Research and analysis

HSE risk assessment methodology

Published 10 December 2024

© Crown copyright 2024

This publication is licensed under the terms of the Open Government Licence v3.0 except where otherwise stated. To view this licence, visit nationalarchives.gov.uk/doc/open-government-licence/version/3 or write to the Information Policy Team, The National Archives, Kew, London TW9 4DU, or email: [email protected].

Where we have identified any third party copyright information you will need to obtain permission from the copyright holders concerned.

This publication is available at https://www.gov.uk/government/publications/quarterly-reports-on-the-pesticide-residues-monitoring-programme-background-and-methodology/hse-risk-assessment-methodology

HSE assessment of risk

The surveillance programme is designed to enable the regulatory authorities to check whether pesticides are being found at levels within the maximum residue levels (MRLs). This confirms that the regulatory processes are working correctly, and as part of this, that pesticides users are complying with any specified conditions that were part of the authorisation. In addition, this work checks that dietary intakes of residues are within acceptable limits. This may be more challenging when pesticide residues are found in food products that have not been grown in the UK or EU, notably when older pesticides have been used.

When assessments are carried out

HSE perform screening assessments for each residue and commodity combination to identify residue levels that would lead to intakes above the relevant health based reference doses (these are also sometimes referred to as toxicological reference values (TRVs)). Further information on this screening approach is available on request from HSE.

We then present detailed assessments for every case where the actual residue level found could lead to an intake by any group above the reference dose.

If dietary intakes exceed the reference values this does not automatically mean there are expected adverse health effects. However, this acts as a ‘trigger’ for HSE to consider these cases more thoroughly.

Assessing dietary intakes

HSE conduct both short-term (acute) and long-term (chronic) assessments based on the residues found in the PRiF surveys. Each of these is tailored accordingly.

These assessments use information on food consumption collected in UK dietary surveys in conjunction with the residue levels we find. HSE estimate dietary intakes for 10 consumer groups: adults, infants, toddlers, 4 to 6 year olds, 7 to 10 year olds, 11 to 14 year olds, 15 to 18 year olds, vegetarians, elderly living in residential homes and elderly living in their own homes. Occasionally, HSE use additional pesticide specific information on the losses of residues that occur during preparation and cooking of food.

How the assessment is carried out

Short term risk assessment

HSE calculate short term intakes (also called NESTIs) using consumption data for high level (97.5th percentile) consumers, based on single day consumption values and the highest residue found in a food commodity. The residue found is multiplied by a variability factor to take account of the fact that residues may vary between individual items that make up the sample analysed, in order to address exposure to a higher than usual residue in a single item, such as a single apple or potato.

This is why in some of our detailed risk assessments we refer to some of the general variability factors (of 5, 7 and 10) that are applied in short term risk assessments. Sometimes, regulatory assessment of data for a pesticide can support an alternative specific value of the variability factor, and where justified, HSE will apply these to the risk assessment and explain this. The estimated intake is compared to the Acute Reference Dose (ARfD), a suitable health based reference value for effects that could be caused by a single day or one-off consumption of a higher than usual residue.

Long term risk assessment

HSE also calculate long term intakes (NEDIs) for high level (97.5th percentile) consumers, but in this case the consumption data are high level long-term values rather than peak single-day events. Similarly, the residue values used reflect long term average levels (we use the median value across each commodity type) rather than occasional high values. In this case the estimated intake is compared to the Acceptable Daily Intake (ADI), a suitable health based reference value for lifetime.

Where HSE gets the reference doses from

The reference doses (ADI, ARfD or other suitable health based reference values such as Tolerable Daily Intake (TDI) (sometimes used instead of an ADI)) are set following regulatory assessment. In the UK, these values will be reviewed prior to establishment by the Expert Committee on Pesticides (ECP). HSE use reference values from a range of respected sources, including the EU and values set by the European Food Safety Authority (EFSA). Up to 2019, the UK was part of a harmonised approach to the assessment of pesticide substances in the EU, and we have taken part in the peer review of previously established values. As such, HSE’s toxicological reference value database (GB Toxicological Reference Value Database) contains the reference values that are relevant to consumer risk assessment within GB, and this includes EU values that were established up to 2019 as well as any new (GB) derived values. HSE will also use values from other respected international sources, such as those established by the joint FAO and WHO Meeting on Pesticide Residues (JMPR, which evaluates and publishes residues and toxicological evaluations of pesticides) and levels set by regulatory authorities in other countries. For a small number of pesticides, the reference doses used have been determined by HSE (for example, prothiofos, tecnazene). These have not been independently peer reviewed and should therefore be regarded as provisional.

