ARCHIVE - FOOD SURVEILLANCE INFORMATION SHEETS

	\| Help \| Site map \| Search \| Text links \|
	Archive - MAFF

Home

MAFF UK - EPOXIDISED SOYA BEAN OIL MIGRATION FROM PLASTICISED GASKETS

Index to MAFF UK Food Surveillance Information Sheets, 1999

Summary

The latest MAFF/DH Joint Food Safety and Standards Group (JFSSG) survey of chemical migration from packaging into food was carried out on epoxidised soya bean oil (ESBO) migrating from plasticised gaskets in the lids of glass containers used for baby foods. The Tolerable Daily Intake (TDI) was not exceeded for the general population of infants in Great Britain. ESBO was found in 66/137 samples, excluding repeat samples, at levels up to 105 mg/kg. The results were similar to those of a recent survey in Sweden. JFSSG has been informed that industry has been taking steps to reduce the migration of ESBO from lid gaskets into baby foods. The Committee on Toxicity of Chemicals in Food, Consumer Products and the Environment have considered this survey and concluded that it would be prudent for action to be taken to reduce ESBO levels in these products. The Food Advisory Committee (FAC) welcomed the action already being taken by industry with this objective in mind. The FAC noted that the design of the packaging and the use of ESBO in it have provided confidence that the products are safe. The FAC concluded that action by industry must continue and be done in ways that continue to ensure the highest level of safety of the products. It is particularly important that microbiological safety and the tamper-evident nature of such packaging should not be compromised in any way by action that might be taken to reduce ESBO levels.

Background

ESBO is a mixture of substances that is produced by the controlled epoxidation of soya bean oil in which the C=C double bonds are largely converted to epoxy groups. ¹ It plasticises gaskets in the lids of most glass containers used in this country for baby foods. ESBO makes the gaskets flexible and so helps to make tampering with the packaging evident, and prevent microbiological contamination and insect infestation of baby foods. The migration of this mixture of substances into food from gaskets is not subject to harmonised control in the EU. It is controlled in Great Britain under the general provisions of the Food Safety Act 1990.

There have been few surveys of ESBO in baby foods. A recent Swedish survey ² found this mixture of substances at up to 51 mg/kg in 74 out of 81 samples of baby food (made from ham, beef, turkey, chicken, fish, carrots, corn or blueberries). It was not detected in five samples of blueberries or in two samples of other foods. The Swedish work also established that contamination of baby foods with ESBO is due to migration from the lids of glass jars. Earlier work in Great Britain ³ showed that ESBO was not present in the lacquer covering the inside of the lids, but this mixture of substances was found in all 13 samples tested of baby foods (at up to 8 mg/kg) in glass jars, and in the gaskets from the respective jars (at levels up to 36 per cent).

The purposes of the JFSSG survey reported here were to provide up-to-date, more extensive British data for comparison with the results of the Swedish survey, and to derive a worst-case estimate of intake by the overall population of infants in Great Britain. The following specific questions were addressed:

If ESBO is present in baby food on sale in Britain, how do the incidence and levels of this contamination compare with the results from the survey in Sweden?
Is the Tolerable Daily Intake (TDI) exceeded from consumption of baby foods by the general population of infants in Britain? The EC Scientific Committee for Food has allocated a TDI of 1 mg/kg bodyweight/day for ESBO. ⁴ The UK Committee on Toxicity of Chemicals in Food, Consumer Products and the Environment has recommended the same TDI. ⁵
Have levels of ESBO in baby foods, on sale in glass containers in Great Britain, declined since industry started to take action to reduce ESBO levels in these products? (JFSSG has been informed that industry has been taking steps to reduce the migration of ESBO from gaskets in glass container lids into baby foods.)

Sampling and analysis for this JFSSG survey were carried out by PIRA, Leatherhead, Surrey. Work started before adoption of the Guidelines for Planning and Reporting JFSSG Surveys announced in the January 1999 issue of the DH/MAFF Food Safety Information Bulletin. Nevertheless every effort has been made to report this survey to the standards defined in those Guidelines.

Methodology

Sampling

Samples were purchased using plans that were based on estimated market share. Thus different numbers of samples were obtained of various companies’ products. Sampling was weighted towards products with fat contents towards the top end of those in available samples. However some samples with lower fat contents were also purchased as ESBO might be found in these, albeit perhaps ² at lower levels.

Samples were obtained from different areas to ensure that any regional variations in availability of products would not compromise the estimate of intake by the overall population. A wide range of products was obtained, with only a few samples being obtained more than once. Thus overall intake can be estimated but results for particular products cannot be compared between regions.

In the first part of the survey, 49 samples of packaged foods were purchased in triplicate from retail outlets in south-east England. Full details of the samples, including brand names, are given in Table 1 together with the levels of ESBO detected in the food samples. The absence of a particular brand from the list in this or other tables in this report means only that the brand has not been included in the survey.

