Methodological advances to improve the cost-effectiveness of monitoring for mycotoxins in cereal grains

Research output: Thesisinternal PhD, WU


Mycotoxins are toxic secondary metabolites produced by certain types of fungi under favourable conditions, and can be present in several food and feed commodities such as cereals and derived products. Mycotoxins can lead to adverse health effects in both animals and human. Therefore, legal limits and standards have been set for the maximum concentration of certain mycotoxins in various feed and food ingredients and products. An effective mycotoxin monitoring system needs to be in place to check for the presence of mycotoxin in raw materials for food and feed and derived products as part of Hazard Analysis and Critical Control Points (HACCP) programs in the feed and food industry, and for compliance with the legal limits. Monitoring is done by regular collection of samples from batches along the supply chain and the chemical analysis of these collected samples for the presence of mycotoxins. For official controls of both feed and food products, sampling and analytical (S&A) plans for mycotoxins are prescribed by the EU regulations. The monitoring procedures described in the EU regulations are, however, resource demanding because of the high number of samples that need to be collected and the detection methods that need to fulfil strict performance criteria.

Mycotoxins are often heterogeneously distributed in cereal grain batches; in the major part of the batch, the concentration is low, whereas only a few spots are highly contaminated. Because of this heterogeneous distribution, samples collected at different locations in the batch have different mycotoxin concentrations. One of the challenges in mycotoxin monitoring is to collect the right number of samples to reflect the mean concentration of the batch. Another challenge is to accurately determine the concentration of the mycotoxin in the final sample. However, the most accurate detection methods for mycotoxin analysis are usually also the most expensive and time consuming methods. The resources available for mycotoxin monitoring in a cereal supply chain are often limited and, therefore, choices have to be made to effectively distribute available resources between sample collection and the chemical analysis of the samples. The more resources are spent on the chemical analysis, the less is available for the collection of the samples, and vice versa. The overall objective of this thesis is to design methods to design cost-effective monitoring plans for mycotoxins in cereal grains.

Chapter 2 estimated the financial losses of the 2013 aflatoxin incident In Europe. In the summer of 2012, maize from the Black Sea area and highly contaminated with aflatoxins was transported and processed into dairy cow feed in the Netherlands. As a result, milk produced in the Netherlands exceeded EU limit for aflatoxin M1. The financial losses of this incident were estimated for several stakeholders in the Netherlands and its neighbouring countries Belgium and Germany. The estimated financial losses were in the range of tens of millions of euros for all stakeholders in these three countries. The largest percentage, about 70% of the costs, were for the maize traders importing and selling the maize. The remaining 30% of the costs were for the feed processing industry. These estimated costs gave an indication of the indirect costs of imperfect monitoring in the upstream stages of the maize supply chain.

Chapter 3 performed a literature review describing the methods previously used in the life sciences to develop cost-effective monitoring strategies, including monitoring for animal diseases, plant pests, hazards in the soil, water, food and animal feed. The pros and cons of these methods, as well as their applicability to food safety were assessed. Chapter 3 concluded that several methods were previously used to develop cost-effective monitoring strategies such as: simulation models, optimisation models and statistical models with a combination of using deterministic and stochastic input data.

Chapter 4 developed an optimisation model to design cost-effective S&A plans for mycotoxins in a cereal batch. Two case studies were used, being deoxynivalenol (DON) in a wheat batch and aflatoxins in a maize batch. This chapter concluded that the sampling step had the highest influence on the effectiveness of a S&A plan whereas the performance of the detection method had limited influence on the effectiveness of a S&A plan. Therefore, a detection method such as ELISA, which is lower in costs and easier to use than instrumental methods, was a suitable detection method given the money saved on the analysis was spent on collecting more samples to be analysed. Adding to that, given that one is interested in the mean mycotoxin concentration of a wheat or maize batch, Chapter 4 concluded that aggregating incremental samples and extracting a sub-sample from the aggregate sample for the chemical analysis is more cost-effective than analysing incremental samples separately for both DON and aflatoxins.

Chapter 5 developed an optimisation model to identify cost-effective monitoring strategies along the maize supply chain. This chapter focussed on the sample collection step of S&A plans only. This chapter concluded that, for aflatoxins, a mycotoxin produced both in the field and during storage in case of sub-optimal conditions, monitoring all batches in the downstream stages of the maize supply chain, thus the transport and storage steps, was cost-effective. The optimal number of samples to collect from the batches depended on the mycotoxin concentration at the time of sampling: when the mycotoxin concentration was slightly higher than the pre-set limit, more samples were needed than when the concentration was lower or much higher than the pre-set limit. The results of this chapter suggest that collecting less samples than prescribed by the EU regulations was cost-effective when considering a flow of multiple batches during a longer time period, e.g. one or more years. For individual batches, collecting less samples led to a high probability of misclassifying the batch in case the concentration was slightly above to the pre-set limit and, hence, collecting as much samples as prescribed by the EU regulations is advisable.

Chapter 6 compared the financial effects of no-monitoring with monitoring schemes for aflatoxins collecting either 1, 18 or 100 samples per imported maize batch in the Netherlands. This chapter integrated the results of Chapters 4 and 5 with the results of Chapter 2. Chapter 6 showed that the net positive financial effect was higher when monitoring for aflatoxins was performed, compared to no monitoring at all. Therefore, independent of the monitoring plan applied, monitoring all incoming maize batches was cost-effective. Collecting 18 or 100 samples per batch, aggregating all these 18 or 100 samples, and analysing one sub-sample, led to a higher positive financial effect than collecting only one sample per batch. However, the difference between collecting 18 or 100 samples was small. We can, therefore, conclude from this chapter that collecting between 18 and 100 samples per batch, which is less then prescribed in the EU regulations, is cost-effective when a flow of imported batches are considered.

The main conclusions of this thesis are:

  • Undetected contaminated maize batches (false negatives) in the upstream stages of the maize supply chain can lead to large economic consequences in the downstream stages of the maize supply chain (Chapter 2).

  • It is cost-effective to allocate most of the budget of the S&A plan for mycotoxins in wheat and maize to the sample collection step, rather than to the chemical analysis (Chapter 4).

  • Aggregating incremental samples and analysing a sub-sample from the aggregated sample is a cost-effective method for monitoring aflatoxins in maize and deoxynivalenol in wheat (Chapter 4).

  • When an entire flow of batches during one or multiple years is considered, collecting less samples than prescribed by the EU regulations is cost-effective for monitoring deoxynivalenol in wheat and aflatoxins in maize (Chapters 4, 5 and 6).

  • When the aflatoxin concentration in a maize batch is 1 to 1.6 times higher than the pre-set limit of 2.5µg/kg, 10 to 100 times more samples are needed to correctly classify batches than when the batch concentrations are lower or higher than 1.6 times the pre-set limit (Chapter 5).

  • For deoxynivalenol, a mycotoxin produced in the field, monitoring in the upstream stages of the wheat supply chain is cost-effective whereas for aflatoxins, mycotoxins that can be produced both in the field and during storage, monitoring in the downstream stages of the maize supply chain is cost-effective (Chapter 5).

Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • Wageningen University
  • Oude Lansink, Alfons, Promotor
  • van der Fels-Klerx, Ine, Promotor
Award date11 Dec 2019
Place of PublicationWageningen
Print ISBNs9789463951739
Publication statusPublished - 11 Dec 2019


Dive into the research topics of 'Methodological advances to improve the cost-effectiveness of monitoring for mycotoxins in cereal grains'. Together they form a unique fingerprint.

Cite this