The OECD MRL Calculator
Abstract:This paper provides a review of the regulatory statistical procedures currently in use in the EU and NAFTA for the calculation of the maximum residue levels (MRLs) based on pesticide residue measurements obtained from supervised field trials. Some theoretical and practical issues with these procedures are highlighted. We conclude that both the EU and the NAFTA procedures are scientifically based and reliable. Recommendations for improvement and harmonisation are also included. In order to set a maximum residue limit (MRL) for the use of a pesticide on a particular crop, a residue dataset is assembled by measuring the pesticide level in samples taken from field trials carried out according to the pesticide label. These residue datasets are usually left-censored (i.e. truncated at the limit of detection (LOD) or the limit of quantification (LOQ) levels), right skewed (i.e. asymmetric, having a long right tail) and contain outliers (extreme values that appear discrepant from the rest). Censored values represent loss of information and can greatly affect the calculation of certain statistical measures like the mean and the standard deviation. A long right-tailed dataset with values seven or eight times the size of the mean can complicate the classification of extreme values as outliers. There is great diversity in the appearance of residue datasets. Some are reasonably fitted well by a normal distribution, also called a bell-shaped or Gaussian distribution. Others are reasonably well fitted by a lognormal distribution (the logarithm of the residue values would be reasonably fitted well by a normal distribution), which is a right-skewed distribution. Finally, some residue datasets are so erratic that they cannot be fitted by any known distribution.
Document Type: Research Article
Publication date: April 1, 2011
More about this publication?