Research report

Exploring improvements to models used in risk assessment of the Scottish monitoring programme for marine biotoxins in shellfish harvested from classified production areas.

1 documents for this subject

Summary

Biomathematics and Statistics Scotland

Holtrop, G.

The aim of the study was to examine the statistical model used to inform the risk assessment of the Scottish monitoring programme for marine biotoxins in shellfish from classified production areas and consider alternative statistical models. The original simple statistical model was based on a limited data set of three to six years results which were available at the time. There is currently over 15 years of results for biotoxins in mussels. 

The current study explores the possibility of using a smooth model that uses the actual date of collection as opposed to aggregating the data by month. The effect of a smooth model on monitoring frequencies was also considered and compared to the simple model.

On the whole there was good agreement between the suggested monitoring frequency derived from the simple model and those derived from the smooth model.

Key findings:

  • Lipophilic toxin  results could be fitted to a smooth model, this was not possible with the simple model.
  • The smooth model was very time consuming in comparison to the simple model
  • For both models there were occasional times when the model did not capture the trends in the biotoxins data
  • Propose a visual display of predicted trends from modelling and observed prevalence to give a more comprehensive  picture of information available for consideration in the risk assessment.