Further explanation of the models HSE use and application to the PRiF work

Deterministic models

The assessments HSE use are ‘deterministic’ which means a defined level of input (such as a median or highest residue and 97.5th percentile dietary consumption values is used). More information on the deterministic intake assessments is available on the HSE Pesticide website. Here you will find information and further links. See below for an explanation of probabilistic models, where the inputs into the assessments can be varied and more realistic assessments reflecting a range of possible scenarios can be modelled.

Detailed risk assessment work is carried out before pesticides are authorised

The fundamental full complement of risk assessment work for pesticides is done at the pre authorisation stage considering the residues data packages when trials reflecting the label uses of the pesticides are assessed. These trials profile the highest likely residues that might arise when the pesticides are used as intended (crops and permitted doses of use). MRLs are set on the basis of these data sets, and the post approval monitoring work then serves as a check for whether residues found are in line with this prior expectation. The risk assessments supporting the MRLs assess the highest residue observed in each crop (or animal product) for short term assessment. For the long term MRL assessments, the median residues, across all crops and animal products are taken together to assess combined intakes over the long term as ‘total dietary intakes’, taking account of all possible food exposures. Authorisation for pesticide products can only be granted where these assessments of dietary intake do not exceed the health based reference values.

MRL considerations

Although MRLs are not safety levels, an MRL would not be established if the residue concentrations measured in the supervised trials used to support the MRL would give rise to health concerns. In most cases residues present at the MRL result in intakes below the ARfD and the ADI. So even if the MRL is exceeded this does not always lead to an intake above the ARfD or ADI.

What happens if we find a pesticide intake above an ADI or an ARfD

In addition, an estimated intake in the monitoring work that exceeds the ADI or ARfD does not automatically result in concerns for consumer health, because a protective approach is used in setting the ADI and ARfD. In the unusual circumstance of an intake exceeding the ADI or ARfD, HSE undertake an evaluation of the toxicological data, and we present details of this assessment.

When we present the outcomes of risk assessments, we provide a conclusion on the possible impact on human health based on the degree of concern following the HSE assessment of risk. These conclusions keep to the following order severity, from lowest to highest risk: “Effects on health are not expected” (toxicological reference values not exceeded), “unlikely risk” (of effects on health), “low risk”, “higher risk” (exposures are undesirable)[footnote 1]. Most detailed consumer intake assessments that we present with the quarterly reports are for short term exposure rather than chronic exposure. This is because in most cases the monitoring data show the majority of samples tested contain residues below the reporting limit and so chronic exposure would not present a concern. The reporting limits are set at suitable low levels based on analytical laboratory procedures so that dietary intakes are typically far below the health based reference values.

Monitoring data presented in the quarterly reports provides a ‘snapshot view’ of the residues found in specific foods in a survey usually conducted over a 12 week period and limited to around 30 samples for most commodities. It is important not to use isolated findings of higher than expected residues in small surveys to make judgements on long term effects over a lifetime exposure.

For monitoring work, long term exposure assessments, HSE use median residue levels, rather than the highest residues found. For quarterly assessment (data obtained over 3 months only) we currently only assess long term dietary assessment commodity by commodity and not as total dietary intakes across commodities. Even where a number of samples in a commodity survey do contain the same pesticide, it is very rare that the ‘screening assessment’ we undertake leads to the need for a more detailed assessment to be presented (only where the dietary intakes exceed the ADI or TDI).

We do not see a high number of monitoring samples for any pesticide that exceed an MRL. This means that, over time, median residue levels found in the monitoring do not often exceed the median residues used in the trials assessed at the time of MRL setting and so do not often require consideration of long term effects. When HSE does need to assess long term effects, it is likely to overestimate exposure to an assessed pesticide residue in a single food item. For a single commodity risk assessment, we assume high level (97.5th percentile) consumption at the median residue level in that food for each day of lifetime. Although the HSE long term assessments by their quarterly nature are indicative only, the assessment we perform is conservative. Furthermore, alternative published assessments (for example, those considering trends and large bodies of data) are available which further consider the long term exposure to pesticides[footnote 2]. Some pesticides contribute more significantly to long term dietary intakes across commodities based on their toxicology and prevalence (such as chlorpyrifos, cyfluthrin, deltamethrin, diazinon, dieldrin, dimethoate, dithiocarbamates, fenamiphos, fipronil, imazalil, lambda-cyhalothrin, omethoate and pirimiphos-methyl), based on the chronic exposure assessments presented in EFSA, 2020[footnote 3].

