存在易造成对食品中致病菌整体污染水平的估计产生偏差。对检测过程中出现的删失数据进行分析研究已逐渐成为食源性致病菌定量建模工作的重要内容之一。本文对国内外相关研究进行综述，介绍了食源性致病菌污染检测中删失数据的分类，比较了替代法、参数估计法、非参数估计法和多重填补法这4 类常用分析方法，简述了不同特征的致病菌污染检测数据集及相关统计学方法在食源性致病菌污染水平估计中的应用。最后，基于目前食源性致病菌污染水平估计中存在的问题进行探讨，指出降低估计结果不确定性的同时不可忽视检测数据的变异性，并对未来的风险监测、风险评估及风险交流相关研究作出展望。

Censored Data Analysis in Estimation of Foodborne Pathogen Contamination: A Review

The determination of foodborne pathogen concentration is an important prerequisite for quantitative microbial risk assessment (QMRA). The existence of censored data readily causes bias in the estimation of the contamination levels of pathogens in foods. The analysis of censored data has gradually become an important part of the quantitative modeling of foodborne pathogens. This article presents a comprehensive review of recent related studies conducted in China and across the world, introduces readers to the classification of censored data in the detection of foodborne pathogen contamination and compares four commonly used methods for censored data analysis, namely substitution, parameter estimation, non-parametric estimation method and multiple imputation. This article gives a brief overview of the application of pathogen contamination datasets with different characteristics and related statistical methods in the estimation of foodborne pathogen contamination levels. Finally, it discusses the problems currently existing in the estimation of foodborne pathogen contamination levels. While great efforts should be made to reduce the uncertainty of estimation results, the variability of detection data should not be ignored either. This review concludes with an outlook on risk monitoring, risk assessment, and risk communication in the future.