R02008: Statistics of radiological monitoring results - how to deal with sets of data where some are below detection limits

Study Duration: June 1998 to May 2001

Contractor: National Radiological Protection Board

The Agency organises a large monitoring programme that measures the level of radioactivity in foods and other substances. The programme mainly ensures that discharges of radioactivity from major nuclear sites do not lead to unacceptable risks from eating local foods.

The Terrestrial Radiological Monitoring Programme (TRAMP) deals with nuclear sites, while the Food and Agriculture Radiological Monitoring (FARM) Programme deals mainly with areas affected by the Chernobyl fallout and samples from different regions of England and Wales to assess the general level of radioactivity in food.

Levels of radioactivity in food are generally very low. The methods of analysis are very sensitive and are capable of measuring very low levels of radioactivity in food. However, reporting levels are established for each radionuclide of interest to ensure analytical costs are not excessive. In some cases, the radionuclides are present at levels below the reporting levels. Thus the lower values are often not known and hence the data may be considered an incomplete picture of the true spread of values.

To avoid underestimating doses when levels of radioactivity in foods are below the reporting levels, they are assumed to be at the reporting levels for calculations. This means in these situations the calculated dose is likely to be higher than the real dose and that any averages taken could also be artificially high.

This project aimed to suggest techniques that can use data more effectively, including those below the reporting level, and so improve dose estimates. In particular, the study reviewed the techniques available for establishing improved estimates of statistics of the spread of levels of radioactivity in food samples. This project considered the various assumptions that can be made about the likely spread of results and the likely validity of these assumptions, by reviewing the processes and mechanisms contributing to the actual spread of data found in practice. The aim was to provide advice on the most appropriate spread to assume when estimating relevant summary statistics from the data.

The approaches presented cover a range of options from so-called parametric techniques, through robust methods to fully non-parametric approaches. The techniques discussed also cover a range of complexity from simple methods like substitutions to highly advanced statistical methods such as Bayesian methods.

A number of techniques have been tested on data from the Radioactivity in Food and the Environment report (RIFE). As a result of this testing recommendations have been made on simple-to-use approaches to dealing with data sets containing reporting level data.

This study reviewed and investigated the mathematical techniques available to provide more realistic, but still conservative, estimates of the spread of data containing values below reporting levels. Guidance is also provided on the selection of techniques to ensure that the Agency is making effective and efficient use of the information available from the environmental samples collected.

By adopting this guidance, the Agency will enhance confidence in results derived from environmental surveillance without incurring additional costs. The Agency will also gain a better understanding of the levels of radioactivity in the environment and, if required, be better able to direct effort to minimising the most significant uncertainties in these estimates.

Final report is available from the FSA Library and Information Centre. To obtain a copy, please contact the Enquiry Desk, Dr Elsie Widdowson Library and Information Services, Food Standards Agency (Tel: +44 (0) 20 7276 8181/8182 or by email to: library&info@foodstandards.gsi.gov.uk).

Contact: Paul Tossell
Tel: +44 (0) 20 7276 8750
Email: paul.tossell@foodstandards.gsi.gov.uk