Article
Sample size calculation for dioxin measurement in eggs: Is it possible to reduce the number of required measurements by pooling of samples?
Search Medline for
Authors
Published: | August 29, 2017 |
---|
Outline
Text
How large is the dioxin intake for consumers using organic eggs bought at the market? How large is the intake for consumers who prefer barn eggs from discounters?
To answer these questions, representative consumer specific pools of eggs should be sampled and analysed, applying a minimal number of measurements that is still appropriate for estimating the expected value of dioxin contamination with a given precision. For this purpose, the effect of pooling egg samples was studied by adapting methods for sample size calculation in microarray experiments as described by Kendziorski et al. (2003). Estimates from a standard, non-pooled design and estimates from pooled designs were evaluated. Different pooled study designs were investigated by changing the number of pooled eggs per aggregate sample, the number of repeated measurements per aggregate sample as well as the number of aggregate samples.
Exported data of the German federal dioxin database were used to generate assumptions on distributions of dioxins and PCBs in chicken eggs produced in different housing systems. Considering the prescribed study design for measurement of dioxins in food monitoring as the standard design, exported data were used to estimate the variance between samples of different origin. Furthermore, data of a validation study for analytics of dioxins in eggs were used to estimate the variance between repeated measurements. Using these estimated variance components, results show that the number of measurements can be reduced by pooling egg samples with keeping mean estimates and confidence intervals similar to those from non-pooled designs. The level of reduction depends on the relation of the variance between samples and the variance between repeated measurements, which varied between the different housing systems investigated in this study.
Die Autoren geben an, dass kein Interessenkonflikt besteht.
Die Autoren geben an, dass kein Ethikvotum erforderlich ist.