We will continue to have focus on residues of consumer relevance, when they are found, in both UK and NI produced and imported produce. As pesticide use changes, including when there are impacts of regulatory action on pesticides, the profiles of residues in the monitoring can change over time. For example, chlorpropham, which was previously found at levels well above the reporting limits in potatoes, is no longer permitted for use in the UK and EU, and this pesticide will no longer have dietary intakes which take up a substantial portion of its health based reference values.

Implications for health

Where intakes exceed a reference dose, it is necessary for the underlying toxicological studies (animal studies) to be considered to enable the significance for the consumer of such an exceedance to be understood. Toxicological studies supplied by the registrants in the regulatory data packages are conducted using different doses to determine the nature of any ill health effects as well as the levels at which such effects can be expected to occur.

Toxicological studies that we refer to and use in the HSE risk assessments are conducted using test animals to identify the highest experimental dose that causes no detectable adverse effects (the NOAEL). Where there is more than one relevant toxicological study, the lowest appropriate NOAEL for the most sensitive adverse effect is typically used. There is some uncertainty in extrapolating between animals and people and it is therefore important to use a ‘safety factor’ to account for sources of variation. This safety factor is incorporated (by dividing the NOAEL by the safety factor) in deriving a reference dose, either an ADI or an ARfD, to which consumer intakes are compared. A safety factor therefore extrapolates from the animal testing to the general population. Factors in the order of x100 are commonly used, x10 for animal to man, and x10 for within human population differences in sensitivity. However, toxicologists may propose different values (for example, from 5 to 1,000) based on scientific reasoning in accordance with study designs and the quality of the data that has been generated from the studies.

In order to ensure exposures to pesticides do not pose unacceptable risk to humans a wide range of investigations are performed. Most of these are performed on experimental animals because the only endpoints that can be examined in human volunteers are those involving observation or blood and urine sampling. Human volunteer studies involving pesticides are not generated in current regulatory work. There is debate at the international level as to whether human studies that have been generated should be used for risk assessment purposes. The UK policy is not to use these data in pre authorisation assessments which support the registration of a pesticide; the JMPR chose to apply judgement in the appropriate use of these data if available. The HSE risk assessments will usually refer to test animal species, such as dog, rat, and rabbit. All toxicological work is undertaken based on principles of minimising animal distress. Where scientifically valid human data are available the risk assessments will refer to these as they reduce the uncertainty in the assessment. Therefore, human data is only referred to in more limited circumstances.

Acute (short term) toxicology is not a concern for all pesticides, as some are not acutely toxic. In terms of the pesticides that have been found in fruit and vegetables through the surveillance programme an acute risk assessment would not be necessary on the following examples: maleic hydrazide, diphenylamine, kresoxim-methyl, and quintozene.

If we understand that a pesticide has a risk of genotoxicity (able to damage genetic material), we will include this in the commentary. During regulatory assessment, careful consideration is given to any pesticides that may exhibit any potential to be genotoxic in live animals. In the monitoring programme we note residue types that have been shown in the toxicological data sets to have genotoxic potential or those where data are suggestive of genotoxicity but not certain. There is a small number of cases of older pesticides, likely found only in imported foods, that might be genotoxic. These are examples where modern data to investigate the true genotoxic potential are not expected to be made available. In such situations, we might conclude on a precautionary basis that any findings of these pesticides are undesirable due to the uncertainty regarding genotoxicity, and at low residue levels any risks of adverse health effects are low due to the limited levels of exposure anticipated. HSE use low reporting limits for these pesticides to detect these residues even at very low levels, as we know they are of particular interest to consumers.

Consumption data and refining the risk assessment

Consumption values

As the surveillance programme monitors residues in all types of food, from raw commodities (for example, potatoes) to processed (for example, wine), dried (for example, dried fruit) and composite foods (for example, fruit bread), consumer risk assessments are specifically tailored to address processed and mixed food products. Sometimes this can be affected by availability of consumption data. For example, for pâté, we assess this using consumption data for liver (all types of liver), and for fish pâté we use consumption data for fish (all sources and types of fish). However, we use specific consumption data where the Food Standards Agency (FSA) have provided data to us (for example, data on orange juice, dried grapes, and bread). Consumption data are available for most raw commodities, but where data are limited then we will suggest using alternative data. This may involve considering other commodities (for example, using potato data as ‘surrogate’ for sweet potato), or alternative sources of consumption data such as EU PRIMo or JMPR consumption and dietary assessment models, to consider items that do not currently feature in UK data sets. Where alternative data are used in our screening and written assessments we explain this in our presentation of the risk assessment work for each quarterly report.