On completion of the first part of the survey it was clear that there was a very extensive range of baby foods in glass containers available in shops. Therefore the survey was extended to a second part ( Table 2), in which 116 samples were purchased in triplicate from retail outlets in Great Britain (England, Scotland and Wales). The following groups of samples were obtained in the second part of the survey: products not included in part 1 of the survey; multiple samples of some products to test how much variation there is in ESBO levels within batches of a product; and further samples of products analysed in the first part of the survey (for comparative purposes). In the effort to obtain some identical samples in both parts of this survey it was apparent from details on labels that changes had been made to several products between the first and second parts of the survey.

Analytical work

Throughout this work the glass containers of baby food were stored upright and unopened at ambient temperature before analysis. Where several portions of food were analysed from the same glass container, these portions were taken at the same time.

The samples of food were analysed using a published method. ⁶ The entire contents of each jar was removed and homogenised. Each homogenate was then extracted and treated by transmethylation with subsequent derivatization of the epoxide groups to form dioxolane derivatives. Analysis of these was by GC/MS. Ions were monitored at m/z 277 and 309 for the derivatized ESBO component and at m/z 305 and 337 for the internal standard. Calibration was checked using standards prepared independently by a second analyst. A blank sample and standards corresponding to 3.4 to 119 mg ESBO/kg food were stored in a freezer. For foods containing lower declared levels of fat, an addition of olive oil was made prior to transmethylation to match the standards in total fat content. The limit of detection was 2 mg/kg and the limit of quantification was 7 mg/kg. (The limit of detection was taken to be three times baseline noise; the limit of quantification was taken as 3.33 times the limit of detection.)

Analytical quality assurance was carried out by:

duplicate analyses of the first 25 samples in part 1 of the survey and one duplicate analysis per analytical run in part 2;
including control samples and a solvent blank in each analytical run in both parts of the survey; and
‘blind’ analysis of two ‘spiked’ samples provided by an independent laboratory (Central Science Laboratory).

The analytical procedures were being carried out sufficiently well for the contractor to carry out analysis effectively in the survey. Thus agreement between duplicate analyses was good and the results for the two ‘spiked’ samples were satisfactory (98 per cent and 114 per cent respectively of the levels of ESBO measured at the laboratory, Central Science Laboratory, that provided the ‘spiked’ samples and developed the method of analysis - the 14 per cent apparent difference between measurement and amount added could be explained by variation in levels of ESBO in samples before they were ‘spiked’).

The survey results were corrected for recovery (92 - 106 per cent in part 1 of the survey, and 92 - 109 per cent in part 2). The variability of results in part 1 was 5.6 per cent, expressed as a relative standard deviation and calculated from control data, giving a measure of analytical uncertainty of 2.2 mg/kg at the 20 mg/kg level. In part 2 the relative standard deviation was 5.0 per cent and the measure of analytical uncertainty was 2.0 mg/kg at 20 mg/kg.

Results

The results of the survey are given in Tables 1 to 4:

ESBO was found in just under half of the samples (66/137), excluding repeat samples, at levels up to 105 mg/kg ( Tables 1 and 2).
ESBO levels were similar in samples from the same batch of a product ( Table 3). This may help industry in its work to reduce ESBO levels.
There were statistically significant, positive correlations between ESBO levels in baby food samples and the levels of fat in the samples (p less than 0.005), pack size (p less than 0.0001) and the recommended ages at which the products should be consumed (p less than 0.00001). Again this may help industry in its work to reduce ESBO levels. The possible association of ESBO and meat in baby foods ² was not testable because samples contained meat with other foods.
Although the number of samples with identical information on the label in both parts of the survey was limited, there was some evidence that ESBO levels did not consistently decrease between the first and second parts of the survey ( Table 4). Indeed samples with higher levels were found more often in the second part of the survey. For example more than 50 mg ESBO/kg was found in 9 per cent of samples in part 2 and in 4 per cent in part 1. These figures take into account duplication of samples. This apparent greater incidence of samples with higher levels in the second part of the survey may have been due to the different samples taken in the two parts of the survey.
The TDI of 1 mg/kg bodyweight/day was not exceeded for the general population of infants in Great Britain. Intake was calculated in a similar way to the Swedish work ² which estimated intake at 0.38 mg/kg bodyweight/day. The estimate of intake made using the data in Tables 1 and 2, excluding repeat samples, was 0.56 to 0.63 mg/kg bodyweight/day. This estimation used a 97.5 percentile value for consumption ⁷ (375g), a mean value of bodyweight ⁸ (8.8 kg) and a mean contaminant level. Upper and lower bound values were calculated to take account of results that were less than the limits of quantification or detection (loq and lod respectively). Upper bound values were calculated by taking results that were less than the loq as being equal to the loq, and ‘not detected’ results as being equal to the lod. Lower bound values were calculated by taking results that were less than the loq as being equal to the lod, and ‘not detected’ results were treated as being zero. As a double-check intake was also estimated by applying Monte Carlo analysis ⁹ to the data from the first part of the survey. This gave a value of 0.44 mg/kg bodyweight/day (using 97.5 percentile consumption). This was similar to an estimate of 0.29 to 0.39 mg/kg bodyweight/day calculated by using 97.5 percentile consumption and mean bodyweight, and the mean contaminant level from the first part of the survey.

