90
N. Wesolek et al.
the enterophathogens: Vibrio parahaemoliticus or Bacillus cereus, which are both
routinely found in bivalve molluscs. Misdiagnosis is due to the fact that DSP toxins
and enteropathogens have similar symptomatologies. Thus there is no accurate
information linked to the annual DSP human poisoning episodes (Gestal-Otero
2000). However, some poisoning events have been reported in detail. Thus in
France, in 1984 and 1985, cultured mussels caused DSP-like symptoms in 10,000
and 2,000 people respectively (Durborow 1999). Because of the extent of the
problem, appropriate sampling plans are required in order to monitor bivalve
production areas to check for the presence of these biotoxins, knowing that a
species with the highest contamination rate can be used as an indicator species
(Regulation 854/2004/EC). Because mussels have one of the highest accumulation
rates (Vale and de Sampayo 2002; Suzuki and Mitsuya 2001), it is considered as the
best indicator species. A sampling plan validation method, primarily developed by
Whitaker (Whitaker et al. 1972) and widely applied (Whitaker et al. 2007a, b), is
used to compute probabilities of acceptance. Knowing that the true contamination
level of a lot is never known, as only sample analytical results can be obtained,
the probability of acceptance is defined as the probability that the sample analysis
results are lower than the food safety threshold concentration level for OA. The
probabilities of acceptance, as determined by different sampling schemes, are
plotted against mean OA lot concentrations. The curves obtained, which are referred
to as Operating Characteristic (OC) curves, enable one to quantify consumer and
producer risks. Consumer risk is the probability that a lot having a true concentration
above the threshold (unsafe lot) is authorized for sale and consumption. Producer
risk is the probability that a lot at a true concentration lower than the threshold (good
lot) is rejected for sale. Then a best fit sampling plan can be proposed, taking into
account the two risk types, as well as considering the practical feasibility of the
sampling plan.
Material and Method
The sampling plan validation method developed by Whitaker consists of a series
of calculations on contaminant concentration data from mussel samples taken
from various mussel lots. The sample concentrations from a lot are adjusted to
a theoretical distribution by a goodness of fit test. This process is repeated on a
few lots. Furthermore, the variability between sample concentrations within a lot
is studied, in order to predict this variability for the mean concentration of any lot,
within a given range of concentrations. Both the theoretical distribution and the
prediction of concentration variability between samples of the same lot is used to
calculate the probabilities of acceptance of lots for the sampling plan tested. All
these steps are further explained in the following sections and used to evaluate
different sampling strategies designed to detect potentially harmful levels of okadaic
acid in mussels.
N. Wesolek et al.
the enterophathogens: Vibrio parahaemoliticus or Bacillus cereus, which are both
routinely found in bivalve molluscs. Misdiagnosis is due to the fact that DSP toxins
and enteropathogens have similar symptomatologies. Thus there is no accurate
information linked to the annual DSP human poisoning episodes (Gestal-Otero
2000). However, some poisoning events have been reported in detail. Thus in
France, in 1984 and 1985, cultured mussels caused DSP-like symptoms in 10,000
and 2,000 people respectively (Durborow 1999). Because of the extent of the
problem, appropriate sampling plans are required in order to monitor bivalve
production areas to check for the presence of these biotoxins, knowing that a
species with the highest contamination rate can be used as an indicator species
(Regulation 854/2004/EC). Because mussels have one of the highest accumulation
rates (Vale and de Sampayo 2002; Suzuki and Mitsuya 2001), it is considered as the
best indicator species. A sampling plan validation method, primarily developed by
Whitaker (Whitaker et al. 1972) and widely applied (Whitaker et al. 2007a, b), is
used to compute probabilities of acceptance. Knowing that the true contamination
level of a lot is never known, as only sample analytical results can be obtained,
the probability of acceptance is defined as the probability that the sample analysis
results are lower than the food safety threshold concentration level for OA. The
probabilities of acceptance, as determined by different sampling schemes, are
plotted against mean OA lot concentrations. The curves obtained, which are referred
to as Operating Characteristic (OC) curves, enable one to quantify consumer and
producer risks. Consumer risk is the probability that a lot having a true concentration
above the threshold (unsafe lot) is authorized for sale and consumption. Producer
risk is the probability that a lot at a true concentration lower than the threshold (good
lot) is rejected for sale. Then a best fit sampling plan can be proposed, taking into
account the two risk types, as well as considering the practical feasibility of the
sampling plan.
Material and Method
The sampling plan validation method developed by Whitaker consists of a series
of calculations on contaminant concentration data from mussel samples taken
from various mussel lots. The sample concentrations from a lot are adjusted to
a theoretical distribution by a goodness of fit test. This process is repeated on a
few lots. Furthermore, the variability between sample concentrations within a lot
is studied, in order to predict this variability for the mean concentration of any lot,
within a given range of concentrations. Both the theoretical distribution and the
prediction of concentration variability between samples of the same lot is used to
calculate the probabilities of acceptance of lots for the sampling plan tested. All
these steps are further explained in the following sections and used to evaluate
different sampling strategies designed to detect potentially harmful levels of okadaic
acid in mussels.