Fruit and vegetables with removable peel

For fruit and vegetables that have peel or skin that might not be consumed we present alternative risk assessments for ‘without peel, flesh only’ where peel versus pulp residue distribution data is available. As standard, we present an assessment for ‘all of the peel’ consumed. It is not expected that consumers will always eat peel, so these standard assessments are considered to be highly cautious and not necessarily realistic. Further data are being generated to better understand whether some people do eat the peel of these, and if so how much of the peel they tend to eat.

At the time of MRL assessment, the agreed international approach applied to the risk assessment is to assume that the peel is removed for certain types of commodity that are designated as having ‘inedible peel’. In this way when the MRLs are agreed, if there are suitable data on distribution of residues between peel and pulp (flesh), the risk assessment supporting the MRL can use a peeling factor which removes the higher residue associated with the peel. The MRLs are set for the whole fruit (with peel on), and as such in the monitoring work there can be examples of residues found well below the MRL that lead to dietary intakes that exceed the ARfD, if it is assumed that a consumer will eat all the peel.

For transparency, in the monitoring work, HSE present the alternative assessments: ‘all of the peel consumed’ and ‘without peel, flesh only’.

Dithiocarbamate residues

Dithiocarbamate residues are determined as carbon disulphide which is a common product from different dithiocarbamate pesticides. For the risk assessment we take a precautionary approach. For short term assessment, the worst case dithiocarbamate residue is calculated by assuming the residue is derived from thiram (a molecular weight conversion is applied to estimate the level of residue based on thiram) and this is compared to the ARfD for thiram. If the ARfD for thiram is exceeded, it might be possible for additional laboratory analysis to determine if the residue could have resulted from either thiram, propineb or ziram. Where it can be confirmed that a specific dithiocarbamate was applied the equivalent residue of the specific active substance is estimated and the intake compared to the appropriate reference dose, we select an alternative worst case dithiocarbamate if the laboratory analyses indicate absence of specific types. We only present a detailed risk assessment when dietary intake exceeds either the thiram or other suitable reference dose.

The analysis of dithiocarbamates is further complicated by an expectation that some types of crops, such as members of Brassicaceae (for example, watercress) and Caricaceae (for example, papaya) might contain natural sources of sulphur compounds that could be also determined as carbon disulphide during analysis in the laboratory. HSE will consider and explain in the report whether residues reported as dithiocarbamates could be from natural sources or whether they have arisen as a result of fungicide treatment.

Residues that are isomers

The routine methods used for pesticide residues analyses are not always able to distinguish between different isomers of pyrethroid pesticides because these isomers are very similar in their physical and chemical properties. Isomers share the same molecular formula and differ only in the spatial arrangement of the molecular structure. Pyrethroid pesticides can have many isomers and are marketed either as pure individual isomers or as a mixture of them. It is possible for isomers to have different toxicological properties.

For such residues, during the risk assessment work undertaken, HSE will use available information on the toxicology, an understanding of the pesticide marketed in the country in which the pesticide was used for the sample in question and whether the analysis can distinguish between different isomers. For example, residues of lambda and gamma-cyhalothrin are not distinguishable analytically. Gamma-cyhalothrin is the more toxic form (than lambda-cyhalothrin), but we might have information on whether gamma-cyhalothrin or lambda-cyhalothrin has been sprayed on the crop. In terms of cypermethrin, alpha-cypermethrin is more toxic than other isomers of cypermethrin. The analytical data can sometimes be used to determine whether the residue may have arisen from use of alpha-cypermethrin. In addition, we might have information on whether alpha-cypermethrin has been used on the crop.

In considering these refinements to the risk assessment, if case by case it is not possible to determine which isomer may have been used, we will assume that the most toxic form was used. As explained above, we only present a detailed risk assessment when dietary intake exceeds the relevant reference dose.

Probabilistic modelling

The standard ‘deterministic’ calculations of consumer exposure used in regulatory assessment and the HSE risk assessments for monitoring work use realistic consumption data and residue levels. However, they tend to overestimate intakes in most circumstances. This is due to the assumptions used, fruit and vegetables would contain high levels of residue in an individual unit and that these would be consumed by high level consumers. They do not take into account the possible range of residue levels and consumption distributions that may occur in reality. These possible combinations of residues and consumption levels can be taken into account using modelling or simulation techniques to produce probability distributions of residue intake levels to indicate the range of consumer intakes, presented as a probabilistic assessment of consumer exposure. These techniques are not yet routinely used to estimate dietary intakes of pesticide residues.