Interpretation and conclusions

The JFSSG survey confirmed that ESBO was present in samples of baby food on sale in Britain ³ and gave similar results to a recent survey in Sweden. ² Although ESBO was found in a smaller proportion of samples (48 per cent) than in the Swedish work (91 per cent), with similar detection limits, it was detected over a broader range of levels (up to 105 mg/kg in comparison with up to 51 mg/kg in the Swedish survey). Intakes of ESBO from this source were similar in Britain and Sweden. Extensive surveys have not been published in other countries.

The TDI was not exceeded for the general population of infants in Great Britain. Indeed two different methods of estimating intake gave similar results, both below the TDI. Estimated intakes of ESBO were similar in Britain and Sweden.

JFSSG has been informed that industry has been taking steps to reduce the migration of ESBO from gaskets in glass container lids into baby foods. This is despite the absence of a maximum limit (as opposed to TDI) for ESBO in baby foods, and the exclusion of gaskets in metal lids of glass containers from EU controls on migration from plastic materials and articles in contact with food. It is understood that ESBO migration might be reduced by using static rather than rotary retorting in sterilising the packaged baby food or perhaps by minimising the area of exposure of the gasket to headspace above the baby food. It is particularly important that microbiological safety and the tamper-evident nature of such packaging should not be compromised in any way by action that might be taken to reduce ESBO levels. It is also understood that industry may be carrying out research on plasticiser-free gaskets and on possible alternatives to ESBO. Both of these approaches would need to be taken forward very carefully. For example any substance used instead of ESBO should not be less extensively researched than ESBO in its toxicology, efficacy or potential to migrate.

The comments of the Committee on Toxicity of Chemicals in Food, Consumer Products and the Environment, and the Food Advisory Committee on this survey are provided at Annexes 1 and 2 respectively.

Companies were provided with the results for samples of their brands included in this survey and asked to comment. All of the companies replied. Their comments are provided at Annex 3.

References:

Working Party on Chemical Contaminants from Food Contact Materials: Sub-Group on Plasticisers (1990) Plasticisers: continuing surveillance. Food Surveillance Paper No. 30, publ. HMSO.
Hammarling, L., Gustavsson, H., Svensson, K., Oskarsson, A., and Karlsson, S. (1998) Migration of epoxidized soya bean oil from plasticised PVC gaskets into baby food. Food Additives and Contaminants 15, 203-208.
Castle, L., Mayo, A., and Gilbert, J. (1990) Migration of epoxidised soya bean oil into foods from retail packaging materials and from plasticised PVC film used in the home. Food Additives and Contaminants 7, 29-36.
Commission of The European Communities, Directorate General III (1999) Synoptic Document: provisional lists of monomers and additives notified to the European Commission as substances which may be used in the manufacture of plastics intended to come into contact with foodstuffs. Updated to 23 February 1999.
Committees on Toxicity, Mutagenicity, Carcinogenicity of Chemicals in Food, Consumer Products and the Environment (1994) Annual Report, sections 1.10 -1.12, p. 8.
Castle, L., Sharman, M., and Gilbert, J. (1988) Gas-chromatographic-mass spectrometric determination of epoxidised soya bean oil contamination of food by migration from plastic packaging. Journal of the Association of Official Analytical Chemists 71, 1183-1186.
Mills, A., and Tyler, H. (1992) Food and nutrient intakes of British infants aged 6-12 months. publ. HMSO, ISBN 011 2429068.
Department of Health Report on Health and Social Subjects 41, Dietary Reference Values for Food, Energy and Nutrients for the United Kingdom. Report of the Panel on Dietary Reference Values of the Committee on Medical Aspects of Food Policy (1991).
Tennant, D.R. (1998) Quantifying exposure to natural toxicants in food. pp. 265-283 in Natural Toxicants in Food, publ. CRC/Sheffield Academic Press.

Further information

Units of Measurements
1 milligram (mg) = 10 ^-3 gram
1 kilogram (kg) = 10 ³ grams

The report of the survey is held in the MAFF Library at Nobel House, 17 Smith Square, London SW1P 3JR (Tel: +44 (0)20 7238 6573). If you wish to consult a copy please contact the library for an appointment giving at least 24 hours notice or, alternatively, copies can be obtained from the library: a charge will be made to cover photocopying and postage.

Further enquiries should be addressed to:
Dr David Watson
Head of Branch B
Additives and Novel Foods Division
JFSSG
Room 212, Ergon House
17 Smith Square
London SW1P 3JR

Tel: +44 (0)20 7238 6250
Fax: +44 (0)20 7238 6124
e-mail: d.watson@fssg.maff.gov.uk

Spreadsheet tables

Tables of Results
Click here to download the Excel 5.0 version of Tables 1-4
Click here to view the .pdf version of Tables 1-4 (if you have any other spreadsheet package)

Return to Index to Surveillance Information Sheets, 1999 page
Go to top of page

These pages were last updated on 15 September 1999