Multiple residues and other developments in risk assessment for pesticides

The risk assessment process is not standing still. HSE is aware that some consumers are concerned by the possible implications of residues of more than one chemical occurring in, say, a single portion of fruit or vegetables or the interaction between mixtures of pesticides and veterinary medicines. The possible implications to health of multiple residues is sometimes called the ‘cocktail effect’. Where more than one pesticide residue is found in a sample, we consider the need for further assessment. The question of which pesticides should be assessed together remains a challenge due to the complexity of the mixtures. In the monitoring work currently, we consider some combinations that we think are a priority (based on toxicological profile or prevalence of the co-occurring residues that are related to one another chemically). If more than one of the following combinations are found together, in the same sample, we will undertake an additional risk assessment:

  • triazoles
  • organophosphates and, or carbamates
  • captan and folpet
  • BAC and DDAC
  • chlormequat and mepiquat

In a ‘first step’ screening assessment approach we will consider whether the sum of the dietary intakes of each pesticide taken together in that commodity (when expressed as a percentage of its own reference value) exceeds a total of 100. If this value is not exceeded, then we do not anticipate that there would be an effect on human health and the assessment is not considered further. If this value (of 100) is exceeded (in the initial screen) then we would present a more detailed risk assessment, including a table to show the dietary intakes of each of the pesticides within the group, in the report. Further information is available on the Consumer Exposure section of the Data Requirements Handbook on the HSE Pesticide website.

International research is aimed at improving the regulatory assessment of mixtures of residues to help understand whether there are any health implications from any observed combinations of pesticide residues in food. In our work, HSE aim to keep our assessments at a high regulatory standard taking account of current knowledge in the field. We will look to adapt as new risk assessment approaches develop. We are keen to ensure our reports reflect consumer concerns.

FSA asked the Committee on Toxicity of Chemicals in Food, Consumer Products and the Environment (COT) to assess these concerns. Their report Risk Assessment of Mixtures of Pesticides was published in 2002.

The Committee concluded that the probability of any health hazard from exposures to mixtures is likely to be small. Nonetheless, it identified areas of uncertainty in the risk assessment process and made recommendations for further work. These fell under the broad headings of regulatory, surveillance, research and public information issues. An action plan to take forward the recommendations was published by the FSA. A number of research projects were commissioned by the FSA to help progress the action plan.

Further to the work done by COT in 2002, combined assessment methodology has been taken forward at the international level, especially EFSA to develop methodology.

Much of the existing recent work on cumulative exposures to pesticides uses probabilistic models and large EU wide monitoring data sets. Notable work includes the EFSA publications on cumulative exposure cited below. On the basis of the work to date, including consumer assessment case studies, EFSA concludes, with varying degrees of certainty for all the population groups assessed, that consumer risk from dietary cumulative exposure is below the thresholds established by EU risk managers. Further information can be obtained from EFSA’s publications, news updates, and FAQs:

Further publications on topics related to consumer risk assessment that are under development are as follows:

HSE (UK) is participating in a number of international initiatives related to residues and risk assessment. This includes:

  • the Organisation for Economic Co-operation and Development (OECD) working group on residue definitions
  • the ongoing JMPR programme of evaluation work/attendance at the Codex Committee on Pesticide Residues (CCPR)
  • participating in JMPR and CCPR discussions of a technical nature on general considerations for risk assessment

Further advances in risk assessment methodology will be taken into account in developing the approach to risk assessments in the future.

  1. Furthermore, HSE will always conclude that on a precautionary basis any findings of genotoxic substances in food are undesirable (please see the explanation regarding genotoxicity in the section on ‘Implications for health’). ↩

  2. Total diet studies (for example, those performed by US FDA (US FDA total diet study program), various Nougadère A et al., publications in Environment International journal on TDS in France). See also EFSA evaluations of chronic exposures to pesticides (2020 and 2021 examples included in the list of EFSA publications on cumulative exposure to pesticides outlined in this section (under ‘Multiple residues[…]’)). Annual EU monitoring data is published on the EFSA website: The 2018 European Union report on pesticide residues in food. ↩

  3. EFSA (European Food Safety Authority), Medina-Pastor P and Triacchini G, 2020. The 2018 European Union report on pesticide residues in food. EFSA Journal 2020;18(4):6057, 103 pp. ↩

Back to top